Vaccines, Blood & Biologics
Resources for You
December 9, 2010 Transcript: Product Development Program for Interventions in Patients with Severe Bleeding Due to Trauma or Other Causes
U.S. FOOD AND DRUG ADMINISTRATION
CENTER FOR BIOLOGICS EVALUATION AND RESEARCH
PUBLIC WORKSHOP: PRODUCT DEVELOPMENT PROGRAM FOR
INTERVENTIONS IN SEVERE BLEEDING DUE TO TRAUMA OR OTHER
Bldg. 10, National Institutes of Health
8800 Rockville Pike, Bethesda, MD, 20894
Thursday, December 9, 2010
LIST OF PARTICIPANTS:
Center for Biologics Evaluation and Research, FDA
JAY EPSTEIN, M.D.
Director of the Office of Blood Research and Review
KEITH HOOTS, M.D.
Director of the Division of Blood Diseases and
Resources at NHLBI
COL. DALLAS HACK, M.D.
Director of the U.S. Army Combat Casualty Care
Research Program and Chairman of the joint
technology coordinating group related to Combat
BASIL GOLDING, M.D.
Director of the Division of Hematology at the
Center for Biologics, FDA
MITCHELL FROST, M.D.
Medical Officer at the Center for Biologics
JAN SIMAK, Ph.D.
Senior Investigator at the Center for Biologics
GEORGE SHASHATY, M.D.
Medical Officer at the Center for Drugs
ROXY HORBOWYJ, M.D.
Medical Officer at the Center for Devices
MARTIN SCHREIBER, M.D.
Chief of Trauma, Critical Care and Acute Care
Surgery at the Oregon Health & Science University
JOHN HOLCOMB, M.D.
Chair of the Department of Surgery
Director of the Center for Translation of Injury
Research at University of Texas at Houston
LENA NAPOLITANO, M.D.
Chief of the Division of Acute Care Surgery
Director of the Critical Care at the Hospital of
University of Michigan
NIGEL KEY, M.D.
Director of Hemophilia Treatment Center
Distinguished Professor of Medicine at the
University Of North Carolina at Chapel Hill
DAVID HERR, M.D.
JOHN HESS, M.D.
JERRY LEVY, M.D.
PETER RHEE, M.D.
Trauma Surgeon at the University of Arizona
AMY SHAPIRO, M.D.
ANTHONY PUSATERI, PH.D.
Research Coordinator of the Hemorrhage and
Resuscitation Group and the Combat Casualty Care
Research Program at the U.S. Army Medical Research
and Materiel Command
CHARLES WADE, PH.D.
Professor at the Department of Surgery
University of Texas Health Science Center
TIMOTHY BILLAR, M.D.
University of Pittsburgh Medical Center
BIJAN KHEIRABADI, PH.D.
U.S. Army Institute of Surgical Research
DANIEL MYERS, PH.D.
JOHN SCOTT, PH.D.
JEFF LAWSON, M.D.
HASAN ALAM, M.D.
ANDREW CAP, M.D.
MITCHELL COHEN, M.D.
JEREMY PERKINS, M.D.
BRYAN COTTON, M.D.
RICHARD WEISKOPF, M.D.
RICK DUTTON, M.D.
JANICE ZIMMERMAN, M.D.
SARA GOLDKIND, M.D.
MICHELLE BIROS, M.D.
JILL BAREN, M.D.
LAURA BROSCH, M.D.
REBECCA PENTZ, M.D.
MARY SOUTHWORTH, M.D.
|OVERVIEW OF AVAILABLE PRODUCTS||20|
|PLASMA AND COAGULATION FACTORS||21|
|PLATELETS AND SUBCELLULAR PARTICLES||25|
|SMALL MOLECULES USED TO AID IN HEMOSTASIS||29|
|DEVICES FOR TREATMENT OF BLEEDING||34|
|CURRENT PRACTICE OF MEDICINE FOR SEVERE BLEEDING||42|
|PATHOPHYSIOLOGY OF TRAUMA||56|
|DEFINITIONS OF BLEEDING||79|
|PANEL DISCUSSION: DEFINITIONS OF BLEEDING||98|
|ANIMAL MODELS FOR PRE-CLINICAL EVALUATION OF|
|PANEL DISCUSSION: ANIMAL MODELS FOR PRE-CLINICAL|
EVALUATION OF HEMOSTASIS PRODUCTS
|CLINICAL TRIAL STATISTICS||202|
|CLINICAL TRIAL STATISTICS- QUESTIONS AND DISCUSSION||215|
|CLINICAL EVALUATION OF PRODUCTS|
|ENDPOINTS FOR CLINICAL EVALUATION OF PRODUCTS|
FOR BLEEDING INTERVENTIONS-LOCAL AGENTS
|PANEL DISCUSSION: ENDPOINTS FOR CLINICAL EVALUATION|
OF PRODUCTS FOR BLEEDING INTERVENTIONS-LOCAL AGENTS
|ENDPOINTS FOR CLINICAL EVALUATION OF PRODUCTS FOR|
BLEEDING INTERVENTIONS-SYSTEMIC AGENTS
|ETHICAL CONSIDERATIONS OF CLINICAL TRIALS|
|PANEL DISCUSSION: ENDPOINTS FOR CLINICAL|
EVALUATION OF PRODUCTS FOR BLEEDING
* * * * *
DR. VOSTAL: Good morning and welcome to the National Institutes of Health in Washington, D.C. My name is Jaro Vostal, and we're going to get started in a couple of minutes. I just wanted to run through a few logistics issues.
So there will be several coffee breaks and lunch throughout the day. For that there is a cafeteria downstairs and a cafeteria upstairs. There is also a coffee shop if you go straight back to the main entrance of the new hospital. The restrooms are on the -- in the lobby on the left and right side. And in terms of audience participation, so we try to design this workshop to get as much discussion of the different viewpoints as possible.
So we're going to have some -- several panel sessions which will be addressing talking points that you'll be able to see on the screen. If you have some questions, you can raise them at the end of the panel sessions either through the microphone or there will be 3x5 cards which are -- have been placed in your folders and you can pass those to the side and our volunteers will pick those up. And the panel moderator will then have an option of taking those questions or modifying them as he chooses.
And finally, we also designed this workshop to have it so we would have some kind of a practical output at the end of the 2 days that we're going to spend discussing this topic. So tonight, the panel chairs and the panel members are going to get together after the meeting to summarize what was presented today and that will be presented on the second day in the last session. So for those people who are going to be doing the summaries, we will meet at the outside of the doors and go to a separate building where we have conference rooms for discussions.
With that I would like to introduce Dr. Jay Epstein who is going to start off -- start us off with some welcoming remarks. Dr. Epstein is the director of the Office of Blood Research and Review at the FDA.
DR. EPSTEIN: Thank you very much, Jaro. It's a great pleasure to be here and to offer a few welcoming remarks. As you know from your program this is a cosponsored workshop which has been developed cooperatively by the FDA Department of Defense and the NIH, National Institute of Heart, Lung and Blood. And our goal here is to see if we can discuss the issues that would govern a rational scientific and regulatory program for the development of products for improving outcomes of severe bleeding related to trauma or other causes.
So it's a fact of life unfortunately that severe trauma occurs both in civilian and military populations and it's been recognized that despite major advances in therapies there is still very high mortality and severe trauma related specifically to bleeding and consequences of bleeding and that this is a public health issue both from a civilian and a military readiness perspective. So there is a strong interest to see if we can make progress in this area.
Those of you familiar with the field who I assume are the majority of those here today already know that the physiology of severe bleeding is complicated that despite millennia of dealing with severe bleeding, we're still in a process of trying to understand the overlapping effects of blood loss, coagulopathy, tissue hyper perfusion, and shock. And that indeed treatment of severe bleeding is part of the disorder in the sense that large volume transfusion particularly crystalloid solutions can contribute to coagulopathy and hypothermia and indeed complicate the pathophysiology.
The issue of product development though is itself complicated because of these complexities of the disorder and part of the problem has been a lack of consistent definitions of trauma -- of associated pathological states, which has made it difficult to identify target populations especially in prospect and therefore to get good scientific answers from studies.
So this workshop is a multidisciplinary effort, which as I say has been jointly sponsored by several involved organizations. It builds upon a previous workshop that was convened by the NHLBI in April of this year, which focused on identifying research opportunities in this field and of course funding opportunities. The objective of this workshop is to develop a practical approach for FDA so that we can be more science-based as we advise candidate product sponsors on the design of trials.
I think it's self-evident FDA regulates the devices and the medical therapeutics for treatment of severe bleeding among other medical conditions. Of course, we need to be consistent in our approach to product evaluation especially for products that have similar indications and so we're interested in developing a transparent and of course, science-based approach to the evaluation of products, which have the intended use for treatment of severe bleeding, whether due to trauma or other causes.
So in this workshop -- and I should just mention that a multi-organizational and multi-disciplinary scientific committee has worked very, very hard over approximately the last 6 months to bring together the necessary expertise to shed light on the issues in this field including hematology, surgery, critical care, and indeed ethics so that we can try to make progress.
So as I said we have a joint effort today among cooperating organizations and these brief remarks on my part will be followed by Dr. Keith Hoots, who is director of the Division of Blood Diseases and Resources at the NHLBI and then some opening remarks also from Colonel Dallas Hack, who is director of the U.S. Army Combat Casualty Care Research Program and the chairman of the Joint Technology Coordinating Group related to Combat Casualty Care. So with that I'll just invite Dr. Hoots to make his own welcome. Thank you very much. Glad to see such a good turnout for such an important issue.
DR. HOOTS: Thank you, Jay. Welcome to everyone and I probably should say for -- probably the majority of the people in the audience, welcome back to NIH. And it's our pleasure to be a cosponsor of this meeting with the sponsor FDA and with our other cosponsor, the Department of Defense and Department of the Army. And Dr. Epstein told you about some of the groundwork that has been laid beginning at least with the -- this workshop that we put together in April to -- that was focused on trauma-induced coagulopathy.
And obviously, we and NHLBI and in the Division of Blood are very focused on a number of issues relevant to today's discussion in particular transfusion, in particular the coagulation defects that come from traumatic injury. And it's my pleasure to say that we've also, as a result of these efforts to put together some of the programs that you've all participate in -- are participating in today that as an outgrowth of this a number of new collaborative initiatives are developing.
We are working closely with the Department of the Army on a couple of these and with preexisting infrastructure from NHLBI and from the military putting together programs in both transfusion and the Resuscitations Outcome Consortium that we host at NHLBI.
So this is a really important issue for us and we're delighted to be a part of this process. And so I don't want to belabor this point other than to say that we truly value your input and obviously that's why we have these meetings. We know how busy all of you are. We appreciate your coming here, in many cases great distances to provide us your insight, your advice, and also to work collaboratively with your governmental agencies to try to push forward progress in the field of response to acute injury in the traumatic venues of the military and civilian lives. So with that I'll follow up with Dr. Hack -- Colonel Hack.
DR. HACK: Thank you, Jay and Keith, and thank you all for coming. Welcome and I really appreciate the effort that you all have made to come and help us to understand and discuss the issues related to this. This is a somewhat difficult field because in our profit driven, you know, Wall Street driven health care system now essentially it's hard to get some of the bigger companies to get invest -- to invest in this area because of essentially the transient nature of the problem. The people have this and they go away, it is not a 30-year course of taking a blockbuster drug.
And so we need to work very creatively to figure out how it is to move this field forward to reduce that -- some of that risk by making it more predictable for people to work in this area and to actually work at really doing again, as Jay eloquently mentioned here, reducing some of the deaths that occur that -- we believe that can be prevented if we get some new products and techniques in terms of taking care of those folks.
I really look forward to the discussion and to the outcome -- the report that comes. I also -- I'd just say that it's been good; I see many of you folks at some of the conferences we hold and just again let you know we'll be moving our ATEC conference this year. We'll be moving to Fort Lauderdale still at 15th -- 18th of August and it's a really good opportunity to get together with the people actually out there, the trauma surgeons and those folks actually out there actually saving the lives. And so again and welcome you to that and I look forward to working with you on this the next couple of days. Thank you. With that, you'll introduce --
DR. VOSTAL: Thank you very much for those welcoming remarks and our next speaker will be Dr. Basil Golding who is the director of the Division of Hematology at the Center for Biologics, FDA.
DR. GOLDING: Thank you, Jaro, and welcome to everybody. As you all figured out by now, we're talking about product development programs for interventions due -- for bleeding, severe bleeding due to trauma or other causes. Raison d'etre, why are we here? The main reason is -- are outlined here that there are and you've already heard that there are a significant number of deaths due to severe hemorrhage and severe bleeding and this occurs both in the civilian and military populations and importantly the idea is that some of these may be reversible. And the important aspect of this meeting is to try and figure out how to develop products to treat this condition. And FDA does anticipate that products for this indication will be submitted and we need to know how to regulate and how to assist these products.
The difficulty and the challenge here is that they are -- there is a lack of clear definitions, lack of clear entry criteria, lack of clear endpoints and a path forward for drug development. So the main purpose of this workshop is to try and get answers to these challenging questions. The focus of the workshop was discussed in a lot of detail with the steering committee and a lot of thought I think went into deciding how to organize this meeting. And what I've listed here are some of the causes of severe bleeding. I've listed trauma, surgical, elective surgery such as cardiovascular or liver surgery, and medical causes such as clotting factor deficiencies, patients on anticoagulation, and anatomical causes such as GI bleeding, intracranial bleeding. And of course there is also obstetrical bleeding. Now when thinking about the different causes of bleeding, it's clear to us and you'll see the numbers during the presentations that the most important cause in terms of numbers is bleeding due to trauma. So an active decision was made to focus this meeting on trauma since this seems to be, appears to be, a more urgent public health issue both in the civilian population and in the military.
Other conditions may be considered if they are informative to trauma. And we will -- hopefully there will be some discussion of other conditions such as elective surgery. And in a drug development program there may be important lessons to be learned from other types of severe bleeding, which can be informative and can lead to improved study designs for phase III studies in trauma.
So the critical questions that are going to be asked at this meeting, and in fact the sessions correspond with these questions, so what is severe bleeding? What -- in other words, what is the definition and what are the eligibility criteria for enrolment into a trauma study for severe bleeding. So the items one and two that I've listed here should be covered in session one, which relates to the definitions of severe bleeding.
The second session deals with animals. So the question there is what animal studies are relevant. And of course, it is important to try and develop animal models, which mimics the target population. And a lot could be learned from animal studies. But there's an additional point about animal studies that when you read the regulations of Waiver of Informed Consent, it clearly states in those regulations that animals studies and other evidence are required to show a prospect of benefit before you can embark on a study with Waiver of Informed Consent. So that overlaps with the ethical issues, which will also be discussed in detail in a session in this meeting -- at this meeting.
So what are the gaps? There are gaps in technology, and you'll hear some of the newer methods that are being used to assess trauma victims early on and quickly so that this can be used. The critical point here is that this can be used in deciding on enrolment and monitoring of these patients. And the other large gap is the evidence of correlation with the outcome.
So you'll be hearing the trauma surgeons and other experts in the field discussing all the parameters that are used to assist these patients and to the -- and what are the -- what is the thought process that goes into deciding on what endpoints should be used. And the issue there is how much evidence is there to correlate the parameters that you have with the actual outcomes -- mortality or other outcomes.
So in terms of what do we expect, what does FDA hope to achieve with its cosponsors at this workshop? We are hoping that the information that is gathered here and gained here will suggest how to move forward for drug development programs. We need to be realistic that not all the answer -- not all the questions will be answered, but that we will be able to, I hope, at least identify where the gaps are. And by identifying them, we will -- hopefully, we will be able to figure out ways to resolve those issues so that then this field can move forward.
We're also hoping after the meeting that -- there will be a transcript, and we hope to publish the summary of the meeting. So I'd like to end and I have special thanks to the steering committee. As pointed out, their worked -- invested a lot of time and effort in developing the program and I think we are very fortunate to have such a talented group of people in the field here today and presenting their studies and presenting us with their ideas of how to answer these challenges.
And I'd also like to thank our cosponsors, Health and Human Services, National Heart, Lung, Blood Institute, and the Department of Defense. And additional thanks go to our colleagues in CDRH and CDER who will also be presenting here. And a special thanks goes to Jennifer Scharpf and her elves who have put together this meeting and they have done a wonderful job. Thank you for your attention.
DR. VOSTAL: Okay, thank you, Dr. Golding.
OVERVIEW OF AVAILABLE PRODUCTS
DR. VOSTAL: So our next set of speakers are going to be representatives from different centers of the FDA and they are going to talk about the products that are currently on the market with the indications for severe bleeding or therapeutic treatment of bleeding. So our first speaker is Dr. Mitchell Frost who is a medical officer in the Center for Biologics.
PLASMA AND COAGULATION FACTORS
DR. FROST: Good morning. Thank you, Jaro. As Jaro said, I'm Mitch Frost. I am a medical officer in CBER, Office of Blood, Division of Hematology and I'd like to list for you the plasma and coagulation factors that are currently available for the treatment of bleeding along with their major relevant indications.
So to begin with, in the context of the next two days, I'd like to just start with the statement that to-date there are no biologics licensed in the U.S. for use in a trauma situation to control bleeding or that carry the specific indication for use in severe bleeding. Available products for use in the treatment of bleeding include plasma and cryoprecipitated components.
To begin with, it's fresh frozen plasma, which carries the indications of management of preoperative or bleeding patients who require replacement of multiple plasma coagulation factors such as in liver disease or in DIC and also patients undergoing massive transfusion who have clinically significant coagulation deficiencies. Also in this category is cryoprecipitate, which carries the indication of control of bleeding associated with fibrinogen deficiency and to treat Factor XIII deficiency.
Continuing with the available products in this category are coagulation factors which can be either plasma-derived or recombinant. To begin with is the Factor VIII products which carry the general indication for prevention and control of bleeding in patients with hemophilia A. Examples of these products include Advate, Helixate FS, Recombinate, Xyntha, Kogenate FS, Hemofil M, and Monoclate-P.
In the Factor IX category of products, the general indication is for prevention and control of bleeding in patient with hemophilia B. Examples of these products include BeneFIX, AlphaNine SD, and Mononine. Continuing is the category of Factor VIII and von Willebrand factor concentrate products, which carry the general indication for treatment of bleeding in patients with von Willebrand disease. Examples of these include Willate, Humate-P, and Alphanate.
And then lastly in this category is activated Factor VII and prothrombin complex concentrate products or bypassing agents, which carry the indication for treatment of bleeding episodes in hemophilia A or B patients with inhibitors. Of these are NovoSeven which is a recombinant Factor VIIa and FEIBA which has both Factor VIII inhibitor bypassing activity and prothrombin complex factors.
Also available are topical agents to control mild and moderate bleeding during surgical procedures. The first category is the thrombins and the indication here is as an aid to homeostasis, whenever oozing blood and minor bleeding from capillaries and small venules is accessible and control of bleeding by standard surgical techniques is ineffective or impractical. Examples of these are Evithrom, which is derived from human origin, Recothrom is a recombinant product, and Thrombin-JMI is derived from bovine.
Continuing with topical agents are the fibrin sealants, which among other things contain plasma-derived fibrinogen and thrombin. TISSEEL carries the indication as an adjunct to homeostasis in surgeries involving cardiopulmonary bypass and treatment of splenic injuries. And then there is EVICEL and TachoSil which carry the indications as an adjunct to homeostasis for use in patients undergoing surgery when control of bleeding by standard surgical techniques such as suture, ligature, or cautery is ineffective or impractical.
So in summary, the current products available to treat bleeding include FFP, cryoprecipitate, the thrombins and fibrin sealants, replacement factors, and bypassing agents for congenital conditions such as hemophilia A and B, and patients with inhibitors. And as we have seen, these products are generally intended for bleeding due to specific congenital factor deficiencies and as an adjunct to homeostasis for control of mild to moderate bleeding during surgical procedures.
And once again, to date, there are no biologics licensed in the U.S. for use in a trauma situation to control bleeding or that carry the specific indication for use in severe bleeding. Thank you.
DR. VOSTAL: Thank you, Dr. Frost. Our next speaker is also from Center for Biologics. The speaker is Dr. Jan Simak, who is a senior investigator and he will cover products related to platelets and subcellular particles.
PLATELETS AND SUBCELLULAR PARTICLES
MR. SIMAK: Good morning. Let me make a brief overview on platelet transfusion products. Let's start with current FDA licensed platelet products. We have, in general, two types of products. The first type is aphaeresis platelets containing more than 3x10 to the eleventh platelets per unit. They are mostly stored in plasma, and new now are also available, stored in plasma with additive solution.
The second type of product, platelet product, are whole blood derived platelets prepared by platelet-rich plasma preparation methods. They contain more than 5.5x10 to the tenth platelets per unit and they are available for pre-storage pooling or post-storage pooling. They can be leukocyte reduced or irradiated. And with regard to storage in open systems, the storage is allowed up to 4 hours in closed system. It depends on the products, 24 hours up to 5 days and at room temperature 20 to 24 degree Celsius.
What about CBER regulatory mechanism? Blood components are biologics and their clinical investigations are performed the same way as drugs -- for drugs by Investigational New Drug Application under IND. CBER also, Center for Biologic also regulates devices, the transfusion devices for collection, processing, and storage and testing the transfusion blood products and these studies are performed under Investigation and Device Exemptions, IDEs.
With regard to licensing, blood components are licensed by biologic license applications, BLAs. Anticoagulants and additive solutions in bags are approved by new drug application mechanism, NDA and transfusion devices are approved by mechanism of pre-market approval or cleared by pre-market notification.
We have two guidances related to this topic available. The first guidance is for collection of platelets by automated methods from December 2007 and we also have a draft guidance for platelet testing and evaluation of platelet substitute product from May 1999.
What about potential future products? And those informations are in public domains. That's -- what I'm talking about are only published studies. They are not -- these products are not FDA approved.
So we know that there are many studies available on process, platelets or pathogen-reduced platelets or cold stored platelets and there are studies on platelet substitutes or platelet-derived homeostatic agents like cryopreserved platelets or freeze-dried lyophilized platelets. Cryopreserved platelets are prepared by different preparation methods mostly by a DMSO or mostly stored in DMSO, about six percent. There are different methods of washing or no washing prior to infusion.
We know from studies that these platelets are extensively altered compared to normal platelets. They have low in vivo recovery. They have activated procoagulant phenotype and high in vitro procoagulant activities. And there are several clinical studies in small group of patients and anecdotal reports.
With regard to freeze-dried platelets, they are prepared usually by mild aldehyde stabilization, lyophilization, or rehydration. They retain partially the ultrastructure and morphologic and functional characteristics. They have enhanced procoagulant properties. And we have a couple of studies newly -- a good study tested in animal models recently published in the journal Thrombosis and Haemostasis.
Conclusions about cryopreserved and freeze-dried platelets they are possible -- they have possible useful indication in hemostatic agents in severe bleeding in trauma and surgery. However, manufacturing process needs to be optimized, product characterization, QC, and potency assays need to be developed.
They are potentially thrombogenic and therefore the benefit-risk balance has to be evaluated and it's critical. And safety and efficacy need to evaluated in adequately controlled surgery, trauma, clinical trials. Thank you for your attention.
DR. VOSTAL: Thank you, Dr. Simak. Our next speaker is Dr. George Shashaty, who is a medical officer in the Center for Drugs. And he will talk about Small Molecules Used to Aid in Hemostasis.
SMALL MOLECULES USED TO AID IN HEMOSTASIS
DR. SHASHATY: My name is George Shashaty. I'm a medical reviewer in the division of Hematology Products in CDER and will be presenting a very short list of drugs that are approved for purposes of controlling bleeding.
Aminocaproic acid was approved in 1964. Its mechanism of action is in its ability to inhibit plasminogen activator and plasmin. It has a short half-life. And it's excreted mostly unchanged in the urine.
The indication statement is very old. And because the indication is written with examples of conditions where fibrinolysis may be contributing to bleeding could probably be administered in a host of hemorrhagic situations, the approved dose of the drug, whether given orally or intravenously as shown on the slide. Because of its short half-life, the dosing is almost continuous. Wrong pointer.
Tranexamic acid was approved in 1986. Similar to aminocaproic acid, its mechanism of action is in its ability to inhibit plasminogen activator and plasmin. It also has a short half-life and is excreted mostly unchanged in the urine. In contrast to aminocaproic acid, its approval indication is extraordinarily narrow. And its use for the approved indication probably represents a tiny fraction of its use in the United States.
Although it was formerly available in oral form, the commercial sponsor for the Reference Listed Drug discontinued the oral form several years ago. However, there is a newly approved oral form of tranexamic acid for the indication of heavy menstrual bleeding. The approved dose of the drug is shown on the slide.
Trasylol or aprotinin was approved in 1993. It is a polypeptide protease inhibitor that has antifibrinolytic and systemic immune response activities. It has a half-life of about 2-1/2 hours and is filtered and reabsorbed in the kidney leading to renal accumulation.
Its specific approved indication is for the prophylactic use to reduce perioperative blood loss and the need for blood transfusion in patients undergoing cardiopulmonary bypass in the course of coronary artery bypass graft surgery who are at increased risk for blood loss and blood transfusion. The approved dose includes a loading dose, a pump prime dose, and a maintenance dose.
In 2006, several publications describing adverse reactions associated with the use of Trasylol led to a reconsideration of its benefit risk assessment. After publication of the BART trial in which the benefits and risks of Trasylol were compared to those of aminocaproic acid and tranexamic acid, the sponsor voluntarily withdrew Trasylol from marketing. The drug is available under a compassionate use IND.
Progress in regulatory activities has advanced considerably over the last few decades particularly as to the evidence required to assure the efficacy and safety of a drug prior to approval. This includes well-defined enrollees in the study population, agreement to an evaluation of clinically meaningful endpoints, the use of randomization and blinding to reduce bias, sufficient population size to assess adverse reactions, formalization of data collection to validate observations and pre-specify statistical analysis plans.
For example, earlier approvals of antihemorrhagic drugs were based on evidence found in published literature and personal testimonials, whereas approval is now based on data submitted from adequate and well-controlled trials. Approval for use of a drug is restricted to the patients who are similar to those included in the trials since what is effective and safe in one group may not be so in a different group.
In contrast to earlier studies, the sponsor must now demonstrate a benefit on a clinically meaningful endpoint rather than on a soft endpoint or a surrogate end point. For example, Trasylol was approved because of adequate and well-controlled trials demonstrated that its use in the approved indication led to a decrease in the number of units of blood transfused per patient and the percent of patients requiring transfusion. Most late-phase trial designs require randomization and blinding to minimize bias in the assessment of efficacy and safety endpoints.
FDA does not regulate the practice of medicine. After a drug is approved by FDA, a physician is permitted to use that drug for purposes which he or she believes are clinically useful for a patient. Off-label use, therefore, is very common. For example, tranexamic acid use is not restricted to patients with hemophilia who are undergoing tooth extractions. Oftentimes novel dosing and administration schemes are developed and promulgated in the literature. For instance, although the dose and administration of Trasylol remained fairly standard, its use was being extended to other surgical populations in whom large blood loss is frequent such as spinal surgery, liver transplantation, and orthopedic procedures.
The dosing of tranexamic acid in clinical use has varied by a factor of as much as 10 based on publications and virtually no one uses oral route of administration for the indication. Results from the CRASH-2 trial are likely to lead to an increase in the use of tranexamic acid in trauma patients.
As a result of these further studies, most of which are not submitted to the FDA for review, the efficacy and safety of the use of a drug is blurred. And this makes it less likely that a rational benefit safety assessment can be established for a particular use in an off-label use. Thank you.
DR. VOSTAL: Thank you, Dr. Shashaty. Our next speaker is Dr. Roxolana Horbowyj, who is a medical officer in the Center for Devices.
DEVICES FOR TREATMENT OF BLEEDING
DR. HORBOWYJ: Good morning. I'm a general and critical care surgeon practicing locally and a medical officer in CDRH. This presentation will briefly describe to you medical devices intended for use or indicative for use to control bleeding or achieve hemostasis. This presentation will also go over some of our background regulations, definitions, examples, and points to consider for device evaluation for indication for use to control bleeding or achieve hemostasis.
The reason that I'm going through some of this history is because while the FDA can trace its roots back to about the 1800s and in 1938 the Food, Drug, and Cosmetic Act was passed, this did not address devices. And the Medical Device Amendment Act was enacted in 1976, has provided FDA the authority to ensure reasonable assurance of safety and effectiveness of devices through classification, pre-market notification, pre-market approval, and post-market controls.
And this Medical Device Amendment provided us a definition of a medical device, which is listed here and I'll read it quickly. "An instrument, apparatus, implement, machine, contrivance, implant, in vitro reagent, or other similar or related article, including any component, part, or accessory, which is intended for use in the diagnosis of disease or other conditions or in cure, mitigation, treatment, or prevention of disease, in man or other animals, or intended to affect the structure or any function of the body of man or other animals and which does not achieve its primary intended purpose through chemical action within or on the body of man or other animals and which is not dependent upon being metabolized for achievement of its primary intended purposes."
This amend was the basis of our Code of Federal Regulations and devices are in 21 CFR Part 800 to 1299. Our regulations describe the device types that existed prior to May 28, 1976. So they're referred to as pre-amendment devices. And these were classified by the 1976 classification panels. They're appropriately 1,700 general categories of classified medical devices in our regulations.
Our regulations also provide us a definition of an implant. An implant means a device that is placed into a surgically or naturally formed cavity of the human body and is regarded as an implant for the purpose of this part only if it is intended to remain implanted continuously for a period of 30 days or more unless the commissioner determines otherwise in order to protect human health. The regulations also give us a basis for device classification and -- for three risk-based regulatory controls. Class I uses general controls, class II special controls, and class III pre-market approval.
A medical device is assigned a product code. Our product codes are unique three-letter constructs used by CDRH to differentiate device types within a given regulation across our center. New indications for use or new technologies are assigned new product codes that are placed under the original regulation.
An important aspect of these are our definitions of Indications for Use and Intended Use. The Indications for Use is defined as a general description of disease or condition the device will diagnose, treat, prevent, cure, or mitigate including a description of the patient population for which the device is intended.
This description usually also includes whether the device is a prescription device or an over-the-counter device if -- where the part of the body or type of tissue applied to or interacted with, its frequency of use and its physiologic purpose, for example, does it remove water from blood or does it transport blood.
We also have a definition of the Intended Use, which is distinct -- somewhat distinct from the Indications for Use. This is defined as the objective intent of the persons legally responsible for the labeling of devices determined by such persons' expression or may be shown by the circumstances surrounding the distribution of the article.
So this definition of intent and Intended Use is a statement, an overall assessment by FDA that encompasses all aspect of how and for what purposes and under what circumstances the device is intended to be used. So now looking at devices intended to control bleeding, you'll find that devices intended to control bleeding are found under various regulations, they're also found in different procodes and some are found by their intended use.
So for example, these are two regulations that we use. The upper regulation is for an absorbable hemostatic agent. It's identified in its regulation as a device intended to produce hemostasis by accelerating the clotting of blood. It's an absorbable device. Its class is class III, so it needs pre-market controls.
This is a regulation for gastrointestinal tubes and accessories identified as a device that consists of flexible or semi-rigid tubing used for instilling fluids into, withdrawing fluids from, splinting or suppressing bleeding in the alimentary tract. And this is regulated by class II regulations, which require special controls.
Devices intended for the use of -- intended for use of control bleeding are also found then under dental devices, ear, nose, and throat devices, general plastic surgery devices, neurologic devices, and ophthalmic devices.
Some of our devices are not classified under regulations, and they go by procode, for example, procode FRO is for dressings, procode MGB device haemostatic vascular unclassified device.
The other devices that may be indicated for use to control bleeding or achieve hemostasis may be found in this series of regulations for non-absorbable gauze and sponges for external use, hydrophilic wound dressings, implantable clip, removable skin clip, non-absorbable gauze for internal use, implantable staple, removable staple, manual surgical instruments for general use and nonpneumatic tourniquet.
Many of the devices then are regulated for general use to control bleeding or achieve hemostasis. But devices that may be indicated for use to control bleeding or achieve hemostasis for specific bleeding severity, duration of use, and anatomic site maybe looked at this way. For minimum and minor severity of bleeding, we have absorbable and non-absorbable devices that are of various class. They may be prescription and non-prescription.
Devices that are indicative for use for mild to moderate severity of bleeding in order to control that bleeding or achieve hemostasis in those conditions or absorbable devices either implanted, class 3 and available only by prescription.
For moderate to severe bleeding, currently marketed devices are non-absorbable, temporary, external use only, unclassified devices available by prescription and non-prescription. At this time there are no marketed devices in the U.S. for severe to massive bleeding.
So points to consider for device indication for use to control bleeding or achieve hemostasis across our regulations and across our procodes consider things that are in the following list. So while we have some well defined statements for our indications for use or intended use in other aspects that you saw and, for example, for what is prescription versus what is non-prescription, some of the other terms are not very well defined as you have heard by other speakers, but are important to consider.
For example, the conditions or disease to be treated, we consider the bleeding severity and co-morbidities, the parts of the -- or type of tissue -- the parts of the body or type of tissue applied to interact with external versus internal. A better definition could probably be found here; physiological purpose, frequency of use, whether multiple exposure or single exposure, duration of use.
While we have a definition for an implant being more than 30 days, temporary or otherwise is not formally defined. And then it's important to consider, in looking at the indications for use, the patient populations' characteristics; whether the patient is an adult or pediatric patient, whether the patient is found in an open field or under controlled sterile conditions, for elective or emergency health need and has compressible or non-compressible bleeding sources.
And our proposed indications for use are expected to be supported by the date of presented. So we look forward to this workshop and your input from it. Thank you very much for your attention.
DR. VOSTAL: Okay. Thank you, Dr. Horbowy. So that concludes the summary of the products that are FDA approved and on the market. To get an idea of what the real world practice is, we asked Dr. Martin Schreiber, who is the chief of Trauma, Critical Care and Acute Care Surgery at the Oregon Health & Science University to come, tell us about current medical practice.
CURRENT PRACTICE OF MEDICINE FOR SEVERE BLEEDING
DR. SCHREIBER: Good morning. It's a pleasure to be here this morning, and speak to you about some systemic hemostatic agents and non-FDA approved uses. I think the first point I want to make is that these are all adjunctive measures. The most important thing we have in trauma is a qualified surgeon with a $0.50 silk suture and hopefully that does not require FDA approval.
In terms of the importance of the matter this has been alluded to, but I just want to show you a little bit of data. This is a very nice study that came out of San Antonio that looked at 753 consecutive deaths after trauma, including those that didn't make it to the hospital as well as those that did. The most common causes of death are actually CNS related. About 70 or 67 percent of deaths after trauma are related to CNS events including brain injury and spinal cord injury, but hemorrhage contributes to about 37 percent of all trauma deaths.
And it turns out that both in the military and in civilian practice hemorrhagic death is the most common cause of preventable death and therefore the most important to us in terms of what we can do for patients. Now where most of the money that comes from NIH goes is really in this area, multiple organ failure and that makes up about 9 percent of deaths.
In terms of how quickly do patients die, I think this is a very important point, this is also from the paper by Stewart from San Antonio, each of the little dashes on the bottom of this graph is 12 hours. And what you see here is that about 50 percent of the patients who die after trauma are dead within 12 hours. And that number quickly goes up to 70 percent by 36 hours. So the interventions that we are using to save lives need to be available early after injury and rapidly invoked.
I'm going to talk about the agents that are listed here. These are the ones that we are currently using in practice in the United States. I'm going to talk about them in the order that they are listed because this is the order in which we have the most experience. I'm not going to talk about hemostatic agents particularly hemostatic dressings that are approved primarily for external use, but also being used by trauma surgeons across the nation for internal use. I'm not going to talk about fresh whole blood or platelets derived in theater that are currently being used and also not FDA approved.
So the drug that we have the most experience as American trauma surgeons is recombinant Factor VIIa. VIIa is a naturally occurring coagulation factor. It functions on the surface of injured endothelium where tissue factor is exposed and de-encrypted. VIIa binds tissue factor, activates Factor X to Factor Xa which cleaves prothrombin to thrombin.
Thrombin then results in activation of platelets, and these activated platelets serve as a surface upon which this process can be magnified hundreds of thousands of times resulting in a thrombin burst which is occurring at the site of injury, which then results in fibrin production and theoretically this is all occurring where it's most needed, where the injury is occurring.
So recombinant Factor VIIa is approved for use in the United States in haemophiliacs with inhibitors to either Factor VIII or Factor IX. The first reported use in trauma occurred in an Israeli soldier who had a gunshot wound to the inferior vena cava. Shortly after this paper came out, there was a series of retrospective studies, and it was really felt by many American trauma surgeons this was the golden bullet, this was the thing that we've been looking for, for millennia; this is what we needed to save patients' lives. We all started using it. It turned out that in our hospital, Oregon Health & Science University, this drug was the drug that we spent the most money on in 2004. So, rapid accelerated use of this drug across the country with actually very little data.
We do have randomized clinical trials in multiple areas, trauma and traumatic brain injury, spontaneous intracerebral bleed, retropubic prostatectomy liver transplants, surgery -- cardiac surgery and spinal surgery, all of these have been studied in a prospect to randomized fashion.
In terms of the first trauma trial that was done around the world, there was a prospect to randomized trial. The drug was given after the eighth unit of packed red blood cells was transfused. The dosing was 200 micrograms per kilogram, followed by 100 micrograms per kilogram at 1 and 3 hours. These are larger doses that have been previously studied. And blunt and penetrating traumas were analyzed separately. There were 266 patients in this trial. The trial was stopped for futility. The primary endpoint of the trial was mortality at 28 days.
These are some results of the trial. There was no difference in mortality in either the blunt or penetrating trauma groups. However, there was a significant reduction in blood products used, particularly in the blunt trauma group. There was no difference or significant difference in reduction in blood products used in the penetrating group.
This was felt by many to be a negative study because of the absence of difference in mortality, but there was a reduction in blood products used. Also significantly in the blunt trauma group, there was significantly less massive transfusions given to those patients who received Factor VII.
In terms of complications, many complications were studied. The only complication that was interesting, that was different was there was a reduction in the ARDS rate in the blunt trauma population; again, no differences in the penetrating trauma population.
Again, this was felt to be a negative study due to the absence of difference in the primary endpoint being 28-day mortality. So a second trial, the control trial, was performed and actually just recently published in the Journal of Trauma. The difference between this trial and the prior trial is that the drug was now given between the fourth and eighth unit of packed red blood cells with the thought being that if the drug is given earlier, there is a greater opportunity for it to be efficacious.
This was also a larger trial, again, with a primary end point of mortality at 28 days. This trial was also stopped early for futility due to the absence of difference in mortality. Again there was no difference in 30-day mortality in either the blunt trauma population or the penetrating population.
Once again there was a reduction in blood product used primarily in the blunt population, but actually in the penetrating group there was some reduction as well. There were no significant difference in any morbidity parameters including multiple organ failure or single organ failure. However, multiple organ failure at day 30 in the blunt group very close -- was very close to reach significance.
There have been several trials that have been coming out recently that are meta-analysis looking at complications. This was a -- this is a New England Journal paper that actually just came out within the last couple of months. In this paper they were particularly concerned looking at thromboembolic events. And you can see here there was no difference in overall thromboembolic events between those that received recombinant Factor VIIa and those who received the placebo. Again this is a meta-analysis for a very large trial that includes all the prospect to randomized trials that I alluded to earlier.
However, if one looks at arterial events, there were six significantly more arterial events in the group that received recombinant Factor VIIa. If one goes further and breaks the population down into various age groups, you can see that for those patients who were greater than 65 years old, there was a significant increase in serious arterial events particularly high risk populations like those with either cerebrovascular disease or coronary artery disease. These patients have a higher risk of having stroke or MI.
I want to move on now to talk about tranexamic acid. You heard it mentioned earlier today. Tranexamic acid is a synthetic derivative of lysine. It has a high affinity for plasminogen which blocks the binding side of fibrin and prevents the breakdown of fibrin clots. That's the mechanism of action. The approved uses again are for tooth extraction in hemophiliacs that was in 1986, and then for menorrhagia in 2009 and that's with the oral form.
There are several clinical trials that talk about the use of tranexamic acid particularly those with total hip replacement and cardiac surgery which show reduced blood loss. These are prospect to randomized trials. They are very small trials on the line of about 40 patients. A Cochrane review which recently came out stated that there is insufficient evidence to either support or refute use and further randomized clinical trials are needed.
This is the CRASH-2 trial. This trial came out while I was in Afghanistan; the day after this trial came out, I received multiple phone calls and e-mails from the Pentagon and high ranking generals asking me why we weren't using this in our soldiers. And the reason I was getting these e-mails and phone calls because I was a Joint Theater Trauma System director that was responsible for making the clinical practice guidelines for use in theater.
So the CRASH-2 trial is a extremely large trial. It involved 274 hospitals in 40 countries, 20,000 patients, a huge trial. How was this trial done? Adult trauma patients who were within 8 hours of injury who either had significant hemorrhage or who were believed to be at risk of significant hemorrhage were eligible for the trial.
Randomization was by indecision. This is stated in the paper itself which means that if the physician thought the patient was likely to need an antifibrinolytic agent, they were given that antifibrinolytic agent and the patient was not included in the trial.
Alternatively if the physician felt that the patient was not a candidate for antifibrinolytic agent, the patient was not given the agent and again they were not eligible. It was only in situations where the physician was unsure whether or not the patient should receive the drug that the patient was randomized.
And they either randomized the tranexamic acid or a placebo. The loading dose was 1 gram given over 10 minutes and the maintenance dose was 1 gram over 8 hours. Again this was a huge trial; 20,000 patients were randomized and what you can see here is very few patients were lost to follow up. So at the time of follow up, there was over 10,000 patients in both groups.
The most important finding of this study was there was overall about a 1.5 percent reduction overall mortality and with this huge trial that was strongly, statistically significant at the 0.0035 level.
In terms of the causes of death, there was a major reduction in bleeding causes of death in the patients who received tranexamic acid. There was no difference in vascular occlusion deaths, multiple organ failure deaths, head injury deaths or other deaths in this population.
Again though there was about a 1.5 percent reduction overall mortality in this prospect to randomized trial. Interestingly, there was no difference between the groups in thromboembolic events regardless of if you look at DVT, PE, MI or stroke or any thromboembolic events. And in fact there were trends towards less events in the patients who received tranexamic acid.
Overall, there was no difference in blood transfusions given to the two groups with about six units transfused both the tranexamic acid and the placebo group, this being interesting of course this drug is proposed to be working by reducing blood loss yet there was no difference in transfusions given. One possible reason for this is the fact that if a patient survives they are likely to have greater blood loss.
Moving on in the prothrombin complex concentrate, this is the drug that we as trauma surgeons in the United States have the least experience with. The drug that is available in the United States is Profilnine. This is a three factor concentrate, it contains the vitamin K dependant Factors II, IX, and X. There is minimal Factor VII present and the Factor VII that is present is not clinically significant.
Many other countries such as Australia, Great Britain, Israel are using the four factor concentrate and that is widely available in other countries. The three factor concentrate is approved for use in hemophilia B, but there is increasing use that we are seeing in the United States in patients who are on coumadin.
