9:10 a.m.


Friday, November 16, 2001








Conference Room 1066

5630 Fishers Lane

Rockville, Maryland


Science Board Members


Marion W. Anders, D.V.M., Ph.D.

Michael P. Doyle, Ph.D.

Owen Fennema, Ph.D.

Robert S. Langer, Sc.D., Chair

Robert M. Nerem, Ph.D.

Cecil M. Pickett, Ph.D.

Jose Principe, Ph.D.


FDA Participants


Norris E. Alderson, Ph.D., Acting Senior Advisor for Science, FDA


Dennis Baker, Associate Commissioner for Regulatory Affairs, FDA


Celeste F. Bove, CCC-A, Acting Executive Secretary, Office of Coordination and Communication


David W. Feigal, Jr., M.D., M.P.H., Director, Center for Devices and Radiological Health (CDRH)


Ajaz Hussain, Ph.D., Deputy Director, Office of

Pharmaceutical Science, CDER


David Lepay, M.D., Senior Advisor for Clinical Science, FDA


Joe Levitt, Director, Center for Food Safety and Nutrition (CFSAN)


Bernard Schwetz, D.V.M., Ph.D., Acting Principal Deputy Commissioner, FDA


Steven Sundlof, Director, Center for Veterinary Medicine (CVM)


Linda Suydam, D.P.A., Senior Associate Commissioner for Communications and Constituent Relations


Janet Woodcock, M.D., Director, Center for Drug Evaluation and Research (CDER)


Kathryn Zoon, Ph.D., Director, Center for Biologics

Evaluation and Research (CBER)

Invited Guests and Speakers


Frances Bruttin, PrincewaterhouseCoopers, Pharmaceutical Sector Team


Alexa I. Canady, M.D., Co-Chair, CDRH External Science Review Committee


Doug Dean, Ph.D., PricewaterhouseCoopers, Pharmaceutical Sector Team


Steve Hammond, Manager, Process Analytical Group, Pfizer


G.K. Raju, Ph.D., Executive Director, MIT Pharmaceutical Manufacturing Initiative


Norman Winskill, Ph.D., Vice President, Manufacturing, Pfizer, Inc.


Public Comment Speakers


Robert Chisolm, International Technology Manager,



Gideon Kantor, Ph.D., personal statement


Nouna Kettaneh-Wold, Umetrics


Gabi Levin, Ph.D., Brimrose Corporation of America


Scott C. Ratzan, M.D., M.P.H., M.A., Editor, Journal of Health Communication


Svante Wold, Ph.D., Umetrics




Call to Order

Robert S. Langer, Sc.D., Chair     5


Introductory Remarks

Bernard A. Schwetz, D.V.M., Ph.D.     7


Update on Action from April 2001 Meeting

Norris E. Alderson, Ph.D.     22


Emerging Science Issues:  Pharmaceutical Manufacturing

Janet Woodcock, M.D.     28

Doug Dean, Ph.D.     46

Frances Bruttin     57

G.K. Raju, Ph.D.     74

Norman Winskill, Ph.D.     111

Steve Hammond     123

Ajaz Hussain, Ph.D.     149



FDA Science Board     166


Open Public Comment

Nouna Kettaneh-Wold                 194

Gideon Kantor, Ph.D.     198

Scott C. Ratzan, M.D., M.P.H., M.A.     205

Gabi Levin, Ph.D.     221

Robert Chisolm     230


Update on CDRH External Science Review

Robert M. Nerem, Ph.D.     241


Response to Update

David W. Feigal, Jr., M.D., M.P.H.        260



FDA Science Board     286


Emerging Issues in FDA's Oversight of

Clinical Research

David A. Lepay, M.D.     298



Science Board     324


     CHAIRMAN LANGER:  Good morning.  I would like to call the meeting to order.  My name is Bob Langer.  I wanted to do two things.  First, my major task is to remind all the Board members to pay Donna $8 for lunch during the break.  And, secondly, I thought we could just maybe go around the table and introduce everyone just briefly.

     DR. PRINCIPE:  My name is Jose Principe.  I am Professor of Electrical Engineering.  My expertise is biosignal processing.  I work at the University of Florida.

     DR. PICKETT:  I am Cecil Pickett.  I am Executive Vice President of Research and Development at Schering-Plough Research Institute.

     DR. FENNEMA:  Owen Fennema, Emeritus Professor of Food Chemistry, University of Wisconsin.

     DR. NEREM:  Bob Nerem from Georgia Institute of Technology.  I am Professor of Mechanical Engineering and Biomedial Engineering.

     DR. CANADY:  I am Alexa Canady.  I am not on the Science Board but was on the Review Committee, Co-Chair, and I am Professor of Neurosurgery at Wayne State University.

     DR. DOYLE:  I am Mike Doyle.  I am Director of the Center for Food Safety at the University of Georgia.

     DR. ANDERS:  Greg Anders, now a Professor Emeritus of Pharmacology and Physiology at the University of Rochester.

     DR. SUYDAM:  I am Linda Suydam, FDA Senior Associate Commissioner for Communications and Constituent Relations.

     DR. SCHWETZ:  Bern Schwetz, Acting Commissioner of the FDA.

     CHAIRMAN LANGER:  I am Bob Langer.  I am Chair of the Science Board and a Professor at MIT in Chemical and Biomedical Engineering.

     DR. ALDERSON:  I am Norris Alderson.  I am the Senior Acting Director--I am Acting Senior Advisor for Science.  I'll get it right.  It's a good job.


     MS. BOVE:  I am Celeste Bove.  I'm the Acting Exec. Sec. in the Office of Science.

     MR. BAKER:  I am Dennis Baker.  I'm the Associate Commissioner for Regulatory Affairs.

     MR. SUNDLOF:  I am Steve Sundlof.  I am the Director of the Center for Veterinary Medicine.

     DR. WOODCOCK:  Janet Woodcock, Director, Center for Drugs.

     CHAIRMAN LANGER:  Okay, so now I would like to turn this over to Bern Schwetz, who is the Acting Principal Deputy Commissioner, to make some introductory remarks.

     DR. SCHWETZ:  Thank you, Bob.  I'm in the precarious spot of beginning to talk and my notes are still in my office over in the other building, which is a position I try not to get caught in very often.

     Let me follow up from where Norris was going with his title.  When the Science Board met last time, Liz Jacobson was still with us as the Acting Senior Advisor for Science, and in the meantime Liz has retired, and is still helping us in some ways.  But she is no longer in the position that she was in when we met last time, and instead Norris has agreed to come in and help in this regard.

     Norris has spent many years in the Center for Veterinary Medicine, and in his most recent position--recent in the last 15, 16, 17 years--has been the Director of the Office of Research within the Center for Veterinary Medicine.  And in consultation with Steve, who reluctantly agreed to let Norris out of CVM for a while, Norris agreed to come and fill this position as the Acting Senior Advisor for Science, and I am very happy to have Norris helping me in that regard.

     There is another change that has been made within my immediate Office of the Commissioner that I want to update you on.  Dr. Mac Lumpkin is now serving as the Acting Deputy Commissioner of the agency.  Mac was in the Center for Drugs with Janet for many years, and more recently has served as the senior medical advisor to me during this time when I have been the Acting Commissioner.

     So we have now given Mac a more permanent title, if you will, of the Acting Deputy Commissioner, so I am glad to have his help.  It helps to have, when you're not a physician in charge of the agency, it's nice to have a physician as a right-hand person helping with the day-to-day running of the agency.

     I also want to thank our two new members of the Science Board.  Dr. Cecil Pickett, as he has already identified himself, is from Schering-Plough Research Institute, the Executive Vice President of Discovery Research, a cell biologist by training, and was on the Science Advisory Board of NCTR.  And I had a conversation, a couple of conversations with Dan Casciano about the possibility of having Cecil work on the Science Board at this time, instead of on the NCTR Science Advisory Board, and Dan agreed.  So, Cecil, we're very happy to have you with us.

     And Dr. Principe is also new.  We welcome both of you to the Board.  A professor of electrical engineering, has an interest in something that is important to a lot of us, computational neuroengineering.  That is something that a lot of us have talked about in one way or another as it relates to drug research or a new approach for looking at the nervous system function.  That is something that I'm hoping we will be talking about more as we go.  So welcome to both of you.

     I have in my notes to talk about where we are from the standpoint of a new Commissioner, not that I can bring you information, but in hopes that maybe you have some information.


     Because while I don't think we're in a different position that we were several months ago, where names were being mentioned but there isn't any specific action.  So as far as I am in this position, the agency is still stuck with me, and for how long we don't know.  But at any rate, things continue as they have been.

     I want to talk a bit about counter-terrorism and the impact of the recent two months on the priorities and the operation and the activities of the agency.  Everybody says things aren't the way they used to be.  Things haven't gone back to normal.  I can assure you that things haven't been very normal in the agency since then.  And I'm not going to go into detail on anyone of these things, but I just want to kind of give you a flavor of the kinds of day-to-day issues that we've had and what has been consuming our time and energies at a time when we're trying to keep the usual business of the agency flowing.

     We have very thoroughly reviewed the emergency preparedness plans that we have within each of the Centers, because they all have them, for dealing with emergencies that relate to the products that are in the domain of that particular Center.  We have certainly spent a lot of time backing up the rest of the department and the other agencies from the standpoint of the availability of vaccines and drugs and medical devices, because whether it was the terrorist actions in New York City or anthrax since then, we needed to have everything from blood to skin, drugs, vaccines.

     Many of our products were out there and of concern in one way or another, so we spent a lot of time reviewing product security, and how would we know if our products were the subject of some terrorist action or not?  To say nothing of the fact that we have an adequate supply in the event that they are needed in case of some action.  So product security has been reviewed.

     Food security has been a big issue, and a lot of us have spent time talking to the press, talking to the public, talking to the industry, communicating about where we are on food security as a possible means of terrorist action.

