FOOD AND DRUG ADMINISTRATION
NINETY-FIFTH MEETING OF THE
CARDIOVASCULAR AND RENAL DRUGS ADVISORY COMMITTEE
Thursday, January 17, 2002
8777 Georgia Avenue, N.W.
JEFFREY BORER, M.D., Chairman
Director, Division of Pathophysiology
Weill Medical College
525 East 68th Street, Room F467
New York, New York 10021
JAIME HENRIQUEZ, Executive Secretary
Advisors and Consultants Staff
Center for Drug Evaluation and Research
Food and Drug Administration
5630 Fishers Lane
Rockville, Maryland 20857
MICHAEL F. ARTMAN, M.D.
Professor of Pediatrics
New York University Medical Center
530 First Avenue, FPO Suite 9-V
New York, New York 10016
BLASE A. CARABELLO, M.D.
Chief, Medical Service
Veterans Affairs Medical Center
Medical Service (111)
2002 Holcombe Boulevard
Houston, Texas 77030
SUSANNA LEE CUNNINGHAM, PH.D.
Professor, Department of Biobehavioral
Nursing & Health Systems
School of Nursing, Box 357266
Seattle, Washington 98195-7286
THOMAS FLEMING, PH.D.
Professor and Chair
Department of Biostatistics
University of Washington
Seattle, Washington 98195-7232
COMMITTEE MEMBERS: (Continued)
ALAN T. HIRSCH, M.D.
Associate Professor of Medicine and Radiology
Minnesota Vascular Diseases Center
University of Minnesota Medical School
420 Delaware Street, S.E., Box 508 FUMC
Minneapolis, Minnesota 55455
BEVERLY H. LORELL, M.D.
Professor of Medicine
Beth Israel Deaconess Medical Center
330 Brookline Avenue
Boston, Massachusetts 02215
STEVEN NISSEN, M.D., F.A.C.C.
Vice Chairman, Department of Cardiology
Professor of Medicine
Ohio State University
The Cleveland Clinic Foundation
9500 Euclid Avenue, F15
Cleveland, Ohio 44195
SPECIAL GOVERNMENT EMPLOYEES:
JEFFREY KOPP, M.D.
Kidney Disease Section
Building 10, Room 3N118
National Institute on Diabetes,
Digestive and Kidney
National Institutes of Health
Bethesda, Maryland 20892-1268
ANDREW S. BREM, M.D.
Division of Pediatric Nephrology
Rhode Island Hospital
593 Eddy Street
Providence, Rhode Island 02902
FOOD AND DRUG ADMINISTRATION STAFF:
RAYMOND LIPICKY, M.D.
ROBERT TEMPLE, M.D.
JUAN CARLOS PELAYO, M.D.
MELISA COOPER, M.D., M.S.
BRIAN F. DANIELS, M.D.
LLOYD D. FISHER, PH.D.
JULIA A. BREYER LEWIS, M.D.
EDMUND J. LEWIS, M.D.
KANNAN NATARAJAN, PH.D.
HANS-HENRIK PARVING, M.D., DMSc
MARC A. PFEFFER, M.D., PH.D.
C O N T E N T S
NDA 20-757/S-021, Avapro (irbesartan)
Sanofi-Synthelabo (c/o Bristol-Myers Squibb)
for the Treatment of Hypertensive Patients
with Type 2 Diabetic Renal Disease
* * *
AGENDA ITEM PAGE
CONFLICT OF INTEREST STATEMENT
By Jaime Henriquez 6
By Dr. Brian Daniels 7
By Dr. Edmund Lewis 10
IDNT Efficacy and Safety
By Dr. Melisa Cooper 54
IRMA 2 Efficacy and Safety
By Dr. Hans-Henrik Parving 159
By Dr. Edmund Lewis 201
OPEN PUBLIC HEARING 134
COMMITTEE DISCUSSION OF QUESTIONS PRESENTED 212
P R O C E E D I N G S
DR. BORER: I'd like to welcome you and begin the 95th meeting of the Cardiovascular and Renal Drugs Advisory Committee of the FDA.
Before we begin the formal presentations, we'll hear the conflict of interest statement by Jaime Henriquez, the Executive Secretary of the committee.
MR. HENRIQUEZ: The conflict of interest statement. The following announcement addresses the issue of conflict of interest with regard to this meeting and is made a part of the record to preclude even the appearance of such at this meeting.
Based on the submitted agenda for the meeting and all the financial interests reported by the committee participants, it has been determined that all interests in firms regulated by the Center for Drug Evaluation and Research present no potential for an appearance of conflict of interest at this meeting with the following exceptions.
In accordance with 18 U.S.C. 208(b)(3), a full waiver has been granted to Dr. JoAnne Lindenfeld for her unrelated consulting for one of the sponsors. She received less than $10,000 a year. And to Dr. Alan Hirsch for unrelated speaking for the sponsor. He received between $5,000 and $10,000 a year.
A copy of the waiver statements may be obtained by submitting a written request to the agency's Freedom of Information Office, room 12A-30 in the Parklawn Building.
In the event that the discussions involve any other products or firms not already on the agenda for which FDA participants have a financial interest, the participants are aware of the need to exclude themselves from such involvement and their exclusion will be noted for the record.
With respect to all other participants, we ask in the interest of fairness that they address any current or previous financial involvement with any firms whose products they may wish to comment upon.
DR. BORER: Thank you.
We'll begin then. This morning we're going to consider an NDA related to irbesartan, Avapro, for the treatment of patients with hypertension and type 2 diabetic renal disease. I think the presentation will be introduced by Dr. Daniels.
DR. DANIELS: Thank you and good morning, members of the advisory panel. It's a pleasure to be here today to discuss information about Avapro and its use in the treatment of type 2 diabetic renal disease. I'm Brian Daniels, the Vice President for the Pharmaceutical Research Institute at Bristol-Myers Squibb.
Briefly let me review our agenda and speakers. After my introduction, Dr. Ed Lewis, a Muehrcke Professor of Nephropathy and Director of the Nephropathy Section at the Rush Medical College, will present important background information on type 2 diabetic nephropathy with an emphasis on the endpoints used in our renal studies. Dr. Lewis was a principal investigator for the IDNT study.
Then Dr. Melisa Cooper, a nephrologist and a vice president in the Pharmaceutical Research Institute at Bristol-Myers Squibb, will give the efficacy and safety data for IDNT.
Then Hans-Henrik Parving, Professor and Chief Physician at the Steno Diabetes Center, Denmark, will present the efficacy and safety data for IRMA 2. Dr. Parving was an investigator for the IRMA 2 trial.
Finally, Dr. Lewis will then return to discuss the overall risk-benefit profile of irbesartan in the treatment of type 2 diabetic renal disease.
We have some additional consultants that the panel can use to answer their questions. Dr. Julia Breyer Lewis is a Professor of Medicine at Vanderbilt University. Dr. Lloyd Fisher is a Professor Emeritus in biostatistics at the University of Washington, and Dr. Marc Pfeffer is a Professor of Medicine in the Cardiovascular Division at Brigham and Women's Hospital. All three were involved with the conduct of the IDNT trial.
Just to remind everyone that Avapro is an angiotensin II receptor blocker, active at the AT1 receptor subtype. It's current indication is for the treatment of hypertension. It's available in over 79 countries, with over 3.6 million patient-years of experience worldwide. It's recommended starting dose for hypertension is 150 milligrams daily with a maximum dose of 300 milligrams daily.
Two complementary trials constitute the Avapro development program in type 2 diabetic renal disease. Together they studied over 2,300 patients along the continuum of type 2 diabetic renal disease. These trials were designed as specific renal studies using endpoints appropriate for the severity of renal disease being investigated.
The irbesartan diabetic nephropathy trial, or IDNT, investigated renoprotection in 1,715 hypertensive patients with type 2 diabetic nephropathy, defined as overt proteinuria. Irbesartan microalbuminuria in type 2 diabetic subjects, or IRMA 2, studied 590 hypertensive patients at an earlier point along their disease continuum, specifically those patients with microalbuminuria.
The proposed indication for Avapro for patients with hypertension and type 2 diabetic renal disease: Avapro is indicated for the treatment of type 2 diabetic renal disease.
Now I would like to introduce Dr. Ed Lewis who will give important information on the background of type 2 diabetic renal disease. Dr. Lewis?
DR. EDMUND LEWIS: Thank you. Good morning.
Type 2 diabetic nephropathy is a growing problem worldwide. Type 2 diabetes is epidemic and approximately 40 percent of patients with type 2 diabetes will get nephropathy. Currently approximately 45 percent of patients entering our dialysis transplantation programs in this country are entering with the primary diagnosis of diabetic nephropathy, and the cost of this is enormous.
The natural history of type 2 diabetic nephropathy does not differ greatly from that of type 1, and that is that there is an inexorable progression from early nephropathy to overt nephropathy with progressive structural and functional changes which ultimately lead to a decrease in the glomerular filtration rate and end-stage renal disease.
Virtually all patients with type 2 diabetic nephropathy have hypertension. One difference between the course -- and there are many differences in terms of the patient populations -- of the patient with type 2 diabetic nephropathy as opposed to type 1 diabetic nephropathy is that there is cardiovascular morbidity and mortality throughout the course which represents a second system involved, a competing endpoint, which has to be taken into consideration in the design of any trial.
We can define renal failure as a decrease in the ability of the kidney to carry out its primary function of filtering impurities in the blood, and this is measured by measuring either the glomerular filtration rate, the creatinine clearance being one approximation of that, or evidence of the retention of filterable molecules, particularly creatinine.
As you know, the creatinine clearance represents the estimation of the amount of blood cleared of a molecule in a unit time, so that the numerator in the formula is the concentration of that substance, creatinine, in the blood and the volume of urine per unit time, and the denominator is the serum creatinine so that when one plots the clearance of creatinine against the serum creatinine, you have a hyperbolic function.
Now, creatinine measures the glomerular filtration rate because it is freely filtered and it is not reabsorbed by the kidney. However, there is some secretion of creatinine so that it is not an ideal measure, but it is clinically the most convenient measure of the glomerular filtration rate. The ideal measure is a molecule that is not secreted at all.
Given the fact that we have a hyperbolic function, it's important to note that whenever the serum creatinine doubles along this curve, the creatinine clearance halves. Early then in the course of renal disease, a relatively large change in creatinine clearance is associated with a relatively small change in the serum creatinine. However, doubling means halving of the clearance. Late in the curve, relatively small changes in clearance are associated with large changes in the serum creatinine because it's a hyperbolic function.
So, in designing a trial, the goal is to have an entry criterion where patients enter in an area where changes in the glomerular filtration rate are reflected by readily measurable changes in the serum. Later in the course of renal disease, small changes cause large changes in the serum creatinine, and so again in the design of a clinical trial, we're looking at the changes here being reflective of the phenomenon that we are measuring.
A number of years ago, we, the collaborative study group, carried out the study of ACE inhibition with captopril in type 1 diabetic nephropathy, and we used, as an endpoint in that study, doubling of serum creatinine, meaning halving of the glomerular filtration rate relative to the baseline. When we compared the doubling of creatinine to the clearances in those patients who had doubled, we looked at iothalamate clearance and creatinine clearance. Now, iothalamate happens to be a molecule that is ideal for measuring the glomerular filtration rate. It is freely filtered. It is not reabsorbed and it is not secreted. As you can see, among the patients who doubled their creatinine in that study, there was at least a halving of the glomerular filtration rate. So, we felt that that justified our use of doubling of serum creatinine for that definition of halving of the glomerular filtration rate.
End-stage renal disease is the clinical requirement for renal replacement therapy. The Medicare definition of end-stage renal disease for patients with diabetic nephropathy is a serum creatinine of greater than 6 or a creatinine clearance of less than 15 mls per minute, so that in order to use an objective definition and get away from variances in practice of nephrology in terms of the use of dialysis in patients with kidney disease. The objective definition of end-stage renal disease is taken as the federal Medicare definition of serum creatinine of greater than 6, and that again comes into the design of the study that you'll be hearing today.
This relationship then between the creatinine clearance and the serum creatinine actually defines renal function. Creatinine parameters are not surrogates; they are not substitutes. The creatinine parameters define the ability of the kidney to filter the blood.
To reflect further on the type 1 diabetic nephropathy trial which preceded our type 2 diabetic nephropathy trial, when we looked at the so-called hard endpoints of death, dialysis, or transplantation in the captopril trial of type 1 diabetic nephropathy patients, there was a risk reduction of 50 percent for that endpoint among the patients who received captopril as opposed to placebo.
When we looked at the Kaplan-Meier curve for doubling of serum creatinine, we had the same risk reduction and approximately the same curve. And the reason for that is that the median time from a halving of the glomerular filtration rate to end-stage renal disease was only 9 months so that a halving of the baseline glomerular filtration rate in diabetic nephropathy, with its inexorable downhill course, is a very important milestone. And as you will see, this is true in the type 2 patients as well. And that explains why a doubling of serum creatinine correlated so well with the hard endpoints in the previous study.
Now, I want to remind you about the structure of the glomerular filter because what we're really studying is the function of the glomerular filter. What this graphic shows is that the glomerular capillary bed -- glomerular capillaries having three layers basically, endothelial cell, basement membrane, and epithelial cells, so the filtration is going on here -- and it is built on an architectural structure of connective tissue which is known, here in pink, as the glomerular mesangium.
Now, in a normal glomerulus, the black here is silver staining of the glomerular basement membrane, and you can see these beautiful, graceful basement membranes of the capillary loops, and one can barely see the architectural structure on which these glomerular capillary loops lie.
This is the face of the enemy for a nephrologist. This happens to be from a biopsy that was taken during a pilot trial that the collaborative study group did prior to the irbesartan diabetic nephropathy trial that you'll be hearing about. In this pilot trial, we utilized entry criteria which approximated those which are used for the IDNT, and we did renal biopsies on these patients with type 2 diabetic nephropathy in order to determine the nature of the glomerular lesion which we would be studying.
This is quite typical of what we found. As you can see, typical of diabetic nephropathy, there is a marked increase in the connective tissue of the glomerular mesangium, and it is the progression of that connective tissue which leads to scarring and obliteration of the glomerular capillaries which is ultimately responsible for renal failure in diabetic nephropathy.
In this case, as is true in the average case entering the IDNT, the patient had already lost 50 percent of their renal function so that in the irbesartan diabetic nephropathy trial, what we have are patients who have this advanced abnormality and what we're trying to do is prevent further progression of that abnormality.
In the design of a therapeutic program, given the goal of preventing progression of the established lesion, it would seem appropriate to not only try to tell doctors to try to prevent progression of that lesion, but it would appear appropriate to tell doctors to treat a patient so that they can prevent this advanced lesion from ever occurring. And for that reason, it is logical to study early diabetic nephropathy to see whether you can, in fact, tell a physician that they can treat a patient to prevent them from going on to advanced disease.
Now, there are a number of abnormalities that can be measured in early diabetic nephropathy. Structurally the abnormal connective tissue metabolism which occurs in diabetic nephropathy occurs very early so that there is marked thickening of the glomerular basement membrane in patients who are biopsied very early with the first evidences of diabetic nephropathy and there is expansion of the glomerular mesangium, which you saw. If one wanted to look for the signal for connective tissue metabolism, say, mRNA for type IV collagen, you will find an increase in the signal very early, and I'll show you some data about that.
Structurally, functionally the earliest evidence of diabetic nephropathy is an alteration in the selective permeability characteristics of the glomerular capillary wall, which means that the normal permeability characteristics, which means exclusion of the filtration of macromolecules, begins to be breached, and from a clinical point of view, a reliable and reproducible way of measuring that is to measure small amounts of albumin which begin to appear in the urine at the earliest stages of diabetic nephropathy.
So, the first functional alterations are associated with increased filtration of albumin, and some of that is reabsorbed by the renal tubules. That which is not reabsorbed is excreted in the urine, and that is referred to by the term microalbuminuria.
Now, microalbuminuria is defined as a urinary excretion of abnormal quantities of urine, more than 20 micrograms per minute, which is approximately 30 milligrams per 24 hours, and less than 200 micrograms per minute, which is approximately 300 milligrams per day.
Now, the reason for the upper limit is it is somewhat artificial but it is the borderline between microalbuminuria and the definition of overt proteinuria or overt nephropathy. The reason that it's the borderline is in fact that is where the routine clinical tests that doctors use for proteinuria, the dip stick, turns positive. So, if you want to find out whether there is an abnormality in the selective permeability characteristics of the glomerular capillary earlier, you have to do specific tests for albumin. The clinical dip stick is negative. So, that's what defines overt nephropathy.
This is from a biopsy of a patient who has very early diabetic nephropathy and microalbuminuria, and as you can see, there is basement membrane thickening here, but there is a beginning of the increase in glomerular mesangial material which ultimately leads to the florid lesion that you've already seen.
Most studies of microalbuminuria or early nephropathy have been done in type 1 patients for a variety of reasons. I'll be glad to answer questions about that later. However, it has been consistent to find thickening of the glomerular basement membrane of the order of magnitude similar to patients who have overt nephropathy in patients with early nephropathy, and there is a beginning, as you saw, of mesangial expansion in these patients.
In a study carried out by Sharon Adler, she looked at normal patients who were living related donors, biopsies from normal patients, biopsies from patients with diabetes who had normal albumin excretion, meaning less than 20 micrograms per minute, patients who had microalbuminuria, and patients with overt nephropathy. So, as you can see, among this group the serum creatinine would naturally be normal in normals, it's normal in patients with diabetes and no microalbuminuria, and it's normal in patients with microalbuminuria because early in the course of diabetes, you don't have changes in the ability of the filter to function. Of course, in overt nephropathy, the creatinine is going up because your glomerular filtration rate is going down, as you see here. The albumin excretion rate in the microalbuminuric patients is 56 and in overt nephropathy 4 grams.
And she measured the glomerular collagen mRNA for type IV. As you can see, the signal in microalbuminuric patients is elevated the same as in overt nephropathy, so that the biochemical abnormality for the development and continuation of diabetic nephropathy is there early, giving us good reason to intervene as early as possible.
If one were to study early diabetic nephropathy then, it is not practical to study the structural abnormalities which occur early. It would mean doing renal biopsies on hundreds, if not thousands, of patients, and these are not easy patients to biopsy. They are very obese.
The functional abnormality of altered capillary wall perm selectivity is what we are left with in order to study early diabetic nephropathy. And as I've said, the macromolecule of clinical relevance which can be measured and is accurately measured and reproducibly measured is albumin.
Now, the very quantity of microalbuminuria in a study could be measured and a conclusion made from the study, or a study can use as an endpoint the movement from the microalbuminuric state to the overt nephropathy state, meaning that the patient has crossed the border of 200 micrograms per minute. So, they have progressed.
The importance of that is that in this progression of diabetic nephropathy, that's what happens. Patients go from small amounts of albuminuria to clinically overt proteinuria/albuminuria and then the glomerular filtration rate starts going down. You cannot get a patient having decreased glomerular filtration rate going to end-stage renal disease without them going through the period of overt proteinuria. So, in the design of a trial -- and the IRMA 2 trial is so designed -- what we are looking for is an endpoint which accurately measures the movement from microalbuminuria to overt nephropathy.
The rationale for the clinical development program then was to determine whether inhibition of the renin-angiotensin system is renoprotective in type 2 diabetic nephropathy just as it is in type 1 diabetic nephropathy. And renoprotection is the term applied to the the effect of a drug in protecting the kidney from progressive renal disease which is independent of other systemic effects that that drug might have such as blood pressure lowering.
Now, there are several reasons why interruption of the renin-angiotensin system could be renoprotective in diabetic nephropathy. The glomerular capillary tuft is an arteriole portal system, meaning that the capillaries have an arteriole feeder and an arteriole drains the capillary tuft, so that the pressure within the glomerular capillary tuft is under the control of changes in the two arterioles.
In a normal kidney, there is autoregulation of the blood pressure within the capillaries, meaning that in a normal person, if your blood pressure goes up, there is constriction of the afferent arteriole so that the pressure within the glomerular capillary tuft remains constant. In the diabetic state, which has been directly measured in animals, but of course not directly measured in humans, there is deficiency of that autoregulation meaning that any elevation in the systemic blood pressure is more directly transmitted to these capillaries and there is the barotrauma opportunity there. So, any drug that lowers the systemic blood pressure will lower glomerular capillary tuft pressure because of this deficient autoregulation.
More importantly in the diabetic state, there is constriction of the efferent arteriole for reasons that are not clear, and that is under the influence of angiotensin II. So, in the diabetic state, there is an increase in glomerular capillary tuft pressure directly measured by Dr. Brenner and associates many years ago because of this increase in the tone of the draining arteriole. So, inhibition of the renin-angiotensin system at that level relieves that pressure and diminishes barotrauma.
Furthermore, abnormal matrix metabolism -- and I've shown you histologic examples of that -- is under the control of angiotensin II modulation through TGF-beta which controls collagen type IV metabolism in the kidney, as well as other connective tissue proteins.
Lastly, there is an issue of whether increased amounts of protein trafficking through the kidney is nephrotoxic and angiotensin II, in addition, does decrease glomerular filtration of macromolecules. So, it is possible that a decrease in proteinuria traffic is also a protective mechanism. So, there is good reason to believe that the therapeutic interruption of the renin-angiotensin system can be renoprotective in this disease.
So, we will be presenting studies of early diabetic nephropathy, the IRMA 2 study, the goal of which was to show whether one could stabilize the perm selectivity abnormality in the kidney so that the patient did not go on to overt proteinuria and nephropathy and the irbesartan diabetic nephropathy trial which looked at the therapy of the advanced lesion to see whether the progression of advanced nephropathy could be inhibited.
The goal then of the irbesartan diabetic nephropathy trial is to take patients whose substrate glomeruli looked like this and prevent the progression of this advanced lesion.
With that, I'm very pleased to introduce Dr. Melisa Cooper who will review with you the results of the irbesartan diabetic nephropathy trial. Thank you.
DR. BORER: Thank you very much, Dr. Lewis.
Before Dr. Cooper begins, I want to determine if there are any committee questions. We have some new people on the committee and some guests, so I'd like to set some early ground rules and make a statement.
One of the over-arching issues we're facing here -- and I think you've outlined it really extraordinarily for us -- is to determine what the drug may do, once we hear the data, that causes clinical benefit, makes a patient feel better or live longer, versus what are pharmacological effects, that is, what makes the tests look better but may not have an impact in a significant way on making the patient feel better or live longer. So, that's an over-arching issue that we're going to have to consider because we really need to see some evidence of clinical benefit, which we may well.
With that in mind and with that lovely presentation having been given, I want to ask if there are questions of Dr. Lewis, and I'd like to structure that just a little bit. I want to begin with the committee reviewer, JoAnne Lindenfeld, and then we have two nephrologists who are ad hoc members of the committee today, Dr. Kopp and Dr. Brem, Dr. Kopp from the NIH and Dr. Brem from Rhode Island Hospital. So, after JoAnne, I'd like to hear from the two nephrologists and then we can ask if anybody else has any questions about the presentation of Dr. Lewis.
DR. LINDENFELD: I'd like to echo that was a lovely presentation. Thank you.
Just some questions I have about clarification. One, could you tell us something about the progression of renal disease in diabetics in blacks and Hispanics compared to whites?
DR. EDMUND LEWIS: Yes. Well, of course, in type 1 diabetic nephropathy, it's basically so few patients from those ethnic groups that we don't know a lot except that the few patients who are black, African Americans, who have type 1 diabetic nephropathy do progress faster than whites.
In type 2 diabetic nephropathy, overall patients who are Hispanic certainly tend to have a more rapid progression than patients who are white, and I think that the relevant literature on this actually is Native American literature. It's the Pima Indian data because I think genetically the Hispanic problem for type 2 diabetic nephropathy is probably based there as far as we can tell from the course of that disease. So, it occurs earlier in patients who are Hispanic and it is certainly inexorably progressive.
In terms of whether the absolute rate of progression is worse, that is not entirely clear. Reflecting on the Pima Indian data, the rate of progression of early, meaning microalbuminuria to overt proteinuria, actually approximates the white population and the rate of progression of the disease itself also might be a little faster, but it also approximates the white population. So, it's a bigger health problem, but in terms of what one can expect from the course, other than its occurring earlier in the life of a patient, the courses aren't that dissimilar.
Again, there's less information about African Americans, but I think as a general statement one can say that kidney disease, not just diabetic kidney disease, but hypertensive kidney disease and the like, appears to be more progressive in the African American population and more refractory to any therapies.
DR. LINDENFELD: So, to just follow up on that, in a study that evaluates progression of renal disease, you would like to see those groups, blacks, Hispanics, be equal in all treatment groups.
DR. EDMUND LEWIS: Well, I suppose that you would like to see that, but the problem is that you have to find these patients. I think it would be a more accurate reflection of what I would say is that in a study of type 2 diabetic nephropathy, it would be appropriate to make every effort to get minority groups in the sample. There's no doubt about that.
To get parity I think would be extremely difficult. I think that as you will see in the IDNT, when you start talking about multinational studies and so forth, the representation, for example, among the blacks in the United States in that study was certainly equivalent to the relative population of blacks and so forth, but then when you start to get European involvement, there are no blacks. So, it's harder to construct a study where you have parity there. I don't know exactly how you would come to that.
DR. LINDENFELD: I guess my question relates to whether or not one would design the study for stratify for race, for instance, to make sure that different races were equally represented among the groups.
DR. EDMUND LEWIS: Yes, I don't know. I think conceivably one of the biostatisticians in the group might want to address that. I think that the pre-stratification of a clinical trial such as this I think brings in certain complexities, not the least of which is you expand your sample size tremendously and make the study even harder to do.
There are other issues about bringing minorities into clinical trials which also are a little difficult, and I think the AASK trial at the NIH showed that it's hard sometimes to get minority populations into clinical trials such as this.
DR. LINDENFELD: Right. I'm not just talking about recruiting, but rather making sure that the minority groups are equally represented among the treatment groups.
DR. EDMUND LEWIS: Oh, yes. No, I agree with that. Absolutely.
DR. LINDENFELD: And I think that's going to be an important point. At least the literature would suggest there is an increased rate of progression of diabetic renal disease in minority groups, suggesting that you'd want those to be equal.
Just a second point just for my own understanding. Can you tell us if there are any commonly used drugs -- and we use a lot more drugs in these patients now than we did when the captopril study was done -- that affect the secretion of creatinine or the absorption of albumin?
DR. EDMUND LEWIS: In this trial -- actually it was true in the captopril trial too -- in order to control blood pressure, in addition to the coded medications, at least three antihypertensives and diuretics were used. So, the treatment of hypertension, which of course, actually both from a cardiovascular and a renal point of view, is terribly important in this patient population, is a polypharmacy issue and that is a very relevant question.
None of the antihypertensives -- actually we had this data in the type 1 study because it was a much smaller study. We had iothalamate clearances in those patients, so we were able to determine whether drugs altered creatinine secretion better because we had both the creatinine clearance and the iothalamate GFR. We can compare those.
The antihypertensive agents generally used, which is what was used in the study, and the diuretics generally used would not alter significantly the creatinine secretion course. And I think this sort of goes back to your first question. The randomization of these patients and the fact that all of these drugs were being used in all patients would sort of cancel things out if there was a minor difference, but to our knowledge there is no difference.
And in terms of albumin excretion, I think all that you can say about that is that there is certainly a relationship between the variance in albumin excretion and the systemic blood pressure so that if you lower the systemic blood pressure, you will have less albumin excretion over a very broad range of albumin excretion. Therefore, in designing a study where one's endpoints are albumin excretion, you have to account for the blood pressure lowering effect.
DR. LINDENFELD: And then just one final question. In the type of patient that was entered in the IDNT trial, a patient with gross albuminuria and elevated creatinine, how much would you expect the initiation of diuretics to change the serum creatinine?
DR. EDMUND LEWIS: I wouldn't. I think what we found is that -- you see, it's hard for me to answer this question for the trial because in a very complex group of patients like this, physicians were using more and less diuretics according to how much edema the patient had. We were really all over these doctors in terms of controlling blood pressure and stuff. So, there were variances in dosing even of diuretics.
But the only direct answer that I can give you about that is that we did have a protocol about elevation of the serum creatinine early because what we were concerned about was whether, using an agent like irbesartan, something that interrupted the renin-angiotensin system, a patient with bilateral renal artery stenosis would go into acute renal failure.
Now, as it turned out, that didn't happen during the study, but there were patients who raised their serum creatinine early in the study because they suddenly had their blood pressure controlled, and you will see the data on that. Most of the patients coming into the study were way out of control relative to any standards, and once they had their blood pressure controlled, which included diuretics, there would be a bump in creatinine in a number of these patients, and we at the clinical coordinating center of the collaborative study group would be advised about these patients I think generally, usually. Certainly if they doubled their serum creatinine, we would, but if they raised it by 25 percent, we would be advised about that, and we would talk through the clinical problem and invariably the creatinines came back to normal once diuretic therapy was modulated. So, these are very complex patients.
I think that we had the appropriate feedback to figure out that this was happening, and it was not a study-long issue. It was an issue that would occur early in the study when these patients were getting their blood pressure controlled.
DR. BORER: Dr. Kopp?
DR. KOPP: Thank you. I'd like to echo a second time that that was an excellent presentation of a complicated topic.
I'd like to get your thoughts about a topic that I'm sure will come up again which is the role of macroproteinuria as a surrogate endpoint for both diabetic nephropathy and in the future nondiabetic nephropathy. We know that the level of proteinuria represents a graded spectrum of risk for rates of progression. Do we know quantitatively what level of reduction in proteinuria is clinically significant, and is there data in terms of a similar quantitative reduction in risk of progression?
DR. EDMUND LEWIS: Well, you know, I think nephrologists are on the same wavelength on this issue, and I think that the wavelength that we're on is that I think we're all beginning to understand that the more proteinuria you have, the worse your course will be. I think that we can all agree with that.
I think the other thing that we have to say for certain is that no clinical trial has been designed to test the answer to your question. You can tell me if I'm wrong on this, but I think that it would require a design where you're actually shooting for two different levels of reduction of proteinuria, for example, and that hasn't happened. So, all of the data that we're working with is post hoc.
Having said that, I think that when one looks at a given disease like diabetic nephropathy, given the problem of constraints of how long you're actually going to be able to follow these patients in a clinical trial, I think the best that we can say is that one group did or didn't progress in terms of their proteinuria more than the other, implying that the patients who had greater progression of proteinuria are at greater risk of continuing renal damage.
I think that almost for certain any study of kidney disease where the patient does well, well being progression or regression of renal disease, the proteinuria goes down. And any patient who does poorly, that is, their GFR keeps going down, the proteinuria is likely to go up. But it becomes a chicken and egg thing then because is the proteinuria going up or down because you're treating the glomerulus or is it going up or down because proteinuria is a determinant of nephrotoxicity let's say. And I don't think that any of these trials, including the ones you'll be hearing today, necessarily -- Dr. Parving might have different feelings about this, but I don't think they necessarily help in terms of answering your question.
DR. KOPP: So, I guess I hear you saying perhaps it's not quite time to begin to use proteinuria as an endpoint in and of itself. Is that the implication?
DR. EDMUND LEWIS: No, I'm not saying that because what I'm saying is that I think that -- well, the first thing that I have to say is that -- and I think again nephrologists understand this in general together well. When you're studying a filtration system, there are so many things you can study, and it doesn't matter whether you're an industrial engineer or a bioengineer studying dialysis membranes or a renal physiologist. What you can study is either the capacity of the membrane -- and in terms of the kidney, it's the glomerular filtration rate -- or the selective permeability characteristics of the membrane, which in glomerular disease is proteinuria.
So, I think that it is time for us to recognize that if one prevents going from low amounts of protein excretion to high amounts of protein excretion, certainly we have enough correlations there to be able to say that that is progression of the renal disease. So, I would argue that a study, the goal of which was to show that you didn't go from one stage of the disease to the next stage of the disease, more proteinuria, is a valid study of the intervention in the course of renal disease. But that's just my opinion.
DR. KOPP: Thank you.
DR. BORER: Dr. Brem?
DR. BREM: I'd like to ask again a question about glycemic control. One of the things that people have stressed in the past is adequate glycemic control for patients and that that is a major factor in progression of disease. Yet, there wasn't any discussion about that in your presentation. I was wondering if you might comment a bit about that and perhaps how it may affect outcome.
DR. EDMUND LEWIS: Well, glycemic control in either the type 1 or type 2 patients is certainly not easy, but in the type 2 patient, it is extremely difficult because of the fact that you can't just give them insulin and get the response you want. I think we have the UK PDS, for example, which says that a glycemic control is important.
In the IDNT, there was a tremendous range in terms of hemoglobin A1C's which narrowed over the course of the disease. However, there was still a range. These people are extremely hard to control.
One of the investigators who was on our executive committee, Dr. Rudy Bilous of Great Britain, who's I think a well-respected diabetologist worldwide, looked at our hemoglobin A1C data relative to data that they had gotten in the United Kingdom of control of type 2 diabetes and found that basically the distribution of our hemoglobin A1C's was exactly what was the case in the general population of type 2 diabetic patients.
More important to your question is that irrespective of how difficult it is to control hemoglobin A1C's, the level of hemoglobin A1C throughout the study in all three treatment groups was equal.
DR. BREM: Right. Well, I guess the question I was asking is if the hemoglobin A1C were in the lower range, did those patients progress more slowly in all the different categories of treatments from these different studies, sort of an analysis of variance.
DR. EDMUND LEWIS: I think that neither the collaborative study group nor Bristol-Myers Squibb has looked at quartiles or quintiles of hemoglobin A1C and the rate of progression. I think we just haven't looked at that. I think that it is an interesting question, but I think that for us the two really burning issues were: one, was our glucose control what is seen in patients in the wild, which was true; and two, was it equivalent in all three groups. Of the many, many analyses that we've done through, I'm sorry to say we haven't done the one that would satisfy you for that question.
DR. BREM: The other was a minor thing I guess in terms of the creatinine doubling. That I guess is assuming that the creatinine in most people is 1. As a pediatric nephrologist, I would point out that many children have creatinines considerably below 1 and perhaps small adults have creatinines that are below 1 as well, as creatinine reflects muscle mass. If the creatinine is below 1, for instance, and doubles, it may go into what's still considered a normal range and yet be doubled and, in fact, probably represents a 50 percent reduction in renal function. Does that 9 months apply to those patients?
DR. EDMUND LEWIS: Yes. No, I agree with that although I just want to expand on that for non-nephrologists who don't think about creatinine clearance on an hourly basis during the course of the week. The hyperbolic curve that I showed you, that particular curve would have been 100 over the serum creatinine. In very muscular people, the daily creatinine production would be much higher, which would shift the entire curve to the left, but it remains a hyperbolic curve with the same shape and so forth. In small people, children, very elderly people and so forth, the curve might be shifted to the left rather than the right because they're making much less creatinine but it's still a hyperbolic curve.
In our study -- and you'll be hearing more about this -- the creatinine entry was such that you could not double your serum creatinine and remain in the normal range. If you doubled your serum creatinine, you were in the high 2's or 3's. I think a woman could have a lower creatinine and come into the study, but still, when they doubled, their creatinine was quite elevated. So, we don't have information about patients who doubled their serum creatinine and it's still in the normal range.
I think because of the hyperbolic curve, which would be much steeper in a child, it would be much harder to know exactly where you've doubled and halved your creatinine clearance because you're really on that down slope which is why, in the design of the clinical trial, we're going to the linear part of the hyperbolic curve not up the vertical axis.
DR. BREM: So, those patients probably already had evidence of significant renal disease or impairment at the start perhaps of their study.
DR. EDMUND LEWIS: Yes. The mean GFR coming into our study was 50 and the mean urine protein was 900 milligrams. The glomerulus that I showed you was not the worst glomerulus that I picked out of 30 renal biopsies. We really were studying advanced disease.
But I think that you hit upon the issue which we just discussed, and that is the patient with early diabetic nephropathy, which is where you really want to intervene, is in many ways analogous to the patient who is a child. That is, you start to get evidence of renal disease, but you can't actually measure it accurately by measuring the glomerular filtration rate. So, all that is left for us is measuring the other parameter of filter function which is permeability. I mean, that's all that's left.
DR. BORER: Before we go on to other questions, let me ask Bristol-Myers Squibb to sort of make a bookmark because you may have some data which we haven't heard yet, so I don't want an explanation now, relevant to Dr. Brem's question. My recollection is you did Cox model analyses on these data, so you could at least tell Dr. Brem and the rest of us whether the effect on proteinuria and the other endpoints is independent of the effect on glycemic control or on glucose or on hemoglobin A1C even though you may not be able to give specific data. Don't tell us now but when you present the data.
Are there any other questions from people around the table about the pathophysiology of renal disease? Ray, you had a question?
DR. LIPICKY: I guess I'd like to just clarify something as a non-nephrologist. I think I heard you saying two things, and maybe you did and maybe you didn't. But you're saying, I think, that if you understand kidney disease, the creatinine is not a surrogate measure of anything. It is a measure of disappearance of functional glomeruli, and consequently although patients don't feel anything and there is no morbid/mortal consequence that is associated with any creatinine, it is a direct measure of how many functional nephrons you have and that may be just an exaggeration. So, that's part one.
Then part two is that although progression from microalbuminuria to overt proteinuria again is not a symptom, that if you understand the nature of the disease, that is a sure sign that something has happened to the glomeruli, and if you do not see that happen, then that's a sign that nothing has happened to the glomeruli. Did I say that in the right way?