What are the advantages over FFP? This drug is rapidly available. It can be given very quickly. It does not require that we give large volumes of plasma to patients who are on coumadin, these tend to be elder patients who cannot tolerate large volume transfusions. There is a decrease infectious risk because the drug goes through multiple viral inactivation steps. There is also a decreased TRALI risk.
TRALI is the number one cause of death after transfusion. It is primarily associated with FFP transfusion. The TRALI risk is decreased due to the lack of anti-HLA and anti-granulocyte antibodies. Three thousand international units will increase factor activity by 40 to 80 percent very rapidly. However, there is inadequate prospective data to support the use of prothrombin complex concentrate currently.
What are the reported uses? Prothrombin complex concentrate has been used to reverse patients with bleeding with vitamin K antagonists. It has been used in surgical procedures in hemophiliacs and there are studies available looking at cardiopulmonary bypass as well as massive bleeding. Based on the data that is available our institution has developed a protocol for the use of Profilnine or prothrombin complex concentrate.
We are using this drug on patients on coumadin who present with an INR that is greater than 1.45 who are either bleeding or require an emergency procedure. We are giving 4,000 units or 50 units per kilogram for patients who are under 80 kilograms. In light of the fact that there is minimal Factor VII present, we are giving this drug with one milligram of recombinant Factor VIIa, and we are also giving 10 milligrams of vitamin K at the same time because of the short-term effects of the prothrombin complex concentrate and the longer terms effects of vitamin K. The INR is checked after the infusion and this process is repeated if the INR remains greater than 1.5 with 2,000 units of dosing.
So basically, I have gone over three drugs in which -- which are being widely used in the United States, none of these are FDA approved for the uses that we are using them for. And I think that this really highlights the fact that the off-label use of systemic hemostatic agents in the United States is actually very, very common.
Interestingly, and I think we learn this very, very well from the Factor VII process, the use of these drugs may not be justified based on the data that we have. And I am truly hoping that we don't go through the same process with prothrombin complex concentrate and tranexamic acid that we did with Factor VII and then we get good prospective randomized trials we find out that these drugs are not -- really are not indicated.
And I think the bottom line here is that that we really haven't found a way to beat that qualified trauma surgeon with the $0.50 silk suture. There are no magic bullets when it comes to drugs for stopping bleeding. Thank you.
DR. VOSTAL: Thank you Dr. Schreiber. Our next talk is on pathophysiology of trauma and it will be given by Dr. John Holcomb, who is a chair of the Department of Surgery and the director of the Center for Translation of Injury Research at University of Texas at Houston.
PATHOPHYSIOLOGY OF TRAUMA
DR. HOLCOMB: Thanks. Good morning the -- so pathophysiology of trauma that is a pretty broad topic. I choose to make it a little bit broader to talk about systems as well. The pathophysiology of trauma is interesting when you start thinking about it. You can have structural issues of organ specific -- and I will show a picture of that -- and obviously, a variety of issues from nutrition to psychological, everything in between is encompassed with the assigned topic that was given. I will not attempt to cover all of that, we will be here probably for the next year.
And it's -- from a research point of view, it's obviously a double-edged sword. My wife is an endocrinologist and so she studies organs that are about this big that have beautiful cycles and there is not a lot of other stuff that goes along with that. Now, that is my view of what she does. She might have a different opinion of what I do probably.
The trauma -- if you get injured you affect every part of your entire body, every organ system everything is absolutely affected and how those interact is actually relatively unknown. Now, that is what we like about it. For those of us who work in this area, we like that. But it also I think makes it difficult to be a reductionist and narrow it down to one organ system and spend your whole life working on one organ system. And so it is a double edged sword.
The structural part -- I'll have two slides on. There is laparotomy, thoracotomy -- those physical things that certain anesthesiologists, emergency medicine, the clinicians do that actually put their hands on trauma patients and try to fix an organ. We use external fixaters, packaging, suturing, et cetera.
And then I will have my gratuitous gory slide because I am a trauma surgeon and you expect that. This is a patient I took care of a couple of weeks ago. He was -- he had a thoracic abdominal gunshot wound, arrived in profound hemorrhagic shock, three liters of blood in his abdomen, and that is a Foley catheter sticking out of his liver. So I used the balloon on the Foley catheter, not in his bladder but in his liver to help stop the bleeding. That is -- we are talking about other agents for the rest of the discussion.
So there are structural issues, but there is many more other issues in the pathophysiology of trauma that will be covered in this symposium. Injury is a really big deal, I have a couple of extra slides in the handout, but -- there is about 10 percent of the U.S. population comes into emergency centers every year, 10 percent of the U.S. population every year. So there are a lot of injured people. It's a $500 billion problem annually across the world. It is projected to be the leading cause of death by 2020 by WHO, not malaria, not HIV but injury.
In the United States it is the leading cause of life years lost, far, far outweighing cancer and heart disease because it is a disease of young people. It is a disease of young people, thus the life years lost from CDC.
And then obviously, we have a military component here. Many of us in the room are either currently in the military or retired military and we have the DOD as a cosponsor. From my time in the military -- and we have looked at this specifically the reaction to injury the pathophysiologic response to injury in the military population is exactly the same as in the civilian. The mechanism of injury is different whether you have an IED or a high velocity gunshot wound or a motor vehicle crash and a low velocity gunshot wound but the response is physiologically the same.
And so we strongly feel that the research solutions, be they military -- discovered in the military environment or the civilian environment, will apply equally across both spectra. We dove into that in great detail in a burn patient and found it to be exactly the same.
So in summary of those slides, injury is a really big deal, it's actually bigger than most people realize; huge opportunities for large studies. I will go into that in a second as well. A huge societal problem that is really has not been investigated fully. There is great opportunities, I think, for really large advances fairly quickly because there has been so little done in a high quality area.
Now on the truck (phonetic) system, again from a pathophysiology point of view I choose to kind of drive this into systems. In a second I will tell you why we did this. Ellen MacKenzie has really two similar papers. This one says that if you go to a trauma center you will have a better outcome if you are seriously injured than if you don't go to a trauma center.
Now, if you are not seriously injured, it probably doesn't matter as much. If you just have an isolated fracture to your femur, probably you could go to just about any place and they can take good care of you. But if you are seriously injured you have an improved survival rate at level one trauma centers.
And then the next paper she wrote a couple of years later was that it's actually cost effective. So not only are the level one trauma centers improve outcome, but they are cost effective as well. The -- this -- from the committee on trauma through the American College of Surgeons has spent the last 10 or 15 years really organizing and putting together registries, and we'll show some of this in a second.
The National Trauma Data Bank has millions of patients entered from 600 hospitals; there's about 600 to 900 patients going every year. There are about 20 trauma centers in the United States that admit more than 3,000 trauma patients a year; so 20 centers admit 60,000 patients. If we use the percentage of massive transfusion in this study we just completed at 10 trauma centers in the country, that's 900 massive transfusions a year.
That's an important number, remember, because that is the number that were -- shows up over and over and over again in many of the studies that we think we have a great opportunity to decrease the preventable deaths. So in 20 trauma centers, 900 massive transfusions in a year, lot of sick people; the mortality in this group of people is anywhere from 40 to 70 percent -- huge variability, but it's a great target to get after.
Just an example of 2009 data, and there's more in your packet -- over 600 -- almost 690,000 admissions. And after about age 15, men do worse than females as is out there. And so the registry data out there really allows us to drive into the epidemiology, into the fine data sets and targets for research in a great area. This slide is interesting from an ISS point of view.
This slide shows that that mortality in patients with a -- ISS greater than 24 which is injuries to various score -- these are sick patients, has mortality of about 30 percent. That's our target. Now the problem is, we don't calculate ISS in the emergency department. It's calculated at the end. But again, there are a lot -- data sets can allow us to hone in on the right kind of patients.
I think some of the reasons that the trials that Marty Schreiber discussed were considered failures, was because we didn't select the right kind of patients in the emergency department, and then didn't use the committee consultation process to intervene early. We waited too late on patients that were bleeding too slow.
More data from NHTSA. This is data collected by and put together by Dr. Charlie Wade and Dr. Champion. Retrospective data of deaths and it looks at time to death and location of death in almost 130,000 fatalities. As Marty Schreiber showed earlier from the single-center study from San Antonio, exactly the same curve is that all death after injury is essentially over and done with by about 150 minutes.
So you have about 2 to 3 hours to intervene. So whatever your study is going to do, it has to be done very early. You can't be waiting 4 or 5 or 6 hours to intervene; needs to be done essentially on admission if not pre-hospital. And the number of deaths in their urban environment almost exactly the same -- they're 50/50.
Now, the military death rate is a little bit different. It's probably more like 75 percent pre-hospital and 25 percent in the hospital. I think that is in direct relation to the severity of the injuries from explosions. Where are these deaths another way of looking at this in time? I think this is a fascinating slide from a research point of view.
If you're -- these are the type of data that are available to design research studies and decide if you're going to go pre-hospital or in the hospital with your intervention. And then of course time to death -- and again an interesting slide from a study design point of view just emphasizing this fact that even in the hospital the time to deaths are almost all over with by 2 to 3 hours.
Now, why is all that important? Those are fairly current data. This paper from Don Trunkey in Scientific American in 1983 talked about when patients die and what they die from. This is a seminal paper that really drove, I think, lots of work for this -- for the next 25 to 30 years, and in fact is probably still driving some of the philosophy between our community consultation process now.
If you notice, there's three humps. There's a very early, there's a middle at several hours, and then there's a third hump out there that's largely multi-organ failure. Now move forward to 2005 from 1983, in the paper from L.A. County in the Journal of the American College of Surgeons in 2005. And I'm not going to go over the details of this, but what this paper says is the third hump has gone.
That trimodal distribution is gone because of the work that's been done on multi-organ failure. As Marty Schreiber spoke earlier, that 9 percent number is where everybody has really focused their efforts. And why is that? Well, it's relatively easy to study multi-organ failure. You don't have to do it at 2:00 o'clock in the morning. You can do it -- start at 8:00 and go home at 4:00, they're in the ICU, it's convenient.
But it's not the largest part. It's actually the -- it's actually one of the smaller parts in mortality. And so while in 1983 that was clearly there, in 2010 that third hump is gone. We published a couple papers about the causes of death in the military environment as well is exactly the same -- hemorrhage and head injury. Multi-organ failure in the military population is about a 2 to 3 percent problem.
The -- so I want to go back to the previous slide. The -- and then this slide again from Dr. Demetriades -- he had a -- the trimodal distribution of death is gone. This was out of L.A. County in the previous year. And then what he says in this paper is that the deaths caused by hemorrhage, chest or abdominal trauma peak at 1 to 6 hours and by head injury at 6 to 24 hours, and suggested strongly that these findings may help to focus research.
So again, a change in epidemiology over the last 25 years. These are data from our trauma center trying to look at this trimodal distribution as well -- almost 2,400 deaths over 9 years. And if you notice, all the deaths were in this area, they're almost all over and done with by 24 hours, certainly by 72 hours, no trimodal distribution of death at all.
So again reiterating what Marty said, we did not actually compare our notes together, but time -- if we're going to design these trials, they need to -- the interventions need to be on very early, either pre-hospital or on admission. And every minute that you wait actually, death rate continues to go up.
And if you're intervening after about 3 to 4 hours, almost 90 percent of the deaths are done that going to happen. So we really need to get after it. Now, converting -- over to more systems point of view, the committee on trauma publishes lots of publications. These are the two extremely important ATLS that people know about, and Resources for Optimal Care of the Injured Patient that's updated every several years, as is ATLS.
These two books drive how trauma systems, individual centers in care are organized not only in the United States but around the world, around the world. ATLS is really an international phenomenon. Now, it -- and then from a guideline point of view, Eastern Association for the Surgery of Trauma probably has the best guidelines out there describing what to do with the levels of evidence, and they're widely used.
The DOD has actually on purpose copied this system of care to include the guidelines. The guidelines are on the ISR website and are widely utilized in the current theater. (Inaudible) is a relatively new phenomenon that's come about in the last couple years; so using that registry data, creating a subset that's cleaned and honed even more to look at quality.
And this slide is from a recently -- a recent publication looking at observers of expected deaths. And what it just shows these data set -- we have systems of care, we have guidelines of care, we have large registries, we have subsets of the registry that allows us to look at quality and quality outcomes. Now, ATLS is actually supported by very little data. So it's widely used around the world.
But much like some of the guidelines for American Heart through ROC we've actually learned a lot about cardiac care and CPR. And it's -- you know, it's interesting how little data actually support those recommendations. And then a lot more recently with CPR ROC and other funded studies. But the trauma group has really lagged behind in that. There are very much consensus.
The transfusion examples -- I'll talk a little bit about that -- or some great examples on how poor the data are that have guided care. And then the ABCs which are a big -- airway, breathing, circulation -- a big component of ATLS, are absolutely based upon consensus and no care. Dr. Schreiber and Dr. Rhee are at least two of the members of this study that was published in 2008, went through the 23 recommendations in ATLS and the five levels of evidence.
You know, there are many ways to rank levels of evidence, but they're all kind of essentially the same idea. There actually are a fair number of level one and level two, but the -- what's interesting -- we dive in the details -- so they're actually focused on almost the same recommendations. So while there are 20 -- there's 30 level one and level two recommendations, they center on two to three areas.
And so they're all grouped in that one layer -- spinal cord and CT scanning of the neck. All -- the majority of the recommendations are in the level four and five, which essentially four guys in a bar getting together and having a conversation and coming up with a consensus. That is really if you go -- and we've had those discussions with those four guys.
That's actually how things really work for the majority of the trauma population for the leading cause of potentially preventable death in the United States and in the military. Now, I think that leads to this finding by Joe Minei. And this comes from the Resuscitation Outcomes Consortium funded by NHLBI. Huge variation in outcome amongst trauma centers -- huge variation.
Because the quality of the data are not very good, there's huge variation in outcome from one center to the other. So the summary of this part of the pathophysiology is that there are robust systems of care in place, there are documented improvements in care through sophisticated national registry data with millions of patients in there and describing outcomes of interventions and also observers of -- versus expected outcomes between institutions.
Death occurs very, very early and the epidemiology has absolutely changed over time. The focus -- head injury and hemorrhage with hemorrhage being the leading, potentially preventable. Many of the head injured deaths are non-preventable, but many of the -- a large portion of the hemorrhagic deaths were potentially preventable.
Huge variability in outcomes between institutions, and the quality of our guidelines really needs improvement. Now, going into a little bit more of the classic physiology, one of the things that will be discussed -- was discussed in the April symposium quite extensively is this coagulopathy of trauma.
And this is -- this scheme that was worked -- many of us were taught and we taught for a long time is that patients go into hemorrhage after injury shock. There is iatrogenic factors -- what we do to people may actually cause injury. And then you have transfusion. There is obviously comorbidities. And the patients develop hypothermia and acidosis because of their shock, and because of that develop coagulopathy.
That is a reaction, if you will, to all of these events. Now, the hypothermia part is obviously a big deal. There are people talking about therapeutic hypothermia and there certainly is a role for therapeutic hypothermia in specific isolated organ injuries. However, in trauma patients right now, the data would suggest uncontrolled rather than controlled -- but uncontrolled hypothermia, that which was -- comes after your shock -- is associated very strongly with a very bad outcome.
But it remains to be seen whether therapeutic hypothermia for multi-system trauma patients in shock will be beneficial. That's an interesting research question. And then from an acidosis point of view, these are patients -- 900,000 patients from the National Trauma Data Bank showing -- again confirming what others have shown, is that early description -- the base steps is that in the emergency department is associated with a bad outcome and the higher the base steps, that the higher the bad outcome.
So the coagulopathy issue is extremely important. Brohi and Macleod and -- Karim and Mitch Cohen published this paper in 2003 and Jana Macleod in 2003 as well from Miami. Two separate publications in the same year, you know, dealing with trauma, talking -- saying -- really focusing on coagulopathy that it was independent of hypothermia and acidosis.
It certainly could make coagulopathy worse by having acidosis and hypothermia. But the coagulopathy itself was an independent bad actor, and this paper correlated from Brohi said that the ISS and PT -- if that was your measure, INR was correlated with increasing mortality.
Wenjun Martini, 2 years later, looked at this in an animal model and showed that if you had the independent contributions of hypothermia and acidosis on decreased thrombin burst from a coagulation point of view, hypothermia sculled slowing of that thrombin burst while acidosis lowered the thrombin burst. The two together you got both -- that were additive -- so very bad actors.
And then John Hess was the first author -- also in the audience -- of this paper, a large consortium -- the folks actually in international consortium -- trying to put together a new way of thinking, kind of update of that Denver paper from before, focusing on coagulopathy with all of the different impacts.
None of these arrows are weighted; none of the contributing factors are weighted. Really what this paper says, published in 2008 just 2 years ago, is that the coagulopathy is a big problem. I'm not exactly sure what causes it. But it is a big deal and we should really get after it.
In relating -- in describing the acute coagulopathy of trauma shock down here, Karim Brohi and Mitch Cohen again have really spent a lot of time, done very nice work and are proposing that the -- this coagulopathy of trauma is modulated through the protein C pathway in its -- the endothelial response, the systemic endothelial response to hemorrhagic shock. And we firmly believe that they're on the right path and will continue to do beautiful work here.
The early cytokine production -- again going back to the pathophysiology, cytokines are all elevated. There are patterns of elevation; there are times of elevation. And using cytokines -- if we wanted to do this in the emergency department, I think we could absolutely predict who was going to be in multi-organ failure or not.
Again, multi-organ failure, once you get this small problem in trauma patient, it is a very bad problem. Patients -- many patients die from multi-organ failure because there's no effective treatment. So again, death occurs fast. That lethal triad of hypothermia, acidosis, and coagulopathy is present on admission, uncontrolled hemorrhages are leading potentially preventable cause of death.
While multi-organ failure is rare, it is predictable and lethal. The cause of coagulopathy is still unknown. So our conundrum -- well, all of this is which one of these products do we use. We have -- this is what we have in our emergency department -- we have plasma and red cells, we have normal saline, lactated Ringer's, and we have any number of artificial colloids.
The problem is from a data-driven point of view outside of tradition, we don't know which one to use, when to start, how much to use, and when to stop. That's really the summary of the resuscitation data in a data-driven fashion. You should be very careful driving home from the meeting today, right?
DR. HOLCOMB: You should think about that. Just looking at normal saline and lactated Ringer's for a second -- and I -- obviously that's a little hyperbole, but it really drives home the point. Even something as simple as normal saline and lactated Ringer's which are frequently used interchangeably, there's frequent discussions that you can't use lactated Ringer's when you use blood products.
That's actually not true. A nice publication in the anesthesia literature this year show that myth not to be true. And that Jonathan Waters published in 2001 that in a randomized prospective blinded fashion, one of the places where there are level one data in aortic aneurism patients, so they bleed quite a bit, that lactated Ringer's decrease the amount of total blood products given.
This is a randomized blinded study. So lactated Ringer's in bleeding patient versus normal saline, two bags that many people consider interchangeable are not. That to me is a pretty important philosophical stepping-up point for these kind of discussions. Let me go back to Dr. Counts' paper in 1979 -- extremely influential paper.
This is just to point out how really poor our data are. Dr. Counts is a really smart guy. I was actually provided to give one -- a talk at his retirement and talked to him about this. What he wrote with 27 patients without a control group in 1979 is that when you used modified whole blood, you don't need to use plasma to resuscitate patients.
Now, whether people use this paper still today is they don't resuscitate patients with plasma. To do so is unnecessary and wasteful. But how many of you all use modified whole blood in the resuscitation of patients? Nobody does that, to include Dr. Counts in Seattle. So we really need to go back to these papers and look at what they say what products they used.
And almost all of the papers published in the '70s, '80s are what's referenced today, because nobody studies this anymore. And none of those products are used. Certainly if they look similar, the additives are very different, and there is no clinical outcome data associated with any single additive that's used in any of these blood products.
So the quality of data within which we used to drive resuscitation today -- and it's actually referenced in ATLS -- is extraordinarily poor. And that's the point of that discussion. Now, we talked -- when people talk about plasma or any of these products, they always talk about the coagulation proteins. In our lab we kind of look into this a little bit.
The coagulation -- the amount of the coagulation proteins may not be quite as important as what these bag of proteins do. FFP has been on the publication 1,000 to 2,000 to 3,000 proteins, you know, many beyond the coagulation proteins. And they have great effects on the endothelium, on permeability, and many other issues.
So we are suggesting that some of these potentially beneficial effects of plasma, if they're real, may not be related to coagulation proteins and may be related to those other 980 or so proteins that are in that bag. Now, in my last slides -- will -- if we do studies -- and that's obviously what we want to do is do high quality studies -- will clinicians pay attention to them and change their practice?
And this study in 2008 is not a transfusion study, but it's a follow-on to the ARDSNet trials. And what it shows in this complicated graph is over time clinicians did change their practice based upon the ARDSNet data. So clinicians will change their practice if there are quality data. In summary, trauma centers and all those systems of cares are out there to improve outcome.
The systems do -- are in place to propagate guidelines. The data that we -- generated data can be put in manuals and discussed at -- and implemented. There is huge variability because our data is really poor. There are really extraordinary large numbers of trauma patients, and there's a lot of bleeding occurs in trauma patients.
Deaths occur very early, so any intervention is going to have to really utilize the community consultation 50 24 process. Incredible lack of quality of data -- we don't understand the mechanisms. And because of that, we have a hard time applying specific focused interventions in clinical data and clinical studies. Clinicians will implement guidelines in the systems of care in place to do so. Thank you very much.
DR. VOSTAL: Okay. Thank you, Dr. Holcomb. So that brings us to our first session with a panel discussion. And the way these sessions will work is that we'll have an introductory speaker and then the panel moderator will moderate the discussion based on the talking points that were selected before the workshop.
So this session deals with the "Definitions of Bleeding." An introductory talk will be given by Dr. Lena Napolitano who is the chief of the division of acute care surgery and a director of the trauma -- surgical -- clinical -- critical care at the hospital of University of Michigan.
DEFINITIONS OF BLEEDING
DR. NAPOLITANO: Good morning, everyone. I was tasked this morning with reviewing severity of bleeding issues in clinical trials with particular regard to inclusion criteria for the trials. So first review -- what definitions are you certifying severe bleeding in trauma, the hemorrhagic shock; ATLS classification is used clinically, it is not used as clinical trial injury criteria.
As John mentioned, massive transfusion is utilized. The definition that is commonly utilized is 10 units of blood in 24 hours. In terms of systemic agents, this is an ideal cohort for the reasons that John explained. Here is the ATLS severity of bleeding scales that are again not used as inclusion criteria.
And again to review -- I knew this -- to review that the mortality rate for massive transfusions in trauma patients is very high. This is a recent review that shows that the mortality rates range anywhere from 25 all the way up to 71 percent. In a recent retrospective review of the German Trauma Registry where they looked at ratios of FFP to packed red blood cells, you can see that data for the mortality rates based on 6 hour in white, 24 hour in gray, or 30 day in black.
You can see that the mortality rates still are substantial, ranging again from 30 to 50 percent. This is John's study of 466 massively transfused civilian trauma patients that were entered into a prospective cohort study at 16 academic trauma centers. And I want to focus on two graphs here, one at a 24-hour survival rate and one at a 30-day.
The 24-hour survival rate, as you can see here, ranges from 40 percent mortality up to 80 percent mortality, again dependent on the high FFP to high platelet to packed cell ratio or the low FFP to low platelet packed red cell ratio. Notice that the curves again diverge very early. They diverge at approximately 2 hours after injury.
And then at the 30-day survival, these changes in the curves diverging again very early at the 5-hour time point do continue out to the 30-day time point. All of these studies show us that this massive transfusion cohort again has a very high mortality rate, and are the best patients in terms of enrolling in trials for systemic agents.
So the challenge is how to identify these patients early in their trauma course, identify them in the pre-hospital setting or identify them on admission to the trauma bay. I want to next review three scores that are utilized -- utilized in terms of study, but again not utilized clinically. The TASH score, the McLaughlin score, and the ABC score.
These scores are utilized to predict patients -- trauma patients who are at risk for massive transfusion. The TASH score was developed out of the German Trauma Registry. It includes the eight variables that you see here -- hemoglobin, base excess, systolic blood pressure and heart rate, physiologic variables, and then some clinical variables specifically related to injury, hemoperitoneum by FAST, pelvic fracture, femur fracture in male gender.
The score then ranges from 0 to 28 points. And you can see here that a TASH score of 18 correlates to a 50 percent risk of massive transfusion. And the development data set for TASH by the German Trauma Registry ranging from the years 1993 to 2003, a decade, about 6,000 patients were looked at in that trauma data set and 49 percent of the German Trauma Registry patients were excluded because they did not have one variable base excess.
So this is an important thing to consider -- not always are VBGs and ABGs obtained in the trauma setting. This recent publication this year of a revalidation data set is quite intriguing. It looked at 3 years -- 2004 to 2007 -- again about 6,000 patients from the German Trauma Registry. And again you see that about 50 percent of the cohort was excluded because they did not have a base excess.
Now, this shows the importance of looking at the data over time. The TASH score of 16 in the development data set had a 50 percent risk of massive transfusion. And in contrast, the new validation cohort from more recent times -- 2004 to 2007 -- now that has a risk of 35 percent massive transfusion. So again we are doing better at caring for our trauma victims, and the data will change over time.
Let's look next at the McLaughlin score. This is a score that was developed in combat casualty care. It included a cohort of 302 trauma patients who received at least 1 unit of blood. There was a 26 percent chance of massive transfusion in this cohort with a mortality rate of 29 percent. And the risk factors for massive transfusion were mostly physiological variables.
They were tachycardia and hypotension, acidosis and anemia. You can see the sensitivity and specificity here at 60 and 77 percent. And when you had a McLaughlin score of 4, you had an 80 percent chance of massive transfusion. But it's important to note that patients in this cohort who had none of the four values still had 11 percent chance of massive transfusion.
So this tells us again that these scores are not very specific and do not capture all of the patients that in the end up receiving massive transfusion. Those two scores that I just reviewed, the TASH and McLaughlin include laboratory testing. And we have to consider that this will delay the enrolment of patients in clinical trials, particularly again if you use lab parameters not using point of care testing again for the initial ABG or VBG.
So there is one additional score, the ABC score -- the assessment of blood consumption score that uses no laboratory testing. And here vital signs -- penetrating mechanism and a positive FAST for hemoperitoneum are used. So this score developed by Bryan Cotton and colleagues looked at four components that we just reviewed. You score 1 point for each.
And if you look at the scores of 0 to 4, you can see that there is a direct correlation with degree of massive transfusion and with degree of mortality. So it looks very reasonable and very fast and quick to do in the trauma bay. What's a little concerning about the ABC score is that there is not differentiation between a score of 2 or 3 for risk of massive transfusion.
You can see here 41 percent versus 48 percent. Whereas if you have a score of 4, 100 percent of that cohort had massive transfusion. A very good thing about this score is that all of the factors that are included here contribute equally to the score being sensitive for massive transfusion. And the area under the receiver operator curve is 0.85, very similar to the TASH score of 0.84.
That initial study of Dr. Cotton and colleagues was a single-center study. They have also validated their study in a three-institution -- Vanderbilt, with Pittsburg and Hopkins study just recently published this year. And once again, the ROC for the ABC score was 0.9. So it fares very well.
Now, I just want to spend a word on bleeding severity scales that are used in other clinical trials that we should be aware of, that are completely not usable for trauma trials. For cardiology studies, VTE prophylaxis trials, studies on anticoagulants, platelet inhibitors, other fibrinolytic drugs, there are a number of scales that are used, and these are in your handout.
There is a World Health Organization severity bleeding scale. There is the ISTH scale, fatal, symptomatic, or fall in hemoglobin, and then studies in myocardial infarction and PCI studies, interventional studies in acute MI, the TIMI bleeding classification scale, the GUSTO scale. These are again all out there in the literature that are used in those specific clinical trials, and they are not relevant to trauma trials.
And then one word on topical agents and severity of bleeding. Topical agents have been studied in surgery mostly in patients that are not having very severe bleeding, but bleeding that is unable to be controlled locally. And they're used as an adjunct as we heard this morning. This is one of the cardiac surgery fibrin sealant studies where the group was treated with fibrin sealant or a control and the endpoint here is control of bleeding within 5 minutes.
And you can see there was a statistically significant reduction. And the second endpoint was the absolute time to cessation of bleeding. So here you're not talking about severe bleeding, you're talking about local bleeding that occurs in the operative setting where topical agents are utilized.
I want to end here with looking at the systemic trauma trials some of which Marty Schreiber just reviewed for you, but focus on the inclusion criteria. So here is the Factor VIIa study. This is the ex-U.S. study not done in the United States. It was a phase II study focused here only on the inclusion criteria, severe trauma requiring 6 units of blood within 4 hours of admission.
When we look at this study, again the Factor VIIa was administered at zero time point after the completion of 8 units of blood, and then an additional dose given 1 and 3 hours later. And the primary endpoint was the total number of units of red cells transfused from dosing to 48 hours. This enrolled 300 patients, and very mindful to note that this did include a severe bleeding cohort.
These patients were massively transfused on average 22 units of blood in the first 48 hours. They already had established coagulopathy. You can see prothrombin time is elevated here in the blunt cohort, it is elevated here in the penetrating cohort as well. so this group clearly is a massive transfusion cohort.
And Marty showed this slide of 33 percent massive transfusion rate in the blunt patients who did not receive Factor VII and 14 percent in those that did. Reflect again that the patients that were enrolled were severe trauma patients requiring 6 units of blood within 4 hours of admission. And overall, this cohort received over 20 units of blood within 48 hours.
He showed that the red cell transfusion rate was reduced by 2.6 units of blood, and there was a 50 percent reduction in the massive transfusion rate. And here is the Kaplan-Meier curve so that you can see that again. This did occur -- the separation of the curves did occur very early in the blunt transfusion cohort.
The survival curves were no difference. Again at 30-days survival, no difference in mortality. And the secondary endpoints which were multiple-organ failure from ARDS were no different. Now let's look at the same thing for the CONTROL trial and again focus solely on the specific inclusion criteria. These are the differences in the study.
These were again blunt or penetrating adult trauma patients. They had to have received 4 units of blood, but completed -- but not yet completed 8 units to try to get to earlier administration. And they had to have acceptable markers for active bleeding as you see listed here, continuing hypotension acidosis or continuing ongoing fluid resuscitation.
So basically the CONTROL trial, the Factor VII U.S. trial just moved up the administration of Factor VIIa to the fourth unit of blood rather than the eighth unit of blood, and otherwise it was a very similar trial. In the end here, 573 patients were enrolled. Many more blunt trauma victims in this trial -- 481 versus penetrating only 92.
There was no differences at baseline in the study cohort. But again, this study did enroll a massive transfusion trauma patient cohort. And you can see here the lactate was elevated at 3 to 5 and the INR was also elevated at 1.6, coagulopathy clearly present here in this cohort. But a very significant difference in this trial was the overall mortality rate.
Here if we look back at the first Factor VIIa trial, ex-U.S., the mortality rate for blunt trauma victims was 25 to 30 percent. Those patients were rapidly bleeding and again received, you know, 20 to 30 units of blood within the first 48 hours. In the U.S. trauma trial it's a very different cohort.
The overall mortality rate in the blunt trauma patients are 11 percent, so it's 50 percent reduction. They're bleeding more slowly, and overall the total amount of blood in the massive transfusion cohort was significantly less.
And then the third trial here is a CRASH-2 trial that again you've just heard about by Marty Schreiber. The inclusion criteria here are solely physiologic criteria -- systolic blood pressure less than 90 or heart rate greater than 110 or both; so solely physiologic criteria, no requirement for ongoing fluid resuscitation. This trial, very large -- unfortunately, about half of the cohort that was enrolled received no transfusion.
And again I think we don't want to -- for enrolling patients for systemic agents we don't want to include these patients. Only 50 percent of the patients received blood transfusion. On average they received 6 units of blood per patient. As John just mentioned, in the bleeding cohort, deaths occur early.
So in this trial in the CRASH trial, only 5 percent of patients had bleeding at their cause of death and these deaths occur very early here. They are in the red bars. You can see by 24 to 48 hours the bleeding deaths are no more. Traumatic brain injury was the leading cause of death in this trial at 6 percent of the cohort.
So getting back to the inclusion criteria just using systolic blood pressure less than 90 is problematic. You're going to have a large cohort of patients that will be traumatic brain-injured patients. And then this graph nicely shows that 28-day mortality is not a very good endpoint for bleeding risk in massive transfusion patients.
Again, particularly for the use of systemic agents, it would be much better to look at an earlier mortality endpoint. In these subgroup analysis of the CRASH trial, this is a very important finding. If you look at the patients who are enrolled who had systolic blood pressure less than 90 -- that was the inclusion criteria -- tranexamic acid really had no effect.
You can see the relative risk here is 0.94 with a confidence interval of .78 to 1.14. So the cohort that had the greatest signal for benefit in the tranexamic acid trial was in the most severe shock group with systolic blood pressure less than or equal to 75. Here the relative risk is .87 and you can see that the confidence interval is less than 1.
Finally, the tranexamic acid trial does have some problems and issues with it. Factor VIIa was co-administered in a number of patients. There was no standardization of resuscitation or clotting factors. There was no data regarding massive transfusion or coagulopathy.
And as John shared with you just a minute ago, we know from this nice systematic review and many other studies that the administration of fresh frozen plasma and platelets is associated with a reduction in mortality. It is also associated with a higher rate of ARDS and acute lung injury, and has a signal for reduced multiple-organ failure.
So again administration of FFP and platelets will be a confounding factor for many of these clinical trials. Nonetheless, we're left with here the final conclusion. All-cause mortality was significantly reduced, risk of death due to bleeding was significantly reduced, and that tranexamic acid safely reduced the risk of death.
But again I have a hard time believing this and believing by which mechanism if there was no difference in the blood transfusion rates and no difference in coagulation studies. So I'd like to end with the ROC trauma shock trial. This study is not yet published. It is in press. This is the ROC out-of-hospital hypertonic saline trial. It was done from 2006 to 2008, so a very recent trial, multi-center, double-blind, three-armed trial.
Again focus on the inclusion criteria here, adult trauma patients. This is a very different inclusion criteria, pre-hospital, systolic blood pressure less than or equal to 70 or a composite systolic blood pressure 70 to 90 with a heart rate greater than 108. But the difference here is this is not in-hospital, this is pre-hospital, so very different inclusion criteria.
Where did they come -- where did this inclusion criteria come from, where did they derive from. Eileen Bulger did their first hypertonic saline trial and used the pre-hospital systolic blood pressure less than or equal to 90 as their inclusion criteria for their first study of hypertonic saline and hypovolemic hemorrhagic shock.
And this study showed no difference in ARDS free survival at 28 days. And part of the problem is they had a very small number of patients -- only 24 and 15 patients that required massive transfusion in this study. When they looked at it, 65 percent of the deaths occurred within the first 48 hours after injury. This single pre-hospital systolic blood pressure less than 90 was not a specific marker for hypovolemic hemorrhagic shock.
And in this study using again pre-hospital systolic pressure less than 90, 45 percent of the patients enrolled did not require any blood transfusion in the first 24 hours after injury. So following the completion of this study, their conclusion was that futility of this trial was due to failure to enroll patients likely to require massive transfusion and hence came the inclusion criteria for the ROC study of systolic blood pressure of 70.
This study had three cohorts -- saline, hypertonic saline, hypertonic saline-dextran or normal saline. And the primary endpoint was 28 days survival with the secondary outcome measures of the usual multiple-organ failure 28-day ARDS-free survival and multiple-organ failure. This study enrolled 895 patients. It is remarkable to look at the pre-hospital inclusion criteria in this study.
The mean systolic blood pressure was 55 to 59 in the pre-hospital setting. And so indeed they did enroll the ideal cohort that they were deriving. It's also very remarkable to note that the admission systolic blood pressure, once the trauma patient arrived at the hospital, was greater than 100 in all of the cohorts. So again you can see that pre-hospital resuscitation is playing a big role here.
In this study we identified that 80 to 90 percent of the patients had persistent lactic acidosis, so I believe that this is very good indicator of hypovolemic and hemorrhagic shock, and 30 of this cohort required emergent hemorrhage control.
Now, what about massive transfusion? Sixteen percent of these patients required massive transfusion, so a much larger number than the initial hypertonic saline study, and the on average the pack cells within the first 24 hours was five, it's still low.
In finality, the conclusion of this trial, it was close early for futility and concerns related to increased mortality in the post-randomization subgroup. There was no difference in overall 28-day survival for the three cohorts, but there was this signal of a higher 28-day mortality in the subgroup of patients who received no blood in the first 24 hours. And where you see it is not on the 28-days survival or mortality curve, this looks like no difference in any of the cohorts, but where you see it is actually on the 24-hour mortality curve, you can see here that the hypertonic saline group in red had a higher mortality, and that was particularly in the cohort that received zero blood.
Finally here, death within 6 hours in the group that received massive transfusion comprised approximately 6 percent of all of the deaths in the massive transfusion cohort. And, overall, there was no difference in the secondary outcome measures. So I think we have to think of a number of things of our challenges, this is a heterogeneous patient population, we need to consider pre-hospital vital signs rather than in-hospital initial admission; we need to think about maybe merging physiologic criteria with shock data in terms of lactater-based deficit considering the ABC score, and the number of units of transfused is a bit problematic.
And then finally here in terms of our outcome measures, many of them are measured only as packed red blood cell units. We're not including a number of other things such as time to cessation of hemorrhage or time to shock resolution, and I think we need to consider earlier mortality endpoints as John mentioned. And finally, I think it's difficult to determine the effect of the systemic agent alone because many of the other things that have already been discussed, a good surgeon, efforts to seize hemorrhage and aggressive coagulation of coagulopathy will all factor in in terms of confounding variables in these studies. Thanks for your attention.
DR. VOSTAL: Thank you, Dr. Napolitano. So at this time I like to invite the panel members to come up to the podium and take a chair. Please not that to activate your microphones you'll have to push the button and draw the microphone close to yourself.
PANEL DISCUSSION: DEFINITIONS OF BLEEDING
The panel moderator for this session is Dr. Nigel Key who is a director of Hemophilia Treatment Center and a distinguished professor of medicine at the University Of North Carolina at Chapel Hill.
DR. KEY: All right, well, thank you very much. I'd like to thank the FDA for the opportunity to moderate this panel and to get involved in the organization of this workshop which I think is going to be very interesting, so then today and tomorrow. We're off to a pretty good start. We have a mixture on the panel here of people with variety of backgrounds, starting from the far end, Amy Shapiro, who like myself is a hematologist but a pediatric hematologist; Dr. Peter Rhee from University of Arizona who is a trauma surgeon; Dr. Jerrold Levy from Emory, a cardiac anesthesiologist; Dr. John Hess who is a trauma blood banker, if I can call him that, and a hematologist; and Dr. Herr, Dan (sic) Herr from Maryland who is a intensivist.
So just a couple of ground rules, we -- you do have in your packets 3 by 5 cards, and if you'd like to write questions we can collect them and bring them up periodically and I'll try to triage them here to the audience. We wanted to get the -- have this as a sort of a semi-structured panel initially and get some discussion among the panel members. Towards the end we can have direct questions from the audience also using the microphones.
We're ahead of schedule, we're doing well. And I -- just to frame the discussion, just to remind you, really our focus is on essentially inclusion criteria for patients entering these trials, we're not going to deal at this point with outcome. So please hold your questions on outcomes until that later session.
So Dr. Napolitano, I thought -- or will be joining us on the stage, and -- oh, there she is, okay, thanks, Lena -- I think gave us a very nice introduction and a very nice framework on which to begin the discussion of how to consider the variables. And can I have the next slide up here, please? Okay.
So these are really the questions that the FDA would like us to address, and I think we'd like to see addressed and will summarize tomorrow. Specifically, I'll go through them quickly, what are the parameters or eligibility criterias I mentioned for entry into these studies, and by implication what is the definition of severe bleeding.
And in a sense we're dealing with three different areas here. We might be, as we had some of the conference calls, we were discussing that really we're dealing with our ability to directly measure bleeding which is difficult, very difficult in the situation. We might be dealing with clinical biomarkers whether it be a lactate-based deficit, whether it be an INR for the coagulopathy of trauma, whether it be a whole-blood assay method that can be turned around rapidly. We heard from Dr. Napolitano about the importance, I think, of time and from Marty Schreiber as well. We don't have the luxury of sending most of these labs to the coagulation lab and waiting for an hour because of the rate at which the deaths are happening.
And then third sort of general area in addition to the biomarkers and the direct measurement of bleeding which it maybe very difficult is really using essentially the physiology, the physiological effects of bleeding as entry criteria for these studies.
And so finally, number four here, where are we with the existing scores. I must say that when Lena sent me her slides to review not being a trauma surgeon myself I thought that there was significant -- there are significant data. I think some of these scores do have -- are being developed in such a way the ABC scores, the TASH scores and so on that may certainly have utility in being able to predict some of these issues.
So if I can start off. Lena, I think that just to essentially perhaps if we had them up in front of us, but I think you can tell us your viewpoints of your last two slides in terms of summarizing where you think the field should be going in terms of getting, selecting patients for these trials.
DR. NAPOLITANO: I think I concluded two things, one is that systolic blood pressure less than 90 I believe is not useful, and I think the ROC trial has shown us that the systolic blood pressure less than 70 or equal to 70 in the pre-hospital setting is a very valid physiologic inclusion criteria that again by the time patients reach the hospital the vast majority had systolic pressures of 100.
I personally would like to see a combination of systolic blood pressure of 70 in the field and lactic acidosis on admission in the trauma bay. We use point-of-care testing. I think virtually all trauma centers use point-of-care testing for their VBGs, ABGs in the trauma bay, so we get that back within a minute, our initial blood gas.
I think the ABC score has not yet been used a clinical inclusion criteria in a trial, and that needs to be tested. The one concern I have there is again that the physiologic criteria in that is systolic blood pressure less than 90. So again I think moving towards a lower blood pressure would be more meaningful.
DR. KEY: Okay, Peter, maybe I can ask you as a trauma surgeon here to follow that up.
DR. RHEE: Well, I think the problem is that you can't separate the entry criteria without talking about what your outcome variable is. You know, if you're talking about mortality, which I think, you know, 30-day mortality I think is the worst thing to study in this thing, and I think that's -- we're going to always get back to that problem.
If you're going to look at bleeding then you need to -- and you're going to do something to stop bleeding then that should be your outcome variable. So while Lena is correct, if you want to look at mortality as your overall result and then -- and it's a long discussion about this mortality thing whether it's the right one or not. But then you need -- it needs to be sufficiently poor so that you can see a divergence from a statistical and clinical study analysis point of view. But, you know, as Dr. Cotton's study shows and -- not Dr. Cotton's but Brian Eastridge's paper shows, you know, if you're looking at therapeutic interventions blood pressure at 110 may be more important for that. And what we're trying to do, and the reason why I think this session is going to be so difficult is because what we're trying to do is say which predictor is going to eventually show that they require a lot of blood, and that really results on what therapies are being done.