     The threat of anthrax in our mail rooms had a major impact.  It's one thing to talk about having drugs available for those other people who might be exposed to anthrax, but when we have anthrax that supposedly--we had the presumptive positives in five of our mail rooms.  After routine sampling and culturing, the word came back that we had presumptive positives, and then there were many days before it was confirmed negative.

     Well, in the meantime our people, like others, were put on antibiotics, and thousands of questions, many hours spent in front of our employees trying to provide perspective and answer questions, and hope that we wouldn't get positive confirmed results back, and what were we going to do in the event that they did?  So I think that was another test of our ability to deal with an internal emergency that certainly had ramifications on the outside, as well.

     Again, the expertise that we have that comes into play at times like this, an example is the recent interest in the irradiation of mail as a way of getting rid of anthrax or some other organisms through the mail.  And in the Center for Devices and Radiological Health we have the expertise to be able to help answer those questions.  So our people have been in there with the Postal Service and the others on a daily basis trying to figure out, is that an effective and safe way, and what kind of operating conditions would you have to have?  What are the capabilities?  Is it just flat surface letters, or is it boxes?  How thick can something be?

     Another wrinkle to this is, if we go into the irradiation of mail aspect, what about things, products that we regulate that are shipped in the mail?  And that's everything from drugs to biologics to devices to food, a number of other things that end up going in the mail, that if it was irradiated, wouldn't be effective if you used it.  So there are a whole bunch of other questions.

     So then do we have to go back to the manufacturers and say, "In the event that your product is shipped in the mail, is it stable under these conditions?"  And what do you have to tell the consumer to worry about?  So it's amazing how many ripples come out of this kind of thing.

     Some of this has translated into a supplemental budget that we have put in, and from the standpoint of the money that has been requested by the administration, of that $20 billion that is earmarked so far, we have a request for $106 million.  $61 million of that would be used to help protect us against things that would be imported, so a lot of this would go to the field.

     A lot of it has to do with food security, but Joe Levitt isn't getting it hands-on.  It's really going to the field operations, so Dennis Baker and his people are the ones who will get a lot of this to help improve our presence at the border, as well as our presence, our capacity behind the border for doing the laboratory work that is increased in the event that we have activities at the border and have to do a lot of sampling.  Or the domestic side of it, where in the event that we have to do a lot of sampling because of questions that have arisen from domestic supplies, that we need the capacity to be able to do the lab work that this kind of increased action drives.

     So that's $61 of the $106 million, and the rest of it has to do with other parts of the agency where we're trying to increase the stockpiles, make sure we have the materials that are needed in the stockpiles that CDC manages and the rest of the department manages for emergencies.

     We have spent a lot of time in the product centers--we, I mean generously, the people who are in the product centers have spent a huge amount of time communicating with the manufacturers to be sure that products that are critical are in fact not only available but in a supply that would be sufficient to meet the needs that we might anticipate under emergency conditions, antibiotics and vaccines in particular, and the possibility that we would develop new capacity for producing vaccines.  There have been a lot of discussions with the industry that hasn't been manufacturing vaccines, or at least the ones that we need for anthrax, smallpox, and so on.

     So there has been a lot of discussion, and the pressure that we get to get out there and make it tomorrow and have it ready is one that is very difficult for us, because the last thing we want to do is to approve a vaccine that's coming from a new source, that isn't safe or isn't effective, and to build up hopes only to find out that it wasn't what we thought and we may have created a worse problem than we hoped to solve.

     So at a time when everybody wants to be protected right now and have everything in place, we are trying to make sure that in the event that we increase the capabilities for manufacturing or developing new products, that when they are made available, they are safe, but do it as fast as we can so that we are not seen as "business as usual" and there's an emergency and we stood in the way.

     We of course have continued to raise attention to the fact that we have some pathogens in laboratories of ours, and so does the industry that we regulate that's developing vaccines and other products, and to get out and find out the security of all of those pathogens that could be of interest to somebody else.  So that has been a focus of ours, and part of this money that I'm talking about goes to increase our own internal security.

     And of course the amount of coordination with other agencies, inside and outside the department, has consumed hours per day of a lot of us, just talking and being sure that we keep the Secretary informed, or other members of our department who are out in the press informed, or to be sure that as decisions are made on how we should deal with these kinds of issues, that the best thoughts are put on the table to help focus the discussion and make some of these decisions.

     One of the things that we have developed internally as a result of this is what we're referring to, and the name isn't necessarily nailed down yet, but a new Crisis Management Center.  Dr. Woodcock has agreed to help pull this together for us, so for now, Janet is detailed to the Office of the Commissioner and Dr. Steve Galson, who is her new deputy, is helping to run the Center from day to day, but Janet is nearby.  But Janet is helping us develop a plan to have a Crisis Management Center.

     The agency has decades of experience in dealing with emergencies.  That is one thing that has been part of our menu all along, because we have an average of two to three tampering incidents per week, and those tamperings might be anything from a hoax, some of them have been real, some of them have been not only real but very serious.  And the possibility now on our mind is that you never know when a tampering--this has been true before, but especially true now--when one of those tampering incidents might be the beginning of a terrorist activity.

     And because we see a lot of these on a day-to-day basis and the Centers are well-prepared to deal with food poisoning issues and other product tamperings, there is a threshold of that at which we need to engage as a whole agency, as we have for the anthrax, as opposed to the Center for Foods dealing with a food crisis that they have dealt with in a normal way all along and it doesn't come to the attention of the whole agency.

     So this management center will help to keep us communicating and help to sort out those more routine variety of emergencies that the Centers take care of all the time, as opposed to one that could rise to the level that the whole agency needs to deal with it and be prepared to work with CDC and maybe the USDA and EPA and other agencies.  So we appreciate that Janet has agreed to help bring this center to fruition.

     Let me switch gears now.  We end up spending a lot of our time talking about terrorism and preparedness for it, but I want to talk about some other things, too.

     Peer review.  The CDRH peer review has been completed to the stage where Bob and Alexa are going to comment on the review that has been done of CDRH.  We are happy to see that this has been a productive process, and Dr. Feigal is going to respond to the comments that will be made.

     We continue to talk to Dennis in the Office of Regulatory Affairs, with the expectation that a review of ORA--because of the science within the field operation--would be the next peer review, and then CVM and CDER will follow, so one-by-one we keep moving on this.

     Just a very small comment on budgets from the standpoint of the 2002 budget.  We are under continuing resolution.  Both the House and the Senate have passed conference reports, and the agency is doing pretty well for 2002.

     And we are pleased to know that we are receiving the pay increase, the salary increases as an item this year, as opposed to in past years having to take the required pay increases of our employees out of our operating budget.  Well, this year the operating budget is there and, in addition, there is money for the pay increases, so that is a significant step forward.  And if we had had that for the last five or six years, we wouldn't have been in the trouble we were programatically.

     In addition, I mentioned the supplemental budget request of $106 million.  That will help.  One of the things that we realize is that while we have spent a lot of money on the food safety initiative in the last three or four years, a lot of the investment that we've made in food safety is very important from the standpoint of food security, so it has been a good investment.

     And it goes the other way, as well, so that as we get money for people at the borders, as we get more money to help with vaccines and drugs and what not, that helps, the security money helps safety surveillance and maintenance as well.  So it isn't that we are investing in something that has a use only under those peculiar circumstances, so that's good for us.

     We're working on the 2003 budget, but that is something that we will continue on for the next couple of years.

     As I close, let me come back to Board members.  Two of our members are here for the last time.  In fact, only one of them is here.  Dr. Marian Nessle couldn't be here today, but today is her last meeting.  But in addition, Greg Anders is here as his last meeting this time, and I very much appreciate the help that both of you have been during this time.  Greg helped me as a member of our Science Advisory Board at NCTR for a number of years while I was still there, and I convinced him to come and be part of our Science Board here, as well.  So, Greg, it has been a long time that you have committed your attention to us, and I really appreciate it, and thank you a lot.

     In addition, I would remind you that over the next year--and please don't stand up and cheer when I say this--we do have more members whose term will be coming to an end, so I would ask you to be thinking about other colleagues of yours whom you would like to recommend as replacements, as we look at five new members coming on next year.  So we can deal with that down the road, but be thinking of names.

     Bob, I think I'll turn it back to you.

     CHAIRMAN LANGER:  That sounds great.  Let me now turn it over to Norris Alderson to give us an update on the action items from the April 2001 meeting.  Norris?

     DR. ALDERSON:  Thanks, Bob.  Let me tell you that, first, I'm looking forward to working with all of you at the coming meetings, assuming I'll be here.  That is still an unknown.  But I promise that the next time we meet, I will have the title correct.

     I do want to remind you that Dr. Schwetz mentioned the two Board members, that this would be their last meeting, we have selected two members to replace them and I want to tell you about those.

     First is Dr. Jim Riviere.  He's from the College of Veterinary Medicine at North Carolina State University, and he's a veterinary pharmacologist.  The second is Dr. Josephine Grima, who is Director of Research and Legislative Affairs for the National Marfan Foundation.  Her expertise is in biochemistry, molecular biology, and cell biology.

     Another little update is, I'm sure all of you remember Sue Bond.  I'm glad to tell you that Sue and Rod Bond are the proud parents of a baby girl born in October, and so Sue is out on maternity leave at this time, and she will be back in February.  In her absence, Celeste Bove, to my right, has been filling in extremely well, and I'm extremely proud of the fact that we've had the staff to fill in behind Sue.

     Two other people I do want to recognize that have helped in this, and that's Donna Mentch and Monica Spence.  And particularly Monica today, if you've got problems with travel and things like that--I already heard one person that needs some help--Monica can help you take care of those problems.  She will also be dealing with you in getting your reimbursement for your expenses, so don't forget that.

     Now, an update from the last meeting.  Bern has mentioned the peer review, so I won't mention that anymore.  One other suggestion that came out of the April meeting was that we have an Ethics Advisory Committee, and remembering back, this came up in a discussion, I believe by Dr. Zoon, on tissue engineering and cloning.