DR. EDMUND LEWIS: Let me think about it for a second.
Now, one thing you should understand, Dr. Lipicky, is that in the profession we consider you a nephrologist.
DR. EDMUND LEWIS: So, I just want to make that clear.
In terms of the creatinine parameter, yes, I think that you state it correctly, and that is in terms of kidney diseases, in terms of the fact that there's disease going on in the kidney, not just type 2 diabetes, but a whole variety of kidney diseases are silent. And if you are trying to measure the progression of renal disease, you are left with measuring the functional ability of the kidney as a filter, and creatinine is a direct measure of that filter. That's why I say it's not a surrogate because it is measuring the function.
And in terms of the proteinuria question, yes, once again, I think that in proteinuria studies you have to be careful because there are, on a day-to-day basis, many factors which can alter the excretion of protein, including if you run up to the top floor of this hotel and run back down, you will excrete more albumin than if you were just walking around here. You have to be careful about that because there are alterations in the tendency of that filter to leak protein under a variety of conditions, chain smoking cigarettes for a while, running around, running marathons, and so forth. But when you get fixed increases in the level of protein excreted and you start crossing borders, like the border into overt dip stick positive proteinuria, you are then talking about changes which reflect the early changes in the course of disease.
DR. BORER: Any other questions? I think Alan and Steve each had a concern. Alan?
DR. HIRSCH: My question is again a follow-up to Dr. Kopp's and my nephrology colleague's question. When we talk about surrogate markers, obviously, we have to place some kind of value on the surrogate, and later today we'll be talking about combined endpoints and value to the patients. I want to come back one more time.
With some surrogate markers, there's a percent reduction in LDL cholesterol. I know pretty well what that does to the patient in terms of any cardiovascular risk. There must be some threshold below which intraocular pressure decrease will prevent blindness, some nadir wedge pressure change which alters shortness of breath and mortality.
What I struggle with as a non-nephrologist is what level of microalbuminuria change has any impact down the road in some time frame on a clinical outcome. Do we have any information, or is it merely at this point a qualitative improvement in the natural history?
DR. EDMUND LEWIS: Tell me if this is adequate or not. What I would like to do, since Dr. Parving is probably the most logical person in the world to discuss this topic, I really want you to hear his opinion about this.
But to just go back to my answer to Dr. Lipicky, I think the problem is the goal being early intervention. I think that your confidence about measuring something that doesn't have a symptom and that is a point in time during the course of a disease is dependent upon how much information you have about the natural course of this disease. One of the things that you can be sure of in diabetic nephropathy is that the course is inexorable, so that when you start to see increases in urine protein excretion, you can be certain that that will progress if there is not an effective intervention. I think we know enough about the course of diabetic nephropathy to be able to say that, but to a certain extent, I'd like to defer to Dr. Parving who's done I think most of the really truly valid publishable studies in this area.
DR. BORER: Maybe we can hold that for Dr. Parving's presentation.
Final question, Steve?
DR. NISSEN: The cause of mortality in these patients of diabetic hypertensive disease, if I'm correct, isn't about 80 percent of the mortality cardiovascular?
DR. EDMUND LEWIS: Yes.
DR. NISSEN: Myocardial infarction and stroke being the most common.
DR. EDMUND LEWIS: Yes.
DR. NISSEN: So, would not one test of this surrogate of doubling creatinine be the relationship between the ability to affect the doubling of creatinine and the ability to affect cardiovascular mortality, death, myocardial infarction, nonfatal infarct, stroke, et cetera?
DR. EDMUND LEWIS: See, I think it's not a one to one. It's a relative increase. And we're talking about populations now.
You get into a very interesting, complex issue, and these are interesting and complex patients let me tell you. There's no pleasure to do a clinical trial with this group of patients.
Microalbuminuria in the nondiabetic population -- let's say the hypertensive population -- is a marker of cardiovascular disease. People, for example, with hypertension who have microalbuminuria have a much worse prognosis over the next 10 years in terms of myocardial infarction, cardiovascular death and so forth than people with hypertension who don't have microalbuminuria. And we don't know why that is. We don't know what the vascular issue is that explains that.
But I think that at this point I have to say that because there are a number of clinical states associated with decreased perm selectivity and microalbuminuria, it doesn't negate the importance of that parameter in diabetic nephropathy just as decreased glomerular filtration rate is seen in many diseases, it doesn't mean that studying that in diabetic nephropathy is not valid.
So, the microalbuminuria means that, indeed, that is a population of patients who have increased cardiovascular risk. Obviously, type 2 diabetic patients have increased cardiovascular risk. But I don't think that one can draw the conclusion that you can use a renal parameter in patients with overt or even latent diabetic nephropathy with a cardiovascular index and say the cardiovascular event is the hard endpoint even though the albumin is the renal parameters --
DR. NISSEN: But if that's what happens, if your renal function gets worse, you ultimately go on and die a cardiovascular death, then wouldn't one want to see that a drug that slows the development of end-stage renal disease would have a beneficial effect on the hard cardiovascular endpoints? What I'm getting at is, as a way of validating the surrogate, whether or not we ought to see such a relationship.
DR. EDMUND LEWIS: You know, you're undermining my concluding statements.
DR. EDMUND LEWIS: The thing is that if you look at the cardiovascular course of patients with type 2 diabetic nephropathy, certainly there are excess cardiovascular events throughout nephropathy and those patients with type 2 diabetes who have proteinuria have many more cardiovascular events than those who don't, and those who have a decrease in GFR have more still than those who don't. And when they go on dialysis programs, the cardiovascular events go way up. Of course, that is why, in terms of preventing cardiovascular events, the one dramatic thing that we can do is prevent them from going on to end-stage renal failure.
But I think that in this patient population, what I've come to see is that the cardiovascular disease in patients with advanced renal disease -- so, we're talking about the IDNT patients -- is so advanced when you start to study those patients that I don't think that you can use a cardiac endpoint to indicate that you've done something as far as -- you know, that altering the progression of kidney disease can alter that. I think that that's what it comes down to.
DR. BORER: I made a misstatement. We have one final question from the far side of the table there.
DR. TEMPLE: The previous discussions are interesting. They go to the heart of surrogacy and all kinds of things. I would say we've certainly accepted the idea that creatinine doubling is an anatomical finding that has something to do with whether you're going to have renal failure. That's not a big stretch in many ways for reasons you just gave.
It would be true, though, that something that had a physiologic effect or a pharmacologic effect on creatinine might not be very persuasive because what you're saying is when you see a creatinine doubling, that's really an anatomic effect. You're describing the state of the glomeruli. So, something that had a transient effect wouldn't be nearly as persuasive. You wouldn't know what to make of that. JoAnn was sort of asking about that before.
My question goes to the microalbuminuria. Do we know whether any of the drugs being studied here might have a sort of physiologic effect -- I'm not sure what that would be -- that would decrease the amount of albumin but not really reflect the state of the kidney?
Just by analogy when people wanted to say that use of ACE inhibitors at the time of an infarction would prevent remodeling, we always said, well, that's nice but just showing a change in ejection fraction while still on drug is not very impressive because that just may be that you're a vasodilator. So, that doesn't prove anything. Take the drug away and show us that you still have an impact on ejection fraction. That would be convincing.
So, my question is how, does that apply to the microalbuminuria findings here? Is there anything these drugs might do that could be fooling us about whether they're really making an anatomic change or just sort of changing the hemodynamics in the kidney to alter protein excretion? What's known about that?
DR. EDMUND LEWIS: I think that that's the important question for you. It's the important question for us when we're designing trials, and in a way it is very, very difficult to come up with a concrete answer unless you follow these patients for 10-15 years. So, we do have the constraint of coming up with a parameter within the period of some reasonable clinical trial.
I think that Dr. Parving will address this because in the IRMA 2 trial, the higher dose angiotensin receptor blocker actually was associated with continued decrease of urine albumin excretion even after the drug is stopped. And I think that that's probably the best that you can ask for if you want to say it's physiological. It's not physiological.
I think in terms of the preamble to your question, there's very, very little known about tubular reabsorption of albumin, and I think that you will see in the IRMA 2 trial data with two doses of ARBs and so forth. I don't think that there is a reason to believe that decreased albumin excretion is because the same amount has been filtered and more is being reabsorbed. That certainly does occur for sure with lowering the blood pressure and that has been accounted for in this trial. So, I think my goal here is to get off this podium.
DR. EDMUND LEWIS: After Dr. Parving's talk, I hope you will grill him about this.
DR. BORER: This is really the final question. Tom?
DR. FLEMING: Well, I think my colleagues have asked a lot of the key issues here, as I've been thinking about it, but I think Dr. Temple just got at something that I've been thinking about as I've been listening to you.
You had mentioned creatinine clearance is, in essence, not a surrogate. It is truly the clinical event of interest. And listening to your presentation, it strikes me that what would be the truer measure would be something that's fundamentally structural progression, structural abnormality versus functional abnormality. I'm motivated to ask the question by Bob's question because it seems as though there are more factors that could influence the functional abnormalities. Wouldn't we best be served, although it may not be so achievable, to be looking at something that is directly structural progression, structural abnormalities?
DR. EDMUND LEWIS: Well, no. I think in an ideal world -- and I think it's not unreasonable to make that demand. You know, this is coming out of a life where my focus has not been diabetes. It's been lupus actually. So, we're more interested in structural and functional issues there.
First of all, I think that is to me not conceivable that one could do a study of multiple biopsies in this patient population. This happens to be a dangerous population for renal biopsies, and I think we were very fortunate in many ways that we did the pilot trial and that was fine because this is a very obese population of patients and they have hypertension. So, their risk with renal biopsy is greater than the usual patients whom we biopsy.
The ethics of doing multiple biopsies I am not sure that any IRB would approve of, but I can't speak for IRBs in the future and so forth. And I know that comment probably doesn't mean anything in terms of what's going on here, but that is my opinion.
The other thing about that -- and we've certainly seen this in doing multiple biopsies in other diseases like lupus -- is that there is a sampling issue so that if you want to find a difference between two biopsies, certainly there are morphometric ways of measuring things, but in the end, even though it sounds like that might be the gold standard, the fact of the matter is that the accurate and reproducible way of studying renal function is the functional issue which is the ability of the kidney to filter and not the morphologic issue which in this case, especially with the advanced disease, would mean that you would be trying to show stability. That would become a real statistical issue in terms of morphology.
So, in answer to your question, I think that ideally certainly at the bench with experimental animals, that's what you do, but in terms of our ability to actually study clinically patients with type 2 diabetes, I don't think we could do it.
DR. FLEMING: Well, I can readily be persuaded with what you've said, that measuring these functional abnormalities may be more measurable and even potentially more reproducible. My concern is more uncertainty about what is the magnitude of effect, duration of effect, and other factors that could influence those functional abnormalities that aren't necessarily integral to what it is that we're trying to do here.
DR. EDMUND LEWIS: Well, let me just ask you, are you referring to the perm selectivity issue, which is the proteinuria issue, or are you referring to the filtration issue?
DR. FLEMING: Actually my concerns would apply to any of these markers.
DR. EDMUND LEWIS: I guess the key term here with type 2 diabetic nephropathy is "inexorable." As you will see, using the serum creatinine as a direct measure of renal function, you can expect progression, you can expect doubling, indicating having the glomerular filtration rate. Shortly after that, you can expect the patient to get to a level of renal function where they require dialysis and transplantation, and that is progressive, and I think that you will see in our data that that in fact is what happens.
So, if one uses doubling of serum creatinine, as we have, as the index of significant loss of renal function, those patients invariably progress to the hard endpoint, if you will, which is requiring end-stage renal failure management. And it's there where it's undeniable that you've actually got a clinical event.
So, we are not talking about a measure. We're not talking about creatinine as a surrogate any longer; we're talking about it as a measure. But we're not talking about it as a measure that doesn't have serious clinical significance; we're talking about it as a measure that ultimately we can expect a hard endpoint, if in fact we were to follow the patient long enough.
DR. BORER: JoAnn?
DR. LINDENFELD: Just one final question. You showed very nice data in the captopril trial that creatinine clearance and iothalamate clearance were exactly equal. Do we have any data at all like that in this type of patients before the institution of therapy and after the institution of therapy?
DR. EDMUND LEWIS: No. Well, I'm not sure I get your question.
DR. LINDENFELD: To be sure that secretion is not an issue.
DR. EDMUND LEWIS: No, we don't have --
DR. LINDENFELD: It seems like that's a physiologic measure that would help us understand that, as Dr. Temple brought up, we're not seeing sort of a physiologic change that's not reversible. So, that kind of measurement would be enormously helpful to show that, before and then after treatment, those two things don't change.
DR. EDMUND LEWIS: Yes. Again, I want to emphasize that patients entering the IDNT were patients who had really advanced disease. As you will see, their blood pressures were high even on antihypertensive medication before they got to us, and this is not a clinical situation where we can just stop drugs and do clearances. I don't think that it's something that is a practical thing in this patient population. I don't believe that one can get the data that you're asking for, which is creatinine dynamics off the drugs that these patients are going to have to be on. So, it's a problem there. What I'm saying is I don't think there's an answer to your question.
DR. BORER: Dr. Lewis, I want to thank you very much. I must say I wish you had been speaking about this to my class when I was in medical school.
DR. BORER: We'll go on and while we're doing that, I want clarification that requires only a yes or a no from Dr. Cooper or maybe from Dr. Daniels. Is it true that the proposed indication is for the treatment of patients with type 2 diabetic renal disease, not for the patients with hypertension and type 2 diabetic renal disease? Is that correct?
DR. COOPER: With hypertension.
DR. BORER: Because that's not what was given to us. So, we have to make that clarification.
DR. COOPER: In both studies all the patients had hypertension.
DR. BORER: I know they did, but the proposed indication, your slide A-6, doesn't say that. That's why I'm asking. But now you've clarified it. You're asking for approval for treatment of patients who have hypertension and type 2 diabetic renal disease.
Having clarified that, let's move on. Dr. Cooper?
DR. COOPER: Good morning, Chairman, members of the advisory committee and the FDA and invited participants from the academic community. I have been involved with the irbesartan diabetic nephropathy trial since its inception working with Dr. Lewis and the collaborative study group to design this trial between 1993 and 1995. I am here today to share the results with you.
The presentation is divided into four segments: the study design and conduct, the demographic and baseline data, the efficacy results, and the safety.
The irbesartan diabetic nephropathy trial, or IDNT, was designed as a single trial that tested two hypotheses. Does interruption of the renin-angiotensin system with the angiotensin II receptor antagonist irbesartan provide renoprotection in subjects with type 2 diabetic nephropathy independent of blood pressure lowering? Specifically, would irbesartan be superior to placebo in the primary comparison and would irbesartan be superior to amlodipine in the secondary comparison?
The primary endpoint was a composite of doubling of baseline serum creatinine, end-stage renal disease, or death. The design of the study was carried out according to the principles that the collaborative study group had established in the type 1 diabetic nephropathy study with captopril. An irreversible doubling of serum creatinine is a direct measure of the decline in the kidney's ability to filter blood and corresponds to the loss of 50 percent of renal function. When a subject reached doubling of serum creatinine as an endpoint, coded medication was stopped to allow the study investigator to treat the subject outside of protocol. Verification of doubling of serum creatinine as an endpoint required submission of two consecutive samples for measurement of serum creatinine by the central laboratory at Rush Presbyterian Hospital after all corrective actions defined by the protocol had been undertaken to confirm there were not reversible causes.
End-stage renal disease was defined as renal transplantation, the need for dialysis, or a serum creatinine equal to or greater than 6.0 milligrams percent. This threshold for serum creatinine was selected because it is the trigger for initiating dialysis in diabetics as endorsed by Medicare.
All-cause mortality was included in the primary composite endpoint due to the competing risk of cardiovascular disease in these type 2 diabetic subjects.
The secondary endpoint involved cardiovascular events that affect these subjects: cardiovascular death, nonfatal myocardial infarction, hospitalization for heart failure, permanent neurological deficit attributed to stroke, and amputation.
All primary and secondary outcome measures were adjudicated by the outcome confirmation and classification committee or the mortality committee without knowledge of coded medication assignment. These committees were independent, non-BMS entities.
In order to qualify for study entry, subjects had to be 30 to 70 years old with type 2 diabetes, hypertension, as defined here, and a urine protein excretion exceeding 900 milligrams. Serum creatinine was between 1.0 and 3.0 milligrams percent in women and 1.2 and 3.0 milligrams percent in men to assure that renal function was on the linear slope of decline.
Subjects from 209 sites located in 27 countries were randomized to one of three treatments: placebo, irbesartan, or the calcium channel blocker amlodipine.
When the trial was first designed, the relative importance of blood pressure lowering alone versus unique benefits of antihypertensives with mechanisms of action, other than interruption of the renin-angiotensin system, in type 2 diabetic nephropathy remained to be determined. Published reports of studies in experimental models and in patients with either microalbuminuric or proteinuric diabetic renal disease suggested that administration of calcium channel blockers could be renoprotective. Furthermore, amlodipine at that time was the most frequently prescribed antihypertensive used in diabetics.
In order to test the two study hypotheses, aggressive management of blood pressure control was essential. Multiple antihypertensives, with the exception of those disallowed by the protocol, angiotensin II receptor antagonists, ACE inhibitors, and calcium channel blockers, were added for all subjects to ensure that the target blood pressure level, 135 over 85 millimeters of mercury, was reached. An independent committee of physicians, the clinical management committee, reviewed data periodically in a blinded manner to ensure a blood pressure lowering to target levels for each subject and across the three treatment groups.
Subjects were followed for an average of 2.9 years and were seen every 3 months until the end of the study. A data safety monitoring committee reviewed unblinded safety and efficacy results periodically throughout the study.
1,715 subjects were randomized to one of the three treatment groups and all were included in the intent-to-treat analysis. 16 subjects did not receive study drug. All of the 1,699 subjects who received at least one dose of study drug were included in the safety analysis.
408 subjects discontinued study drug early. Of these subjects, 161 reached one of the endpoints and 121 were followed until study closure without an endpoint. 118 subjects were missing measurement of serum creatinine at study closure. Dialysis, transplantation and mortality status was known in 89 of these subjects. Mortality status was known in the remainder. 8 subjects were lost to follow-up. The remaining 1,291 subjects completed double-blind therapy as defined by the protocol.
The incidence of discontinuation of study drug was similar across the three treatment groups.
The baseline characteristics of all randomized subjects is demonstrated here. It was similar across the three treatment groups. Subjects were close to 60 years of age, predominantly male and caucasian, with type 2 diabetes for an average of 15 years. In response to one of the earlier questions, distribution of the races across the three treatment groups was similar.
Consistent with the natural history of the disease and the duration of known diabetes, subjects had mild to moderate renal insufficiency with a mean serum creatinine of 1.7 milligrams percent, and notice here the creatinine clearance at baseline was 57 to 59 milliliters per minute. Normal creatinine clearance in this population would be considerably higher.
The mean urine protein excretion was close to the nephrotic range.
Blood pressure measurements at baseline were also similar across all treatment groups.
Here are the mean systolic and diastolic blood pressures plotted over time. Reductions in systolic and diastolic blood pressure from baseline were observed in all three treatment groups. The attained blood pressure levels were clinically indistinguishable in the irbesartan group, which is in yellow, and the amlodipine group in blue. There was a 3.9 millimeter of mercury and a 2.7 millimeter of mercury difference observed between the irbesartan group and the placebo group in pink in the mean systolic and diastolic blood pressures, respectively. While these differences are statistically significant, analyses of the primary efficacy endpoint, to be shared with you shortly, confirm that these differences are not clinically meaningful in this study.
On average, two to four antihypertensives were required to achieve this level of blood pressure control. The most frequently prescribed antihypertensives were beta-adrenergic blockers, central adrenergic agonists, and peripheral adrenergic blockers. The use of all classes of agents was slightly more common in the placebo group. The majority of subjects used either thiazide diuretics or, as renal disease progressed, loop diuretics.
As you recall, the primary efficacy measure is the time to the composite endpoint of doubling of serum creatinine, ESRD, or death. This slide shares the primary results of the study. As seen here, irbesartan in yellow significantly increased the time to the primary composite endpoint when compared to placebo in pink, demonstrating a 20 percent relative risk reduction, with a p value of 0.023. The treatment benefit was apparent as early as 18 months and was maintained throughout the study.
In the secondary comparison with amlodipine in blue, a 23 percent relative risk reduction was observed. Again, this difference was statistically significant with a p value of 0.006. This treatment effect was seen in the setting of clinically indistinguishable blood pressure levels.
To confirm that the blood pressure differences between irbesartan and the placebo groups were not clinically meaningful, the primary analysis was adjusted using blood pressure levels as a time dependent covariate in the Cox regression model. The results for the primary efficacy endpoint were similar with a relative risk reduction of 19 percent and a p value of 0.035.
A similar analysis, adjusting for the levels of hemoglobin A1C, was also conducted, and once again, the results for the primary composite endpoint were similar.
Lastly, the amlodipine group behaved similarly to the placebo group with no observed benefit in the primary composite endpoint.
This slide displays the Kaplan-Meier curves for the renal outcomes, a predefined endpoint consisting of doubling of serum creatinine or ESRD. Treatment with irbesartan in yellow significantly delays the progression of diabetic nephropathy compared with placebo in pink with an observed relative risk reduction of 26 percent. This was statistically significant with a p value of 0.012.
For the secondary treatment comparison with respect to amlodipine in blue, a 34 percent relative risk reduction in favor of irbesartan was observed. This again was statistically significant with a p value less than 0.001.
The Kaplan-Meier curves suggest that the treatment benefit was observed as early as 18 months and was maintained for the duration of the study. The Cox regression analysis confirmed that the observed renal benefit of irbesartan was independent of blood pressure lowering.
Together, these results prove that blockade of the renin-angiotensin system with irbesartan delays the progression of diabetic nephropathy and that these benefits were in addition to blood pressure reduction alone.
Data on the next two slides provides insight into the relationship between doubling of serum creatinine and ESRD.
Patients with proteinuria who double their serum creatinine have advanced to the stage of the disease characterized by progressive and irreversible loss of renal function. This is evident in this analysis showing the cumulative rate of reaching ESRD for subjects who have doubled their serum creatinine. The median time to ESRD, defined as renal transplantation or the need for dialysis or serum creatinine of at least 6.0 milligrams percent, once halving of the GFR has occurred, was 9.8 months and is similar to that observed in the captopril trial of type 1 diabetics, which was 9.3 months.
The relationship between serum creatinine and ESRD is further defined on this slide showing dialysis and transplantation events that occurred in subjects after doubling of serum creatinine or in subjects who experienced ESRD as defined by a serum creatinine of at least 6.0 milligrams percent as a first event. Of the 322 subjects who doubled their serum creatinine, 133, or 41 percent of subjects, underwent dialysis or transplantation during the period of follow-up. In contrast, only 5 percent of subjects who never experienced a serum creatinine event reached ESRD. These results indicate that progressive decline in renal function increases the risk of subsequent outcomes.
Of the 71 subjects whose first event was ESRD, as defined by the serum creatinine, the overwhelming majority, 59 or 83 percent of subjects, went on to dialysis or transplantation, and this occurred in a relatively short time frame. The mean time until dialysis was initiated in these subjects was only 2.5 months.
Based on these results, it is reasonable to conclude that with longer follow-up, all subjects who doubled their serum creatinine would reach ESRD unless death intervened. Furthermore, these results mirror practice in the nephrology community. The standard approach to the treatment of diabetics with advanced nephropathy is to periodically monitor serum creatinine and initiate dialysis once the serum creatinine reaches 6.
The next series of slides portray the results for the components of the primary composite endpoint and the secondary endpoint analysis, cardiovascular morbidity and mortality.
This slide displays the relative risk reductions of the primary composite endpoint and the individual components. In order to assess the impact of treatment on the individual components, all occurrences of that component event were included in the time-to-event analyses. When a subject reached doubling of serum creatinine as an endpoint, coded medication was stopped to allow the study investigator to treat the subject outside of protocol. Thus, the intent-to-treat analyses presented for each of the components include events which occurred in subjects who were no longer on coded medication.
The first panel displays the risk reductions for the comparison of irbesartan and placebo and the second panel for the comparison between irbesartan and amlodipine. The observed benefit of irbesartan, when compared to placebo, was driven primarily by the two renal outcomes, doubling of serum creatinine and ESRD. The consistency of the results are apparent in the comparison with the second control group. The relative risk reduction of 23 percent was also driven by the renal outcomes, doubling of serum creatinine and ESRD.
For all-cause mortality, the point estimates are close to 1 for each comparison, suggesting that treatment with irbesartan had no adverse effect on subject safety.
As you'll recall, the secondary composite measure was time to cardiovascular morbidity and mortality and it was evaluated to assess potential risk in the type 2 subjects given the competing risk of cardiovascular disease and to exclude evidence of harm. There was no difference observed between any of the treatment groups.
The sample size here, less than 600 subjects per arm, was smaller than has been typically required to detect differences in cardiovascular events due to blood pressure lowering using drugs with different mechanisms of action. These results reinforce the benefits of optimizing blood pressure control.
The next slide displays the relative risk reductions of the secondary composite endpoint and the individual components for the comparisons between irbesartan and placebo and irbesartan and amlodipine. Cardiovascular events which occurred in subjects who were no longer on coded medication were included in these intent-to-treat analyses. Furthermore, by protocol, cardiovascular events that occurred after ESRD was reached were not captured because the initiation of dialysis and other therapeutic interventions are known to influence cardiovascular risk factors.
There were no statistically significant differences in the comparisons between irbesartan and placebo for any of the individual events, indicating that there was no overall increased cardiovascular risk associated with treatment with irbesartan.
In the comparisons between irbesartan and amlodipine, the result for hospitalization for heart failure favored treatment with irbesartan. The point estimates indicate directional trends for cardiovascular death, nonfatal myocardial infarction, and stroke in favor of amlodipine treatment. However, the confidence intervals for these risk reductions overlap 1 and did not reach statistical significance.
In view of these results, a post hoc analysis combining the renal and cardiovascular endpoints was conducted to assess the overall benefit/risk of therapy. This combined composite endpoint provides equal weight to both the renal and the cardiovascular events and assesses the time to the first occurrence of any detrimental outcome, whether it be renal, cardiovascular morbidity, or all-cause mortality. In this analysis in which the cumulative event rate approached 80 percent, irbesartan retains its treatment effect compared with either placebo or amlodipine, a 19 percent relative risk reduction compared to placebo and a 21 percent relative risk reduction compared to amlodipine, thus suggesting that the overall benefit of treatment with irbesartan is preserved.
The final segment of this presentation will focus on irbesartan's safety profile. In general, treatment with irbesartan in this patient population was safe and well-tolerated and resulted in few discontinuations. This table presents the incidents of adverse events, serious adverse events, discontinuations due to any adverse event, and death. There were no substantial differences between any of the treatment groups in these important safety measures.
Just to address one of the earlier questions, there were approximately 260 deaths reported in the study. Slightly greater than 50 percent of them were due to cardiovascular events.
The next slide includes those adverse events of special interest that are likely to occur in subjects with renal disease which resulted in discontinuation of study drug: hyperkalemia, inability to control blood pressure, edema, orthostatic symptoms, and the early rise in serum creatinine.
It is well known that agents that interfere with the renin-angiotensin system increase the risk of hyperkalemia due to hypoaldosteronism. Although the incidence of hyperkalemia due to these agents is infrequent in patients with normal serum creatinine, in patients with impaired renal function that continues to worsen, the risk of hyperkalemia will increase.
As expected, subjects treated with irbesartan experienced a higher incidence of hyperkalemia compared with either placebo or amlodipine. This resulted in permanent discontinuation of study medication in 12 of the 577 subjects. Periodic monitoring and appropriate intervention reduced the severity of this electrolyte disturbance. No subject with documented hyperkalemia attributed to treatment with irbesartan experienced death associated with hyperkalemia.
Inability to control blood pressure was a concern in this patient population because of the severity of the hypertension. Discontinuation of coded medication for this adverse event occurred more frequently in the placebo arm.
Edema, requiring discontinuation, occurred more frequently in the amlodipine arm.
Orthostatic symptoms were also a concern because of autonomic neuropathy, and discontinuation of study drug due to orthostatic symptoms was similar across all three treatment groups.
Lastly early rise in serum creatinine, a well-documented risk in patients with bilateral renal artery stenosis treated with ACE inhibitors, only occurred in one placebo-treated subject.
In summary, irbesartan significantly reduced the time to progression of advanced diabetic nephropathy as demonstrated by the beneficial effects on the composite endpoints, renal outcomes and total mortality. There was a 20 percent reduction in the primary endpoint compared with placebo and a 23 percent relative risk reduction with respect to amlodipine.
Importantly, renoprotective benefits of irbesartan were independent of blood pressure reduction.
Finally, in this patient population, irbesartan was generally safe and well-tolerated.
Before I introduce Dr. Parving, I guess I wanted to know if you had any questions.
DR. BORER: Yes, we will, and I don't think we'll be able to complete them all before the break that I'm now told is mandatory for FDA people. Other parts of the Government sometimes tough it out.
DR. BORER: But I'm told that this group can't compete. In just about 5 minutes, we will take a break, so we'll have a few questions first. Then we'll complete after we come back from a 15-minute break.
But I would like to ask you a question now so that, since you may not have the answers readily available, during the break you can try to pull the relevant data together.
Granted that ESRD as the first event was relatively uncommon compared with doubling of the serum creatinine, nonetheless you show us an evaluation with ESRD as first event that suggests that this occurred earlier, not quite significantly earlier, but earlier, in patients who were not on irbesartan than in patients who were on irbesartan. Therefore, since all the patients who doubled their serum creatinine were allowed to receive drugs that are presumed to prevent the progression of renal disease, which I guess would in virtually all cases have included an ACE inhibitor or an AT1 receptor blocker -- you can correct me if I'm wrong about that -- I'd like to know, first of all, what drugs were they put on.
And secondly, what happened to the rate at which ESRD developed in those patients who went from placebo to a presumably effective drug or from amlodipine to a presumably effective drug compared with the rate that was seen before the cut point in the patients who were still on randomized therapy at the time that they hit their first endpoint, being ESRD? That may be sort of complicated and maybe I didn't say it quite right, but I think you get the idea.
If you don't have those data right now, that's fine, but I'd like to know what those results are after the break. Do you have any idea of that right now?
DR. COOPER: No. I'd prefer to take a break and we will compile the data, to the best of our ability, to address your question.
DR. BORER: Okay, that's great.
I'll tell you what. Rather than have at you here, is it okay if we break 5 minutes earlier than you said? Yes, okay.
DR. FLEMING: Given, Jeff, that you're putting on the table issues that we might discuss after the break so they have time to get it, one thing I'd like to see is the numbers of people who had dialysis or transplant, and so specifically two analyses: dialysis-free survival analysis and dialysis/transplant-free survival analysis. I.e., the first being events are either death or dialysis; the second events being either death, transplantation, or dialysis.
DR. BORER: We'll take a break now and we will begin again precisely at 20 minutes of 11:00.
DR. BORER: I assume that the requirement of the FDA that it get a break will also mean that people who work for the FDA want to have lunch. That's another thing we don't often do in other parts of the world. But to be able to stop in time to do that, we're going to have to start right now. So, let's sit down, get together, and begin the questioning of Dr. Cooper.
Where do we want to start here? Dr. Kopp?
DR. KOPP: Dr. Cooper, I had two questions. You may have mentioned this and I may have missed it. But were beta-blockers similarly used in all three groups, the issue being those also have an antirenin effect.
DR. COOPER: The use of all agents was slightly more common in the placebo group. So, the beta-blockage use in the placebo group was approximately 50 patients of the patients; in the irbesartan group, it was 43 percent; and it was a little bit less in the amlodipine group. So, it was slightly more common in the placebo group.
Your second question?
DR. KOPP: The second question actually -- well, I guess I'll launch in -- is the issue that Tom Hostetter raised in the editorial in the New England Journal, which was the noncomparison with ACE inhibitors. I guess one of the issues here is that as a practitioner with a patient with type 2 diabetes, you can look back on the type 1 diabetic study and see that captopril had a 50 percent reduction in doubling of creatinine, or you can look at this agent with a roughly 25 to 30 percent reduction, and you have to choose. Do you go with an agent that might potentially be more potent or go with the agent that has been used in the particular subset that you're looking at, type 2 diabetes?
So, the question would be you must have given thought to the use of ACE inhibitors. Any comment about why that arm was not used?
DR. COOPER: Yes. Can I please have subtalk 2.4?
There's no data in type 2 diabetics with renal disease as to what class of drug, whether or not it interrupts the renin-angiotensin system, could be effective. This was the first trial conducted in this patient population. There was much discussion, especially because we were doing this study in collaboration with an academic group, about the choice of the comparator.
For using a calcium channel blocker as a comparator, there were three points. The first was we wanted to evaluate blood pressure lowering due to a different mechanism of action other than interruption of the renin-angiotensin system. The second point suggested that calcium channel blockers could possibly be renoprotective, and at the time that the study was designed, between 1993 and 1995, there was a fair amount of literature and much discussion about calcium antagonists in all patients with renal disease. And lastly, because amlodipine was the agent of choice for this patient population, we wanted to assess whether or not this drug could be renoprotective.
Specifically addressing your question about why we did not select an ACE inhibitor as a comparator, there were three points. The first was that we would be testing a mechanism of action that's similar. It's similar but it's not the same. With irbesartan, which is an angiotensin II receptor antagonist, you have complete blockade of the angiotensin I receptor. With an angiotensin converting enzyme inhibitor, you have other pharmacological activity, and specifically you have an entire series leading to potentiation of such things as bradykinin that were not yet tested in this patient population. No one could make any assumptions without data and without evidence that type 2 diabetics with this extent of renal disease would do well with an ACE inhibitor.
And lastly, we're just being very pragmatic. Should we have conducted a study with an ACE inhibitor, we would have had to conduct a non-inferiority study, and the sample size would have been prohibitive.
DR. BORER: Let's keep on around this side of the table here. Bev?
DR. LORELL: I'd like to hear a bit more information about the actual strategies that were used in the trial when an increase in creatinine, albeit later found to be transient, occurred. Clearly in the real world, certainly in treating heart failure, the major reason for stopping an ACE inhibitor and probably also ARBs in patients for whom ACE inhibitors clearly reduce mortality is seeing transient rises in creatinine. That will impact the use of your drug in the real world.
What strategies were actually used and what was the mean and median absolute magnitude of transient bumps in creatinine that were addressed and reversed?
DR. COOPER: In answer to your last point, we do not have specific data about mean or median transient increases in serum creatinine. What I can share with you is the protocol that was used.
There were approximately five reasons that were identified that could lead to reversible changes in renal function. The investigators were all instructed to first repeat serum creatinine measurements and determine whether or not any of these five reasons could be contributing to a transient increase in serum creatinine. They then needed to wait an additional 4 weeks before sampling the blood again. If there was still a transient increase, the protocol actually allowed for dose reduction of study drug to determine if there was some dose-related effect. If the increase in serum creatinine was sustained, then measurements from the first aliquot and from the second aliquot were subsequently sent to the central laboratory for confirmation of the serum creatinine. So, on average, there were approximately 4 weeks between the first serum creatinine being drawn and the last serum creatinine being drawn to protect against the possibility that we weren't dealing with a situation which was reversible acute renal failure.
DR. LORELL: But during those 4 weeks, did the investigators embark on the protocol of interventions on those five potential factors?
DR. COOPER: Yes.
DR. LORELL: They did.
DR. COOPER: Yes.
DR. BORER: Blase?
DR. CARABELLO: Let me try to understand better what happened to patients who doubled their serum creatinine. At that point, coded drug was stopped, and they were treated openly and presumably aggressively. How were they then treated statistically? Were they censored from the initial group that they were in, or did they continue on in that group? What happened to them in terms of follow-up?
DR. COOPER: In order to address that question, I'm going to ask the statistician responsible for the results from Bristol-Myers Squibb, Dr. Natarajan.
DR. NATARAJAN: Hi. My name is Kannan Natarajan from Bristol-Myers Squibb. I'm in the Biostatistics Department.
To answer your question, we treated them as intent-to-treat, so we did not actually discard any events that might have happened after they stopped coded medication. All of these patients were analyzed as they were randomized.
DR. BORER: Steve?
DR. NISSEN: Yes. I want to come back to that in a minute.
But first, I wonder if you could put up your slide C-16. Is that possible?
DR. COOPER: Core slide C-16 please.
DR. NISSEN: There are a variety of endpoints listed there, and those of us in cardiovascular medicine tend to think of the hard cardiovascular endpoints as being the composite of cardiovascular death, MI, and stroke. Now, one interpretation of the data -- and I want to see if you concur with this -- is that in the comparison with amlodipine, you saw a 23 percent decrease in the risk of reaching your renal endpoint, but at the cost of, at the expense of, a 36 percent increase in the risk of cardiovascular death, a 51 percent increase in the risk of nonfatal MI, and an 86 percent increase in the risk of stroke.
Each of those point estimates overlap a relative risk of 1, but the hard cardiovascular endpoints, if you lump those together, my guess is -- and I actually did some statistics here myself and Tom probably could do it very quickly. I got a p value of around .01. So, it looks to me like there's actually stronger evidence for an increased risk of hard cardiovascular endpoints than there is evidence for a beneficial effect on the softer endpoint of an increase in creatinine. Is that an accurate reflection of the data?
DR. COOPER: Is Dr. Pfeffer here?