So if had Dr. Holcomb up here and I take the patient who is bleeding, who is obviously bleeding from, you know, things, you see a pumper in front of your eyes, belly is extended with blood, lactate's high, blood pressure is, whatever you want to do, and it doesn't reliably always predict, you know, lot of blood transfusions because, I mean, Dr. Holcomb might be good at putting this clamp on the vessel, and we don't know which one of these patients are going to get controlled in their hemorrhage and which are going to have sustained uncontrolled hemorrhage which is going to determine who requires blood transfusions.
That's why we're always frustrated in our hospitals when we say, you know, you walk into a room, into an ER and you see your trauma surgeon there hypertensive patient with blood hanging, you know that person is in trouble and I'll ask, well, did you activate the massive transfusion protocol for example. And they did.
But if we're looking at it from a monthly basis, you know, most of our, two-thirds of them, activations didn't require it but that was because we got the patients to stop bleeding. And so I think we go to keep that in mind, and that's why this topic is going to be very difficult to discuss from a clinical-study perspective.
DR. KEY: So more about this question of I guess, you know, and it's the trial design in terms of the patients who is bleeding versus the patients who is at risk of bleeding and we saw difference in terms of how the CRASH and the VIIa trials are structured and when the intervention is given. I mean, what should be the optimal, and the trial may be designed around that, and maybe that you increased your specificity by introducing your agent after the addition of several units of blood.
Anyone like to comment on that as opposed to upfront? We're looking for an agent that if I'm coming at you with a pro-hemostatic agent should I be looking at a -- trying to narrow down on the patients that are at a high risk of bleeding and treat them prophylactically or should I be designing my trial to wait and sort of self-select some of the patients?
DR. NAPOLITANO: Well, I'll be the first to say something. I think late is not good. So late administration is bad. And I think that's what we've recognized in clinical practice, maybe not so much in the clinical trial results, but when you allow these patients to get so coagulopathic, so thrombocytopenic, it can be difficult to salvage them. So I think early administration of whatever systemic agent we're going to study is going to be very important.
The question gets back to what Peter said is finding that balance of not giving and prothrombotic agent to a patient who is not going to need a large amount of blood and yet giving it early enough to the patient that is bleeding. And I think that's the tension and challenge.
DR. KEY: Jerry, I -- from a cardiac surgery standpoint you've done a lot of trials and like to hear your perspective.
DR. LEVY: You know, the interesting thing is that we talk about procoagulant, but I think intervening early with antifibrinolytics is a bit of a different paradigm because not only does plasmin do a lot of things in terms of adding to the coagulopathy, causing breakdown products of fibrinogen that create not only D-dimers but fibrin degradation products that have almost kind of (inaudible) like effect preventing platelet cross-linking, the plasmin can cleave glycoprotein 1b and 2b/3a receptors causing additional coagulopathy.
But plasmin also activates thrombin. So you sort of amplify the coagulopathy and generate thrombin further contributing potentially a DIC and early hemostatic issues. In the -- this, the multitude of studies that have looked at antifibrinolytic, preemptive prophylactic therapy is important. And there are rumorings about some of the CRASH data, and reanalysis is looking at timing in terms of outcomes as an issue.
The interesting thing in the CRASH study was that they loaded up to eight -- what was within six hours of injury, six or eight hours, six hours, which is interesting because the earlier the intervention the better off you are. I think whatever therapy we end up it's really got to be a multimodal approach, because for instance look at the red cell substitutes that's great, but when you hemorrhage a lot you also have coagulopathy-need factors.
So a lot of things into this -- into the mix. But I think the role of antifibrinolytics is very interesting. And if you look at the paradigm of measuring bleeding there are a couple of issues. One, chest tube output or drainage is problematic especially in the surgical patient with all the things that are done in a surgical irrigation and elsewhere, so you have to be a little bit cautious about drainage and how you measure blood loss per se.
And second, as a non-trauma person who has looked a lot at this literature and the data there has got to be some way that we somehow quantify trauma at tissue injury because a lot of that micro-vascular injury, development of all the liberation products may be very important contributing. And again the idea of using early antifibrinolytics I think is an important part.
DR. KEY: So Jerry, I know you wrote the editorial for the CRASH trial, and how do you see this issue of the discrepancy between survival and blood transfusion requirement, do you see this as a specific property of antifibrinolytics, improving survival through other mechanisms or is it a general effect of a prohemostatic agent.
DR. LEVY: You know, Nigel, those are really great questions. But in the several earlier editions of CRASH-2, if you thought CRASH-2 was radical you've just seen earlier versions where they wanted to make it mandatory that everybody got tranexamic acid. And my concern is that how can you say that, and there are lot of issues. I think you nicely illustrated some of the concerns.
Also look at the appendix and where the data came from, it came from a lot of third-world countries where you even wonder if they had blood products available, 20, 25 percent of the patients were from Indian, from countries that I can't even pronounce correctly. I mean, it's problematic in the sense that, you know, the blood-blood issues that may have influence.
And a study of that size is pretty dramatic. But there are a lot of quirky aspects. I think it's a great paradigm and food for thought moving forward, but it also -- what I like about it, it amplifies the concept that in any kind of therapeutic algorithm of early intervention not only do you want to generate clot and clot formation but you need to inhibit its breakdown.
And clearly lessons from TISSEEL, lessons from other perspective that not only generate clot, there is a massive response when thrombin is generated through these TPA to generate fibrinolysis, and that needs to be interrupted because plasma not only lysis clot but it also causes incredible proinflammatory response generating thrombin, et cetera. So I think again it's a great lesson from clinical studies that I've been involved with from laboratory data that we published that really emphasizes the key role of antifibrinolytics in a multimodal response.
DR. KEY: Go ahead, Peter.
DR. RHEE: I think you're asking specifically about timing of when to do it, so if we're going to put product X on wound, we got to give, you know, fluid Y into the patient or drug, whatever you want to do, if you're going to give something because you're in trouble, at six hours it occurs, but if a person is alive at six hours they're probably going to do pretty well. And the kind of people that I'm concerned about are the ones that I'm trying to bleeding immediately, that came in and within the first hour or two you're already in trouble, those are the guys that die.
The guys that are still alive at six hours is good because you are on your own luck (phonetic) and you've given lots of stuff and they get into a delusional coagulopathic stage. So, you know, I would like to be able to give the drug as soon as they hit the door if I think they need it.
Now, obviously, you know, let's call it a drug whether it's a fluid or dressing or whatever else, it needs to be effective, it has to stop bleeding, it can't be harmful to you, it has to have no side effects or very minimal side effects, and it has to be very cheep, if it was -- you know, if Factor VII didn't cause thrombosis and it was $0.13 a piece, I'd give -- you would be giving it to everybody with a shaving injury, right?
So that's the reason why we're trying to do these studies because of the cost portion of it and we don't know really what the complication rates are, you know, and then we got to figure out whether it works or not. But if you want to know when to give it, you want to give it, you know, before you get into trouble, not afterwards.
DR. KEY: All right, Lena.
DR. NAPOLITANO: Yeah, just to remind everybody, I showed the data from the ROC trial because it's not published yet, one-third of the deaths in the massive transfusion cohort occurred in six hours, so Peter is absolutely right. I mean, you have to give it early, otherwise you will lose the ability to intervene and reduce mortality, and one-third of the cohort that we're interested in studying.
DR. KEY: So Dan, please go ahead.
DR. HERR: I spent my first part of my career resuscitating trauma patients and I just resuscitate cardiac patients which were the same. One of the things though that I have -- I have said (inaudible) been talking about bleeding definitions and, you know, the definition for bleeding in my mind is I know it when I see it, it reminds me of something else in life. But, you know, we really have to -- if we're going to have definitions then we have to put our definitions based on the product that we're going to study because everybody can say I know when it's bleeding but I don't know when to use the products. So if we develop definitions based on the product that we want to study or the -- not the input but the actual pathophysiology that we're going to study and that the product is going to interfere with, that's the definition we should use.
We all know that bleeding is bleeding and we've tried these big generalized trials. I think, you know, we've gotten sophisticated enough now that upfront we can probably pick our phenotype for the product and do a better job. I understand that, you know, you might not have that time in the field, and I think the field stuff is very good to start looking at the patient and start because when you look at the ethics of this you really have to decide who is going to be in this study.
So if you have those general definitions I think it is. But you can't -- we can't go back to the artificial blood trials and put dead people in trials. I think that's the problem.
DR. KEY: Okay, we'll -- if you have questions just hold them up. I think maybe, Jaro, we are ready to take some in and we'll use the microphones later. But if you have 3 by 5 questions please just hold them up and we'll collect them. So Amy you had your hand up.
DR. SHAPIRO: I'm not a trauma surgeon, so I apologize for giving a comment here. But -- see -- it seems to me from looking at this data and listening to all these wonderful talks that there are some good scores out there that could be utilized to try to select out high-risk group of patients. But the scores are not necessarily sensitive enough to determine of that high-risk group which ones actually require massive transfusion to evaluate a particular intervention to determine its outcome, it includes patients who don't necessarily end up requiring massive transfusion.
So it would seem that in the design of this trial, these scores might be reasonable to utilize but that the size of the trial should be large enough to include subset analysis of those who actually require massive transfusion to evaluate a particular intervention and to look at that in a comparative sense, in a control group or data that's just collected for purposes of evaluation, those patient's outcomes who don't have an intervention. It also seems that we ought to be banking samples, you know, lactate blood pressure obviously very important, hematocrit anemia. But it depends where on the timeline you are. And I also think that there are probably some markers that if we banked samples and used them across studies that we might be able to identify something that could help us with improved interventions.
The endothelial cell injury that was discussed I think is very important, the protein C pathway isn't being evaluated, there are -- there is a protein C concentrate available, there could be other interventions that might be helpful to these patients when administering pro-hemostatic agents as well. I also think we need to look at the groups of patients by age, the geriatric population is different in their response and sensitivity to injury, and the pediatric patients are different as well in terms of their resilience and their vital signs.
DR. KEY: So I'm glad you brought that up. Jerry if you'll excuse me. One second, go ahead with your comment because I'll change the -- go ahead.
DR. LEVY: You know, one interesting thing, if you look at the -- Marcel Levi and I reported in the New England Journal about three, four weeks ago the recombinant VIIa data. And what's interesting is that in the older patient population there was a higher of arterial not venous thromboembolic issue. But, you know, lot of the data also came from the intracranial hemorrhage studies where there were a higher -- you know, at higher doses.
And I think the problem there is that if you look at, you start normal hemostatic systems, normal fibrinogen, normal platelets, everything is functional and you give a major pro-hemostatic agent, it's a lot different in the population that Dr. Holcomb and the gang and everybody is looking at, which is -- comes in as you -- acidotic, coagulopathic, hemorrhagic, it's a total different scenario just as the, you know, the different ISTH bleeding scores, four units of red cells and intracranial hemorrhage.
Well, wait a minute, how do we get four units of cells and intracranial -- these are two totally different issues. So I think this is a great opportunity to look -- move forward to develop a de novo or modify one of the trauma scores.
DR. KEY: So if I can go back to a point that Amy was making, I think, and that we sort of touched on, and I think was a major focus of the workshop in April was really this getting at the pathophysiology or the coagulopathy perhaps, perhaps that appeals to me as a hematologist, but maybe John has I could ask you, we've seen a couple of your publications that John Holcomb showed and the change in thinking.
And it -- despite the issues of point of care, prothrombin time and so on and so forth, it doesn't seem like rocket science actually to be able to get a point of care prothrombin time, how do you see it as a -- is that a separate of patients, the ones that come in that Karim Brohi identified, is that a good biomarker of the -- that should be in these trials in predicting outcomes, bleeding particularly.
DR. HOLCOMB: You have nanomolar concentrations of factor VII in your body. You have 5,000 molecules of tissue factor on every smooth muscle and cell and fibroblast, and, therefore, if you just do the calculations you have to macerate about two-thirds of a pound of tissue to soak up every bit of factor VII in your body. And so, you know, you can often just look at these patients and know that they will be coagulopathic and the test actually doing it provides some additional specificity.
You know, the problem the trauma surgeons have is, they want to be sensitive, they don't want to miss anything, therefore they are not specific. You know, Peter says the specificity for their standard test, they used about 30 percent. We just looked at 835 people who came in and got a unit of blood in the first hour out of our emergency refrigerator, and the number who went on to get massively transfused was 37 percent.
You know, even with the best judgments that we can hasten available and start giving blood, we're not particularly good at picking up the number of people who will be massively transfused. Many of them, you know, John puts his clamp on and they're fine. Others, you know, somewhere along the line they drop out of that process.
DR. KEY: Okay. Peter, go ahead.
DR. RHEE: If you get a guy with a AAA in front of you, that hasn't bursted yet, his base deficit is still okay, blood pressures are okay, and he is not coagulopathic yet, right? Then he bursts loose and he is still not coagulopathic in the beginning portion. So we all know, you know, everybody in the audience knows about shock, and if you go without perfusion for a while then your pH goes down and your base deficit and lactate go up and your blood pressure eventually goes down.
One of the major problems we have is that we have no way of telling, you know, whether, how much they have bled. We actually have tests that can show you what your blood volume is, and then we can do that in the emergency department. But we don't know what a person should have either and we're so rudimentary in that phase because the guys who made up the ATLS table did it in a hotel room and they did it off the top of their head. And so there is no -- because if you see a person and it looks like they bled a lot, we teach all our medics there is no way for us to know exactly whether that bloody shirt is a liter or 250 ccs, and that's why we rely on physiology.
But when we rely on physiology we don't even have the fundamental animal work that shows this is what your vital signs are this percent of time when you bleed this much. You know, the studies that they did in the trauma trials where they show -- of the people who are hypotensive, pressure less than 90, you know, a third of them don't even have a -- have tachycardia. So if you want to score -- to use a score in all those things, we need to make an inclusive score that captures a lot of these people, go with blood pressure.
And then if they happen to have heart rate or something like that, fine, if they also have a base deficit stick that in there. If they're coagulopathic, that's the sound of badness going on. It will probably be the group of guys that need to have some intervention.
DR. KEY: All right.
SPEAKER: I think that's a very important point. And if you could use a particular entry criteria as entry but collect as much data as possible, relevant to some of these other scores and what they've looked at so that you could evaluate that database for any particular parameter's relevance for patient age, type of injury, rate of bleeding, it might be helpful or informative later.
DR. NAPOLITANO: Yeah, that's essentially what the ROC trial did. So the ROC trial had predefined subgroups that they would analyze and in the end you can identify again meaningful -- meaningful both inclusion criteria along with the endpoints. So I think we'll learn a lot from the ROC trial in terms of the subgroup analysis which we don't have yet.
DR. KEY: Okay. I think we'll open it up for some questions. And I'll take a couple from here. They are, not all the questions at this point I think fit into the particular issue that we're addressing, so if you send a question, we don't address it, it may come up later. Let's start with a good one here.
Hypothetically, if fresh-frozen plasma was invented today, what studies would be required for approval, end points and design? I don't know if anyone would like to comment on that, but I think we're talking about the same issues we're talking about for any other product. Anyone like to take that one?
SPEAKER: Well, somebody would do a very narrow study on a situation in which fresh frozen plasma appeared to work and which they could get the FDA to agree to and we would get it approved for something and then use it widely because it appears to have many --
SPEAKER: Maybe use a off label.
DR. HOLCOMB: And then we'll use it off label, I mean, it's -- you know, it's the problem we've gotten into with trying to do, you know, there is thousand proteins in plasma and many of them have useful effects and we would find one that we could get a license.
SPEAKER: Would you -- if you were bringing FFP into market today you couldn't get it approved because the FDA would want you to show a 28-mortality advantage.
DR. NAPOLITANO: You know, I think the bigger question would be now that we have studies albeit the retrospective that all show the signal of mortality benefit with plasma, with early plasma transfusion, who would be willing to randomize patients to a perspective randomized trial? I would have a concern about that. I'm not sure that that would be ethical.
SPEAKER: I want to also remind you if you look at the 2006 New England Journal of Medicine article on aprotinin issues, table two, the risk of renal dysfunction with aprotinin has a odd ratio of 2.4. And if you look at FFP it has an odds ratio of 2.41. Packed red blood cells, 1.6. There is a lot of interesting data about renal dysfunction, other issues. Again it depends on the patient population, the way you look at it. And look at the most common cause of TRALI, in the current era the use of FFP to reverse warfarin. So it's yin yang and you have to be little cautious.
SPEAKER: Yeah, I was just starting to mention the TRALI issue, I'm not sure FFP would ever be approved, I think that would be the problem.
DR. KEY: Well, there is quite a few questions here, and this -- we may come back to this in the next 2 days, but they focus on essentially what other biomarkers, if we want to call it that, could be used to identify. So I'll just quickly, I'll group these ones together, what is an acceptable clinical test for signs of platelet dysfunction in early trauma to use as an indication for platelet transfusion other than platelet count. And again if you want to think about your answers to that.
Should -- what other coagulation factor parameters should be measured, is asking the panel to characterize bleeding. I think it says only PT is reported and not complete enough to define, so I think that's a question of the global nature of the PT versus specific clotting factors.
There is one about a device. Would a device providing real-time continuous output of skeletal muscle pH and oxygen saturation along with hematocrit non-invasively be superior to reliance on blood pressure and heart rate? It might be from somebody with a -- from a company.
Should determinant, determination of eligibility include evaluation, e.g. with ROTEM or TEG device in the trauma center and then target specific intervention. So I think they are sort of thematic here that I think Lena made the point that we have the issue of timing, sure, but perhaps one the issues that come out of this might be as it was in the April meeting as what do we need going forward here to look at these patients and categorize them.
SPEAKER: One of the -- I'll go first. One of the I think most beneficial and to me makes most common sense although the test is not perfect yet is the TEG. The TEG is the only thing I have that I can use to point to the product that I need that actually might work, obviously the TEG has its problems because it's not a very rapid turnaround. But I think what John's studies have show, the fibrinolytics and the fibrinolysis that's going on is a very important issues.
And the TEG is the one place where I can get a good look on how much fibrin or at least I'm -- whether or not I'm having fibrinolysis, and I do have a product that I can actually treat that with. So I'm a TEG fan now, I wasn't before, and I think it's actually a pretty good study. So I think it should be included especially if you're going to be honing down on what your bleeding problem is.
SPEAKER: Well, there are two machines in the world, the thromboelastogram which is the TEG and there are rotational thromboelastometry which is the ROTEM studies that look at these whole blood clotting tests, the ROTEM you can do intrinsic, extrinsic and fibrinogen-type related, they're called FIBTEM, they are very much influenced by fibrinogen which is important because I think that's -- from my perspective an important part of when you look at how to correct the coag defect.
Regarding the platelet abnormalities, we've been struggling with this for years because most of the point-of-care platelet function and even whole blood or PRP-type aggregometry requires a critical platelet mass, and our patient population with dilutional coagulopathy none of these platelet function tests work. The TEG and ROTEM still work to some extent but again it's highly influenced by fibrinogen and by things like von Willebrand factors, things that are in cryoprecipitate and fibrinogen.
Any point-of-care test, any testing you do in a massively bleeding transfused patient prevents you from going from empiric to more objective therapy. At least in the surgical literature and cardiac surgical literature we use alginate blood. The trauma literature, there is a lot of interest in these tests, the news, and I'm going to let someone who know more about that than I do, but these are two very interesting tests that provide a novel paradigm but you can't do platelet function testing with dilutional thrombocytopenia and dysfunction that occurs commonly in our patient population.
SPEAKER: The TEG is great and I've been promoting it for a while on my slides. And -- but it depends on whether you're doing research and if you have money to do the TEG. What I mean by that is that on Monday morning the cardiothoracic surgeons schedule a case, and they have a lot of people in that room and they play this orchestrated event which is, you know, controlled hemorrhagic shock and that -- they use TEGs very effectively in that type of scenario. The problem is, when I work everybody goes home except for my team, okay, the nurses that are working in the ER, they turn off all the lights in whole hospital expect for down in the trauma bay, all the anesthesia, everybody is kind of gone.
You know, so in the front lines when they come in at Friday night at 2:00 in the morning, I don't know how to get that test done quickly and easily. And I know that there are some guys that have used it in the lab and then brought it to the bench work, but the average clinician the TEG is just too difficult to use and it takes too much manpower and takes too much time.
SPEAKER: I will echo what Peter says, it can't be used as an inclusion criteria into a study for a systemic agent. But I think regarding what Amy said, it can be used, meaning if you have it available or you include it as part of the study protocol to look at afterward to do a post-hoc analysis, I don't think it's meaningful to include as an inclusion criteria.
DR. KEY: Go ahead, John.
DR. HOLCOMB: That may change in the future. There are many different ways to do what a TEG does, one can use ultrasound, one can use capillary viscometry, this would mean you could have a TEG that could turn upside down and still have work that would be cheap to manufacture, that could be in throwaway cassettes, there are just many. And I want to thank the U.S. Navy for actually putting out the SBIRs that have brought many of these products, you know, the people are working on them. Just going to have to see how they pan out, but they are -- you know, one of the nice things about capillary viscometry is you can actually measure the platelet count while you're doing the platelet function. And so you can actually get a per platelet function level.
DR. KEY: Okay. Actually for the last 10 minutes if people want to come to the microphones we can certainly take questions. I'll just setup the last -- some -- if your questions are not answered here I'm going to hand them to Jaro perhaps for the future sessions, because some have to do with management and algorithms and so on which we don't want to deal with. But there's two questions here that might go together, and I think it will come up again later. But if the panel feel like discussing quickly at this point, one is defined "Entry at EMT or ER." The other is given the focus on inclusion criteria, shouldn't there be a larger focus on post-treatment exclusion criteria to refine the dataset to the patients of interest in the appropriate arms.
So this is really to deal with trial design. So -- a few people shaking their heads. If you'd like to comment on that then we'll take a few questions from the floor.
SPEAKER: Just regarding the ROC trial, the entry criteria were really initially evaluated by the EMTs in the field. So if a patient had a blood pressure of 70 in the field, they could be enrolled in the clinical trial. Again very different than waiting and not considering the patient at all in the pre-hospital phase and waiting until they arrived at the trauma center and the trauma bay. Again, I showed you in the ROC trial that most of the patients that were enrolled in the ROC trial that had a blood pressure of 70 in the field, by the time they go to the ED, their systolic blood pressure was 110. So if we're trying to define patients that are requiring active resuscitation that may be the place to start.
DR. KEY: Anybody else like to comment on that?
SPEAKER: Well, just that secondary analysis, you often lose the safety signals.
SPEAKER: And finally when you're concerning entry criteria with our elderly trauma population coming to us and the amount of platex (phonetic) being used, and the two new anti-thrombin drugs that are going to oral and not morphine that will take over the market, you're going to see a lot more massive bleeding, and the real problem is we don't have an antidote.
DR. KEY: That's a good point. Dr. Hoots?
DR. HOOTS: So Dr. Rhee was talking about 28 or 30 day mortality and the problems with that, and then Dr. Napolitano presented data from ROC about traumatic brain injury impact at least as one of the sub-markers. So a lot -- many, many years ago we showed in a large prospective neurotrauma trial that currently the effect on coagulopathy could be segregated based on whether TBI was there or not in terms of polytrauma. So I guess my question is particularly as long as we are considering mortality endpoints should we be stratifying these trials according to whether their TBI is present or absent?
DR. KEY: That's clearly very important. Please identify yourself and go ahead. We'll go here and then here. DR. GOLDING: Dr. Golding from FDI. You know, there was some intriguing papers about NIRS to measure oxygen saturation in muscle, and I'm not sure that it's being fully validated, but what do members of the panel think about that test or any other test that could be used point of care repeatist (phonetic) for measuring profusion?
SPEAKER: Well, that -- so I participated in that trial, this NIRS study, NIRS the Near Infrared Spectroscopy. It's like a pulse oximeter goes on your feet and muscle and shows hyperperfusion basically. And what I try to tell the company that you shouldn't try to -- their problem was they were marketing that tool as a predictor of multiple organ failure. And I was trying to get to convince them to sell that tool as a noninvasive way of measuring base deficit, because that curve is like right on. Okay, it mimics perfectly well.
The multiple organ failure problem with that study was that 15-20 years ago when we were drowning people with crystalloids, multiple organ failure was our number one cause of death. Since we started using plasma and we stopped using crystalloid, that's gone away now. So they did the study in rural seven centers. It took them years to try to find the sickest patients and they kept redefining the patients, because they couldn't find sick enough of a patient. So they got these hypotensive patient with tore torso injuries, with lung, bone injuries that required blood and all these things and showed that a lot of the patients showed that they got multiple organ failure.
There was only 50 people that got multiple organ failure in that study over like 3 years, okay? If a guy comes in and he's that sick and dying, hypotensive, tachycardic, getting blood and all this stuff, and you can't tell that he's going to multiple organ failure, you should not be in medicine.
DR. GOLDING: Right? You've got all these guys coming in perfectly well and you've got the sickest guy there, do you need a tool to say that number's low, he's going to multiple organ failure? Come on, let's come up with that. So the problem with that then you said I think is a noninvasive way of adding data that, we need to say. And if it's a low number, yeah, that guy's probably hyperperfused. But somebody with an arterial injury, for example, their lactate level may not be up, you know, versus a venous injury who bled long, you know, or pelvic fracture that's been bleeding slowly, those peoples' lactates go really bad. But someone an arterial bleed, the blood pressure falls quickly because there's a hole in your arterial system and they haven't had a chance to have a buildup of their lactate levels. So all of these information is useful to put into the entry criteria, but none of them are going to be that predictive.
SPEAKER: Yeah, just a comment. I think the problem with it is that we identify all these patients who are in hemorrhagic shock by very simple measures, blood pressure, heart rate. So that device did not add anything else in the hemorrhaging trauma victim. That's the simple issue.
DR. KEY: Question over here.
DR. MOORE: Yeah, I've two comments; Moore from Denver. I'm a trauma surgeon. Number one, I think as you look at the data here and certainly have looked at it over the years, it's clear that the ball game is won or lost in the first 2 hours. Most people die within the first 2 hours, so intervention has to be early. But in terms of endpoints and I'm always mystified why we continue to talk about a massive transfusion at 24 hours. It seems to me a much more rational definition of massive transfusion is 10 units in 6 hours.
Number one, that's when we give most blood in trauma. Number two, that's when the pharmakinetics of all the factors, et cetera, which seems to make more sense in a clean definition of massive transfusion. My second comment is on identifying the patient at risk for coagulopathy. We have used the TEG routinely in our trauma center for over 3 years. We obtain a rapid TEG at them time the patient enters the trauma bay and have a result within 10 minutes.
The fact is, and we have a manuscript in press right now, the TEG at the time of admission to the emergency proper predicts massive transfusion better than any constellation of vital signs or other biomarkers you can use. While the TEG might not be perfect, I suggest if we're going to enter patients into a study to study coagulopathy, then we have to identify the problem at the time they reach the hospital. Thank you.
DR. KEY: Okay, thank you. Marty.
DR. SCHREIBER: Thanks, Nigel. We're spending a lot of time talking about inclusion criteria which of course is critical in a study, but we haven't talked about methodology at all and I think methodology is equally critical.
As an example we talked about the CRASH-2 trial which used randomization by indecision, which introduces a large amount of bias. There's been a lot of discussion about the ROC trials. I want to make a point about the ROC trials is that is we didn't really hypertonic saline versus normal saline. We studied an initial bolus of hypertonic saline which was 250 ccs, but in the pre-hospital setting those patients received over a liter of normal saline all three groups.
So we studied hypertonic saline plus normal saline versus normal saline. And I think that introduces a huge problem for numerous reasons. One of the reasons why the study was done is because we wanted to look at was hypertonic saline a good fluid to give in the field in a military setting. Well, medics don't carry two liters of normal saline with their 250 cc bag of hypertonic saline. So I think that's a problem. Furthermore, that early mortality I believe is secondary to the fact that these patients were over-resuscitated early, which potentially could even result in a pulmonary edema or congestive heart failure.
So I'm very interested in the panel's comments and their thoughts about this. Was this trial completely done incorrectly and therefore we have a negative result.
SPEAKER: I was on the ROC -- the protocol review committee and also on the DSMB. We anguished over those results and there were many goals of that. You know, to think that you give a little saltwater in the field and that's going to end up in survival 28 days later I think is being very naïve. And we're looking for that silver bullet, but if fluids can be given in a different way, and it keeps the same amount of people alive in a day, that's useful data and information.
If it can keep people -- more people alive for a few hours that's useful information. But waiting just 28 days to say one dose of saltwater in the ambulance makes a difference and I think that's the problem is that they were forced to use 28-day mortality for that study. So when we sub-divided all of these bins, we found one bin of people who didn't get blood that had a worse outcome if they got some sort of hypertonic saline. And I think it's probably because it basically showed what Dr. Mattox showed in the past, lots of fluid, you know, 2.5 liters of fluid is what the hypertonic saline is versus just 250 of placebo, this probably showed that it might be killing people by giving that too early.
But on the other hand if you look at the people outside of that bin, hypertonic saline actually tried to reduce, you know, multiple organ failure. But we also canceled the study because there wasn't going to be enough of a divergence and outcome at the end, and we thought we were going to be wasting a lot of government money. But the military wanted to know from equivalency, can we use this, because there's a lot of weight difference for the medics, and that data clearly shows it's probably okay, but you don't want to be going crazy with that fluid in the beginning.
DR. KEY: Okay, I think for the sake of time, we'll just take the last two questions here. I'm sure there will be an opportunity to catch up later. John, please go ahead.
DR. HOLCOMB: Holcomb from Houston. I like the panel to discuss the conundrum of sensitivity and specificity which is what these entry criteria are with clinical care and research. So they're actually extremely different. You want to maximize one for clinical care, and you want to maximize the other for research purposes. And then the reason I showed 900 massive transfusions in my slide up there is you can balance that sensitivity and specificity from a research point of view. You'll -- we could miss half the MTs, but your actual MTs in your enrolled patients in a study could be as high as 80 percent. John Hess or somebody from the panel?
DR. HESS: Yeah, John, you have seen our data. You know, we looked at that over a 5-year period and used the surgeon's decision to hang the first unit of blood out of the open trauma refrigerator. And again the -- you know, depending on what else you want to add to those patients, you can increase that specificity, but you know, again just by looking at the patients and making aware that decision is made. But you know, it's clinically the people are thinking in terms of the sensitivity, they don't want to miss anybody.
DR. KEY: Okay, last question, Colonel Hack?
DR. HACK: Many trauma patients, particularly our trauma patients are young and relatively fit. Therefore I caution you to be so simplistic in your inclusion criteria of blood pressure and pulse. They frequently don't -- they're able to compensate, so they frequently don't have a -- raise their pulse rate or lower their blood pressure until it's too late. And we need to find other methods to identify those patients early, and I would like your ideas on what we can use in that area.
SPEAKER: Can I make a comment about that, because there's a intensivist and working a lot in the cardiac surgical patients, you know, echocardiography is like main stay. And any patient in the ICU post-op issue, the first thing I'll do is try to get a transapical view of the LV. Is the ventricle full or empty? I mean, it's just -- a picture is worth a thousand words. The patient becomes hypotensive, do I have acute papillary, ischemic papillary muscle dysfunction, I might have seen ischemic mitral regurge. I mean, it's an amazing tool.
TEE is great, but I mean, I think you can get most of the time a transapical window if nothing else to look at LV and RV, you know, dynamics, effusion things like that. I think it's time to think about redefining it. I'm glad you brought that up, because as somebody interested in shock from a hypersensitivity anaphylactic septic perspective and the mediator just -- so many things cause it, but the question really is are we empty or are we full? And a transapical view might be a really nice way to cutely pick it up. It's technology that's being used. And I don't know, Dan, in your cardiac ICU if you guys have that ability, but it's -- I think it's a great thing to think about.
MR. MYERS: Yeah, we definitely use it in -- we're back to lifting the legs up and watching the ventricle fill or not fill. Simple as that.
SPEAKER: I think it's an incredibly important point not just in the young patient population, even in the elderly population, some patients are very fit. I think one option to consider is if we want to use continue to use, systolic blood pressure a 100 or 90, then in the end if we enroll 50 percent of patients that have zero blood transfusion, those patients should be excluded from the analysis at the end. We don't want to be studying patients that in the end get zero blood. And I think, you know, what I showed in review was that a lot of these trials enrolled many patients who were hypotensive at some point, but had no transfusion, and in the end they're included in the analysis and it's unreasonable.
I think it's a very important point. I personally would like to see a lower blood pressure, and maybe it needs to be still at 90, but with lactic acidosis or base deficit, and I think that would take care of that issue.
DR. KEY: Okay, well, thank you very much. A great discussion. Put it back to Jaro.
DR. VOSTAL: Thank you very much. Thank you Dr. Key, that was very nice. And thank you for the panel members for the lively discussion. So we've come to a point in the workshop where we have a coffee break. So we are now -- the next session will be at 11:00 o'clock, so please be back by 11:00. Thank you very much.
ANIMAL MODELS FOR PRE-CLINICAL EVALUATION OF HEMOSTASIS PRODUCTS
DR. VOSTAL: Thank you. All right, so we're going to continue with our session theme and session number two will be on pre-clinical animal models. And our first speaker, the introductory speaker will be Dr. Anthony Pusateri who is a research coordinator of the Hemorrhage and Resuscitation Group and the Combat Casualty Care Research Program at the U.S. Army Medical Research and Materiel Command. Dr. Pusateri?
MR. PUSATERI: Good morning ladies and gentlemen. Thanks Dr. Vostal. It's a great privilege for me to be here today to talk with you a little bit about animal models to study or to -- in the development of products for severe bleeding. What I'm going to express are my opinions, not the government's, and specifically what I'd like to do is discuss key features and selected animal models for use in development of products for bleeding and trauma by citing these individual models and making these recommendations, not trying to eliminate other specific models. Also I want to point out that this discussion does not apply to development of products for congenital deficiencies.
The primary purpose of the studies that I'll be discussing is really to provide what you might call pivotal animal proof of concept for hemostatic efficacy and also provide additional evidence for safety to support movement at clinical trials. These studies are not meant to replace the specific models that are used during development to provide preliminary data and proof of mechanism or dosing information. Also these are not meant to replace ADME and standard toxin safety studies, although we might see in the future that we want to alter some of those safety studies for these types of products.
The models are suitable for studying pathophysiology, but that is not the specific purpose of the studies I'll be discussing. Also these products -- these models are often used in studying other treatments for shock, for example, oxygen carriers. That's not what we'll be discussing today, but I just want to bring up one point that's often overlooked is that those studies should also consider the hemostatic mechanism.
There have been some review articles that I think if pointed out possibly a hundred factors to consider in standardizing hemorrhage models. I just want to highlight a few that are very specific in this case for models for -- to control severe bleeding.
First is the relevance of the hemostatic challenge, in other words the coagulopathy or the vascular defect or both. I've also looked for a proven, reproducible -- reproducibility in the model, evidence that the model has been standardized, the severity of the model, and absolutely critical for this discussion is the ability to demonstrate hemostatic efficacy. The surrogate markers such as thromboelastography or other measures of coagulation and hemostatic function are very valuable. Also survival is a very valuable endpoint. But in these studies it's critical that you can demonstrate that a product improves hemostasis directly, for example, reduce blood loss, decrease time to hemostasis, reduced incidence of free bleeding. If you can't do that in this study, it's really unlikely you would be able to do that in a clinical trial.
Remember, these products are specifically designed, the ones we're talking about, to control bleeding. And also these animal models, you have to be able to apply the product as it's designed. In terms of safety or what I call hemostatic safety we want to be able to assess thrombotic risk, the relevant thrombotic risk and tissue reactions.
For this discussion, severe bleeding is defined as bleeding that will lead to significant mortality or morbidity if it's not controlled. So in the situation for a patient or for the -- for the patient, the bleeding can't be controlled by the body's natural mechanisms under the patient circumstances. There's a number of ways to categorize products for severe bleeding, topical and systemic, internal use, external use. What I've used for this discussion is external use in the pre-hospital or pre-surgical setting versus internal use in surgery. And then that can further be broken down to topical use in surgery or systemic use in surgery as I mention those other combinations.
Next is to look at what does the hemostatic product need to be able to do. You can have severe hemorrhage resulting from a large vascular defect that overwhelms the body's normal system, or you can have severe hemorrhage due to an impaired hemostatic function so that the body is not able to seal vascular defects that normally would not be lethal, or you could have both. For the external use product category, pre-surgical setting, the predominant cause is the vascular defect. Therefore the product should be able to control bleeding from large vascular defects, so that's the type of a model we would need to use.
For internal use projects -- correction, products, the predominant problem is impaired hemostatic mechanism. So the product should be able to control coagulopathic or non-surgical bleeding. Now, in actual use there may or may not actually be a coagulopathy in the situation, but it can never be ruled out. So we will use the -- placing that requirement in the animal model will ensure the worst case scenario. And there are different animal models for each.
For external use products, I'd like to highlight some of the selection criteria. In looking at a model, the factors that determine severity of the hemostatic challenge involve vessel diameter, the blood pressure, and the time from the beginning of uncontrolled hemorrhage to the application of the product. Some general assumptions is that we will begin with a normal hemostatic mechanism. Bleeding will be accessible to the application of the product. The hemostatic product has to function without vascular control -- without otherwise having vascular control. The severity is such that there can be rapid exsanguination and in the general concept for these products is that it provides temporary vascular -- temporary hemostasis to allow time for transport and surgical repair. They're not an end in themselves.
First to comment on vessel diameter. Of course that's specific to the blood vessel and to the animal species. The tension within the vessel wall is related to both the pressure and the radius of the blood vessel. To stop bleeding, the product must be able to form a barrier that can hold against the tension that's characteristic of that vessel under those conditions.
Just as an example on the left-hand side, the table shows the different blood vessels, human blood vessels, example of blood pressure is radius and then you can see on the right side tension in the vessel, across the vessel wall. Just -- I want to focus in on why this is important in species selection. So you can see in the top-line the characteristics of the aorta for a human, and the tension in that -- across that vessel wall. Now, to a rat or a mouse their aorta is just as important in terms of physiologic function, but in terms of a hemostatic challenge for a topical application, it's much lower.
If we just use as an example, the arterial that's listed here, even allowing for a slightly higher blood pressure, you can see there's a greatly different hemostatic challenge as characterized by tension in the vessel wall. So in this -- species is important here and this leads us to swine being the preferred model based on size of vessels. Another important point is the timing of application following the induction of uncontrolled hemorrhage and I want to point out here that these -- we're using uncontrolled hemorrhage models to assess whether something can stop bleeding or not.
On the -- in this idealized blood pressure diagram, on the vertical axis is blood pressure in millimeters in mercury, across the bottom is time. So you start out with a 90-millimeter mercury mean arterial pressure. At the time of vascular injury, there's rapid bleeding and rapid reduction in blood pressure. Now, you could see that application of the product at 30 seconds represented by the upper arrow versus 5 to 10 minutes will result in a very different tension -- very different hemostatic challenge for the dressing.
In fact when the dressing is applied at the nadir of blood pressure, there may already be a hemostatic plug formed. So -- and as I'll mention in a couple of these -- one of these -- some of these models, timing of application is a very important factor to control.
What I want to discuss is a generalized -- is swine model of groin injury. And just as a general procedure, all these models are anesthetized, the animal is in a normal state including temperature. Groin injury is induced. There's a free bleeding phase, dressing application phase, resuscitation and observation. And the main measures are blood loss, time to hemostasis, incidents of rebleeding.
And you can see just as a general look on the left-hand side what the blood pressure profile would be for that, beginning at time zero with induction of the injury, decrease in blood pressure to about in this case around 5 minutes or so application of a dressing, followed by resuscitation, and return of blood pressure.
There's two leading variations for this model. One is a 6-millimeter punch biopsy in a single-wall of the femoral artery plus the use of lidocaine to eliminate the effects of vasospasm and retraction of the vessel. This model is more lethal than the other leading model which involves trans-section of the femoral artery and vein. That does allow retraction and vasospasm. Of course it's also more -- a common injury pattern when relating to the actual patient and it's a less lethal model.
This shows a cut-down view of the femoral vasculature showing the femoral artery and vein. The data line shows the approximate point where the trans-section would be made for the trans-section model.
Arnaud et al recently published two studies studying the same -- the exact same 10 hemostatic dressings using the trans-section model and the puncture model. The tables here show percent survival through the study period and average survival time. And a very interesting finding was that there was almost no change in the ranking of the dressings, which is shown across the bottom. The top four were exactly the same. The bottom six there's only one or two changes in order. So they did yield similar findings. However, it was -- it did appear that there was a greater hemostatic challenge for the hole-punch model although either could may be necessary based on the format of the product to be tested.
I'd like to focus in on the DOD consensus which incorporates the punch biopsy and the paper is in press for that. The DOD -- the objective stated in that paper is to demonstrate that a test agent is significantly more effective in controlling hemorrhage than a control or some standard of care. Primary endpoints are post-treatment blood loss, time to hemostasis, blood pressure, and of course survival. Secondary endpoints include a number of physiologic parameters including the ability to gather quite a bit of data on hemostatic functioning including TEG and that's another benefit of using swine given adequate blood volume to take repeated samples.
Some points that may seem small right now, but when really considering these models these are important distinctions; in this model, the consensus model, there's no splenectomy. The artery is -- femoral artery is bathed in lidocaine for dilation to minimize vasospasm. There's a start -- required starting mean arterial pressure before the vessel is damaged. Then there's a -- in the process of forming the punch, the artery's clamped proximally and distally, then a 6 millimeter diameter arteriotomy is made using a vascular punch.
Clamps are released and there's a 45-second free bleeding period. Then there's -- in the bleeding -- the free-bleeding wound, the dressing is applied with standard number of laparotomy sponges and a standard pressure for 3 minutes. Resuscitation involves 500 milliliters of Hextend followed by up to 10 liters of Lactated Ringer's, going back to the pre-injury mean arterial pressure as a resuscitation target.
Following monitoring CAT -- CT scans are done to examine vessel patency and the leg is mobilized to look at the durability of the hemostat simulating movement of the patient and then tissues are collected. This shows the cut-down view of the femoral artery in the upper left, then forming the vascular defect, the free bleeding phase, and followed by dressing application. You'll notice two blue probes; those are temperature sensors, which brings me to a -- my next point. And that is something that I think has been relatively overlooked, and that is the hemostatic safety of these products. In internal use products, I think this has been fairly consistently done. For external use products, I think it's been overlooked.
Now, due to -- we're talking about severe bleeding. So due to the nature of the injury, we have to make the assumption that the hemostatic material will come in contact with the endothelium, surrounding tissues, flowing blood. And so important considerations include the local tissue reaction, the vascular reaction, the potential for hemostatic particles to be moved through the vasculature, thrombosis, and possible requirement for extensive debridement.
In the last maybe 5 to 7 years, there've been a couple of fairly significant issues. One was where an approved topical hemostat caused pain and thermal injury. Another approved topical hemostat caused endothelial damage and possibly thrombosis, and there was some evidence of trans-location of hemostatic particles.
The issues have now been addressed and the companies are working on these things. The information has been published, but the point is that those were revealed after approval of the product. So my -- to me I think more attention needs to be placed on examining hemostatic specific safety for topical agents and really for all agents based on the characteristics of the product. And there have been some recent studies that looked at this systemically and are noted here.