     The leadership council of the FDA has looked at this and decided to establish a list of ethicists that we will maintain in the Office of Science Coordination and Communication for our use.  We looked at that and felt this was the best way to address it.  So we took your consideration and we put something in place to have that for us when we need it.

     The second item is, if you recall last meeting, Dr. Skulnick made some comments regarding institutionalizing our peer review system.  Since he is not here today, we'd like to hold off on discussing that because I think he has a big interest in that, so we will bring that up at future meetings.

     I do want to bring your attention to the upcoming FDA achievement awards.  If you will recall, that's a process that you are the final determining body on who gets those awards.  Our in-house review committee has completed their review of these.  We will be forwarding to you in the next two weeks two candidates for each of the categories, and you'll get the opportunity to vote, I hope by December the 7th, on your recommendations for each of those awards.

     These awards will be presented at our Science Board in February.  So this is very key to us.  We've got a lot of good scientists in the agency.  This is one way we are able to recognize those, and we think it's very important to them.

     One final point.  We communicated to all of you this summer about coming in early for one of the meetings to visit our laboratories.  I think all of you responded to that in an affirmative way.  You've told us before that you wanted to hear more about how we establish priorities for our research, and this is one way we can start that discussion with you, is with you going out to our laboratories and really talking to the scientists and the managers there to address this issue.

     So this will require you to come in a little earlier.  We would like to do this the afternoon before this meeting.  So at the same time we've had to change the date of our April meeting because of a conflict with the Food and Drug Law Institute meeting.  So we want you to respond back to us as soon as possible about your available dates, April and May time frame.  If we don't hear from you shortly, we will take the initiative to get back to you.

     So that's my comments, Bob.

     CHAIRMAN LANGER:  That's terrific, and very concise.  That's very helpful.  Thanks a lot.

     Well, I think we've got a very exciting morning ahead of us, and the issue that we're going to discuss is the emerging science issue of pharmaceutical manufacturing, and Janet Woodcock is going to lead that discussion.  I'll really turn this over to her.  Janet?

     DR. WOODCOCK:  Thank you.  Good morning, everyone.  If I can get the shift P--shift key?  All right.  Now, what else do you want me to do?  You'll do it?  All right.

     While we're getting our audiovisual stuff working, what I want to talk about today, and the program we have put on, relates to science issues in the regulation of pharmaceutical quality.  Now, pharmaceutical quality we think is really--it's the roots of drug regulation for the FDA.  It was quality problems that, in the early part of the 20th century, that really led to the formation of the Pure Food and Drug Act, the Food, Drug and Cosmetic Act.

     CHAIRMAN LANGER:  Janet, microphone.

     DR. WOODCOCK:  Oh, microphone, sorry.  Okay.  That led to the formation of the FD&C Act eventually.  There were improper constituents, there were impurities based on manufacturing and so forth that led to various tragedies.

     Since that time, drug quality is really felt to be the basis of our findings of safety and effectiveness, because if the quality of the product fluctuates, the findings from the clinical trials of safety and effectiveness can't be relied upon to project into what the drug actually is.  So this is a fundamental issue for the FDA, but we feel that is currently being challenged, and let me explain how.

     How do we regulate drug quality right now?  Well, I'm going to go over this a little bit because some of you may not really be aware of this.  The pharmaceutical industry manufacturing sector is extremely tightly regulated by the FDA.  That's the first fact.

     As most of you may know, before a product is approved by the FDA, before a drug product is approved, there has to be pre-review and approval of both the process, the synthesis, the manufacturing process and so forth, all the documentation that has been assembled by the firm, and aspects of the facility submitted.  All of that is reviewed in a prior approval way before a new drug would get onto the market.

     In addition, the facility is inspected by FDA inspectors and everything is gone through extremely carefully there.  And once a drug is on the market, if there are changes to the product or the process, these usually have to be submitted to the FDA and reviewed prior to being instituted.

     And in addition the ongoing manufacturing facility is subject to FDA inspection and has to conform to standards called Good Manufacturing Practices.  And according to the statutory framework, that is supposed to be done at least every two years, so every firm should be inspected by the FDA every two years.

     So this scheme, which has been in effect for many decades, is how a pharmaceutical quality is currently regulated by the FDA in the United States.  Now, we think, one of the reasons we are coming here today is that we think there are some issues regarding this.

     The goal of regulating pharmaceutical quality this tightly is that the product be of the highest possible quality.  In other words, that's why we do this regulation, to ensure the highest possible quality of the pharmaceutical products that are regulated, but we feel that we may not be totally achieving this goal in some ways.

     We are seeing at FDA an increasing trend toward manufacturing-related problems.  These include things like recalls.  When there are manufacturing problems, it may disrupt manufacturing operations, and we may see shortages or loss of availability of important drugs.  And this also can have, manufacturing problems or issues can have a negative impact on getting new drug approvals out.

     So that is one issue.  Although we think the products now are of high quality, we will have some presentations today that discuss the fact that the pharmaceutical manufacturing sector may have low manufacturing and quality assurance process efficiency.  And this is an issue because the cost of drugs is a real issue for health care in the United States, and this contributes, a lack of efficiency in this sector contributes to the cost.

     Now, not a lot of people have talked about this.  I don't think it's a real popular subject.  So I think this may be a somewhat controversial set of presentations that you're going to hear this morning.

     And, in addition, we find, we feel that innovation, modernization, and adoption of new technology in this sector has been slowed.  And what do we base that assertion on?  Well, we know that in other countries where products are manufactured not for the U.S. market, there has been a more rapid introduction and adoption of new technologies of various kinds.  And that might be for the same product that is actually manufactured for U.S. use where those technologies are not employed.  And these technologies we're talking about usually lead to a better level of quality assurance.

     Now, the final issue for us in the current system of how we regulate pharmaceutical quality is, it really does place a high burden on FDA resources.  Our regulation of pharmaceutical quality is resource-intensive for the FDA.

     We get about 4,000 supplements submitted yearly.  These are what I talked about earlier.  When a change is going to be made in the manufacturing process or in the product in some way, a supplement must be submitted to the FDA.  Some of these have to be reviewed by us.  Some of them are simply noted by us.

     Our FDA inspectors are unable to meet, currently, the biennial GMP inspection requirement, so we are not in the plants every two years, as Dennis very well knows.  And unfortunately for the non-domestic industry--and globalization is causing a major shift, especially in the bulk drug manufacture, to overseas facilities all around the world--our presence is even less there.

     And some of the cause of this is the financial issues that I think Bern and others have introduced the Board to, some of the financial challenges that the FDA faces.  However, nevertheless, whatever the cause, the bottom line is, we are not in the foreign plants even as often as we are in the U.S. plants, which is much less often than we are required to be under the statute.

     This graph just shows, from the previous 10 years, sort of the rate of increase of us getting these manufacturing supplements.  You can see as we approve new drugs, very frequently they go through a lot of changes in the next few years.  Especially with current drug development, where things may not be completely worked out or optimized at the time the drug goes on the market, we're seeing a lot of supplements filed, and I think this reflects that.  It also reflects the success of our generics program.  We're getting a lot of generic drugs on the market, and they have a lot of manufacturing supplements filed to them.

     Now, due to the efforts of our chemists in developing new guidance, many of these in recent years have changed from requiring pre-approval, where the firms have to wait when they submit these until they're approved by the FDA, they have changed to what we call changes being effected, where they simply can notify us.  They simply can send in the supplement, notify us, and go ahead and implement the change.  But nevertheless you can see this is an increasing burden for the FDA at a time we don't have additional resources to deal with this.

     Now, how did this system--how did we get here?  Well, this whole system evolved starting about 30 to 40 years ago, at a time when the sectors of industry, many of them, by no means universal, but there were parts of industry that really lacked rigorous manufacturing procedures, so it was totally appropriate, the changes in regulation that were implemented at the time.

     But more importantly than lacking SOPs, there was really--at that time the science had not advanced to the point where there was some predictability in what factors affect a formulation's performance, if you follow me.  So it was really what I call sort of a "know nothing" approach.  We know nothing about what impact a change will have on a formulation.  Therefore, we must control and check everything, because we cannot predict, we cannot model what's going to happen if we make a change.

     And subsequent to that, the science and technology base--and again, this is going to be controversial--it hasn't evolved as quickly as in other manufacturing sectors, and there are probably a number of reasons for that, which I'll go into a little bit.

     Now, the standards that I referred to earlier, the Good Manufacturing Practice standards, which are the standards that we impose for inspecting plants, process standards for manufacturing facilities, these are empirical standards.  They are not scientifically based standards.  And the reason for that, again, is that we didn't have the underlying science to be able to predict what factors were important.

     Now, over the last 10 years, the Center for Drugs and the Center for Biologics have worked on the International Conference on Harmonization, which is a sort of international standards harmonization setting body that developed consensus-based standards in a number of quality areas.   And most of these had to do with the review aspects, what we review in terms of stability, how you test for stability and things like that.  So some of those may have a more modern science base to them, but nevertheless they were primarily consensus-based, based on the experience and standards of the three regulatory agencies, Japan, EU, and U.S., that actually were putting these standards together.

     And finally, I have to mention this because I think it's a very strong factor, why has the science base and technology base--and I see people nodding in the audience--not evolved rapidly like it has in other sectors, aerospace or, I don't know, computer chips or what have you?

     Well, for the innovative pharmaceutical industry, the most important thing is to get the products on the market rapidly, and we all know that.  Manufacturing is really an impediment, in a way, something that shouldn't get in the way of that happening.

     And so the industry, in the face of this intense regulation, I think has been very risk-averse in introducing new technologies or in challenging the FDA standards, because the bottom line was to get the product on the market or keep the product on the market.  And this has sort of played into, I think, the issues around adoption of new technologies.