DR. JULIA LEWIS: The FDA has asked for a hold on the question. They're with Dr. Pfeffer right now. Members of the FDA are with Dr. Pfeffer.
DR. COOPER: Okay. Can we come back to addressing your question when Dr. Pfeffer returns? Thank you.
DR. BORER: I wonder if you've had time to look for the data that I asked about earlier and that Tom asked about?
DR. COOPER: I'm going to begin with the second question that you asked specifically about transplantation and dialysis, with the caveat that in order to produce specific slides with time-to-event analyses, et cetera, we're actually putting those together now and we can share them with you probably after lunch. So, if we could start with subtalk 5.8.
The first slide displays the actual number of events that occurred within end-stage renal disease. So, you have the number of dialysis, transplant, and serum creatinine events. This is all events that occurred, not just those that were part of the initial composite component.
If you look at the irbesartan events, you'll see that 77, 73 plus 4, events occurred in the irbesartan group, and if you look at the placebo group, you'll see that 88 plus 6, or 94, events occurred in the placebo group. So that the incidence of occurrence of transplantation or dialysis events was less in the irbesartan group than in the placebo group.
I do not have a risk reduction or a p value for that result, but I would be happy to remind you that in the FDA questions, the actual risk reduction of time to dialysis was included, and that's .8. Confidence intervals do overlap 1.
Given that in this study, the progression of renal disease, you would have needed to follow these patients for a continued length of time in order to observe a statistically significant result. Also, the composite endpoint was a composite, and none of the individual components were powered in order to achieve a significant result.
The next slide. This is the total incidence of the events, dialysis, transplantation. Of course, we include serum creatinine since it was part of the ESRD definition for the components. So, we have 77. This is the same slide. 77 and 94.
Let's move on to slide 361 which is the Kaplan-Meier curve of time to ESRD. For the Kaplan-Meier curve here, this is ESRD. This includes serum creatinine of 6. We're trying to pull together the other Kaplan-Meier curve now. The relative risk reduction for irbesartan versus placebo was 23 percent. This was not statistically significant. It was .07, but it trends in the appropriate direction.
Once again, if you look at irbesartan and amlodipine curves -- excuse me -- if you look at the amlodipine and placebo curves, they're superimposed on each other, indicating that there's no difference in the event rate in those two groups.
DR. LINDENFELD: Could I just add something here? Correct me if I'm wrong about this, but in the captopril trial, there was a 50 percent reduction in end-stage renal disease, and they did not use a definition of creatinine greater than 6.
DR. COOPER: Right.
DR. LINDENFELD: In this trial, all of the difference in end-stage renal disease is in creatinine greater than 6. None of it is in transplants or dialysis. So, there would be no reduction in end-stage renal disease if one didn't use creatinine greater than 6. Is that correct? I believe it is.
DR. COOPER: No, I disagree with that. As I shared with you before, the incidence of transplant and dialysis events was lower in the irbesartan group, and at least for time to dialysis, there appears to be a relative risk reduction in favor of treatment with irbesartan. And that's your last point.
In the comparison of the data between the captopril and the irbesartan trials, there are a couple of points that I think are important to communicate. The first is the relative incidence of death in the type 1 patients who, at the time of the study, were 35 years old and not 58 years old, was very, very different. We had just a handful of deaths, and that's one point.
The second point is in the discussion of the captopril study, in the communications with the FDA as we were designing the study, the feedback that we received at that time was that we needed to have as firm a definition of ESRD as possible.
When you consider that this trial was conducted in 27 countries and the number of investigators, all of whom need to make a decision about when to initiate dialysis, there is no standard in the nephrology community on when to initiate dialysis. We felt very strongly that by including a serum creatinine of 6 or greater as part of the definition of ESRD, we were making that endpoint less arbitrary, and it was a clear definition. That's reinforced by the results that I shared with you earlier in the presentation where the time to dialysis following a serum creatinine of 6 was only 2.5 months.
DR. LINDENFELD: I understand the reasons that you said, and other people may want to comment on this. But in fact if you exclude the creatinine of 6 and use transplant or dialysis, it was 22 versus 24. So, there was not even a trend to a change. I'm just saying this is different.
DR. COOPER: That's incidence.
DR. LINDENFELD: Right.
DR. COOPER: Okay, we need to have the data for time to.
DR. LINDENFELD: Right.
DR. BORER: Tom?
DR. FLEMING: I think part of what you're saying, JoAnn, is my understanding, and I think what Dr. Cooper has said is in part my understanding as well. Let me just get that out and see if we have a consensus here.
In the captopril trial, they did specifically look only at transplantation, dialysis, survival. It did show a 50 percent reduction and p was .006. I'm still interested in knowing what the results of that endpoint would show in this trial, specifically what does transplant, dialysis, death, as a composite endpoint, show in this trial.
I agree with Dr. Cooper. My understanding is the contributions of the elements will be different. In the captopril trial, only 30 percent of those endpoints in the composites were death, although death did show a reduction. There was a 43 percent reduction in the death rate, although it was less of the dominant contribution. There were proportionately more dialysis/transplantation events. Here we would see in the composite endpoint, which I'm still waiting -- we still haven't been shown it -- we will have more dominance by death.
My understanding is where you're right, JoAnn, when you look at time to the primary endpoint -- and dialysis is the first event -- there's no evidence of a reduction there, 24/22. But if you continue to follow people past creatinine increases and look at whether or not this translates into a reduction in dialysis -- I believe it's what we're seeing now, which is the data you're showing us -- there's evidence of a 20 percent reduction, but it's not significant.
But clearly when you get this composite endpoint of transplantation, dialysis, death, that relative risk reduction is going to be a fair amount less than 20 percent and not at all close to the 50 percent reduction of captopril in that corresponding analysis. So, I'd still like to see that analysis.
I'd like to move on to a related point, but did you want to say any more about this?
DR. LINDENFELD: No. Go ahead.
DR. FLEMING: Steve brings up another very key point and that is if we're looking at clinical endpoints looking at the aggregation of clinical endpoints, certainly it's appropriate to focus on those that are renal related. Certainly it's appropriate to look at a transplantation/dialysis-free survival endpoint separately.
But it's also very clinically relevant to say, especially if we're going to compare to amlodipine, what is globally happening here that's really clinically important? When we keep seeing these meta-analyses, we keep seeing the creatinine changes included in those, and of course, they continue to dominate.
There's no question there is a difference in time to doubling. There's no question, and amlodipine doesn't provide that benefit. But if we look at how that translates into true, tangible clinical outcomes, looking first with a focus toward renal, i.e., dialysis, transplantation, death, we haven't seen it yet, but my guesstimate is it's going to be a reduction of 10 to 12 percent relative risk, compared to 50 percent with captopril.
We haven't at all yet seen an analysis that's, in essence, in the spirit of what Steve wants to see, which is let's look at all events that really matter. Let's look at transplantation, dialysis, survival, but then also factor in those important cardiovascular events, such as MI and stroke, where it would appear that there's no longer an advantage over amlodipine. And in fact, it's not clear to me whether there's a disadvantage. It would certainly be important at some point soon to see those two composite analyses.
DR. COOPER: So, if I understand you correctly, Dr. Fleming, what you're requesting is the time to a combined composite endpoint, excluding the serum creatinine events, that focus just on dialysis, transplantation, cardiovascular events, and death.
DR. FLEMING: Indeed, because essentially what we're looking at here is a continuum. What we're looking at in the primary endpoint is an endpoint that is dominated by time to doubling of serum creatinine. We've seen, however, that there's only about a 9-month lag from that endpoint to end-stage renal disease, and in fact a large number of people have achieved end-stage renal disease endpoints. But those endpoints are still heavily influenced by having serum creatinines hitting 6. And we're told that that, in fact, is a trigger for intervention, although interestingly there is some lag in when that intervention occurs.
But if it's in fact a short lag, then we presumably should be fully adequately powered to see the tangible effects. Does this translate in tangible effects in terms of reducing the renal-focused endpoint, which is transplantation, dialysis, death? So, let's look at that composite endpoint, numbers of people that had that endpoint, relative risk estimates.
Then looking more globally, as Steve had pointed out, let's look at the more global clinical consequences, because we've acknowledged that in this setting cardiovascular events dominate what are the bad things that happen to people. So, at least I would like to know what is the relative outcomes in bad things. Take your secondary endpoints and add transplantation and dialysis or take Steve's three endpoints, which are stroke, MI, cardiovascular death, and add dialysis and transplantation, and let's see. There's a lot of data here on these clinical endpoints. Let's see what those results show.
DR. COOPER: Dr. Fisher, would you like to comment?
DR. FISHER: Yes, I'd like to make a few comments. I'm a little bit shocked, for example, to hear Dr. Fleming think that 50 percent in type 1 diabetics, who are not required to be hypertensive, by the way, for that trial, so that a substantial proportion were not -- so, the concurrent therapy was very different and anything you observed could, in part, be related to hypertension as well because it was a placebo-controlled trial. And there were 35 and this is 58. So, I don't really understand the relevance.
A second point I'd like to make -- and I'm not a clinician.
DR. FLEMING: Lloyd, the relevance of what?
DR. FISHER: The relevance of the captopril data in young type 1 diabetics to demonstrate nephropathy, everybody hypertensive type 2 diabetics. I mean, granted we are treating diabetes and things are somewhat --
DR. FLEMING: You can put captopril aside if you wish. The interest in looking at what are the direct clinical outcomes stands on its own as being intrinsically of interest.
DR. FISHER: Just a second. We've heard about the cardiovascular death, and I think that's very relevant. However, the overall death rate is essentially unity, if you take into account deaths from all causes, total mortality. So, I personally would focus on that. I don't think the patient is too concerned about why they died. Well, they're not concerned about why they died actually, I think it's fair to say. The patient survivors are probably not too concerned about why the patient died but whether in fact there is an excess risk. Of course, that point estimate, including everything, is there.
There may be analyses that haven't been run, but certainly the fairly strong trend, when you look at -- and I imagine there are curves. I can't remember. There are umpty-doodle backup slides. Does anybody know if the cardiologist speaker is going to be allowed to speak?
DR. BORER: Not for a bit. There are some issues that have to be resolved first. So, we'll have to hold that.
DR. FISHER: Okay, because that's very important. It's not as if nobody thought of these issues. There's a very nice presentation by a person involved in the classification of that, a card-carrying, well-known cardiologist, who indeed could address these issues and is prepared to address the issues.
But one of the points he makes, in case this doesn't get through, is he was surprised, when they got done, that there were many more renal endpoints than cardiovascular endpoints. Both are very important. And if you put them together, this is a sick population.
But I don't know if the sponsor has every analysis Tom would desire, but there are a number of analyses that can be presented with backup slides looking at those endpoints.
I would only like to point out the study was not designed nor powered for longer-term follow-up. Maybe it should have been. But I think it's a little unfair to say, well, gee, if you didn't reach the components that I personally like, then you know, it doesn't mean much. That's kind of a stretch to me. You may say, well, gee, it was a great trial, but unfortunately it didn't have the best endpoint for the state of the science at this point in time, and I could understand that and that would be somewhat defensible.
But all the additional evidence, while not totally persuasive at the same significance level with fewer events, points that everything does go on as you've seen. You can throw things together and it looks nice and so on and so forth.
DR. BORER: Ray?
DR. LIPICKY: Well, but I guess to my mind there is some relevance of the captopril trial in the sense that there's this elegant schema for understanding the progression of kidney disease, and that if you can look at the captopril trial, you see that there is a clinically relevant endpoint that is easily met, and you should accept the creatinine as not a surrogate but a real thing. And the trouble here is, it seems to me, that the clinically relevant stuff that was measured sort of undermines that basic philosophy. So that although FDA has said doubling creatinine is an endpoint that is okay, FDA may be wrong, and perhaps one shouldn't accept that as a reasonable thing.
DR. COOPER: Can I just make a couple of comments?
We're going to do our best to collect all the data and be able to respond. But there are a couple other comments here that are pertinent to the conversation, and then I'd like Dr. Lewis to be able to comment as well.
The first comment is that, Dr. Lipicky, if you will recall the first advisory committee on captopril, most of the cardiologists at that time were concerned that all of the benefit for captopril was because of its effect on heart failure, and that was driving the results of the study, which is why we felt very much that it was critical to include heart failure in the analysis.
The second comment -- and I think Dr. Lewis will be able to address this -- is with respect to the delay between serum creatinine of 6 and initiating dialysis. There are clear explanations for why that, quote, apparent delay would occur.
DR. EDMUND LEWIS: Well, I've spent the last 8 years discussing the captopril trial, so I don't see why today should be any different.
First of all, I want to make sure that we're all on the same page as far as who was studied in the captopril trial and who is being studied here because the common denominator may be diabetes, but we're talking about two very different trials and two very different populations of patients. So, let me just establish that first and then we can go on from there.
We have slide 10-1, please.
So, as you can see, we have a population of patients that's 24 years older. They are obese, whereas the type 1's were slim. Their blood pressures were considerably higher, particularly the systolic. However, I also want to emphasize that in the trial that we're talking about, IDNT, 100 percent of the patients were hypertensive, and in the captopril trial, 75 percent of the patients were hypertensive. So, 25 percent of the patients in the captopril trial had a very different course, particularly the ones in the placebo group.
In addition, the type 2 patients that we're studying had a significantly worse level of renal function with a mean serum creatinine of 1.7 compared to the patients in the captopril trial.
So, we're talking about two different populations of patients here. They're older. They're obese. They all smoke. They have an enormous history of cardiovascular disease, as you can see, 45 percent having had a cardiovascular event. You couldn't get into the captopril trial if you had a cardiovascular event. And their blood pressure is a problem over years. These patients have chronic hypertension compared to the type 1 patients.
Can I have subtalk 48-6 please? Yes.
DR. LINDENFELD: Dr. Lewis, while you're on that subject, the levels of proteinuria were the same, though, between the two trials.
DR. EDMUND LEWIS: Well, actually those are the geometric means. So, we've had a little interaction here because, of course, I represent BMS today, but we are the collaborative study group, and if you look at our paper, our actual means, not geometrical, but the actual means of urine protein excretion in the irbesartan trial is considerably higher than it was in the captopril trial, just meaning that we had more patients with a lot more proteinuria which kind of evens out when you do geometric means. So, you'll have to take my word for it on this. The patients in the type 2 trial on average had higher proteinuria. We had more patients with massive proteinuria than in the type 1 trial.
Now, in terms of doubling of serum creatinine and ESRD, one thing that I do want to point out to the panel: times change, as well as issues about various diseases. And the reason why our hard endpoint in the type 1 study was death, dialysis, or transplantation -- and my recollection is there were 22 deaths in the type 1 study. It was 14 in the placebo group and 8 in the captopril group, which was not statistically significant, but it was that trend.
And the reason we bundled those in the type 1 study was because when we designed the type 1 study, we included death with dialysis and transplantation because at that time it was very difficult for a patient with end-stage renal disease due to diabetic nephropathy to actually get on a dialysis program. So, we saw those deaths not being as cardiovascular deaths but as renal deaths, which is no longer an issue because, as I say, 45 percent of patients on our dialysis programs today have diabetes.
So, it was a different time, and that's why that design was put in. But I think that it points out that you can't really exactly take even definitions such as death as being identical between the two studies because we're talking about the 1980s as compared to now, and things have changed.
Now, if you look at doubling of serum creatinine, which I hope we have established as being a very important clinical event in this course, which presages end-stage renal disease -- I mean, this isn't an episodic disease. This is a continuum here. So, if you look at that, you can see that in fact in the two arms here, we have a substantial decrease in the likelihood of reaching that milestone.
Now, one of the things about this and the apparently stronger results in the captopril trial is that, first of all, we did not have cardiovascular death as a serious competing endpoint in that trial. People are dying during this trial before they ever have a chance to double or go into end-stage renal disease, for that matter. And in addition, the placebo group in the captopril trial was losing renal function at such a rate that it was easier to show a difference between the two groups because in those days blood pressure was not controlled as rigidly, and that group of patients, the placebo group, certainly was losing renal function faster.
So, in terms of comparing the two trials, from my point of view, having been the PI for both of these trials, the only thing that I think that really can be said about the two trials is that the results for both trials are strongly in the same direction. To compare the numbers from the two trials I really personally don't think is valid.
Now, as far as the end-stage renal disease or death issue, I think that to a certain extent, of course, we've said that we had a composite endpoint. These people could have a renal endpoint or a cardiovascular endpoint, and the cardiovascular endpoints were not statistically significant. Looking at the published data in, say, just hypertensive populations, the blood pressure trialists collaborative meta-analysis, clearly irrespective of what agent you're going to use, it's whether you lower the blood pressure or not that's going to really determine what your cardiovascular events are. So, we're not too surprised about the cardiovascular deaths really determining this.
And as far as the ESRD is concerned, I think that you can see from our data about the way people are moving, doubling serum creatinine, getting up to 6, going on to dialysis, that I think that we're really talking about in this trial an issue of length of follow-up. I can't imagine how we can be talking about end-stage renal disease -- our data not showing that we have a serious effect in altering the course of renal disease because we're altering the course in a very positive way with irbesartan all the way up to renal disease. The only reason we don't have a significant p value with that is because we didn't follow them quite long enough. I think that you can assume that everybody who doubles is ultimately going to reach 6, and then they'll go on dialysis in a very short period of time.
So, in looking at the course of renal disease, not looking at the specific p value numbers at each stage, I believe that what we're showing here is not really different from the captopril trial when you take into consideration patient population and all of that kind of thing.
DR. BORER: Dr. Lewis, before you sit down, can you clarify something for me? I know you didn't collect these data, they're not reported, and it's not going to be a primary basis for decision making, but just so I can understand. When somebody reaches a creatinine of 6, let's say he doesn't get dialyzed, are there lifestyle changes that we can infer would occur? For example, is the diet very restricted? Are there other limitations? Can you tell us something about that?
DR. EDMUND LEWIS: Yes, I would be glad to address that.
The reason actually for the Medicare definition of a creatinine of 6 in this population is that the goal of the nephrologist is to get the patient on dialysis before they have uremic complications because once they start having uremic complications, for example, just pericarditis, the road back is a long road. So, what we're trying to do is to prevent the adverse effects of uremia which are systemic by putting the patient on dialysis before they get any of this.
One of the things that has not come up, which I would point out to the committee, is that the Medicare criteria of a creatinine of at least 6 or a creatinine clearance of less than 15 mls per minute applies to the population of patients with diabetic nephropathy. What that means, which is important to the nephrologist, of course, is that Medicare has no problem paying for dialysis when people have reached that level. They have made the decision that that is an appropriate level. It prevents complications, hospitalizations, nausea and vomiting, further inanition or whatever is occurring because the patient is feeling sick, plus the anemia and all of that. They will pay for that.
Now, if a patient does not have diabetic nephropathy and has advancing renal disease, the Medicare definition of end-stage renal failure is not the same. The Medicare definition of renal failure is now creatinine equal to or greater than 8 or a creatinine clearance equal or less than 10. And the reason for that is it is recognized in the community and by the federal government that patients with diabetic nephropathy are, in fact, sicker than patients with chronic renal failure due to other diseases and, therefore, deserve to be dialyzed earlier. I think that my last statement there probably answers your question.
DR. JULIA LEWIS: Can I add a comment to that? I have an advantage as a younger nephrologist. These are still all the patients in my clinic. At a creatinine of 6, the patients are fatigued. They've lost their sleep cycle, and very importantly, their serum albumin as a key marker of nutrition has begun to fall. The serum albumin is actually the single most important predictor of survival in a dialysis patient. So, they've already begun to have signs and symptoms that they complain of. Within 2 or 3 weeks of initiating dialysis, most of my patients, both diabetic and nondiabetic, will say I feel better than I have felt in a year. So, they've had a gradual decline in energy level, nutritional status, and other factors.
DR. BORER: Thank you.
DR. COOPER: May I just intercede here? One of the other observations has to do with hospitalizations, and Dr. Pfeffer is now available and can address some of the questions we have about the cardiovascular events.
DR. BORER: Let's just follow through on this idea. We'll come back to that. There are several interlocking issues here.
DR. LORELL: I'd like to ask a question that may be a segue to Dr. Pfeffer's comments. I am not a statistician, but I am a cardiologist that deals all the time with incidents of death and interventional cardiology trials in heart failure.
I guess I would like a comment from one of the card-carrying statisticians either on our panel or elsewhere. If you look at the incidence of cardiovascular death in this population, it is actually remarkably low. It's about 8 percent in the placebo group. It may be in part because they're being treated with antihypertensives and the cardioprotective class of drugs of beta-blockers.
So, looking at the incidence of cardiovascular death over a 57-month treatment period, if I were going to design a trial with the primary endpoint of reducing cardiovascular death, I would suspect that would be a trial that would need several thousand people in the treatment and placebo arms. And perhaps before Dr. Pfeffer, representing the company, speaks, we could hear a statistician's comment on that.
DR. BORER: Lloyd, Tom?
DR. FLEMING: If you're asking about whether it would take an enormous trial if one were focusing only on cardiovascular death --
DR. LORELL: That's my question.
DR. FLEMING: -- it would take a very large trial, unless one were targeting a very substantial reduction in the death rate on that.
DR. LORELL: Thank you.
DR. BORER: Lloyd, you had a comment?
DR. FISHER: Yes, you are right. It would take a larger trial. Actually, from what I have been informed, it's not as large as Tom would think probably. And the reason is that the cardiovascular event rate really goes up when the people hit dialysis. Now, I'm not familiar with that literature, but everything that I've been hearing, as we've been rehearsing for this meeting, assuming that's true -- and the independent people brought in here could discuss that. So, if you followed long enough, if you're willing to let a lot of people get to dialysis and so on and so forth, and not feel you had to intervene to prevent that in every way you could, then actually surprisingly not just the death rate but the cardiovascular event rate would go up more than you would think.
DR. COOPER: And in that situation, we would have continued coded medication throughout the study rather than discontinuing it at the first event.
DR. EDMUND LEWIS: If I could address that, just to finish Lloyd's statement, the mortality rate, once a patient reaches dialysis, hasn't changed much over the last several years, and it is much greater in patients who have diabetic nephropathy than it is in patients with other diagnoses on dialysis programs. The one-year mortality for these patients is 25 percent, and the two-year mortality is 50 percent. So, the goal is to prevent the patient from going on to dialysis as long as possible because they're not dying renal deaths, they are dying cardiovascular deaths, and whatever it is about dialysis that does this, these patients do very badly.
DR. BORER: Steve?
DR. NISSEN: I just want to make sure I understand whether any of the cardiovascular endpoints were censored in this trial. Am I or am I not correct? When they reached ESRD, from then on were the cardiovascular events included or were they censored?
DR. COOPER: They weren't captured.
DR. NISSEN: They were not captured.
DR. COOPER: Right.
DR. NISSEN: They were captured or captured and censored?
DR. COOPER: They were not captured. The patients were no longer on study drug, so there's wasn't a safety effect that we were following, and because of the interventions associated with ESRD and the change to the patient's status as a result of those interventions, we did not capture any cardiovascular events that happened once a subject reached ESRD.
DR. NISSEN: Okay. Well, maybe I'll have more to say in the discussion period, but I'd sure like to see that data.
DR. COOPER: That's the design.
Can we have Dr. Pfeffer now?
DR. BORER: Alan, did you have one question first here?
DR. HIRSCH: This may also just relate and maybe Dr. Pfeffer can answer it as well.
In transition again from the balance of renal benefit to cardiovascular benefit, I want to go back to Steve's point on figure C-16 where you see a reduction in heart failure events with irbesartan, but a relatively favorable effect on the ischemic events in the amlodipine group. You've shown us baseline data for many renal parameters. I just want to make sure there wasn't any misallocation or randomization imbalances. Do you have data on clinical coronary disease, myocardial infarction history, heart failure in the three cohorts you can share with us?
DR. COOPER: We didn't collect data at baseline to that level of degree, but the frequency of prior cardiovascular events at the time of randomization was similar in all three treatment groups.
DR. HIRSCH: I saw that. I was wishing to break that down a little bit.
DR. BORER: Marc?
DR. PFEFFER: I'd like to start with an apology for some of the confusion. I am a member of this group, but my tenure was supposed to start after this meeting because I obviously knew I was working on this project since 1995, and I knew this date. When I was invited to join, I asked that my tenure start after this session. And apparently my paperwork went through faster than anticipated. So, I apologize to --
DR. BORER: A first.
DR. PFEFFER: But my history with this trial I think is relevant because it goes back to the design phase. Dr. Lewis and the collaborative group had been working with the sponsor -- and this is relevant to the difference between cardiovascular and renal -- to design a renal study in a patient population that had never been tested with a new class of agents that had never been evaluated.
At that time, I came in and had discussions with Dr. Lewis and the sponsor and said how could you not look at cardiovascular events. That's what will happen with these patients. And he said, Marc, you have to understand. We're getting these people at the point of the spectrum where they're more likely to have renal events, but why don't we prospectively look at cardiovascular events too but as a clear secondary. As a matter of fact, all the alpha in this project is on the renal events. So, this was a renal study known in a population with a high likelihood to have a propensity for cardiovascular events.
Now, given that, the sample size was based on the renal events. So, it was a sample size of approximately 600 per group with three active comparators. So, there wasn't a chance to talk about cardiovascular death.
If I could have the first slide. We built a composite. Why does one build a composite? First of all, this is a secondary endpoint. And we built a composite knowing that with 600 people, three groups, two comparisons, to get a signal that there was an ability to influence a cardiovascular outcome, we would need as many what we thought were clinically important events as possible.
So, as you've heard, it's cardiovascular death plus nonfatal MI, and I would say prospectively we even built in an ECG core lab where the baseline ECG was looked at 6 months, 1 year, and approximately 6,000 electrocardiograms were looked at. Hospitalization for heart failure required a hospitalization and an adjudication committee, as did neurologic deficit, and the amputation was clear, above the ankle. So, we felt this is a smorgasbord of bad news cardiovascular events, and let's see, if we have a signal that in these three active comparators, if we can see something.
To give you an idea of where we stood, we also said 600 might not be enough. Let's broaden the definitions and now let's call this a tertiary. That's clearly a definition of where we are. We're in the exploratory phase, but we didn't want to miss something with this new class of agents in this important population.
So, what we added to what you had seen before was nonfatal MIs called by the site. So, if a site called it, we'd add that. We also added revascularization procedures. We now added heart failure that didn't quite require a hospitalization, but the investigator said I'm not comfortable here. We're going to start an ACE inhibitor or an angiotensin receptor blocker, and we also added a different level of amputation and peripheral vascular procedures.
And the results were surprising to me that with that smorgasbord of cardiovascular events adding all these together, in only 2.9 years -- everyone on active blood pressure control, and blood pressure levels are going from about 160 to approximately 140 systolic -- we had a 25 percent event rate overall. Contrast that to the renal where it's 37. So, Dr. Lewis was right. These people were more likely to have a renal event.
But that didn't mean that we didn't prespecify and look at these things. This is the actual numbers. The most common event that happened to one of these randomized patients who was then followed for a cardiovascular event, the most common event was the development of a hospitalization for heart failure.
If we look at the overall composite, I think the conclusion is that this therapy, these three arms, that there's no distinction in the overall cardiovascular event rate.
Now, again, all the groups are receiving antihypertensive therapy. There's a central committee blindly working with all the investigators to try to get the pressures down, not knowing the assignment, and this was the overall.
Now, when the investigators presented this -- and the first time that was done was in Stockholm at the European Congress of Cardiology -- our conclusion was that there was no difference in this prespecified composite, lumping all cardiovascular events.
When you have a composite, I think it's fair to look at the components for hypothesis-generating information, and we did that. And what that showed was the most important line is the first dot, which is the narrowest confidence interval, which is the overall predefined, and you can see that that is right around the nil, which is what that Kaplan-Meier showed.
But then when you break it down into what were the components, the only thing that really leaves the line -- and we are not making a point of this because it's one component of many -- is this hospitalization, but it's counterbalanced by other factors. The event that we had the narrowest confidence interval, of course, is the overall, and we choose to make the statement that there's no influence on cardiovascular events, some very interesting things here that will need further study.
The tertiary analysis, which is even broader, just confirms what I've just said, and once again, the components go back and forth. Really no difference and nothing that you would say we found something here in this one of six subanalyses in a tertiary analysis, but interesting observations that will require larger studies, which are already underway. There are large studies comparing ARBs to calcium channel blockers. VALUE has approximately 15,000 patients; LIFE has 9,000 patients. That's what's going to be required.
Post hoc for the combined -- you've seen this -- was let's add the renal bad news to the cardiovascular bad news and see if it's a shallow victory. Are we just offsetting those renal benefits by more cardiovascular adverse events? And that wasn't true.
But I think an even more important analysis to some of the points that I've heard raised appropriately today, what about the patient? The patient doesn't care if they're referred to the nephrologist, the neurologist, or the cardiologist if they had something happen to them. This isn't a "who's my specialist here." It's "how am I doing?"
We looked at the hospitalizations. Now, this is also skewed in a way that the data collection stopped at the development of end-stage renal failure. So, censoring from the time of development of end-stage renal failure means that we had slightly longer exposure in the irbesartan group. With that slightly longer exposure, there were fewer hospital admissions and the time in the hospital was reduced. I think that's a global measure.
Now, of interest, the cardiovascular component of the hospitalizations was not changed in these three arms with all active therapy. So, our conclusion would be that although we did not show a measurable impact on cardiovascular disease, we did show a measurable improvement in global health, best measured I think by the total hospitalizations.
DR. BORER: Steve or Tom, do you have any other points you want to make?
DR. NISSEN: I tend to look at these events in more of a hierarchical way, and I guess that's why I focused so narrowly on what we would consider the hard cardiovascular endpoints of cardiovascular death, nonfatal MI, and stroke. I would really like to see an analysis where those hard endpoints are looked at. And the reason I say that, Marc, is that most of the "benefit" on the irbesartan versus amlodipine comparison comes from the hospitalization for heart failure, and we all know that amlodipine tends to produce some peripheral edema and that patients with peripheral edema are much more likely to get into a hospital with a diagnosis of heart failure. So, what you're trying to do is equate a soft endpoint like hospitalization for heart failure with much harder endpoints.
And I really want to know what the statistical significance would be if one looked at -- and I recognize it's exploratory and I recognize it's not prespecified, but in terms of looking at overall benefit, I think you have to look at cardiovascular events in that kind of hierarchical way because they have different importance in terms of the overall benefit to the patient. Do we have such an analysis?
DR. PFEFFER: Well, Steve, I think if we could prespecify the importance of a nonfatal event, then give it a rank, we'd all be in much better shape for designing trials. Your bias is that having a nonfatal MI, you'll do better than getting hospitalized for development of heart failure. Well, there are nonfatal MIs and there are nonfatal MIs, and there are developments of heart failure. And I think that's the whole problem with once you get below death, how do you rank these things. Even with the diagnosis of an MI, sometimes it's a triponin leak versus, wow, this person is not going to get out of their chair again. So, I think that's treading in an area that we can't do within this study or that most studies couldn't do. Therefore, we chose to give you the whole global smorgasbord and let you interpret that.
I think the hospitalizations are a very important component of this.
DR. NISSEN: One follow-up and that is --
DR. JULIA LEWIS: Could I comment?
DR. NISSEN: Sure.
DR. JULIA LEWIS: On the adjudication committee -- and Marc can speak to this too -- we were very sensitive to that issue of peripheral edema associated with amlodipine use that you mentioned. In fact, as we adjudicated the heart failure hospitalizations, we required the patients to have other manifestations such as rales, a chest x-ray that showed pulmonary congestion, wedge pressure. I mean, there had to be more to it than swollen ankles.
DR. NISSEN: Sure.
Let me just ask one more question, and that is I want to know the justification for not collecting the cardiovascular event data once they got to dialysis. I'm very troubled by that because we don't have data that I think we should have.
DR. EDMUND LEWIS: Well, once a patient goes on to dialysis, their caregiver, their environment, everything really changes. Plus, their clinical course changes in a highly expected way. So, that data was not collected because of that, because in fact the way we looked at it, requiring end-stage renal disease was the endpoint here. And the high mortality rate of these patients, while it would be of interest to know the exact number, I agree, but we didn't anticipate that it would be any different than any other type 2 diabetic nephropathy that reached end-stage renal disease. They, after all, had not been on coded medication for some considerable period of time. They may have had their blood pressure controlled better than the average patient, so maybe they had a more benign course. But we did not feel that having detail of that stage of the patient's life would actually contribute meaningful information to what we were studying, and what we were studying was does our intervention prevent the patient from requiring dialysis according to what the course of things would be.
DR. NISSEN: But an intention-to-treat analysis says you continue to collect the data as the endpoints occur. I mean, I think it's an unusual approach. I can understand why you might argue that the data might be censored, but I certainly would like to see the data.
DR. JULIA LEWIS: I just want to make two quick comments to add to the reasons why we chose not to do that in the design committee, and that's because there are two ongoing trials, one sponsored by the NIH and one sponsored by a pharmaceutical company, looking at elements of the dialysis membrane interaction with the patient and looking at phosphate binders and certain things that we use to manage them once they're on dialysis that are thought -- the hypothesis is that those things actually impact on cardiovascular events. So, we really thought this was a fairly contaminated population.
Also, recall we only start out with 1,715 patients at the beginning of the trial. Our other feeling was that there were going to be so few patients for a cardiovascular outcome analysis that actually reached dialysis that it wasn't the appropriate setting in which to do a study in what happens to cardiovascular events in ESRD patients.
DR. BORER: Tom and then Bob and then I have some final questions for you before we break for the FDA-mandated lunch.
DR. FLEMING: There's much to say here. It's in a certain sense philosophically troubling to me because we are -- and I can accept this in a certain sense -- arguing that we need to follow patients long enough to really be able to see the full clinical benefits achieved by an intervention that is effectively extending the time to doubling of creatinine. Yet, at the same time we're hearing, gee, when you get out far enough, there's such a myriad of complicated phenomenon influencing the outcomes of these patients, that we don't really want to capture all of these events because it's difficult to interpret them.
In essence, what I want to understand is what are the true clinical consequences of an intention to deliver an intervention versus not and follow all the patients forward in time. And it may not be possible to expect statistical significance on all the cardiovascular endpoints. That doesn't mean it's not very informative to understand whether there's a pattern here that is suggestive of benefit or lack of benefit. So, it's a simple question.
Marc, you've indicated that you were a bit surprised that cardiovascular events were about two-thirds what the renal events were. Maybe that's what it is. I have trouble knowing whether that's what it is because we stopped systematically following the cardiovascular events at certain points in time. So, it's a little difficult to understand that.
What I would like to see, Marc, about three slides from the end, you threw something up that is getting at, at least indirectly, what some of us have been really struggling to see. Could you put the slide up again that shows the actual number of documented events of each type when we're looking at the secondary endpoint? And I'd like to have this left up for several minutes so at least we can make some notes as we go on to other discussions.
Fundamentally, what I'd like to see -- descriptive or inferential isn't critical to me. What I want to see is what the data show about the difference between the three intervention arms in the fraction of patients that have the more renal endpoints here, death, dialysis, survival. Show me what that analysis is.
And then it is relevant to be able to see more globally how those renal and cardiovascular outcomes pool not that I have to prove statistical significance or not. I'd like to understand what the data show about the actual influence of the strategies here in impacting both renal and cardiovascular outcomes. So, at some point before we get into voting, I'm really hoping someone can put those specific analyses before us.
DR. JULIA LEWIS: Can I make just a quick comment? I know you're cardiologists and I know that heart attacks and cardiovascular deaths are really important outcomes for you. But again, as a nephrologist, I have to tell you whether or not you have to go to a dialysis unit three times a week is also a very important outcome, and if the government ran out of money, 100 percent of those people would be dead without dialysis. So, we don't have renal death because we're rich and fortunate in our country. It's a huge factor for patients. Many of them are more frightened of it than they are of a heart attack.
DR. BORER: Bob?
DR. TEMPLE: I guess I have a couple of observations. Maybe this should be left for the discussion, but it seems to me the discussion is bearing on them.
This was not a trial to describe which the best antihypertensive is. A trial of 40,000 people is attempting to do that. We don't know what success it's having. But you really wouldn't expect a trial of this size to be able to pin down the question of whether amlodipine is better at preventing heart attacks than irbesartan. There are mountains of data on that question. Most of it, I admit, is ACE inhibitors not A2 blockers. But it's obvious that trials go every which way. I mean, a big trial in diabetics -- not so big -- the ABCD trial sort of suggested that calcium channel blockers are death and ACE inhibitors make you live, and then other trials don't show the same thing.
It doesn't seem surprising to me that in people, all of whom are treated apparently appropriately for their blood pressure, you see twists and turns, and I'm not sure how much you can make out of a trial of this size on those endpoints when hundreds of thousands of patients have not allowed anybody but certain individuals to reach a conclusion about whether calcium channel blockers are better or worse. So, I wonder how much one should make of this. So, that's one observation.
The second is -- people have said this repeatedly but I'm not sure whether everybody buys it -- that when you reach a creatinine of 6 or something like that, you are on your way to dying or going on dialysis, although this trial didn't follow that long enough. So, there seems to be a minimization of that because you didn't die or go on dialysis yet. I wonder about that because the contention is at least you're on your way there. If we followed you another year or two, you'd definitely be there. But those are not counted as serious events because they didn't quite happen yet. So, I wonder about that. It seems to me worth discussing. Does any disagree with that?