I'd like to move to animal models for internal products and first I'll be covering systemic application. Once again, for internal use the product should be able to control coagulopathic bleeding. Therefore the appropriate animal model should incorporate a relevant coagulopathic component. Some other key features is once again similar vessel and organ size to humans, the ability to quantitate hemostasis, and the ability to induce shock and fuse fluids, transfuse, and apply the treatments. And once again, the leading models are swine models.
Parr et al reviewed coagulopathy animal models recently and identified six key initiators of trauma-associated coagulopathy, and as was mentioned earlier, we're just beginning to understand this phenomenon. But these key indicators are known to be involved, and those are tissue trauma, shock, hemodilution, hypothermia, acidosis, and inflammation. Animal models for testing internal use hemostats -- hemostatic agents should incorporate one or more of these.
The models I'd like to highlight are isovolemic, normothermic, hemodilution models. And these have been especially successful in some European studies. Some common features for these models include a 65 to 80 percent hemodilution with hetastarch, either normothermia or mild hypothermia, replacement with washed red cells to prevent early death from anemia, standardized injuries -- either liver, bone, and spleen injuries have been used ; the liver's the most severe.
There's no additional manipulation of the injury, no aggressive resuscitation required. Then there's a 2-hour study period and then the opportunity which has also been published to look extensively at laboratory parameters of both thrombosis and hemostasis.
This shows the -- on the left, the splenic incision model, 8-centimeter by 1-centimeter incision. On the right shows the -- what's considered a venous bleeding model by drilling a 3-millimeter hole in the femur -- at the head of the femur to the point of penetration to the bone marrow.
I'm summarizing a few papers here, but this just shows the status of these animals prior to the hemostatic defect, prior to the injury. So at the top -- this varies from lab to lab, but the mean arterial pressure is in the 50s or 80 possibly due to differences in anesthesia, but this is a mildly hypothermic or normothermic model. You can see the evidence of dilution in the platelets and hemoglobin and fibrinogen. This is a normal ph model, no acidosis, but there is extended PT, expanded -- extended clot formation time in the ROTEM and decreased maximum class strength measured by ROTEM. Skin bleeding time is extended, SBT, but that's not useful for assessing hemostatic products.
The spleen bleeding time is extended, the bone bleeding time is extended, and these are all significantly -- spleen blood volume loss is greater and so is bone blood volume loss. I'll just review some -- very, very briefly studies that we're actually able to use in these models to show a difference between hemostatic product and a control.
Dickneite et al published a study looking at prothrombin complex concentrate compared to placebo, found a decreased spleen bleeding time, decreased bone bleeding time and decreased bone blood loss. They published another study looking at more than one product, but the main point here is that PCC once again decreased spleen bleeding time. In another study comparing PCC to fresh frozen plasma, spleen bleeding time, bone bleeding time, spleen blood loss, and bone blood loss were all reduced with PCC compared to fresh frozen plasma. So the point here is not to highlight a specific product, but to show these models have been successful in differentiating hemostasis among products.
Now, this is a more severe version of the same model. I want to just skip directly down to the first major difference and that's the fourth bullet, the standardized liver injury, which is in this case, 12 centimeters long by 3 centimeters deep. The blood loss in this study looked at fibrinogen concentrate plus PCC versus saline. Blood loss was significantly reduced and you can see the increased severity of the model here with almost 2 liters of blood loss in the control group.
Survival was improved and another greater additional evidence for the severity is that there was zero percent survival in the saline group. Of course, all coagulation parameters that were measured in general improved in all the studies I've mentioned so far using this model, and there was no thrombosis detected. I want to now transition to another model that is fairly popular, and just to say upfront, I would not recommend this model for systemic product. This involves hemodilution plus hypothermia in a severe grade 5 injury. This has been published a few different ways. Hemodilution with 6 percent hetastarch, either with a high or low molecular weight hetastarch depending on whether it was done in Europe or the U.S., or albumin.
Hypothermia has been standardized. There's very severe injuries induced that's lethal in less than 30 minutes if untreated; requires packing so that the product must be an adjunct to packing. It's a very aggressive resuscitation regimen and there's been varying results with different -- in different studies.
So I would say at this point this is not well standardized and I would not recommend this for use for systemic products. However, it has been very useful for topically-applied internal use products and there have been a number of papers published. The one cited here is actually a product that I believe is now in clinical development. So I would recommend it in that case.
Something else I want to bring up in terms of topical internal use and that's the importance of duration of efficacy. You can imagine how this would be important, for example, if a product -- topical product were used during damage control surgery awaiting final definitive repair. It would be very important to understand how long that product would function. A very nice study was published that used the model of infra-renal aortic punch. You could see in the lower left-hand corner the severity of the bleeding there, followed by a standardized compression dressing application in the upper right-hand corner. And in addition to physiologic parameters and survival monitoring, there was repeated CT scans over 4 hour -- 4-day survival period, and what that allowed was the ability to -- in addition to monitoring survival, monitor the pattern of failure of the dressings.
I've talked about models that incorporate one or two components known to be involved in trauma-related coagulopathy. There's another model that uses multiple components, and I -- this actually -- the reference here should be Hess (phonetic) et al 2008. And that's a recently published model of polytrauma shock, coagulopathy, and uncontrolled hemorrhage. And going from left to right, the sequence in this model involves a femur fracture, 60 percent hemorrhage, hypothermia, a 30-minute shock period followed by resuscitation with normal saline at three-to-one times the blood volume lost, then a stabilization period. And that establishes a very consistent coagulapathy -- a very consistent and relevant coagulapathy.
After that is a -- in the models that that have been reported thus far is a grade V liver injury with free bleeding phase and then treatment, and a 4-hour observation period. On the right-hand side, you can see the severity of both the femur fracture and the grade V liver injury.
This is just up to that point where I said the coagulopathy is standardized. And I'd like to show you -- this has been published separately. This is a very, very consistent standardized that was actually used -- had been used at three centers and it was so well-standardized that a multicenter study was possible. Actually that was a transfusion study. But as you can see at the top, starting from left to right, you see the period of shock fall and then the resuscitation presumably with some reperfusion; core temperature is hyperthermic. There's evidence of hemodilution based on platelets in hematocrit.
Then there is an acidosis induced and you can see that the PT is extended -- PTT is extended. The maximum clot strength judged by thromboelastography, the MA is decreased. And another point is that in this coagulopathy induction phase and the polytrauma phase it involves about 20 percent mortality which further reinforces the severity of this model.
There's been six reports in the literature so far. The studies have been examining transfusion products or other treatments for shock. They've consistently shown improvement with transfusion and coagulation parameters. One study showed improved hemostasis between treatments. But these studies weren't really designed to test hemostatic products. So I think that the ability of only one study showing hemostatic improvement, specific hemostatic improvement measured by bleeding time or blood loss is due to the study design.
I think there's a lot of potential in this model. I wouldn't say that right now it's a -- we can say that it will be a very -- it could be used immediately for hemostasis studies, but I think there's tremendous potential here for studying the pathophysiology of a coagulopathy of trauma -- trauma, shock, et cetera, but also specifically for hemostatic products.
We have the standardized trauma in hemorrhagic shock phase which I've mentioned, but following that is the opportunity to vary the injury, which may be necessary based on the product that's going to be used, of course, to vary the treatment. And then something that I think is important is this 4-hour observation period. That to me represents a period where we can get a very good look at the thrombotic potential for a product.
If you just -- if you think about this, these models induce a coagulopathy that may mask in some cases thrombotic potential. What's necessary is it then allowing a window when coagulopathy is corrected and there is exposure to the thrombotic risk and in this case it's a very relevant thrombotic risk because of the shock and reperfusion in femur and polytrauma.
So this might be something that would be very useful to for example mimic the period from cessation of bleeding to the time when prophylactic antithrombotics can be used in the clinical setting.
I'd like to mention that there are some excellent animal models and the rabbit is a recommended animal model for these types of studies. The studies have in general -- some of the best studies involve hemodilution with or without hyperthermia, 50 to 60 percent hemodilution. There is -- these animals are large enough to allow some repeated blood sampling, to get coags and some more sophisticated variables included a limited number of TEGs.
It's both -- both of these models have produced -- reproduced. We've been able to differentiate among hemostatic treatments, which is important in these types of studies. Also we have the ability to match these models with well-established thrombosis models in rabbits for both venous and arterial thrombosis.
We get very -- swine models are very popular and often we get comfortable thinking that human products will work in swine. But there is a species-specificity issue. Recently Howksworth et al published a study examining hemostatic efficacy of a product derived from human platelets in a swine model of hemorrhagic shock. There were problems with thrombosis that were almost certainly related to the xenographic reaction, human platelets immediately aggregating when they come in contact with pig blood.
This may be a case -- this type of situation is a case where non-human primate model may be used. And there's currently a study going on using these types of products with a laparoscopic liver laceration, the shock phase, resuscitation phase, hospital phase. That study is in progress, but it has a number of the characteristics that I've identified as being important in some of these other models.
So to bring this together, swine models are recommended for these sort of pivotal studies that demonstrate hemostatic efficacy. Femoral artery hole-punch model is recommended for external use products. Dilution coagulopathy models are recommended for internal use products. The polytrauma model of shock, hyperthermia, acidosis, and hemodilution has a lot of potential in a number of areas including the study of hemostatic products.
Rabbit models are recommended for small-animals models. Primate models may be necessary in certain cases where either the product is incompatible with swine physiology or maybe there are specific safety issues that need to be addressed. And I'd like to encourage a more systematic approach in the animal safety studies for hemostatic agents.
Thanks for you attention. I look forward to the discussion.
DR. VOSTAL: Thank you, Dr. Pusateri. Now I'd like to invite the panel members to come up to the podium for a discussion and this session will be moderated by Dr. Charles Wade, who's a professor at the Department of Surgery, University of Texas Health Science Center.
PANEL DISCUSSION: ANIMAL MODELS FOR PRE-CLINICAL EVALUATION OF HEMOSTASIS PRODUCTS
MR. WADE: Good morning. I'd just like to introduce the panel members. The first one is Dr. Tim Billar from the University of Pittsburgh Medical Center, Bijan Kheirabadi from the U.S. Army Institute of Surgical Research, Dan Myers from the University of Michigan Medical School. That's Tony Pusateri you just heard from and finally Dr. Martin Schreiber from the Oregon Health and Science University.
I'd just like to make a quick comment. One of the things that -- those of us who design animal models have a very important role, especially in this -- when community consultation is required because the rules do state that you must have adequate animal data to demonstrate possibility of benefit.
So in conducting animal experiments, those of us who are tasked with designing models need to know some things from the clinicians, and I want to make these points to start with for the panel members. First of all we have to define the patient population in order to build the animal model or construct the animal model that best mimics that particular population.
We have to understand the clinical endpoints. Most of these are acute -- the models we've designed are acute studies. In some cases, those of us who've done larger animal models do carry them out for days or weeks post, but most of the studies have been very acute that is within a usually 24-hour period.
And then finally the safety issues have to be defined which we should concentrate on, look at based on clinical observations. I'd like to open the comments up to the panel members and give those a chance to comment on Dr. Pusateri's presentation specifically on his choice of a swine model. And we'll start with Tim down there, since you work on the lower end of the species group here.
DR. BILLAR: I thought Tony gave an outstanding overview and agree that based on anatomic considerations and other considerations such as sampling possibilities of the swine model likely represents the most cost-effective and optimum model, and I'm particularly impressed with the possible utility of the polytrauma model that's been most representative of what we encounter clinically.
MR. KHEIRABADI: I think swine model is a terrific model of bleeding and creating a very severe bleeding whether artery or venous. But I don't think it should be always used as a single model. I think we can learn a lot from smaller animals such as rabbit and do a lot of development part of the drugs to that stage and then perhaps at the various -- final stage confirm them on the large animals. The expenses of the large animals and the number of animals that are needed to get the product to confirm and get to FDA is a huge amount and a lot of that work can be done in a small animal and then take it in the confirmatory action relative to a swine.
I totally agree with Dr. Pusateri, the pressure that we see, the arterial bleeding that we see in the swine is at least visually is much greater than what we see in the rabbit. But at the same time something that stops the bleeding in the rabbit has a chance also to work. But if you don't have any preliminary data, it just does not seem to be -- futile, does not seem to be making too much sense to start from large animals when you don't have really data to support in small animals.
SPEAKER: Those are excellent points and I thought it was a phenomenal presentation that we just listened to. I agree that if you have proof of concept, if you go with the rodent or the non-rodent rabbit models, genetically and antigenically they're much more further away from the humans as when you get to the swine and of course a non-human primate.
I agree with your point about cost-effectiveness, but with our minipig and you know, just regular market pig numbers, you know, an average market pig is around $250 versus a non-human primate which is about 5k. You know, you've to take those considerations and then you've to take into considerations the test and your analysis and the amount of tissue and sample size you can use to come up with a, you know, a plan that -- because ultimately you want to mimic what's going on in the human condition.
SPEAKER: I think the real issue here is what we just heard was the goal of creating these animal models is to replicate to the best of our ability true human trauma. So that involves a number of things. First of all, there has to be tissue injury. There is no such thing as a human trauma that doesn't involve tissue injury. Tissue injury changes the milieu. It creates an inflammatory response. That's an absolute mandatory part of any model.
Second of all, there has to be blood loss. And that blood loss needs to be uncontrolled blood loss. And I think you know there's been a lot of -- throughout the history of studying hemorrhage and shock, you know, people have used controlled hemorrhage models and these do not reflect the way humans bleed after injury. So I think the combination of tissue injury and uncontrolled bleeding are absolutely critical.
So that brings us to what model should we use. Well, the swine model is -- works nicely because swine have adequate blood volume to, number one, have uncontrolled hemorrhage, and number two, do numerous repeated assays. The smaller animal models don't have the adequate blood volumes to do those types of studies.
Furthermore, we're here talking about bleeding and hemostasis. The coagulation system of a swine is actually quite similar to that of a human. Functionally they're somewhat hypercoagulable, but compared to lower phyla of animals it's vastly more similar. So in many ways I think the swine is the ideal model and for the reasons that Tony put out.
But I think we have to remind ourselves repeatedly that our goal is to replicate human trauma and those elements have to be included.
SPEAKER: Okay. Tony in your talk you brought up this idea that you had to have some type of coagulopathy or the coagulopathy was an essential component on the evaluation of these products. And I'd just like to get your comments to clarify that. Do you really believe that or do you just have to have bleeding in general and then subsequently if you do have the coagulopathy it's a final nail in the coffin of the model, a little bit.
MR. PUSATERI: Well, I think, for the -- for external-use models, I don't think coagulopathy is necessary. I think that those would be used in the category -- the way I categorize those, those would be used in the pre-surgical setting or pre-hospital setting. So the patient is likely to be starting with normal hemostatic system, you know with the exception of whatever kinds of drugs they may be on.
But in normal hemostatic system which will -- which will likely -- bleeding continues to develop into a coagulative -- coagulopathic system. But in this case, the hemostatic agent is applied to severe bleeding from a large vascular defect that does not necessarily have to be -- that doesn't have to be coagulopathic to be effective or it doesn't have to work in a coagulopathy to be effective. I think the situation though for an internal use product is very different. The large defects most often would be controlled surgically and the problem then becomes that you have severe bleeding that can't be controlled or structures that really can't be surgically repaired or surgically repaired fast enough.
I think that you may -- there may be situations where there is no coagulopathy, but it could never be ruled out, and so I think that the bar should be higher in the animal model for internal use to include a coagulopathy. I also think that coagulopathy should be -- involve a relevant component of a trauma-related coagulopathy understanding that we don't know the -- we don't really understand the pathophysiology of that phenomena. But for example a heparin-induced coagulopathy would not be appropriate in that case.
MR. WADE: Anybody else? Marty?
DR. SCHREIBER: I do think that coagulopathy is important, and the reason why I think that it's important is because of the papers that John Holcomb reviewed this morning.
The acute coagulopathy to trauma has now been recognized and we know that about 25 percent of severely-injured trauma patients have acute coagulopathy and trauma and those are the patients that die. We know they die early. Therefore, again in order to replicate what's happening in humans, we need a coagulopathy in these animal models because those are -- that's the scenario that we're trying to identify and treat. So I believe it's very important.
SPEAKER: I think you might take it one step further and say not just the coagulopathy, but the coagulopathy shouldn't be a dilutional coagulopathy but one that's associated with the full systemic response and the other factors that we've been talking about such as tissue injury induced inflammatory response. So some of the models Tony talked about were from -- primarily driven by a dilutional component.
MR. WADE: I'd like to make a little comment on that. I've been trying to graph this coagulopathy issue and we talked about dilution a little bit. But I think the etiology of the coagulopathy of trauma is still ill-defined. So building an animal model is very complicated. Do you want something that has hyperlysis or do you want something that has, you know, differences in the initial stages of clot formation or do we want something that platelets are involved in.
So I mean there's a wide range of work still to be done in this area. So, Tony?
MR. PUSATERI: I agree with both comments and I think that based on our current knowledge and the current status of models, the hemodilution models have been very repeatable and used very well. I think maybe 5 years from now, we will have other models that are repeatable and incorporate other components of the coagulopathy trauma that might be more -- maybe we would consider more relevant. But for the time being based on the literature, really those are the most reproducible models.
MR. WADE: We have a couple of questions that have come from the audience. If -- the questions deal with regulatory components. What would the FDA look at? But I want to know what models do you feel -- if a product came up today -- let's just use FFP as an example, if it was brought in to market today, what models would you recommend?
SPEAKER: So this question relates to your other question about -- somebody asked, well, if FFP was going to be coming forward to the FDA, what would it have to go through. Well, we actually can sort of answer that question because the process for getting approval to use lyophilized plasma which is simply, you know, plasma that's been dehydrated to a powder form is underway and that lyophilized plasma is going through the standard process that any of the other drugs was, phase I trials onto phase II trials and ultimately, phase III trials.
So I think that the answer to that question exists. Now what model would I use? Well, I really think that the best model that we have available to us currently is the multiple trauma model that was described by Tony that we developed in conjunction with the Institute of Surgical Research in Harvard.
And I believe that's the best model, because of all the models that we've seen up until this time, it's the model that most replicates severe human trauma that has not only soft tissue injury but bony injury as well as blood loss, hemorrhage control and resuscitation. So I think of the models that are available to us today, that is the best models.
SPEAKER: Well, actually that was one of the only -- that model was actually used to show I think that the lyophilized plasma plus red cells was better than something else. So it's kind of a slam dunk in this case because that model has already been successful in that. But if I just said something maybe more generic that might be a combination of a hemodilution model, which is very reproducible, combined with maybe using the trauma model as a safety component and really focusing on that 4-hour recovery period possibly taking it out to 8 hours.
If -- so instead of going directly to that model to show hemostatic efficacy, you may be able to use two different models, one for efficacy and one to show safety. And to be honest, I really don't see a model with a more relevant thrombotic risk exposure than that model.
MR. WADE: In the validation of the animal models, there've been a number of questions here about validating them in relationship to the human condition. How are your approaches and -- I hate to comment and I remember a few years ago that there was an article review of sepsis models and then the comment at the end of the article was that no sepsis model mimicked the clinical model and there was no good model for sepsis that could be used to validate products in animals. But we still wanted to do all the different models. And so I would like to see if you have a similar approach to that.
SPEAKER: What I think is an interesting theme about the discussion of models is that the model development is evolving, and what's nice to see is that the model evolve -- evolution is occurring more quickly much as our understanding of the pathophysiology of trauma.
So this model development is going to continue and I think the validation point is critical. Something that we know that works well, and at least has a signal that humans should work in the model.
And I'm not sure we're at that stage completely, but you know, I think some of the problems with sepsis related to necessarily being able to correspond in terms of therapeutic efficacy. But I think we have an opportunity in developing these models to set the standard that there should be a positive control if you will. Something that we know at least has a signal in humans also appears to have a signal in the model of similar magnitude.
MR. KHEIRABADI: I always have this problem that -- and I think several of my papers been revealing that is the model that you're creating really represent the wound that we've seen in the trauma. And problem is always that in the trauma, you have so many different things going, so many different type of wounds. How can I have one standardized model that can replicate every one of those? And that's impossible. You have to stand -- in order to compare things, in order to test the efficacy you've got to have to have a standard model.
The standard model by definition is a reproducible system which you don't really have in trauma. In trauma you have different bleeding, different injuries, internal, blunt. So it's a difficulty that we always encounter. The way I felt that if I'm looking at a hemostatic agent, I want to see does it stop the bleeding or not.
And that's my first initial -- the most important purpose of that model. Does it stop the bleeding or not and that should be apart from all the physiological thing that can come into the process of stopping bleeding. If it's a vasoconstriction that causes the bleeding, I've got to take that one out in order to find out whether that product works or not. If the hypertension made it part of the process then I have to prevent that in order to really know whether that product really stops the bleeding or not. But that by itself being the situation then it's, oh, this is not really a physiological model then. But that's how it is. That's how we have no other choices in order to find out whether something works or not, if we're looking at one of the specific aspects of it.
MR. WADE: Comment?
SPEAKER: I think there are several critical elements of any model that can make a huge difference in outcome. So I'll just use the external bleeding model, either femoral trans-section or a hole-punch model. So several questions. You make an injury. Which injuries did you make? How long do you wait until you apply the treatment? How long do you wait till you give fluid? How much fluid should you give? How long should you hold pressure on that injury? Every one of those decisions, every one of those critical decisions affects the outcomes of the model and easily can affect whether one product does substantially better than another.
So we've seen with advance hemostatic dressings, reductions in bleedings when pressure's held for 3 to 5 minutes, which is what is recommended on each of those packages. But if you stuff a gauze in a wound -- and we've published this -- if you stuff a gauze in a wound and don't hold any pressure, it works as well if not better than the advance hemostatic dressings. In the field of battle, that's how it's going to be used. The dressing's going to be shoved in a wound and no one's going to stand around holding pressure for 3 or 5 minutes, because they need to fire back at the enemy.
So the models themselves, which may or may not reflect reality determine which are the better dressings and not the other way around. And I think that's a very critical part of this.
MR. WADE: Okay, we'll take some questions from the floor.
SPEAKER: Thank you, Charlie. Gentlemen another variable --
MR. WADE: Identify yourself, Rick, sorry.
MR. DUTTON: Rick Dutton from Baltimore. As far as replicating the clinical condition, one of the -- my biggest problems with animals models has always been around the effects of anesthesia and that the animal starts anesthetized, often deeply anesthetized, vasodilated and is then injured and bled. Whereas the human trauma patient starts with the injury, the bleed becomes intensely vasoconstricted and then maybe gets some anesthesia later. How do we manage that effect in animal models?
SPEAKER: That's an excellent point.
MR. MYERS: As a veterinarian in the University of Michigan, I don't think that will be -- pass through your Animal Care and Use Committee, and I think that's a big hurdle. That's an excellent point. Different anesthesias affect different surgical planes, different traumas and different injuries. So it is a definite compound there. And that would a tough one to get around with your Animal Care and Use Committee.
SPEAKER: I'd like to make a point on that. One of the things is when we have an animal model, we do have to understand the effects of the constraints that are put on those animal models. And that's one the things why we do have to do the sham controls, et cetera, just to look and see how much the physiology is altered by our manipulations, et cetera. And so I think it's a point that the control and sustaining those controls in the initial phase of animal model development are very important.
Someone else (phonetic).
DR. ALAM: Yeah, a quick comment about the models. These models have got some strong points, for example looking at hemorrhage control or reversal of coagulopathy. But there are some glaring deficiencies and we just need to know that. We convert this polytrauma modeling into not a central liver injury, but peripheral liver injury, splenic injury, switch the femur to rib fracture to make it into a survival model, looking for lay complications. And we've just presented, the data is going to come out in print and the striking thing is you can't get multiple organ failure in these animals and you don't have any delayed thromboembolic complications.
And if you think about any complication that has got an incidence of 4 or 5 percent, let's say arterial thrombosis, you have to do hundreds of animals and survive them to show any statistical difference. So you've got to realize that we should not focus on these delay complication and keep our focus on stopping the hemorrhage early consistent with all the presentations this morning and correcting coagulopathy early and not rely on these models to look for delayed complication, thromboembolic complication, multiple organ failures because these organs will fail you. And the question is whether that's important or not because that was one of Tony's last thing, was that we should modify these variables to look for delayed complications.
MR. KHEIRABADI: I agree with Dr. Hasan Alam, but there was recently a product was -- got approval called WoundStat. Maybe you are familiar with it. Let me just make one thing clear. There are two group of hemostatic agents. Those are systemically infused, injected. Those goes through a clinical trial, as all we know, phase I, phase II, phase III, for safety and efficacy. But then there is a group of hemostatic agent called -- that is used for external injuries. These are just classified as medical devices.
So for them to get approval it don't have to -- they do not need to go to clinical trials. So they're never going to be tested on a patient. So the information that's brought to FDA is essentially a basic toxicity and irritability and sensitivity test on the small animals. And that's how they get pre-market licensing.
Now this product which is called WoundStat was proved to be as good as a previously-approved product called QuikClot granules. And on the surface of it, if we would not have looked at the safety of this product, it looked like one of the most effective hemostatic agents to stop the bleeding. But when you look at a very short survival time afterwards, it turns out that it caused significant damage to endothelial cell, caused local thrombosis and embolism, called thrombosis of the lungs.
So sometimes -- I agree with that that the focus have to be how to stop the bleeding at that moment, but it's very important at least for next 24 hours to find out whether that agent caused any other problem -- any other damage to the tissues and that was never revealed until the animal studies were done.
So for those external hemostatic agent, animal studies is a pivotal role not just for efficacy but really testing their safety of them.
MR. WADE: Tony?
MR. PUSATERI: Well, I guess I was hoping that that post-injury with that polytrauma model would end up with some complications. However, just -- and I don't know -- of course don't know the details of the -- of Dr. Alam's study, but I would really like to talk with you about those.
But I still think that we have to do a better job of hemostatic safety, assessing hemostatic safety. Now the -- to me, I am still thinking that there would still be more relevant exposure to the hemostatic risk because you have to consider that a product may cause the thrombosis even if you're not seeing it that often in the case it doesn't include a hemostatic -- pro-hemostatic systemic product. I -- you know, another alternative and I'm not an expert on thrombosis models, but if you just consider the Folts model that I mentioned, and rabbits in the Wessler model, which is an arterial not a venous thrombosis, that's really exposing these products to an artificial risk and relate it -- it's very difficult to interpret whether a hemostatic product increases an already occurring rate of thrombosis and dissolution of a thrombi, depending on the model.
So it's somewhere in there we need to decide how to use these models to assess hemostatic safety. So I can't say that we have the exact answer on that yet.
MR. WADE: Dr. Moore?
DR. MOORE: Moore, Denver, this has been a wonderful discussion, but it seems like it's incomplete without understanding the basic physiology of animals' coagulation system. As my understanding is from the current literature, the pig is an excellent model to study fibrinolysis, but does not correlate well with the human in terms of coagulation for hemostasis. For example, it has twice the lag to thrombin, half than fibrinogen, twice the numbers of platelets, platelets hyperactivity occurs rampantly in pigs. In my understanding at least from literature is the best model's a sheep.
So my question is is my understanding of literature accurate and secondly then why don't we use a sheep other than the pig in all these studies?
MR. WADE: Comments?
SPEAKER: The ruminant versus the omnivore? Okay, that's an excellent point. As far as the pig, it's a more a -- I think there's a more published history on an established cardiovascular model. The sheep has its problems with size, being able to keep them effectively in most research labs. You may have to contract with the farm. So those are just logistics. As far as it being an animal model, it has been used as a number of thrombosis models. I mean, Dr. Greenfield did a lot of his early Greenfield filter work in sheep, blood pressure volume is about the same. So it is an excellent model, but I think size and logistics which kind of take it out of favor.
As far as the best antigenically and closest to us is probably an upright model, like a non-human primate. But once again like I said earlier they're very expensive. You have a lot of ethical issues that you have to deal with.
SPEAKER: And having worked with both species, one of the things is that most of us have started our careers working with volume expansion product and in the sheep, with the rumen, you get a lot of mobilization from that extravascular water component and it's a very interesting model to play with. Instead of dealing with three compartments, you're dealing with four compartments for fluid flux, et cetera. So it's got it's problems in that area and I think the experience is not been garnered yet for looking at the coagulation profile in the sheep model, where most of us were doing volume resuscitation shied away from it, so.
DR. MATTOX: Ken Matt, Vermont. Most of the discussion has really dealt with producing a hemorrhagic associated trauma and then coming up with an agent to try to clench that hemorrhage process. But what has not been discussed extensively is the use of animal models basically to understand and stratify what's going on in trauma. And I'm wondering in particular with respect to the polytrauma model, what's known about the progression of process with respect to coagulation defects, fibrolytic activation, inflammatory syndrome, cessation and identification and the interaction between those various events, which can be studied in animal models and not conveniently studied when extending it to human studies, by virtue of accidental occurrence. Thank you.
SPEAKER: Ken, that's a great question. What we know certainly is that the combination of femur fracture and fluid infusion results in a fairly substantial coagulopathy. We chose to use normal saline in this model because we wanted to make the animals really sick. So when you give your patients normal saline think about that as well. But what that does is it creates a -- not only is there a lactic acidosis but there is a hyperchloremic acidosis.
Unfortunately we did not -- we were not able to differentiate the coagulopathy that's caused by the femur fracture and soft tissue injury from that that's caused by the fluid resuscitation. We do, however, know that there is a very significant increase in pro-inflammatory mediators, particularly IL6, IL8, TNF-alpha that's very easy to recreate, that follows both again the femur fracture and the resuscitation. So something that we want to look at in the future is we want to look at what degree of the pro-inflammatory response in the coagulopathy is caused by the femur fracture versus as that is caused by resuscitation. We haven't separated those and that's something I think is very important.
The model is uniformly fatal if in the absence of resuscitation, but it is a very good -- it -- we do very much see the recreation of the lethal triad in a very distinct pro-inflammatory state with that model. And we have done a number of animals without resuscitation as control animals and interestingly the animals that do not receive any resuscitation actually got more inflammation than the animals that do receive resuscitation.
MR. SACKNER-BERNSTEIN: Hi. Jonathan Sackner-Bernstein from the Center for Devices at the FDA. I notice some really wonderful science being presented about animal models, but if we're talking about looking forward, I wonder whether we should be debating animal models or perhaps thinking about models in the context of what's going on in technology and science today. There are labs in this country that are performing tissue engineering, experiments that are creating hollow (phonetic) organs, blood vessels from human stem cells. There was recently a project funded and completed by DARPA to create a tissue-engineered immune system that could be used to assess a human response to vaccines and other antigens.
With all that science going on, are we focused a little too narrowly here or is there a place to create tissue-engineered models from human cells that would allow us in a more appropriate fashion because it's looking at human responses, understand that the effects of dilution, of coagulopathies and of some of these agents that are being considered for development?
SPEAKER: I think that we're at the stage right now in our knowledge of coagulopathy-associated with traumas that everyday we realize there's more that we don't know. And there's a -- a lot of people in this room are involved in some very significant efforts to understand the natural history from all aspects, endothelial, enzymatic, cellular, et cetera. So I think that that may be possible in the future to model these things and develop model systems, but I think right now our knowledge is inadequate to establish models that would be that useful.
SPEAKER: I think you've made an excellent point and a very important point. But I do not see this as one or the other. I think the issue of hemostasis resuscitation is a very complex puzzle and I think the stem-cell issue and tissue engineering part that you've -- that you've risen right now is an important part of the puzzle, but I do think that in order to adequately study interventions you need an intact animal-type model or human model.
But I don't think -- I think that tissue engineering, regenerative medicine, a lot of money is going into that, a lot of work, a lot of very important work is coming out of that. So the two are not mutually exclusive, but in fact they're complementary.
SPEAKER: And I agree with that statement. I don't think you can at this point in time get away from the whole-body system. I think it would be an excellent complement and as technology advances, you can marry the two together. To get at some of these issues that would be wonderful.
MR. WADE: And then that leads to one of the questions I have up here and I'd just like to clarify it. It says, "Is it really true that animal models are necessary?" And the point they use the Bickell study in the New England Journal of Medicine on the hypertensive resuscitation study. I would like to clarify that also had an extension animal background behind it before that study was undertaken because he did that in my lab at the time, so I'm very familiar with that.
But that goes back to the question, do you guys feel that animal models are necessary prior to going in to clinic especially when you have community consultation?
DR. SCHREIBER: I think the answer is a resounding, yes. I don't think there's any question about that. There are limit -- there are vast limitations, we've alluded to many of them, in performing human studies. The degree of regulation is quite significant, studies -- many of the trauma studies -- essentially all of the trauma studies that have been done in the United States have been stopped for futility. Their problem is with the design of studies. I think it is very important to start with animal work to give us an idea of what we should study and how we should study it and how to best design models. I think resoundingly, yes, animal models are absolutely necessary.
SPEAKER: I would agree with everything Martin said plus the opportunity to learn from the models. Mechanism is clearly at -- still has to be at the basis of this and you have to confirm mechanism and learn as much about the mechanism, but also learn about the pathophysiology and the process of working with the animal models.
MR. KHEIRABADI: Just to add to this, as I mentioned, the systemic agent go through the clinical trial safety and efficacy, but the hemostatic agent that I use externally and -- for temporary is stopping the bleeding do not see any clinical data. In other words, they -- all they have to do is just a very brief establishment of toxicity test and they can be marketed. So, the animal studies is really the only thing that actually test those efficacy and safety of them before they actually be given to the patient. So, for that group of at least agent those that are used for external bleeding which categorize as hemostatic devices, the animal studies is the only data that we will have in terms of efficacy and safety until they actually put it on somebody.
MR. WADE: Question?
DR. MCPHEE: Yeah. Martin McPhee (phonetic) from STB. I just -- couple of sort of comments on two points, I'll try to be brief. But, you know, one key thing about a model choice and the elements of a model that you need is the mechanism of action of the product that you're trying to evaluate. And we're sort of used to seeing devices and most device type products as Bijan points out, they're topical and what they're going to do is they are going to try and cause the patient to produce a clot to stop the bleeding. If that's the mechanism of action of your product and then you better have a coagulopathic component to your model because otherwise you're going to be let astray by it.
Similarly if you're trying to use a systemic agent and normalize a coagulopathy, you need that in your model. But if your product's mode of action is that it brings everything to the party and provides everything, there is no influence of the clotting system of the patient on it, then you don't need that in order to asses how well it's going to work. So the classic of those are the fibrin sealant products, but you can look at things like cyanoacrylates and so forth and understand that it didn't -- wouldn't matter if the patient is coagulopathic, whether that actually sealed over the -- hid the hole that stopped it working.
And, so I think you have to decide what's your mode of action and that your model has to have what elements here. And the last thing I'll say briefly is you don't have to go all the way forward to tissue engineering. We showed over a decade ago that you can develop very effective X-fugal (phonetic) models because you can take the tissues out of the animals, injure them and then you can measure again how well something is going work. If it's a mode of action that provides all of the components of the physiology, if it's dependent upon the clotting system of the patient those kind of X-fugal models don't work terribly well. But there you do have two different things here to consider depending on what's your product.
MR. WADE: Thank you. Dr. Holcomb?
DR. HOLCOMB: Holcomb from Houston. Dr. Wade, as the moderator, you asked the panel of animal model folks if animal models were important, they all said yes, so be careful with that.
MR. WADE: I'm biased.
DR. HOLCOMB: Not an unbiased answer. And then, you know, comment actually I want to take off Dr. Moore's comment. Animals, their coagulation systems are incredibly different and they're also incredibly different than humans. So saying that you have to have a positive result or an exactly similar result for an animal model before you go into human trials with community consultation is, we need to be a little careful with that. I believe the way Marty does as well, that yes, we should have animal models before doing community consultation. But the animals are different with amongst themselves. They're incredibly different from people having tried to make a coagulopathic animal model for a long time and have done literally thousands of animals, that's a difficult transition to make.
And from a regulatory point of view thinking that you've going to have exactly the same pathophysiology mechanisms transitioning from animals to a phase III community consultation definitive trial is a huge leap that's really not backed up by science, maybe required by regulation, but the science is not there.
MR. WADE: Comment? Tobi (phonetic)?
MS. PARKS: Yeah. Tobi Silvmen Parks (phonetic) Al Consulting (phonetic), formerly FDA. A question to the panel.
If there's no one animal model, perhaps a combination of animal models. I'd like to hear a comment about combinations of models that might be informative for particular situations.
SPEAKER: Well, I'll agree that there's not -- and I wasn't trying to say that there should be in the presentation -- an identical pathway for every product. So, I think that's still going -- that you still remain specific to the product and I could certainly see combinations of models. For example, you may need to go through a certain series of models to prove -- to have proof of mechanism. But I think at some point you have to have a model that is -- that has some component replicating the coagulopathy that you'll be dealing with, for example.
So I think you could get there with a series of animal models. It may be that you would show efficacy in different species with a final safety demonstration in primates, for example, if it's not possible to use pigs, for example, with these human platelet derived products. So I guess the main point is I'm not trying to say personally that there is only one pathway.
I would like to just go back to a different point really briefly and that is I do think that the coagulopathic state is important even taking into consideration the mechanism of action of a product, just back for example, this goes back to Dr. McPhee's comment, I don't think that we knew before we were working with dried fibrin sealant dressings that they would overcome hypothermic coagulopathy adequately to stop bleeding. So I still think the coagulopathy component is important.
MR. WADE: Last question?
MR. MCDONALD: Yeah, Vic McDonald (phonetic), the U.S. Army, Usound (phonetic), that's our advanced development wing.
I think in moving forward with these models there is -- as we develop new blood products we have to keep in mind about the species specificity with the -- and a potential immune response to cellular components is one thing. But oftentimes we're -- what that forces us into is having to, for instance, take out those blood components from that particular species into some sort of a storage system, we're going to create this model and then we're going to give it back after some sort of an alteration.
Keep in mind that pig red cells are not human red cells, pigs platelets are not human platelets. That also goes for mice; that also goes for rats. We have to keep them -- I think we really have to keep that in mind. And oftentimes if we're not -- if we're going to give back some sort of a molecular component in order to try to achieve hemostasis in building these models, we're not just giving them back with an -- with saline or some other crystalloid, we're going to be giving that back with other cellular components, usually red cells -- you know, red cells, plasma and platelets. And if you're going to take those and you're going to apheresis -- do apheresis on a pig and try to store their platelets, they're going to store differently.
I think those are all things that we have to keep in mind as we move forward with this. It doesn't mean that it's impossible to do, it means that it has to be validated before you do it.
MR. WADE: Thank you. I'd like to thank the members of the panel.
DR. VOSTAL: Thank you very much, Dr. Wade, and thank you for the panel members. That brings us to lunch and we will get back together at 1:00 o'clock. Thank you very much.
(Whereupon, a luncheon recess was taken.)
DR. LINDSEY: The afternoon session, so if you would please make your way to your seats. Right, so my name is Kimberly Lindsey. I'm a medical officer in the Division of Hematology for CBER and the workshop co-chair. We're going to get started with our afternoon session. And our first speaker is Dr. John Scott, who's a mathematical statistician from CBER, and he will be talking about statistical considerations for clinical trial endpoints. After his talk, we will have a 10-minute discussion period.
CLINICAL TRIAL STATISTICS
MR. SCOTT: Thanks very much. It's a great privilege to be able to address you today. I don't know what sadist scheduled this statistical talk for after lunch. I -- please accept my apologies in advance.
What I'm going to be talking about today are just very general statistical considerations for clinical trial endpoints. Hopefully this will set the stage for the more specific discussions of severe bleeding endpoints that will come after this.
And although this is very general, I've tried to focus on some of the issues that have come up in discussions we've had at CBER with sponsors who want to pursue bleeding indications.
So as an outline of what I'm going to be talking about, first just defining our terms, what we mean by endpoints, what we mean by primary and secondary endpoints. Next, I'll be talking about situations in which there may not be one endpoint that's adequate for a clinical trial including the co-primary endpoint situation. After that, I'm going to talk a little bit about composite endpoints and rating scales. And then finally, a slide or two on additional study design considerations and conclusions. I also want to note, I'm sure many of you are very familiar with most or all of this, so I hope you'll bear with me as we just try to sort of set some context.
So by the endpoints of a clinical trial, I mean the pre-specified measurements or observations of subjects which are used to determine the outcome of the trial. And I would highlight pre-specification there, endpoints really have to be clearly defined in prospective protocols and where appropriate in statistical analysis plans. We usually divide endpoints into safety and efficacy endpoints, although there can be considerable overlap. And that's particularly true in serious and life-threatening conditions like severe bleeding and trauma, where for instance, lack of efficacy is a safety issue.
We also divide efficacy endpoints into primary, secondary and exploratory or tertiary, and sometimes safety endpoints are divided into primary and secondary as well. So, the primary efficacy endpoint of a phase III clinical trial forms the principal basis for determining the effectiveness of an intervention for licensing purposes. It's the principal basis; it's not the sole basis. The term of art we use at FDA is that we always review the totality of the evidence. So we'll look at additional efficacy variables and we also always weigh efficacy against safety.
We use secondary efficacy endpoints to first of all provide supportive evidence of efficacy to bolster whatever we saw in the primary efficacy endpoint, and also to justify additional labeling claims after success has been shown on a primary efficacy endpoint. And exploratory endpoints are just used to generate hypothesis for future studies.
So when you're talking about choosing a primary endpoint for a clinical trial, I like this quote from the ICH-E9 guidance, which states that the primary efficacy endpoint should be the variable capable of providing the most clinically relevant and convincing evidence directly related to the primary objective of the trial. So, what I would underline in there is first of all clinical relevance. The endpoints have to be clinically relevant; they have to mean something to patients and to practitioners. But then within the class of clinical relevant endpoints, you would want to be looking for the variable that's most capable that has the best chance of showing the advantage of the product you're testing.
So the key criteria for evaluating a prospective primary endpoint would be first of all that it be clinically relevant and interpretable, that it be easily measurable, that it be sensitive to change or sensitive to intervention meaning that your treatment actually has a chance of affecting this endpoint. And I think what we heard a lot this morning is that there's a lot of skepticism that 20-day -- or I'm sorry -- 30-day mortality would be sensitive to change for a lot of interventions.
We would want the endpoint to be statistically reliable, meaning measured with as little noise as possible and whenever possible, we want primary endpoints to be objective.
It's always ideal when we can have a single discrete primary efficacy endpoint that adequately captures a clinical entity that's ideal for interpretability and for statistical simplicity. But there are cases where that can -- that ideal can be met. First of all, in many symptomatic diseases and syndromes, there may not be one endpoint that really adequately captures all the relevant clinical dimensions of a disease or condition. And even in situations where one clear, clinically relevant endpoint may exist, it may not be possible to adequately power a study to use that as the primary efficacy endpoint. And I think severe bleeding in some senses can fall into both of those pitfalls.
So when there is no one adequate primary endpoint, one approach to the problem is just to look at multiple primary endpoints, just to test two or more maybe a whole panel of endpoints and declare that your treatment is efficacious if it shows a statistically significant advantage on any one of the endpoints you tested.