     Now, where are we now?  That's how we got here.  Where are we now?  Well, you know, the drug discovery revolution, which I'm sure you all have talked about quite a bit, has really increased the early pipeline, and there's really not that many barriers to developing candidates, molecular candidates, so there's a push at the very early part of the pipeline.

     There are a lot of ongoing efforts in companies to improve how drugs are developed, the pre-clinical to early clinical paradigm, to move that along quickly.  What we are seeing, I think, is that as the clinical drug development time shortens, as there is an emphasis on speed, there is less and less attention paid during the clinical drug development phase to formulation development, manufacturing process development.

     Now, all of this we think is feeding into some of the issues I talked about earlier, some of the problems that we're seeing, so that we feel it's probably unlikely that the innovator industry will slow down drug development in order to get their formulations finished, and so forth, and perfect.  So what we really think is needed is some innovation in manufacturing process R&D, introduction of new technologies, so when these products come onto the market, we can all have confidence that they will perform reliably, and that innovation can continue to be incorporated into the manufacturing sector.

     Now, the challenges for FDA in this regard are--and this is our challenge everywhere.  It's our challenge in the clinical area and everywhere, in regulating an innovative industry.  It's how do we encourage innovation while ensuring that the quality is maintained.  That's one of our big issues.

     We know, or we certainly all hypothesize, I think, that successful adoption of new technologies will actually improve overall quality of pharmaceuticals, and also probably efficiency, and you'll hear about this in the presentations.  But how do we do that?  How do we enable the introduction of new technologies while maintaining the quality standards?  And this is the essential issue that we're dealing with.

     In particular, in this case, how do we successfully shift from empirical, the art of manufacture based standards, to science-based standards for manufacturing process quality?  How are we going to do that?  It's a huge challenge.

     In addition, we need to try to further decrease reliance on pre-approval review and on the physical, actually getting in there and touching the product and the lines and so forth, not because that's a bad model, but in fact we don't have the resources to do it, and that has been clearly demonstrated over the last decade.  We are not getting in there all the time.  So what other ways can we use to evaluate quality?  And, finally, how to recruit and train a scientific work force that would be proficient in the application of these new technologies.

     Now, today's approach, what we're going to do, we're going to present the problem to you from a variety of perspectives and go into much more detail, and we're going to use Process Analytical Technology as an example of new technology.  By no means are we saying that this is the only new kind of technology that needs to be introduced, but we felt that we needed some firm example that people could look at to see what we were talking about, and so we'll be using some process analysis technologies as an example of the kind of new technologies that could be helpful.

     Our speakers, we're first going to have Doug Dean and Frances Bruttin from PricewaterhouseCoopers talk about it.  Then we're going to have G.K. Raju, who is from academia at MIT, their Pharmaceutical Manufacturing Initiative, so an academic perspective on the study of manufacturing processes.  Norman Winskill and Steve Hammond from Pfizer are going to give you the industrial perspective from their point of view, and Pfizer has been adopting some of these technologies, not in their regulated lines but in other aspects of their R&D.  And finally we'll have an FDA perspective from Ajaz Hussain, who is the Deputy Director of our Office of Pharmaceutical Science.

     Hopefully, by presenting from these various perspectives, you'll get a broad view of what the problem is and also some of the various approaches that we might take to this problem.  For the Science Board specifically, we'll be asking you at the end of the day if you are able to support our approach and what your comments are on this approach.  What resources, external, scientific, academic or whatever, what resources do you suggest we would draw on to bring this about, assuming you do support the approach?  And, finally, are there additional aspects to the regulation of pharmaceutical quality that we should focus on, that we are not adequately highlighting in this presentation and approach?

     So, with that, I'll call on the first speaker.

     DR. ANDERS:  Janet, do you have time for a quick question?

     DR. WOODCOCK:  Certainly.  Go ahead.

     DR. ANDERS:  In your third slide, you--I'm trying to quantify the magnitude of the issue.  So you talk about recalls and loss of availability of essential drugs.  Can you educate us a little bit, what's the magnitude of this problem?

     DR. WOODCOCK:  All right.  Well, sometimes it's difficult for us to talk about these things, which is unfortunate.  But I think even in the era of bioterrorism, for example, there are certain antibiotics and other drugs that used to be manufactured, that are in short supply or are not manufactured.

     One of the contributing factors to not manufacturing drugs is that the process, it is felt it would be so expensive to bring those processes up to the modern standards, do all the validation and all the other work that is required to be done, that companies abandon the manufacture, abandon that product.  And so we get into situations where we have more single-source manufacturers, where we have manufacturers, we have products where people abandon the manufacture.

     DR. ANDERS:  Is it 5 percent of the drugs that are recalled, or 50 percent?

     DR. WOODCOCK:  Oh, you mean what is the relationship of recall to manufacturing problems?

     DR. ANDERS:  What's the magnitude of this problem?  And then, again, the loss, I guess you just now addressed the loss of essential drugs that a manufacturer may say, "Well, I'm going to quit making this compound," and so we have one provider.

     DR. WOODCOCK:  Right.

     DR. ANDERS:  But what's the recall?  What are recall numbers like?

     DR. WOODCOCK:  We can get you the numbers.  They have been rising over the past few years, but that's only one aspect.  Let us go through the entire presentations, set of presentations, because that's just one facet of this problem.

     I'm not saying that manufactured pharmaceuticals are of low quality now.  That's not really the issue.  All right?  But getting into problems that lead to recalls and other shortages and so forth, that represents a problem, a system problem, you know, and we are constantly dealing with that.  We're dealing with shortages that are generated by different manufacturing problems.  I can tell you that we are constantly dealing with this.  And I can't give you a figure, like how often it happens, but it's a constant theme that the FDA has to deal with.

     All right, we'll move on to the first speaker.

     DR. DEAN:  Good morning, everyone.  My name is Doug Dean.  I am from PricewaterhouseCoopers.  I'm based in Basel, Switzerland.  And I am not an accountant.

     You're going to get a double act this morning.  I'm here with my colleague, Frances Bruttin.  Together we have been working in the pharmaceutical sector for a number of years, myself for about 25 years in pharmaceutical manufacturing, and we're going to share with you some of the observations that we have had working with our clients all over the world for the last decade or so.

     And I really like this image to start with, because I think it sums up where we're coming from.  It's a combination of cost, time, and regulation.  Before we get into this, I'd like to just declare some biases here and make sure that you understand the perspective that we're coming from.

     First and foremost, we're engaged by our clients to solve business problems.  This typically means looking for ways to improve the way the business operates and generate greater return to shareholders.  We've been doing this exclusively in the manufacturing sector, largely looking at new processes, new ways of working, and new systems to support those ways of working for the past decade or so, primarily focused on working with R&D based large pharmaceutical manufacturing organizations.

     I think what we'd like to shortly talk to you about today are the perspectives that we have seen in the past decade and the conclusions that we're reaching, basically that the way things currently are in the sector and in the industry is something that can't be sustained going forward.  I think we've already heard from Janet this morning a couple of comments about the state of pharmaceutical manufacturing in general, and I'd like to emphasize that.  Compared to other sectors, the capabilities and the state of pharmaceutical manufacturing is actually quite poor.

     We are going to show you how the way that we traditionally go about measuring our performance in manufacturing actually hides the opportunity for improvement, and we need to consider some different ways to measure it.  The infrastructures that we put in place to support the need for the regulations that we have to work within tend to be not very economically feasible.

     So there's opportunities there for improvement, and that through the introduction of new technologies, but I think critical to emphasize, technologies that are not put in in isolation.  And I would like to draw that point out very clearly today.

     And we will conclude by showing you that there is a massive potential for win-win, both for the consumers, for the industry, and for the shareholders, chiefly through four different aspects here.  We're going to show you how we can reduce risk; increase the effectiveness of our ability to be compliant with the various regulations that we have to deal with.  In so doing, we will show how it's possible to dramatically reduce cost in manufacturing; and by doing that, to give some increased return to the shareholders.

     And these things combined together, then, are going to create a win for the regulators and the consumers and a win for the business, and we'll show you how these things are strongly linked.

     I think it's important to understand the environment that we are currently working in.  I mean, here big pharma.  If we look at what has happened in the industry over the past 30 years, we've seen a dramatic decline in growth, from double-digit growth 30 years ago to what seems to be a level of growth in the market that's leveling out at about 6 percent per year.  Every economist has got a different opinion, but when you talk to them, they are generally in agreement that the industry will continue to grow in an environment of about 6 percent per year.

     If we look at the total annualized return to shareholders over a five-year period, and look at that over the last number of years, it's been steadily declining.  It's the investment from the shareholders that provides the capital to grow the business, to look for new products, so it's important that the shareholders do get a return.  But I think it's important to draw out here that in this environment of reduced growth, there has been reduced return to those shareholders.

     If we look in the engine room of the industry here, the innovation and the discovery and bringing to market of new drugs, we've seen a dramatic fall-off overall in the productivity in research and development.  We as an industry are pouring tens of billions of dollars into research and development but we're getting less and less out of it, measured in terms of the number of NCEs or NMEs that are brought successfully to market each year.

     When we are bringing new drugs to market, what we're finding is that over the past 30 or so years there has been a dramatic decrease in the window of opportunity that we have to get a return on that massive investment.  So decades ago one could enjoy a window of therapeutic exclusivity on the order of years, but lately we've seen that shrinking down to a matter of months or even weeks.  So bigger investments, market growing more slowly, less opportunity to make a return on those investments, emphasis on time, time, time, get to market more quickly.

     When we look in more detail at manufacturing in general, and I would say this is across the board in virtually every dosage form, whether we're talking about active ingredients or secondary production, we generally see that the levels of asset utilization in the sector are very, very low, typically about 15 percent.  But because of the way that we measure the way we use our assets, we often fool ourselves into thinking it's a lot higher, and I'll draw this out more clearly for you later.

     We tend as an industry to accept the fact that we're going to have to throw away 5 to 10 percent of everything that we produce, or we're going to have to rework it, and we plan it in.