Then, of course, the other observation is that there are two comparisons here. One is against placebo which actually translates to a wide variety of other drugs, but not including calcium channel blockers or ACE inhibitors or something like that. And that doesn't show this funny thing on cardiovascular events. So, it's not clear what to make of that.
You might say that these data certainly don't tell you you should always use irbesartan instead amlodipine in everybody because those other events seemed to go the wrong way and it's ambiguous on that. But does that interfere with reaching a conclusion about the effect on renal function? And I think those are somewhat separate questions.
DR. BORER: Thank you.
I have three final questions for you before we break. No discussion, just give me an answer if you can, and they'll probably come up again as we go through the discussion of the formal questions later.
I asked you before about what happened to the people once they were taken off their coded drug. That question had several components. First of all, what were they put on? How were they treated after they were taken off the coded drug, number one? And number two, what happened to their rates of progression compared with the rate of progression in the first portion of the trial before they were taken off the coded drug? So, that's one set that I'd like to hear an answer to.
Second, I want to know something about the exclusions beyond that point at which people were taken off their coded drug. There were several other people who were analyzed one way or another that I'd like to hear about.
And third, you made a point about blood pressure differences not being important, and I think it's useful that Dr. Kopp is here because I think that the data that exists might not support that statement and it may be important for us to know about that.
But we'll go through them one at a time. First of all, what about the patients who stopped their coded drug? How were they treated and what happened?
DR. COOPER: Can we have the first slide on concomitant medication on double-blind therapy please?
This slide displays the use of the different classes of antihypertensives in this patient population during the double-blind period. As you recall, earlier I was asked a question about beta-blockers, and you see that the frequency of use was 52 percent in the placebo group. In most of the classes, placebo patients by and large received more antihypertensives.
We do not have specific information about the use of agents once patients reached double of serum creatinine because there's no approved indication and it was up to the investigator to decide what to continue to use. Our feeling is everyone was very committed to maintaining blood pressure control and the use of these agents most likely continued subsequent to discontinuing coded medication.
DR. BORER: So had you replaced the coded medication to maintain the blood pressure? By increasing the doses of these others?
DR. COOPER: I don't have that information. We didn't collect that level of detail of information.
DR. BORER: At some point it would be important to know, because I'd like to know if they were put on ACE inhibitors or ARBs. If they were, you'd interpret subsequent data one way; if they weren't, you wouldn't.
DR. COOPER: In the second slide that I'd like to show -- and I believe that this slide is on an overhead and not on a projector, so if we could have the overhead set up. The reason why halving of GFR as measured by a doubling of serum creatinine was considered a clinically relevant outcome was because the study investigators felt that once you've lost half of your renal function, you needed to allow the study investigator to treat the patient with whatever therapy, even though there's no approved indication, should be used to delay the progression of renal disease.
Interestingly enough, not all investigators put their patients on an ACE inhibitor. I don't have the exact percent, but it's certainly not all. And what this slide shows you is the rate of progression to end-stage renal disease after doubling of serum creatinine in subjects with and without ACE inhibitors following the endpoint. So, with ACE inhibitors is on the lower curve, and there is data here suggesting that if you treat them with an ACE inhibitor, you are going to delay their progression of renal disease.
And subjects who did not receive an ACE inhibitor. And there could have been many reasons for why the patients weren't treated with an ACE inhibitor. These patients could have had severe hyperkalemia because of their progression of disease as an example. The rate of progression was more rapid.
DR. BORER: Okay. That's not the way I would interpret those curves, but I can be corrected by any statistician sitting here. It looks to me like those lines are parallel. They just have a different 0 offset. Am I wrong about that?
DR. COOPER: If you look at the medians that were calculated until ESRD, it is shorter for those without ACE inhibitors, 6.4 months, rather than those with ACE inhibitors. It's 12.9 months.
DR. BORER: Perhaps we need a little bit more evaluation. Lloyd, can you clarify that for me?
DR. FISHER: I agree with Dr. Borer. What he is saying is the offset are the people who at the time they doubled already were at ESRD, according to the creatinine criteria, reinforcing the point these are different populations. But if you put the offset together mentally, it's not nearly as impressive. So, it's not really clear whether there's benefit or not from these data.
DR. BORER: Well, I'm not sure how much we can infer from this, but I would have been happier to see a real difference between the people who actually were put on renin-angiotensin system affecting agents after the coded drug was stopped than not, and I don't really see that. So, I'm not sure what to make of that.
MR. WILLIAMS: George Williams from Bristol-Myers Squibb.
I think we have to be careful in these kinds of interpretations of different therapeutic events for cohorts, as described here. These are certainly not randomized comparisons.
DR. BORER: Right, I understand.
DR. COOPER: I do have one more slide to show and that's the slide that shares the rate of progression to ESRD by treatment group in subjects who were not put on an ACE inhibitor. So, if they weren't treated with an ACE inhibitor or an A2 receptor antagonist, that's the closest we have to looking at whether or not there was some preserved benefit after study drug was discontinued but they had halved their GFR.
So, you see irbesartan in yellow, placebo in pink, and amlodipine in blue. There is no real difference here statistically, but if you look at the trends, the rate of progression for irbesartan seems to be -- I don't want to say similar because I can't show you the corresponding curve before doubling of serum creatinine, but it is less than it is for the other two groups.
DR. EDMUND LEWIS: May I add something?
DR. BORER: Yes, Dr. Lewis.
DR. EDMUND LEWIS: I just wanted to remind the panel of the hyperbolic relationship that I showed you between creatinine clearance or GFR and the serum creatinine because now we're talking about a period along that curve that is at the tail where very small changes in glomerular filtration rate are associated with very large changes in the serum creatinine. So, if you actually wanted to have a valid study of anything, ACE inhibitors or where the patient was randomized first and so forth, those changes in GFR leading to large changes in creatinine on your hyperbolic curve are so large that you would really need a lot of patients to get anything other than the sort of identical curves that we're showing you here.
DR. BORER: Well, perhaps it's just not evaluable because the study wasn't designed to do this, but you've shown us the data.
What about the exclusions? Now, you've told us what happened or what you know about what happened to people after they stopped coded drug when they doubled their serum creatinine. What about the others? There were patients who never received any treatment. There were patients who had ESRD and creatinine doubling at the same time and were counted one way rather than another way. Can you tell us what you did about, for example, the patients who never received treatment? How were they handled?
DR. COOPER: Dr. Natarajan?
There were 16 subjects who did not receive a dose of study drug even though they had been randomized.
DR. NATARAJAN: Again, Kannan Natarajan from Bristol-Myers Squibb.
The 16 patients were analyzed as per the intent-to-treat guidelines, in essence, actually as they were randomized.
Can I have that slide for the 16 patients please?
These are the 16 subjects who were randomized but never got a single treatment, never treated. These 6 patients were on placebo, 2 patients on irbesartan, and 8 patients on amlodipine. All of these patients were treated as if they received study drug and they were analyzed by the intent-to-treat principle.
Some of these patients did have an event very soon after the randomization and were counted as having an event. If we were to do a sensitivity analysis, counting in a more demonic way, the irbesartan subject is the only one who is actually going to have the event. Still, it does not change your conclusion.
DR. BORER: How about the people who were lost to follow-up? There were 13, as I recall, or something like that.
DR. COOPER: There were 8 subjects lost to follow-up for which we did not have mortality status, and we have a sensitivity analysis for those 8 subjects as well.
DR. NATARAJAN: Can I have the slide for the 8 subjects?
Again, 8 subjects were lost to follow-up. We did not get any information on these subjects at the time of the study closure. There were 2 placebo subjects, 4 irbesartan patients, and 2 amlodipine patients. In the sensitivity analysis, we again considered the worst possible scenario in which all the placebo subjects, as well as the amlodipine subjects, didn't have an event. However, all irbesartan subjects did have an event. As you see, the primary composite endpoint is still very similar.
DR. BORER: Okay, that's great.
You also had patients who had some events known but their mortality status wasn't known, and how did you deal with them?
DR. NATARAJAN: Can I have the 19-patient slide?
There were 19 subjects who had variable follow-up. There were 11 patients for whom we had the mortality status known. Most of these subjects had withdrawn consent and the only thing that we know of is actually whether they were dead or alive at the end of the study. One subject died during follow-up and is included in the ITT analysis. Assuming the other 7 subjects had a primary event, this is how -- and again, this is in the worst case scenario which is highly unlikely to happen in the sense that it's more of a demonic way of looking at it. The placebo subjects and the amlodipine subjects didn't have any event. The irbesartan subjects alone had an event, and this is how it will turn out to be.
DR. BORER: At least we have the data in front of us, and I appreciate that.
The final question before we break. You suggested that although there was a 2 to 3 percent difference in blood pressure between the placebo group and the irbesartan group -- forget for a moment the amlodipine group because I'm going to suggest to you that that issue may or may not be relevant since we haven't considered the possibility that amlodipine might do something bad. But if you just think about the placebo patients versus irbesartan, there was a 2 to 3 percent difference in blood pressure favoring irbesartan, and you suggested that though that was statistically significant, it wasn't clinically relevant.
About 6 months ago, we sat at a meeting listening to data from ALLHAT, and there was a rather formidable presentation, suggesting something very different from what you said, that is, that 2 to 3 millimeters of mercury could account for a lot of difference. And I wonder if either you or someone from the committee who's familiar with ALLHAT or with the relevant data here can talk about that a little bit.
One might infer that the better results in the irbesartan group versus the placebo group had something to do with the difference in blood pressure control rather than some independent effect of blockade of the renin-angiotensin system. How would you respond to that?
DR. COOPER: Well, that was the reason why we did the Cox regression analysis using blood pressure levels during the study to adjust for the primary composite endpoint, and in that analysis, the relative risk reduction, 19 percent, is similar to what was observed without that analysis. It was 20 percent.
I guess I'm interested in your comment about comparing the amlodipine and irbesartan group because amlodipine could have been doing harm. One of the points is that the amlodipine event rate was similar to the placebo event rate, and it is our interpretation that it is unlikely that amlodipine was doing any harm with respect to this composite endpoint.
DR. BORER: You may well be right. I'm cognizant of the fact -- and in fact I had come to the same conclusion that Bob stated -- we had two different comparisons here, and we're asking several different questions.
I don't want to lose my train of thought here before we close. Yes, I do remember now.
I don't know technically how one makes adjustments with the Cox model. I don't know how valid it is to say there was 20 percent and 19 percent and whatever. What I would be willing to accept is that there is an independent effect of treatment even when you consider blood pressure differences, which I assume is what you found. Maybe you can expand on that.
DR. NATARAJAN: Yes. Can I have slide 354?
What we did is basically address the issue of the differences in the blood pressure between the treatment groups whether it's clinically relevant or not. From a statistical point of view, we adjusted in a time-dependent manner and these are the results of the analyses, both unadjusted, as well as adjusted for time varying mean arterial pressure. As you can see, the risk reduction change is very small, from 20 percent to 19 percent, and the significance still exists. And with regard to amlodipine, we did not see any difference in the blood pressure, and thus the estimate did not differ, nor does the p value.
DR. BORER: Tom, can you comment on this?
DR. FLEMING: Well, I think what's been attempted here with the time varying covariate is a reasonable approach. The question is how interpretable or convincing is it really.
Essentially -- and I assume this is what you've done, although there are lots of variations to how you might do this -- what you're saying is we know at baseline that blood pressure is predictive of risk of many types of events, renal and cardiovascular. So, what we'd like to do to fully capture that influence, particularly if there's a difference in the blood pressure profile over time across two different regimens, is put a time varying covariate in that says anytime there's an event, what is that person's blood pressure at that point and adjust for blood pressure, not just at baseline but as it's varying over time.
DR. NATARAJAN: That is correct, yes.
DR. FLEMING: And that's a very reasonable approach to take here.
There are some pretty significant assumptions we're making, though, and that is the way in which blood pressure truly is influencing outcome is fully being captured by whatever that latest measured blood pressure was at that point.
I've attempted these kinds of adjustments in other trials in which we have seen evolving differences in outcomes and evolving differences in blood pressure levels, and we haven't also been able to explain these differences by the time varying covariate analysis. So, I consider what they've done as a reasonable approach, but it certainly doesn't reliably allow us to conclude that there are not differences in these event rates that could well still be impacted by the difference in blood pressure control between the arms.
DR. JULIA LEWIS: If I may add something. I'm an investigator in the African American study of kidney disease and hypertension, which has been presented at the American Heart Association, and I sit on the writing committee.
We in that NIH-sponsored trial randomized African Americans with kidney disease and hypertension to a mean arterial blood pressure of 102 to 107 versus less than 92. We achieved between a 10 and 11 millimeter mercury difference in mean arterial blood pressure. By any measure of renal function, including time to event and iothalamate GFR, we were unable to demonstrate a beneficial effect of being randomized to the lower mean arterial blood pressure group of less than 92.
Although that's a different group -- it's African Americans with high blood pressure and kidney disease -- I thought I would share that piece of renal data with you, which may suggest that the renal bed is somehow perhaps different than the cardiac bed in its response or that we're in a range of the continuum where it's less of an impact.
DR. NATARAJAN: I would like to just add one more thing. Whether or not we adjust and do this time-dependent analysis, the thing to keep in mind is that there was no difference with respect to amlodipine and irbesartan, and that would actually suggest that that is independent of the blood pressure lowering.
DR. FLEMING: Although you're making assumptions there about what other mechanisms of action could differ between the two that might offset a difference that would be attributable to blood pressure lowering.
DR. BORER: I think that's been a very informative presentation. I really want to thank you, Dr. Cooper. You've been very clear and concise and given us a lot of numbers.
DR. COOPER: I'm a nephrologist.
DR. BORER: Yes. Well, when I was in medical school, our physiology department was primarily skewed towards renal physiology because Robert F. Pitts was the chairman. Knowing that I would be a cardiologist when I grew up, I was very excited when one of the teaching fellows said that he had a grant from the American Heart Association. So, I said, what are you doing relative to the heart? He said, nothing of course. The only purpose of the heart is to pump blood to the kidneys. Everybody knows that.
DR. BORER: So, I understand what you're saying here.
In any case we will take a break until 1 o'clock when public comment will be possible, and then we'll finish the last two formal presentations and go on to the questions.
(Whereupon, at 12:10 p.m., the committee was recessed, to reconvene at 1:00 p.m., this same day.)
DR. BORER: We'll begin again.
The meeting is open for public comment. There were no applications for public comment, but are there any individuals who have comments that need to be made?
DR. BORER: If not, we'll go ahead. We have some additional questions and I believe the sponsor has some responses first to Tom Fleming's questions, and once you do that, we have some more questions from JoAnn and from Tom before we get on to the next phase of the presentation.
DR. COOPER: Thank you.
What I'm going to present now are the data to respond to the remaining questions. Can I have the first slide please?
Part of the earlier discussion focused on the effect of treatment on end-stage renal disease, specifically transplantation and dialysis. I did want to share with you the results of the time-to-event analysis for end-stage renal disease, including serum creatinine of 6.0 milligrams percent and greater.
In this Kaplan-Meier curve in which irbesartan is displayed in yellow and amlodipine and placebo are in blue and pink, respectively, a treatment effect was observed with a 23 percent risk reduction in favor of irbesartan. Once again, this is all patients who had an ESRD event even after they discontinued study medication.
Next slide. This is the first of the analyses that was requested, time to dialysis, transplantation, or death comparing placebo and irbesartan. And in this analysis, the relative risk reduction was 13 percent, the confidence interval between .7 and 1.09.
The next slide --
DR. FLEMING: Could you leave that up for a few seconds please?
DR. COOPER: Yes.
DR. FLEMING: Thank you.
DR. COOPER: The next slide contains the time to event for dialysis and death. Once again, a relative risk reduction in favor of treatment with irbesartan was observed, 11 percent. Confidence intervals are on the slide.
DR. COOPER: Next slide please. This is the analysis of time to event of ESRD or death. In this analysis, serum creatinine of 6 is included. The relative risk reduction in favor of treatment with irbesartan compared to placebo is 15 percent.
DR. COOPER: And lastly, the new, redefined, combined composite endpoint, which includes dialysis, transplantation, death, and cardiovascular events. The relative risk reduction in favor of treatment with irbesartan observed is 14 percent. It's not expected that any of these post hoc analyses would be statistically significant. They're not powered to detect differences between treatments for any of the components.
DR. FLEMING: Do you happen to know what the amlodipine total is?
DR. COOPER: We do. That's slide MC-128 please.
For the same combined composite endpoint, the relative risk reduction was 10 percent in favor of treatment with irbesartan.
DR. FLEMING: This is much of what we wanted to see. I guess the last was this specific analysis, but only driven by the death, dialysis, and cardiovascular. Did you happen to do that?
DR. COOPER: Death, dialysis, transplantation, cardiovascular events. This does not include a serum creatinine of 6.
DR. FLEMING: This does not.
DR. COOPER: No.
DR. FLEMING: Okay.
DR. COOPER: Our conclusions from these results are that treatment with irbesartan is renoprotective. When you consider this patient population and the fact that you need anywhere between two and four antihypertensives to optimize their blood pressure control, we feel that these data support the use of irbesartan to protect the renal function, but we also recognize that this does not exclude the use of another antihypertensive to protect the heart.
DR. BORER: Thank you very much.
JoAnn, you had several additional questions.
DR. LINDENFELD: Just to make sure I understand here, when a patient was admitted for heart failure or admitted for another event, was that creatinine included in the events? For instance, if a patient was admitted for heart failure in between the routine evaluations and their creatinine had doubled, was that then evaluated as a doubling of creatinine assuming that it was reproduced?
DR. COOPER: All events were adjudicated by the outcome confirmation classification committee. If a patient had two events at the same time, they would have been adjudicated independently of each other, and all of the criteria would have had to have been fulfilled. Certainly there could have been -- and probably were -- a number of patients who, because of their compromised renal function due to diabetic nephropathy, at the time of the hospitalization where there are other insults to their system, if you will, could have been pushed over into end-stage renal disease at that time and had incurred doubling of serum creatinine.
DR. LINDENFELD: I guess what I'm asking is whether or not patients who had worsening heart failure were sampled more often and they're more likely to reach the endpoint of doubling of creatinine earlier because there was this significant excess of heart failure. That could have made a difference of a number of endpoints.
DR. JULIA LEWIS: Let me see if I can answer the question for you. In order for a patient to reach a doubling of serum creatinine, they had to have a persistent doubling of their serum creatinine on two measurements up to 4 weeks apart. During the time interval between the first measurement of doubling of serum creatinine and the second, the investigator would be encouraged to treat any reversible causes of the rise in serum creatinine.
So if, for example, a patient was in the hospital with heart failure and had a transient rise in their serum creatinine, when they came to their next study visit, the study coordinator would be checking their serum creatinine as part of that visit. If at that visit, the serum creatinine was doubled, it would then fall into the usual path of confirming the doubling.
Did that answer your question?
DR. LINDENFELD: I think so.
I'd like to just get back to this issue of race just a little bit. Could you show me what the doubling of creatinine was in blacks versus whites? I recognize there was a small number of blacks in the study, but is there a substantial difference?
The reason I ask that is there were 98 doublings of creatinines in the irbesartan group and 135 in the placebo, so a difference of 37 patients. There were 78 black patients in the placebo group, 63 in the irbesartan, and 87 in the amlodipine group. This is several percentage points difference because the total number in each group was slightly different.
So, I'm just concerned. If, as we've heard, the progression of renal disease is significantly different in minorities and there's a difference in minorities, whether or not that's a significant point here.
So, I think to just start that, could you tell me was there a difference in the doubling of creatinine in black patients compared to white patients?
DR. COOPER: The answer is yes. Do we have the subgroup analysis specifically for blacks versus whites?
The subgroup analysis shows that the point estimate for doubling of serum creatinine favors treatment with irbesartan. The effect in white patients is greater with a relative risk reduction of 25 percent compared to nonwhite patients with a relative risk reduction of 5 percent. This is a subgroup analysis with a very small number of patients, and once again the results are going to be driven by the number of patients.
DR. LINDENFELD: It doesn't look like there's enough difference in the baseline to make a difference.
But there is an under-representation of, I think, blacks in the irbesartan group compared to both the placebo and the amlodipine group. There's under-representation in the entire trial, but I think there's about a 5 percentage point difference in the number of blacks here. It just concerns me because if blacks are likely to progress at a higher rate, then a small difference could make an event difference of 10 or 12 in that group, which could change the total number of 37 events substantially. Again, I think this is a problem in this kind of trial of not stratifying for the groups that are more likely to progress.
DR. COOPER: Dr. Lewis, do you want to address the rate of progression of renal disease in black patients?
DR. EDMUND LEWIS: Well, not any further than what I've said before. I don't think we have information about the rate of loss in black patients in the study in terms of delta creatinine clearance for blacks versus whites. So, I really can't expand very much on what was said before. I think that here you have this data, and I don't think that there's any further that I can say about that, although I will point out that the effect of race on the outcome -- clearly patients who are white had more irbesartan effect, if you will, than blacks. But I really can't comment about your other point.
DR. JULIA LEWIS: I think I can comment a little bit more about it. Can I have my slide 1?
This is true that if you look at the age-adjusted incidence of ESRD, African Americans have a higher age-adjusted incidence of ESRD based on the USRDS data.
Also, there's data out there that suggests that an African American with high blood pressure and kidney disease compared to a white person with high blood pressure and kidney disease, between the ages of 20 and 45, has a 20 times increased risk of developing kidney disease.
However, in the African American study of kidney disease and hypertension, sponsored by the NIH that I alluded to earlier, 1,094 African Americans were randomized in a three-by-two factorial design. I've already commented on the results of the blood pressure randomization. I will comment that the ACE inhibitors protected their kidneys, but I would also like to comment that the average rate of decline of renal function in this study in African Americans with high blood pressure and kidney disease was 2 mls per minute per year, which would mean that if you started out with a normal kidney function, a GFR of 100, it would take you 50 years to reach end-stage renal disease. So, some of our conceptions based on epidemiologic data on the rate of decline of renal function in African Americans may reflect the fact that unlike in clinical trials, their blood pressure is not under exquisite control.
DR. LINDENFELD: I don't think that helps me much because it isn't a diabetes trial.
Again, actually what you've said concerns me just a little bit more because if there's an over-representation of blacks in the amlodipine and the placebo groups compared to irbesartan and blacks progress faster, then to me that biases the study a bit in favor of irbesartan.
In addition, if irbesartan doesn't appear to work in blacks and it does in whites, from the data you've shown me, then that's an additional problem more than just the fact that renal function progresses more rapidly in blacks. So, that data actually concerns me.
I'm really concerned because these are not very large numbers. I'd be interested. Maybe I'm overdoing it here, but doubling of creatinine is everything here, and there's only 37 patients difference in the doubling of creatinine. Yet, between amlodipine and irbesartan, there are 24 more blacks and 15 more between placebo and irbesartan. You know, just a few patients here makes a difference between a significant and a nonsignificant study.
DR. NATARAJAN: I agree, but I wanted to actually caution the committee in terms of actually over-interpretation of subgroups of very small sizes. Again, yes, actually there are very few blacks, and the differences among the treatment groups, though numerically there, they were not statistically significant. They were not actually any different between any of the treatment groups.
DR. LINDENFELD: Well, I agree with that, but I'm not talking about that. What I'm talking about is the fact that it seems to me from a cardiologist's standpoint, if you just go back and review the reviews, that blacks progress more rapidly and other subgroups too. Maybe Hispanics. But it appears that the literature suggests -- you may have some data that I don't have -- that the progression is more rapid. Again, just help me with this. Maybe you can explain why that's not a concern.
DR. FLEMING: I think JoAnn is talking more specifically about race in its role as a predictor and hence a potential confounder rather than as an effect modifier, although both are issues. But if in fact it's a predictor such that blacks would have a poor outcome and they are under-represented in the intervention group, then you would have some level of confounding. It's not a serious imbalance, but her point is the strength of evidence here is marginal at the level of significance, so even a minor imbalance could somewhat compromise the convincingness of results.
DR. LINDENFELD: That's exactly right, but it could also be an effect modifier if in fact irbesartan -- there are not enough numbers -- but if it's less effective in blacks than whites, then it becomes an effect modifier too in a sense. They are small numbers, but the study is based on very small differences in numbers.
DR. EDMUND LEWIS: I don't know about helping you. All that I can say about the preconceptions of how much faster or how more malignant a course patients with type 2 diabetic nephropathy have does not take into consideration the fact that these patients do not have their blood pressure controlled to the recommendations that are made. And I think that in this study, we get the closest that has come actually to blood pressure control recommendations in both the hypertensive and type 2 diabetic population.
So, I think that the assumption that because somebody is African American, they therefore are going to have a much more malignant course is really not exactly an accurate assumption. It's based on literature where the black population, the African American population, has more problems with blood pressure control than does the white population.
I think that once you get into the subgroup analyses, you're getting into small numbers. I personally don't think that our outcomes are that small of numbers. But when you get into the subgroup populations, you are getting into small numbers, and I think that before you make certain assumptions, you have to have the data.
And I don't really believe that you have data showing that the black population -- the natural course of their diabetic nephropathy is that much worse. The natural course of diabetic nephropathy, as we're showing you, is bad and that's with blood pressure control. That's in everybody.
So, I think there's a certain assumption here. I just don't think we can go any further with it because I don't think that there's a number we can put on it. I don't think that you can say because somebody is African American, their chances of doubling is 1. something compared to a white because that information really doesn't exist.
DR. LINDENFELD: No, I agree, and I certainly don't pretend to be an expert. But if one just goes to the reviews, the reviews all strongly suggested that progression is substantially faster without specific data. But the reviews suggest, from studying the data, that it's faster in blacks and certainly American Indians and perhaps non-white Hispanics.
DR. EDMUND LEWIS: Well, I would be glad to go through whatever data you're picking out of your reviews because I think that I will have no problem showing you that there is a blood pressure issue.
I will say, though, that the Hispanic population and the best study population which is relevant here is the Pima Indian population. The NIH has supported longitudinal studies. 50 percent of the total Pima Indian population gets type 2 diabetes and most of them get nephropathy. And the time from birth to onset of diabetes is very well-known. From the onset of diabetes to proteinuria is very well-known. From proteinuria to decreasing renal function and end-stage renal disease is very well-known in this patient population. They've even had serial renal biopsies, frankly. And I think that you will see, if you look at that population, which genetically really represents the problem in Hispanic populations in North America, that the curves to those various events are the same as those reported from Germany in type 2 diabetic nephropathy.
So, when you start studying populations carefully, controlling for things like blood pressure and so forth, you won't necessarily come up with the conclusions that you get out of reviews. That's all that I can say.
DR. BORER: Bev and then Dr. Kopp?
DR. LORELL: Thank you.
In a little different subgroup, I'd like you to address the issue of the subgroup of women. The point estimate for women for the primary endpoint is even closer to unity than for non-whites, .98. And perhaps you can address that for us.
DR. COOPER: Yes. Dr. Breyer Lewis?
DR. JULIA LEWIS: I'm going to first remind you of the statistical results, and then I'm going to comment on putting it in some perspective.
First, I would remind you that we were not powered for exploratory subgroup analysis. There were, for example, nearly 900 white males in this trial and only 91 black females.
However, as you can see from this, the common point estimate of .8 crosses all the confidence intervals, suggesting a common risk reduction between males and females.
Similarly, the confidence intervals overlap, again suggesting that there's not a statistically significant difference.
Lastly, when looking at multiple subgroup comparisons, it's important that the point estimate favors irbesartan. That's the statistical response to that question.
In terms of putting it in perspective of what's known about the impact of gender on women in hypertension and renal disease, I would first remind you of the hypertension studies. I'll remind you that when the first studies were done, examining whether or not treating people with hypertension with basically beta-blockers and diuretics versus placebo was of benefit. The first three trials, the MRC done in England, the hypertension detection and follow-up program done here in the United States, and the Australian therapeutic trial, when the subgroup analysis was done in women, not only did they either not find a benefit or actually found that the women had a higher mortality rate in the treated group.
Now, although there was concern expressed when these studies were done, no one at that time advocated strongly that women should be withheld from the treatment of hypertension. Subsequently, the INDANA analysis, which has incorporated seven clinical trials looking at the treatment of hypertension, has concluded what in fact has been I think common medical practice, and in fact women benefitted from the treatment of their hypertension, although not in all the categories as did men, but in key categories, including main cardiovascular events.
In the area of renal disease, I'll just review only clinical trials that have in the definition that you would accept that are randomized, double-blind with sufficient numbers of patients enrolled to have power to look at the group as a whole. But again, the analysis of the impact on women is a subgroup analysis in each of these trials. I also selected trials that have outcomes similar to the one used in IDNT.
The first, of course, is the captopril trial, which you are familiar with, in type 1 diabetics. Males and females had equal outcomes. Males did not have a worse rate of decline of renal function, nor was there any difference in efficacy of the ACE inhibition for males.
A study done in nondiabetic patients with proteinuric kidney disease in Europe found that males had a worse outcome and that there was worse efficacy of ACE for males. Excuse me. It was better efficacy for ACE in males.
Another study done in nondiabetic proteinuric patients in Europe. Worse outcome for males, no. But better outcome for males.
If you look at the MDRD, which was not a study, looking at an intervention with a specific antihypertensive agent, but at other interventions, males had worse outcomes.
So, in fact, the subgroup analysis in the available renal studies is all over the map, suggesting that perhaps all of these studies are not powered for these exploratory subgroup analyses.
DR. KOPP: I just wanted to make a comment about the progression of diabetic nephropathy in blacks versus whites. My understanding is very close to what Dr. Lewis said, that in the setting of diabetes, blacks are something like 2- to 3-fold more likely to develop nephropathy, but I'm not aware, although he may have come across something that I don't know about, that once diabetic nephropathy appears, that the rate of progression is different.
DR. BORER: Why don't we move ahead with the presentation, and we'll get to any residual -- we have another speaker down at the end of the table.
DR. TEMPLE: A short question. Going back to the possibility that the greater number of blacks in one group influenced the results, isn't that answered by the breakdown into the black and white populations that we saw in which the effect was larger in the white population? That doesn't seem to be compatible with the whole result being driven by the excess of blacks. Isn't that right?
DR. LINDENFELD: If you think that the progression -- I don't want to make a huge issue of this. I just think these are small numbers. But if you assume that progression is greater in blacks and there are more blacks in the non-treatment group, that group would progress faster.
DR. TEMPLE: No. I agree with that. But then they showed separate results for the white population than the black population. The effect, if anything, was larger in the white population.
DR. LINDENFELD: Right.
DR. FLEMING: One has to be careful, Bob, in keeping these issues of confounding and effect modification separate. If you look at page 73, for example, in the sponsor's briefing document where they present this summary, if you look in the control arm, non-whites have a somewhat higher event rate, 43.5 percent, from whites at 37.3 percent. And as a result, if you end up with some excess of whites in the intervention arm, then there would be a small level of confounding. I don't think this is a very large risk of confounding. It would be very modest. But I think JoAnn's point was in a setting where the significance is very close to the border, this could have some influence.
A separate point entirely, also a relevant point, is is race also an effect modifier, not only is it a predictor such that it appears that non-whites have a somewhat higher rate, does treatment effect differ by race, which is an entirely separate phenomenon from whether race is a confounder. And it also appears here, that in addition to it being a potential confounder because non-whites have a somewhat higher risk, it's also true that the effect seems to be greater in the whites than it is in the non-whites. Two separate issues.
DR. BORER: Without --
DR. NATARAJAN: Could I address that?
DR. BORER: No. Just one moment because I think we may have gone as far as we need to go with this.
There may be some differential effect based on race. It seems plausible to me. We know that renin levels are, by and large, a lot lower in the black population with hypertension than in the white population with hypertension, and here we're blocking the renin-angiotensin system.
I don't want to go into the realm of speculation. These are the data. Indeed, there may be some comments, if the drug is judged to be approvable, about labeling issues or maybe not. But let's deal with that when we get to the questions, and let's hear the remainder of the data that we're going to hear.
DR. FLEMING: Jeff, there were two quick questions that I wanted to raise.
DR. BORER: Sure, absolutely.
DR. FLEMING: If I could go back to the sponsor's presentation, slide C-13. What you had done there is you had broken out the relationship between having had a creatinine event versus dialysis or transplantation. Surely there is, obviously as this shows, a relationship, as we would fully expect. It's interesting that 26 percent of these events of dialysis or transplantation occur in that right-hand column, people that didn't have a serum creatinine event.
I guess I have two questions. One is trying to understand why a quarter of these people would have gone on dialysis or transplantation without having had a creatinine event is question one.
Question two is how do those 69 break out by intervention arm? Hopefully not with more of them on the irbesartan group.
DR. COOPER: The answer to your first question is, by and large, when patients presented requiring dialysis, it was because they either missed visits and so we were no longer able to measure their serum creatinine and then assess whether they had had a doubling or an ESRD event, as determined by the serum creatinine, or the other most frequent reason was because these were patients who were on a rapid slope of decline of renal function, hadn't yet doubled, or achieved a serum creatinine of 6, were hospitalized because of an intercurrent and severe illness that compromised whatever remaining function they had left and pushed them over into permanent end-stage renal disease.
DR. FLEMING: So, there's not a fully consistent relationship, at least in terms of documenting a doubling versus having dialysis occur. There are a substantial number that will actually have dialysis occur before you document.
Can you tell us how those 69 broke out in the three groups?
DR. COOPER: I honestly do not remember. Do we have that data?
DR. FLEMING: If you don't have it now, I can wait and we can get it at the end of the next presentation. It would be useful to know that.
DR. BORER: Yes, let's try and get it.
Just a clarification in response to Tom. If I understood what you said correctly, it's not that there may not have been doubling or far greater than doubling of creatinine in many of these patients who didn't have a doubling event or a 6.0 creatinine before they went on dialysis, but that they went on dialysis in the context of a situation which was not the time at which these events were measured and therefore they wouldn't be captured that way in the data set. They may well have had a creatinine of 6 in the context of another disease, but you didn't count it that way. They just had to go on dialysis. Am I correct in saying that?
DR. COOPER: Yes, that's correct.
DR. BORER: Steve?
DR. NISSEN: I don't want to belabor this, but we talked about other subgroups. I was very struck by the North America versus non-North America data, and I would really like a comment because we've seen an awful lot of studies, particularly recently, where drugs didn't seem to have any effect in the North American population but did in the out-of-U.S. population. I personally am troubled by that, and I want to know if you have any comments or thoughts or can you help me understand why there was only a 5 percent point estimate for the North American population.
DR. COOPER: Part of that is driven by the fact that all of the black patients that were enrolled and randomized were actually in North America, so a number of those events were in that subgroup. There is no other biological explanation for why the rates of progression would be different or the treatment effect would be different between the different regions. Certainly in Latin America and in the Pacific region, the event rate that was observed -- and there are minority populations in those regions -- was very consistent with what was seen in Europe.
Any other comments?
DR. BORER: Dr. Temple.
DR. TEMPLE: Some people have expressed some degree of nervousness about not having a sort of ultimate endpoint on the people who got their creatinine to 6 or doubled it or something like that. A fair amount of time has now elapsed since you published and collected data. Would it be of interest or a possibility to find out when all the people in the trial went on dialysis?
DR. COOPER: We could certainly do that for all subjects who didn't withdraw consent and for all subjects that didn't participate in a site that was closed. Actually ascertaining dialysis or ESRD is more difficult than mortality because with mortality you can just access a death certificate. But yes, we can do that.
DR. TEMPLE: Even if the site is closed, they could find out when the person went on dialysis. I'm not saying it would be easy and I'm not even saying it's necessary. I just wondered if you could do it.
DR. COOPER: Yes, we can make an attempt to do that.
DR. FLEMING: Just a follow-up on that. Why wouldn't that be a compellingly obvious thing to do in the sense that what we're hearing is clearly there is a real relationship between creatinine elevation and dialysis, the latter being an obvious clinically important endpoint, the former being at least debatable as to the level of surrogacy that it actually presents? But the answer is there in this database, and the answers that we have in this database are marginal, even if you focus on the primary endpoint, and if it's just a matter of time, which we keep hearing, then wouldn't it be potentially very informative to know what an updated data set would say?
DR. BORER: We could ask the sponsor to do that.
I just want to make one quick comment and then we must move on. We shouldn't get into a debate here. I have no problem with recognizing that somebody with a creatinine of 6 is sick. I'm just a cardiologist, but even I know that. They don't feel well and Dr. Lewis actually recited the problems that are associated with creatinine at that level left untreated by dialysis. So, I don't think we should spend too much time talking about whether these people are sick or not. If they're not dialyzed, they're real sick, and we can debate that later when we go through the questions. But I don't think that's a key issue.
DR. JULIA LEWIS: Could I expand on that for just one second? In type 2 diabetes, there are 135 million. By the year 2025, there are going to be 325 million, a 100 percent increase in the third world, if you'll forgive me for referring to it, the Asian countries. For them an elevated creatinine is death. We wouldn't be able to go count them going on dialysis later. They're dead because it's not an available therapy.
DR. FLEMING: Then we should be able to see in those people a survival in that as well.
DR. BORER: We should.
DR. COOPER: Right. We will make every effort to collect that data.
It is with great pleasure that I now introduce Dr. Parving --
DR. COOPER: -- who will be discussing the IRMA 2 study. You will be hearing about the ability of irbesartan to alter the course of diabetic nephropathy earlier in the disease so that patients do not advance from the stage of microalbuminuria to proteinuria, the onset of which in diabetes heralds the inevitable decline in renal function. Thank you very much.
DR. PARVING: So, I'm supposed to say good afternoon from Denmark.