The problem with that is that if we perform each test at the 0.05 traditional significance level, there is a greatly inflated risk of a false positive result for the trial of accidentally approving a drug that doesn't actually work; we also call that Type I error. Now, we can adjust the significance levels to accommodate the inflation in Type I error rate, the most widely used approach is the Bonferroni adjustment, where if for instance you were testing two primary endpoints, you could divide the significance level 0.05 by 2, test each of them at 0.025, and that would control the overall false positive rate of the trial. But the problem with that is that when you adjust the significance levels downward, you increase the false negative rate or the Type II error rate of the trial.
You can sometimes compensate for the increase in false negative for the lack of power by increasing the sample size. But there are often practical limits on just how many patients can be recruited.
Another somewhat different approach is co-primary endpoints where again we test a number of different endpoints, but here we declare that the treatment is only efficacious only if it shows a statistically significant advantage on all of the endpoints we tested. The advantage of this approach is that it doesn't inflate the Type I error rate. You can perform every test at the 0.05 significance level and the false positive rate is no more then 5 percent, but there is still a loss of power. Every endpoint you test, you have a chance of getting a false negative and one false negative means a failed trial.
In some cases, we can form a new composite endpoint by smashing together in some way in one variable, multiple distinct clinical outcomes. And there are several ways of doing that. We can form composite event rates, composite time to event endpoints. In some indications, there have been a sort of consensus development process that's led to more complicated constructs specifically tailored to those diseases, and there also rating skills.
So some types and examples of the composite endpoints I just mentioned -- composite event rate means the proportion of patients who experience any one of a number of events within some fixed timeframe. So in the Center for Biologics, one endpoint we've looked at in critical limb ischemia is 12-month rate of death or major amputation. So if a patient experiences either of those events, they're considered a treatment failure in the analysis.
A composite time-to-event endpoint -- instead of just measuring whether or not any one of several events happens, it measures the time from randomization to the first of several events happening. So, cardiovascular trials widely use some variant of time to death, stroke, or MI as an endpoint. And as I mentioned, in some indications there's been a validation and a consensus development process for more complicated constructs. In rheumatoid arthritis, the ACR-20 is a very widely used efficacy endpoint, and I just mention it here as an example of a well-validated scale. It's defined -- a patient is defined as a success on the ACR-20 if they show at least a 20% improvement in the number of swollen or tender joints, plus at least a 20% improvement on at least three or five additional parameters, being patient global assessment, physician global assessment, pain, disability or an acute phase reactant.
Rating scales can summarize the presence or absence and often the severity of any number of signs and symptoms into a single total score. That score is often just a sum or a average of the individual symptom scores but in some cases there are more complicated scoring formulas. Some familiar examples might be the MOD and SOFA scores or the SF-36 which measures general, physical, or mental well being.
So, when you're talking about constructing a composite endpoint when you're thinking about constructing a composite endpoint, the first thing to consider is which components are going to go into that composite, and the components really should satisfy the same criteria as good primary efficacy endpoints. They should be clinically relevant, measurable, reliable, sensitive to change and wherever possible objective. But, in addition to standing on their own in that way, seen together the components shouldn't be wildly different in importance. It's very difficult to interpret a significant result on a composite endpoint when you have components in that composite that are very different in importance. The components also shouldn't show large heterogeneity in treatment effect size. To some degree you won't necessarily know this until you've performed your clinical trials but you want to have reason to believe that your intervention is going to move all of the components in the same direction in approximately the same magnitude. And composite endpoints should be carefully evaluated against these criteria prior to going into phase III trials.
So some advantages and disadvantages of composite endpoints, the biggest advantage the reason sponsors often want to talk to us about composite endpoints is that relative to single event endpoints they can usually increase rather than decrease the power and that increase in power comes without inflating Type I error. Basically, if you're power is limited by the number of events you see by looking at different kinds of events you can increase the total number of events that will be observed. But this advantage in power, I should note, only holds if your treatment actually works for all of the components in the composite. You don't increase power by adding in components for which your treatment has little or no effect.
In addition to that composite, endpoints can provide meaningful outcome measures in situations where no one endpoint adequately captures the effect of treatment. The biggest disadvantage with composite endpoints is interpretability, once you've won on a composite endpoint, once you've shown a statistically significant advantage, the question is what does that actually mean for patients and depending on what components are in that endpoint that can be a very thorny question to answer.
For instance, you could see success on a composite that's driven by one or two components which is masking a lack of a factor even a deleterious effect on other components. And those other components could include mortality. There have been cases where we've seen a win on a composite but subsequent investigation has shown at least a trend toward a mortality disadvantage for a product.
It seems like you could just test your composite endpoint and then test each of the components individually to sort out the interpretability, the problem with that is that it brings back exactly the difficulties we were trying to solve in the first place. It brings back Type I error rate inflation and low power for each of those individual comparisons. In addition composite endpoints, especially rating skills, require validation.
So a word or two about validation, rating skills in particular require validation before being used as efficacy endpoints in phase III trials. So when you come in with your phase III trial protocol it's too late for -- to be proposing a new rating scale, that should have happened before. Validation includes establishing statistical reliability which essentially means that you're measuring something -- you're measuring something with a sufficient absence of noise that you get a reliable number. Validity, which means you're measuring what you actually want to measure, and clinical utility meaning that again your scale is able to be moved by your intervention, you can actually affect a change in this rating scale score.
I wanted to mention that FDA has a fairly recent guidance document on patient reported outcome measures not all rating scales are patient reported outcomes but I think there is some very useful advice that might be more generally applicable there and I have a link to that in my final slide.
Finally, one additional topic the sponsors often want to talk to us about are adaptive clinical trials, and adaptive clinical trial designs can provide advantages and efficiency over traditional designs. Those advantages can -- would usually take the form of either shorter trials and duration or a fewer number of subjects required for the trial. Adaptation can take many forms and that can include dynamic changes in eligibility, and randomization, and sample size, and in treatment regiment, but the sort of take-home point that every FDA speaker on adaptive design seems to like to make is that the adaptations have to be pre-specified, you have to say in your protocol exactly when and how your clinical trial will adapt.
We often have sponsors who come in midway through a trial and they want to make some change to enrollment criteria or sample size and if it's not pre-specified that's a very difficult decision for us to agree with. And FDA has a draft guidance on adaptive design which is open for public comment right now and I have a link to that in my final slide.
So some overall conclusions; I would urge everyone working on these products to start thinking about their phase III endpoints at the very beginning of the product development process, even before their first in human trials if possible. If you're thinking about the endpoints early then you can use your early phase I and phase II trials to learn as much as you can about those efficacy endpoints and to help plan a successful phase III trial. And I would also encourage everyone to consult with their FDA review team early and often about their endpoints including in pre-IND and IDE meetings. And that's it. Thank you. These are the links I mentioned, that's it.
DR. LINDSEY: We actually have some time for questions, if you have any questions, would you please make your way to the microphones.
CLINICAL TRIAL STATISTICS- QUESTIONS AND DISCUSSION
MR. GULLS: Yes, Steve Gulls (phonetic) from Chicago. I enjoyed your talk, can I press you on composite endpoint, maybe it's me but I always struggle, I like the example you gave of death and amputation, it's the same as MI and death, or stroke and death. So it is theoretically possible to have fewer total events in one arm but more deaths, how do you deal with that? I mean that's always difficult for me to --
MR. SCOTT: Yeah, that's an exceptionally difficult question I wish I could give you a very direct answer on that. I think major amputation is in that setting is considered a sufficiently bad clinical outcome and I think also sufficiently predictive of rate of mortality that considering those two together isn't crazy. But you definitely don't want to consider mortality with something that's clearly far less important than mortality.
And after the win on the composite, there are difficulties with drilling down in the components. But we will drill down into the components. And if there's reason to believe that there's a mortality disadvantage and that that disadvantage is real, that would be cause for further investigation, I think.
MR. GULLS: Okay. Thanks.
SPEAKER: Scott, you brought up the 28-day versus 24-hour mortality in these hemorrhagic shock studies and the changing epidemiology of injury care over the last 30 years. So can you elaborate on that a bit more?
MR. SCOTT: I cannot elaborate on that, I'm sorry. I -- unfortunately, I probably know less about bleeding then anyone else in this room. So I'll leave that to the further speakers. All I was saying was that there appears to be a sense that there's more noise in the 28-day mortality measurement because there are many more causes of death that are coming into play, so you're measuring it with less precision.
SPEAKER: Okay, well I was -- I was going to ask you what composite outcomes you viewed as meaningful in the context of this discussion, but given what you just said, I'll skip that.
MR. SCOTT: I'd rather punt on that.
SPEAKER: I'll let someone else --
MR. SCOTT: Sure, thank you.
SPEAKER: -- maybe comment on that. But I would like to ask you what your thoughts about using global tests for multiple outcomes because when you look at the disadvantages of composite outcomes, they really do address several of those disadvantages. And you can still get drilled down, and if you require statistical significance before you drill down, you have weak protection of alpha. So I'm not sure what the advantage of a composite test would be over a global test.
MR. SCOTT: That's an interesting question. I actually -- I share your feeling; I think global tests are potentially valuable. But I don't think there's a lot of precedent at FDA for seeing global tests in pivotal phase III trials.
SPEAKER: They did accept the global test in the TPA trial and it is being used in a large Parkinson's trial right now, so --
MR. SCOTT: Yeah, so I think if the statistical properties of the endpoints are appropriate, it's certainly worth proposing and discussing.
SPEAKER: Just a question about statistical considerations. You know, death and amputation are great because we know they are -- when it happens and when it's done, something like the absence of bleeding is really a subjective endpoint. I mean, I -- for -- as a clinical observer of bleeding, it's a subjective endpoint. And we want to have a way to quantify something that really is inherently subjective. Do you have any statistical insight into that?
MR. SCOTT: That's a good question. I don't have statistical insight into that specific question, but what we -- in those situations, what we try to hope for is that the definition of the cessation of bleeding would be protocolized and as specific as possible in the protocol. And we can't force all clinicians to feel exactly the same way in exactly the same situation. But in the best of all possible worlds, what differences -- what deviations from the protocolized definition would exist would be balanced out between the treatment arms. That would be the best we could hope for.
DR. HERR: I'm Dan Herr (phonetic) from Maryland. We certainly all heard this morning that certain live trauma surgeons and intensivists who don't like the 30-day mortality outcome mark. And I think it's probably worthless in massive transfusion trials. It's an opinion. But I think the 24 or the 48 might be. But what about a composite between, and I was just thinking about this, and I don't know if it's a question or a statement, but what about a composite between -- being an intensivist, the 24 death and then some type of neurological outcome assessment, like we did with hypothermic trials. That would -- would that be a reasonable composite for a massive transfusion trial?
MR. SCOTT: Without answering that, the question very specifically as to whether that specific endpoint would be an appropriate composite, I think something along those lines is certainly possible. Some combination of mortality and major morbidity is something we would definitely be interested in considering. That's --
SPEAKER: If I could just make more -- one more comment about that. Many of the neurologic scales would automatically be coded as worst if a person died. So you really wouldn't need a separate variable for death because, you know, unless you were doing a time-to-event type of thing and combining that with the neurologic outcomes it would be taken account -- into account. For example, if your stroke scale -- if you're talking about stroke it would have to be 40 if you were dead because everything would be gone.
MR. SCOTT: Although that -- there's always an issue with rating scales and cases --
MR. SCOTT: -- where the patient dies. I'm not sure about setting a complicated modeling framework where we're simultaneously modeling an event and a (cross talk) outcome or something --
SPEAKER: Yeah, I mean there's --
MR. SCOTT: -- but it -- the scale would max out in the case of death, but that wouldn't necessarily capture the importance we attach to death in the analysis. SPEAKER: I mean, and you could still put death in there. MR. SCOTT: Right.
SPEAKER: Well, let me open a can of worms. While you talked about the endpoints, you didn't talk about the analysis; do you want to talk about non-inferiority?
MR. SCOTT: So --
SPEAKER: Because I think this is an important topic for trauma as well.
MR. SCOTT: Absolutely.
SPEAKER: And I know it's a very large topic, but if you want to close with a few comments --
MR. SCOTT: Sure. I'm happy to make a few comments about non-inferiority in general. Just for those who may not know, the purpose of a non-inferiority trial instead of showing that a novel intervention shows us a statistically significant advantage over a control, you try to show that it's -- with certainty that it's not too much worse than a control. And the big issues that come up with non-inferiority trials, I think, are non-statistical. They're first of all choosing the comparator and then especially choosing the margin by which you'll accept inferiority. And those are very difficult discussions, and I can put in a plug for a draft guidance FDA has on non-inferiority as well, right now. But I think I'm -- what? No, are there anymore questions?
DR. LINDSEY: Okay. We'll now move to session three, which is the clinical evaluation of products and we'll first have the talk on "endpoints for clinical evaluation of products for bleeding interventions; local agents," to be given by Dr. Jeffery Lawson, who's the associate professor of surgery and associate professor of pathology at Duke University.
CLINICAL EVALUATION OF PRODUCTS
ENDPOINTS FOR CLINICAL EVALUATION OF PRODUCTS FOR BLEEDING
DR. LAWSON: Thank you for that introduction. We get to move away from something elegant and complex like statistics and talk about relatively low-brow things like topical hemostasis. So I have a disclosure upfront to make. One is that I started life as a coagulation enzymologist and my thinking was contaminated by Dr. Ken Mann, who it's nice to see in the audience, and then somehow mutated into a vascular surgeon.
And the lens that I have today would be sort of reflective of those two worldviews. The one thing that is different about a vascular surgeon and a trauma surgeon is trauma surgeons start with bleeding as their phenotype and then try to correct it. Vascular surgeons actually usually start with thrombosis as their phenotype and atherosclerosis, and use anticoagulation as a therapy to get your way out of trouble and then figure how to correct that. So philosophically, there's a different set of tools and a different way we approach it. And that's illustrated by some of my slides also.
As far as formal disclosures, I have a lot of relationships both scientifically and with industry. There are four that are relevant to this discussion; that is with Baxter, Johnson & Johnson, ZymoGenetics and NovoNordisk where I have served as a consultant and advisor over the last 4, 5 years. And my laboratory is funded by the American Heart and the National Institutes of Health.
Thought I'd start with a sort of another way to approach the problem or illustrate the issues around both bleeding and thrombosis, and I thought I would use our own institution, Duke University, as a sort of a way to reflect on this. So a few years ago, I got a little frustrated by the fact that none of my colleagues thought bleeding or clotting was particularly relevant in surgery. And they thought it just was -- existed, it's just what you have to deal with every day. So I pulled our D&C or MNM (phonetic) list for over a 2-year period and collected data on about 350 patients, and actually added up how many of those complications that were presented -- and it's a bias or a single institution and some people don't report their complications. But of those that do, I was able to -- we pretty quickly tally up at least 55 percent of the complications that were presented in our institution for a surgical D&C, were related to either bleeding or thrombosis in the post-operative period.
And if you try to drill into surgery like many us have alluded to today and try to figure out what really happened about half the time we don't really know what happened. So, if you try to adjudicate the data you probably throw out about a hundred cases because it went so bad you couldn't decide if bleeding or thrombosis was really the culprit or if it was some other mal-aligned problem. So when you adjudicate the data and you drop about a hundred cases that you can't figure out what happened, you come up with a bleeding and thrombosis rate of our D&C are complication that idles at around 80 percent of the cases.
And if you look at how many of these are bleeding complications and how many are thrombosis complications, it turns out about 50 percent of the complications presented were related to bleeding, and about 30 percent were related to thrombosis -- pretty significant on both ends of the scale. And if you drill into who of those patients that were presented actually died, there were 67 deaths in this single-institution series of which 27 percent of them -- 27 percent of the deaths related to bleeding, 31 percent related to thrombosis. So while bleeding is probably more common at least in our institution with respect to general surgery, trauma, vascular and transplant, thrombosis is probably just as lethal if not more.
So, what are the issues around studying hemostasis in surgery and I have a couple of reflections. One is, I think we still do a poor job of identifying when we're thinking about studies or we're thinking about outcomes, who's likely to bleed or clot too much. We tend to treat everybody as if they were the same and in fact we're governed by probably hundreds of polymorphisms that change the set point for which one would be even under the best of conditions and when you put someone in a stressed environment like surgery how they react to that stress is probably profoundly different and varies greatly.
We do a marginal job of understanding how to optimize or when is there optimal physiology, it seems rather trivial to just talk about pH, ionic strength, calcium levels when global physiology of the patient is critical. To enroll a patient in a clinical trial around bleeding or a topical agent if you haven't really understood or assessed their physiology, it's a great question.
What I'm going to drill in a little bit on is these topical agents and the truth to be told there's almost very -- there's very little objective science around the effectiveness of these topical agents if they've been studied. We have very little insight into what biologic agents are useful, when they're effective and how much to use as illustrated by the trials and tribulations of studying Recombinant Factor VIIa. And then I would argue that we still are struggling with even how to design a clinical trial about surgical bleeding because we can't really define the front end when do you start a bleeding trial, and I think sometimes we struggle with defining the backend which is what are the outcomes, is it death and amputation or is it the absence of bleeding.
So with that why is this such a hard job, well, we have to take very complicated patients that sometimes get dealt to us. And we have to take them through sometimes heroic operations and while that's all glamorous and dramatic we have to do this in the context of very complicated cell biology, and I would think relatively elegant biochemistry. And probably the area where we understand things the best is in the biochemical mechanisms of how at least clotting proteins and platelets interact with each other, and that's still a long away from true knowledge. And we have to do all of this in a very dysfunctional environment. And, so to struggle with a complexity of the patients, the environment, and the biology it's not surprising that this area is one that's challenged to understand and regulate.
So I would say that most of the problems in both patient outcomes and study design lie at this interface between biology, which again, we have limited insight to. Clinical skill, its one variable that nobody likes to talk about but there are clinical assets in the operating room some of which are better then others, and it's a variable if you're going have a study. Medical therapy, did my patient just walk in on plavix because they have a drug eluting stent in their coronary artery, I don't know. Are they going to be in a new direct thrombin inhibitor, and then how sick our patients are, is it a 24-year old or 25-year old guy that just got hit with a missile in a battlefield or is it a 65-year old guy who just crashed his truck and is going to have a heart attack on the table. These are all complex variables that lead to this being a very difficult area to study.
Sometimes we like to trivialize this issue into something as simple as a teeter-totter and people and myself included will talk about the issues between bleeding to death and clotting to death and I would say if you're a trauma surgeon, your issues around bleeding to death are much more profound. Maybe for me, a ruptured aneurysm is a vascular surgeon, but the majority of my patients actually die of clotting to death -- thrombosis, myocardial infarction, pulmonary embolus, stroke. So I think that hemostasis in surgery is more complicated than the unidirectional thought about stopping bleeding. It's about controlling bleeding. It's about controlling clotting. It's about timing -- and I have a couple of slides to allude to that. And it's about balance. Because if you either bleed too much or clot too much, you both -- either way you end up dead.
So I think a more appropriate structural way to think about this is not as teeter-totter or unidirectional process, but one for which the center is given credit for being the safe place we live. As I have outlined, there is normal hemostasis and there are a number of systems which will push and pull on this. And if you're designing a trial that's going to test a procoagulant molecule, say like a Recombinant Factor VIIa, then you would expect the effect to manipulate this part of the pro-coagulant biology.
It's a very different outcome than one might be -- what then one might see within antifibrinolytic for instance. And so it's very important to think structurally about how you would push and pull on this system, not just what direction it ends up in.
I think one other thing that's important is to think about how this system changes over time and particularly with the stress of injury or trauma. When we take our patients we like to think of them initially in this safe little place of normal with a boundary of thrombosis and a boundary of hemorrhage. And they tend to -- or all of us probably by virtue of being alive, live in this corridor. As we take people through operations, we put them through a period of stress that probably does engender some low level coagulopathy. The magnitude of that coagulopathy is probably due to how good your anesthesia is, how big your operation is, and how good your surgery is.
Post operatively there is an inflammatory phase that probably leads to some or engenders one at risk for thrombosis and then we recover by the virtue of healing. What is the most common complication that I see in my patients, hemostatic -- hemostasis device or not. It is to dip into this period where I will anticoagulate them and I will drop them as low as I can, but post-operative thrombosis in the vascular path. That's very different than this model which is one might see as a trauma surgeon or one is fearful of which is this area where you dip into hemorrhage you might get yourself out of trouble because what you're trying to do is avoid this. And I would say if you're designing a trial you'd want to test the efficacy of an agent as a hemostat topically or systemically, you want to find out when the patients are in this environment and when you're going to apply your therapy. Because if you give them Recombinant VIIa and they're already up here you're going to have no effect. And if you give them NovoSeven and they're already down here, you're going to have no effect. So it's the timing and the direction they're headed that are important.
Well, so just a couple of final comments about operative management and the issues that are important when you're designing both hemostasis trials and looking at outcomes. One is, there is an opportunity for good surgery and good physiology which is staying out of trouble in the first place. There is this window for topical hemostatic agents and I'm going to address those or at least show you some samples of what these agents are. I think many are familiar with them but I have kind of a list and an organizational structure to that. And then there is this topic of systemic biologics and how they potentially even marry into the topical agents as I think an interesting discussion and one that needs to be vetted at some point.
So what are the -- my assignment was to discuss or talk about topical hemostasis in the context of surgical bleeding and then what are some of the endpoints.
Well, what are the hemostatic devices that we have available? We have an array of them. They come in the simple forms of mechanical tools, absorbable products, thrombins, goops, pastes, putties and then systemic biologics and I'm going to review at least the first three.
So as far as mechanical devices, they've been around as long as people have been bleeding and I would say to date if you're pretty rigorous about this area. It's hard to find a clinical trial that demonstrates any topical hemostatic devices better than a finger in the right place and a good piece of gauze. And that's something we all have to struggle with because the finger in the right place for the right amount of time will get you out of trouble more often than not.
Cautery laser, radiofrequency energy, argon beams, clips, sutures, CUSAs, harmonic scalpels are all mechanical devices that have been developed to be used in the operating room and they're used sort of a random array of availability in the operating room. Sort of depends sometimes what your hospital administrator let's you buy more than what has truly been demonstrated to be effective.
It's a very interesting technology. I only highlight this not because I support it in any way other than I think intellectually it shows a little creativity in this area, as far as just a mechanical cautery. This is a device called the TissueLink, and in fact what it uses is it uses radiofrequency energy as opposed to conventional heat associated with electrocautery.
Why do I think this is interesting is it actually invokes a mechanism that causes small capillaries and blood vessels to shrink by the radiofrequency energy, collapsing the collagen structures as opposed to just simply burning and charring tissue. And one might envision if you were doing a clinical trial on this topical device that it might be relevant on the degree of tissue injury surrounding the device after surgical use.
It's very interesting if you could cut through a liver like this and this is just a photograph of doing a liver dissection with this type of radiofrequency energy. And I'm not sure how to quantify that that's effective, but I know it's effective when you see it. As far as other devices, absorbable hemostatic devices, now we're into the context of things that we often leave behind and there is a couple of relevant points here. Something as simple as an oxidized cellulose, it's made of cellulose, what we call -- commonly called things like Surgicel. It's interesting, it has a very low pH. If anybody has ever seen Surgicel turn black, you might wonder why is it turning black? Well, it's lysing the red cells and oxidizing the hemoglobin. It's relevant because that's a pro-coagulant function of the cellulose. You might also ask the question, I wonder what that does to the (inaudible) tissue, the blood vessel I just put on. It's probably relatively toxic. It's never been studied.
Gelatin sponges, great carrier for things like thrombin. Neutral pH -- there is some suspicion sometimes that they can be left behind and be a nidus for abscess formation. It's a very important drug, but gelatin sponges, and then we have our collagens which are very good platelet activators; things like Avitene, a very potent activator in general of platelets. Never really been studied in the context of all our new antiplatelet therapies. But these are biodegradable matrices, they absorb blood, usually will activate platelets, and through that mechanism will induce coagulation and they're all thought to be acceptable to leave in the body or leave behind in an area that's packed that you don't have to go take out.
We're one step further up the line into what I call now biologic hemostatic devices. It's interesting. These are approved as devices. I still struggle with the concept that thrombin is probably one of the most potent biologic enzymes that I know of and it gets approved as a device. But that being said, bovine thrombin has been around for over about 40 years. Initially started as a relatively semi pure preparation of cow thrombin and due to concerns raised about antigen sensitivity and immunologic concerns around bovine thrombin, it has now been pushed through a series of purifications. However, the fundamental issues about how it sensitizes the immune system has never really been vetted.
FloSeal is an interesting component. That was originally a bovine thrombin-derived mixture that was sort of made into a slurry or a paste with a gelatin sponge. It's a very easy to use. Originally, it was a bovine thrombin derived product that has now been converted to human thrombin based on the immunologic concerns noted above. And CoSeal, I list here, not because CoSeal is a hemostatic device -- and it is a hemostatic device, but it is not biologic in the sense it has a thrombin component. But it is an interesting glue. It's a polyethylene glycol mix. It sort of serves as almost a cork. It doesn't appear to have a strong biologic effect doesn't it? It has a mild adhesive effect. There is an interesting molecule itself unclear how to study that in further detail.
Recently, we've had to look more carefully at recombinant thrombin in this case and plasma-derived thrombins where there has been comparative studies between the efficacy or time to hemostasis between a bovine thrombin and a human thrombin. These plasma and recombinant thrombins appear to have activity like other thrombin -- like the bovine thrombin that -- it was their predecessor. To date they've appeared to have a reasonable immunologic profile and they still have to address the additional risk of recombinant technologies and/or plasma isolation technologies with respect to viral exposure and stipulative things like salines.
There is another level of complexity. Now where you have the tissue sealants, in this case the fibrin sealant that I call Tisseel, Crosseal, Evaseal. It's usually a human thrombin and a fibrinogin mix, usually contains some antifibrinolytic carrier in there that spiked in either a bovine, aprotinin, or transmenic acid. Usually it's resorbed within 14 days, or 7 to 14 days and some would argue that as far as its biologic effect it's a better sealant than a hemostat and is used sometimes to seal ducts and other tissues as opposed to directly address bleeding.
Next, the BioGlues that's a bovine albumin and gluteraldehyde mixture. It's very good at tanning tissues by the nature of the gluteraldehyde that's in it. But if you tan tissues and you look at secondary effects around those tissues like nerves they can be associated with nerve injury and one must ask if you're going to test a device like this how are you going to measure a nerve injury when your endpoint is hemostasis. It's easy to use and has been very effective for things like aortic dissection, but others have raised concerns about systemic emboli. It's also a relevant question when you look at something like a fibrin sealant is -- does the basic biologic polymer or the structure matter?
So just two photomicrographs of the physiologic polymer of a fibrin or a fibrin clot compared to one that's made of Tisseel and the only thing that I could think of is relevant is how the Tisseel as a device would absorb or dissolve and is it modified by anything else I might be using in the surgical field. Never really been studied but -- and then there are those that now will add things to the Tisseel or the fibrin sealant glues. Things like antibiotics or other medicines and there is little insight into how that affects the structure of these kinds of polymers.
There is also an array of devices that these types of glues and polymers are passed through for almost every surgical environment one could envision now. And is there a rational way that these devices when they get coupled to the biologic -- is there a rational way that they need to be evaluated.
Recently, we've had some experience with the fibrin bandage. Fibrin bandage appears to be very effective for local wound bleeding. It's a pretreated matrix with fibrinogen and thrombin. It's applied directly to the site of bleeding and the blood itself activates the procoagulant matrix. It's left behind. This device has suffered from the lack of definition of what is severe bleeding in bringing it into further clinical utility.
Recently, there has been the approval of the cynoacrolites. They have -- they were around in the '60s and '70s and then went away for concerns of causing side effect to vessels like malignancy. But recently they've come back through clinical trials and have been relatively effective at direct anastomotic hemostasis. They appear to have no immunologic effect, no obvious biologic effect to date, but they have not been measured in vivo for some length of time and the question is how long would you have to wait before you find out if your blood vessel turned into some kind of sarcoma. That being said they've recently been approved by the FDA and will be -- begin marketing and availability in the very near future.
As far as novel agents, the sort of things that we're going to have to evaluate in the coming years, things like procoagulant snake venoms, other procoagulant chemicals, new matrix polymers, thermal reactants, and extracted elements from blood itself that may be harvested in the operating room, mixed up in some new way and then used in the wound as a topical hemostat. And I think that's the challenge going forward as how are these things effectively evaluated to demonstrate the utility and also evaluate their safety.
So as far as clinical endpoints this is the shortlist of the talking points meant to be more rhetorical, just as I am closing. If you're looking at clinical endpoints for local agents, and I've just listed a number of them, I think the first things you have to consider is what type of bleeding are you looking at. Is it arterial? Is it something that's aortic hemorrhage? Is it venous? Is it a saphenous vein injury? Is it an iliac vein injury in the pelvis? Is it a solid organ like a liver or spleen? Because I would argue that the biologic response of the device would be very different in those three different surgical environments.
Type of surgical injury with respect to trial design as well as efficacies. Is it planned surgery? You can predict when you're going to do it. Is it a random traumatic event? Is it anastomotic? Anastomotic, is it going to be bleeding through your PTFE or your Dacron suture line. And then what -- how will your anastomotic biology be affected down the road with respect to things like vein graft proliferation or PTFE effects of the outflow tract.
Patient risk factors things like coagulopathy -- do you have to have somebody with a coagulopathy to enter, do you want to exclude them if they have coagulopathy? Temperature, systemic medications, all very relevant to the efficacy of any trial design for these kinds of products.
With respect to the rate of bleeding, do you have to quantify the rate of bleeding before you actually have somebody in a trial? Do you have to consent and enroll them and then find that if they're going to bleed a lot then you get to use the device in the OR? Is it audible bleeding? Those of us that are surgeons in the room know what that means. Is it just significant? Is it constant? Is it a dribble? Or is it none?
Evaluation of success -- what do you mean, like, you actually were effective in the operating room? Is it time to homeostasis? Is that subjective when I stand up and I say we're not bleeding anymore? Complete homeostasis -- and how long do you have to have complete homeostasis for?
Blood loss -- almost impossible to quantify with any meaningful accuracy in the operating room? And if you're going to do something where you sample periodically and you're going to peek, you're going to look under your little math (phonetic) of stuff, does that affect the reaction in and of itself?
Finally, safety -- safety endpoints for the patient and/or the user. When I say the user, the operating room is a very complex environment, certain devices like lasers or certain biologics have some potential toxicity to the users. And for the patient, not just stopping bleeding today.
But if I stop the bleeding today, put thrombin on their vein graft, induced prior receptor activation of the smooth muscle cells of the vein graft and induced vein graft proliferation and failure. I may have succeeded in stopping bleeding today, but had their vein graft clot off in 6 months. Those things need to be evaluated.
With respect to biocompatibility, do these things actually biodegrade if you leave them behind? Do they form an abscess? Are there issues about immunology that are relevant? And if you have an immunologic reaction to a type of protein or cell or tissue, can you re-expose the patients to it? If you're going to re-expose, how do you document their first exposure?
Is the re-exposure going to be an anaphylactic-type response which we've seen with certain things like bovine aprotinins? Or is it going to be some bizarre coagulopathy like it's been suggested with things like bovine thrombin?
And then other factors. There is going to be issues or tissue or trauma to surrounding tissues as might be suspected for gluteraldehyde affecting local nerves. Is there risk for systemic embolization? And could the device or drug itself induce coagulopathy, either hypo or hyper coagulopathy if the extravascular device actually leapt into the circulation?
With that, I'd like to thank you for your attention. And I think we're supposed to have a panel discussion for questions.
PANEL DISCUSSION: ENDPOINTS FOR CLINICAL EVALUATION
OF PRODUCTS FOR BLEEDING INTERVENTIONS-LOCAL AGENTS
SPEAKER: So panelists for this session, please make your way to the stage. And the moderator for this session is going to be Dr. Grant Bochicchio, professor of surgery in University of Maryland School of Medicine.
DR. BOCHICCHIO: Okay. Before we bring -- start this session, we need to introduce our panel members. We have Hasan Alam from Mass. General. We have Major Andrew Cap from the U.S. Army Institute Surgical Research at San Antonio. We have Mitch Cohen from the University of California in San Francisco. And we have Lieutenant Colonel Jeremy Perkins from the Walter Reed Institute of Research in D.C. here.
I think it's important to understand at least by sitting in this session in the early morning that we think about -- talk about topical hemostats in the way they're used and how we're going to plan clinical trials. It is actually important to go back to the animal work that was presented earlier, because what ends up happening is when we get a topical hemostat -- and I spend 50 percent of my time working on animals and doing bleeding, injury models, and then the other half actually doing testing in human subjects.
And what we find is we actually end up designing our human trials based on what we find in the animal work. Because in other words, if you have a product that may cause ill effects, for instance, on the lung, you may not want to work -- use it in the chest. Or if it causes adhesions in the abdomen, you may not want to use it in the abdomen.
The next thing that works that you think about it is some topical hemostats actually work to increase clotting at the local area. Others work strictly as a sealant or as a stick-on like a cytosine, really not a hemostatic area, but it really is working as a sealant to cause sticking. So you have to start putting this all together when you think about clinical trial endpoints and local hemostats, because it's a little bit more complicated than you think.
And we really need to kind of go back to the animal work to say, okay, is it going to work in specific conditions like non-coagulopathic versus coagulopathic, issues on vessels versus structures like solid organ. And those are important factors. So when we start off about -- you know, bring questions to the panel, the first thing I think we have to talk about are local systemic factors.
And the question I first put out out there is anatomic factors. Is it different when we're going to plan a clinical trial, if we're going to study something in the chest whether we're looking at the lung versus the liver, versus the spleen? How does the panel look at that? We'll start off with, say, Hasan. What would you think about that?
DR. ALAM: At some level, you would think that those factors are important. But at the end of the day, most likely, I would say no. Most of the complicated clinically relevant trauma bleed comes from a variety of different sources. You get some arterial bleeding, some venous bleeding, some soft tissue maceration, some capillary using this -- all this coagulopathy, there's all this tissue injury.
So some of the nice little boundaries that we draw about purely arterial versus purely venous versus purely bone stuff, that becomes completely irrelevant when you're talking about a clinically massively bleeding polytrauma patient. And somebody who's got just a tiny hole in a blood vessel, you know, all you're going to do is put three stitches in.
I mean if -- we're not talking about local hemostatic agents, but this is for big, destructive wounds, and they bleed from a variety of different sources. So from a clinical standpoint, I think it's not all that important to stratify that. You need that information to figure out which dressings work better for a certain kind of injury. And I think there would be differences.
What's more important in the -- is to look at the clinical endpoints. And the way I look at it is whether the dressing works and -- come to that in a second -- and whether the patient lives. And they're mutually exclusive -- dead patients don't bleed. So the dressing got to work and the patient got to live. And sometime the dressing works is pretty obvious.
And all these simple things about cc of blood loss per minute -- and again, not very meaningful, but we can always look at surrogate markers, whether they got more blood transfusion, whether they got additional interventions for hemorrhage control. So the short answer to you, I think, is yeah, we should keep them in back of our mind, but I won't make it criteria for clinical study.
DR. BOCHICCHIO: About the issue of named versus non-named vessels and a little bit more specifically artery versus vein, would you grade that any differently?
DR. ALAM: I think they're clearly important. But again, if you're thinking about looking at agents to control bleeding, a pure arterial hole, truly like a 3-millimeter or 4-millimeter hole in the aorta, complete aortic transection. Let's think about complete aortic transection or cable (phonetic) transection. There is no hemostatic agent that's going to work on that. It won't, I mean it's as simple as that.
So we're talking about -- yeah, it's important. But again, if we have to set criteria for selecting agents or defining success, I won't make that into a thumbs-up or thumbs-down criteria.
DR. BOCHICCHIO: Dr. Cap, what did you -- what would you look at -- say, for instance, if we have a device that can temporarily stop a bleeder bleeding from the aorta, you know, was that -- is that going to be a different set of qualifications to say yes, you have an aortic injury, it can be used for a certain period of time, but then you have to remove it? Does that have military implications? How would you comment on that?
DR. CAP: Well -- so I guess you're thinking about how you would study something like that and you have to get to the aorta. So there is a lot of work being done on sort of an external abdominal clamp acting as a tourniquet. So that is one technology that the military is currently assessing. And clearly as a compression device, timing does make a big difference --
DR. BOCHICCHIO: Just -- for the purpose of this, like, local agents, like, topical how would you --
DR. CAP: So you -- you know, I think that that would have to be, you know, in the context of a larger polytrauma, because you have to actually get down to the aorta and expose it. And I think if you're doing that, you're doing it in a surgical setting probably unless, you know, the entire abdomen is -- anatomy is disrupted.
So I'm not really sure how you would study something like that, frankly. I think it's just not a realistic scenario. I don't know.
DR. BOCHICCHIO: Think about it in a true real life damage-control situation. If you can really make your way down to the aorta, you're looking at it, you'll put a clamp on it or -- fully in it or put a shunt in it or do something. And at the end -- and if you can't do it, the patient will die. It's as simple as that. It's carrying 4, 5 liters per minute of blood through that organ, that conduit. It's when you --
SPEAKER: But what if you had a device? What if you had a topical hemostat?
DR. BOCHICCHIO: Sure.
SPEAKER: And these are out there in the animal world right now that if you had a hole in the aorta, you could take this 4x4, place it on there temporarily, and it'll stop the bleeding. And then you can then get your proximal and distal control and then go back and repair that. Is that something that, you know, would be something that we look at in our trial design as something that, you know, it can serve as a temporary.
There are, in other words, animal data from the ISR that demonstrate pseudoaneurysms after placing some of these devices down the line. But how would you look at that as trial? Is there a role for temporary hemostats in some of these injuries?
SPEAKER: I think Tony did a very good presentation or summary this morning. And some of those models are designed not to be -- to replicate clinical scenarios, but to challenge the dressing. So you can have a high-volume high-pressure bleed, high-volume low-pressure bleed, and a variety of different combinations.
And the dressing has to work in a high volume -- preferably high-volume high-pressure bleed. And your implication is that if it works in that really challenging scenario that it would work in some other scenarios as well. If you're really looking at a 6-millimeter hole in the aorta in clinical practice, not too many people would like, you know, use hemostatic dressing on it.
You'll put a finger on it or clamp on it or stitch in it and so on and so forth. So those models were not replicating clinical reality, but they were creating a challenge for the dressing itself.
DR. BOCHICCHIO: Dr. Lawson, what do you think about the issue of named vessels, larger vessels, the cava, the artery versus, you know, branches of vessels using these type of products on? Is there a difference? Certain trials, that's going to deviate you or prevent you from wanting to use certain of these products on a larger vessel versus smaller vessels or using it, as you say, after a -- you're finished with your anastomosis and you're going to put something on there at the end.
DR. LAWSON: So first, I like the name "vessels." I work with them daily. (Laughter)
DR. LAWSON: I actually think -- I'll take a sort of contrary view to this. I think it's critically important from a trial design, not so much -- not just what we use, you know, in the OR and clinical day in, day out. You may use certain things you become comfortable with.
But if the goal is to design an experiment that you can successfully execute, the more variability you can take out of that within the clinical trial, I think the better outcome you will have. And so for instance, the cyanoacrylate that I referred to at the end of my presentation, that you dispense, you know, one to two drops of the stuff around the suture line of a vascular anastomosis that is strictly looking at homeostasis in the setting of a vascular anastomosis.
It's an important little model that tested. It's very different than taking a fibrin bandage and saying I'm going to apply this to a liver laceration with a certain amount of pressure for a certain amount of time. And I think that's in contradistinction to something, let's say, like a fibrin sealant or maybe some kind of procoagulant snake where you -- the more you can define it from a clinical trial standpoint to have the model be reproducible, I think the more likely you will be to get a result that will be positive.
If you just throw this stuff into everything from a hip fracture to a ruptured vena cava, to a, you know, distracted, you know, thorax, you're going to get a random array of responses. And I think that's the challenge from a clinical trial standpoint. That doesn't mean you don't want that device to be effective in all those scenarios.
And you -- we may use them clinically in all those scenarios, but that's different than testing it scientifically.
SPEAKER: Can I add on to that? I think that makes in particular these studies difficult to do in the very heterogeneous trauma population where our patients are heterogeneous and our situations are heterogeneous. And it means that we have to extrapolate a large amount of data from very controlled animal experiments and very -- and more controlled elective open vascular surgery, for example, experiments.
And it may be blasphemy to say this to a crowd of hardcore statistics people in a FDA-sponsored workshop. It means that our clinical trials and trauma need to be adaptive. And I use that in the real meaning of the word and the loose meaning of the word. And it means that at some point we have to define our endpoints and our groups as we go. It's blasphemy to say that.
But because things are so heterogeneous and because you may apply a dressing to an aortic tear that really is only a few millimeters and really requires just a few sutures and then it's a success and in the next patient you may apply the same dressing in the same inclusion criteria to a complete transection in a patient who will never be salvageable and you get very different outcomes.
It means that the data ends up being a disaster. And unfortunately, as others have said, it's almost a you-know-it-when-you-see-it kind of criteria. And in order to do a setting like that, you need to know it when you see and group those patients later, combine that less statistically perfect data with animal data and more controlled elective data and then you get some sort of outcomes and know what you can do.
DR. BOCHICCHIO: Focusing on the question -- the comment you said, you know when you see it, the issue of defining what bleeding is, severe bleeding, rate of bleeding, how does the panel feel about defining mild to moderate or severe bleeding or catastrophic bleeding in this area of trauma? Our next panel member?
SPEAKER: Just working our way down the line, you know, there are definitely different rates of bleeding. And it's a little bit different when you're talking about animal models where you're able to measure it fairly objectively. Estimated blood loss is very easy to measure in a pig. It's not very easy when you're trying to measure it off of what's in the suction pot, and you know, the sponges.
So the -- depends on whether you're talking about analyzing something in an animal model or whether you're talking about using it clinically. It's going to be very difficult to measure that clinically at times, especially if you have internal bleeding. And now, this was all for topical agents so we're, you know, assuming that there is some exposure either surgical or it's accessible --
SPEAKER: Yeah, you're seeing it right in front of you. You're seeing it right in front of you.
SPEAKER: Yeah. So yes, the rate of bleeding is important. It kind of -- what was the more specific question that you --
DR. BOCHICCHIO: Well, I guess the question is then how do we -- how do we differentiate in terms of products that a product is approved for severe or catastrophic bleeding versus mild to moderate. In other words, you have Surgicel -- you know, we -- everybody uses Surgicel -- some use Gelfoam, Thrombins for mild bleeding.
But what about when we really take it up to the grade V liver in a coagulopathic patient? How do we get our hands around defining that severe in an objective way rather than I know it when I see it? Is that good enough for a clinical trial?
SPEAKER: Can I just tackle that? Because some of these questions were front and center when all these advanced hemostatic dressings were designed and deployed. So 2004 -- 2002, 2003, 2004 all these dressings were being tested and approved and all the stuff and that they should be deployed or not.
And some of the same questions were brought up. I mean should you be putting on every laceration, every wound. And we have to come up with a practical way of stratifying patients rather than just saying, you know, we're going to look at the grade of the injury and so and so forth -- just totally impractical in exsanguinating patient.
So I propose a fairly straightforward algorithm, and this is about the same kind of algorithm we all agreed on at that time. You have a bleeding that's bad enough that it fails your conventional treatment. You can't apply tourniquet on it and you try your usual thing which is gauze, wrapping it, putting pressure on it.