     It typically takes, in a new product introduction environment, a good deal of time to actually get the scaled-up commercial level processes effective, working to the levels that we would like them to work at, and this is taking far, far too long.  And it's basically accepted, I would say.

     And as we'll show you, very typical across the board to see a total cost of quality approaching 20 to 25 percent in some cases of period costs in a given pharmaceutical manufacturing plant, and this is accepted as being just the way it has to be in the business.  So that's the environment that we find ourselves in.

     Conclusions here is, it's tough and it's going to get tougher, so there's going to be an increasing, an intensification of competition within the business for resources.  So if we're asking for capital investment to improve manufacturing, well, R&D are asking for more capital as well, so we have to have and be able to demonstrate a very good return on that investment.

     And I think the other key conclusion here is that manufacturing has been really regarded as a Cinderella function in the industry, the poor stepchild, and there has been a "cause no problems" mentality which has really led to what I can only characterize as benign neglect of the need for higher and better performance in manufacturing operations.  Steven Wheelwright did a number of studies showing that manufacturing as a function has to stop being internally negative or internally neutral and become an external supporter to organization strategy.  So manufacturing is going to have to contribute more than it currently is.

     Now, when we start looking at why is it like this and where do the problems come from, we find that actually the problem happens downstream.  It happens, begins to happen long before we ever get to commercial-scale production.  We find that processes are transferred into manufacturing that are in most cases not well understood, and in many cases are not capable at the scales that we have to manufacture at.  They are capable in laboratory or at clinical scales but not at commercial scale.

     We find that the emphasis and the focus on new product introduction has resulted in masses of data that are going into the CMC section, but somehow we are missing a lot of critical information that helps people to actually understand the processes to be able to operate them effectively in a production environment.

     We have done a number of studies that have showed that long before we get to Phase III clinical trials, approximately half of the manufacturing costs are already locked in, and therefore we're not able to do anything about it in terms of reducing them when we finally do get to production.

     And, finally, we find that there is little basis, little scientific basis to take a decision to trade off the pressure to get a new compound to market more quickly, to trade that time off in exchange for maybe a month or two more to spend more time understanding the process in order to enjoy the benefits of increased quality downstream.

     Just to give one example here, a client that we worked with a number of years ago, a very simple example, this is an emulsion product.  The critical quality attributes relate to the size of the droplets in the emulsion, and it's measured in an in-line process environment by shining light through and looking at the degree of absorption of the light.

     And there were some upper and lower control limits set on this process in order to get the emulsion the way that it should be.  The usual validation studies were done, three batches at exactly the middle of the upper and lower bounds there.

     The problem was, the process was not well understood, and in fact it had been noted earlier in clinical manufacturing that the function of absorption as a function of the control parameter created a situation where we were actually out of limit in one of the quality attributes, but this was somehow not understood in transferring the process to manufacturing.

     So we've now got a situation where we have a process in manufacturing, we've got an upper and lower control limit, we have operators who are doing their best to keep the process within those boundaries, but it's fundamentally flawed.  Even though we're within those boundaries, we're still producing a product that does not meet its quality attributes and will have to be scrapped or reworked.

     What we'd like to show you, and I'll turn over to my colleague France Bruttin here to take you through this, three key things here in order to start to address this problem.

     One is that in looking at manufacturing operations, we need to clearly understand where the value-adding activities are and where the non value-adding activities are.  We find that non value-adding activities add cost and time but they don't add value to the product.

     We'll take you through one or two examples that show how, as an industry, our traditional approach to measuring performance in manufacturing hides the potential to improve and reach higher levels of performance that we really need.  So as a non-accountant, I can blame the accountants for this, and we'll take you through some of that.

     And we'd like to show you some rigorous approaches that are being used in other industries to apply a more rigorous and scientific-based approach to determine the ability of a process to be right the first time.

     I'll turn it over to my colleague Frances Bruttin now to take you through this.  Frances?

     MS. BRUTTIN:  I think, building first of all on a point that Janet was trying to bring across, where she was saying that the manufacturing, the pharmaceutical manufacturing tend not to be taking opportunities of the new technologies and the innovation and bringing them into their daily work hangs around them being adverse to change, hangs around them being afraid of bringing change into the manufacturing process because of the impact with the regulators and the amount of paperwork that they would have to go through, so they prefer to stay with today's inefficient processes and ways of working and continue to get the product out the door.

     I think we are going to tell you a little bit today that it goes a little bit deeper than that; that in reality, within the basis of manufacturing, there are a lot of inefficiencies, there is a lot of emphasis on trying to test quality in and prove quality, whereas if we take a few steps back and go back to basics, there are some fundamental things that we can change, and we will get quality and regulatory compliance out as a consequence.  So I'm going to take you through some of them.

     First of all, if we look at what actually happens in the plants and we distinguish between value-added activities and non value-added activities, this is an example of a dispensing process.  The process actually takes three days to be completed.  Of those three days, 1 percent of that time is actually value-added.  That is the weighing of the material before it goes into the hopper or the dispenser.  The rest of that time is dedicated either to transportation of material; scanning in bar codes.  It's waiting while the QA come back with the results of the bar code.  It's moving the pallets from one area to another area.  So over three days, 1 percent of all that time is actually value-adding.

     If we take all of the subprocesses that make up the process from the raw materials to finished goods, the actual value-added time as we go through dispension, granulation, compression, coating, and packaging, over a 35-day process, three days are actually value-added.  And these are real numbers.  These are numbers from the studies that we have done actually in the shop floor, on the factory floor.

     To make it a little bit more realistic, we've actually taken photos of where these delays can be observed.  what we have here is, after dispensing, the material is actually stored.  In this case it was actually in the alleyways between the various different rooms where the production processes were taking place.  So here we have delays.  They were staying here for maybe typically five or six days.

     Work in process, again, waiting until the next piece of equipment was ready; waiting for results coming back from QA for the in-process controls.  Again, all this material is captured, which has been held up; it's work in progress.

     And a very typical photo of what turned out to be 100 percent inspection.  There was a problem here with the blister pack, and so these ladies over three shifts for a period of two or three weeks did 100 percent inspection for a single batch.  This is non value-added activities.  This is not contributing to the quality of the product, and it's certainly not contributing to the health of the final patient.

     If you measure what is actually value-adding and non value-adding, and the way we do this is through an activity-based analysis--so we look at what the people are actually doing, we're looking at what is happening to the product as it moves along the process--you can identify those processes that are not actually adding value, you can reconfigure them, and you can make a much more effective manufacturing process.  The best we have seen is about 50 percent.  That means the ratio between non value-added activities and value-added activities is 50-50.  So in this example it is possible to go through the entire manufacturing process in 6 days as opposed to 35 days.

     So value-added/non value-added is one aspect.  The next aspect--and I have to say this is probably due to our colleagues accounting--is how we measure the effectiveness or the output of a factory.  We use standard accounting methods.  This is also driven by the various MRP systems that have been in place.  We talk about standard costs.  We talk about standard utilization.

     This is typical of what you will see.  For asset utilization, what is defined as the total availability of an asset?  We have here 80 hours a week.  That represents two shifts per day, and there is of course the scheduled downtime and the scheduled conversion time.  But since the accountants know that there are traditional losses and other expected losses and delays and waits, then they also schedule in a certain amount of time that they know will not be operational.  Because they schedule it in, it's planned in, so suddenly we lose visibility and transparency of it.

     So the people who come later say, oh, well, the allocated operational utilization is set at X percent.  That X percent has already taken in the fact that the actual equipment is sitting idle.  And therefore if they hit 90 percent of that X percent, they feel they're doing a good job.

     If we look in reality at what is happening--and remember these assets that we have, we have capital investors in those assets, so those assets are alive, let's say, 24 hours a day, 7 days a week, 365 days a year.  So when we're calculating asset utilization, we should be calculating it on that basis.  So instead of it being 80 hours a week, it's actually 160-hour-a-week availability.

     If we then look, we find out that we have quite a high percentage which is unscheduled downtime, so downtime due to problems on the line, due to problems on line switch-over.  Then we have operational time losses which are due to, in some cases to poor planning; to delays, again, delays with suppliers, delays with other material coming in.  And then which I think is actually quite sad, when we look at what we're doing with our machines, we're actually spending quite a significant amount of time producing scrap, i.e., all those batches that do not meet the quality control that we've set on them, or reworking that scrap.

     So at the end of the day, the actual effective up time is actually a very, very small percentage.  From our studies this has come out to be around 15 percent asset utilization.

     So we've looked at value-added versus non value-added, we've looked at the difference between measuring for accountancy purposes and measuring for performance.  So some people may be saying now, "Well, that's fine.  I can blame the plant manager on all of that.  This has nothing to do with our scientific approach to manufacturing."

     Let's get to the main point.  The main point is around the pharmaceutical process.  We have been arguing that this process is not well understood.  It's not well understood when it's transferred from R&D over into manufacturing.  And what we have here is a way of measuring the processes, a way of comparing processes from one process to another.  Obviously it's sigma.  It's not rocket science.  It's based on standard deviations.

     It is a measure that has been used in various other industries.  For example, if we start from manufacturing, Motorola and Siemens, around consumer electronics, but also in service industries, GE Capital, so for people who provide financial services, and also Caterpillar.  So it's a way of measuring lots of processes, not only the pharmaceutical process but all the processes, so the steps that people go through to support manufacturing, how people take those let's say manual steps, how they started, and how repeatable those processes are.

     What this lets us do is see how well we are performing, see how repeatable our processes are through the plant.  Obviously a higher sigma value indicates better performance.  If we look here in terms of defects a 2 sigma process has around 300,000 defects; a 6 sigma process has 3 defects.  This is what you may have heard about in terms of zero defects manufacturing.

     If we see where the various industries are, semiconductors are between 5.5 and 6.  Baggage handling in airports is a 4 sigma process.  So where do you think pharmaceutical manufacturing is?  Probably if I asked a CEO, he would tell me, "Yeah, we're somewhere between 4 and 4.5."  Some people are saying uh-uh.  We're actually about 2.5 consistently.  Consistently 2.5.