I am pleased to give you information on the early course of diabetic kidney disease and I'm going to present the data from the study called IRMA 2. It's dealing with irbesartan in type 2 diabetic patients who are suffering from persistent microalbuminuria.
I will present data in two segments, one giving you some background information and then go to the presentation of IRMA 2.
First, the background. This is actually linking up to what you have just been told, dealing with the IDNT study. It's a Kaplan-Meier estimate of the primary composite endpoint in the IDNT study in relation to quartiles of albumin excretion rate at baseline. The message from this baseline estimate is the following. If you have levels of albuminuria, low 1,000 milligrams, the event rate is approximately less than 20. If you go the very high rate, the upper quartile, then you have an event rate of nearly four-fold higher, at least suggesting that the level of proteinuria or albuminuria is reflecting the underlying cause of the kidney disease as first demonstrated by Bright in 1836 in England.
This cartoon is giving you information on the different levels of albuminuria. We have a log scale and we are dealing with the overnight albumin excretion rate, and the reason why we are dealing with the overnight albumin excretion rate is in an attempt to standardize the collection. So, we are avoiding the marathon runner, we are avoiding other special activities of standing up and lying down because we have that phenomenon. So, we are standardizing it by using overnight collection. So, you collect all the urine during the nighttime.
Normal albuminuria is defined as an albumin excretion rate below 20 micrograms per minute. If you have an excretion rate between 20 and 200 micrograms, we call it microalbuminuria.
This range of albuminuria is usually not depicted by the dip stick test. You need to develop sensitive tests in order to pick it up. This was described the first time 20 years ago by Viberti, Mogensen, and myself as something important in relation to diabetic kidney disease. And we actually have an anniversary this year.
DR. PARVING: Then about the 200 microgram level, we are speaking about overt nephropathy.
You will also appreciate that while the IDNT studies carried out in patients with overt nephropathy having an excretion rate of more than 2,000, way up here, IRMA 2 is a study carried out very early in the course of diabetic kidney disease because, as Ed Lewis already told you, microalbuminuria is an abnormality in the glomerular capillaries leaking protein. So, it's the earliest clinical sign we have of an underlying diabetic kidney disease. So, IRMA 2 is carried out in this range.
Very important information is that 60 percent of our type 2 patients will never, ever develop kidney disease. Unfortunately, 40 percent of our patients in America, in Europe, and in certain parts of Asia, even higher, will develop this devastating kidney disease.
It's also true that in order to develop the disease, you are progressing through the level of microalbuminuria, the earliest state of diabetic kidney disease, into overt nephropathy.
Important to note is that the GFR, meaning the glomerular filtration rate -- the drop in normal man with normal albumin excretion rate is 1 ml per minute per year. If you have microalbuminuria, the drop in kidney function is ranging between 1 to 3 mls per minute per year and that is based on all the available data. You'll appreciate later on in my speech that the rate of decline in the IRMA 2 is 2 mls per minute per year.
If we go to overt nephropathy, the rate of decline is increasing, and it's ranging between 2 to 20, with the average rate reported in the literature of 10 mls per minute per year. Actually the level of decline from the IDNT study -- we haven't discussed that today -- was 6.5 mls per minute per year. So, you can clearly see if you go from this level of proteinuria to this level, you have a progressive worsening of the kidney function. Consequently the aim of IRMA 2 is to keep the patients within that region. We don't want to get them out of the box.
More background information about microalbuminuria in type 2 diabetes. As already stressed by Ed Lewis, it is an early marker of diabetic kidney disease. We have structural lesions too, and we have discussed that already. That's the alternative. If you don't like the clinical physiology, you need to do repeat biopsies. You have no other alternative if you want to evaluate the lesions. We have biochemical evidence suggesting abnormalities, as also seen later on in the disease.
Important to note, again based on all available literature, 5 to 10 percent of these type 2 patients with microalbuminuria will convert into overt nephropathy every year.
As already stated, as long as you have microalbuminuria, the rate of decline in the glomerular filtration rate is ranging between normality and slightly elevated, but a very low rate of decline as compared to what happens if you are running with overt nephropathy.
Then I think in all fairness it should be mentioned that the American Diabetes Association and the International Diabetes Federation actually are advocating that we are screening for microalbuminuria, and if we detect it persistently, we need treatment.
The hypothesis in IRMA 2, the earliest study of irbesartan in diabetic kidney disease, is exactly the same as in IDNT. The objective is to evaluate the renoprotective effect of irbesartan above and beyond the blood pressure lowering effect on the progression to overt nephropathy, and we are comparing that with conventional antihypertensive treatment and it's done in hypertensive patients who have type 2 diabetes and persistent microalbuminuria. So, in essence, this is not a blood pressure trial. This is a trial aiming at evaluating the blockage of angiotensin II. Is angiotensin II nephrotoxic? Yes or no. That's the answer you'll have from this trial.
The study design was carried out in the following way. The patient was run in placebo for at least 4 weeks. We saw them every week at the clinic. They were then randomized either to receive placebo, irbesartan 150 milligrams once daily, or irbesartan, the yellow one, 300 milligrams once daily. We used 4 weeks in the titration period.
The aim of the three arms was to obtain blood pressure equivalence. So, we were actually trying to get exactly the same blood pressure in each and all of the arms, keeping the blood pressure below 135 over 85 millimeters of mercury.
In the placebo arm, it was not allowed to use ACE inhibitor or receptor antagonist. Neither were you allowed to use dihydropyridine calcium antagonists. The reason for that was that in some of the past literature, this kind of compounds, the dihydropyridine calcium antagonist, was reported to elevate proteinuria, and we consequently felt that it was unfair then to use it.
The primary outcome in IRMA 2 is time to the first occurrence of an albumin excretion rate of more than 200 micrograms per minute and an increase of at least 30 percent in albuminuria from the baseline level, and that had to take place at two consecutive evaluations. This endpoint has been used for the last 10 years within trials trying to evaluate the importance of preventing the occurrence of disease in type 2 and type 1 diabetes. So, we are using exactly the same endpoint that other colleagues have applied in the past dealing with this question.
The secondary endpoint is the changes in the overnight urinary albumin excretion rate, and finally, we are looking also at the changes in estimated creatinine clearance. This is based on the so-called Cockcroft-Gault formula, which is an old formula based on measurement of creatinine knowing the sex of the patient, knowing the weight of the patient, and the age, and then you can calculate this formula. Actually the formula has been validated by ourselves in patients with diabetic kidney disease. It works.
The baseline characteristics in the IRMA 2 study. The good news is that the three arms are balanced dealing with demographic data, dealing with clinical data, and dealing with laboratory data.
We have the same age, 58 years of age.
We have a male preponderance, as we should have in this disease.
We cannot discuss ethnicity in our study from the point of view we don't have any other than whites because it was done in Europe. I complained about that. Next time we have to do it around the world.
BMI was, as it should be. They're obese.
The known duration is identical.
Hemoglobin A1C -- we were discussing that earlier this morning -- was also at the same level, and that level was equivalent to a mean blood glucose of 8 millimoles per liter. I hope you know millimoles per liter, because I'm not able to convert it so speedily into milligrams per deciliter. It should be all right.
Microalbuminuria, the level in micrograms per minute is also at the same level in the three arms. Of course, there are small differences, and we will adjust for that when we do our risk ratio measurements.
Another important issue compared to the IDNT, which you just heard about where there was already, when they started the study, the GFR was down to 58. So, it was in harmony already when they started the study dealing with IDNT. In this study they had well-preserved kidney function, 109, 109, and 108.
The blood pressure was equivalent in the three arms. The same systolic, 153 in all three arms; 90 in the placebo; irbesartan 90, and 91. There was no statistically significant differences at baseline.
So, what happened to the blood pressures during the trial? If we started at the bottom line, you'll remember that there was identical blood pressure at baseline. They went down and they stayed down during the study. On average, the blood pressure in the three arms was identical. It was 83, 83, and 83 millimeters of mercury. If we go to the mean blood pressure calculated the usual way, the placebo group and the irbesartan 150 group had a mean blood pressure of 103, 103, and the irbesartan 300 group had 102. And that was significant. It was only small reduction but there was a significant difference in blood pressure.
If we then go to the top, the systolic blood pressure. Again, you'll remember that it had identical values at baseline. The values during the 2 years of observation in the placebo group was 144 millimeters of mercury. In the group receiving irbesartan, the green one, 150, it was 143, so there was a 1 millimeter difference. And finally in the group, the yellow one, receiving irbesartan 300 milligrams once daily, the systolic blood pressure was 141 millimeters of mercury, and that was definitely lower than in the placebo group. Again, we adjusted for that in our hazard estimation.
This is the main outcome of IRMA 2. This is the cumulative event rate, a Kaplan-Meier plot, of the development of diabetic kidney disease, defined as earlier. First of all, I would like to tell you why this has this bumpy appearance. It has to do with the fact that albuminuria at these different time intervals -- it's not measured continuously. It's measured after a certain number of month, 3 months, 6 months, 12 months, and so on, and consequently you can only have events at these time points.
After 3 months, you always see a separation, and the separation is actually persistent during the 2-year study period. At the end of the 2 years, 15 percent in the group receiving placebo on top of standardized treatment -- treatment reduced the blood pressure to nearly the identical level as in the two other arms. There was 15 percent of these patients who progressed to a level of more than 200 micrograms per minute. In the group, the green one, of irbesartan 150, it was 10 percent, and finally in the group of 300 milligrams of irbesartan once daily, it was 5 percent.
At the top you will see the relative risk reduction. The relative risk reduction unadjusted was 70 percent for irbesartan 300 versus placebo, with a p value equal to 0.0004.
If we adjust for the differences in albuminuria at baseline and the blood pressure during the trial, then the relative risk reduction goes down from 70 percent to 68 percent.
If we then look at the dose of irbesartan 150 once daily versus placebo, we had an unadjusted relative risk reduction of 39 percent. It was not statistically significant with a p value of 0.085.
If we adjust for baseline differences and blood pressure difference, the relative risk reduction actually improved. It was 44 percent and the p value was equal to 0.05.
During the study, patients who developed diabetic kidney disease were discarded. So, when you hit the endpoint, you were out. That's important to understand this slide because this is the percentage change in albuminuria. The expectation for each and all of us would have been a rise in albuminuria, but you have to remember that the bad guys are out. So, when they hit more than 200 micrograms, they leave the study.
The message is then from those who received placebo treatment on top of standardized treatment, there was no major difference in albumin excretion rate during the 2 years of study. At the end it was 9 percent above baseline.
If we look at irbesartan 150, the mean reduction in proteinuria was 24 percent, and this was highly statistically different from the placebo group.
Then if we look at my favorite, irbesartan 300 milligrams once daily, you see the reduction, and the reduction is actually continuing during the trial period. So, at 2 years, the difference is 54 percent compared to the placebo group.
Another important issue is that if we compare the mean reduction in albuminuria, it was 38 percent compared to the 150 with 24. There was a highly statistically significant difference with a p value of 0.001.
Estimated creatinine clearance. Again, remember that the patients who developed diabetic kidney disease are leaving the study. So, what you're seeing here is, in essence, what is happening in those who remain microalbuminuric. You are seeing a picture of a so-called biphasic response because we have the estimated creatinine clearance, and you see the initial response from time 0 to 3 months when blood pressure is going down. There is rather a steep drop in kidney function, but before giving you that figure, I will just mention that the initial value of creatinine clearance was identical in the three arms, 108, 108, 109. If we look at the initial drop, it was 5 milliliters in the group receiving irbesartan 300 and it was 3 in the group receiving placebo or irbesartan 150.
So, initially when blood pressure is lowered, we are losing filtration power. That is a well-documented phenomenon when blood pressure is lowered. From 3 months and onwards, we are dealing with a so-called sustained rate of drop in kidney function, and the good news is that the rate is flat.
Actually when we look from here to here, the drop in kidney function in the irbesartan 300 milligram group and in the irbesartan 150 milligram group was only amounting to 2 mls per minute per year. You'll remember that the normal drop, just by getting older, is 1 ml per minute per year.
In the group receiving the placebo, the drop was 1 ml per minute per year, and if you remember that we initially started off having a GFR of 110 and you're only losing approximately 2 mls per minute per year, as long as you're microalbuminuric, then it will take you approximately 40 to 50 years to know the states which we have discussed this morning, if that continues. And of course, you will ask me that, so I will give you an answer later on.
The safety profile in the IRMA 2. We are very early, so we don't have a lot of concern actually. If we are looking at the most important one, the serious adverse events, we had 22.8 percent in the placebo group. In the group of irbesartan, it was less, 15.8, and the group of irbesartan 300, it was 15 percent. So, there were actually less severe events. The number of deaths was equivalent.
So, in summary then, the IRMA 2 study had two messages.
The first is that by using irbesartan in patients with type 2 diabetes and microalbuminuria, we found a 70 percent risk reduction in the progression from microalbuminuria to overt nephropathy.
The second one is that the risk reduction was dose-dependent, meaning that 300 milligrams was superior to 150, which of course is very important when you treat patients, as I do every day.
Furthermore, the effect was an effect above and beyond the blood pressure reducing effect.
And finally, as already stressed by Melisa, it's a safe and well-tolerated compound in these patients.
Thank you very much.
DR. BORER: Thank you very much. Again, a really clear and lucid presentation.
Do we have questions here? We'll start with our two nephrologist members here, if you have any issues to raise. Dr. Brem?
DR. BREM: I was wondering if you could just clarify one point on the slide D-12. The placebo group. If I understand it, the patients who met the endpoint were not included in this slide for GFR or were they included in the GFR?
DR. PARVING: The patients meeting the endpoint, meaning the development of diabetic kidney disease, were excluded when they met the endpoint. They were in the study until they met the endpoint.
DR. BREM: So, on this particular slide, D-12, were those placebo patients demonstrating a decrease in GFR included in this slide or were they not?
DR. PARVING: All patients, as you can see from the numbers here, were included, but when they developed diabetic kidney disease, they were no longer included because, by design, they had to leave the study. Consequently, we could have no additional value on them. So, we are following each and all of them until the time point where they developed diabetic kidney disease and then they are left out.
DR. BREM: Do those patients have an accelerated decrease in GFR in your particular study in the placebo group as you described in the general population?
DR. PARVING: That's a very important question which cannot be answered from this study because this study, as you see, is only running for 2 years. When you go from microalbuminuria, having a very low level of progression, to overt nephropathy, you need a couple of years in order to actually be able to pick up the signal that the rate of decline is worsening.
DR. BREM: But I thought you demonstrated at the beginning -- correct me if I'm wrong -- that patients with overt proteinuria have a decline in GFR of about 6 or more milliliters per minute --
DR. PARVING: That's correct.
DR. BREM: And these patients have a GFR decrease of approximately 2. So, if you counted all the placebo patients who reached overt proteinuria in this particular analysis, wouldn't you expect to see a more pronounced decrease in glomerular filtration rate in that population?
DR. PARVING: It's completely correct, as you state, in relation to the initial slide. The drop in kidney function in these patients is minute, and when they go from one level of proteinuria to a next level of proteinuria, it's correct that the rates start to go up. But you need a certain time interval in order to pick up the signal. Actually most nephrologists suggest that you have to follow the patient approximately for 2 years in order to be pretty sure that you have that signal.
DR. BREM: So, you'd have to follow these patients in this group for an additional --
DR. JULIA LEWIS: The overt proteinuria patients are no longer counted. Once they hit overt proteinuria, you're no longer measuring their creatinine clearance.
DR. BREM: Right, but my thought would be, if you did, you would be able to demonstrate efficacy in GFR much more convincingly. And I'm wondering why you wouldn't have included them in this particular analysis because it would be more supportive of your argument.
DR. PARVING: We have data from the literature. I would like to share them with you. I think we can show you 4-3.
This is the review of the literature available to each and all of you. That's a retrospective study, the first one. All the remaining studies are prospective.
The message from this slide is the following. First of all, the conversion from microalbuminuria into overt nephropathy is ranging from 4 to more than 9 percent.
If we look at the drop in kidney function -- and you have to remember that these studies were followed prospectively for 4 to 5 years, so they had a much longer observation period than we have in IRMA 2. The rate of decline in the study from the Pima was 1, from East Welley, it was 2, from our study group at Steno it was 3.2 percent, from India it was 1 percent, from the ABCD trial it was 1 percent. These patients were treated with blood pressure lowering agents here and here, while they were untreated here because they were normotensive. So, in long-term observational study, the rate of decline, as I depicted to you, is between 1 to 3.
And I can expand on that because we have a follow-up of the Steno study, and that will be 4-37.
The Steno study we published a couple of years ago. It's a study dealing with type 2 diabetic patients who have microalbuminuria, the same way as in IRMA 2. They're very much the same. It's a study where we were looking at the potential importance of multifactorial intervention. In this study, we were actually measuring the glomerular filtration rate using an isotope technique initially and during the study period. We reported the data after 4 years, and now just before leaving Denmark, I had the possibility of looking into the data after 8 years. We have 129 patients followed now for 8 years.
Those patients who remained microalbuminuric in this prospective, randomized trial had a rate of decline every year during the 8 years of 3.2, while those patients who developed overt nephropathy, using exactly the same definition as I gave you earlier, had a drop in kidney function of 4.7, again showing to you that when you pass from one category to the next, then you start to have worsening of kidney function.
But again, the important issue in this very early state of diabetic kidney disease is that you have to do long-term studies, and this is an 8-year study while IRMA 2 is a 2-year study and we could not pick up that signal.
DR. BORER: JoAnn, do you have any specific questions?
DR. LINDENFELD: In the GFR substudy that's mentioned in the briefing booklet -- you've explained the reason for that not changing. But after 4 weeks of withdrawal of drug, certainly in the 150 group, the proteinuria went right back up to baseline. And after 2 years, do you find that disturbing that that doesn't suggest that there's been a persistent change? And then again, it also went up in the 300 milligram group, but not as much. I'm wondering what we would make of that?
DR. PARVING: In the literature, several studies have been carried out and after carrying out the studies, some of us, at least those of us from Denmark, have stopped the treatment and then see what happens. That has been done in type 1 patients, in type 2 patients, early and late.
So, in the IRMA 2 trial, we actually did the same. We followed in a subset of patients kidney function during, of course, the 2 years of observation, and then after 2 years, we stopped all treatment. So, at this time point, all kind of blood pressure lowering treatment was stopped and we are now looking at the change, after stopping treatment for 1 month, in albuminuria. In the placebo group, the pink one, actually the level goes back to baseline and no change. If we look correctly at the 150 milligram irbesartan group, there's actually a huge rise of 80 percent going back to normal, which may suggest that a major part of the effect in that particular arm was hemodynamic.
The good news, however, is that in the yellow group, the irbesartan 300 group where we saw a significant reduction in development of diabetic kidney disease, we only saw a rise of 13 percent, and this in my mind is one of the first times ever where we have demonstrated that by stopping this kind of treatment, we are not regaining what we expect. Actually it seems to suggest that there is a residual effect of the irbesartan 300 milligrams, but again, that has to be proven in a larger number of patients. So, all in all, it may suggest that the effect of our compound in the high dose has residual renoprotection.
DR. BREM: Were those data controlled for blood pressure?
DR. PARVING: Blood pressure rose. As you may remember, the blood pressure was identical in the three arms, more or less. The diastolic was identical.
DR. BREM: But at the end.
DR. PARVING: All of them had a rise in blood pressure when you stopped treatment because you stopped all kinds of medication. So, in essence, I think if I recall correctly, the rise was less in the placebo group, was biggest in the two groups who no longer had the treatment with irbesartan. So, it's not a blood pressure phenomenon.
You'll also remember that despite the fact that blood pressure was reduced to the same level in the placebo group, as in the irbesartan 150 milligram group, there was actually no difference. There was a huge 24 percent difference in albuminuria. So, even though there are minute changes in blood pressure, this can definitely not explain that. But there was a rise in blood pressure.
And there was also a rise in kidney function, the way it should be, meaning that when you stopped that kind of treatment, you see a regain in kidney function. As I said to you initially, there was a drop the first 3 months, this biphasic pattern, and this is due to a blood pressure drop.
Actually give me the option of mentioning something, which some of you may remember and some of you may have forgotten, and that's the story about malignant hypertension. The story about malignant hypertension is dating back to the 1950s. When you had malignant hypertension, the survival until end-stage renal failure was 2 years, if you didn't die from stroke before that. It was a devastating condition.
However, when blood pressure lowering was initiated, we saw the following pattern. When you lowered blood pressure, you saw a rise in creatinine and, of course, that always indicates that you may do some harm to the patient. However, the creatinine level then stabilized, stabilized, and stabilized, and the patient no longer went into end-stage renal failure.
So, this initial phenomenon is actually well described more than 50 years ago in malignant hypertension and is documented in each and all of the major trials dealing with kidney outcome. The MDRD study, the captopril study, all of them have this initial drop.
DR. BORER: I'd like to go back to slide D-12 again. I don't want to try to over-interpret data that have been processed in a certain way with all the dropouts that you explained for reaching overt proteinuria. But I'm struck with the observation that the people who were left in the trial in the placebo arm and the irbesartan 150 arm, even though they more frequently had overt proteinuria, as we can see from the numbers below, and therefore dropped out, even though there were more dropouts, more overt proteinuria, those who are left in seem to have a slightly better response over the 24 months than the group with irbesartan 300. That may be an artifact of all the processing of these data, but I'd just like your comment about that.
DR. PARVING: I think definitely it's fair. From the point of view that I stated initially, that actually the initial drop in kidney function is the important player here because the drop in the irbesartan 300 milligram group, those who gained most in relation to avoiding development of nephropathy, was 5.7 milliliters. So, the absolute drop was 5.7 here and it was 3 here.
You will also appreciate that the level initially was 1 milliliter lower. So, that's another issue.
At the end of the study, after 2 years of observation, the difference between irbesartan 300 milligrams and the two other groups was 3 mls. So, most of the difference is actually explained alone by the fact that the initial drop in kidney function in irbesartan 300 was bigger than in the two other arms. There was no difference in the irbesartan 300 group and in the two arms dealing with the slope. It was identical.
DR. BORER: Should we draw any inferences from the fact that the drop was greater in the 300 milligram group? I mean, is that a bad thing?
DR. PARVING: You could say that, unfortunately -- luckily, it's the opposite. Actually it turned out that in several studies that those who have the biggest initial drop in kidney function have the best long-term prognosis. That has been demonstrated from our group and from the group in Groningen in Holland and also from Italy. It actually suggests that the initial drop in kidney function is reflecting probably the drop in glomerular pressure which is elevated in these patients, at least in animals, as demonstrated by Barry Brenner.
DR. BORER: Blase?
DR. CARABELLO: It's just that those data seem different from the table that we have in our book, table 7, where the initial drop in the irbesartan 300 group in GFR was 2.3 percent and then the late drop at 24 months was 12 percent. This is on page 10 for anyone that doesn't have it. It seems to be the reverse of that.
DR. PARVING: This is the intention to treat. I don't know what you have there.
DR. CARABELLO: This is a GFR substudy.
DR. PARVING: Oh, don't do that.
DR. PARVING: Because now we are mixing everything together, but it's all right. I will clear it up.
This is dealing with all patients enrolled in IRMA 2. The substudy is only dealing with a subfraction, actually approximately 130 patients, who participated in the substudy. The substudy was not a random pick because the substudy was dominated by Dr. Parving and his group because we had 50 patients in the group after the 130. So, the substudy is in no way representative or a random sample for the whole population.
Consequently, this is the important player. This is the whole group and all the data based on the Cockcroft-Gault. It's not the substudy.
DR. CARABELLO: So, was the substudy done to confuse cardiologists?
DR. CARABELLO: Or was there another purpose?
DR. PARVING: I need to be honest now. We were actually asked by the FDA to do it.
DR. PARVING: And I'm pretty sure that the FDA did not want to confuse anybody.
DR. PARVING: What the FDA really wanted us to look at was the effect when stopping the drug. That was actually the aim of the FDA. They would like to see what is the effect when you stop your treatment after 2 years. In the high group, in the group of 300 milligrams of irbesartan, there seemed to be a persistent effect, at least the proteinuria did not go up. And I'm pretty sure that the FDA will be pleased to see that.
DR. PELAYO: I'm the primary reviewer for irbesartan diabetic nephropathy.
I think the issue for the subgroup study -- that was the wrong question to ask because regardless how many patients you study, it doesn't matter for how long you are going to follow them up after you stop the medication, those studies have no -- you can't interpret them because there are multiple scenarios that I could create. So, I would totally disregard those studies not only as the primary reviewer but also as a nephrologist. And I say that to confuse everybody. So, that was the wrong question to ask because there is no answer that can be interpreted.
DR. PARVING: Unfortunately, I happen to disagree slightly because we made a paper a couple of years ago where we actually demonstrated in type 1 patients with microalbuminuria, a randomized trial carried out in Denmark for 8 years. Don't shake your head.
For 8 years we did the study. We published it in the British Medical Journal, and what we did after 8 years, we measured glomerular filtration rate during the 8 years in these type 1 patients with microalbuminuria, and then we stopped the treatment and remeasured glomerular filtration rate. And the outcome of this study, with the first author as Elizabeth Mathiesen, was the following.
8 years of treatment with the drug called captopril in type 1 patients with microalbuminuria stabilized kidney function. There was no drop whatsoever during 8 years when we reevaluated after stopping the treatment. So, I think actually that FDA was very smart asking us to do that.
DR. KOPP: Just one other question about that. In the GFR substudy group, did you drop patients out who had met the proteinuria endpoint?
DR. PARVING: Sorry. Once more.
DR. KOPP: In the substudy group that we're talking about --
DR. PARVING: In the substudy group, if they developed diabetic nephropathy, they were out.
DR. KOPP: You were out in that study as well.
DR. PARVING: Yes, exactly because the aim of the substudy group was actually to evaluate what happens when we stop the treatment. So, they had continue until 2 years, and then we stopped the treatment.
DR. LINDENFELD: But in this substudy, without belaboring it too much, there really wasn't any difference in the dropout rate.
DR. PARVING: No.
DR. LINDENFELD: So, again I think it does point up that in this study GFR didn't change and you can't explain the substudy on the fact that the patients that developed proteinuria dropped out.
DR. PARVING: No, no, no. The important issue dealing with the glomerular filtration rate from IRMA 2 is the following. I will never, ever dare to claim that there is any difference in the drop in kidney function in these patients. The message is the opposite. The message is as long as the patients stay microalbuminuric, you are only losing 2 mls per minute per year. In other words, it lasts many, many years. If we calculate this, it will take 40 to 50 years to go to the department of nephrology asking for dialysis, and that is the message.
DR. BORER: Ray, you had a question?
DR. LIPICKY: Yes. If you could show slide 4-197 again because I think you said a few words, and I probably missed it.
So, in the very last end there, week 4, there are three data points, and we were sort of led to believe at the beginning of today that you developed proteinuria when all these things build up in the glomeruli. So, in 4 weeks, in one of the two possible things, the green and the yellow, all of that mass of stuff must have reversed? All of those bluish globs that we saw early in the day went away -- or came back? I'm sorry. Came back.
DR. PARVING: No. I will be very pleased to answer that.
First of all, there is a rebound phenomenon. That is well demonstrated.
DR. LIPICKY: Yes.
DR. PARVING: You have to remember that the expectation is actually the following. When you have 2 years with microalbuminuria, you'll assume, if we have not treated the patient, that it will be up here. However, there was a rebound to the baseline suggesting that a major part of the lowering in this group was due to hemodynamic effect. I quite agree.
DR. LIPICKY: But you have the placebo group there.
DR. PARVING: Sure.
DR. LIPICKY: So, you don't have to refer to something way up there.
DR. PARVING: But the placebo is not a placebo group left untreated.
DR. LIPICKY: No, no. I understand.
DR. PELAYO: You know what is the problem? To interpret that, you have to understand that the antihypertensive medication was discontinued. If it was discontinued in the placebo, it was discontinued in in 150 and it was discontinued in 300.
DR. LIPICKY: No, I --
DR. PELAYO: Wait. Let me finish.
Then if you discontinue the antihypertensive, the blood pressure will go up. That in and of itself can affect glomerular permeability. If you stop abruptly the inhibition of the angiotensin II system, that also can modify hemodynamics and glomerular permeability. Therefore, to me the data is not surprising.
But it still is the wrong question to ask because it doesn't matter what happened after you discontinue the antihypertensive. What matters is what happened before because all this could be due just to a functional effect.
I think Dr. Brenner, who is sitting on my right, could explain this in a more elegant way and without an accent. Dr. Brenner, do you care to enlighten the audience with your knowledge about proteinuria, glomerular hemodynamics, and antihypertensive treatment, and the wrong question and how you can really interpret the data?
DR. BRENNER: I don't think I can improve on it.
DR. PARVING: Could I have a chance to answer the question?
DR. PARVING: And will you answer all the remaining questions for me? That's all right.
DR. LIPICKY: Well, it's getting mixed up. Let me ask the question again.
We were supposed to take the decrease in proteinuria, effect of ARBs, as inducing a morphological change in glomeruli and that the proteinuria occurred because of some morphological effect. And what you have there, at least in one data point, is what looks like something morphological happened in 4 weeks that negated 2 years of therapy. That sort of is mysterious.
DR. PARVING: No. I think that the question raised is quite on target, and I will definitely be very pleased to answer it.
First of all, it's important to realize that this kind of kidney complication is not something which is done overnight. It takes a number of years to develop that kind of lesion, even the early one, with mesangial expansion as increased basement membrane. There's also good reason to assume that the number of years it takes to get rid of it is probably the same. Why should it be different?
We have one marvelous example from the literature, and that is the beautiful biopsy study carried out by Michael Mauer from Minnesota. He took type 1 patients, biopsied them, and then he gave them a new pancreas. They had this new pancreas working for 10 years with completely normal blood glucose values. No insulin, no nothing. He rebiopsied after 5 years. There was no significant difference. But after 10 years, he saw that there was a significant reversal of the structural damage. I think to my mind that this study demonstrates that it takes a long time to get rid of it.
I think what we are doing here, we are only doing a short-term study. Of course, the message from this study is that the patients are put on the treatment. We will never, ever stop them. We'll continue of course. That's the kind of treatment you need if you need if you want to get reversal of the kidney structural damage. But I think, in essence, that everybody has to realize that it's a very slow process and you need to do it for many, many years in order to gain.
DR. BORER: May I ask? I'll try not to muddy the waters a little bit more. It sounds to me like this is a two-component model. Number one, the morphological changes in the kidney, and number two, superimposed upon that, acute hemodynamic changes that can change the expression, if you will, of the effects of the morphologic changes.
You took away drugs at week 24 and you said the blood pressure went up. Presumably it went up relatively rapidly, and we saw here the effect of what was sitting there with a new blood pressure level on top of it. And it looks worse for at least two of the arms.
If you followed along further, would you have expected -- or let's say you lowered the blood pressure further with some other drug. Would you then have expected the green point to come back down a little bit again?
DR. PARVING: It depends on what kind of compounds that you're aiming at because all blood pressure reduction will eventually reduce the albumin excretion rate, each and all of them. But some of them are more potent and those that are more potent are those that are blocking the effect of angiotensin II. That goes for ACE inhibition and for receptor antagonist.
DR. BORER: I understand that. I didn't want to get into that. I see that the irbesartan 300 doesn't seem to reverse nearly so much as the others, and that suggests that there's some residual effect, et cetera.
I was only asking the question if you took away the new hemodynamic load on top of the morphological change that already existed, would you see some tendency towards improvement. That's all.
DR. PARVING: Sure.
DR. BORER: And you would.
DR. PARVING: You will see that.
DR. BORER: Tom?
DR. FLEMING: I'd like to just, one more time for my own sake, go through the interpretation you gave to your slide D-12 just to make sure I understand what you're telling us your interpretation is.
We have had explained to us today a biological progression that occurs with microalbuminuria first, leading to proteinuria, leading to glomerular filtration rate changes, leading to end-stage renal disease, which in essence is dominated by dialysis, transplant, and renal death. This is the progression.
In the IRMA 2 trial, we're really going back to this earlier stage. We're looking at whether or not we can delay this progression to clinical proteinuria. Often what we would want to do, from a statistical perspective to get a sense of the validity of that as a surrogate for ultimate clinical benefit, is to see how that translates into effects on other tangible phenomena that are downstream in terms of clinical consequences. Of course, the next one in line is the filtration rate.
What you're saying is there's not an apparent benefit here, but you're saying be careful because it's going to take more years of effects before you're going to get to a point where you're going to expect to see effects on GFR. Is that correct? Is that a correct interpretation you're giving to why one shouldn't be too concerned about this?
DR. PARVING: I think you're right.
DR. FLEMING: And if I take that then as the interpretation, I can be -- I'll give my interpretation. Jeff, you can comment. Or go ahead. I'll give my interpretation after you.
DR. BORER: No. I just want to ask is that really what you -- I mean, if I'm understanding what you said correctly, it's not that you were waiting for an improvement in GFR. You were waiting to see a worsening and it didn't happen.
DR. PARVING: Exactly.
DR. BORER: You want to, as you said, keep them within the box.
DR. FLEMING: Sure, but I would like to see some evidence of a net benefit relative to placebo. And I'm understanding that we would need to have many more years of effect on delaying progression to increases in proteinuria to be able to expect then, when we look at this phenomenon downstream, we'll see an effect.
DR. BORER: Maybe yes, maybe no. You dropped out and didn't collect data on the people who developed proteinuria.
DR. FLEMING: Yes, but my initial sense is, looking at the numbers of dropouts -- and it's only a speculation. Unfortunately the trial wasn't designed to truly answer this. It's not clear that that would have reversed this perspective.
DR. PARVING: But could I give an answer? Because exactly what you are saying is that if we should have picked up the signal, then of course we should have kept the patients who developed diabetic kidney disease in the trial. And then we should have told you that those who developed diabetic kidney disease and were followed for 5 or more years did worse than those who didn't. Unfortunately, the design of the study was so that those who were the bad-doers were actually leaving the study.
DR. FLEMING: We understand. We understand that.
But I guess the bottom line conclusion to me is I could be persuaded that the lack of tangible evidence of a benefit on the next phenomenon could, in fact, be that we're looking too early. But at a minimum, I'm left without any substantive basis to say I've got evidence to validate my surrogate. I know natural history. I know the progression in natural history, but I don't have any direct tangible clinical evidence to show that when I've intervened and achieved this effect in delay of proteinuria, that this is the magnitude and duration of effect that will reliably translate into ultimate clinical benefit downstream. It may or it may not. But I'm left with much less evidence of validating a surrogate here than I would typically expect to have.
DR. JULIA LEWIS: If I could have the slide from the hyperbolic curve from the other Dr. Lewis' talk.
If you'll remember -- and I'm sure it will be up there in a second -- the IRMA 2 trial was not intended nor anticipated nor would we ever design a study to look at the change in rate of decline of renal function in patients who are in this area of the curve for two reasons.
One, in order to, for example, double your serum creatinine, you have to lose a gigantic amount of renal function to measure that.
In addition, in this area of the curve, the measurements of GFR -- the scatter in the measurement itself is almost equal to the rate of decline in renal function when you're in this area of the curve in microalbuminuric patients.
So, in early diabetic kidney disease with microalbuminuria, you simply can't reliably measure the changes in renal function in those patients, which is why we worked in this area of the curve in IDNT. So, we didn't anticipate to see changes in GFR in IRMA 2 or to be able to detect changes in IRMA 2 in those early microalbuminuric patients over a 2-year period.
DR. LIPICKY: I guess I wanted to follow up on what Tom said because he said what I was trying to say much better, but I'd like to try just one more time to say what I meant.
And that is that the data from IRMA 2 are consistent with the hypothesis that was forwarded. They do not prove the hypothesis that was forwarded. In fact, they sort of don't help it very much.
Then secondly, the creatinine doubling is consistent with the notion that people get into trouble in the first trial, but the actual number of events that were observed don't go along with that. So, this is the nature of the beast that you're evaluating.
DR. PARVING: Could I then add? I will not disagree but I would like to say that the natural history of diabetic kidney disease, where you have such a slow rate of progression in kidney function, which is actually what causes the death of the patient, if you lose filtration power, you're a dead man. Initially the drop is so slow --
DR. LIPICKY: The only thing I said is that if you're a believer, you believe and you don't need data. If you're not a believer, you need data, and that's going to be the problem that's discussed.
DR. PARVING: I think you have the data. I think you have the data telling you that you only have a drop of 2 mls per minute per year in those who are microalbuminuric. That's exactly what you want to know.
DR. BORER: Bev.
DR. LORELL: I think that in this study that you're describing, it seems to me it would be extremely problematic to follow the patients even longer-term to see a more rapid decline in creatinine clearance. And the reason I say that, as a non-nephrologist, is if it is true in the diabetic, that the development of macroalbuminuria precedes this more rapid decline in creatinine from an ethical standpoint for the individual physician investigators, it would have been impossible I think, once a patient reached the point of microalbuminuria, not to use the best data available, which would have been the captopril study, and to have said at this point ethically -- admittedly they're not the same population, but best data available -- I must move to treating with an ACE inhibitor. So, I don't think even today it would be possible to do this study to carry it out long term to look at events.
DR. LIPICKY: This was done today. What do you mean?
DR. LORELL: No. I'm saying we have another study here now looking at irbesartan and data from losartan.
DR. LIPICKY: So, you think it's ethical to do things you don't know anything about rather than find out whether something works or not?