If it works, that by its biology has sorted itself into controllable bleeding. If that doesn't work, your standard treatment doesn't work, then whether it's the nature of the wound or the structures hit or patient's coagulopathy or a combination of all of those things, that injury has put itself into the uncontrolled category.
And that's when you move to the next tier of intervention which would be advanced hemostatic dressings, or in this setting, the new local hemostatic agent.
DR. BOCHICCHIO: Now, interesting that brings up the question in animal trials we always design our trials as primary hemostats. So in other words, we do the grade V liver clamp injury model. We don't throw sutures in that. We actually go ahead and apply the dressings for immediate hemostasis.
Is that something the panel think that should be different in the local hemostat world because it's not mimicking -- we're not mimicking the animal trials and human trials. What we're doing is we're using it as an adjunct if failures work. And now we lose time. And as many of the previous speakers said, as time goes by, you know, our mortality goes up.
So is there a difference in looking at mild to moderate severe bleeding? If you know a product does work, or you feel confident that in the lab in a grade V liver injury, it works 100 percent of the time; immediately in 3 minutes you have hemostasis, versus now you're filling sutures, you have continuing bleeding, you're throwing pack, and you're doing the things that you conventionally do. Is that a disservice if you have a type of material that might do it instantaneously?
SPEAKER: Sure. I think from a clinical trial's -- at least design and outcome standpoint, because it's -- again, just to separate clinical utility of something that we would use in the OR everyday versus getting something through an FDA trial to demonstrate efficacy. So I agree with the idea about you could have failure of conventional therapy.
I think that's one reasonable than now you've got -- you know you've got something back. The other alternative which has not been done that often but can be done is to film the event and then have it adjudicated. So you use this -- the enrolled patients -- you're going to put the fibrin bandage in every patient that comes in with some injury criteria.
And then on the backend you adapt your stratification based on -- yeah, you know, couple guys sitting in a room watch the video tape of what happened and say yeah, that was really bad. And couple other people say yeah, that was kind of moderate. But it's subjective, but you can have it done in an adjudicated fashion.
SPEAKER: And you have to adjudicate what the other treatments were and what was the reason for the efficacy, right? If you apply the hemostatic bandage but it really turns out that getting a clamp on the vessel or having a spleen in the bucket is what stopped the bleeding, then maybe that's not efficacy of the local hemostatic agent and vice versa.
And so -- and I'd be very interested to know what other people in this room think about how you would do. You know, this video taping is an intriguing idea, but how you would do that sort of adaptive -- and maybe that's not the correct word -- grouping later, or you know, post-hoc grouping is really what it is.
SPEAKER: Yeah, I mean I haven't seen that used in a trauma model. It has been used for vascular anastomotic bleeding where each -- when you finished sewing the thing together and you took the clamps off, somebody started rolling the camera and objectively quantifying how much bleeding there was from the side and then how much -- you know, I mean independent of the surgeon doing the case, because you know, we are alive and our patients do great.
So you need somebody who is outside the room who can review the tape and isn't biased by the immediacy of the situation if you're going to do something where you're going to apply some objectivity to a subjective event.
SPEAKER: That's not going to work for trauma, because presenter after presenter in the morning talked about how the patients die pre-hospital. The medics are putting it in, it's a chaotic situation. And if they are alive to be filmed, they will live.
SPEAKER: Well, I think it begs the question as a non-(inaudible) trauma. I mean I use limited traumas when, you know, the femoral artery gets shot with a bullet and I go and fix it. But is trauma -- although the relevance of trauma can -- you know, is obvious, is it the right model to study bleeding from a clinical trials design standpoint?
Because of the heterogeneity of it, I would argue that it might not be the best model from a clinical testing standpoint. And control trauma, which is elective surgery, is maybe a better model design --
SPEAKER: And then leaving as a segue as, you know, most of the clinical trials that are done right now for approval, are done in elective cases. And whether, you know, predominantly, you know, large cancer cases where you're going to have more bleeding in those cases and the question then really comes up. If you have a topical product, how do we look at time to hemostasis?
How does the panel -- is it 3 minutes that you have to apply a product for? Is it 5 minutes? Is it you look at it every minute? How does the panel feel like -- you know, is there a predetermined time for each product? Or should you look at it every minute? And how does that work as far as -- as Dr. Lawson said, you peek at it every minute?
Maybe the pressure -- by taking the pressure off of the dressing yourself, you're actually causing some disruption of the clot or the pressure itself acts to help the -- or induce homeostasis. So --
SPEAKER: You're talking about a clinical trial or animal trial?
SPEAKER: Clinical trial.
SPEAKER: Clinical trial, I think it should be whether the bleeding stopped in a truly -- I mean, it's not the peeking thing, and the time to stopping the bleeding is not as important, it's whether the bleeding stops. So what I propose that it should be patients who otherwise have failed. These are the patients who are just proceeding on to die.
You put the agent in -- agent x and your criteria is whether the bleeding stop and secondary endpoint is that the patient lived. And if the bleeding stopped, sometime you can quantify it by yes/no answer. But often enough is surrogate markers where they required more blood transfusion, where they required further intervention, where they required something else, so -- where they took 3 minutes or 15 minutes.
SPEAKER: But most products you're going to have to apply -- you're going to apply it manually. Most of these are --
SPEAKER: Sure, sure --
SPEAKER: So there's got to be a -- some kind of time period that if you're going to apply the whole pressure on it --
SPEAKER: Yeah, I think this --
SPEAKER: -- for a certain period of time. So the question is, you know, predetermined -- 3 minutes good enough, 5 minutes good enough.
SPEAKER: It's difficult to imagine that that would apply, though, because while you're pressure if you have a predefined x number of minutes with this dressing, you're doing other things, right? I mean you're trying to dissect those vessels out to get real control. You're working on other things. Patient has multiple other injuries. You're trying to get --
SPEAKER: Well, for instance, we're not -- let's move aside from trauma for a minute, just -- like, in an elective trial that you say you have a defined bleeding site. You're done an APR, you have bleeding from the pelvis --
SPEAKER: Oh yeah, perfectly clean --
SPEAKER: -- it's mild to moderate bleeding --
SPEAKER: I mean if you have a clean model like that, it's not -- that model is not that many steps away from an animal model where exactly you define what your, you know, your endpoint is -- bleeding, no bleeding, and your amount of time depending on the -- how the product should work is, you know, x number of minutes and then you can probably quantify bleeding in that case --
SPEAKER: So is it -- so the question really -- what I'm asking is it 3 minutes? Should it be 5 minutes? Should it be something that's worked out, and you prioritize as far as what the time limit should be? SPEAKER: Depends on the product and the application is.
SPEAKER: How about looking at it from the standpoint of sort of what you're going to do next, which is a little bit of a -- you know, where Mitch was going with this. I mean even in an elective surgery setting, you know, after you've done whatever intervention you're going to do, there is usually several follow-on steps that, you know, maybe delayed by trying to control bleeding.
And if you can reduce the delay of bleeding compared to patients who don't get, you know, the hemostatic intervention, that could be an endpoint. So if you spend a whole bunch of time in your case trying to control this, you know, oozing from your APR or giving blood products, let's say, and you reduce that, that could be useful.
But the other thing I just wanted to mention about the problem between looking at elective surgery settings compared to trauma with regard to homeostasis is that you're leaving out the coagulopathy problem. So you know, you may get adequate control of your bleeding with a topical agent in that setting, but if you don't have shock and coagulopathy, well, then you're not really capturing the unique setting of the trauma.
SPEAKER: So the question then comes out should products then be specifically approved for, you know, elective, non-coagulopathic patients. And then -- and as an additional step for patients who are coagulopathic, that would be more resistant to hemostasis, so -- a routine hemostasis.
SPEAKER: Well, that would be one way to look at it. The other thing I want to mention is -- you mentioned cancer surgery. And the baseline sort of inflammatory and hemostatic profile of a cancer patient is dramatically different from sort of a normal person who gets a bypass. And that's also true in the liver disease and so forth.
I mean these are very different diseases that affect the hemostatic and inflammatory systems uniquely. And I think that needs to be kept in mind. What works in a liver patient may not work as well in somebody who has cancer or in a normal person who suffered a trauma.
DR. BOCHICCHIO: So going back to the original question as far as the time to homeostasis -- (pause) -- so is there an ability today say that, you know, whether it's trauma or non-trauma, you know, typical trials right now currently for local hemostats are -- usually it's hemostasis at 3 minutes or 5 minutes. And then there's a second look at, you know, the secondary outcome at 10 minutes or at closure.
Is that something that is -- the panel feels is reasonable? Does that need to be changed as -- do you need to look at that local site more often than 3 minutes, or you know, holding pressure a 3-minute standard or is it really product specific?
SPEAKER: I think for trauma patients -- it's almost funny to talk about it. If we do a damage control and pack the, I don't take the packs out at 3 minutes, 5 minutes. I don't take him out for 2 days. I mean if that thing -- it stopped bleeding, I don't want to peek at it. It stopped bleeding. So all those trials they're talking about 3-, 5-minute peek and all that stuff, those are fine for suture lines, stable patient.
I just want to close it definitively and I'm looking at those three drops of blood that are coming through. I see it, I hear -- I mean -- and if the trial is designed in non-trauma patients, then I think those are debates that should be had. But for a trauma patient where we're talking about massive bleeding, you just control the bleeding and you don't peek at it again.
This morning all the presentations again emphasized that we had a number of products that are available for mild, moderate bleeding, for oozing, so and so forth. For big audible trauma bleeding, nothing is approved. And if that's the trial that needs to be done, difficult as it may be, then we have to come up with more meaningful criteria for that.
SPEAKER: But I had some --
SPEAKER: But just want to say that Peter Rhee has been standing there for -- at least I've been watching him for about 15 minutes. His legs must be getting tried now.
DR. BOCHICCHIO: Peter?
DR. RHEE: I just wanted to just get that -- but -- so if you're -- our new biotech company and you come to the agency with a new Thrombin-like molecule that you think is going to behave similar, and all the precedent has been time to hemostasis because that's what the -- your preceding companies have done to get approval and you have to compare your drug to their drug.
And now to the point that Grant is trying to make is that the trial design itself affects the outcome, because you have to take your stuff that you put on the anastomosis, for instance, and peek at it. And when you do that, you disrupt what you -- same reason we don't like to take this stuff out is you make it bleed again.
And so by developing the data that the agency wants, to compare your product to their product, you're forced to do something that you wouldn't normally do, that is sort of the nature of how difficult it is to do clinical research in this area.
SPEAKER: You know, for those things -- I mean then you send the company back home and say, you know, take all these models that Tony Pusateri talked about, this morning about the grade V liver injury, coagulopathic animals, and so forth, and compare your product to products that are already being used. So that thing gets done in a preclinical arena in a robust animal model. And those are totally appropriate model to test those 3-minute, 5-minute, 10-minute questions. The other scenario where you can do it safely is elective surgical patients where they're oozing from the bone or bleeding from the suture line. I completely agree with you.
I mean those things can be done either in an animal model or in a controlled clinical trial. But I agree and that --
SPEAKER: At the end of the day -- I mean to translate most of those things that work in controlled settings really don't work in massive audible bleeding. And what we really need, what is missing right now in our armamentarium is not the putties and dressings that's -- control oozing, but something that controls big bleeding. And that has to be tested in a big bleeding model.
DR. BOCHICCHIO: Well, I think there's two -- but there's two main questions and we can get the Peter's questions, which is there's a time element, how fast does your product work, because if it takes 20 minutes or 30 minutes to work versus 2 minutes, in a trauma world that's too long. So I think that that clearly needs to be an endpoint in your clinical trial.
The second thing is how often does it work or how often is -- what is its success rate in terms of -- is it what -- when you are using -- it doesn't work 100 percent of the time or does it work 90 percent of the time, does it work 50 percent of the time. The question is if you have a product, what's the minimal acceptability rate or efficacy rate for it to be approved.
Does it need to be at least 80 percent, 90 percent? Keep in mind there are technical issues. As we all know, there's application errors by surgeons no matter who it is that may impact that. But that's something that needs to be also addressed. And Peter, we'll get to your question.
DR. RHEE: Peter Rhee from Tucson, Arizona. Of course starting with animals is always the right thing to do. We eat billions of billions of animals, so a thousand animals a year is cheap and nice way to screen things. But what we have now is a lot of products which are already FDA-approved and are being sold and marketed.
And a lot of soldiers are carrying stuff and they think that this is life-saving work. So we really have to go to the next stage which is human trials, because we have virtually no human trials on these new local hemostatic agents that are out there. And we know in these animals everything works; everything works well.
And when we have the right model that -- and you're looking at the intracorporeal portion of it that we have, you know, coagulopathic animals and it works well on them. And then we have extremity type of injuries and you put these things and you shove them in your -- put gauze in there, they seem to work well.
But when I carry these products in my hospital, in my butt pack and I use them, none of them work. So we need to design trials in humans now and try to figure out which one of these products are better than the other, because these companies are selling a lot of these things and advocating them to work when there is not a single bit of human data that says so.
There are some survey stuff and case series that says a lot of people think it works and that they have seen it work. But we really need to design a good prospective randomized trial on these things. And when we are talking about intracorporeal use, that's a trial specifically for surgeons.
And within that, there is two separate categories. One is for vascular or cardiothoracic which care about suture line, and trauma patients where we don't care about suture line. The primary endpoint for trauma patients is, like, livers and stuff like that where we can't get them to stop bleeding in those people -- whether you re-operate or there's bleeding at re-operation or survival.
But what we also need to do is we need to have a local hemostatic study where the paramedics and the medics in the field are just putting it on wounds. And that's the hardest study that really designed, because I don't know how to measure that. But I know when I got somebody with an open wound just from an open fracture, all the stuff that are in the market right now, I've tried every single one of them, and I can't get them to stop the simple minor of bleeding from the open fractures.
So I think that's what we need to focus on, not necessarily whether that suture line is going to bleeding for another 3 minutes or not.
DR. BOCHICCHIO: So that brings up the point that -- questions from the audience was is there a rule for a local hemostat to achieve primary hemostasis in trauma. And again that goes back to the question -- I think that there is an issue right now where there is a lot of products or several products that are in the pipeline that are not approved for animal -- not approved for human trials that actually can achieve hemostasis in grade V coagulopathic liver trials.
Or you can even pretty much cut a hole in the ventricle with some of these new devices and stop it with them in a minute. So when you're talking about severe hemorrhage, these products will work. But the question is then how do you get these things to market, how do you approve them, and what are the surrogate endpoints. Is it time? Is it 3 minutes or 5 minutes?
Is it a coagulopathic patient? How do we define the severity of the bleeding? Because we don't really have a way of defining it's more than mild to moderate.
We have a feeling for that, but how do you say yes, the FDA is going to approve this product that for a patient who does have a multiple gunshots, who -- a grade V liver injury, who's got audible bleeding and you want to use that maybe as a primary hemostatic packing where in the lab the day before you put your packs on and it stopped it, are you willing to go ahead and do that in the patient who's sitting in front of you that's dying? And John, you're probably the perfect person to answer that question.
SPEAKER: I don't know if I'm going to answer. I'll just make a comment. That's actually a little bit more complicated that you laid out, Grant, because usually these patients have multiple sites bleed. So suppose the animal trials which are nice and convenient -- the trauma patient frequently is bleeding from multiple sites and one site may be bleeding more than the other.
As the picture I showed this morning, the guy had three bleeding sites. His major bleeding site, which was audible, was from his liver. The other two sites were more, like, irritating. And so -- Jeff, I put your stuff on the irritating sites, you know, that they really were bleeding cells, it wasn't that big of a deal. And they stopped and it really doesn't matter at all.
So you really have to be able to separate these things out. I think the trauma studies are more complicated. I do think that the trial design has been extraordinarily different from the two drops in anastomosis, Jeff, that you talked about is extremely controlled. You're not ever going to have death from that anastomotic bleeding, you're not going to have transfusion. Those are things that we talk about.
So if you don't have death and transfusion with trauma bleeding, who cares? It doesn't matter to the patient. It may cause an effective hematoma to take back from elective vascular surgery. That's not what our endpoints are and we won't ever be concerned with those. And I know you know that and feel the same way for the most part.
The -- I think that the regulatory pathway of time to hemostasis is interesting. Nobody talks about it, and it's not clinically relevant. We talk about death and transfusion rates, rates of red blood cell transfusion. On rounds if you go to all 1,200 trauma centers in the United States and around the world, that is what's discussed.
And I think the conundrum here -- and the slides that we heard earlier from the regulatory folks of bringing in clinically relevant endpoints, need to really be -- we need to pay attention to that.
SPEAKER: Shall I -- you know, I just got to respond to this sort of intoxication of just being a trauma surgeon and saying everything else isn't relevant.
SPEAKER: Because you know, there is a lot of surgery that goes on everyday that isn't trauma that causes a lot of bleeding and clotting out there -- just a lot of things in hospitals, you know. And I agree with you that I -- probably fundamentally a trauma trial design should be fundamentally different.
And you may never be able to get a hemostasis in that complex of an experimental preparation. I mean just -- it just may be undoable. But you know, with the hemostasis or hemostatic products, you know, if I'm doing a fem-pop and putting a couple of drops of stuff on your thing, yeah, it might be interesting to see the time to hemostasis.
But it is really relevant if the fem-pop thrombosis -- and they're going to cut your leg off. So it's not to know if it causes the anastomosis to fall apart and you have a groin blowout. I mean, those are critical life-threatening situations too that then it does turn into a trauma case. But just to -- not to rebuff that a little bit.
SPEAKER: And you know, I think that for trauma studies -- I mean you're not going to have, you know, a primary hemostasis endpoint for audible bleeding. And that -- it's going to be an adjunct to surgery. They have a proline deficiency not a hemostatic agent deficiency. And so, you know, that's one thing that you have to really think about.
And you know, when you're talking about trauma as complex as all the different injuries are, you also have to keep in mind when you're doing a clinical trial that there is going to need to be some consistent way of managing these patients. You know, you have one place that resuscitates patients in one way and another place that resuscitates them differently or two surgeons that resuscitate differently.
You're going to get different outcomes. You're going to get different outcomes, and so I think key part of any clinical trial would need to kind of have some sort of approved on way of managing patients, you know, without getting into the specifics of injuries.
SPEAKER: And that brings up the question of covariates. And you know, we just recently finished a trial looking at -- I mean, you actually used time to hemostasis in a trauma trial where we actually did this prospectively at our institution where once the patient went to the operating room and they had bleeding identified as a surgeon, then the clock would start and we would time it from the time the bleeding started until the surgeon said that hemostasis was obtained whether -- regardless of how it was.
And that actually did work in -- and I think the case will be presented at East (phonetic) next month. And it actually showed that blood transfusions were less with decrease in time to hemostasis and the surrogate endpoints. So the next question that really comes up with that is -- what we talked about time to hemostasis -- should blood transfusions be a major covariate in a trial as far as outcome.
We heard previous data earlier about a lot of these trials, there wasn't a lot of blood transfusions in some of these trials. So now in order to get those big blood transfusion groups you have to get to them earlier, which will make it more difficult to get those severe patients. So --
SPEAKER: I think you have to consider them covariates. You can try to reduce the variability all you want, but ultimately -- and what I was talking about this adaptive -- for lack of a better word -- design is exactly that, figuring out what all the covariates are, which makes it difficult to know going in in a trial, exactly what the indication and the outcome is going to be.
Because really what you're going to figure out is within these covariates this particular device or this particular -- either systemic or local treatment works in these circumstances of injury demographics, patient type, what the surgeons did, and in the context of everything else. And that's the kind of data -- that's the best data you're going to get out of these trauma -- to these trauma trials and probably how these trauma trials need to be designed.
I will also just add one thing that I think that we as trauma surgeons we talk about it a lot, but it's an endpoint that we could talk more about and agree upon. We talk about a lot but we don't quantify it, the end of bleeding in the operating room.
And if that were something that were better taught and better codified and of course there'll be variability in it, but the surgeon's saying, okay, bleeding is done, now we're just mopping up or sewing (inaudible) together or whatever, it would be a very useful endpoint that could be, you know, hopefully less variable, you know, from surgeon to surgeon and place to place.
SPEAKER: Some of these issues that we talked about, the nature of the injury, time taken to achieve hemostasis, the amount of blood products and all this stuff, in a trial situation those would be things that you would look at post trial in an analysis fashion. The certain agent might work better for one or the other.
But in a -- going into the trial as inclusion-exclusion would become -- I mean since these are rare injuries that we're talking about; this is not every patient who's coming in. To include or exclude somebody based on certain predetermined criteria and listen till they're very logical, would become little impractical.
SPEAKER: What about -- I just want to throw this out there -- the idea of constructing a composite endpoint that would look at the rate of development of coagulopathy or how bad the coagulopathy became, as well as alteration in physiology, like, severity of shock and so forth as a -- as an endpoint to studying the interventions so that, you know, a patient who got topical hemostatic agent x, you know, how do we know we really controlled the bleeding?
Well, you know, just talked about all the problems and looking at time to hemostasis, but if you go back and look at -- you know, they had the severe shock, they had less coagulopathy, I think that could be a sort of harder and more objective --
SPEAKER: But is it -- I guess I'll just throw the question out. If you have -- if you see audible -- and that's where I think the big difference between topicals and the IV formulations. In topicals you're going to give in early, you're not really seeing where the bleeding is. When you're using the topical hemostat, you see the bleeding in front of you.
So if you're able to kind of in a sense put the finger on the dike and stop it, then you've kind of -- you know, you've seen or you've done what you need to do. All those other factors may occur or not. But is that good enough to say okay, we stopped that bleeding there. How long did it take and how long did it last for? And then complications down the line.
Is that a good enough endpoint for a trauma trial or even a ruptured aorta trial where you have a patient coming with ruptured aorta with severe bleeding and you open up the abdomen it's bleeding, you take a fibrin patch or whatever you want to call it, and slap it on the aorta and hold pressure and it stops bleeding and now you can get proximal distal control and do your operation?
SPEAKER: The -- I think a complication there is that -- you know, we talked a little bit earlier about trying to move such trials to a pre-hospital setting. And I think the, you know, sort of hand off of information of, you know, if the medic in the field puts a dressing on something plus or minus holding pressure for who knows how long during transport, you can't really control that.
And so the question is did -- were those dressings applied by the people in the field effective or not. And at the end of the day, you're only going to know when you sort of look at the physiology of the patient and say yeah, they probably lost less blood because they're looking better.
DR. BOCHICCHIO: Well, I guess that -- before I get to questions here, that's a very good question. Is there a difference -- or should there be a difference in product approvals for military use versus in-hospital use or pre-hospital use versus in-hospital use depending on the provider who is placing it, their experience in the ease of placement of the device, and is it temporary or not or does it cause you issues.
For instance, if you have a powder that's used in the field and then it stops bleeding there, but then you get to the patient in the hospital and as a trauma surgeon you're trying to figure out what's going on and you have powder everywhere that's kind of blowing the scene, and you know, that kind of, you know, then throws you off.
Is that a good thing? I mean -- so I throw that -- should there be different approval processes for military field use versus civilian use?
SPEAKER: Clearly, yes. Clearly, yes --
SPEAKER: Or at least pre-hospital.
SPEAKER: You know, the expertise of the person who's applying it, the environment, in the middle of the night, up back in the Humvee, in a operating room with all the equipment, sure. I mean, if I have two nurses to mix something and heat it up for 15 minutes before they hand it to me, I mean it's totally different environment there like in a dish. Of course.
DR. BOCHICCHIO: So we're -- so we're clear. Jimmy, you're -- are you in agreement to that there should be different approval processes for military field use and with paramedics placing it versus in-hospital?
SPEAKER: Military field medics placing agents is the same thing as pre-hospital.
DR. BOCHICCHIO: So you would lump those two together.
SPEAKER: I would lump those together.
SPEAKER: Yeah, I wasn't arguing for military-unique labeling or anything, I was just saying that, you know, that -- what Hasan was saying is that, you know, difference between an operating room and any ambulance anywhere.
SPEAKER: Yeah, conceptually speaking, I -- that's how I would think of it, you know, how would you do a clinical trial. You know, if that's where the military wants to use it, then that's a pre-hospital study.
DR. BOCHICCHIO: Questions.
DR. MATTOX: Ken Mattox from Houston. I'm a cardiovascular surgeon by training with little bit of hobby in trauma.
I rise to raise a word of caution. With what I have heard the last 30 minutes, no topical agent would ever be approved. If you define things so widely, Surgicel would not get through the approval process. Define what we wish for a topical hemostat and the topical hemostats, by almost definition, are a damage control product.
If it isn't bleeding I'm not going to throw anything on that wound, I'm just going to close it. If it's bleeding I'm going to throw everything on it. And if it stays white, if the dressing of whatever I use was white I'm going to be happy and I'm going to say fine. If it turns red, I'm going to put more on or take it off and try something else or throw in a vascular clamp.
Define what we wish with a topical hemostat very narrowly so that the company can make that narrow-defined product and the marketplace will determine whether or not it works or not. For most bleeding pre-hospital, outside of the IED explosion, is going to be stopped with a simple little dressing, especially in civilian practice.
I've probably been through 15 topical hemostats that are now in the historical section of the medical museum that nobody uses anymore. Surgicel works the same way a little 4 by 4 would work. So define what we wish to of product development very narrowly and then let the marketplace determine does it work or does it not. Don't over-engineer our endpoints or our definition.
Don't try to stop the transected aortic bleeding because it's never going to work. And those people have, for the abdominal wounds, a 70 percent mortality anyway and we're going to define failure. We shouldn't define our definition -- construct our definitions to cause the failure of an ultimate product. DR. BOCHICCHIO: Thank you. Yes.
DR. MANN: Ken Mann, Vermont. I'm the ultimate reductionist in this room because I don't work with anything that's alive. When I -- I like to use football analogies, and if you will follow football you'll realize that if you're going to win games you got to have both a running game and a passing game.
And, you know, in the previous section we discussed animal models and their appropriateness or inappropriateness to basically be models for bleeding with all kinds of objections. We're hearing a discussion here that goes from the untimed trauma event and the circumstances around that event and the utility of a defined surgical intervention that causes bleeding that can be arrested with some kind of agent perhaps. I think we have to keep in mind and be open-minded that this process is going to be a continuum from the bench to the bed side and to the defined surgical intervention and the trauma intervention situation with all the peculiarities and difficulties that recur.
Remember that all the interventions that you're using right now came out of basic research, the ones that work, thrombin, VIIa, tranexamic acid all came out of basic research, and they were translated into the environments from basic to cellular, to animals, to defined intervention, and, hopefully, some of them or their alternatives will also work on trauma.
But I don't think you want to basically go only with the situation that exists in the trauma, emergency room or the situation that exists in an animal laboratory. We have to have all of those events together to get to the endpoint in a rational fashion that has worked in every other area.
DR. BOCHICCHIO: Thank you. So I'm going to cast a panel here today, give me succinct answers. We'll start with you, Hasan. Timing hemostasis, give me a -- is it 3 minutes, is it 5 minutes, initial after you apply your product?
DR. ALAM: Instantaneous. I'll be happy as long as it works. I just want to take about 10 seconds to just make a comment. Maybe we -- maybe it's impractical to do a clinical trial. In our trauma call room, which is right next to the OR, we have a toolbox, it's red in color to match the bleeding and it's full of haemostatic agents. There are about 10 different agents in there from Combat Gauze to HemCon to QuikClot to everything else, none of them have been tested in a clinical trial and we use them all the time for desperate measures for bleeding. I mean, most of the time Gauze --
DR. BOCHICCHIO: Well, I guess I'm asking if I have a product right now that I say works in every military, animal model there is get me -- give me a time.
DR. ALAM: Three minutes.
DR. BOCHICCHIO: Next.
SPEAKER: Sure, 3 minutes sounds good.
SPEAKER: I would say the -- being a little more abject or obtuse here; that the time that is important for the given agent is relevant to the magnitude of the bleeding and/or the potential risk for an adverse event. So if you're going to pour something that's --
DR. BOCHICCHIO: I want one-word answers for my camp.
SPEAKER: I'm giving you a rational sort of -- or at least my -- that if I got something that's really effective, really fast for really bad bleeding, that's great. But if it's poisonous to blood vessels or poisonous to the liver or something and I would only use it in catastrophic situations then I want something that's going to have a relevant time point that might be slower, more gentle and more biologically compatible.
SPEAKER: I would say either 3 minutes or enough time to allow me to find some other way to stop the bleeding, use something else in our armamentarium to give you enough time to come up with a plan B because as trauma surgeons we're constantly thinking of plan A through Z while we're working on plan A and Z, just need time to get to that next step.
SPEAKER: When you're looking at time to bleeding cessation you have to say compared to what, so you have to say compared to doing nothing, compared to a control, you know, I would argue that time to bleeding is a continuous variable not a yes/no at 3 minutes. And so you measure did the bleeding stop at 2-1/2 minutes versus 3-1/2 minutes. And that's the difference. So it's a continuous variable, not -- it's not a yes/no, it's not dichotomous.
DR. BOCHICCHIO: Okay. Well, I think we're out of time. I thank everybody for their participation.
DR. LINDSEY: Okay, we're going to have a 15-minute coffee break, so you could be back here at 3:00 o'clock so that we can start promptly.
ENDPOINTS FOR CLINICAL EVALUATION OF PRODUCTS FOR BLEEDING
INTERVENTIONS-SYSTEMIC AGEN- - TS
(Tape starts abruptly)
DR. LINDSEY: -- seats. We'd like to get started with the next session. So our next panel discussion is going to be on endpoints for clinical evaluation of products for bleeding interventions for systemic agents, and our first talk is going to be by Dr. Bryan Cotton who is an associate professor of surgery at the Center for Translational Research at the University of Texas Health Science Center.
DR. COTTON: Good afternoon. I don't know what the relevance is but the single conflict of interest I might have with particular to this talk is serving as an adjudicator for ongoing CSL Behring trials determining whether or not hemostasis has been achieved in some of their drug studies.
I want to go ahead and start with what I think is a very, very relevant trauma model and is actually a real case that had occurred in the last couple of weeks at University of Texas Health Science Center at out hospital. We had a 17-year-old male status post-motor vehicle crash sustained severe blunt torso trauma.
He arrived to us incredibly tachycardic, heart rate 140, systolic of the 80s, obtunded, and respiratory distress, he underwent rapid intubation. He was found by chest x-ray to have a pneumothorax, and his ultrasound of his abdomen or the FAST, Focused Assessment for the Sonography of Trauma, was positive for fluid or blood. On getting -- he was intubated, had a chest tube placed, was rushed to the operating room. Upon emergent laparotomy, he was found to have numerous injuries, most significantly a grade 5 liver injury, which is pretty much liver in floating pieces. He also had a severe splenic injury.
During the repair of the liver he lost his vital signs, his entitled CO2 dropped precipitously and the patient was coded. He was placed rapidly into Trendelenburg with the right side up, which made the Sternotomy even more difficult.
A Sternotomy was performed, a 18-gauge angiocath was placed into the right atrium pulling back the air embolism that was suspected and vital signs returned.
They completed the case, take him to the ICU, stabilized over the next 12 hours. He had a full workup, again showing multiple injuries, subarytenoid hemorrhage, multiple rib fractures, long bone fractures. The next day he was -- had his chest closed, another 24 hours later had his abdomen closed.
He developed a pneumonia at hospital day 6 and a PE at hospital day 13. At hospital 17, he was discharged to rehab. So again, we presented our death in complication conference somewhat to what we talked about earlier, a ventilator-associated pneumonia and a pulmonary embolism. And these would of course be endpoints to be evaluated to whether we were efficacious in our care of this patient.
My thoughts to this are, good, I'm glad this guy who died in the operating room developed a pneumonia, I'm glad this guy that died in the operating room. And one of our talented surgeons saved his life, I'm glad he got a PE later. Again, he lived long enough to get the complications.
So multiple endpoint is what we're going to be talking about, Dr. Weiskopf and the panel coming up here in next few minutes. Thirty-day survival, is that a potential endpoint, is 60-days survival a appropriate endpoint. Organ-free failure days, ventilator-free days, even maybe even cognitive or executive function at 90 days are relevant. Well, in this study by Annane and colleagues looking at steroids in severe sepsis, that's absolutely reasonable endpoints. In the intensive glucose control trials, looking at those and pro and con, those are absolutely relevant endpoints, in the original ARDSNet trials and those following up, all of these, 30-, 60-day cognitive function, at 90 days, all of these are relevant endpoints.
In these patients, I don't think they are. I think they are absolutely ridiculous as endpoints and we need to be searching for something more relevant, which is again the point of most of this discussion as well as this conference.
So what about 30-day? Again, deaths from injury are declining, we know this. However, what I found very interested in getting ready for this talk is that deaths from hemorrhage for percent are not, they are not getting any better. Now, our mortality curves are being skewed even more to the left, more people dying in the pre-hospital setting. This is one out of Australia looking at differences in survival from San Francisco in the '70s, San Diego in the '80s, Denver in the '90s and then Australia in the 2000s, looking at the time to death.
And if you look at it, although they are improving and changing, they are still very much skewed. We've gotten our pre-hospital stuff down pretty good, we've gotten much better with aeromedical extrication and rapid transport, they are dying and they are dying early when they do die, but they are not dying late. Again, a change in the amount of CNS, CNS used to be the problem, the central nervous system, head injuries that used to overwhelm us. The percentage of those or the ratio of those to hemorrhage is decreasing. The incidence of multi-organ failure like we talked about earlier, Dr. Rhee pointed out as well, are decreasing, they are becoming non-existent or at least more a nuisance than anything.
Looking at 30-day mortality, again the rates have stayed the same, 30 to 37 percent across the board. How about 24-hour, perhaps that's a better endpoint, perhaps we throw away the 30, we aim for 24. Hemorrhage again in the first 24 hours is the number one cause of death. It's also the number one cause of death in the first hour. The ones that die in the operating room, they die of hemorrhage.
Looking at again Kaplan–Meier curves going out to 24 and 30 days, you will see again very much skewed, the deaths are early; they are on the left side of the screen. Other studies looking at different ratios, again showing skewed survivals, even in the 24 hours are skewed even more to the left.
Other ones coming out of the Holcomb's evaluation of platelet rations and plasma ratios, again being skewed incredibly to that left. Whole blood study out of the military again skewed to the left. And then a recent study by Fred Moore looking again at massive transfusion patients, skewed to the left. So perhaps maybe even 6 hours is about as far out as I really care to look for mortality.
And in that kid that we had, the 17 year old, perhaps whatever thing that we did that saved his life, in this case air aspiration out of the right heart to save his life and get vitals back. If he lives 6 hours and whatever it was that you randomized him to saved his life, is perhaps that not the more relevant endpoint, we'll talk about that.
How about hemostasis? This is probably, if you ask anyone or majority of trauma surgeons this is our endpoint, hemostasis, when bleeding stops. So problem is how the heck do you define when bleeding stops. I mean, like Rick Dutton and Mitch Cohen were talking about earlier, you know it when you see it but can you measure it and how can you quantify that. And it's difficult.
Is it a blinded adjudication forum held to someone like me at the -- what I'm doing now with CSL where I look at something and I see if the patient stop bleeding or is still bleeding at the time. It's kind of hard on a piece of paper to put yourself in that person's shoes who is managing that patient. Is it the surgeon in the operating room, is it the anesthesiologist on the other side of the curtain who kind of gets a feel that bleeding has stopped.
Is it when you stop transfusing patients and then you take a couple of steps back and say I haven't transfused for X amount of time so bleeding has stopped. I don't know. We'll hopefully settle that today.
How about hemodynamic instability or stability, to me that's dirty, there is too much going on with that. The blood pressure, the heart rate, those are things that we measure. Again on arrival we measure. In the operating theater we have a dense data capture system right at certain places where you can get data on a more rapid and more continuous variables.
So do you measure it every hours, every 30 minutes, or do you go on some of these dense data capture systems that measure it four times a second. I don't -- what is the answer and what is your cutoff. So, again 90 served as well when we came up with our ABC algorithm as part of our project for the MPH. However, 70 probably is just as relevant. Is 110 relevant? A lot of that is -- there is so much noise involved with it that I don't know that those should be the absolute surrogate and endpoints -- perhaps for inclusion, but again for endpoints I don't know that, to me it's too dirty.
Again, if you've a 55-year-old female or 55-kilo female that comes in who's got a heart rate -- blood pressure 90, is not the same as a 6'5" guy, 100 kilos coming in that's 70 that's got a blood pressure 90, those are two different physiologies.
How about other endpoints, lactate, base deficits, things like that? Well, we know that they are good predictors. They do predict who survives and lives. However, just absolute clearance or correction of those is not really the good endpoint because even in the Scillia's (phonetic) work from the early '90s showed that a significant people, number of people who did not clear their lactate, never cleared their lactate still survived. And some that cleared it still died. So it's not really there as far as my opinion for a good surrogate.
How about the number of transfusions. If we can decrease the number of transfusions, perhaps that's a relevant endpoint. If we can get it down to components that are decreased, if we'd used less platelets with this strategy or less FFP, perhaps that's our endpoint. And again, we know that every single unit carries a real risk. It's not dichotomized, it is absolutely continuous. Every single unit is a real risk factor. So if we can decrease it from 10 to 6 perhaps that's an adequate endpoint.
We know that that's -- again with not just with multi-organ failure but with ventilator-associated pneumonia as well, perhaps every single unit reduction is important and that is maybe potentially one of our potential endpoints. This is one study that we did where we looked and found with a different implementation of a well-constructed, predefined massive transfusion protocol a reduction in transfusions. And to us that was a real endpoint.
That to us was four, five unit reduction relating to a significant decrease in cost and decrease in exposure to again a potential injurious agent such as blood products. Kera Tokey (phonetic) when he was at -- when he was at Southwestern and Parkland showed the exact same thing, a reduction in products with implementation of a protocol. So whatever it is we're studying perhaps if we can reduce the risk and exposure to these products, those are important.
How about complete avoidance of transfusion as an acceptable endpoint. I don't think that discussion has any place in this arena, my personal thoughts. I don't think that you can really say that in these patients we're talking about right here that we're going to aim for absolute avoidance of transfusion. The only people that are going to really be absolutely avoided transfusions are the ones that die before you can spike the bag of blood to hang it.
So again it brings back to potentially massive transfusion as an endpoint which we talked about earlier in the day. Again we have looked at that. And after implementing certain components of what's called damage control resuscitation which is what lot of centers are implementing and using now we've show a damage -- we've shown that with this implementation process that our emergent laparotomy massive transfusion rate has gone down and our massive transfusions in our damage control laparotomies have gone down. So is that a significant endpoint. That's something else that will be discussed.
How about composite endpoints? Composite endpoints measuring how well we're doing in trauma care and in this specific case severe bleeding. Now, again the -- you know, the most well-defined or at least widely accepted what we published really have no bearing on severe bleeding if you look at the composite endpoints that are there. And so we'd obviously have to come to an agreement as to what is a relevant, what is a durable evaluation when it comes to a composite score.
Is it something like we looked at with massive transfusion where we look at how well we were doing the type and screening, how well we were getting the ratios, whether we were wasting products with our thing, whether we were utilizing certain products. I don't know that those are the appropriate endpoints, but is that what we're talking about, and then we have to get agreement with, you know, a consensus on that.
How about avoiding or reversing coagulopathy? If I give drug X or treatment X and they become non-coagulopathic at time Y, is that relevant? We know that massive transfusion patients are patients in severe bleeding populations whether it be military or whether they be civilian are at a much higher risk of death with just an alternation of a blunt endpoint such as an INR.
Just looking at INR PT the mortality goes up dramatically if their arrival INR is elevated versus non-elevated, controlling for injury severity, controlling for other things, the mortality goes up dramatically. It ventures in again anywhere up to 67 percent mortality in the coagulopathic population versus as low again as 5 percent in the non-coagulopathic population.
How about other surrogates, length of stay, ventilator days, organ-failure-free days, like we talked about before, those are things that are clinically meaningful, but are they relevant to the intervention that we're studying. In other words, like our guy earlier, is it relevant or is it important or do I care other than having to get and preparing for a mortality and morbidity conference.
Do I really care that he got a phenomena? Yes, I care, I'm his physician, I care about that. But at that same time I'm -- I really don't really take that home with me. What I take home is the guy at 6 hours, at 24 hours was alive and breathing and able to communicate.
Other things again, looking at organ failure, looking at different needs for renal replacement therapy, looking at whether you get the belly closed or not, whether he gets such a massive transfusion of blood products of crystalloid, or resuscitation that he couldn't get his abdomen closed versus you gave treatment Y and you got his belly closed, but he still had the same ventilator-associated pneumonia, the same PE rate and the same organ failure rate. But if we got him closed and that guy goes back to driving a truck in six months, to me that's -- that might be successful.
So again other surrogate endpoints are out there, they are available, they've already been talked about, and I'm sure we'll talk about it more in just a moment. But the point is aren't these outcomes really more a statement of just a survivor's curse, truly living long enough to get these complications versus actually these complications being associated with the actual intervention versus the disease itself. So we've talked about and covered many primary and secondary endpoint. I'm trying to keep this as brief as I can so that we can open up a little more discussion with the expert panel that's coming up. We've got a lot of surrogates to talk about and again we'll bat those around. But what we know from this talk and what we know from looking at it is that hemorrhage occurs, death from hemorrhage occurs early, and it is a lethal, lethal model.
We need to focus more on the early time points, we need to stop the bleeding, and that's what we really want. We want to know when bleeding stops. But how to measure that is, you know, above my pay grade and I'm happy to hear any interest from the audience.
And then again making it out of the OR alive, to me that's a big thing, making it to the ICU alive with a patient is tremendous. Perhaps maybe it's even making into the ICU, warm, non-acidotic and non-coagulopathic, and then if they die later, is that really a measure of how good intervention X was at the early time point, and I don't know that. So hopefully we will be discussing that and talking about that as we get ready to start the expert panel here. Thank you.
PANEL DISCUSSION: ENDPOINTS FOR CLINICAL EVALUATION OF
PRODUCTS FOR BLEEDING INTERVENTIONS - SYSTEMIC AGENTS
DR. LINDSEY: Okay. Our moderator for this session is Dr. Richard Weiskopf from the Cardiovascular Research Institute.
DR. WEISKOPF: Thank you. I going to invite all the panelists to come up please and while they are doing that I'll introduce them. You've just heard a very nice outline of the problem from Dr. Cotton at the University of Texas in Houston, and our other panelists are -- maybe each on of you can raise their hand as I call out your name, Rick Dutton from University of Maryland; John Hess, who you saw earlier, University of Maryland; Lena Napolitano, University of Michigan; and Janice Zimmerman from Weill Cornell Medical College, although that's in New York, believe it or not, she is actually at Methodist Hospital in Houston. I don't want to begin to try and explain that.
We have a daunting task not only for the topic but given the time of the day if most of you are like me, this is about time when you get a major sinking spell. So I'm going -- I'll try and keep the panelists provoking you and making sure that you are staying awake.
In terms of conflict of interest, my consult for Sangart and CaridianBCT and although -- and they don't have any real direct interest in bleeding, but they do have interest in allied fields, and I was an employee for Novo Nordisk in Denmark from 2005 to 2007. And they of course do have an important interest in bleeding, but I have no current association with them.