     Two ways of seeing this.  You can measure the actual processes, and you come up with a 2.6, 2.5 measurement.  Or you look at the cost of quality, or should I say the cost of poor quality.  There is a direct correlation between the two, and as Doug had previously mentioned, we are all the time coming up with somewhere between 20 to 25 percent of period costs are costs that are related to poor quality, not only the costs for the QA and QC function, but it's the costs of rework, it's the costs of scrap, it's the costs of prevention appraisal, 25 percent.

     Obviously everybody here understands standard deviation and sigma, but just very briefly, obviously there's two aspects that you're trying to control.  One is the deviation around the particular process, between its upper and lower limits, so the spread.  And the second is the ability to maintain the particular process between those limits over time, so the shift.  So what you want to do is obviously keep the process with precision and with accuracy.

     Now, if we take all of these together, what does this mean for the pharmaceutical manufacturer.  Well, one point is, obviously it's unit costs.  Taking these four together, we can have a dramatic influence on bringing down unit costs.

     We start off by material costs.  Obviously the reduction in scrap and waste, I think Doug mentioned is between 10 and 15 percent, so you can immediately reduce that down to I think somewhere--2 percent is the max in terms of scrap and waste.

     With the reduction of non value-added activities, there is obviously a possibility to reduce the period cost.  Part of this is reducing those costs of compliance, so reducing the amount of internal and external failures, and then once you are confident with the high level of quality you're getting out because you know your processes and they are highly capable, you can then reduce the amount of money that you're actually spending on prevention and appraisal.

     On the other hand, your efficiency is going up because you're increasing the process yield due to the less amount of scrap, and your plant volumes can go up.  Why?  Well, if we look at the argument around non value activities, previously that process was taking 35 days.  It's now taking 5 days.  If you can absorb the capacity, you can now do five runs in the time it was taking you to do one run previously.

     So the net result of this, if you address these factors, is it's possible to significantly bring down the unit cost of production.

     Let's shoot forward to , I don't know, five years, to the company that is performing, that's outstanding, that's leading terms of pharmaceutical manufacturing.  They are operating at a 5 sigma level, so what does that mean?

     Well, their quality and compliance costs are down to somewhere between 3 and 5 percent.  The unit cost of production is 60 percent lower than the competitor's, who are still operating at 2.5 percent because they weren't at this presentation today.  Cycle time has gone down to 5 days, so obviously their productivity is up, their yield is up.

     And this point, again, newly introduced processes are effective immediately.  We saw again when Janet showed that slide with the number of supplements that were coming in, that's all the--the process has been transferred from development into manufacturing, and then we tweak it.  We get it a little bit better and a little bit better and a little bit better, and we blame those people in development because they didn't give us the right process.

     What we're saying is, if you can use process capability and 6 sigma concepts already in development, so that you know that you understand the process, and the process is capable and is able to stay within the limits that you have set, you can pass this then on to manufacturing, who then know that from day one that process is effective.  They can then do some fine-tuning, but you shouldn't be seeing all of these supplements because the fine-tuning is based on a statistically stable process which development and manufacturing understand.

     Key enablers to make this move?  First of all, I think I would really like to focus on process understanding, understanding what is happening in the process.  Second point is understanding the parameters that influence that process, those parameters that you are going to measure, those parameters that you are going to control to keep that process within the defined limits.  The process capability hurdle is already in development to ensure that the processes that manufacturing get are actually effective and capable from day one.  Obviously technology is a huge enabler in this area, as well.

     And just some examples, we'll hear more later from my colleagues:  Near infrared analysis of raw materials and in-process controls; continuous high-volume microwave sterilization; on-line measurement of variables, and supported by sigma tools, so you know exactly where your process is.

     With these technologies there is also a need for encompassing enterprise technologies.  So, for example, once you have all of these points, they need to be integrated into electronic batch records, so that again you're not wasting time waiting for the results to come back from QA or putting the whole batch record together manually.  There is no point of having these technologies in place and then taking the paper results and pasting them into your batch record.  And I won't even go near 21 C.F.R. 11 at this point.

     And then finally, obviously, electronic document management solutions to enable you to share information between development and manufacturing.  And, oh, yes, the thought of the manufacturing people being able to contribute early into the development of the CMC section, before it goes in for the submission, rather than achieving that as it's thrown over the wall from development and being forced to live with consequences of what happens in development.

     So what's the upside of this?  All of those elements contribute to something which we call the compliance infrastructure, the organization, the people, the procedures, the policies in place to ensure compliance with internal quality management systems.  Let's not forget that we're doing this to ensure the quality of the end product.  And the second point of compliance is with the regulators.

     What this can do is fundamentally shift the cost of compliance curve.  What we see, the blue curve is where it is today in terms of your typical 2 sigma performance.  By bringing in these new changes, you can move that curve down to 5 sigma performance, and you immediately get direct cost recovery but you also get a compliance gain.

     What does this mean for the industry?  Well, in a win-win situation the industry will be moving not only towards quality to meet the regulatory needs in terms of, if we take point 3, if that was the paradigm, then point 3 would be going straight up.  But here what we want to do is a balance the need for quality and the need for productivity.  So in the move of moving from 3 sigma to 6 sigma in terms of doing the right thing right, we get quality and we get productivity, so we end up with a win for the regulators, a win for the consumer, and a win for industry.

     So, to summarize, the industry needs to measure for productivity and not for the accountants.  The solution is not just a collection of technologies.  It's more complex than that,.  And there is a win-win.  The economics of compliance, where is the point on the cost of compliance where there is a win for the industry and a win for the regulators and the consumers?

     Thank you.

     DR. WOODCOCK:  Perhaps we can hold questions as we go through, if that's okay with the Board.  If you have specific questions, perhaps you would like to ask them now, or clarifications.  Otherwise, we're going to have a long discussion period.

     All right.  Thank you.  Now, the next speaker is Dr. Raju from MIT, and he is going to talk about pharmaceutical manufacturing from an academic perspective.

     DR. RAJU:  I'm going to try to follow up on a number of things.  What I'm going to try to summarize in the next half an hour or so is a set of research activities we have done within the MIT Pharmaceutical Manufacturing Initiative.  We are a technology university and a business university, and so we try to use some of those skills to try to understand the opportunities and communicate them to society, the regulators, and the companies who fund a large fraction of our research.

     The purpose of the Pharmaceutical Manufacturing Initiative is to describe and capture the opportunity to improve pharmaceutical manufacturing.  Let me see what that means.  What that means is, if this were the pharmaceutical industry, and the industry and the academicians are focused on many different aspects of it, we're going to choose to focus on this aspect of it.  And really it's very rare to focus on that aspect, and that's going to be the subject of my talk for the next few slides.

     This is something that has been the course of many years of working, and we have had a chance to work with almost every brand name company, a large number of biotech, and some branded generic companies as well, and so over the years we feel we have had a reasonable set of experience base that makes us feel that maybe we can begin to make some conclusions.

     We were very excited to start this manufacturing initiative at MIT, and if this was our goal, it seemed like a good place to start is to find out what pharmaceutical manufacturing really was.  So we went around to the vice presidents of manufacturing, to the companies, to society, to each other, and said "What is manufacturing, what is this piece here?"

     We were all excited.  What we saw was something very different.  What we saw was an organization that was told, "Don't be on the critical path."  "You're not as important as R&D."  "You're a cost."  "Don't stock out."  "Be sure you do it the same way you told us for the next 12 years, and show compliance, and we'll come every couple of years to your plant when we can to see how well you're doing."  And that's the investigators, as well.

     From a society, from the rest of the organization, there was a message that says, "Just don't screw up."  "You're a cost center."  And I can't think of anything worst for somebody to tell me, because it means I don't add value to society.  The vice president of manufacturing says, "How can I win?"  And the definition of a win almost seems nonexistent.

     The result of that shows up in the implementation of the technologies, and we'll show you some of the consequences and some of the opportunities of those technologies as we go forward.  But if this was pharmaceutical manufacturing, then either starting this initiative at MIT was something we shouldn't have done, or maybe this problem, so-called defensive mind set, is really the opportunity.

     And we said yes, that's what it is, let's try to understand why, let's try to understand the drivers of the system, let's get to the common things of science and technology, and then we'll change the results that come out of the system.  We then brought a significant number of vice presidents of manufacturing, who all said "How can I win?" and said "Let's all get together.  Let's begin to formulate a winning strategy."

     This is for the MIT Sloane School of Management, which has a very leading management of technology program, the Department of Chemical Engineering, and the Department of Industrial and Physical Pharmacy.  Although disputable, the Business School, Chemical Engineering School, and the School of Pharmacy are rated to be one of the best in their disciplines.

     And we said, here is a set of vice presidents who are trying to figure out how they can win.  Science we think should help society, and science could be a win-win situation.  Let's begin to listen to their concerns.  Somewhere along the way, since this is a regulated industry, we will have to figure out what we do with the FDA, but let's keep them in the equation.  Let's figure out what we want to do, what the opportunities are.  And somewhere along the way we'll start talking to the FDA, when we feel we are ready, but we have said we've got to do it sometime.  And I have been fortunate to be involved in all those three different disciplines through formal affiliations.

     If that was that tiny little box that we call pharmaceutical manufacturing, and we all want to analyze it together to understand why it looks the way it is, let's standardize on a few boxes around that box.  And if you think of pharmaceutical manufacturing as this 12 to 15-year process, sometimes more, sometimes less, over time when you have lab scale, pilot scale, and manufacturing scale work being done, and over space where you have the chemistry or the biology of the system done in the active ingredient, then the dominant physics around the components, and then the pill finished, and then the packaging which is the paper around it.