DR. LORELL: No. I would argue that from the point of view of the physician investigators caring for individual patients --
DR. LIPICKY: But to do something where you do not know what you're doing is correct.
DR. LORELL: I didn't say that.
DR. LIPICKY: Well, why didn't you say that?
DR. LORELL: I'm sympathetic with the stopping endpoint in this trial, given the other data that was available, albeit in type 1 diabetes. So, I actually think it would be very problematic to keep a patient, once they had developed macroalbuminuria, on one of these three treatment arms.
DR. PARVING: That was actually the main reason why that was decided because it was tested out at several of the safety committees in Europe, and they would not allow us to go ahead. So, that was quite simple. So, what you're saying is at least the notion in Europe. They may be wrong, as Ray Lipicky is saying, but that was the notion.
DR. BORER: Steve?
DR. NISSEN: Bev, I don't agree with you. If it's ethical to take patients at a later stage in the disease, as was done in IDNT, and give them placebo for a long period of time, it certainly would be ethical to do so in an earlier part of the disease curve.
This is an important question because no matter what we decide here, clinicians have to know when in the course of the disease might an intervention such as this be useful. So, we're looking for a signal here that says, well, gee, maybe if you start this therapy very early when you get that first microalbuminuria, you can prevent this whole cascade. So, we're all kind of looking to find some evidence that that's the case, and there isn't any evidence, unfortunately, in the data.
DR. EDMUND LEWIS: Hopefully, I'll get back up there. I should be wearing armor the next time I get back up there, but hopefully I will get back up there to try to convince you otherwise.
There are complex issues here that involve the ethics. First of all, in early type 2 diabetic nephropathy, microalbuminuria, the study by Ravid from Israel using enalapril showed clearly that ACE inhibitors absolutely stabilized microalbuminuria over a period of -- I forget. I think it was 5 years, and that for that reason, ACE inhibitors were appropriate therapy. So, the physician, seeing the patient with early type 2 diabetic nephropathy using evidence base ethically should be using an ACE inhibitor at least.
Now, in our study, the irbesartan diabetic nephropathy trial, it is not fair to say that since they had two arms that didn't inhibit renin-angiotensin, what about the ethics of that compared to this? Because we were facing an entirely different clinical problem. The clinical problem that we were facing was not only do type 2 patients with diabetic nephropathy perform the same as the type 1's, when their renin-angiotensin system is blocked, but also we're dealing with a much older population of hypertensive patients.
So, the question was also not just benefit, but risk. That is, in this patient population, when you use a renin-angiotensin system antagonist, is there enough bilateral renal artery stenosis to cause serious adverse events with acute renal failure, and do they have enough hyporeninemic, hypoaldosteronism to cause much more severe hyperkalemia during the course of the study than the type 1 set?
So, we had no problems with the ethics of that because before we started the study, we had no idea. The angiotensin receptor blocker, even if it was -- and I think it proved to be -- as effective in treating the glomerular disease of type 2 diabetes, as captopril was in type 1, if that effect was going to be offset by these adverse effects of severe hyperkalemia and acute renal failure, then you had an entirely different clinical decision to make. For that reason, there was not an ethical problem in doing our study, but that does not mean that patients with microalbuminuria shouldn't have been treated with ACE inhibitor in the clinic at least because there was data to say that that was the case.
DR. BORER: Again, thank you for a very illuminating presentation. Maybe we can go on to the risk/benefit summary and get to our questions.
DR. PARVING: It is my great pleasure to introduce Dr. Edmund Lewis who will give the summary with the risk/benefit.
DR. EDMUND LEWIS: Thank you. It's a pleasure to be back here.
DR. EDMUND LEWIS: We propose that therapy with the receptor blocker irbesartan alters the continuum of diabetic nephropathy throughout its course, that the IRMA 2 study showed that, in fact, going from microalbuminuria to overt proteinuria was significantly affected and that phenomenon diminished by the drug, and that in the IDNT, the events associated with decreasing glomerular filtration rate and end-stage renal disease were significantly diminished.
Now, I want to emphasize that the endpoints that we chose were not arbitrary endpoints. The disease is a continuum. I think there's nothing in the literature or anywhere else that would suggest otherwise. However, in designing a clinical trial, you have to identify points in the course where you can try to tell a difference. Therefore, doubling of serum creatinine or a creatinine of more than 6 is not arbitrary. We chose those, I hope that you can see, and we'll present more evidence I guess, to show why that was the case.
Now, I don't think that one can underestimate the importance of trying to prevent end-stage renal disease, and I hope that ultimately in the presentation of the data what you saw as cardiologists -- today I'm becoming much more sensitive toward the cardiologists -- is that what we were doing, in terms of the renal system, was not putting the cardiovascular system at greater risk, that the same drugs that one uses to treat cardiovascular disease in these patients would be used, that blood pressure would be managed, and that overall the mortality rate, the endpoints that you saw were not significantly altered. So, we're not saying treat these patients' kidneys at the risk of allowing harm to their heart. That is not the message, and I don't think anything about the data indicate that.
Can you put up 1.2? There you go.
Now, the reason why we nephrologists are so concerned about that renal endpoint -- we treat the whole patient, which includes the patient's heart actually. The reason we are so concerned about this -- and this is published annually in the U.S. Renal Data Survey. This happens to be a publication which shows exactly the same as the USRDS. In fact, it's showing you USRDS data I guess.
Here this orange curve is the survival rate of patients with type 2 diabetes on renal replacement therapy, on dialysis. As I've said, here at 12 months the survival rate is a little under 60 percent. At 24 months, the survival rate is 40 percent, which is very close to pancreatic cancer over that period of time, and it's almost identical to your consensus class IV heart failure. End-stage renal disease is a dreadful thing to happen to a type 2 diabetic, and that's why we are trying so hard to prevent it.
Up to now, there is no proven therapy. There is no data. I've told many, many doctors to treat with ACE inhibitors until the data become available, but going back 8 years, there was no question about captopril being approved for type 2 diabetic nephropathy on the basis of type 1 diabetic nephropathy data. This was the best you could offer. But there is no study that ever said in the overt nephropathy patient that ACE inhibitors were as effective, to this day.
So, what we're talking about here is not only the diagnosis that causes the most patients to go on dialysis, 45 percent, but actually the proportion of patients in end-stage renal disease programs in this country with the diagnosis of diabetic nephropathy is increasing every year more than any other diagnosis. So, it's an increasing problem, and certainly in countries where they can't afford $65,000 a year to try to keep these patients alive on dialysis, it is death. It is death.
The IDNT, in terms of benefit, allows us to look at two active comparators. It's not two trials, but it's more than one trial, and we keep seeing irbesartan versus placebo, irbesartan versus amlodipine. And at this stage of the disease where we can measure the functional capacity, the filtration capacity of the kidney, you can see that of course the primary composite endpoint is very positive, but the all-cause mortality part, the cardiac part of the primary composite, does not reveal efficacy for irbesartan. And I don't think we would have expected that in this study. I think we've discussed that at length. It's the renal endpoints of the composite that are the important issue: the doubling of serum creatinine and end-stage renal disease.
I think it is clear from our data, both our renal data and our cardiac data, that the calcium channel blocker did not have an adverse effect in this patient population. Amlodipine appears to be a perfectly good drug to control blood pressure in this patient population, and according to our data, you would not get excess either cardiac events or renal events above that using other antihypertensive agents.
The doubling of serum creatinine, then, we are putting forward as the event that is early enough in the course of the disease where we could say that this is a very important endpoint. But actually it has still allowed doctors to treat patients afterwards, to try to prevent end-stage renal disease, so that there was an ethical issue there.
However, when one looks at the doubling of serum creatinine, you see that the median time to end-stage renal disease was 9. something months, and these patients go on to your hard endpoint of end-stage renal disease. Now, some of them don't because they doubled so late in the study that they don't have time to go on to end-stage renal disease. But it is inexorable. They are going to go on to end-stage renal disease.
You can talk about doubling of serum creatinine as a surrogate for that. It is not a surrogate for renal function. It is a measure of renal function. And it's the renal function that is continuously going down.
So then, what we show is, relative to placebo, a 33 percent risk reduction -- the pink is placebo -- with a p value of .002, and a 37 percent risk reduction versus amlodipine, .003, and amlodipine versus placebo, there's no difference.
Now, I know that members of the panel -- in general people look at these Kaplan-Meier curves in a vertical way. But this is a time-to-event analysis, and from the point of view of the physician and the patient, it's actually the horizontal way that counts because what we can tell our patients, on the basis of this study, is that if they take the irbesartan -- here's the average follow-up of 31 months and here is your point of your event rate for doubling of serum creatinine, for those who have doubled, with either placebo or amlodipine, and here's the shift to the right -- you will not have your doubling event for 11 months.
This isn't just a point chosen nonrandomly to show you this phenomenon. If you look at other points along the end of this curve, you see the same, more or less, 11-month delay.
And 11 months doesn't sound like much perhaps, but believe me, when you're on dialysis, it's a lot. And if your mortality rate is going to be 25 percent during that 11 months on dialysis, then being off dialysis sounds like a very good thing. So, I think that it is not only the relative risk reductions, it is also this delay in the important event that is very, very important.
And this is a very conservative issue because, remember, we're starting with patients who have already lost half their renal function. I don't think that it is overstating the case to say if they had only lost a quarter of their renal function when they started, that delay might be longer. That was certainly the case with the ACE inhibitors, that our most conservative result was in the captopril trial. Once we really started to use these drugs and test more potent ones and higher doses and better blood pressure control and so forth, we got much, much more dramatic results.
When you adjust for blood pressure, of course, irbesartan versus amlodipine, there is no difference in terms of your renal endpoints because blood pressure control was identical in those two groups. So, you cannot imagine antihypertensive effect of irbesartan being the reason why we got these risk reductions. We had to adjust against placebo because -- not surprisingly, the hypertension of type 2 diabetic nephropathy is very hard to treat. Anyone who has tried knows that, and when you have to treat their blood pressure without ACE inhibitors, without ARBs, without calcium channel blockers, you have something on your hands. So, the fact that we had a few millimeters difference is not surprising, but it's really heroic that that was all that we had as a difference. Nevertheless, the difference between adjusted and unadjusted for blood pressure is not significantly different.
So, in terms of the face of the enemy, please remember that patients entering the IDNT had advanced renal disease when they started, and so our result is in patients who are very far along.
In terms of the IRMA 2 study, you have seen this, and our evidence is in these patients they do not enter the definition of overt proteinuria when they're on the irbesartan. You've had the entire discussion of microalbuminuria, but what I want to remind you of is that in type 2 diabetic nephropathy, you do not get to the stage that we saw them in the IDNT without going through a long stage of increasing proteinuria. So, that is a clinical phenomenon. There are no clinical signs and symptoms, but it's a clinical phenomenon that is significant. And when you adjust for blood pressure differences for the two treatment groups, there is no difference in the relative risk reduction of actually going onward to a positive dip stick, which is an important clinical event. And please note with this 300 milligram group, the relative risk is reduced by 70 percent. 7-0 percent.
So, in terms of benefit, our conclusion is that irbesartan retards the progression of both early and overt nephropathy in type 2 diabetes mellitus with nephropathy by a mechanism which is independent of blood pressure control.
Treating 15 patients with advanced diabetic nephropathy entered into the IDNT trial for 3 years, you save one clinical event. That's in terms of renal outcome. Treating 10 patients in the IRMA 2 trial for 2 years, you save one event which is microalbuminuria going on to the positive dip stick overt nephropathy.
I think that it is important to state -- and I'm not presenting this as part of the BMS irbesartan application -- that there are not two clinical trials in the literature to which the medical profession has been exposed to. There are three trials in the medical literature with that exposure, and in the very least what you have to say is that on the basis of that, there will never be an ethical trial looking at ARBs versus placebo in the future or ARBs versus non-renin-angiotensin inhibitor in the future because we've got all of this information.
The last trial published by Dr. Brenner one page after mine in the New England Journal --
DR. EDMUND LEWIS: The RENAAL trial was 1,500 patients. The design varied very, very little from our trial. The outcome events were identical. The patient population was basically identical. The baseline characteristics of the populations were identical.
So, I just want to point out to you, just in terms of the totality of information available as far as the prevention of progression of this horrible disease is concerned, that if you look at the risk reductions of that trial using losartan versus our trial using irbesartan, you see they used the same primary composite endpoint. They got essentially the same reduction with the same variance. If you look at the renal endpoints, basically the same reduction. If you look at doubling of serum creatinine, hardly any difference. If you look at the occurrence of end-stage renal disease, about the same. If you look at all-cause mortality, the same. So, I believe that that trial supports what we are saying with our application, although it's not part of our application.
Collectively these results demonstrate that the renoprotective effects and the benefits of irbesartan across the continuum of diabetic renal disease, we hope you agree, has been demonstrated.
In terms of risks, you've heard Dr. Cooper address the side effects and the risks. I won't go on about that. Certainly the overall risks of this drug are very well known to you. I don't think it was a problem for your primary reviewer, and the specific risk, in terms of hyperkalemia and so forth, is no different than that which the medical community has a concern about and has to treat with ACE inhibitors. So, the risk/benefit assessment we believe favors the use of irbesartan across the continuum of renal disease.
Collectively the data are what they are. We hope you agree, we hope you concur with our statement that this drug should be approved for the treatment of type 2 diabetic nephropathy throughout its continuum. Thank you.
And with that, I'll stay here for questions.
DR. BORER: Actually, I don't think we'll have many questions at this point because we have questions that we have to go through. But I want to thank you very much for, again, a wonderful presentation. To orient everybody, I want to thank the sponsor for the presentation in its totality. I certainly and I think everyone on the committee found it very informative and enlightening.
However, at this point, we're going to move on to the questions put to us by the FDA, so there's not going to be any more discussion and no more comments from the sponsor unless we specifically ask for them. So, I'd appreciate it if you keep that in mind.
There are a number of questions here. Some of them we may be able to go through quickly, some not. To try to be most efficient about it, what we'll do, after quickly going through the preamble here, is present the questions to our primary committee reviewer and then see if anybody disagrees with the answers that she gives.
DR. KOPP: One question.
DR. BORER: Yes.
DR. KOPP: Dr. Pelayo described the second study IRMA as a non-IND study. What does that mean and does that have any bearing on how we view that data?
DR. LIPICKY: No. It has no meaning.
DR. BORER: Yes. It is important because this needs to be part of the public record, that as we take a vote, even if we're all agreeing with everything JoAnn says, we have to do that by name verbally into the microphone, and I'll ask everybody to do that as we go along.
With that having been said, we're asked to give an opinion about the benefits and risks of irbesartan for the treatment of nephropathy in patients with type 2 diabetes. I assume, Ray, you may have meant patients with hypertension and type 2 diabetes, or did you not mean that?
DR. LIPICKY: We did not mean that. I may as well start it out. I fail to see the distinction between having hypertension or not having hypertension if they have diabetic nephropathy, but of course there is an empirical difference.
DR. BORER: Okay.
Reviews of chemistry, pharmacology, toxicology, biopharmaceutics, biometrics, and clinical safety present no apparent barriers to its approval.
And we're asked to determine if the strength of evidence for a treatment benefit, relative to the risk, supports approval.
The direct evidence comes from the studies listed.
Question number 1. There were 411 total endpoint events in the placebo and irbesartan groups, 33 fewer in the irbesartan group than on placebo. One of the characteristics of a none-too-small p value is that the result is sensitive to the handling of subjects with incomplete data.
So, 1.1. 16 subjects, 8 on placebo or irbesartan, never received any treatment. How were they handled? How should they have been handled?
DR. LINDENFELD: Well, these as I understand it, were handled as intention-to-treat, and I believe that's proper.
DR. BORER: Is there anybody who disagrees with that?
DR. BORER: Nobody disagreed.
We'll vote on the question at the end, I guess, with a verbal statement.
408 subjects, 275 on placebo or irbesartan, discontinued study drug. These were the people predominantly who reached an endpoint and came off coded drug. How were they handled, and how should they have been handled?
DR. LINDENFELD: These were also handled as intention-to-treat, which I think is proper. I believe the cardiovascular endpoints were followed only until they reached the point of end-stage renal disease.
DR. BORER: Any disagreement? Yes, Steve.
DR. NISSEN: Again, we talked about this earlier, but I'm quite disappointed that the other events were not collected after they reached those endpoints. So, I think that that to me is actually an important issue, and my feeling is that they were not handled as well as they should have been handled.
DR. FLEMING: And just a brief added comment. I agree with both my colleagues. I agree with JoAnn that ITT is the proper way to handle the discontinuations, and I agree with Steve that technically speaking, ITT doesn't mean including all randomized people. It also means including the follow-up of all randomized people. So, it does compromise the ability to at least more clearly understand the impact on those endpoints that were censored in follow-up after end-stage renal disease diagnosis.
DR. BORER: We'll put a bookmark there.
DR. TEMPLE: Some of them were followed, though, because they have mortality data on all of those people, and there's that slide that shows which people who had an endpoint of doubling went on to end-stage renal disease. So, I was a little foggy on what they did and what they didn't follow. I guess strokes and things like that were not followed.
DR. BORER: Maybe we can have a clarification, very quick. Dr. Cooper perhaps can tell us. You followed everyone except the ones who were lost to a mortality endpoint. We know about ESRD because everybody was followed to that event. We just don't know who had a stroke, who had a heart attack after ESRD. Is that correct?
DR. COOPER: That's exactly correct. The company made every effort to follow every patient with respect to ESRD and mortality. We have all of the data with respect to mortality for all patients except for 8, and we have all of the data with respect to ESRD in all patients with the exception of 37. So, we have to go back, as indicated earlier in the discussion, to ascertain the dialysis and transplantation status of those patients. But that's correct. The only data we did not systematically collect after ESRD were cardiovascular events that were nonfatal.
DR. LIPICKY: Can I ask a question? Because it's my impression it was after doubling of creatinine.
DR. BORER: No.
DR. COOPER: No.
DR. LIPICKY: It was ESRD. So, everyone was followed to ESRD?
DR. COOPER: Yes.
DR. LIPICKY: Even if they met the doubling of creatinine.
DR. COOPER: Yes.
DR. LIPICKY: Okay.
DR. BORER: 19 subjects, 13 on placebo or irbesartan, were lost to follow-up. Mortal status is known for 11 of 19, 7 of 13 on placebo or irbesartan. How were they handled and how should they have been handled? JoAnn?
DR. LINDENFELD: Well, these were included when the outcome was known, and as I understand the analysis, there was a specific sensitivity analysis done to be sure that if one attributed all bad outcomes to the irbesartan group, that this still remained, that the difference was very, very small. So, I'm not too worried about this small number of patients.
DR. BORER: Any disagreement here?
DR. BORER: No, okay.
2 placebo group subjects were credited with endpoint events for near doubling of serum creatinine. How were they handled? How should they have been handled? How many other near-doubling events were not counted as events?
DR. LINDENFELD: This is an area we didn't cover, and we can see if people think we should. There were 2 placebo patients that actually were credited with a doubling of creatinine who, when they went back and looked at the initial, by strict criteria the first study creatinine did not actually double. The adjudication committee, as I understand from the briefing booklet, decided to include them in the doubling. I don't believe we know how this was handled otherwise.
I guess one other question would be of the endpoint events, how many were changed in the endpoints committee? I don't think we've seen that data, and perhaps you could just give us a brief answer to that.
DR. EDMUND LEWIS: Yes. With respect to those 2 patients, our protocol design was that the central laboratory had to confirm a doubling of serum creatinine event, which then went to our outcome committee for adjudication.
And in the 2 patients that you're referring to, what had happened was the geographic lab for that part of the world had not declared a doubling. However, duplicate samples were sent to our central lab and we confirmed a doubling. Now, we're talking about tenths of a milligram. But we confirmed the doubling. We sent that information, along with the information from the local labs, on to the outcomes committee and the adjudication was that those 2 patients indeed had doubled according to our predefined protocol determination.
DR. LINDENFELD: Maybe you can give us a quick answer to how many times did this happen in the other groups, the irbesartan group and the amlodipine group.
DR. EDMUND LEWIS: It didn't. Those were the only two cases.
DR. LINDENFELD: They were the only two cases in the entire study.
DR. EDMUND LEWIS: Yes.
DR. FLEMING: Just one refinement of JoAnn's answer to question number 1.3. If I'm recollecting correctly, Jeff, you had asked this morning a series of questions that related to these issues, and if I'm recollecting correctly, in the 1.3, the 19 subjects lost to follow-up, if one did take a worst case analysis, I think the significance technically, if you believe .05 is a magic number, was crossed. It's hard to know what to make of that because a worst case analysis is incredibly conservative.
DR. BORER: Yes. It was .055 something, as I recall.
In summary, what effect have the sponsor's rules for handling these situations on the credibility of the principal finding? JoAnn?
DR. LINDENFELD: I think they've been handled well, and I don't think it should influence the credibility of the studies.
DR. BORER: Steve?
DR. NISSEN: Well, I sill am very troubled by the lack of cardiovascular event data after those patients reached the end-stage renal disease time point. And I'm particularly troubled because prior to that point in time, we saw point estimates for MI, cardiovascular death, and stroke that were going rather strongly in the wrong direction. And if those trends were to continue, they were pretty close, as individual endpoints, to statistical significance. So, those additional events that may have occurred later that were censored could well have led to a statistically significant result with respect to having a worse outcome than the amlodipine treated arm. So, I really do think it undermines my comfort level significantly.
DR. BORER: Ray?
DR. LIPICKY: I think I feel compelled to say, because the question was oriented so that you would, but you didn't, that since there was only a delta of 33 in the two groups, that a difference makes that indeed, depending on what you do with things and one of the conditions did make it happen where you lost the conventional significance, and that that was simply meant to heighten your awareness to where you were.
DR. BORER: Our awareness has been heightened.
Number 2. Of the 411 primary endpoint events on placebo or irbesartan, 58 percent were creatinine elevation and 42 percent were death or need for dialysis. All of the apparent treatment benefit was the effect on creatinine. And now we need to determine what we think about this.
2.1, was this a statistical anomaly, and 2.2, was this because there were just so few clinical outcome events? Was this because effects on clinical outcome would not be expected over 57 months of follow-up? Was this because an effect on serum creatinine is a poor predictor of clinical outcome?
Subjects who experienced doubling of serum creatinine could later have end-stage renal disease and die. When these events are counted, the relative risk of death on irbesartan was .92 and the risk of needing dialysis was .80. Are these data supportive of an effect on clinical outcome?
Why don't you try and take the whole question as one, JoAnn?
DR. LINDENFELD: I don't think this is a statistical anomaly.
It's important to say I don't think the study was a 57-month study. The mean duration of study here was closer to 2. something years. So, it wasn't a 57-month study. If it had been, I'd be far more concerned about the lack of cardiovascular events here.
I can't explain why there was not an increase in cardiovascular mortality. I think when we relate this to the captopril trial, there are several things that come up. One is that was a different population. Those patients had much less well-controlled diabetes. These patients are likely to have been on far better therapy at this point in time. So, I don't believe it's just a statistical anomaly. I think the follow-up may just have been a little bit too short to see substantial differences in cardiovascular outcome.
I think it helps that the relative risk of death is less, but it would be nice if it were significant. So, not strongly supportive.
DR. BORER: Can I just ask for an opinion? I think one of the key elements here that one can infer from this question is that we're being asked whether we believe that there's a clear relationship between doubling of serum creatinine and a progression to ESRD within 9.8 months. And we were shown data about this. Can you comment on that, JoAnn?
DR. LINDENFELD: I believe that there is a clear correlation between doubling of serum creatinine and end-stage renal disease, based on this data, yes.
DR. LIPICKY: Based on this data?
DR. LINDENFELD: I think so.
DR. LIPICKY: What do you see? What makes you say that?
DR. LINDENFELD: Well, we see a substantial difference in end-stage renal disease, if you believe a creatinine greater than 6 as part of end-stage renal disease. We talked about that earlier.
DR. BORER: So, the incidence of end-stage renal disease, 22 to 23, doesn't mean anything.
DR. LINDENFELD: Well, that's if you only use dialysis or transplant. If you use the creatinine of 6 -- and I think what I've heard makes me think that that -- and the other discussion that we heard just about the creatinine of 6 makes me feel that that was probably a reasonable addition.
DR. LIPICKY: But that has to be, right, because if you use creatinine as the one thing and you also use creatinine for the other, it's got to be the same? Isn't it? I mean, does that really convince you?
DR. LINDENFELD: No, it does not convince me, but I think it's supportive data. Does it absolutely convince me? There are two questions. Does the data absolutely convince me? No, the data doesn't. Do I believe the doubling of creatinine is an important precursor for end-stage renal disease which is important in clinical outcomes? All of this data persuades me that that is true, in addition to other data, yes.
DR. BORER: Any disagreement? Dr. Kopp?
DR. KOPP: No. I would say not so much disagreement. I think I agree with what you said.
But I've puzzled during the day about why the rate of dialysis and transplant was so much higher in the captopril study, given that both were about 3 years and the captopril study involved younger patients. But I realized you mentioned one factor that would favor less renal disease in this group, which is glucose control was worse in the captopril study. Another is that blood pressure I think was not as well controlled. And a third I realized is that some people who doubled creatinine and therefore came off study could have then received an ACE inhibitor and postponed the onset of their ESRD. So, I think that might tend to unlink some of this, particularly over a 2.6-year study.
DR. LORELL: Yes, I agree very much with that comment. I would support that.
DR. BORER: Let's move on to question number 3 then.
DR. FLEMING: Another comment, Jeff, on 2 if we could.
DR. BORER: Oh, I'm sorry. I didn't see you.
DR. FLEMING: It might be worth getting into just a little more depth in 2, if I could.
When I think of effect -- and it may be a simplification, but in addition to the marker here which is looking at changes of a certain magnitude in creatinine -- there are at least maybe three fundamental domains of what's clinically important. One that I might say is a direct obvious renal, which is dialysis/transplantation. Then there's the domain of mortality, which includes renal to an extent, of course renal-related deaths. And then there's the third domain which would be the cardiovascular events. That would include the stroke and the MI, cardiovascular death, heart failure.
My sense is what's happening here is when you analyze these data in different ways, you're getting different weightings of these three domains. With the first of those three domains, there's a signal for benefit. The second and third domains, there's essentially an indication of lack of difference or, to put it another way, to obtain evidence that is convincing of small differences that would take a much larger trial.
So, if I could just briefly refer to a series of five analyses that become more inclusive as you go through them, when you look at the primary analysis, you see the 411 events that we were asked to look at here. If you look at a breakdown of that, when you look at the first occurrence, what you find is there are 64 deaths in control and irbesartan. So, there's no difference in first occurrence of deaths. There's actually no difference in first occurrence of dialysis. There's no difference in first occurrence of transplant. The entire difference is in first occurrence of the doubling. But it's misleading to look at it that way in the sense that, for example, for dialysis you're truncating the follow-up there at the first occurrence of the primary endpoint. So, we want to follow on beyond that.
And that leads to the second analysis which is looking at end-stage renal disease. The first analysis we've seen, there's an excess of 33. There are 33 events prevented. And this is a relative risk of .8 and this has the p value of .023. When you look at the end-stage renal disease, you're getting almost the same relative risk reduction of .77 as the relative risk. You have 19 fewer events, and yet not quite significant. So, if one takes the approach that end-stage renal disease is so proximal to dialysis that it's a reliable surrogate, that there isn't an issue about surrogacy, then for this particular endpoint, we're seeing an estimate of a 23 percent reduction with not quite statistical significance and 19 fewer events.
It's interesting if you look at dialysis. What we were shown is that translates into 15 fewer events. So, in fact, it's sort of a confirmation, I might say, that end-stage renal disease is close enough to dialysis to basically refer to it as a reliable measure. But dialysis, like end-stage renal disease, showing 15 to 19 fewer events, is around that area that we would consider convincing. It's about a p value of .07 to .1, something in that neighborhood.
When you add death, the deaths are essentially comparable. In fact, I think most of the deaths that occur -- there were 87 versus 93 deaths. So, there were 180 deaths. Only 36 of those were people who had had a prior dialysis. So, a large fraction of these deaths are occurring to people who had not had a prior dialysis.
So, basically it would be called competing risks, which points out that at least for the duration of follow-up that we had in this population, there is a significant myriad of health challenges these patients are facing and the renal complications are obviously one important part, but there are major complications outside of the renal. And it would appear from these data -- and it may be what people would say we would expect -- is that there's no reduction in those particular deaths.
So, when you go to the next level of analysis, which is dialysis/death, what you're seeing then is still a numerical 13 fewer events, but now relative risk is .89. So, you're only reducing the relative risk by 11 percent. Obviously, very nonsignificant.
And it's interesting that if we compare that, that's the endpoint in captopril that showed a 50 percent reduction. This particular endpoint, we're showing an 11 to 13 percent reduction.
And personally I find it very acceptable to focus on the renal-related phenomena here, death, transplantation, dialysis. But if you then go one step further and you add in what was at least documented for the cardiovascular events in the secondary endpoints, now you're looking at 209 versus 229 or 20 fewer events, corresponding to a proportion of patients who have events, 36 versus 40, so a 4 percent absolute reduction or about a 9 percent relative reduction.
So, my sense is when you look at this to answer this question, one really needs to break apart these domains. And what, at least my interpretation, these data are telling us is if you focus on end-stage renal disease or, correspondingly, dialysis as the only measures you're looking at, you're seeing something on the order of a 20-23 percent reduction, but it's p values of .07. So, it's very close to whether you would say that's convincing evidence.
When you then add in death -- so, you're looking at dialysis-free survival -- because you're adding in almost as many additional events that were not impacted at all in terms of their reduction -- that 23 percent reduction is cut in half to an 11 percent relative reduction, very nonsignificant, although I don't worry about it being nonsignificant. I'm looking more at the magnitude. And then it's reduced to 9 percent relative reduction when you bring in the other cardiovascular events.
So, it seems as though there is -- this is an issue that we have to decide, is there adequately convincing evidence that you're affecting the clinical renal events because there's clearly a signal toward that. But the other events, even if you just go to death, aren't being influenced nor are the cardiovascular events being influenced.
DR. BORER: Well, we'll have to keep all that in mind as we move along here.
May I ask you, Tom, just one thing? I think you've really covered the waterfront. The point is made in the questions that we don't see the curves begin to separate until 18 months. And that's true. Of course, we've heard about the natural history of these diseases, and it's not terribly surprising that we might not see an impact for a while. But I'm impressed with the fact that at least until you get out to 42 months, by which time the numbers become so small that the statistical stability of point estimates would have to be of concern, the curves seem to continue to diverge. It appears that we're having an increasing effect over time. Do you accept that or can you comment on that?
DR. FLEMING: I think what you're referring to is this is what the primary analysis does show when you look at --
DR. BORER: Also the end-stage renal disease analysis.
DR. FLEMING: If I go through my hierarchy of five analyses, those are a tier 1 and tier 2. They're fairly close.
It's certainly an interesting issue. It's relevant. It's going to mean that statistics such as the log rank test will be pretty sensitive to those kinds of emerging effects. It means that it's possible, plausible that if one had continued this for a number of additional years, then the magnitude of the signal may have been more apparent. It comes back to a comment just before the break I think that Dr. Temple had asked about what ability would there be to follow up these patients in this study to see whether there is more data than what we've had presented to us for effects of the signal on dialysis. It's entirely possible that it would show more signal.
DR. BORER: Let's go on to question number 3. Irbesartan reduced the composite event rate compared with amlodipine by 23 percent. Considering the low nominal p value, is this as good as a second study? This p value is smaller than for the comparison between irbesartan and placebo because amlodipine did worse than placebo. How does that confirm a benefit of irbesartan?
DR. LINDENFELD: I don't believe that this is as good as a second study. First, when you look at those curves, they were different. Amlodipine was just slightly worse, not statistically significantly so from placebo, so it's not a surprise that this p value is lower than when compared to irbesartan. So, no, I don't think it's as good as a second study.
Does it help a little bit? It helps me a little bit in that in the amlodipine group, the blood pressure was well controlled, and I think that's a helpful finding, but certainly not as good as a second study.
DR. BORER: I want to ask Tom for a clarification here again. I'm not sure what we can infer from the nominal p values here. The way I would look at it, it's unlikely that the difference between irbesartan and placebo was due to chance alone for the primary analysis, and unlikely that a difference between irbesartan and amlodipine was due to chance alone in the primary analysis. I'm not sure what you can infer about placebo and amlodipine and about the difference in the p values between those two. That seems beyond what we can really draw conclusions about. Am I right about that or am I misinterpreting here?
DR. FLEMING: Actually my sense about this is similar I believe to what I understand JoAnn is saying. When I look at this, there is some level of reassurance about the irbesartan effect against placebo when you look at it against amlodipine and you track that same effect. Of course, the extent to which I can draw that reassurance is based on the assumption that amlodipine can really be viewed as a placebo.
Where I worry is that I don't know whether we can say that amlodipine is a placebo, at least as it relates to the measures on the primary endpoint. Certainly when we get to the cardiovascular measures, if we're going to pool amlodipine with placebo, then I almost feel like, gee, is that really fair not to pool it where it's going to make irbesartan look worse, which are the cardiovascular endpoints. If we do that pooling relative to other measures such as cerebrovascular events, MIs, neurologic abnormalities, if you pool amlodipine and placebo and compare it to irbesartan, it looks like irbesartan is 30 percent worse. Well, I don't believe that either. I think what's happening is amlodipine is better than placebo. So, to then pool amlodipine with placebo in those measures that will make the statistical strength of evidence look better seems to be a little bit, at best, arbitrary. So, there is this clinical issue, can you pool this when you're really having to essentially say, to strengthen your evidence, I'm willing to say amlodipine is a placebo. So, there is that clinical complication.
There's also a statistical complication. If I allowed myself to generate a p value by essentially comparing to the placebo and comparing to the placebo with another arm in the trial and view that whichever one of those p values look more impressive and report that p value as being meaningful, you're going to have an inflated risk of false positive conclusions. You can't conditionally pool something from another arm with my control arm if it's going to strengthen my evidence. It would be interesting.
Would it have been pooled had it weakened the evidence?
Would we still have done the same pooling analysis of amlodipine against control if it would have weakened the strength of evidence because amlodipine itself would have carried some benefit on this endpoint?
So, bottom line is I strongly object to anybody sprinkling p values on such ad hoc suspect analyses to, in a sense, strengthen the interpretation of those.
On the other hand, coming back to what JoAnn said, I think there is some level of reassurance. It's not remotely what I'd call a second trial reassurance, but there is some level of reassurance by saying that the amlodipine arm was similar to the placebo arm and the irbesartan was better than each of the two.
DR. BORER: Steve?
DR. NISSEN: I think I'm agreeing with you, Tom. We can't have it both ways. We can't say amlodipine is placebo-like for one set of endpoints, but then ignore the others. The minute we start to do that, we're creating an anomaly. It seems to me that if we look at amlodipine as placebo, then we're forced to compare what happened with irbesartan and amlodipine with all those other endpoints. Clearly, there are several of them that go disturbingly in the wrong direction.
So, I think you have to look at the totality of the data here, and in that sense, I don't find it reassuring at all because, for a patient, you really have to ask the question. The patient enters a clinic and you have to decide which drug you're going to give them, and I think if some endpoints go in one direction and some go in the other, the net clinical benefit is very hard to establish and certainly doesn't strengthen the evidence against placebo to lump amlodipine in the same category.
DR. BORER: Let's go on to number 4. Comment on other secondary endpoints in IDNT.
4.1. There was a prespecified analysis of time to first cardiovascular death, nonfatal MI, CHF hospitalization, disabling stroke, or amputation. There were 416 such events with no significant difference in the distribution among groups. Is this further evidence of a lack of clinical benefit? Is it comforting that there's a lack of apparent harm? Were there simply too few events, et cetera?
4.2. We discussed part of that here. There was a prespecified analysis of time to first cardiovascular death, nonfatal MI, coronary revascularization, CHF hospitalization, need for ACE inhibitor or ARB for heart failure, disabling stroke, amputation, or peripheral revascularization. There were 518 such events with no significant difference in the distribution among groups. Is this further evidence of a lack of clinical benefit? Is it comforting that there is a lack of apparent harm? Were there simply too few events to show a meaningful effect?
DR. LINDENFELD: Once again, we have to come back. Lack of clinical benefit. I think the primary endpoint here was renal disease and that's where we really want to show the clinical benefit, and we've discussed that data. So, in this study, in this trial, I think we have a lack of a clear-cut clinical benefit in these cardiovascular endpoints, but certainly this doesn't imply a lack of clinical benefit on end-stage renal disease.
Now, again, the point has come up over and over again. If we see this doubling of creatinine, why is it not reflected in these other events? But again, we've discussed that, and these are two different things. This lack of clinical benefit for cardiovascular outcomes doesn't dissuade me that there's a clinical benefit in renal disease, which is real.
I don't believe there were too few events to show a meaningful event. Perhaps the study needed to go longer. Maybe that says that, yes, there were too few events. But I don't believe there were too few events. I can't explain the lack of cardiovascular outcomes here.
DR. BORER: Any other comments here? Tom?
DR. FLEMING: A very brief addition to that. The sense in which I might argue there could have been too few events is we're estimating something like an 8 percent reduction, and that's not significant. It's informative in that it's suggestive that the actual effect, if it's real, is very modest. I'm not willing to say that these data prove that there isn't an effect on cardiovascular events, and in that sense it's too small a trial. We probably would have needed a much bigger study. If we would have viewed, for example, that conclusively establishing that the 8 percent is real, that would have taken a huge study.