You heard from -- you just finished hearing from Dr. Cotton who very nicely outlined some of the issues. In brief format and sort of broad brush strokes, these are the candid endpoints that we -- that he talked about and that we will be talking about. Why are we talking about endpoints at all? What is the point of that?
My view is that this is the clinical translation of an aim and an indication into reality where the rubber meets the road. This is important not only for academia but for sponsors as well. They have an idea that they want to use a product for a specific indication and then that drives the endpoint. But it's interactive, the endpoint then drives a protocol, and at times one might see some problems with that and one goes back and readjusts the -- any indication perhaps based on an endpoint that may or may not have been previously doable.
As we heard earlier, there are very few approved products in this area. By products I mean biologics and drugs and devices. And in part this may well be because of problems that have not only perceived but real with the endpoint of a number of the trials.
One of the overriding issues that I would like the panel to consider in when we discuss these points is how close an association should there be between the endpoint and the agents Pharmacodynamics and Pharmacokinetics? In other words, if we have an agent that works, begins to work or has substantial working within 2 minutes and its half-life is an hour or 2 hours, is it reasonable to have an endpoint that is a long way away from that, say 30 days.
And the other things I would like the panel to consider as they view these various endpoints is that the endpoints must be rooted in reality. They need to be not only scientifically valid, but clinically meaningful, both qualitative and quantitative, that is as been eluded to earlier we might be able to see a decrease in bleeding, but how much of a decrease in bleeding makes it clinically meaningful to be important enough to have approval.
Is it logistically feasible? We need to setup endpoints that studies can actually be done. And in doing so, those studies must consume a reasonable amount of resources and not be unreasonable. By that I mean the cost in terms of money and personnel, both in absolute and in relative terms. By relative I mean many sponsors have multiple candidates that they wish to explore.
And ultimately they decide which candidates they are going to explore based on the cost and difficulty of doing trial A versus trial B and what they view their ultimate return of that. Trauma is a very difficult area and very expensive area in which to do research. And the more difficult an endpoint we setup, the more difficult those programs become.
Yeah, I think we like the other set of slides now that have been prepared by the FDA. Well, while those slides are getting, coming up, I like to start off by asking the panel -- okay. It is often said, and I think we heard it said more than once today that the treatment of bleeding is to stop the bleeding. Given that, how good an endpoint is hemostasis itself? Remember, we're talking here about bleeding, and I don't have any particular point as to who ought to start this conversation, but if there is no immediate volunteer I'll begin to pick on people.
Janice, you're closest. Dr. Zimmerman, say something.
DR. ZIMMERMAN: Well, I have a good excuse, I'm not a surgeon, I'm an intensivist. But listening to the comments that have been made so far today, this is one of the endpoints that I would say is very difficult to use as an objective endpoint. As people said is it when the surgeon says they have achieved hemostasis, is it when you stop transfusing, I think there is too much variability to use this as a primary endpoint for a systemic agent. I think we have better options.
DR. WEISKOPF: John, I know you're sitting next, and I know that you may have a slightly different view of that. Let's hear it.
DR. HESS: You know, I think stopping hemostasis in terms of when you stop giving blood, it's not a perfect endpoint, but if you are -- do it over four, five units it's a very real one because if you stop giving blood and the patient continues to bleed you will have --
DR. WEISKOPF: We'll get to transfusion in a bit, right now we're talking just hemostasis in and of itself.
DR. HESS: Yeah, you know, I think it can be detected clinically, at least in the operating room very well. And, you know, I think it's important. It is what we're trying to stop
DR. WEISKOPF: Rick, I saw you shaking your head, but I wasn't sure on what direction.
DR. DUTTON: That was a yes.
DR. WEISKOPF: Okay.
DR. DUTTON: Yes, I think hemostasis is a good endpoint. It is difficult to define but it can be done.
As an anesthesiologist I've stood around a lot of operating rooms watching a lot of patients bleed. And I can tell when they stop. They stop -- I know they have stopped bleeding because they stopped requiring fluid, they've become hemodynamically stable. I can't start anesthetizing them in the case of trauma patients. And that physiologic component combined with the surgical anatomic component of the surgeon is done trying to fix things. I think those two together are measurable and can define --
DR. WEISKOPF: Okay. Lena, recognizing that almost everything else we will discuss is in some way a surrogate for stopping bleeding what is your view about just the measurement in some way of hemostasis itself.
DR. NAPOLITANO: I agree that time to cessation of hemorrhage or hemostasis achievement is an important endpoint, but I would not consider it as a primary endpoint.
And I'll make the analogy of two patients, one has that grade 5 liver injury that Brian just showed where it's very difficult to achieve hemostasis even in the operating room and likely that patient maybe packed and may go to angiographic embolization versus a patient who has a, you know, grade 4 splenic injury you do a splenectomy the bleeding stops. They are very different patients.
So I think -- although I agree it's a good clinical endpoint, we don't have a lot of data out there.
Grant said he has some data that he is going to share at the East Meeting. I think it's not a primary endpoint.
DR. WEISKOPF: I'm going to push you a little bit on that --
DR. NAPOLITANO: Okay.
DR. WEISKOPF: -- because I didn't really hear you annunciate a satisfactory alternative if you're not going to, look, if you have an agent that purportedly provides hemostasis you're saying that looking at what's supposed to be the action of that agent isn't quite good enough.
DR. NAPOLITANO: Yeah, I agree.
So I would consider alternate endpoints. So maybe re-operation for bleeding, number of units of blood transfused and timed hemostasis, all those three might be good. But again just timed hemostasis, just using an agent, the systemic agent will not take care of grade 5 liver bleeding. I don't believe in totality. There are some surgical issues that have to be taken care of.
DR. WEISKOPF: Well, I didn't say specifically time to hemostasis, there are other measures such as the clinician, the surgeon says this is good enough for me to stop, I don't need to pay attention to this any more at this time.
DR. NAPOLITANO: I haven't thought of it in that way.
DR. WEISKOPF: Okay, this might be, Bryan, a good time for you to step in.
DR. COTTON: And again I think hemostasis is the ultimate endpoint. I think that's what we want, everyone of us, I think it keeps coming in all the talks. However, Ms. Napolitano has hit on what is the crux of it, is it's difficult, at least when I think of it, it's difficult to measure and quantify. You can, you know, dichotomize it like Dr. Dutton was saying, which is, you know, has the bleeding stopped, yes or no in the operating room.
But I think something a little more continuous might be a better endpoint such as, you know, when we've stopped transfusing X number of units over certain amount of hours or when we have a composite of that, which is you have the surgeon and anesthesiologist in agreement and you've stopped transfusing blood products, I mean, some level of that. But again, I think it's the aspiration of all of us to -- for hemostasis to be our endpoint. It's just kind of hard to get an agreement or consensus and quantify it.
DR. WEISKOPF: All right, let me see if I can summarize the rather diverse views of this group. I think I hear is that at least it's something that in and of itself it is a good endpoint, but many think that it is very difficult to assess in an accurate way clinically. But that -- maybe I'm putting -- now, I'm going to begin to put words in people's mouth. If one could find a good enough method of assessment, perhaps some of you might move a bit, move your opinion more to accepting this as a valid endpoint for a clinical trial. Is that a fair assessment of what people are vocalizing?
Okay. In that case, having solved that problem, we'll move on to the suggestion that -- and as I said earlier almost everything else that we're going to talk in this session really is a surrogate -- some other measure of some effect that hemostasis might have because of the difficulties that you just heard in perhaps assessing hemostasis. And the first that comes to mind is that of homodynamic stability or instability whichever side you want to look at it from.
In other words, perhaps providing a patient who is hemodynamically unstable because of lack of hemostasis and in some way measuring hemodynamic stability. I'm going to work from the other direction, Bryan why don't you start it off.
DR. COTTON: I think -- I'll just try to elude to or touch on little bit in slides, I think that vital signs are a little too dirty and a little bit one size done not fit all. Again, you know, 100 kilo, 6 foot 5 guy 70 years old got multiple vascular disease et cetera comes in and his blood pressure is 90, that's a completely different animal than a small 55-, 60-kilo young female who comes in who has got a blood pressure of 90.<.p>
I think there is -- you know, our whole models and our systems and our 90 or a 100 or whatever our physiology criteria is for activation of trauma is meant to over-triage people and capture all the people and not miss the injured population. And so I think, you know, when we go as low as 70 we're going to have a really tight model, we're going to miss some people, but if we have 110 we're going to include a lot of people.
So if we're going to over-triage with the vital signs we're going to have to accept that we're going to have to remove populations of patients, large populations from that study's analysis, I think --
DR. WEISKOPF: Well, somebody earlier today said something like that. And I think by and large that represents a hurdle and analysis that most statisticians would not be too willing to accept. So I think we need to look in another direction rather than a post-talk elimination of enrolled patients.
Lena, I think your opinion might be slightly different from Bryan, let's hear what you have to say.
DR. NAPOLITANO: So what I said earlier was I really do think that enrolling patients more tightly is a better thing because if we look at all the trials that I showed this morning where they enrolled patients with systolic blood pressures less than 90, 50 percent of those patients or more had not relevance to the study, they received no blood transfusion.
So if we just take the last trial that I reviewed, the ROC study, enrolled 895 patients, and in the end they got 135 patients that had massive transfusion that's a lot of money that went to enrolling a very small cohort of patients.
So I personally feel if we're going to go the route of enrolling patients based on physiologic criteria of systolic blood pressure less than 90 that then in the end you should consider excluding those patients from the final analysis --
DR. WEISKOPF: Lena, I'm going to interrupt you because you're not answering the question I asked.
DR. NAPOLITANO: Okay. I thought I was. I was following right with what Bryan said.
DR. WEISKOPF: The question is not whether we ought to exclude patients who didn't really meet the enrollment criteria but in fact are -- is hemodynamic stability a useful endpoint for severe bleeding.
DR. NAPOLITANO: Now, Bryan was talking about inclusion criteria. I just want to be clear --
DR. WEISKOPF: Yeah, he may have done off track, but you continued the train in that direction.
DR. NAPOLITANO: Emphatically no, so no.
DR. COTTON: I was supporting why I thought vitals were a crappy endpoint.
DR. NAPOLITANO: Yeah, emphatically no. You know, we have elderly patients on beta blockers so heart rate means nothing. We have young people who have very high heart rates and they are not in shock, it's just sympathetic tone, so emphatically no for hemodynamic parameters.
Resolution of shock by resolution of base deficit or a lactate, I think, is intriguing. I think there are some problems with it as Bryan mentioned in that. About a quarter of the patients don't have resolution of lactate or base deficit usually related to organ failure or mostly renal insufficiency.
DR. WEISKOPF: John. You're next, Rick.
DR. HESS: You know, I will accept the surgeon's comments on it. I'm really much less able to comment on that.
DR. WEISKOPF: Okay. Rick.
DR. DUTTON: I'll disagree violently with Lena which is our normal state of affairs. I can tell when the patient is hemodynamically stable -- and I think most anesthesiologists, most bed-side ED nurses, most ICU nurses live in this world. ATLS defines it as the transient responder. And the test is very simple.
You give them fluid, the blood pressure goes up and it either stays up or it doesn't, and if it doesn't and you have to give them more fluid then they're unstable. But if it stays up they're stable. And in real life that's how 90 percent of all of our resuscitation decisions are made, and I think that is valid.
DR. WEISKOPF: Janice.
DR. ZIMMERMAN: Well, I don't think hemodynamic stability or instability is very helpful as a endpoint for the various reasons that Lena suggest. And I deal with an older population and I would say things like blood pressure and heart rate are really not that discriminatory for many reasons. I have people walking around with blood pressures of 70. And again I think we go back to what is a good endpoint, it has to be relevant, and this may be relevant but the ease of measurement and the objectivity of measurement is very difficult.
DR. WEISKOPF: Well, I'll throw one more thing into the mix, and I'm surprised that I didn't hear anybody mention, and if they did I'm sorry, and that is many -- vital signs are relatively easy to manipulate by administration of various pharmaceuticals and that could easily complicate trial endpoint issues, does anybody have a view that differs from the one I just expressed.
DR. DUTTON: Well, I'm a chronic patient poisoner, yes, but I know what's supposed to happen when I give that drug. So if I induce Bryan's patient to intubate them in the emergency department and they die, that's a good sign that they are unstable, whereas if they don't die it's a much better sign about their physiology, and it is as I'm saying. This is part of how we know whether they're stable or not.
DR. WEISKOPF: Well, we'll get to counting body bags later. But in practical terms if one has a endpoint that says, just say hypothetically a blood pressure that has returned to 100 or more how does one evaluate that if somebody has given this patient some phenylephrine or epinephrine or some other vasoactive substance.
DR. DUTTON: I don't like pressors, so --
DR. DUTTON: So if you are giving pressors it almost certainly is a bad sign.
DR. WEISKOPF: Yeah, well, perhaps --
DR. DUTTON: -- so -- definition of instability.
DR. WEISKOPF: I would agree with you, but
nevertheless reality would say that that's going to occur in a clinical trial would it not?
DR. DUTTON: Sure. Yeah --
DR. WEISKOPF: Okay, well --
DR. COTTON: And I would follow up the whole vital sign manipulation, I mean, depending on the agents you use or again how many Neo sticks you have in your pocket, in the operating theater. And getting back to one of our other lecturers, you know, just as there is a lot of variability and talent and management of the patient on side of the ether screen there is altered, you know, lot of changes and variability on the other side as well. So, you know, if Rick is one side of the screen versus another colleague, you may have a lot of variability that can be manipulated with those pharmaceutical agents you're talking about.
DR. WEISKOPF: Okay. I think surprisingly enough, we're in more agreement on this topic than I thought we would have reached, although nobody mentioned things like various shock indices, whether those are viable endpoints. If anybody wants to make a quick stab at that, fine; if not we'll move along to the next topic. No takers?
Okay. Let's go to the issue of transfusion of which there are several types, several components as it were. There will be the -- and many trials as we've heard today, have used in one way or another transfusion as an endpoint, primary endpoint for -- primary efficacy endpoint for that trial. We could look at trials have at reduction of transfusion, whether that be one component or all blood components, complete avoidance.
But Bryan, as you carefully pointed out, that's really not possible in the terms of -- the kind of patients we're talking about, although one perhaps could construct a trial that would say avoidance after whatever intervention we're planning. That might be difficult given the patient population we're talking about. And the third would be avoidance of massive transfusion, again, either in toto or only after the intervention.
Rick, I'll let you start this one off.
DR. DUTTON: I do think that less blood is better than more. I don't think anybody would disagree that blood is bad for you, having a blood transplant is bad for your physiology, it's going to worsen your outcomes. The trick is how much and how do you define it and when does the difference become meaningful. One problem we've had in a lot of trauma studies is it's all well and good to measure a transfusion difference as in the factor 7 trials, but you have to know that all of your transfusion decisions are being made in a consistent fashion to the same protocol, and that's very difficult.
DR. WEISKOPF: John?
DR. HESS: I think it's a useful outcome, but as Rick and I have discovered, it's fairly gross and you have to be willing to look for differences that are quite large, three or four units at least.
DR. WEISKOPF: All right, let me approach this another way. Does anybody in the panel think that measurement of transfusion is an inappropriate surrogate measurement for alteration of bleeding? So everybody would agree in some way that transfusion -- measurement of transfusion somehow would be an appropriate endpoint. All right, then let me bring up the issue that John just brought up or asked for expansion by all, how much of a decrease of transfusion is enough to be meaningful?
John posited that something on the order of three or four units would -- is that what you said, John? Okay, three or four units. He'd like to see an endpoint that says a reduction of three or four units as opposed to some percent reduction. Bryan?
DR. COTTON: Yeah, I'm thinking along the order of three or four units as well looking at some of the historical data for what maneuvers in resuscitation of these patients have changed. And it looks to be -- the difference is around three to four units. I think the more important thing is to look at the risk of multi-organ failure, death and pneumonia, those type of things, and look where the inflection point might be as you go from X unit to Y unit in --
DR. WEISKOPF: Well, I'll point out to you as you well know that all those publications are retrospective analyses with many confounders. And I would hate if I were in industry, and planning a trial that would cost in tens of millions of dollars if not more, I would hate and would not power a trial or conduct a trial based on such data. We're -- you're asking industry for an awful lot of faith there.
DR. COTTON: And you're also having to ask and standardize the clinicians taking care of these patients, because there's tremendous clinician variability --
DR. WEISKOPF: Right.
DR. COTTON: -- and transfusion triggers. So even with some pretty darn good data out there about transfusion triggers, there's still a tremendous amount both in the operating theater and the ICU about clinician variability in transfusion.
DR. WEISKOPF: Janice, I haven't asked for your reviews in a while.
MS. ZIMMERMAN: Well, I would say that avoiding, you know, even two units of blood transfused maybe clinically relevant, and when you get to statistical analysis, that's another issue. But there are some prospective trials in the medical population, showing that the complications, particularly infectious complications increase with every unit of blood transfusion. Again, the surgical retrospective --
DR. WEISKOPF: Right.
MS. ZIMMERMAN: -- suggests the same thing, so. And again, you'll capture a lot of those complications as safety events, so you're kind of double-dipping there, but you know, I still think avoiding transfusion and maybe even avoiding two units of blood is clinically relevant.
DR. WEISKOPF: All right. John, I'm going to come back to you. Given that what Janice just said that for every unit there's increased risk, why are you looking for three or four units and not just one?
DR. HESS: Because of patient heterogeneity; when you look at broad bands of blood use, you know, five to nine versus 10 or more, you can actually see differences in reasonable sized patient cohorts. But if you try and band it more narrowly, it blurs.
DR. WEISKOPF: Isn't that what statistics are all about?
DR. HESS: Yes, but reasonable numbers are something we're going to have to live with. We're never going --
DR. WEISKOPF: So you're talking more about powering a trial rather than the medical issue?
DR. HESS: Absolutely.
DR. WEISKOPF: Okay. Lena?
DR. NAPOLITANO: Yeah, I agree that any reduction in blood transfusion is meaningful and particularly a reduction in massive transfusion is meaningful. Where I struggle with this issue is, is the reduction solely related to the systemic agent that you're studying, or is it as Bryan said, related to the practitioner variability and how they transfuse, or whether you have a good trauma surgeon or don't have a good trauma surgeon. So I think this particular clinical endpoint is confounded by a number of other variables.
DR. COTTON: So you know, getting back to that --
DR. WEISKOPF: Well, let me -- I have -- I'll be with you in a sec. Let me ask Lena to clarify what she's saying. I didn't quite get your bottom-line there, Lena. Given that as confounded, are you saying it is or it is not a good endpoint?
DR. NAPOLITANO: It is good endpoint, it is. I just accept that it's confounded by these other variables. DR. WEISKOPF: Okay. Bryan?
DR. COTTON: And getting back to Dr. Hess's point, it's a very relevant discussion about the sample size or the power in the four-unit range versus one-unit or two-unit range. I mean, we just finished calculating sample size for one population in our institution and a 4- to 5-unit reduction in a patient population getting anywhere from 6 to 10 units; 4- to 5-unit reduction, we're talking about a 150 patients. DR. WEISKOPF: Yeah.
DR. COTTON: So I mean, once you start getting in the two range, you're talking about an exponential increase --
DR. WEISKOPF: Yeah. Yeah, those are certainly important considerations as I pointed out in introducing this session, but John let me come back to you one last time and then we'll move on. Suppose somebody in fact wanted to do a trial they could run a trial with sufficient power to what they think is assess a one-unit difference. Would you think that would be acceptable clinically, that you would accept that as a valid endpoint?
DR. HESS: No, I mean, just the variability and the size of people makes a one unit transfusion almost meaningless. You know, we have 100-pound people and we have 250-pound people, and a one-unit difference in them means something very different.
DR. WEISKOPF: Well, doesn't that get us back to saying that gee, maybe this surrogate is not as good as hemostasis itself?
DR. HESS: I agree.
DR. WEISKOPF: Okay. I think we'll terminate this particular topic. And by the way, I'm not sure how much time we have since we started this session late. Fifteen more? Okay, thank you. All right, what I want to do is move on -- and I've saved this for not the end, but to -- after get everybody warmed up a little bit. The issue of mortality that many trials have been focused on mortality as the endpoint. And mortality, by and large, has been a substantial distance in terms of time from the hemorrhagic incident.
The mortality that has generally been looked at is something like 28 or 30 days and as we've heard repetitively today from multiple datasets is that for at least the type of patient we're talking about that the mortality is essentially all over, certainly within the first 24 hours and by and large many hours before that. It's all over except for the shouting.
So I'd like to press the panel about two issues. One, is in fact mortality an appropriate endpoint period. Independent of what time it's assessed at, is it appropriate for an agent, whatever that agent might be, that mechanism of action is to either decrease or stop bleeding, that presumably in patients such as this, there are many potential causes of mortality? And do we not increase the noise to a point that makes it far more difficult to really assess the mechanism of action and the efficacy of the agent, whether it be a drug or biologic device by using mortality rather than something closer to the pharmacologic action.
And I'm not sure who wants to take this on first. I'll take a volunteer if not all -- all right, Janice, you had your hand up first.
DR. ZIMMERMAN: Well, Lena and I were debating this a little bit during the break. And I've heard all the arguments about 48-hour mortality. I will take the position that I still think 28 or 30 ---
DR. WEISKOPF: Well, wait, I'm going to stop you for a sec, because the question I want answered first --
DR. ZIMMERMAN: Okay.
DR. WEISKOPF: -- is mortality an appropriate endpoint --
DR. ZIMMERMAN: Yes.
DR. WEISKOPF: -- for an agent that's -- action is to alter hemostase -- alter bleeding in a favorable way?
DR. ZIMMERMAN: Yes. My rationale here is that I think our health care system demands clinically relevant outcome. Stopping bleeping is good, but if it doesn't translate into something that's relevant to the patient such as a good outcome, whatever that might be defined as such as mortality, I sill think mortality is a good --
DR. WEISKOPF: Okay, Bryan, are you going to take issue with Janice's statement that -- or implication that stopping bleeding is not good for the -- is not sufficiently good for the patient?
DR. ZIMMERMAN: That's exactly what I said.
DR. COTTON: I think bleeding cessation is absolutely good for the patient. And you know, hemorrhage is the number one cause of death in the OR, it's the number one cause of early death. If we can reduce hemorrhage-related mortality and that 6-hour mortality whatever it might be, I think that's it. I really don't -- not say again that I don't care; I always care about my patients. But 28-30 day mortality, good --
DR. WEISKOPF: Well, I'm going to stop you again because we're going to talk about the time in just a moment.
DR. COTTON: Yeah.
DR. WEISKOPF: But you think mortality is an appropriate endpoint --
DR. COTTON: Yes.
DR. WEISKOPF: -- for a hemostatic agent.
SPEAKER: A-mortality (phonetic) --
DR. WEISKOPF: A-mortality, someplace independent of time, we'll come to that. Rick?
DR. DUTTON: Yes.
SPEAKER: No --
DR. WEISKOPF: Okay. Good one-word answer. Thank you. Lena?
DR. NAPOLITANO: No. No.
DR. WEISKOPF: Why not?
DR. NAPOLITANO: I think Bryan said it, but he didn't, you know, point out hemorrhage-related mortality is a good endpoint. All-cause mortality is not a good endpoint. So if you die of traumatic brain injury, it has nothing to do with the systemic agent that you tested.
DR. WEISKOPF: So you would, aside from having say exclusion criteria for TBI for this sort -- these sort of patients, you would seek to exclude them in the analysis. You would go for a modified intensive treatment analysis, or something akin to that, if that passed muster. John?
DR. HESS: Yes.
DR. WEISKOPF: Okay. So every -- almost everybody thinks that mortality in and of itself is a valid endpoint. And when I -- I don't know if we're ever going to get to rating these endpoints as to which -- well, maybe we'll have time at the end. But now, let's talk about the thorny issue of when that mortality ought to be assessed. As I said earlier, traditionally many of the studies have looked at 30-day mortality. We've heard a lot of evidence to say -- to at least point in the direction that that may not be appropriate. And let's start with you, Bryan.
DR. COTTON: I think mortality --
DR. WEISKOPF: Pick a time. Give me a one-clause answer that says if I'm running a study, this is the time I want to assess mortality for an issue that has to do with severe bleeding.
DR. COTTON: Six.
DR. WEISKOPF: Hours? And we can come back to rationale in a bit. Rick?
DR. DUTTON: Twelve hours.
DR. WEISKOPF: Twelve hours.
DR. NAPOLITANO: Twenty four.
SPEAKER: (Off mic)
DR. WEISKOPF: For a while, I thought we were in a bidding contest here.
DR. ZIMMERMAN: I would still go for a longer mortality, 28 or 30 days.
DR. WEISKOPF: Okay. I think that probably reflects your intensive care background rather than operating room background, or do you want to defend that a bit?
DR. ZIMMERMAN: No, actually I will because just getting a patient out the OR is not sufficient. And I know sometimes that's adequate, you know, stop the bleeding, get him to the ICU, but if I can't get him out of the ICU back to home, what have we done? We've saved them for 24-48 hours, but they still die.
DR. WEISKOPF: All right, let me -- so everybody with the exception of Dr. Zimmerman thinks that sometime within the first 24 hours is an appropriate time for an endpoint. Let me raise this to you, Janice, and see how you respond to this that if we save patients, presumably the -- whatever we're doing, whatever the intervention is, it works and that there's efficacy within the first 6, 12, 24 hours whatever it is. We then have a cohort of patients that would have died within that period of time who now come to your care in the ICU for a period of time. And they then -- they obviously are the sickest of the bunch because they would have died if not for the intervention.
And they therefore are probably subject or more likely to get a variety of complications and issues that you might not have otherwise seen, and to which we don't have currently available good therapies. And so those patients might eventually die. But without getting to that point, we have no ability to do further research or conduct or arrive at other interventions. How would you respond to that critique?
DR. ZIMMERMAN: Well, I think that's one possibility. I think the other possibility -- and this is I think the unknown, which is if you stop the bleeding early, I would think theoretically, they should have less complications of, you know, massive transfusion inflammatory response, potentially. And I'll just go back a little bit to like one of the factor 7 trials, the original trauma one. They actually excluded deaths in the first 48 hours because they thought that there was no opportunity to save someone who's going to die the first 48 hours with factor 7. So we're kind of saying both things. My hypothesis would be is if you've done something good and saved them for 48 hours, why are they not going to live? And again, your hypothesis is they're more complicated, but in reality maybe that's not --
DR. WEISKOPF: Well, without getting into details of the effect, the phase 2 factor 7 trial, the inclusion/exclusion criteria were such that patients were enrolled who were in fact darn near dead, if not actually physically dead.
DR. ZIMMERMAN: Hope not.
DR. WEISKOPF: So the panel -- we'll see there what two, four, five panelists and if I include myself, that's six, I think at least for the audience, five of the six of us think that the endpoint, if it is to be mortality, ought to be somewhere within the first 24 hours, and not further on with a -- a rather well-reasoned exception to go to something much further on like what has been more traditional 28 or 30 days.
All right, let's move to the next which are perhaps somewhat even more difficult to deal with. And let's take the next logical consequence, if not mortality, morbidity. And I'll assume for the purpose of this discussion that the morbidity has to be severe, not some minor morbidity.
So first there were two questions here, is morbidity a reasonable endpoint? If so, how severe should it be to be allowed to be an endpoint? I'll start in the middle this time. Lena, you're smack in the middle.
MS. NAPOLITANO: Yeah, I have a problem with morbidity as an endpoint because so much of our clinical practice that is very heterogeneous impacts on this. So you know, need for mechanical ventilation, organ failure, is dependent on adequacy of resuscitation. Prevention of infections is dependent on what you do in the ICU. So I really am not in favor of these as a primary clinical endpoint, would much favor mortality.
DR. WEISKOPF: John?
DR. HESS: Same.
DR. WEISKOPF: Okay, Rick.
DR. DUTTON: I think if you're going to use a surrogate endpoint for mortality, it makes more sense to go back to transfusion.
DR. WEISKOPF: Bryan?
DR. COTTON: One-word answer or more than one word?
DR. WEISKOPF: I'll give you up to sentences.
DR. COTTON: It's the survivor's curse, the complications that occur, and I really don't care with respect to a hemorrhage-related issue. We can deal and come up with a solution and a cure for the survivor's curses, multi-organ failure, things like that later
DR. WEISKOPF: Okay. Janice?
DR. ZIMMERMAN: I would say morbidity is not a good primary endpoint..
DR. WEISKOPF: Okay, I assume you've each been talking about single-organ morbidity. If not, how about multiple-organ morbidity? And while they're thinking about their answers till I pick on them, I haven't seen any cards coming in. Please, this is meant to stimulate thought and questions in the audience. So please do bring some cards forth. Okay, Rick you want to talk about multiple organ morbidity or failure?
DR. DUTTON: I become less enthusiastic, especially when you look at this population of people who are bleeding to death. The organ failure they have is almost entirely pulmonary, so it's not -- your composite score isn't any different than your ventilator-free days --
DR. WEISKOPF: Okay.
DR. DUTTON: -- or single system --
DR. WEISKOPF: Right, as we've seen on the figures that say that third hump that Don Trunkey first showed in that Scientific American article from San Francisco General a long time ago, that third hump has basically disappeared, with -- probably with the exception of pulmonary failure.
Rick, so why would you not accept pulmonary failure as some part of your mix here?
DR. DUTTON: Again, because it's more removed from the proximate endpoint, death at 12 hours, or getting to the ICU in the first place. And I think if you're going to look for a surrogate, as I say, transfusion is the much more immediate one.
DR. WEISKOPF: Fair enough. Lena, you're shaking your head.
DR. NAPOLITANO: I agree fully, and particularly I would not include ALI and ARDS because they are impacted on by a degree of transfusion, they're impacted on about -- by a ventilator -- mechanical ventilator management, and so I disagree with that.
DR. WEISKOPF: Janice?
DR. ZIMMERMAN: I would agree.
DR. WEISKOPF: John?
DR. HESS: Very much so.
DR. WEISKOPF: And Bryan?
DR. COTTON: Agreed.
DR. WEISKOPF: Okay, I have question here which impacts on the discussion we've just been having about mortality and morbidity. Do people think that there is a link from the 24-hour, say, mortality to longer mortality, more distance say 28 days the kind -- what Janice would accept or is there sufficient disconnect that makes -- that pushes you to the 24-or-less hour mortality endpoint?
SPEAKER: It gets back to something like -- you know, they tried to teach me in fellowship which was the longer you stay in the ICU, the longer you stay. And you just -- your exposure days, exposure time is tremendous. So every day you're in there, you're at risk of a resident hurting you or a bug hurting you or something hurting you. So the only way to fix that is to die quickly and not get exposed. So if you live to fight another day with whatever intervention you have, again, I think you worry about curing whatever that morbidity is later.
DR. WEISKOPF: Right. Somebody from the audience asked a question that I will sort of translate into a more broad issue. The issue is if we look at 20 -- say somewhere within 24 hours as an efficacy endpoint, well, what about if we in some way or other alter 28-or 30-day mortality in an adverse way? My view, not to poison the panel, but I'll just give my view for openers is that that is appropriately captured as a safety endpoint. And certainly in any trial that I can conceive of would be captured in a safety endpoint and nobody would want to ultimately be having an adverse impact on mortality. But let's see, other people have different views here.
SPEAKER: I agree with that. In a human clinical trial, you're going to capture all of this data. So you'll always have it available to look at. But if something we do makes you survive in the first 24 hours, but die in the first 30 days, you still have 29 days to figure out a way to prevent that, and you're still ahead.
DR. WEISKOPF: Does anybody have a differing view rather than ask everybody for the same repetitive answer? Okay. Let's move on to an area that was touched upon briefly by other speakers and that is composite endpoints. We've talked about several different types of endpoints here. Does somebody -- would somebody like to promote the concept that more than one of these can be rolled into a composite endpoint? Well, everybody's sitting back in their chair.
SPEAKER: I think you water it down. And I'll listen to what Dr. Napolitano says, but I think you water it down and you lose your signal with did the bleeding stop, yes or no, are they alive, yes or no, and I think you lose with the composite.
DR. NAPOLITANO: Yeah, I would just want to bring up it's been done in other trials. There was a factor 7 trial in cardiac surgery that used a composite endpoint of blood transfusion and another intervention for bleeding, a second intervention for bleeding. And I think we all have trouble identifying if we accept that data is real or not and if we're going to change clinical practice based on that, and it's challenging.
DR. WEISKOPF: Anybody like to express an alternate view? Okay, let me then move to what perhaps is even more difficult issue, and that is we've been talking essentially -- except for when we talked about hemostasis, we've been talking about surrogates. But now we're going into some -- although maybe most people wouldn't recognize those things that we talked about as surrogates, in fact they are surrogates for hemostasis. But now, if we look at what perhaps people might ordinarily classify as surrogates, things such as length of stay in an ICU, length of stay in a hospital, markers of organ function, such as perhaps lactate, various creatinine clearances, things of that nature, timed to another important intervention. I'd like to hear the panel's view about these surrogates, one on mass, whether it is appropriate at all to look at any of these surrogates and if so which would be the best?
SPEAKER: These make wonderful secondary endpoints.
DR. ZIMMERMAN: And I'll also say that length of stay data is very -- looked upon favorably if it's decreased by administrative people. So cost issues come into play. But length of stay in an ICU in a hospital bed is so variable depending on, you know, who makes the decision to transfer out, are there beds available. So I don't think it's a great primary but again it sometimes might help sell something to an administrative; it decreases cost of care.
SPEAKER: To pick up on Bryan's point, if my primary endpoint is decreased ICU length of stay, it might affect which resident I have take care of the patient. (Laughter)
DR. WEISKOPF: John? No, okay.
DR. NAPOLITANO: Just one comment, it has no relevance to the systemic agent unless it met the other primary endpoint, which is reducing hemorrhage and reducing mortality.
DR. WEISKOPF: I'm sorry, Lena, I didn't catch you. Say that again.
DR. NAPOLITANO: It -- decreasing length of stay really has no relevance to a systemic agent trial unless it is related to reducing bleeding, reducing hemorrhage, and reducing mortality.
DR. WEISKOPF: Okay. I could expand that argument, but I don't want to backtrack and open up things that we already seemingly have decided. But this might be a good time to take a couple of questions from the audience. Dr. Moore (phonetic), please. You're going to defer? Okay.
SPEAKER: I'd like to sort of turn this mortality thing, not -- maybe not upside down, but at least a different perspective. And in the context I would say a potent -- pro -- systemic procoagulant you could use factor 7 A or some drug that might be analogous to it. You know, you take an out-bred group of humans and hurt them in a little of different ways that are completely random, and then try to treat them at different times that are random, and then try and test a very potent systemic procoagulant drug and ask the question does it affect the outcome, and you chose the outcome to be obtusely related like did they die.
I can't see a way that trial could ever be successful. And it isn't that the drug doesn't work or it isn't that the drug isn't very potent, but it seems like the logical endpoint would be freedom from hemorrhage at a given time after the event it was tested. I mean, did they die of head injury --
SPEAKER: -- anyway 12 or 24 days later, it's freedom from hemorrhage at 12 or 24 hours and then you look at mortality as the adverse events. And if mortality stacks up in the group that got treated, then you'd say, well, we're doing some bad of these people.
DR. WEISKOPF: We only have a couple of minutes left, so I'm going to ask --
SPEAKER: But so the --
DR. WEISKOPF: -- so I'm going to ask comments to really be brief.
SPEAKER: The choice is, is it freedom hemorrhage as the endpoint?
DR. WEISKOPF: Well, I think we've covered that earlier on. Dr. Moore?
DR. MOORE: Well, I think just for a matter of discussion, I'm surprised that you didn't have D (phonetic) up there, and that is laboratory assessment of coagulation function. Understanding all the controversies, full trauma time, partial trauma time, we're now getting to viscoelastic properties that probably reflect more what we see clinically.
I would argue some of these tests, and they may become more advanced as we proceed on technologically, would be at least a composite circuit.
DR. WEISKOPF: Yeah, okay. I'll take blame for that because that sort of categorizes the surrogates and I didn't explain perhaps that carefully well enough. But let's pose that question to the panel. What does the panel think about surrogate of four hemostasis, a non-clinical laboratory test as being an adequate endpoint?
DR. MOORE: I think it's a great endpoint for your phase 2 trial. In fact I think it's a necessary endpoint for your phase 2 trial, but you have to demonstrate physiologic efficacy. But I don't think it's appropriate to make a clinical decision on -- from your phase 3 trial or your licensing trial.
DR. WEISKOPF: Anybody have a different response? Brief question, please.
DR. RHEE: Peter Rhee from Tucson again. I think if you're going to test something, you ought to see if that worked, meaning that you ought to look at a direct outcome. So for example, it depends on what you're talking about, whether you're talking about FFP or factor 7. If you're giving factor 7 to reduce coagulopathy, then you need to measure whether it did that. While you would like it to do a 28-day mortality, that's always looking for the Holy Grail, or the Bomb for a touchdown every time. But sometimes, you have something that's designed for five yards and not always a touchdown.
DR. WEISKOPF: Okay. Brief question or comment please.
SPEAKER: Yeah, very brief comment. It relates to the lexicon of surrogate endpoint and I think the FDA described it as something that is unachievable. It requires phase 3 clinical trials from multiple sources and compounds. You may want to think about a terminology of disease-modifying biomarker, and if the biomarker is not disease modifying, the chances that it will have any clinically meaningful outcome is very slim. So I think --
DR. WEISKOPF: Yeah, so --
SPEAKER: -- that's what I just want to comment.
DR. WEISKOPF: Well, that gets into an entire additional conversation of validation of biomarkers and surrogates for which we only had 40 minutes here and couldn't cover that. Steve, one more comment?
SPEAKER: Dick, I just want to push a little -- if everybody's accepting mortality, what you didn't get at is what is the magnitude, because that affects --
DR. WEISKOPF: Fair enough, that's a good question.
SPEAKER: Hovering -- logistics cost everything else, and I think you've got to at least spend a couple of moments on that.
DR. WEISKOPF: I think you make a very good point, and let's go to the panel. Let me put this in some perspective. As was pointed out, people are talking about the crash 2 trial, which I think is an extremely flawed trial, and published with a "Mortality difference of 1.5 percent." Would anybody prospectively power a trial for that or people accept that as satisfactory endpoint? Just quick answers, we're running out of time.
SPEAKER: No. No.
DR. WEISKOPF: Okay. Let's have one more answer and that is what percent reduction in mortality would you view as clinically meaningful? Bryan? Just come down the line again.
DR. COTTON: Five.
MS. NAPOLITANO: Five.
MS. ZIMMERMAN: Ten.
DR. WEISKOPF: Okay, we've run out of time and I thank the panelists and Dr. Cotton for his presentation. There's one other thing I'd like to say that in our audience today is Dr. Ulla Hedner. Most of you probably may not know who she is. She is the one who really single-handedly developed -- activated factor 7 and Ulla, would you stand up, please?
DR. WEISKOPF: Thank you. Oh, yeah, okay, sure. SPEAKER: We'll have a 10-minute break and reconvene in about 10 minutes. So please be back here promptly at 25 minutes after 4:00.
(Discussion off the record)
SPEAKER: Just a reminder that the FDA reps along with the moderators, we will be meeting at the back of the auditorium at the end of today's conference to proceed over to Building 29. (Discussion off the record)
ETHICAL CONSIDERATIONS OF CLINICAL TRIALS WITH TRAUMA
SPEAKER: Okay, our last session today is going to be on Ethical Considerations in Trauma, and our speaker is Dr. Sara Goldkind, senior bioethicist in the Office of Good Clinical Practice at the FDA. She's going to be talking about submissions to FDA under 21 CFR 50.24, highlights for sponsors and investigators.
DR. GOLDKIND: We're almost at the end. And I want to thank the Office of Blood in CBER for thinking very early on about ethics so that the planning committee could work diligently and thoughtfully to put this panel together. And what I'd like to do, in my brief remarks, is to be very simple and very didactic when I talk about the FDA regulations 50.24, and that really is the regulation that governs clinical research in the emergency setting with the exception from informed consent. And I want to put this regulation into perspective, and explain that it's the only one that exists under FDA regulations for the purposes of research.
In other words, FDA takes very seriously the idea of doing a clinical trial and enrolling a person into clinical research without obtaining informed consent. The only other existing regulation that is sometimes confused with this one is a regulation that allows emergency treatment of a patient with an investigational agent for the purposes of life saving therapy. That is not a research regulation, that's a regulation of providing an investigational product to someone who is in a life-threatening circumstance for lifesaving measures.
What we're doing here with this regulation is we are providing an investigational product, holding out the prospect of direct benefit. So we are trying to save life. But we're also trying to accumulate rigorous data to better inform pharmacotherapeutics and labeling.
Now, the other thing I want to mention is that I'm hitting on highlights. This is -- 50.24 is a very complicated regulation. We are about to issue, hopefully sooner -- soon guidance in this area; this will be our third guidance. It's quite complicated; we've evolved in our understanding over time since the regulation was adopted in 1996. And what I'm hitting on are some take-home points, I think, for you all and some clarifications of common misconceptions.
So I'm going to look at four different areas, study design which we've talked about extensively today. So I'm going to touch upon it from an ethical perspective as well as within the context of 50.24. I'm going to look at informed consent considerations, which you may think is a misnomer considering that this is an exception from informed consent, but we'll clarify that later; additional protections that are part of this regulation and some regulatory considerations.
Now, when I say additional protections, one of the things that I'd like to clarify is that FDA spent a lot of time thinking about how do we develop and meet this unmet need for developing rigorously studied scientific information on efficacy and safety for emergency treatments, severe blood loss. And our moderator Dr. Biros was one of the people who was in the frontlines bringing together consensus panels to get public input on what would a regulation look like that did not include informed consent, which as you know is one of the fundamental protections for human subjects dating back to 1947 and the Nuremberg Code.
And the regulation really tries to balance this issue of self determination and the principle of autonomy with the principle of beneficence or doing good to someone that we frequently see that same kind of balance in the emergency room setting on a clinical level and here we're talking about it in a research context.
And so when we don't have informed consent from the person directly, we've included in this regulation what we think are additional protections that can help assure the right safety and welfare of the subjects enrolled in the clinical trial and we'll talk a little bit about them later.
So my first and the bulk of my discussion is going to focus on study design. And from the get-go, I will say that this is the most, as you've been struggling today and you probably know, this is the central component of the ethics of these trials is getting a scientifically valid and ethically acceptable study designed.
And a part of that is considering whether or not you can do the research and obtain reasonably important scientific information that might address the question from a consenting population. That might not be the case in the trials that we've discussed, but I want you to think about sort of an algorithmic approach to using 50.24. And the first step is to think about can the necessary scientific information be obtained in a consenting population and how would that affect scientific validity and how would it affect the generalisability of results? Can the risks be minimized by studying a less sick population?
And if you do determine that it's necessary to study a non-consenting population, then you have to adequately justify that in the protocol and why you cannot obtain the same information from a consenting population as well I'll mention that the protocol has to justify the therapeutic window which is an important consideration when thinking about the feasibility of obtaining informed consent.