     And if we say let's look at this over time and see what decisions we make, and over space and begin to understand why, and figure out if there is a win-win, what is the role to transform pharmaceutical manufacturing, we said those are the boxes we're going to talk about for the next few slides.  What aspects of those boxes should we talk about today?  And we said, let's draw a big box around all those boxes and say, let's look at pharmaceutical manufacturing and discuss it in one of those dimensions of performance.

     Now when we talk about cost, that's going to be sensitive.  When we talk about quality, between the regulator and the regulated there are different perceptions of cost and quality and the reasons for them.  Safety is something that's universal.  I think everybody is doing very well there.

     And among these choices, I will decide to choose to talk about time because it means the same thing to all of us.  We all have the same watch.  It's neutral.  The cultural aspects, the accounting aspects, are something that we can deal with within our companies.  And so the rest of my talk, I'll focus on time, but a lot of my work involves cost and quality as well.  We can't do it all.  Let's focus on something that we can all have a win-win on.

     So let's take the simplest possible step, blending and blend uniformity as a central performance measure of that simple step that's supposed to take five minutes.  And being a chemical engineer who comes from the fermentation area, when I began to focus on this five-minute step, I was really disappointed because it's such a simple step and we worry so much about it.  I wanted to find out why and I wanted to find out what the opportunities were.

     So the focus is only on this tiny step as an example of the possibilities over time.  We look at blend process development from here to here and say, off-line, today's technology, versus on-line, tomorrow, day after tomorrow's technology.  Where is the win-win?  What is it?  What's stopping us now?  What can we do about it?

     First, it didn't seem like it was the technology itself.  We were fortunate, we had a professor from the Mechanical Engineering Department who said, "Here is light, here is lasers.  Let's shine them in through the blender, and I think through the window you can look at uniformity as well as that piece that you've been using for so many decades because somebody put their hands into the blender that way."

     So this is MIT's duct tape technology here, not necessarily FDA-approved, on a pilot plant, through a window looking at blend uniformity.  The decision is a very simple one.  The question is a simple one that says, "When are we done blending?" which is "When is the relative standard deviation or the signal at the end below some number?"  And the number is 6 percent, 5 percent, 4 percent, depending on where you draw the line.

     But if that was the simple question, the answer was not very disputable.  We could do it very reliably, depending on where you start, what the technology is.  LIF, one of our inventions, together with NIR, something that brought up together, and a lot more analytical technologies can do this very well.

     The technology was easy to develop, relatively--is pretty well developed today.  This is an opportunity for us to do something about it.  Now what?  We have something with duct tape in our labs and some of our plants.  Who are we going to start talking to?  What do we compare it against?

     And the obvious question that says, compare it against what you have already, and that is the thiefing assay that I told you about, these rods that are put into the blender and then samples taken out physically.  We know it's the thieving itself that drives the variability rather than the on-line sensor.

     And then we say, "Can you compare the on-line sensor with the off-line sensor for different active concentrations to determine end point?  It's clear that on average, you can.  It's clear that the new technology is less variable than the old technology.

     The question then is, do I still have to use that as a benchmark to prove my new technology, when it's the old measurement that's the problem?  If it is, then I'm going to have to collect a lot of data, not because it's a blend uniformity problem but because I'm comparing against an old measurement technology where the measurement is the problem.

     The question then is, how do we begin to formulate if this is the right strategy for us to go forward, or we should look at the process and the product uniformity, which is really what CGMP should be all about, and we all agree it is.

     So what?  We have a sensor, developed pretty quickly, that's mountable technologies that are similar, can do the same thing, may be less variable.  Let's figure out that it means something to all of us, as otherwise it's not worth a purely scientific exercise.

     Let's define cleaning. Cleaning, yes, it's that five-minute step, but it has a lot of steps before and after it.  If we want to understand "So what?" we want to understand what the consequences are on the two sides.

     Blending really has these different unit operations where you clean a blender, you load a blender, you mix, you sample, you transport to a lab, and then you have a result and a decision here.  You can be undermixed, mixed, or then you have to discard it if you're overmixed, and this is a new phenomenon that is quite well described within the industry, called segregation.  You can actually overmix something.

     The point being that while the physical process here of making something is here, the information process of measuring whether you have succeeded is far away in another functional department.  When the material process is disconnected from the physical process, what is the space between the two called?  It is called inventory.  That's the difference between the material process and the information process, and that's why we have all these inventory levels.

     So if we were to transform from on-line to off-line technology, we have to get the information and the material flow to be in the same place.  And so if we can get the on-line sensor onto the blender, on that simple operation, we can change the business processes that were talked about in the previous presentation, and we can put material information flow on the same place.  We can not only do that, but once we know when to stop the blend, we can start figuring out what to do about variability within the blends.

     And so we said, "This looks like it's important.  It looks like it could make a difference.  Let's try to figure out how big a difference it is, because the bigger the difference, more likely we would want to do it, more likely we are going to start communicating, more likely we're going to start sharing data."

     And so we said, "Let's capture all of those steps," the blending steps that I told you about, the cleaning, the charging of the active, the blending, the sampling, the QC, the decision, and the retrieving.  "Let's capture all those steps, collect data from all these companies."  We have a consortium, so we have a basis to collect data.  "Let's try to figure out how we do it today and then ask how can you do it tomorrow."

     This, for example, charge active can be modeled by each of its steps.  You clean the blender and then you load the blender.  You can then go inside the QC lab, and you can see you transport to the QC lab.  You hold in the QC lab, you retrieve in the lab.  You then prepare, test, and analyze.  And then you have the people in the lab, and they're all busy moving from here to there trying to do a measurement of product uniformity based on this old technology.

     Let's look at process development of just blending, now, just blend process development versus the old technology versus the new, and say, let's say that in that day, January 1, 2000, when we celebrated the new millennium, we continued, and we do, to use the old technology.  How long will it take for us to actually get a good blend, based on today's industrial practices?

     And so we start the blend process development process, and you can see the blending taking place, the sampling taking place, going to the QC lab, a decision being made.  Is it uniform?  Is it blended?  You then go to the lab and you have a different organization figuring out whether to approve it or not.

     The arrow there indicating we just got our first blend below the RSD specification, and we're so excited.  We look at our date, and it's the 4th of January in the morning.  We started on the 1st for this 5, 10 minute operation.  Now I've got one.  There's this thing that says, if you can get three in a row, you're all set for the next 12 years, or at least the interpretation of it that says-- one, I've got two.  All of a sudden we're not getting the third.  But I got the third.  It's now five days in.  I go into my lab and I look at what everybody is doing.

     Now you can look at what people are doing, and if it's red, it means that they're very busy, all your QC people.  Now they're really busy, and they're really busy moving the samples within the lab to figure out if it's uniform.  Because I'm not sure this one is going to pass, I'm going to take a few extra samples.

     Since most of our analysis is based on wet chemistry rather than something you can do in the process, and most of our products are solids, you have a whole bunch of wet chemistry based HPLC equipment, red, indicating that they're very busy.  We have a lot of busy equipment and people in the lab, and since we're not sure whether we're going to succeed, a lot of material information leads to a lot of samples waiting in the lab, right there.

     You can go down to our plants, not just the lab, and you can say, "What are the operators doing?"  Fairly busy, not fully busy, because they finished their part.  They're waiting for what to do next.  Blenders waiting to figure out what to do next, because they don't know the result.  The result of information is separated from the physical aspect of the actual making.  And what is the difference between the two called?  Inventory.  That's where the period costs, all the costs show up.  Now what do we do?

     Well, we say we've done three.  There's this argument that's a very common one.  If you look at the industrial average, 27 percent is our cost of goods sold, 73 percent as a result is our gross margin.  We've done three in a row.  We now want to ask the question, should I do a fourth?  Should I do a fifth?  That would become the trade-off of time versus cost versus quality--

     CHAIRMAN LANGER:  Can you use the microphone?

     DR. RAJU:  --versus cost versus quality.  The question, the common response to that is no.  I've done three.  I pretty much understand our process, given the technology that I have.  It's now five days forward, and I'm going to go forward to do this for the next 12 years.

     Now, during the next 12 years you have another argument.  It's not a time-to-market argument now, if you've been approved.  It's now a cost of changes and a cost of supplements argument, and there you have this so-called cost of supplements and paperwork argument that I'll talk to you later.  But before we go into manufacturing, let's stay in process development and do it right, and go back to the same date of January 2000 and see if we might learn something.

     So let's go back to January 1, 2000, the same date again, and say this is exactly the same, exactly the steps.  I'm not cheating here.  Charging the active, loading, and cleaning.  It's exactly the same date.  No practices around changed.  Nothing changed around cleaning, nothing changed around sampling.  The only thing that has changed is you're using the on-line sensor that you took a few months or a few years to discover, although there is a lot of research that was done before it.  Let's get started and see what kind of a trade-off we have across quality, time, and these competing alternatives that make our life difficult.

     So we start at the same time.  It's the 1st of January.  It's still the first day, and we just got our first batch.  A little bit more than a day, we finished two, changing no other practices.  One and a half days, we finished our three batches.

     Now, how much inventory do you think we have?  You've redesigned your process using technology that gave you cost, quality, and time all together.  What is everybody in the lab doing?  Let's look at the QC people, red indicating busy.  The question then becomes, "Oh, my God, do we need those people anymore?"


     And then you have the CEO, vice president, saying "What is the right head count?  Can you benchmark across the industry and tell me whether it should be 20 percent or 30 percent?"  And the opportunity here was, when you had the right technology, you didn't have to have that thing called QC.  The same people, instead of focusing on moving samples which is a doing job, maybe could think about the blend itself and the uniformity self, which is the process understanding job, and maybe they'll have more fun, and maybe you'll keep them a little longer, and maybe you can pay them a little more.

     You just got three.  You're not exactly, in removing some non value activities, potential for value-added activities, you have a whole bunch of equipment that was once busy and now is not necessarily busy.

     You now have another question.  You know, if it took you a day and a half to do three, maybe you can go back to those three and figure out, "You know, I'll have a higher standard."  You can do four and it's still two days.  Take three days, you can do five.  You can do six in a little bit more than two days.