So, this is the third domain that I had referred to in my answer to question 2, and my sense of that is consistent with you, JoAnn, that the first domain is what the intention and the focus was. It's still relevant to know what the third domain showed because these are very clinically relevant endpoints. And what the data show is they suggest that if there is an effect, it's very modest and it would take a much bigger trial to sort out whether there is in fact a very modest effect on cardiovascular events versus no effect.
DR. BORER: Now, the next question we actually have to -- I'm sorry. Go ahead.
DR. LORELL: I appreciate your insights on that. I think they're very helpful. I think it's also very much worth emphasizing that the treatment design in this appears to have had at least two potent cardioprotective interventions that were seen in all three groups. One was aggressive blood pressure control. A second was that a relatively high number of patients were on profoundly cardioprotective beta blockade. We weren't told about aspirin, but I'll assume, unless I'm corrected otherwise, that aspirin use was comparably distributed. So, I would look at the way these patients were treated as having a very powerful cardioprotective intervention that was done in all three groups, and I think that may have partially blunted the ability to see any difference because of the low event rate.
DR. BORER: Bob?
DR. TEMPLE: Yes, I think I had much the same comment. We keep saying there was no difference, but there really isn't any hypothesized difference. They're all on appropriate regimens with all kinds of stuff. There's a hypothesized difference in renal events, but there isn't any hypothesized difference in any of the others. I mean, there is some disturbance about the fact that amlodipine looks a little better on some of those. That's certainly something to think about. But you wouldn't really have predicted an advantage in those events in this setting unless somehow the renal events led to fewer of the other events, and it probably wasn't followed long enough to see that.
DR. BORER: Steve?
DR. NISSEN: Just in response, part of the reason why you might have hypothesized that is if progressing to doubling your creatinine and getting renal failure is a very bad thing, leading to myocardial infarction -- we've all heard that once you get to end-stage renal disease, you've got this terrible cardiovascular morbidity and mortality, and therefore preventing that might be expected to prevent those secondary consequences. So, I guess I think you could hypothesize that. Even though the sponsor didn't necessarily power it for that, I think we wouldn't have been shocked if we saw that.
DR. BORER: The next question we have to stand up and be counted on.
Are the results of IDNT alone an adequate basis for approval of irbesartan for the treatment of patients with type 2 diabetic nephropathy?
JoAnn, why don't we start with you and then we'll go to that side of the table and move around?
DR. LINDENFELD: I would say no to this question. I think that the study shows an improvement in the doubling of creatinine, but we've generally required two studies at .05 or one study at a much lower p value than this. In addition, there's a small number of endpoints. So, just as a standalone study with no other data, I would say no.
DR. BORER: Dr. Brem?
DR. BREM: I would make one comment and it comes to the point you made very early in the discussion and that is, is this for diabetic nephropathy or hypertensive patients with diabetic nephropathy?
DR. BORER: I think we can define how we want to interpret that. Why don't you carry through the thought?
DR. BREM: I think the way it's written here -- Dr. Lipicky, if I'm misquoting you, please interrupt -- I think he's trying to get at an indication for diabetic nephropathy, yes or no, independent of the hypertension. And I'm not sure on one study of this nature that we have enough information to make a blanket approval.
DR. LIPICKY: Then to make it easy, make it with hypertension. So, I'll call your bluff.
DR. BREM: I think even with hypertension, I'm not sure this study alone, in the absence of everything else --
DR. LIPICKY: You've answered my question.
DR. BORER: Dr. Kopp?
DR. KOPP: I had a question about those standards. It's two trials each at .05 or one trial at .00125. Where does that second number come from?
DR. LIPICKY: Both Tom and Dr. Temple are here to amplify on what I'll say. But basically if you just take the common sense view, that if somebody finds something, well, one has arbitrarily by history defined finding something as a p of .05. Usually you say, well, Tom found that. I'd like to know Harry found that too. Almost everybody says maybe Tom is right, but I want to know someone else found it. So, that's another .05.
So, if you now require for your standard of evidence -- and I'm not sure you should; in fact, I am advocating you should, but I'm not sure you should -- two trials of .05, statistically that's .05 squared. So, then you have to divide by 2 because you have to end up in the same distribution of the tails. And that comes out to .00125.
So, you have this various grading then of strength of evidence from the convention, 1 chance in 20 of being wrong, to a really very small chance of being wrong. One has to make the decision where you think this strength of evidence is. And I would maintain that you ought to be closer to the two studies at p of .05 than to one study at a p of .05 because one study at a p of .05 is just too shaky.
DR. KOPP: Clearly then my answer is, according to those standards, this doesn't make it. I agree that I don't think we can consider this two independent studies so we don't have that criteria met.
DR. BORER: Bob?
DR. TEMPLE: I just want to comment a little further. We have just been at a workshop on this discussion.
Historically the agency always said that you need independent substantiation of a finding, basically in the form of another controlled study. The Food, Drug and Cosmetic Act was altered in 1997 to allow us to reach a conclusion on the basis of a single study with what is "confirmatory evidence," whatever that means because that has never been properly defined.
We've written a lengthy document on what constitutes good enough evidence and have generally said a couple of things. First of all, other data from other studies, maybe with a different endpoint, can sometimes help you believe in one study. Obviously, that's a matter of judgment. And we've also said that when really all you have is a single study, it ought to be at a more extreme p value, confidence interval, whatever you care to do. Whether that translates to .00125 or .001 or whatever is again a matter of judgment.
But it is fairly clear that we're allowed to think about -- and the document says this -- data from other sources. Now, this is anticipating later. But you're entitled to take into account such things as the other study showing a different endpoint that may or may not be relevant. How to do that is an intense matter of judgment. I wouldn't try to tell you what to do, but you're permitted to reach that sort of conclusion. You even can think about related drugs, if you want to. But how to do those things and what the precedents are is very iffy, and there aren't very many. So, you're in substantially uncharted territories, but you're allowed to think.
DR. LIPICKY: Just to add to the part of you're allowed to think and nobody knows what the right answer is, it isn't just the p value. Right? It's partly, well, yes, you made a p of .05, but if you change one patient, and now you're at a p of .1, well, geez, that's not really a p of .05, just as in this case, it's a p of .02, but if you lose a few patients, it's .07. Now, that's a big difference. So, part of the question is not prior knowledge or is it a p of .05, but how robust is the data. How likely is it that if you take the numbers you're looking at and act on them, you would be making a mistake? So, it's another part of the whole business, and it doesn't come down to p values only.
Nor is it really one study/two studies. I mean, you could have one study that has a p of .01, let's say, and is so robust that you wouldn't possibly think that it could turn out any other way. Or as Dr. Temple says, you know so much that you would have predicted that, and indeed this now turns out that way.
And so, there's all kinds of this. That's what's being talked about now. Where are you on this continuum of your confidence that what the trial found is real?
DR. BORER: So far, to summarize, we're at 3 to 0 against, in terms of question number 5, and we'll go to Beverly Lorell.
DR. LORELL: Well, picking up on Dr. Temple's comment and on your comment, Dr. Lipicky, I'd welcome some discussion among the committee about their interpretation of the supportive value of the RENAAL study. Admittedly, it's a bit on uncharted ground, but at least to my mind, those data in a very similar design --
DR. LIPICKY: You haven't seen it. I think our proceedings here should be related to data you have seen and where you have seen a whole review like you've just seen of this, and there may be things that you know about that haven't had that degree of stuff and I don't think you should count that. Dr. Temple may think differently.
DR. TEMPLE: Well, I think we've vetted and the committee has vetted captopril data, so you might think that was relevant.
It is hard to take the RENAAL study into account because you haven't had an opportunity to see it, although we have.
DR. LIPICKY: Well, but that is a difference.
DR. TEMPLE: Yes.
DR. LIPICKY: And we would represent to you that the captopril study is as it was. We couldn't make that representation for the RENAAL.
DR. TEMPLE: We tend to be nervous about -- no offense to anybody -- presentations in journals without an opportunity to see the data, even though everybody is trying his or her best.
DR. BORER: I think we'll get to the strength of supporting evidence in the subsequent questions, but maybe we can try and deal with this one first, which is specifically, if you look at IDNT alone, is that adequate for approval?
DR. LORELL: Well, but in response to that explicit question, I would say no.
DR. BORER: Do you want to state a reason?
DR. LORELL: For the same reasons that have been discussed, that it is a single study with a modest p value.
DR. BORER: Dr. Cunningham?
DR. CUNNINGHAM: I would also say no for the same reasons.
DR. BORER: Mike?
DR. ARTMAN: I would say no. I think there are a lot of confounding issues. We really haven't delved into some of the issues related to polypharmacy and whether that was all controlled for, et cetera. I think that the issues related to gender and ethnicity -- there was some hand-waving.
We've talked about the issues related to the black population and we're told that that couldn't account for differences. Then we looked at North American versus European and they said, oh, well, that's because all the black people were on the North American side and that accounts for the difference.
You know, I just am underwhelmed. And the mantra that Ray has instilled into us has been does the intervention make you live longer or feel better, and I don't see compelling evidence for either one of those. So, I would say no.
DR. BORER: Tom?
DR. LIPICKY: You cannot blame it on me.
DR. ARTMAN: Oh, I blame everything on you.
DR. BORER: He wasn't blaming it on you. He was giving an explanation.
DR. ARTMAN: I'm giving you credit. I'm attributing it to you.
DR. BORER: Tom?
DR. FLEMING: Well, issues have been discussed, but essentially when I look at data from a single trial and I'm confronted with the question should this study, at least in my own recommendation to the FDA, be viewed as adequately convincing. We lose a little bit with the single study of the replication concept. That is important. It's not just a p of .025 squared times 2, which is what .001 is. There is that merit to being able to see an independent set of investigators maybe in a somewhat related setting being able to show that the results of positivity could be confirmed.
Having said that, though, I do accept that a single trial in settings could be adequate, and I certainly am influenced a bit by what the strength of evidence is when you say .001, i.e., the .025 squared. There are settings that would move me away from even saying I would need to see that from a single trial, if I'm looking at a mortality endpoint, if I'm looking at secondary measures that are strongly reinforcing primary.
So, in this setting, I completely concur with the sponsor's perspective that the first focus of this is the renal components and dialysis. And when I look at that, I see some p values that are in the neighborhood of .025 to .075, something lurking around .05. When I look at the secondary measures, I don't see that they have to be positive in order to make me view this as a single positive trial, but I do think that when the primary is about .05, I do need to see those secondary measures showing positive reinforcement for this study to be judged in its own right as a single convincing trial. And for mortality and for the cardiovascular endpoints, there's not evidence of benefit.
My view of that is I think this is just on the edge of what I would consider adequate strength of evidence for this to be called, just barely, a single positive trial, but I don't see it as meeting any of those other factors that would bring me to a much more convincing perspective that this study conclusively establishes benefit at the level that I would wish to have as a standard for strength of evidence from two independent studies.
I guess the last point is -- and I don't know what FDA's view about this is -- I would also be persuaded if this was a setting that was a rare setting that would be incredibly difficult to enter patients. This is a setting where this is going to be very widely used, and I think having a standard of being adequately convinced it's effective is particularly compelling in a setting where you've got an intervention that's going to be so widely used.
So, I look at it as a study that just does get into the realm of strength of evidence for being called a single positive study, but I couldn't see an approval being justified based on this study alone.
DR. BORER: Blase?
DR. CARABELLO: I would vote also no. I think it was a single study. I found the amlodipine data helpful in helping me believe that this was not simply an effect of blood pressure lowering, but I was mostly disturbed, despite discussion to the contrary, about its lack of effect in women in North America. I just am bothered by the fact that that subset analysis seemed to be so weak.
DR. BORER: Steve?
DR. NISSEN: Well, one of the questions I ask is, although it is off-label use, almost all these patients now are getting treated with ACE inhibitors. A recommendation to approve will cause a shift in prescribing practices. So, what level of evidence do we want to have to actually cause that to take place?
The p value here is really .035 for the primary endpoint, and if you'll recall, the sponsor's analysis of the blood pressure differences suggested that at least some of the positivity was due to that. So, now we're getting perilously close to even the standard for a single study.
You add to that the confounders, as in race, gender, and location, North America versus not, and now there are just too many confounders that could take this on the wrong side of even being adequate as a single study. So, I just think there's just not compelling evidence from IDNT to approve. So, my vote is no.
DR. BORER: Alan?
DR. HIRSCH: The first time I think I've ever gotten to speak last, and yet I've learned how to use the word opine.
DR. HIRSCH: First, I have to say to my previous instructors, Dr. Brenner and Dr. Lewis, I also heard you and I have absolutely no doubt that ARBs alter the structure and hemodynamics of end-stage renal disease or the kidney proceeding to end-stage renal disease. In other words, the paradigm I understand is important and affects a great number of patients who will ultimately in this country and the world die of their disease. But I opine no as well, and I should justify why.
The same issues. I'll repeat a few of them. Single trial, I think, whose statistical significance is borderline. For me the supporting data and the secondary endpoints and the use of IRMA 2, though they support the pathophysiology, in general don't yet convince me that the single has adequate power.
Like Dr. Lorell, I certainly am aware of published data from losartan and RENAAL, and that helps me but we're not there yet. So, I can't include that in my analysis.
A little bit like you were saying, Dr. Temple, whereas if I were a manuscript reviewer, this is clearly an important trial and significant, our role as advisors to the agency is different. There's a higher standard of evidence because it will change practice. So, I say no now.
I'll go a step further to set up the discussion later I think that you wanted, Dr. Lorell. We do change precedent by how we interpret trials, and I fear that when we take a single trial, as you might imply, and permit the global data to change our analysis, that we permit use of surrogates that we're not all quite comfortable with, number one, that we might permit a somewhat low sample size to be used not to understand why there's no efficacy in North America when, in fact, we're regulating North American use -- or I should say American use. And I worry that then we'll also promulgate incomplete follow-up regarding those cardiac events in future trials.
So, overall, looking at a large potential use with a very, very important disease, it doesn't quite reach that level of significance. So, I opine no.
DR. BORER: Opine is Ray's usage.
Just with regard to the strictly stated question number 5, I'm going to vote no as well, but I want to give some explanatory statements.
First of all, although I agree with the thrust of Ray's suggestion earlier that diabetic nephropathy is diabetic nephropathy, and the presence or absence of hypertension probably -- probably -- doesn't alter the fundamental nature of the disease. Nonetheless, when we're talking about approval of a drug, we have to consider the efficacy and the safety for its intended use and the balance between the two. And I really have no information at all that would allow me to give an opinion about that with this drug in patients with diabetic nephropathy who don't have hypertension. So, just as Dr. Brem suggested, I would limit my consideration of this drug to patients with type 2 diabetic nephropathy with hypertension. Those are the patients we saw where the risk/benefit relationship may be different than in the other populations.
Having said that, I agree that it's a single trial with a level of consistency, indicated by the p value that's relatively close to the margin that nominally we accept, and there are some other issues.
However, and perhaps to presage aspects of the discussion that we'll get into, I really don't have any trouble with a creatinine of 6.0. I'm not a nephrologist, but my understanding of the literature and my clinical experience is that when patients have dramatically subnormal creatinine clearance, as people with a creatinine of 6.0 do, they progress, and they progress relatively rapidly. And if they're not dialyzed, then they will die, and before they die, they'll be very uncomfortable people. I don't need a set of data collected by the sponsor about the effects of pericarditis, the effects of anemia, the number of episodes of nausea and vomiting to believe that because I think it's been well documented in the literature, and I think that nephrologists probably know that and people in other subspecialties may not have the same feeling for it. But I do see this patients with some frequency because of my focus on patients with valve disease who have cardiac surgery. So, I have no problem at all with the endpoint of 6.0 or dialysis or transplant. I think the one is a short step from the other.
And I have really no particular problem with the doubling of creatinine as a pretty solid predictor of the progression to these bad endpoints that we don't want people to get to.
Having said that, I think that I to a lesser extent and the entire committee perhaps to a greater extent would feel more secure. We would have a more secure view of the data and the interpretation of the data if in fact we did have that additional information that Tom had asked for earlier about the progression to dialysis and the progression to transplant beyond the first event ESRD. So, I'd like to see those.
I think if at the end of the day we don't come out voting in favor of suggesting to the FDA that they approve this drug for the requested indication, that those data should be obtained and given for review because they might change the opinion of some of the people who are looking at these data, specifically the kinds of things that Tom was asking for.
I'm really not terribly concerned about the gender and ethnicity issues. I don't want to get into mechanisms. I'm already on record as telling Tom at an earlier discussion at another meeting that I have no idea how any drug causes its clinical benefits, but I can talk a little bit about pharmacologic effects.
I think that there's an analogy here. The gender issue, the ethnicity issue, all the other substudies are indeed substudies. If we're concerned about them, then we could suggest that the FDA say something about that in the label and note the lack of information or the lack of security in certain subpopulations. But they are substudies. They're post hoc assessments. There was no hypothesis being tested there. So, I'm not terribly concerned about that.
And I'm also a little sorry that we got into such a detailed -- I'm not sorry that we got into the discussion, but that the issue of the nonfatal cardiac endpoints seems to assume such great importance because I am convinced that when you look at the totality of major events, that there are fewer major events on drug than without, although there does seem to be a different distribution of some of those cardiovascular events than the renal events which causes you to lose a little bit of confidence in the strength of the overall conclusion.
So, again, in summary, I believe that IDNT alone isn't an adequate basis for approval of irbesartan for treatment of patients who are hypertensive with type 2 diabetic nephropathy, but I'm not as concerned about some of the other issues that have been raised as you've heard from some of the other committee members.
DR. TEMPLE: Jeffrey, I think not everybody who voted clearly referred to this study alone. You just did. The question was designed to not have you consider IRMA yet and just go on this study, but it wasn't clear to me everybody was treating it that way.
DR. BORER: I think Alan didn't and one or two others didn't, but I think most everybody focused on this alone.
DR. TEMPLE: The second observation I want to make is we've been severely criticized for putting a mention of a subset in the labeling, referring to the MERIT trial, where we thought we had better than usual cause. It's just worth observing that here the subsets are extremely small and it would be a miracle if they all went in the same direction. So, we're having some trouble for doing that at all. Just so you know, people are threatening not to include U.S. patients in their trials because we pay so much attention to it, but don't worry about that.
DR. BORER: Well, just for the record, nominally they did go in the same direction. The magnitude of the effect was small, but as you've said, small numbers, post hoc. I don't know what you make of that.
DR. FLEMING: Jeff, one other quick clarification. I think there may be more concurrence in what you were saying than you might have suggested in handling the primary analysis. You were saying you were persuaded that end-stage renal disease, which includes in its definition a creatinine at level 6, would be an adequate clinical endpoint, as would dialysis. I didn't hear anybody disagreeing with that.
And when you referred to my interest in seeing more data, if one accepts that these end-stage renal disease endpoints are clinical endpoints, one gets a p value of .07. If one uses end-stage renal disease as the primary endpoint, and if you look at dialysis, you get a significance level along that line as well. If you use the primary endpoint, as they had defined it, which is a twofold increase in creatinine, then you slip just on the other side of .05 to .023. So, when you said you would accept that, basically, at least in my own comments, when I say you're on the edge of .05, it's accepting end-stage renal disease as clinical endpoints.
DR. BORER: Let's move on then to 6, 7, and 8. IRMA 2 randomized 611 subjects with type 2 diabetes and microalbuminuria to placebo or irbesartan, two doses, for 2 years. The primary endpoint was time to progression to overt proteinuria, and the analysis plan compared each active arm to placebo. The results ordered by dose, but only the 300 milligram dose group was statistically significantly different from placebo.
Number 6. Comment on the handling and implications of premature withdrawal of 166 patients, 27 percent.
DR. LINDENFELD: Well, patients who reached the endpoint of overt nephropathy were withdrawn. The implication, of course, is that that makes it difficult for us to see ultimately effects on GFR.
DR. BORER: What does that do to your level of --
DR. LINDENFELD: That's coming up, I think, in another question. But it makes it difficult on the basis of the data to presume that a reduction in proteinuria reflects a change in creatinine clearance.
DR. BORER: Any other comments on that point?
DR. BORER: No? Then let's go on to number 7.
There was a trend toward a greater increase in the rate of change in serum creatinine on irbesartan than on placebo. Comment on the hypothesized relationship between proteinuria and renal function as evidenced by creatinine clearance.
DR. LINDENFELD: I think I would make the same comment again. This data just doesn't allow us to make a relationship between proteinuria and creatinine clearance.
DR. BORER: That answer certainly stands.
I think -- and perhaps you don't think it's worth doing this, but I think the issue that Ray may be getting to us about is that the 300 milligram dose caused a greater fall in creatinine clearance than the 150 milligram dose or than placebo. And we heard that that may be a good thing. What do you think about that?
DR. LINDENFELD: Well, as I understand the explanation, that was an early effect and then stabilized after that early effect. I'm not concerned about that effect.
DR. BORER: Number 8. A 133-subject subgroup was randomized to have GFR measured at 3 months, at the end of active treatments, and then 4 weeks after the last dose. At 3 months and at the end of active treatment, there were no statistically significant differences in GFR between placebo and either dose of irbesartan. 4 weeks after the last dose, GFR increased in all three treatment groups. Differences from placebo were again statistically non-significant, or perhaps not statistically significant. Comment on the hypothesized relationship between proteinuria and renal function as evidenced by GFR.
I think it might be fair, unless Ray doesn't think it's fair, for us to include in that discussion not just the GFR substudy, but the other data that we saw for the entire group.
DR. LINDENFELD: I would comment here that we've seen a lot of information suggesting that the changes we would see with angiotensin receptor blockers are likely to be permanent changes, or at least if not reversal of the underlying disease, prevention of advancement of the underlying disease. And when one removes the irbesartan, at least the 150 milligram dose, and sees a return right back up to placebo levels, that makes us think that this was a hemodynamic effect of some sort rather than perhaps a clear-cut change which we would expect to see longer. It can't just be a blood pressure change, the fact that the blood pressure was allowed to go up, because we didn't see that same thing happen in the 300 milligram group. So, on the other hand, the 300 milligram irbesartan group did have a persistent lowering.
So, it doesn't help me. It certainly doesn't add to this relationship between proteinuria and GFR, but I don't know that it subtracts from it either.
DR. BORER: How about the relationship between irbesartan and ARB and proteinuria? I think one of the thrusts of the question here may be does the delay in any loss of apparent stabilization of proteinuria with a 300 milligram dose, after you stop the 300 milligram dose, give you any sense of the action of irbesartan compared to placebo, for example.
DR. LINDENFELD: Maybe I'm not quite understanding this question. Why don't you repeat it or rephrase it for me.
DR. BORER: My understanding of these data are that they were shown to us to suggest that because proteinuria didn't return even really towards baseline 4 weeks after stopping the 300 milligram dose, that in fact there was some protective effect that was maintained after stopping the drug, as compared with placebo or the lower dose where things moved back towards baseline. And should we draw any inferences from that finding about the activity or presence of beneficial activity of irbesartan?
DR. LINDENFELD: Well, I think it's marginally helpful. I'm concerned. I would have liked to have not seen the 150 milligram group go right back up to the placebo level. So, the 300 alone -- you know, if we had a 250, what would that have done? It's helpful but it's not enormously persuasive.
DR. BORER: Dr. Kopp?
DR. KOPP: Well, I think the sponsor was careful not to speculate, but I won't be so careful. So, one possibility is that the low dose is operating purely hemodynamically and the higher dose has some additional structural effect, even an antifibrotic, not just a stabilization effect. So, one possibility is that during this time of suppression of angiotensin II activity, TGF-beta, and so forth, there's the possibility for some remodeling to have occurred so that structurally you're better off at 24 months, even without the drug, that you were at the beginning. Obviously, without biopsies, who knows? But it does suggest there's some structural benefit, not just stabilization.
Either that, or 4 weeks wasn't long enough and there's some residual effect that is clearly -- in that situation, I'm not saying there are drug levels around but some change in cellular phenotype has been maintained that doesn't reverse. Of course, it would be nice to see the same thing at 3 months. That wasn't done.
But I think it is favorable that 300 milligrams had a long-term effect even in the absence of the drug for 1 month.
DR. LIPICKY: But you have answered the question I think unless you want to discuss it some more.
DR. BORER: We do. Bev?
DR. LORELL: I think that at first, in hearing the discussion today, there did seem to be some disconnect between the behavior of microalbuminuria and creatinine clearance. But I think, on the other hand, the point was made in the discussion as an hypothesis for which there is support, that the somewhat disparate behavior of creatinine clearance may have been related to hyperfiltration associated with hypertension in removing that component of hyperfiltration.
But I actually did find it both interesting and supportive that, in terms of looking at the primary endpoint of microalbuminuria, that that benefit was not only persistent but appeared to even go in the improvement direction with stopping the drug for 4 weeks.
DR. BORER: Well, let's go on to number 9 and here's another one where we have to make a statement into the microphone.
Are the results of IDNT plus IRMA 2 an adequate basis for approval of irbesartan for the treatment of -- however you want to say it -- hypertensive patients, or if you don't want to be hypertensive, then any patients, who have type 2 diabetic nephropathy?
JoAnn, why don't you start and we'll go around the table again.
DR. LINDENFELD: This is obviously a difficult question. I would answer no to this. I think that the IRMA data is supportive, but it's not quite enough additional data, given the things we discussed in the IDNT study. It's not yet quite enough additional data to make me feel comfortable that all the data we have is convincing.
DR. BORER: Why don't we start on the other side this time. Mike?
DR. ARTMAN: I actually liked the IRMA 2 trial a lot better. I thought the rationale, I thought the data were compelling. And, yes, there is this disconnect between the early direction of the creatinine clearance. That was at the same time that there was the greatest reduction in microalbuminuria. So, I think those data are compelling. However, I don't believe that they're robust enough for me to support the approval. So, I would agree with JoAnn and say no on this one.
DR. BORER: Tom?
DR. ARTMAN: One more comment. I do think that if we are going to approve an indication, it has to be for the population from which the data were gathered. So, I agree strongly that it would have to include the word hypertensive patients, and I'm surprised that Ray, who has taught us again not to stray from the study population, would try to sneak that in.
DR. BORER: Tom?
DR. FLEMING: I think the IRMA 2 trial provides us evidence that there is an effect on progression for microalbuminuria to proteinuria, and there is a lot of evidence. Dr. Lewis gave a very informative presentation about natural history and that this is an important step in the cascade of events that lead to very significant and important clinical consequences.
However, we have a myriad of examples and experiences to know that a correlate does not necessarily make a surrogate, that in fact now having a treatment-induced effect on that marker is reliable evidence of a treatment-induced effect on the clinical events that are down the cascade that we're really trying to prevent. We weren't even able to directly assess the next step, which is GFR rates.
My sense is IRMA 2 is informative. It establishes an effect on an important early phase marker that I believe does provide enhanced plausibility of efficacy. That type of data, though, typically in my view requires confirmation then in studies that would, in fact, more reliably demonstrate the effect.
We have one study which, in my own view, is on the edge of what would be strength of evidence for a single positive study. Now we're talking about a study on a marker. If the first study had been much closer to being convincing, I would have found the two together to be adequate, but a second study that establishes an effect on a marker that does, in fact, provide enhanced plausibility of efficacy, but falls far short of what we would consider as strength of evidence for a single study for establishing benefit, that doesn't add in adequate strength of evidence to the first study to make it a convincing package from my perspective. So, the two studies I believe together wouldn't provide an adequate basis for approval. My vote would be no.
DR. BORER: Blase?
DR. CARABELLO: Essentially we've been told that this disease is a continuum, a trip from New York to San Francisco. And I think we clearly have the piece from New York to Cincinnati, and I believe IRMA 2. We have the trip from Chicago to Denver. I believe that the drug helps to prevent the doubling in creatinine. It's the Cincinnati to Chicago and Denver to San Francisco pieces that aren't there that I wish we had to complete the whole story that would make this a more convincing argument. So, I also would have to say no.
DR. BORER: Steve?
DR. NISSEN: We've got lots of examples where a drug may be effective at one phase of a disease process and not so effective in another phase, and that's why it's hard to put the two together because they don't support each other as much as they might if they were looking at a similar population. So, that's the problem I have, is both studies taken separately are useful, but I find that I can't combine them in making any kind of reasonable decision because I'm not convinced that the process is the same early in the disease and late in the disease without more evidence that that's the case. So, my vote is no.
DR. BORER: Alan?
DR. HIRSCH: It's not often when we all sort of come to similar conclusions.
I think the two studies have great beauty and actually do provide great help in advancing future care for patients with renal disease. I'm impressed. And the goal, of course, is to change outcomes, so I will summarize again.
One, I do think we have a signal of efficacy in two separate trials. Yes, I think that IRMA 2 is supportive of IDNT in the sense that we've shown a signal that's positive. But again, I find these are different signals in different populations, and therefore I really have a hard time combining them.
Again, I would emphasize the natural history continuum. There's been vision in place in these things along the natural history, but I find each of the signals along the continuum to be just weak enough that I can't connect them, making the metaphor to crossing the country by airplane. So, therefore, with two surrogates, an improvement in proteinuria and an improvement in doubling of creatinine, they would need to be linked yet again in my mind to a stronger clinical outcome to achieve approvability on their basis alone.
DR. BORER: Dr. Brem?
DR. BREM: It's difficult to top the community summation, and I certainly agree with it. Again, what's missing is the difference, the leap between advancement of microalbuminuria and change in renal function, which is what we all believe probably occurs, but hasn't been convincingly shown.
Based on just these two studies alone, in the absence of any other information, I would have to agree with my colleagues and vote no.
DR. KOPP: I would vote to approve. I think a couple of a points I'd like to make. One is that this is a continuous process histologically so that if you do a kidney biopsy of somebody in the microalbuminuric phase, it looks exactly like that of somebody in the later phase. So, there's no reason to think that the histology is different. In fact, there's reason to think that it's the same. It simply becomes progressively more severe, as you saw this morning, wider glomerular basement membranes, more mesangial matrix expansion. So, I think it is a continuum of one disease.
Earlier I spoke against the first study, but I think I focused on the glass being half empty. I take it that the glass for the first study is half full. Admittedly, the p value is only .023. Although I found all elements of the primary composite endpoint to be convincing -- I'll go in the reverse order -- death, dialysis, and a creatinine of 6 I have no problem being very hard clinical endpoints as I think most of us did.
For me a doubling in serum creatinine is very worrisome and is, as I think Dr. Lewis was trying to make the case, more than just a surrogate but actually a measure of renal function such that in this disease, in particular, but really in most nephrotic conditions a certain sign that this patient is destined to progress absent further therapy.
So, I took the composite endpoint to be quite convincing, and my only limitation was that the p value was .023 with the caveats that if you argue about particular situations, it might drift to a .05.
The second study with the higher dose of irbesartan had a p value of less than .001. And I am convinced, as I say, that this is the same disease treated at two different points. I do take the comment over here that because we're studying two different points, they are not exactly in support of one another, but I choose to focus on the other side that it's the same disease process we're treating, and so I vote to approve.
DR. BORER: Bev?
DR. LORELL: Thank you. I thought that the IRMA 2 study was really a very well-done and very beautifully presented study. I would view it as a supportive study and not as a second study of the same weight as the IDNT trial in terms of changing practice for a very large number of patients in the United States.
I would vote no. I think that that is very close. I would like to see additional supportive data for some of the harder endpoints that we discussed earlier around the table.
DR. CUNNINGHAM: Well, I think I see the glass as being maybe one-third full. I think it's also supportive and somewhat convincing.
But I have to say, as a person who's sitting in the consumer seat, that what I see as the real problem here is the drug that wasn't study, that being the ACE inhibitors. And I think from the consumer perspective, that's really what we'd like to see the data on. So, my vote is no.
DR. BORER: Just as a point of clarification, we all would like to see that I think, but we're really being asked to judge this application not what --
DR. CUNNINGHAM: I realized that. That's why I said no. But I wanted to put on the record somewhere along the way that that was my view.
DR. BORER: Okay.
I'm right on the borderline, but not to presage any final comment here, as we go down the question, I'm at this point still on the minimally negative side. I agree with everything that Dr. Kopp said. I believe that IRMA 2 deals with the effects of this agent on the same disease at a different point and it's very positive. It would be lovely if we had the GFR data, and as a non-nephrologist, it's probably not appropriate for me to make the jump from proteinuria data to GFR data, although I'd be willing to do it.
I'm not concerned that we don't have biopsies because I don't think we could get them. I think Beverly said it before, and I'm convinced by the information we were shown that we have enough information to be reasonably certain what the biopsies would show if we had them.
So, I think this study is strongly supportive and I think that it gets me right, just about, to the point where I'd be willing to vote for approval, but not quite. I'd like to see just a little bit more. Maybe those data are available or maybe that little bit more will become clarified as we go down through these questions. So, I'll reluctantly, still at this at point, vote no.
With that, let's go on to number 10. A drug with a related mechanism of action, captopril, has an indication for diabetic nephropathy in patients with type 1 diabetes. The primary basis of that approval was the demonstration in a 409-subject 2-year study of a 51 percent reduction, p equals .004, in risk of doubling serum creatinine, and a 50 percent reduction, p equals .006, in risk of mortality or end-stage renal disease. Both effects were manifest in the first few months of treatment. Captopril also reduces the progression for microalbuminuria to overt proteinuria.
10. Are the results with captopril germane to a discussion of irbesartan? In particular, is nephropathy in type 1 diabetes enough like nephropathy in type 2 diabetes? And 10.2, are the pharmacological effects of captopril and irbesartan adequately similar?
DR. LINDENFELD: I believe that we've heard enough today and seen in our background booklets that, yes, the nephropathies in these two types of diabetes are quite similar and would be expected to respond similarly.
In terms of the second point, of course, the pharmacological effects are not exactly the same. But I believe that we've heard some data today and there's some data that exists that the effect on renal function is at least in great part an angiotensin effect. So, I think there are enough similar mechanisms to make the data with captopril helpful.
DR. BORER: Dr. Brem?
DR. BREM: I agree. Although there are obvious differences in the first captopril study that have been well described already in terms of age and blood pressure normalization, I believe that the basic progression of disease is probably similar enough in both models or both types of diabetes that it would be expected that both should behave and respond to treatment in a similar fashion. So, I think they are germane.
DR. BORER: Okay, that's a yes.
DR. KOPP: Yes, I think they are germane.
DR. BORER: Beverly?
DR. LORELL: I agree.
DR. BORER: Dr. Cunningham?
DR. CUNNINGHAM: I don't know if I'm convinced that the pharmacological effects are the same, but I think they're certainly useful.
DR. BORER: Do we need a more specific yes or no there?
DR. CUNNINGHAM: I guess yes then.
DR. BORER: Mike?
DR. ARTMAN: Well, yes, I think the results with captopril are germane, but I take exception with the pharmacological issues. I do not think we can equate irbesartan with an ACE inhibitor. I think there are differences in the pharmacology. There are certainly differences in the stimulation of AT I versus AT II receptors. Whether or not sort of this unopposed action of AT II receptors is good, bad, or ugly, I don't think we know. So, I don't think we can generalize the pharmacology of ACE inhibitors to that of the AT I receptor blockers.
DR. BORER: Tom?
DR. FLEMING: I defer to my clinical colleagues in interpreting the biological parallels. The data are confusing when one looks at them head to head, but I think we'll get into that in future questions.
DR. BORER: Blase?
DR. CARABELLO: Certainly the two drugs have some similarities and also some substantial differences, but I think the similarities probably outweigh the differences, so I would vote yes.
DR. BORER: Steve?
DR. NISSEN: I'm actually a little surprised by this discussion. It's tough enough to look at effect of a drug when you have other drugs in the class and say, well, an effect is a class effect. Now, we're talking about two different classes of drugs, and so I'd want to have pretty good evidence that the effects are very, very similar before I'd extend that across drug classes, let alone within a class. And we've already seen in many examples where drugs in the same class don't have the same biological effect. So, I think it's a potentially dangerous precedent to say that two drugs that happen work through kind of similar mechanisms would have the same effect from two different classes, and I think we ought to be very careful here. So, my vote is no.
DR. BORER: Alan?
DR. HIRSCH: You are a strict constructionist. The words are relevant and germane. So, I think they're not identical, but they're certainly kissing cousins and relevant. I would say yes.
DR. BORER: I'm going to vote yes too. I've been convinced by the discussion that the nephropathy in type 1 and type 2 diabetes is sufficiently similar so that one should be able to draw inferences from one and apply them to the other.
And with regard to the pharmacological effects, I agree with everything Mike and Blase say. Steve, there are a number of differences here between ACE inhibitors and angiotensin receptor blockers. And I'm on record as saying I don't know how drugs cause their clinical benefits.
Nonetheless, I think the fact is that both of these types of agents and both of these agents act on the same general system, and I think that, as Alan says, they're germane and relevant, though not identical. And I vote yes.
Number 11. If the results with captopril are relevant to irbesartan, are the results on protein excretion similar with respect to direction and magnitude? 11.2, are the results on doubling of creatinine similar with respect to direction and magnitude? Are the results on death or ESRD similar with respect to direction and magnitude? And if you say no to any of those or if you say yes, probably we ought to have an explanation of why.
DR. LINDENFELD: I guess the key word here is similar, and I would say yes, they're similar. The effects are greater in the captopril trial, at least they were certainly greater on the doubling of creatinine. I think it was a 48 percent reduction as opposed to 33 percent, and greater in proteinuria and end-stage renal disease, somewhat greater. But the direction is very similar in all of these.