And I would also urge that when you're designing trials of this sort, or any trial really, that you plan ahead and you consider all aspects of the trial upfront, do you want to obtain biomarkers, do you want to do an extra blood draw to get that sample to do a later stage biomarker assay, build it into the trial upfront.
So in order to do a -- to conduct a study in with the exception from informed consent under 50.24, you have to have -- the subjects have to be in a life-threatening situation. They need not be immediately life threatening or immediately near death, but death is likely unless the course of the disease is interrupted and an intervention is required before consent is feasible.
Available treatments have to be unproven or unsatisfactory. So if you're going to have a controlled trial you are in a state of clinical equipoise, you don't know which arm will be better or worse for the subjects who are randomized. And we define unproven as a lack of substantial evidence that a treatment is effective for the condition of interest.
An unsatisfactory harkens back to what the drawbacks are to the treatment. Are there safety concerns, is there a poor survival rate, is it only partially effective, does it take too long to work, does the treatment have administration issues such as is it IV and it's meant to be used in the field?
So those things have to be taken into account when you assess whether you can even go forward in this very preliminary way with the study under the exception from informed consent. And then we'll talk next about the fact that the intervention must hold out the prospect of direct benefit.
And this goes back to the idea, as I mentioned to you, that we incorporate the notion of beneficence into this regulation. And we define -- and what I've done is I have taken the regulation verbatim onto this slide, but I will talk about it a little bit more fully. The subjects are in a life-threatening situation as I mentioned that necessitates intervention and based on the information from appropriate animal and other pre-clinical studies they support the potential for the intervention to provide a direct benefit to the individual subjects.
So when we say appropriate animal or other preclinical studies, again there's a scientific -- these are scientific considerations that go into whether the animal model is appropriate for the -- for -- to look at humans and can be extrapolated to the human setting, whether there are other clinical data that would be relevant perhaps information on bleeding and consenting elective surgical patients, would that be information that can be translated to this more critically ill population with other confounding variables.
And also when we think about the prospect of direct benefit, we have to note that the risks associated with the investigation are reasonable in relation to what is known about the medical condition of the potential class of subjects, the risks and the benefits of standard therapy, if any, and what is known about the risks and benefits of the proposed intervention or activity. So all of those have to be considered and addressed, and I again submit that they should be addressed upfront in the protocol when it's submitted.
So again consider we've talked a lot today about endpoints. For some reason, there has been confusion about whether or not you can have morbidity endpoints under 50.24 research. And the answer is yes, they may be acceptable. What we've always said is that the severe morbidity is clinically relevant and closely associated with the mortality.
And again, there are a lot of scientific judgments that are involved in assessing the ability to use morbidity endpoints. We've heard many erudite discussions today about the merits of certain endpoints over others in terms of morbidity endpoints, but I wanted to clarify that morbidity as well as mortality endpoints are acceptable under 50.24, but they would need to be carefully justified and adjudicated.
So how about study design? Well, you can have an active control trial, as I mentioned earlier, a non-inferiority trial or placebo-control trial, all of those would potentially be acceptable under 50.24. Dr. Scott talked a little bit about non inferiority trials, and I'm certainly not going to discuss them more than to say that they're acceptable with the appropriate scientific support that would be necessary for any non-inferiority trial to be interpretable.
And that -- and I would not only recommend FDA's guidance in this regard, but also ICH E-10 discusses choice of control group and it discusses scientific and ethical considerations that I think that you should all attend to if you haven't.
Now when we talk about placebo-control trials being acceptable, the vast majority of the time if there is standard of care, we're assuming that it would be an add-on placebo-control trial, an add-on to whatever standard of care is unless you're really -- the trial is trying to determine whether the standard of care is in fact useful and other aspects about standard of care itself.
And then finally in terms of study design, there's been a lot of discussion in the past about whether or not the phase -- certain phases of a clinical trial are allowed or not allowed under 50.24. And neither the preamble to the proposed rule nor the preamble to the final rule nor the rule itself is there a discussion about phases of a clinical trial. What is emphasized, as I mentioned before, is the notion that -- is the requirement that the intervention itself hold out the prospect of direct benefit. Having said that, though, we would have a hard time imagining when a PK study could be done under -- could meet the requirements of 50.24 unless perhaps it's a nested PK study in a phase 3 trial.
And generally we would expect that preliminary information that's needed for the design and to better inform the design of a phase 3 trial would be obtained in consenting subjects. But we don't rule out the possibility that a phase 2 trial could be under -- done under 50.24 and that would require a lot of documentation by the sponsor investigator in terms -- to support that proposal and careful early discussions with the appropriate FDA review division.
And with that I want to emphasize and I think it's interesting that both the statistician and I have some of the same conclusions and that is because I think that you can't have an ethical trial if you don't have a scientifically valid trial. I want to emphasize that FDA invites you all to come to the appropriate review division for pre-IND or pre-IDE meetings so that you can consider protocol development early and you can -- and the FDA can comment on the adequacy of the IND or IDE submission, and it provides an opportunity for the FDA and the sponsor to discuss scientific and ethical considerations of the trial.
So now I want to spend a little bit of time focusing on informed consent. And this regulation -- and I want to underscore this -- even though the regulation is entitled exception from informed consent for the -- for emergency research, the regulation requires that an informed consent document be drafted, that there be a description of the informed consent process in the protocol. And there are a few reasons for this.
Now, again the regulation does recognize that the subjects cannot routinely be consented and that treatment would have to start before consent were obtainable from a legally authorized representative or from the subject. And that there's no reasonable way to identify likely subjects prospectively and we can talk about that a little bit more.
But you need to include the informed consent document because there may be, even though you don't expect it, the possibility of consenting the subject or the subject's legally authorized representative. Perhaps not every single subject who's enrolled in the clinical trial could not provide informed consent. And this is sort of a way of again underscoring how important we take informed consent. We also expect that the informed consent document can be used for informational purposes when later informing the legally authorized representative or the family members about the trial. And we expect that it's going to be submitted to FDA for review.
Now, I want to clarify another confusion related to 50.24 and that is this notion of not informing the legally authorized representative or the family member at the earliest opportunity. We have seen some trials where they -- there is the idea that you can justify not providing information about research enrolment early on at the earliest feasible opportunity because the family members may be emotionally distraught and we do not think that that's justification for holding off on providing this important information.
We ask family members all the time to make decisions, for example, in the emergency room, and the surgical waiting room, and the ICU, et cetera, to make decisions when they may be emotionally distraught and we don't think that that's a reason that there should be postponement in early informing in this setting. So some of you may or may not know about community consultation and public disclosure, but it's two of the additional protections that are built into this regulation to help bring in additional viewpoints about the trial in the case of community consultation and to help inform the community about the trial in the case of public disclosure.
By the time you get to the point of community consultation and public disclosure, you have to have worked out all the other scientific considerations and ethical concerns related to the trial design.
This is sort of akin to going out into the community and meeting with representatives of the communities where the studies will be done and from which the subjects will be drawn. It's a two-way dialogue. It's meant to provide the sponsors and investigators and the institutional review boards with the community reactions to this research and we expect that the plans for community consultation or at least we would like to see the plans submitted to FDA.
Now, the additional protections also includes, as I mentioned, public disclosure and there are two components to public disclosure. One is it occurs prior to the start of the study and the second is expected to occur after the completion of the study. And when it occurs prior to the start of the study, it should include the plans for the investigation, the risks, and the expected benefits at a minimum and, as I said, we describe additional recommendations in guidance. After this study is completed, it should -- public disclosure should be done to apprise the community and researchers of the study including the demographic characteristics of the research population and the study's results.
It's meant to be a one-way communication. The information can be posted in a newspaper article, can be described over a radio announcement, et cetera. One additional point of clarification is that as sponsors you're required to promptly submit the materials that you use for public disclosure to the FDA docket and to the IND or IDE file, and if you don't then you're not in compliance with FDA regulations. This seems to be a point that has not been well-absorbed by the sponsor -- by sponsors, but we do expect submission of this information.
And then lastly, I want to just explain to you that these are highly regulated trials. We've had about 77 submissions to the FDA. Under this exception, we at FDA have reviewed them under the appropriate regulations. There are informed consent regulations, institutional review board regulations, if they involve pediatrics under subpart D and under the applicable IND or IDE regulations.
And if they don't meet the regulatory requirements we don't let them go forward. They may fall short of meeting 50.24 requirements or they may not present a complete application. They may not have a statistical plan. They may not -- they may fall short scientifically. And so we've had -- we've granted about 44 trials to go forward under 50.24. Of those we assume that there are less that have gone forward because they have to then -- they're reviewed by institutional review boards and the IRB may or may not approve the trial to go forward.
And I wanted to just describe for you since I know that we've gotten questions about why is it that FDA might not let a trial go forward under 50.24, so I'm going to highlight a couple of areas of our regulations, the clinical hold for drug studies and disapproval or withdrawal of approval for device studies. And some of the reasons we might not let a 50.24 study go forward is that human subjects are or would be exposed to an unreasonable and significant risk of illness or injury, the plan or the protocol for the investigation is clearly deficient in design to meet its stated objectives or it might not satisfy the pertinent criteria of 50.24.
The device regulations largely mirror the drug regulations and they talk about failure to comply with the investigational device exemption regulations. And the fact that there might be a reason to believe that the risks to subjects are not outweighed by the anticipated benefits to the subjects and the importance of the knowledge to be gained, or informed consent is inadequate, or the investigation is scientifically unsound, or there's reason to believe that the device as used would be ineffective.
And then finally I want to let you know that sponsors, investigators, the FDA, IRBs, were all in a collaboration to carefully scrutinize the conduct of trials that involve the exception from informed consent. FDA does conduct inspections of IRBs and clinical investigators that are reviewing or conducting 50.24 studies, and we expect all parties involved to be knowledgeable about the requirements, the ethical and regulatory requirements for the conduct of those studies. And then finally, we expect that the investigators and institutions where these studies are going to be done will have well-developed internal standard operating procedures that show compliance with the regulations. And then if you need help and we invite you to come to us for help; if you know the appropriate review division that's the place to start, but if you don't I've listed contract -- contact information for CDER, CBER, CDRH, and my office. Thank you.
SPEAKER: I'll invite the panel members to come up please and the moderator for this session is Dr. Michelle Biros, professor of emergency medicine and vice chair of research, Department of Emergency Medicine at the University of Minnesota.
PANEL DISCUSSION: ENDPOINTS FOR CLINICAL EVALUATION OF
PRODUCTS FOR BLEEDING INTERVENTIONS SYSTEMIC AGENTS
DR. BIROS: Thank you. While we're waiting for the panelists to come forward, I just want to express my sincere appreciation for the opportunity to discuss this really important issue related to the trials that you will all be involved with. What I want you to do is imagine that you've already decided the best animal model, you've figured out the outcome parameters, you've kind of figured out what you're going to measure, you've done the pre-clinical stuff and now you're ready to have the rubber hit the road and go to clinical trials. Instead of -- the individuals who are involved in this panel will be listed here in a moment on this slide and instead of my introducing them based on their academic credentials what I would like to do is ask each to state why they are involved or what their interest has been in their past experience with the issue of exception from informed consent and emergency research. If we could start with Gretchen, please.
MS. SCHULDT: My name is Gretchen Schuldt Doege and I'm probably like the only nondoctor in the room. My husband suffered a massive heart attack 2 years ago and was enrolled in a non-consensual study.
DR. BIROS: Richard?
DR. WEISKOPF: Richard Weiskopf, professor emeritus in the University of California, San Francisco. Had a long -- have had a long-standing interest in research in this area including ethics. I've written little bit about ethics but I'm certainly not a professional ethicist card-carrying -- or card-carrying ethicist. I've been involved in designing trauma trials including trials that operated within this environment of not in the U.S. but something equivalent in other parts of the world. DR. BIROS: Mary? DR. SOUTHWORTH: I'm Mary Ross Southworth. I'm with the Center of Drugs at FDA. I work in the division of cardiorenal products and our division has several -- has reviewed several exception to informed consent trials not for trauma, so much for ACS and cardiac arrhythmias.
DR. GOLDKIND: And as you heard I'm Sara Goldkind. I'm the senior bioethicist at the FDA and I've been involved with issues related to emergency research since I came to the FDA 7-1/2 years ago.
DR. HOLCOMB: John Holcomb, I'm a retired colonel in the Army. Actually I ran a study of a hemostatic dressing in about six countries surrounding Iraq right before the war started. The dressings were carried into Iraq by medics in their rucksacks under an IND with the FDA. And we had some interesting issues that came out of that combat study. The -- now I'm a civilian. At while in the military, help put research teams into theatre. I'm a civilian and we're actively -- currently doing a 50.24 community consultation process with whole blood versus component therapy and are planning a 50.24 study with multicenter prospective randomized study.
DR. BIROS: Thank you. And as previously introduced my name is Michelle Biros. I'm an emergency physician. I practice at a level one trauma center in Minneapolis. I first became involved with this issue in 1993 when I was serving as the research committee chair of a very small professional organization called the Society for Academic Emergency Medicine.
We were minding our own business when a very reasonable and interested lawyer approached us to say that there was a movement afoot to stop resuscitation research in which informed consent could not be obtained and would we like to represent the academic community in some organizational hearings that were being held by the FDA and other regulatory agencies.
As a result of that particular communication, we developed a coalition of acute resuscitation researchers and through the opportunity of working with a number of individuals from many professional organizations including not only researchers but also patient advocates, lawyers and bioethicists, we worked with the FDA to develop the regulations that are now called the exception from informed consent for emergency research circumstances or 50.24.
I've also had the opportunity to be involved with -- at my institution with seven studies that have used the exception from informed consent and I routinely get calls from many people asking about particular studies and how these regulations can be met. I take this as a really very important and honorable thing that I have done and I am very pleased that we have been able to maintain an interest in this as well as offer the opportunity to discuss these regulations that provide incredibly important patient safeguards to resuscitation researchers who are really interested in saving lives.
So with that in mind we're going to talk about some aspects of the regulations that are always under revision and reconsideration and that I think are important when they're -- as I mentioned before when you get out to the point where you're going to do a clinical trial and the patient that you need to enroll is unable to provide prospective informed consent because of their critical illness or injury. Could we have the next slide, please? Oh, maybe I can do it.
So one of the things that we talked about and Sara mentioned was community consultation and public notification. This happens before the trial begins, but now what I would like to talk about as our first discussion point is what can be said at the bedside or on the street or wherever this critically injured person happens to be in order to communicate with family members or concerned legally authorized representatives about enrolment to a clinical trial. So the question is if you were the physician or first responder treating a trauma victim, what should you communicate to the family member about possible enrolment? Should we start with John?
DR. HOLCOMB: That's a good question. So it says physician or first responder -- I assume that this is pre-hospital.
DR. BIROS: Pre-hospital.
DR. HOLCOMB: First responder.
DR. BIROS: Yes.
DR. HOLCOMB: That means you have one to two people caring for this patient and you're -- and by definition if it's a community consultation trial the patient is critically ill, your focus is on the patient. So my guess is that the answer is nothing because your focus is on caring for the patient and doing care and there is not a lot of people standing around to do the consent process.
DR. BIROS: Yes. Sara.
DR. GOLDKIND: So I would disagree. I think that of course some of it depends on how quickly you need to provide that intervention. If it's a resuscitation trial where you have the therapeutic window is vanishingly small if existent of all, it is extremely difficult to actually -- I don't see how you can obtain informed consent from a family member who is standing there. You might be able to in a very truncated way do something meaningful though and find out whether the family member would object to the person being enrolled in the research.
DR. BIROS: Yes. Go ahead Richard?
DR. WEISKOPF: I understand both views. I think there is a further complicating issue and that is we face this frequently -- well, not frequently but when people show up in an emergency room and let's say, for example, they are Jehovah's Witness and they are in -- they, you know, the only way you know that is that somebody who came with that person says that this person is Jehovah's Witness and the person themselves is perhaps say unconscious, how do you know that the person who is telling you this has the actual standing is who or she say they are. And this I could imagine this easily could occur say in an automobile accident where one person is seriously injured and the other is not and the other claims to be a close relative. Nobody is going to take the time to ask for documentation that this is so. And if that person then says I don't want my loved one in this trial, what do you do under those circumstances?
DR. BIROS: Well, Sara, do you want to address what the regulations would say in those circumstances?
DR. GOLDKIND: The regulations don't really go into detail about who is a family member. And I've, you know, having been -- practiced ethics in a clinical setting as well where we had to talk to family members and obtain information I would say that you should err on the side of not enrolling the subject. And I say this with all due respect to clinical research and how difficult it is to enroll people into clinical trials. But I think you should just simply err on the side of not enrolling that person.
DR. BIROS: Yes, Gretchen.
MS. SCHULDT: I think that's the hardest thing. There is, I mean, there was -- in my case, there was a lot of people standing around and a lot of people not telling me anything about -- that was going on. I think if there is a pressure or a need to enroll then at some point when the family consent is sought, they have the -- the family must have the option, and the patient, to withdraw entirely not at some date set by the researchers. But there are -- as far as I can see, there are no consequences to the researchers for mishandling things. And I think even if the consequence is benign and giving the patient a choice that just needs to be done.
DR. BIROS: Mary?
DR. SOUTHWORTH: I think that clearly you can't make a blanket statement about whether consent can be given or not without careful consideration of your study design; the careful definition supported scientifically of the therapeutic window, how much time are we dealing with, perhaps what could you convey to a patient in a very short period of time? We're not talking about reading a 7-page consent form, but perhaps a card with pertinent information. And then there should be a point where the family members can opt out and that should be well-defined in the protocol as well. These are things that the FDA will be looking at in your protocol.
DR. BIROS: John and then Richard.
DR. HOLCOMB: Yeah, it's a -- it is a -- I think this is a extraordinarily difficult question. I think if looking through here and looking through the questions and talking to Laura Brosch, whose position I'm taking because she had a health family emergency, so much more conversant in this area than I am. I think this question really needs to be worked through very carefully.
The scenarios are innumerable and the family issues in the trauma population happen almost every time and age comes into play. Is this person who is with the patient 14 or are they 18? And it makes a huge difference. Let me tell you, you can't tell folks many, many times and it doesn't matter if they're driving or not because who knows. These -- there are innumerable variations of boyfriends, girlfriends, common-law divorced, regular, on and on and on, all the social issues that go with trauma patients that we're talking about in the civilian environment.
I would say that my answer earlier was consistent with two medics, in an ambulance that's a normal rig, one driving, one caring on the way to the ED and with some unknown number of family people present. If there is a lot of people present that means there's people who can talk to the families. And in the hospital when we do these emergency consultation studies, in our hospital 24X7, they are not called in from home, but in the hospital 24X7 are people whose only job in life is to do that. And the studies that are now going on, it's mandated by the IRB and everybody that there be every 4 hours documented attempts to talk to the family.
And when that family person comes in, then they'll sit down in a way that we can talk and go through this. And unfortunately the consent forms are seven pages long, all of them are. And if there's any opt out -- and I agree with Sara absolutely -- if there's any question of opting out, then that's what happens. And then the follow on, we say the follow on letter that comes needs to be very well constructed. The communications with the families later need to be warm and caring and not cold and heartless.
DR. BIROS: And if I may, just to expand briefly on this. There's been my experience that most IRBs require a documentation of effort to contact family members. That is clearly logged in indicating how those members were identified and who attempted to call them or contact them on a very systematic and quite frequent attempt. If all that fails, then there are also requirements by most IRBs that a letter be sent at some point in time to inform legally authorized representatives of enrolment of this -- of a patient's enrolment if they cannot be contacted elsewhere.
Some studies have been more aggressive in this. In fact, there has been a few that want you to track patients down at home. We've done that. We have tracked down patients at nursing homes or other places. It depends on the rigor of the IRB and I think it depends somewhat on the intensity of the study and how complex it is.
One other question -- one other -- just I'm throwing in the emergency perspective here. It was brought up about a blood study in which a patient arrived who was a Jehovah's Witness. Now that is a real important thing to understand that when you decide to do a study in a particular population, community consultation has to take two forms. One is to talk about this study to the community in general, and a second is to target the potential population of subjects and also get their opinion.
Community consultation is not community consent or community veto. But the issues that come up must be brought to the IRB. We were asked to do a blood study product at my institution, but we're considered a bloodless hospital by a very large Jehovah's Witness population. So our IRB said this was not the hospital to do the study. So remember that community consultation piece sometimes will prevent or direct some of these later problems.
DR. WEISKOPF: Yeah, I agree with what John said. Just for clarification, when I mentioned Jehovah's Witness, I was talking about ordinary care, not a research protocol.
DR. BIROS: Yes, certainly.
DR. WEISKOPF: And with respect to the point that Gretchen raised about withdrawal from a study, I would think in every study that I have seen the protocol clearly should state that the patient, him or herself, or the legally authorized representative can withdraw the patient's participation at any time, not in -- not at specific time periods that are noted anywhere, that there's always -- should always be that absolute right.
DR. BIROS: Correct. And also to -- just to expand upon that, remember that we're not asking for consent for participation. We're asking for consent for continued participation because we have already enrolled those patients. They've already been enrolled. Now we are asking for consent for continued participation in the study. And legally authorized representatives can withdraw at any time.
MS. SCHULDT: I feel strongly that they can be unenrolled and should be if the family or the patient didn't want to be. I mean, it's kind of like the federal government comes in, kicks in your door, takes over your life, puts you in a study, and you lose all control. You lose control over everything. And this is kind of communication because it hasn't been discussed yet today.
The family also ought to be assured that the financial responsibility will rest on the people conducting the study. You can't come in, basically do -- well, you can because you do, do medical experiments on people, and that's what studies are, not understand the consequences because that's what you're doing the study, that's why you're doing the study. And then say whatever happens, family, thank you for your involuntary participation, and by the way, the financial burden is yours.
You know, this is again personal, but I had insurance. My husband lived for 6 days in cardiac intensive care. If I didn't have insurance, I'd be ruined in 6 days. And I don't know if the interventions the paramedics did without telling me kept my husband alive for 6 days, neither do they. I kind of lose the trail particularly when you guys, meaning the medical community, get the results. Somebody down the line might make huge profits off this stuff, and the family's sitting there with gigantic bills and nothing to -- nowhere to go with them.
DR. BIROS: Thank you, Gretchen, for that. Are there any other comments or concerns about this first question about what should be communicated? What about the possibility of opting out as was brought up at the time a patient is considered eligible for a trial? Is it sufficient to say we'd like to enroll your loved one into a trial, do you have objections?
DR. GOLDKIND: Well, this goes back again to how you think about the ethics of autonomy in the situation or at least substituted judgment and decision making. The way the regulation is crafted, if there's a -- if you cannot obtain informed consent, then you should be able to ask for any objections to participation in the research and the family member can object. That's different than opt-out which we can talk about in a moment which is frequently confused.
So again, it goes back to some of the issues we've been talking about before what is the therapeutic window, how much information can you provide about the trial, and the alternatives to participating in the trial in a very short order so that it's an informed objection. But I think the regulation sort of holds the bar of objection lower. That is you don't have to have as much information because you recognize there isn't as much time to -- you can't provide all the information, but the objection should still be respected.
Now opt out is where the trial -- the regulations do not require that trials be designed with what's called an opt-out mechanism. However, some IRBs may require the opt-out mechanism be implemented at that site. And it is a -- some sort of a protocol described pre-designated objection to participating in the trial such as a medical alert bracelet or a wallet card, something that's very easily accessible and visible that declares that the person does not want to participate in this kind of research.
DR. BIROS: Thank you. Let's move on to the second question which has some relevance to today's conference. Clinical trials in the military setting, particularly those involving the exception from informed consent, raise ethical and scientific concerns. What special considerations should apply to trauma trials in this population?
John, you began to allude to some of your trials.
DR. HOLCOMB: Yeah, having done trials in theater, the -- you know, when the investigator carries a weapon and is a colonel, and the subjects that are injured in combat are largely privates, carry weapons also, usually bigger weapons actually.
DR. HOLCOMB: But a huge difference in the rank structure and it's not -- this is totally different. For those who are in the military, you know what I'm talking about, with the colonel and private. It bears no relation to the civilian world actually. There's nothing like it. And you're in a military zone. I -- my experience, this is my personal opinion. This is -- by the way, I told -- Laura and I talked about this.
This is not DOD policy. I'm not representing the DOD. If Laura was sitting here, she'd say the same thing. It's my personal opinion that it really is unethical to do community consultation trials in a combat zone.
DR. BIROS: So it appears as if there is -- if the concern is power differential and what could be perceived as coercion.
DR. HOLCOMB: It's absolutely. I mean, the whole system is set up -- if you think about the colonel tells the private to take the hill and the private is not really supposed to question that too much. Now these are doctors largely and you know, you are not out, I mean, in the firefight. But it's a very different world in an active combat environment.
DR. BIROS: Richard, did you want to add something?
DR. WEISKOPF: Not really. John has the outstanding experience in this -- from this group. I spent a number of years in the military, but not doing trials in theatre. Did trials in the U.S. and -- but certainly can vouch for the, sort of, power structure that John has described. And it's certainly no matter what the piece of paper might say, or a verbal interaction, it is not without unstated coercion, if you will, that there is clearly not an unbiased reception and response.
DR. BIROS: Sara?
DR. GOLDKIND: So can I take this opportunity to broaden the question a little bit. John, I'd like to hear your thoughts on whether you can ever do human subjects research in the military setting? Can you get un-coerced informed consent in that setting with that type of power structure?
DR. HOLCOMB: I think you can. I think it's extraordinarily difficult. Jeremy Perkins who is in the audience -- there is Jeremy -- and maybe, Jeremy, you can go to the microphone and comment. Jeremy Perkins ran one of the research teams in theater, one of the very first research teams in theater and so he has some experience in doing this.
Yes, selected, probably not what we're talking about with patients in hemorrhagic shock in a very different mode than we're talking about. I think there are randomized individual consent studies that can be done and are being done in theater.
I think that the constraints and the issues of coercion and influence with power, if you will, are just so much greater in the deployed military environment. I would add as well that if we're in these hospitals there is no PAs and nurse practitioners and residents, you know, where you might have a faculty of 35 surgeons in my hospital, in a deployed hospital you'll have 5. So everybody is pretty busy taking care of patients.
There's not a lot of extra people around. There is not research nurses and IRBs and all of that, IT people and all the infrastructure that surround to do research because people get killed over there. Nurses and doctors have died in Iraq and Afghanistan and will continue to do that.
And the DOD will not is really love to put a lots of research people in to collect data and do studies because they actually put them at personal risk. And there's not a lot of civilian nurses signing up to be research coordinators.
DR. BIROS: Do you have a comment?
DR. PERKINS: Just a comment. I mean, yes, you can do human subjects research in theater, but because of limitations you're going to have to seek a, you know, an alteration of informed consent and probably limit it to no greater than minimal risk studies. So you know, if you can get that together then, yes, you could do that.
And you know, I would say also human subject research in the theaters is not only limited by the power structure, but also getting contact with the legal authorized representative. And there may be specific tactical issues that, you know, oh, by the way, I'm calling from Baghdad and your husband has just been blow up, do you mind putting him on a clinical trial. It just doesn't quite work well like that. So no greater than minimal risk is kind of the policy that we've taken thus far and haven't tried to push that envelope by any means.
DR. BIROS: I think this -- oh, I'm sorry -- go ahead, Sara.
DR. GOLDKIND: So as a general matter, FDA regulated, you know, investigational products are not minimal risk interventions. So that's essentially saying by and large you would expect that FDA-regulated research would not be conducted in the military setting.
SPEAKER: Thanks. I just would make a comment that the research that is being conducted is all IRB approved. They are -- the protocols are written, they're approved by IRB back here in the United States. All of the infrastructure that we have in the civilian side is done. It's just that there is a large emphasis placed, as you said, a non-FDA regulated studies, community consultation studies higher than minimal risk are extraordinarily difficult and to my knowledge have not been done. I've been out 2-1/2 years. Jeremy, are any higher than minimal risk studies been done, do you know?
DR. PERKINS: (Off mike)
DR. BIROS: I think this last -- these discussion points kind of focusing on the last question here, number four, recognizing that there is not one universally applicable answer, what ethical and scientific factor should be considered when evaluating the potential of alternative trial designs to the conduct of emergency research with exception.
For instance, you can have a study where a control of blood might be done in an elective surgical population, but not in the setting of trauma that allows us to have a proof of concept. But it's not quite sure if that translates to a circumstance in which you have emergency trauma. What -- how should we balance the rigorous scientific design and the ability to obtain consent?
SPEAKER: Well, I think you're going to hear some more about this tomorrow from both John and Dr. Moore. But let me expand your question a little bit, if you will, and that is embedded in this unstated, but I think embedded in this is how do you conduct -- what sort of phase 2 trial can get you information that is adequate to support a phase 3 50.24 trial.
What studies would have an appropriate endpoint in an appropriate population that would say, yes, it looks like there is substantial likelihood of having substantial benefit to the patient population that you seek to study in a 50.24 trial. I think that's a real problem in terms of the way the regulation is either written or interpreted that it may not be actually possible to do that under current circumstances.
DR. BIROS: Comments?
DR. GOLDKIND: Well, we struggle with this every time we get a protocol submission under 50.24. So it's not -- there is really no one-size-fits-all and it really requires a case-by-case analysis I would submit. But I think that it's not just what type of phase 2 trial because I think that's true, but it's also can phase 3 trials in a consenting population how much can you use the information that's acquired in that setting to support prospective direct benefit for perhaps a different indication or the same indication, but a more seriously ill population. And that's difficult to address as well.
DR. BIROS: Okay. One last question here and then I want to open it up to the audience. One of the requirements of the exception from informed consent is public disclosure. This has to happen before the trial starts and then it has to happen after the trial is completed.
But as you know, it sometimes takes a really long time for us to know the results of the trial from an analytical and practical standpoint. So how should this particular requirement be fulfilled? Does anyone have any suggestions or experience with this?
SPEAKER: The ROC group which we just joined, so we did not participate in the Eileen Bulger hypertonic saline studies used a phone-in dialing system. It's a random number dialing system which from a epidemiology point of view is -- it may be better than the 15 groups of paid people that you kind of get together. So there is actually a body of literature growing about this. And which one is better or worse is, I think, a fairly subjective comment.
The -- I think the question is how is it best done, is a great one -- are we doing a great job right now. I don't know. There are certainly been in the 70 trials proposed and maybe 40 approved I think probably a variety or tweaks in evolution and how that community consultation process is done. And it's different as required by every IRB.
DR. BIROS: Correct. And that goes to community consultation. You're correct that there has been some study of random digit dialing. I caution that approach because community consultation needs to be a two-way communication. And unless you know who is calling those people and they're educated in this trial and they're certified to be able to get information from you in subjects, I think you kind of may dilute out the community consultation process somewhat. I do think community -- I think random dialing is a great public notification technique, but it's expensive. So you have to weigh that as well. Sara, did you want to comment about this?
DR. WEISKOPF: I'll address the other question that is what do you do with the data after the study. And just a sentence by way of introduction that is I have a strong belief in autonomy and I spent 2 years in industry and I recognize that a sponsor owns the data of the study. Nevertheless, I think that we are imposing upon our population an alteration of the very strong belief in this country of informed consent.
And we're doing it for noble reasons to advance medical science under circumstances that are carefully defined in the regulation circumstances where the data cannot be obtained in any other way. Nevertheless, when multiple trials are conducted in the same field, same area, and we are not well informed enough about previous trials in that area that had been conducted under 50.24 and yet another trial comes along and potentially makes the same mistakes that were done in a previous trial, we are not serving our population as well as we ought to.
And I thought about this a great deal and I've come to the conclusion that as -- and it may sound extreme to some of you -- that even though industry owns those data if it is industrially sponsored that one might consider that as part of the agreement. I know it's not part of the regulation, so it's not possible right now. But in the future perhaps that is part of that regulation that the price that one would pay for being able to do that study is a complete and full disclosure of the data that's been accumulated and not just what it might appear in a publication 2 years hence because we all know that all the data cannot appear. It's just more data than can possibly be published.
DR. BIROS: Thank you. Sara?
DR. GOLDKIND: So I think that we should think about this question from the standpoint of public disclosure back to the community that either was part of the trial or in which the trial was conducted as perhaps distinct from disclosure of the research results to the scientific community. We now have clintrials.gov which many of you're familiar with which ,and that database is a repository, if you will, of clinical trials -- specific clinical trials of which phase -- these phase 3 emergency research trials would qualify. And some results and adverse event information, et cetera, are posted on that website.
So even if the study doesn't ultimately get published -- and I have to say I'm a great believer in publishing negative results as well as positive results, even so that we can inform our co-researchers, so we don't repeat trials that shouldn't be repeated. That information is somewhat available through that mechanism. But what I think we're trying to get at is how do we actually respectfully inform the communities from which the subjects were drawn.
DR. BIROS: Yes, Gretchen.
DR. GOLDKIND: And I think that's a harder question.
DR. BIROS: Gretchen, why don't you give us your insight?
MS. SCHULDT: I think that's actually not that difficult. I mean, if the Wisconsin Tissue Bank can write to a family and say exactly what went where and who got what, I think it would at least increase buying significantly if survivors of these studies, families were informed. A simple newsletter could be national. You have their addresses, you have their last -- I mean, people move obviously.
And the other thing for all this communication stuff, my past life was as a journalist, go for the big four and work down from there newspapers, TV, radio, and the Internet and then work down from there. Studies like you're looking at now of blood loss through trauma, it's true, if it bleeds it leads; TV saying there are some natural stories.
I'm sitting in the audience today as a total civilian looking up words you're using on my iPhone, so I know what you're talking about. And I'm just fascinated and I don't think getting the word out would be difficult and I don't -- follow-up, follow-up, I think, would be a simple -- fairly simple.
I just -- I mean, maybe some money, maybe you'd have to hire consultant to do some PR for you. I don't know what -- how these studies are put together, who has what. I would hope though that -- absolutely would hope that some of the data would be available from the people who were -- for the people who were enrolled in these studies and they would be allowed to know what -- how things turned out in the end.
DR. BIROS: Mary, did you have a comment?
DR. SOUTHWORTH: I think that it's time for us to probably think a little bit outside of the box. And this has got to be the most socially connected society that we live in today. The FDA twitters, the FDA has a Facebook page. I'm not suggesting that those are ways that we disseminate this sort of information, but information is easy to get these days. So I think that thinking beyond phone calls, letters, perhaps using some other social media, I'm not going to saying it's going to happen tomorrow, but I think we need to think a little more progressively about that.
DR. BIROS: What is our time? Ten minutes. Okay, go ahead.
DR. HOLCOMB: I think most of the organizations conducting these studies now actually have a media component, they have a media committee, and there's a media plan which encompass everything you all just said. There's also most of the organizations and the ones that we're involved with right now actively talk about engaging the local civilian city and county government, mayors, city council, et cetera to make sure they understand what's going to go on their city. We talked to the reporters both the big four and engaged them at the time. The worst thing you can do is have this kind of drop to the lap of the community and they're not aware of it --
DR. BIROS: But --
DR. HOLCOMB: -- at all of those levels and then go back, and so because you have an ongoing dialogue and get updates of the study and then when the study is completed you can go back to those same avenues that are wide open and relate those comments. I'll tell you from a trauma point of view we have -- there's -- the reporters know who we are, we know they are, and when they have questions about things the dialogues go back and forth pretty clearly.
DR. BIROS: Yeah.
MS. SCHULDT: Is there a NIH protocol? I mean, I can tell you I've still known people in the news biz, most are shocked that you can do non-consensual studies. They certainly didn't know this was going on, so I'm just wondering if there is a tool kit for local people involved in the studies?
DR. BIROS: Actually that's -- I'm sorry to interrupt you -- that's a great concept. And most large studies who are looking to use ethic now -- that's exception from informed consent in shorthand -- are now developing templates to make sure that people -- if it's a multi-centered trial the same type of information is passed on to many communities. And one good hint I think, good practical consideration is as you design a trial think about establishing such a committee for human subjects protection as a subcommittee of your research group. Because they can keep track of whether or not it's effective and these communication styles are appropriate for a community. Did --
DR. GOLDKIND: And I think that if you can publish the results of that analysis what community consultation approach has worked best for a given trial that really helps everyone out as well.
DR. BIROS: Yes.
DR. MOORE: Moore, Denver. I've conducted three of these 50.24 trials in the last decade and I can assure you that the community education is the most difficult aspect as you all know if you've done any of this. My question is, how can we empower the potential enrollees in these studies? For example, in our most recent study we provided bracelets, albeit that's discomfort to the potential enrollee, at least, that gives them some power to wear a bracelet in the year in which we anticipate the study to deny entry.
DR. BIROS: Correct.
DR. MOORE: But I'm curious how the panel thinks we can empower our potential research enrollees to feel like they have some role in this research process?
DR. BIROS: Again, that's a great comment and it has to do with community engagement and overall research. The bracelets you're referring to are frequently described in community consultation sessions and they are opt out. So a person attends these community consultation sessions decides that they personally would not want to be enrolled in this and they can opt out by using this bracelet. Clearly, community consultation does not hit the whole community. So we have to be realistic that that's not going to be the best way for people to make their wishes known ahead of this time, and that's why these regulations are so carefully scrutinized and surveillance is ongoing.
One of the trials that I'm currently involved with has an opt out national database. So prior to randomizing a person into a study, the investigator has a certain number of minutes to go on to this database to make sure this individual is not registered to opt out.
Again, in this, we did wide community notification about this. We sent 4 million letters out that type of thing, it's not going to hit the whole community but it is another way to make sure that the community has a voice in these. I also think that responsible IRBs understand that community consultation is an important part of decision making with these studies. And I personally find community consultation not to be the hardest thing but to be the most interesting and the most fun thing because it lets me know what my community wants me to study, and it also lets me know what they think of me as a researcher and how I can improve their image. And also educate researchers to understand what the community knows and does not know. Any additional comments on this? Yes, Gretchen.
MS. SCHULDT: You know, I always prefer opt in to opt out.
DR. BIROS: And that also has been under much discussion as well.
SPEAKER: Can we ask Martin Schreiber? You're in the ROC right now. You want to describe the media plan from the ROC for community consultation?
DR. SCHREIBER: The ROC is a multi-center trial and the plan for that actually is by -- varies by institution. Each institution has to work with their individual IRB to come up with a plan. Our institution actually did depend heavily on the random dialing technique, but, in addition to doing random dialing we've also had information on the Internet, we've done public meetings. We found that the public meetings are very unsuccessful and that we may have a meeting and three or four people will show up. So that does not seem to be an effective way to do this. We've put this in the newspaper, on the radio, and on the television.
I have been under-whelmed with the efficacy of that. I found that essentially every patient that I've seen in one of these trials did not know the trial was going on. So I don't think what we're doing is adequate. I do think that the random dialing technique is one of the better ways to do that because we're actually reaching out to people instead of depending on them to either have access to the mode that we're delivering it in or to come to a meeting. So I do think random dialing has benefits, but overall in my experience not a single patient has -- knew about the study when they're actually enrolled in the study.
DR. BIROS: And that has also been validated by other investigators. So we are in the -- the (inaudible) community is in the process of taking a look at the most effective way and how to measure success with community consultation and public notification. And we have one more question here and then I have a very interesting question on a card here. So --
DR. LAWSON: Jeff Lawson, Duke. How do you deal or what are the contingencies for let's say transfusion protocol, whole blood protocol? And you hit a Jehovah's Witness unknowingly in the field and it's not just oh, they got a medicine but now you've sort of violated a core principle of their belief system. How do you address that philosophically and potentially medically legally?
SPEAKER: So we have a large Jehovah's Witness population in Houston and from a research point of view with their blood product -- the blood products for this particular trial are all approved products, they are A or B approved products. The unfortunate problem with a Jehovah's Witness in hemorrhagic shock without a family member around and comes into bed from the field as you don't know. We've talked extensively with elders about this and we treat them like anybody else because you don't know, they don't have a bracelet on that says Jehovah's Witness.
You give them blood just like everybody else. When the elders and the family show up and say stop, we stop. The last patient I had actually slowly bled to death over about 48 hours. The family was adamant and actually the patient had said I don't want blood as he was being intubated. So if you don't know you just don't know, you treat them like everybody else. When you do know, you stop. You can bring in the elders because each little part, each little community within the Jehovah's Witness has different viewpoints on how they feel, about which blood products they will and won't accept. And you know, I think all clinicians who've been in this position finally respect the elder and the individual patients' viewpoints.
DR. BIROS: Thank you. This is -- I have received a number of card questions, but there is one that I think is -- would be important for clarification. If a patient is enrolled and treated in a single dose or short-term exception treatment study and then later chooses to withdraw consent or their legally authorized representative decides that they would like to withdraw, what happens to the data obtained already? Can the family block the use of that data and can non-interventional follow-up safety data be captured?
DR. GOLDKIND: So FDA has guidance that addresses this issue. It's called data retention when you withdraw from clinical research. And, based on our legislative mandates as well as our regulations we're required to maintain the data that has been collected in a clinical trial in the database. There are a lot of reasons for this and we've articulated them in this guidance which I think you all ought to read. One is, that we don't want to have people become part of research and then through a biased, let's say, or a nefarious withdrawal of the data have scientifically invalid or spurious results.
Now, so we as a general matter if the exception from informed consent regulations are followed appropriately would keep the data in the database. At the point that the subject is able to provide informed consent, or the legally authorized representative, or the family member says they want to withdraw from the research, at that point we would no longer collect data.
Now, in many cases the main intervention may occur in the pre-hospital setting and what you're doing in the hospital at that point is just following their survival, their neurologic status or what have you, which is all non -- no longer providing any additional investigational products. We would still say -- now, this is if the family member says I don't want, you know, if they're withdrawn from the research we could encourage you to at least explain to the family, to the subject, to the LAR what remains to be a part of the -- what remains to be done in the clinical trial and that there are no further interventions. They may say that's okay then, please go ahead and continue the follow-up; they may say no. If they say no, you really have to stop collecting the information.
DR. BIROS: Okay, Richard, one quick comment and then we're going to call it a day.
DR. WIESKOPF: No, I have a follow-up question for Sara. What if the family says I specifically ask you to remove my loved one's data from the dataset for this study? I mean, that would be the only reason for withdrawal if this -- if the intervention was already passed.
DR. GOLDKIND: That's a very difficult situation. If there are ethical reasons that that data should be removed, then we will consider that. As a general matter, we try to explain to the family why the data should not be removed from the trial because, we don't want to -- you know, the enrollment of all the subjects into the research is predicated on the fact that the results -- the research will be able to meet it's stated objective, and it won't be an ethical study if it can't.
And so we are -- we would really discuss that on a case by case basis, but as a -- if there is no ethical reason that would obviate the use of that data then as a general matter as I said based on our legislative and regulatory mandates we'd not remove the data.
DR. BIROS: Thank you. I'm very sorry, I know there are many other questions and comments, but the time has come for us to call it quits for today. Thank you all very much for your kind attention.
DR. VOSTAL: Thank you Dr. Biros. That was very nice. And thank you for the panel members. So that concludes today's session. And please come back tomorrow morning at 8:00 o'clock same place, same time. And for the panel moderators, we're going to continue working through the night to produce the summaries of the different sessions. So if you could come down here and we'll meet down here in front of the podium and go to a separate building.
(Whereupon, the PROCEEDINGS were adjourned.)