DR. BORER: Mike?
DR. ARTMAN: Yes. I think the directions are similar, but the magnitudes seem to be much greater with captopril than with irbesartan.
DR. BORER: Do you draw any inferences from that observation that you'd like to share with us?
DR. ARTMAN: No.
DR. BORER: Tom?
DR. FLEMING: Well, we're comparing results from different studies. That's always hazardous. Yet, I'm not persuaded that they're similar enough that I would say similar. If I chose, as best I could, a comparable endpoint, which would be dialysis, transplantation, and death, we're looking at an 11 or 12 or 13 percent reduction against a 50 percent reduction. That's getting to be an important difference. And the mortality, small numbers, in the captopril setting, but there was a 40-odd percent reduction in mortality and there was more than a 50 percent or about a 50 percent reduction in dialysis, whereas here there's no effect discernible in mortality; dialysis reduction is 20 percent. The setting is different to an extent, but then again, to the extent that the setting is different, it makes me less comfortable to extrapolate results from the other trial.
So, I'm not as knowledgeable as my colleagues about whether the biological phenomenon and pathways and mechanisms of action are truly sufficiently parallel that we can really rely on a different trial and a different agent, but at least looking statistically at the evidence, I see a substantive difference in the magnitude of effects that are being estimated.
DR. BORER: Blase?
DR. CARABELLO: Yes, well, certainly the ACE inhibitors appear more effective, but there's been no head-to-head comparison. It's sort of like saying, well, one team beat another team by 50 points and the other one beat the other team by 20 points, and therefore the difference ought to be 70 points. And that's just not the way it works. So, I don't think I can draw very much from those differences.
DR. BORER: Does that mean that you think that they're relevant or not relevant?
DR. CARABELLO: I think that they are relevant, but I can't draw any differences between them.
DR. NISSEN: Again, I think it's a slippery slope here. You're talking about a disease. One is a disease of insulin deficiency. Another is a disease of insulin resistance. And how that plays out in the vascular system leading to the kinds of events that lead to mortality and morbidity in these patients is probably somewhat different. I think again we've got to be very careful about setting that kind of precedent. I would not want to go on record as saying, well, something that works in type 1 diabetics should be inferred to work in type 2 diabetics because I do think the pathophysiology of the disease, not necessarily the kidney, but the disease overall is very different. I think, again, we ought to be very careful about the kind of precedents we set in these discussions because I think it sends potentially the wrong message.
DR. BORER: Alan?
DR. HIRSCH: Well, let me reemphasize sort of what Steve just said. Whereas I've been stating that I certainly believe they're relevant, we around this table can't ignore the similarities in directional trends. Now I'll go the other direction and say although we've as a group said that the magnitude of benefit in the captopril trial might at that time of history been due to the care given at that time or because less cardioprotective drugs were used or glycemic control was less intense than nowadays, all those things may be true, but I hesitate to make too much of a comparison because it's also possible that the diseases are not identical, that we really do have different molecular entities, we have different potential pharmacodynamic effects. Bradykinin does exist. There are known differences between what ACE inhibitors do and A2 antagonists do in tissue and to mRNA expressions.
And finally, there's the dose question. It's really hard to know at the end of the day how this dose of captopril in this population compares to this dose of irbesartan in this population. It's very hard to bring these together other than to say, yes, they're similar.
Yes, that's a no.
DR. BORER: I'm going to vote yes. I think they are relevant. I think the results are directionally generally similar, and the magnitudes obviously are not. But these are different trials in different patients at different times with different protocols, et cetera, and it's very hard for me to get too excited about that. I think that these results have an influence on the way I think about the results of the irbesartan trials, and I'm not going to quantify that.
With regard to the fact that they're different diseases, the patients had different diseases, type 1 and type 2 diabetes, they did. But, of course, we've been shown data suggesting that the nephropathy in type 1 and type 2 diabetes seems to be pretty similar, and we also have in our books data from the enalapril study in patients with type 2 diabetic nephropathy, though not hypertensive, so I'm going to be drawing a parallel from a different group. But patients with type 2 diabetic nephropathy improved in at least one measure of their renal performance when they were on enalapril which is also an ACE inhibitor.
So, when I put all those facts together, those observations together, I have to say that I am influenced by the captopril data. The question is how much and how much do I have to be, but my answer is yes to 11.
Number 12. Did I miss somebody? I'm so sorry. Go ahead.
DR. BREM: I guess this half of the table doesn't count. Once the cardiologists have spoken, I guess that's the word.
DR. BREM: Obviously, I'll restate what you said. We're not comparing the true efficacy of the two agents with one another. We're just asked a straight question, are they in the same direction and are they consistent with one another? I think the answer is yes, they are consistent with one another. And I would say, for that reason, they're germane and relevant.
DR. KOPP: Yes. Without belaboring it, I would say yes. I think they're relevant and we'll come in a minute to decide are they a quarter of a study, a half of a study, one study.
DR. LORELL: I also believe they're relevant, and I'd like to comment on the two reasons why I think they are.
I think the data presented today and, in fact, the slides that we were shown this morning which described the effect of placebo in the captopril trial and doubling of serum creatinine and the similar slide that was presented for placebo in the irbesartan data are extremely striking in that the event rate is almost identical at 48 months. So, it suggests that although, as you pointed out, one is type 1, the other is type 2 diabetics, that what the kidney is doing and seeing may be remarkably similar.
I think the data are also relevant for the point that Steve Nissen brought up earlier and that is although this may be somewhat disturbing and not ideal, I think the reality in the United States in clinical practice across the country is that patients who already have type 2 diabetic nephropathy are in large part being treated with off-label use with an ACE inhibitor.
So, with those two arguments for relevance, I think it is worrisome that the magnitude of benefit seemed to be so much stronger and more robust in the captopril study, albeit it was type 1 diabetics and non-hypertensives. That influences me perhaps, rightly or wrongly, in wishing to see a more robust data set for irbesartan or any other AT I receptor blocker since I think the impact of approval would be to profoundly change a current, very widespread practice of use of ACE inhibitors.
DR. BORER: Was that a yes or a no vote?
DR. LORELL: It's a yes for relevance. It's a no that I don't think the results are similar in magnitude.
DR. BORER: Dr. Cunningham.
DR. CUNNINGHAM: I would agree. I think they are the same in direction, but the magnitude is very troubling.
Actually since it's my first time on the committee, I'm going to go back and say I do not really think that they're pharmacology the same, that the angiotensin receptor inhibitors are the same as the blockers. I think we don't know that. That actually was two questions. So, I might say yes to one and no to the other for 10.
DR. BORER: Number 12. Now, the key question here. Are the results of IDNT, IRMA 2, and prior expectations derived from the captopril database an adequate basis for approval of irbesartan for the treatment of either hypertensive or not hypertensive patients with type 2 diabetic nephropathy?
DR. LINDENFELD: I believe they are. I would vote yes for this. It's close, though. But I'll tell you what. The IDNT trial is not perfect and it's not terribly robust, but the IRMA trial supports it. I'm helped a little bit by the amlodipine data which at least lowered blood pressure, so we know this wasn't only a blood pressure effect in the IDNT trial.
And I believe that, while I agree with everything that's been said, that the two drugs, captopril and irbesartan, do not have entirely the same mechanism of action, in fact, could be very different, one of the pertinent mechanisms of action here is through angiotensin, and so they do share an important mechanism of action.
So, I am concerned by what Bev said that by approving this drug, we could change the standard of care, and there's a big concern here about the magnitude of benefit. But I'm not sure that can be our concern. If the drug meets the standard of approving, I don't think I can let that change my vote of yes for this.
DR. LIPICKY: Can you clarify a little bit? So, what you're saying is that your priors from captopril are enough to say that when you said no-no to the previous questions, that now you mean yes-yes. Did I say that right?
DR. LINDENFELD: No, you didn't. I said I still would say no-no for the first two questions, but what you've asked here is whether or not the data from captopril, because of at least some shared mechanisms, would be enough to tip me over and say the totality of the data suggests that this should be approved. Then I would say yes.
DR. LIPICKY: But that's because you're convinced from the captopril trial that, in fact, there is class effect on the disease because this is a different class --
DR. LINDENFELD: Right.
DR. LIPICKY: -- so it's not even the same class. And it's a different disease.
So, I'm just trying to make sure I understand what you're saying. So, what you're saying is that although it's a different class, you're willing to buy an ACE inhibitor class effect on the captopril trial. There the delta in clinical events was 18 people I believe. Here it's 0, but in captopril it was 18. So, on that basis, you're willing to buy this also. Is that really what you're saying?
DR. BORER: Ray, always does this.
DR. LINDENFELD: I think what I'm saying is I was very, very close. I think there's a lot of really good data here in two good studies, and we've seen a pathophysiologic sequence for which there's a lot of data which I believe, and the fact that we have an awful lot of data with ACE inhibitors that share a common mechanism tips me over to say that that's just enough more to say that this data now becomes in my view enough to say yes.
DR. FLEMING: Could I just ask for further clarification of this, following up on Ray? Can you give us some insights, just in a precedent-setting manner, of how we have done this in the past? I find this intriguing. We're looking at two pivotal studies and coming to a conclusion, and then we're searching for other relevant data which is certainly relevant to do so, moving outside of the class, though. Essentially is this then saying any agent within these two classes? How much are we extrapolating? Any agent within these two classes then largely would rely on the studies that had been done here, together with some surrogate endpoint data to then be an approval? I'd just like to have a sense of how this is playing out.
DR. TEMPLE: Well, we don't keep good track. First, let's stay within a class. The division pulled together the basis for approval of the various ACE inhibitors in congestive heart failure, and quite consistently we've approved those claims with p values between .05 and .01. Pretty consistently, usually one study. Now, that's because those are all the same pharmacology. So, that's one precedent.
Another might be said to be the recent approvable for Valheft, for valsartan. The committee divided closely on it. We reached a somewhat different conclusion. I don't even want to blame anybody else for it. I reached the somewhat different conclusion based on a subset analysis, but clearly influenced, I would say, by the similar pharmacology and a particularly persuasive subset. So, I don't want to over-attribute it.
But I think the answer is you are allowed to let these things -- think of them as priors or think of them as mechanistic explanations -- influence you. The reason we bring hard questions like this to advisory committees is that it's very hard to pin down exactly what you're doing when you do it. They surely come into the category of what confirmatory evidence might be under the words that the law uses, although we've certainly never pinned down what that means exactly. What I heard JoAnn say was she was sort of here and she got pushed over by the amlodipine comparison and these data, and I think that's how people actually think. They put it all together. Obviously people can disagree on what the right conclusion is.
DR. LIPICKY: If I might contribute to that a little bit because this is really a very difficult issue. For a precedent, we have approved for congestive heart failure captopril on the basis of a single trial for exercise tolerance, a p of .0048 or something like that. So, precedent -- that is, what have you approved things for in the past -- may or may not be useful. I don't think we would do that ever again at this point for that disease because, indeed, there have been things learned.
But indeed, nephrologists, as you have heard today in very elegant presentations, would pull all of this stuff together, including captopril, and have it influence their thinking process. Well, are we to say nephrologists are crazy and they don't think right? I'd be happy to say that --
DR. BORER: Remember that you were called a nephrologist earlier.
DR. LIPICKY: So, this is all a matter of judgment and I think it is not necessary to ask the question what are the precedents because I think the precedents only say what have you done and you may have done wrong things. So, there's the logic of it.
DR. FLEMING: One does struggle, though, to see if there is a logical consistency. Severe sepsis, a major FDA recent issue in December where this issue went in the other direction. A study that looked pretty good, but everything else had been negative, and FDA went ahead and approved, more or less, saying it's this study even if everything else had been negative.
Now we're hearing -- well, we don't know what we're hearing yet, but I guess what we're being asked to discuss is if there is a study that's out there that's positive that shows a considerably different effect, actually more positive, which actually could pull us in the right direction, but it looks very different and it's a different class, that we should be persuaded by that. One would like to be scientifically consistent when one thinks through the strength of evidence you would have to see to approve an agent.
DR. TEMPLE: Sometimes a very strong result in a single study, even in the face of past failures, is convincing. And in the sepsis case you describe, I think that was the basis for it there and the others were not persuasively negative. They were persuasively --
DR. FLEMING: The committee was 10 to 10 in the vote on that one study.
DR. TEMPLE: Well, I mean, obviously they're going to be close. The p was .005. That's either strong or weak, depending on your attitude.
DR. LINDENFELD: One other thing here is we haven't seen any data, I don't think, that suggests to us that this doesn't work. It may not be strong in any single study or any single area, but we haven't seen anything that suggests that it's unlikely to work. And that influences me a bit.
DR. BORER: Steve, a final comment and we'll go on.
DR. NISSEN: Yes. I guess, JoAnn, the problem with that is that if we make a decision to approve, it has consequences, and I think I know what those consequences are and I think we better face that. And that is, that some of the patients currently treated on ACE inhibitors are going to be switched over to irbesartan. Do we think that's a good thing or a bad thing? Do we think there's enough evidence here to tell physicians that we're now going to approve this agent, the first agent to be approved for this purpose, and the standard of care, whether it's right or wrong -- I know it's off-label. But this drug is going to get detailed and people are going to be told, listen, don't give your patient lisinopril. Give your patient irbesartan because we have FDA approval for this indication. Do we really want to do that? And if we do, let's vote for it, but I don't think I want to do that.
DR. LINDENFELD: That's a really important issue, but that issue wouldn't change if we had larger numbers with the same reduction in creatinine doubling and the p value were stronger.
DR. NISSEN: Right.
DR. LINDENFELD: That wouldn't change that. The fact that you think that an ACE inhibitor is better than this drug -- if we just had even stronger data for this drug but it still appeared that it was less effective than captopril, you'd still be in the same bind.
DR. NISSEN: No, but in the presence of weak evidence, then do we really want to change the standard of practice, which is one of the things -- that's one of the effects of what we do, fortunately or unfortunately.
DR. BORER: In fact, the FDA doesn't define standard of practice. Guidelines committees do. And my guess is that the impact will not be quite so great as that on patients who are being treated one way or another way because there are biases in the minds of every nephrologist I would guess. I think all we're being asked to say here is do we believe this stuff works or do we not, as JoAnn says.
Having said that unless, Dr. Temple, you had another comment --
DR. TEMPLE: Well, if there were mountains of evidence that all the other ACE inhibitors did what you want, you could wonder about that. But in fact, what you've got is captopril and everybody uses something else, I'll bet, because they want a once-a-day drug.
DR. BORER: Dr. Brem? We've had one vote yes for approval based on these three separate sources.
DR. BREM: Well, I've heard these three separate sources, but I've also heard references to outside sources, including the enalapril study. So, if one is going to be consistent and use outside sources like the enalapril study, then I suppose we can say we can use the outside source that was the losartan study which is the same class of agent as this, showing virtually the same findings as what was presented all through today. So, I would say if you're going to be fair, you're fair for everybody on both sides. And it would be supportive evidence, albeit it we haven't gone through the same detail as what was discussed today, but it's certainly consistent both in magnitude and direction. And it is further supportive data for approval in my opinion.
DR. BORER: So, is that a yes?
DR. BREM: Yes. It would be a vote for approval.
DR. KOPP: Well, not surprisingly, I say yes again, and I'll stop there.
DR. LORELL: I'm going to address question 12 very narrowly, exactly as stated, and I view that the prior expectations from the captopril database in fact are not an adequate basis for approval. And I'll restate briefly what I said a few minutes earlier, that I think using the terminology that Dr. Pfeffer raised earlier that observations are hypothesis-generating, I think the remarkable difference in the magnitude of effect, as well as the time of appearance of effect, in the captopril studies, the benefit, the curves diverge much earlier. I think it's hypothesis-generating that in fact the two drugs may not be identical and may have quite different magnitude of effects. We don't know that because that study has not been done. So, strictly answering question number 12, the prior expectations derived from the captopril data for me do not push over toward an adequate basis for approval.
DR. CUNNINGHAM: My answer would be no too. I think from the consumer perspective, I really worry that if the standard of practice currently is using angiotensin inhibitors, approval of this could actually move people to use a drug which might be less effective for which we don't have enough data. Unfortunately again, the issue is we don't have the data we need really to help the people who have this. I think having renal failure is a dreadful problem and dialysis is obviously a terrible thing to have to endure. Just we don't have the information we need. But no.
DR. ARTMAN: I'm kind of surprised at this whole discussion, and I wouldn't say no. I'd say, hell, no.
DR. ARTMAN: I think that we're talking now about a different study population with the captopril, we're talking about a different class of drug. It's hard for me to weigh that in, in any sense, to strengthen my decision about irbesartan.
So, if we follow this to its logical conclusion, I guess we could begin to argue that maybe we should be recommending approval of captopril for type 2 diabetic nephropathy. I don't know. It just seems over the top to me, so I would say no.
DR. BORER: Tom?
DR. FLEMING: No. And I share your sense exactly. I do believe that it's relevant, when you're making a judgment about the effect of an intervention, to be aware of and take into account what is available on efficacy of interventions that are studied in related settings. Though, to be giving that substantial weight here, I would have wanted to have had a much more careful discussion about the captopril data, what the studies were, what any other studies were that would be relevant to this decision.
Appropriately we gave a great amount of attention to these two studies, and I believe with that tremendous information we've been provided and in an intensive day of discussion, we have absorbed an awful lot of understanding. And even at that, there are a lot of complexities that are still difficult to fully understand.
To now be reaching out and asking this committee -- boy, in 15 years of being on innumerable advisory committees I can never remember being asked to essentially say if it's no, but now look at external data from other agents studied in other trials with all the complexities of understanding differences that you see across studies, across specific disease areas and classes of agents, that you would actually, without having any direct presentation and discussion of those other data, be asked to revise or reassess your assessment. It's very troubling to me. But I do appreciate the need for FDA to think about this, but it's troubling to me the process, that we're being asked to think about it after having focused almost exclusively on these two trials.
Based on that, I'd say no.
DR. BORER: Blase?
DR. CARABELLO: We are going to get to question 13, I assume.
DR. BORER: Yes, we will.
DR. CARABELLO: Okay. Having said that, then I will vote no on 12. I don't think that ACE inhibitors -- although I think their results are germane, I think there are still enough differences between the two classes that they don't persuade me enough on top of my previous arguments about why I thought the data wasn't strong enough and compelling. So, I'll vote no.
DR. BORER: Steve?
DR. NISSEN: I also will say no. I want to bring up one more time the fact that it's not like all the endpoints are all going in the same direction here. I've got to remind this committee that many patients that come into this process of nephropathy have cardiovascular disease and they tend to die of cardiovascular disease. When I see point estimates for cardiovascular death, myocardial infarction, and stroke, in comparison to amlodipine, go substantially in the wrong direction, I'm troubled. So, that takes away.
We talked about this external study kind of adding to our confidence. Well, there are things that take away from my confidence, and that substantially undermines my confidence in the benefit of irbesartan here.
So, I want more data before I'm willing to stick my neck that far out and say that this is good for people when I know the cardiovascular endpoints are such a prominent problem in this patient population.
DR. BORER: Alan?
DR. HIRSCH: I'll start off with my no, but I'll try to add something new as we as a panel face these decisions in the future.
I think we're all feeling uncomfortable because there are three things we're weighing. We're weighing this intrinsic data set for this particular ARB. You've heard our opinions about that. We're obviously weighing this need to consider precedent-setting if we want or don't want to do that. And then the third thing is how we think it will affect the market.
So, I'll just add, though I care a lot about how standards are set for clinical practice, I very well trust the renal community to make its guidelines. I think guideline committees are where the market and the practice standards will be set. That's not our role, though I care a lot.
Vis-a-vis precedent, I actually do care very much. I think Tom was getting to this, that we think as a group how we look at data and how we set precedent. Ray, we can ignore some bad past ones and maybe make better current ones.
So, with those two things in mind, looking at the first thing we usually do, which is this data set, I'll stay with no.
DR. BORER: I'm going to vote yes. I want to point out to everybody that there's nothing in the label of this drug as it now exists, because it's an approved drug, that would preclude anyone from using it in a patient with hypertension who happens to have diabetic nephropathy. It's a drug for people with high blood pressure.
The issue of whether it actually is beneficial for the kidney disease, over and above that, is what we're talking about here. I'm convinced that it probably is, and I'm sufficiently convinced both by the two trials that we saw, taken in tandem, plus what inferences I'm going to draw from drugs of a different class, it's true. So that I think that in total these data are sufficient to allow me to believe that it's reasonable to treat patients for prevention of progression of their diabetic nephropathy, as well as for their hypertension. So, I'm going to vote yes.
I think that if, at the end of the day, because we have a couple more votes here, we still have a net no, as we do right at this moment, perhaps it would be useful to cite what other kinds of information we might want to see so the FDA could think about that. Maybe some of those data can be drawn from the existing database.
Having said that, let's go on to number 13. It doesn't require a stated vote for everyone. Are there results from other development programs that impact on approval of irbesartan for the treatment of type 2 diabetic nephropathy?
DR. LINDENFELD: Well, the RENAAL study certainly, although we've said that it's difficult for us to talk about that too much because we have not actually seen the data and what's published sometimes, it's been brought up, is not always the data, once we see an FDA analysis.
But again, there are other enalapril trials, other ACE inhibitor trials. While, again, these are different mechanisms, I think that there's a weight of data from a number of other things that there are shared mechanisms of benefit here in other trials. So, this would push me a little bit more, but I'm probably not the one to say too much here because I said yes already.
DR. BORER: Does anyone else want to add to that? Blase?
DR. CARABELLO: Yes. I think the RENAAL trial is compelling. I remember reading it in the New England Journal and saying, gee, isn't this interesting. A sartan works in type 2 diabetes, and thinking, gee, if I saw a second study that said that sartans worked, that I would probably be convinced. I realize the depth of plumbing of the data is different, but I do think it has an impact here and it has an impact on me.
DR. BORER: Steve?
DR. NISSEN: I really have to take exception to that. Again, I'm worried about precedent, and I'm worried about the slippery slope. We don't have the RENAAL data in front of us. One of the things I've learned in the last year or so on this committee is the data isn't always what it seems to be, and until you get a real look up close and personal at the data, you ought to be very careful. I don't know if that trial will ever be presented to this committee, but when it is, we ought to look at it with the same scrutiny and the same microscope we looked at the IDNT and the IRMA 2 data. In the absence of that kind of scrutiny, we ought to be very, very careful about the precedent of making decisions regarding data that is not on the table.
DR. FLEMING: One more comment? One more question?
DR. BORER: Tom?
DR. FLEMING: A question for Bob and Ray. You have said you've seen this data. We haven't had the data presented to us. Do you really want to go around the table and get our vote? You're certainly at liberty, since you've seen the data, to factor it in however you choose. Do you really want to go around the table and get our views on data that you didn't share with us?
DR. LIPICKY: Well, we have not seen the data. It exists but you haven't seen a review.
DR. TEMPLE: No.
DR. LIPICKY: You haven't seen a table. You haven't seen a figure. I haven't see a table. I haven't seen a figure. I haven't seen any pros. I haven't talked to anybody who has actually opened the volume. So, we don't know any more about RENAAL than is in the New England Journal.
DR. TEMPLE: Right. You could tell us what you would think, for example, if after review, that was thought to be a solid study. We can't tell you that we know that. But you could say that. Obviously that might not affect your conclusion because you wouldn't know.
But it raises the whole question of what, which is absolutely not settled, except I did tell you that we have behaved as if ACE inhibitors share common properties when we deal with congestive heart failure. And I would go further and say the beta-blocker approvals probably all reflect the fact that other drugs of the same class have had that effect.
So, you can tell us how you're thinking of it. I don't know that fits into a vote exactly.
DR. FLEMING: Well, in fact, you have asked us this structured set of questions. We have gone through question 12, which basically is based on the two trials and prior expectations from the captopril study, what was our judgment. Now in question 13, you're saying is there other relevant information. I don't have a problem with your asking the committee. I understand why you might want to ask the committee are there other sources of data that we would advise you should look at when you think about your final judgment.
But I don't understand why you're going to press to question 14 and then say, now I want you folks to vote on that. That's an entirely different perspective. Why are you asking us to do any more now than suggest additional sources of information that you should think about?
DR. TEMPLE: First of all, some people have, without thinking about those other studies, reached the conclusion that the answer is one way or another. So, you don't necessarily have to incorporate that. But you might want to tell us I'm at the edge. I'd be convinced if you looked at that study and found it satisfactory. You can tell us whatever you want. It isn't only the vote. We listen to the text also. It would be of considerable interest to me to know if you thought that a favorable finding in a closely related drug should be taken as the priors or the other evidence or whatever. And we can worry about getting access to other data, and then we can also worry about whether we're violating a company's proprietary interest and a wide variety of other interesting questions that we're not ready to talk about here.
DR. LIPICKY: There are people within the agency who tell us not to ask questions where the answer is obvious, and to me the obvious answer to this question would be, of course, not. You're insulting my intelligence by asking me. Sometimes it's useful to ask questions like that so that we can get your reaction. But here we're being told now, because we asked questions we probably shouldn't have asked, that you know, well, the captopril study tips me over. Well, then it's fair to say, what else would tip you over?
DR. BORER: As the hour is moving on here, question 14 -- 15 is going to require some discussion I think. But 14 says, should irbesartan be approved for treatment of nephropathy in type 2 diabetes? Prior to 13, there were 4 yes votes already. So, just to deal with the time issue, I think I understood Blase to say that at 14 he would change his previous vote, but for the record, what's your vote now, Blase?
DR. CARABELLO: That is correct. Yes.
DR. BORER: So, now we have 5.
There are six other committee members who have not been convinced by these data, and so I'm going to ask to go through each of those six. Should irbesartan be approved for the treatment of nephropathy in hypertensive or non-hypertensive patients with type 2 diabetes? Beverly?
DR. LORELL: For the issue of diabetic nephropathy in hypertensive patients, my answer would still be no.
DR. CUNNINGHAM: No.
DR. BORER: Mike?
DR. ARTMAN: No.
DR. BORER: Tom?
DR. FLEMING: I encourage the FDA to look at all information that they would have access to that they judge to be relevant.
I can't go beyond question 12 without specific data put before us with a judgment and with a good insight that we have a level of understanding of that data comparable to our level of understanding of the two trials here. My answer is no.
DR. BORER: Steve?
DR. NISSEN: My answer is also no. And I really do need to say something here, and that is the reason that I don't want us to consider the RENAAL data is there were things that came out in the discussion of the IDNT data that I didn't know from reading the publication. For example, the cardiovascular endpoints are not in the New England Journal of Medicine publication. Those individual endpoints were very persuasive to me. I don't know as a fact that there aren't things in the RENAAL data that we haven't seen yet that would be very negative in my view. So, my answer is still no. I can't consider a study that isn't before us.
DR. BORER: Alan?
DR. HIRSCH: Well, no surprise based on my prior comments. No. But I would, since the text is read, strongly encourage the FDA to gain whatever access to data sets that exist that would permit us to look at this issue again, if that's possible.
DR. TEMPLE: I just want to be sure I understand the last couple of comments, and perhaps other people have the same sense. We have no idea whether we can get the two companies on the same stage or sort of steal one's data to use in the other. Let's leave all that aside.
What I guess I hear you saying is that some of you are not so convinced by information about captopril in a different population, but you might find independent studies of the same drug class in a very similar population more persuasive.
DR. HIRSCH: Yes, and I'll just try to summarize my feelings and maybe speak for others here. This is a very major national clinical problem. We're all seeing a signal of efficacy. We are all very close to the edge, but we're not quite there. So, whatever we can get that brings us closer to a decision would be great.
DR. BORER: Before we go on to number 15, which is the last question, can I ask if anyone has any comments about data that could be perhaps presented to the FDA and if they were presented to you would be useful over and above issues about other drugs and what have you. What, for example, would you think if the company was able now retrospectively to go back and determine cardiovascular events in the patients after cardiovascular events were no longer being counted? And what if the differences sort of evaporated? Would that be compelling to anyone? Would you like to see that?
DR. LORELL: Yes.
DR. BORER: Okay. So, that's one type of information that might be useful if the FDA is not convinced by the data that exists.
There may be others. For example, although Tom suggested that he accepts the creatinine of 6.0, as I do, perhaps if we asked the company to come back with the actual numbers of people who went on to dialysis or transplant and the time at which they did that to confirm the importance of the creatinine level, would that be useful to anyone?
DR. LORELL: Yes.
DR. BORER: Okay, so that would be too. So, there may be some additional data that exists that might be helpful to the FDA in its final deliberations.
Having said that, do the results of the irbesartan development program in type 2 diabetic nephropathy support the use of proteinuria as a surrogate for clinical benefit? And we need everybody to say something about that. JoAnn?
DR. LINDENFELD: I don't think the data quite support the use of proteinuria at this point for a couple of reasons. One, I think that we don't have any idea of how much reduction in proteinuria is clinically important. I mean, if you saw a 10 percent reduction, would that really lead to some sort of realistic clinical change? I don't know the answer to that. So, I'm not ready to say that yet.
DR. BORER: Let's start on this side of the table with the nephrology contingent. Dr. Brem, what do you think of that? Do the results of the irbesartan development program in type 2 diabetic nephropathy support the use of proteinuria as a surrogate for clinical benefit?
DR. BREM: I was just talking to Dr. Temple about the very same issue. For a nephrologist, I believe that it does. And I read the FDA commentary, and I suppose by itself, in the absence of changes in GFR to correlate, it wouldn't. But I believe that it does. So, as a nephrologist I think it does, and if they looked at it correctly, it would correlate with a change in GFR. One thing that would be very helpful and supportive is if the company had similar data or would be able to retrieve that information to demonstrate the connection between development of proteinuria and change in GFR; that would be a very strong selling point.
DR. BORER: I'm sorry, Tom. Go ahead.
DR. FLEMING: I just wanted to clarify to make sure that at least I'm understanding the question. There are at least two different questions you could ask. One is, is proteinuria a correlate for subsequent clinical endpoints? Or there is the much more complicated question, is a treatment-induced effect on proteinuria an adequately reliable predictor of a treatment-induced effect on long-term clinical endpoints, that simply the demonstration of the effect on proteinuria is adequate evidence of an effect on the clinical endpoint? Which is what I would typically think of as the surrogate. Are you asking the first question or the second or both?
DR. LIPICKY: We're asking the second question only. Is it a surrogate? That is, proteinuria would be good enough. You see that proteinuria goes away and you say that people aren't going to get transplanted or have dialysis.
DR. TEMPLE: We believe a lot of nephrologists, as you've just heard, believe that proteinuria is an indicator of the fundamental structure of the kidney going down the tubes, and we have a lot of people who are asking us to use that as an endpoint because it's a very accessible endpoint and it occurs early and you don't do harm to people. So, that's a very important question.
DR. FLEMING: And that's the first issue. And there's a myriad of natural history data, and Dr. Lewis and others have clearly discussed the relationship that is establishing this as an important biological step along the cascade of events, such that in natural history, it's a correlate.
But we have a myriad of correlates across all disease areas, the vast majority of which are not established to be surrogates, and those that are typically are done so after extensive clinical experience where one is looking at the relationships of treatment-induced effects on the marker and the predictiveness of those effects on the clinical endpoint.
We didn't even get from this study evidence of whether there was a relationship with effects on proteinuria to effects on filtration rate, much less the ultimate clinical endpoint. So, how does this study begin to provide us evidence as to whether we don't just have a correlate but we truly have a replacement endpoint surrogate?
DR. TEMPLE: Well, don't yell. Just say no.
DR. BORER: We have his vote now.
DR. KOPP: You specifically asked about diabetes, but I'll make the first point that if you were looking at other diseases, a complete remission I think would be acceptable, that is, somebody with a disease like minimal change that you had an agent that could put people into complete remission, I think most nephrologists would be convinced that a sustained remission, as defined by no proteinuria, would be evidence of benefit.
How about in diabetes? I guess if you had an agent that put somebody into complete remission, no proteinuria, who previously had proteinuria and that was sustained, I would find that pretty convincing.
The problem comes from what you mentioned, JoAnn, which is how about a 50 percent decrement, which is the typical effect of ACE inhibitors. Then how about a 30 percent? Then how about a 25 percent? And as we said earlier, we don't have a graded spectrum of risk table that allows us to evaluate that. So, I think that's where you're coming from.
And I have to say by itself, I don't find it adequate. Now, I did find it adequate in the second study, IRMA 2, because I had another study that had a GFR measure, and in that setting, it's supportive. But by itself as the only measure, short of a complete remission, I have trouble accepting it.
DR. LORELL: I agree with that. I think the evidence is very compelling that proteinuria is both a measure and may even play a role in the progression of diabetic nephropathy. I don't think that that is sufficient to use it as an endpoint for approval or disapproval of a drug.
I think a very poignant example in the recent cardiovascular and even perhaps cardiovascular-renal world is the use of estrogen replacement therapy where it was argued for a long time that the improvement in lipid profiles that were seen demographically and in trials could be used as a surrogate marker to indicate expected improvement in major cardiovascular events. And actual trials that asked that question have given a striking answer in the other direction.
DR. BORER: Dr. Cunningham.
DR. CUNNINGHAM: I think I don't actually know, so I cannot say yes, but I don't know if I can no either. So, you'll have to deal with that.
DR. BORER: It's an abstain.
DR. ARTMAN: I would say no and leave it at that.
DR. BORER: Tom, I think you already said no.
DR. CARABELLO: I think I also would like to be fair and simply say I don't know.
DR. NISSEN: No question that proteinuria is a bad thing, but the question being asked is do we know that a drug that reduces proteinuria is going to have a consistent, reproducible effect on hard endpoints. I would want to see several trials, maybe even with different classes of drugs, showing that we know what the relationship is between the reduction in proteinuria and the reduction in end-stage renal disease, dialysis, transplant, death, and a lot of other hard endpoints. And we haven't come even to the first step. Validating a new surrogate is like hitting an inside-the-park home run, and we haven't even gotten to first base yet. So, I think we've got a long way to go here.
DR. BORER: Particularly in Cleveland where the stadium is now smaller than it used to be.
DR. NISSEN: But the team is good.
DR. BORER: Alan?
DR. HIRSCH: I probably don't know either. Not to disrespect my renal colleagues, I also would like my protein in my urine to go away with treatment anecdotally, but when it comes to data and approval, my problem with surrogates is that we've been burned too often as clinicians, whether it's left ventricular hypertrophy as a surrogate marker -- I can go on and on -- VPBs with dysrhythmic drugs. So, if I was going to use proteinuria as a marker, I need some specific clinical outcomes and some kind of a dose response where I know, like with blood pressure, these many millimeters down affects these many cardiovascular risks, this LDL change affects this clinical outcome. I need some better handle on the dose response, I think as you said Dr. Kopp.
DR. BORER: I'm going to vote no also. With Dr. Kopp, I thought that the proteinuria data were very compelling as supportive evidence with another trial with harder endpoints, but by itself, the issues that everybody has mentioned, the relation of the magnitude of the effect to the outcome of interest to us just is too poorly known to allow us to accept as a surrogate at this point.
DR. TEMPLE: Well, I was very struck by the observation that at least one dose of irbesartan was able to produce a dramatic decrease in proteinuria that obviously had nothing to do with anything anatomical at all. So, maybe the other dose did something anatomical I don't know. But that was impressive, I thought, that it could go away so quickly, and it certainly seems reason for pause.
DR. LIPICKY: Jeff, I think you said that IRMA supported IDNT, and now you say that proteinuria is not a surrogate. Can you explain how you came to those what I think are disparate decisions?
DR. BORER: Yes, I can. I don't think they're disparate at all.
IRMA I believe gives me some information about the pathophysiology of disease and convinces me that the drug has an impact that's positive on the pathophysiology of the disease. That was supportive of hard body count data from another trial. Absent the hard body count data from the other trial, I would not have been in favor of --
DR. LIPICKY: But the body count was 22 and 24.
DR. BORER: Well, in my judgment the data in their totality, for all the reasons that I stated that are going to be on the record when you read it, I thought were sufficient.
DR. LIPICKY: I withdraw my question.
DR. FLEMING: If I could, I'd like to, even though I voted against it being a surrogate, defend that there's logic to what Jeff has said. When I argue it's not established to be a surrogate, my belief is we have not come close to establishing the scientific evidence to be confident that when you simply show an effect on proteinuria, that we know that it's of the duration, magnitude, and all the other things we have to understand to know that it's reliably going to predict an effect on clinical endpoints.
Nevertheless, there's still a level of plausibility of efficacy that you gain from such an insight. Even if you have a marker that's not an established surrogate, you understand natural history, and if you can impact that marker, it provides some level of sense of plausibility. So, if in fact you believed that the first trial was close enough or was quite close, then something that's not a surrogate, but establishes a level of plausibility, could tip you over. I don't know if that's where you were, but it's logically consistent you could have that opinion.
DR. BORER: You said it better than I could.
I think we've completed the questions, but Bob has another.
DR. TEMPLE: No. I just want to say one thing and it's not tooting my horn because they were written by someone else. I think this was a terrific set of questions, and I want to compliment Norm Stockbridge and Ray who did them. It led to what I think is one of the more fascinating discussions I ever remember at an advisory committee that brought out all kinds of stuff.
DR. LIPICKY: Just Norman. I had nothing to do with it.
DR. TEMPLE: Okay, even better.
DR. BORER: Okay. With those disclaimers having been made, we'll adjourn and meet again tomorrow morning.
(Whereupon, at 5:50 p.m., the committee was recessed, to reconvene at 8:30 a.m., Friday, January 18, 2002.)