     You now have done twice as many runs, you have done twice as much, maybe more, process understanding.  You now have a basis to say, "I'm going to do something I understand for the next 12 years, because I chose to open my eyes, to develop some basic technologies."  What would the regulators, to make them understood and communicated?  And while I think we all want together, because actually my uniformity may be better, my measurement is certainly better; I may be faster, I am; I might be cheaper, I am.  And it didn't seem like anybody in this overall societal framework lost along the way.

     That was product uniformity, and we can keep going forward on that.  Let's skip to the next step, because I want to give you another representation as well.

     If you were to look at old versus new, by looking at the business transformation of these processes based on technology, that means looking outward and deciding to work together.  We have fundamentally changed the performance measures, whether you measure it as cost, quality, or time.

     I said I'll focus on time because it's neutral.  The impacts are not 10 percent; it's 10 times in terms of blend process development time.  This is the old technology where you have one, two, or three blenders.

     Whether you do on-line technology versus off-line technology, it is not necessarily the factor of 10 improvement.  It's the variability reduction, because the predictability comes when you automated and understood and then automated your processes, and there are all on that sensor right there, instead of being in a human variability of deciding whether it's Monday, Tuesday or Wednesday, and whether I'm going to wait until Monday to move my samples to the lab.

     But there's an investment that has to take place across all of us to get this to happen.  This is an argument of the potential, that you catalyze us to all work together to make it happen.

     So this is not 10 percent here.  We're talking about 10 to 15 times improvement in manufacturing.  If CGMP is all about variability reduction, we get that certainly.  To become independent of some of the product and organization I think would allow for a lot of generalizability beyond.

     So we took a little piece called blending, it's five minutes, and we saw this huge opportunity.  That was the opportunity across time.  Let's look at the opportunity across space, and say let's look at routine manufacturing rather than process development, and say where is the time being spent on average, first?  And if we look at 6 sigma, it's about looking at below the average, but let's start with the average.  We have to simplify life.  We have to start with averages because that's one number, a summary of reality at that point.

     For the sake of simplicity I will not talk about the active ingredients.  Since we finished this example, I'll focus on this example, and we can have so many different products once we decide to look at horizontally, and if I choose just a high volume product you might say, "You know, you forgot the others," so we look at all the different representatives and look at where time is being spent.

     If you look at our processes, you'll find they look unbelievably similar.  If I go with the standard set of color coding, blue for the process steps and red for the QC testing steps, you'll find that we measure quality very much at the end.  Wow, it takes a lot of time.  Wow, it was pretty expensive.  We don't really necessarily measure significantly in-process tests in a few places.  Very rarely is there a feedback loop.  We often, if you don't feel happy about something, we throw it out.

     There is raw material testing.  While the steps here are all about the chemistry, and this is about the physics, most of our specifications are around the chemistry, even though we really do physics afterwards.  That's one argument.

     The other argument is saying, given this process flow, where is my time being spent?  And so let's draw that same process flow diagram in time, and you'll find here is where I'm spending my time, and here is where spending my time.  And the actual process, as you saw on the previous slide, takes a lot less time than the testing itself.

     Let's look at another process, so that we make sure that I'm representative, and this is the benefits of being in a consortium.  Looks very similar, testing.  Missing here, a lot at the end.  You then say, let me open that up in time, and then you see time, and you see 60 days for just the physical formulating part of it.  You see a lot of that in testing.  And then you expand that out and say, "Let's include the API, and let's look at the time from the beginning to the end."  And then you say, "This is close to half a year now."

     Now, if I was making potato chips and it took me about half a year from the beginning to the end, I don't think I would like what I get at the end.  But if it's tablets, I think clearly we have different kinds of time dimensions.

     But somewhere along the way there's an opportunity to look at time.  And half a year may not be necessarily the place we want to be, but let's look at the drivers behind the time.  And before that, let's make sure we look at complex and liquid products, so that you are convinced that we have looked beyond a few examples.

     Here are examples of the QC time, which is the process time, in a liquid line.  Here is the sterility test that totally dominates the time.  And here is the testing time versus the process time, and if you were to summarize them all, in red are the testing times, in blue are the process times, six examples.  Do they look similar or do they look dissimilar?  The reds look very similar.  The red looks very big.

     Now I began to wonder whether I should have called in the MIT Pharmaceutical Manufacturing Initiative or the MIT Pharmaceutical Testing Initiative, because it seemed like all the time was being spent in testing.  But is it just testing?  What are the drivers behind that testing is the next question, because we've got to understand why.

     And if you look at the drivers under that testing, you'll see in green is the process itself, and here is the actual test itself in blue.  But all the red are the manual transfers, the interruption of the process, the securing of the samples, the documentation, the transferring to the lab, the testing at the lab, so-called non value-added processes.

     So if we were to develop technologies, it doesn't matter whether it's LIF or NIR, what is it that matters?  If it's LIF or NIR, it would only impact this part of it.  What really matters about any technology that we would develop is the word "on-line" rather than the words "LIF" or "NIR", because those are the ones that drive many of these paper operations.  There's an investment we have to put in place to get there.  It seems like it's quite a doable task.  It seems like many of the technologies are very much in place.

     If you then broke that detail down further, you would find that the actual test itself takes 2 percent of the time, and it's the paper aspects of the preparation before and after that takes 98 percent of the time.  You've then got to say, what is the technology opportunity to deal with these, too?  And that will come along the way, but that is surely not the high leverage place at this time.

     If we were to now summarize what did we learn here, quality testing is discontinuous; testing times are large; testing times are more than the process times; and it's the off-line nature of the test times that drives the overall time.  The word "on-line" is a very important part of what we want to do.

     Obviously we want to look beyond the average, and that's where the learning is going to begin.  So let's look at those same processes that I just showed you, those six, and start looking at different aspects of the time rather than some average number of that time.  And you will find that all times are not created equal, and you will find that while you will think your cycle times were 25 days or 30 days, and that's the one I showed you, there is another cluster of batches here in different colors that are totally different from everything else.

     These are the so-called exception batches or variance batches.  And all of a sudden something happened, because you did just three runs, maybe, to some extent; because there's a paper trail, there's a lack of access to all information in the same place.  You now have to find out why you had that exception, and you have to have the same amount of rigor whether you're going to use it or throw it out.

     You now have to say why you got an exception, you've got to say which lots it happened to, why it happened and what is the probable cause, and there's a significant additional time taken for that.  And as I looked at that, I said, "Okay, it was the Pharmaceutical Manufacturing Initiative.  Maybe it should have been the Pharmaceutical Testing Initiative.  But maybe it really should have been the Exception Explanation Initiative, because the consequence of explanation initiatives shows up in the "So what?" category.

     If you look at all these times on average, and you look at how much of an impact does that make, you say the average cycle time is about 100 days; the cycle time, standard deviation, is 100 days, which says if you're talking about 6 sigma, that's a lot of cushion that you have to build in.  And exceptions increase variability by 50 percent, increase variability by 100 percent, on the average by 50 percent.

     So when you told the vice president of manufacturing, "Just don't screw up."  "Don't come in the way."  "Never have too little inventory."  "Don't stock up,"  what does he do?  He builds a plant about 10 times bigger than he needs, about two years earlier than he needs, and he deals with the consequences for the rest of the 12 years.  And as we measure the numbers, they show up in these places.

     We need to have a fundamental technology:  one, on-line sensors, but not only just LIF and NIR, but a way to look at these exceptions in a systematic way, and that's the next technology opportunity as well.

     Let's try to look at asset utilization numbers in a manufacturing sense, very similar to what I showed you, so that we can understand where the time is being spent.  Just like we said "So what?" around on-line technologies, we said "So what?" around routine manufacturing.

     Again, since we have a consortium of companies, we collect from the batch records all the detailed process steps, and we capture all the time spent in every batch for all the batches ever made, and we figure out how much time is spent by the operators versus testing, versus non valued versus value-added, and here is the first process that I showed you in terms of its time.

     We started this batch, and we say, "Let's start pharmaceutical manufacturing given a start date, we're now in the market, and see where is the time being spent."  And you can see the physical product, which is the round product right down here, and the paper product which is the square product going along with it, the so-called batch records.  And you can go inside any one of these steps and you can say, "Where is the time being spent?"

     And you can say very quickly the physical product goes through, the paper product takes a long time, because somebody has to sign, has to figure out that they did what they're supposed to do, and the technology opportunity there would be the electronic batch record opportunity.  That's number one.

     We can now go inside the lab and figure out what tests we do, and we can say here are the different tests we do, the appearance, color, fill weight, and these are very solid, nice tests that have been in place.  And we look at the bottom line drivers within those tests, such as the assay test, and you can find that it's the on-line nature of those tests that is going to impact those tests, and you can see that many of the paper and physical aspects of it drive today's performance.

     As I started this simulation in the beginning of June 1997 in this case, let me figure out where I am, because if I got a batch at the end, I would get a number bigger than zero.  It's about a month now, and I haven't got a batch out of the other end.  Let me try to figure out where those batches are.

     DR. WOODCOCK:  Can you clarify, these are real numbers?

     DR. RAJU:  Yes, these are real numbers.

     DR. WOODCOCK:  From a real manufacturing plant?

     DR. RAJU:  That's right, yes.  These are real numbers from the actual manufacturing plant.

     And you can see that we just got our first batch out and approved.  In the meantime we're collecting finished goods inventory at the end, based on the real numbers, and we're trying to figure out why is that inventory being held up there, and it's being held up and it's growing, and we're now going into July and it's a month since we started.

     And let's try to figure out why those batches are sitting there, and so let's go and figure out what's happening that might make them sit there.  And you would go inside, and you'll find that there is an exception here that's waiting to be decided on, and unless that information process can end up with a result, the physical processes are all waiting for the information process because they have been disconnected.

     There's a technology opportunity here that has not been addressed.  Why?  Because it's so difficult to talk about.  Exception represents what we didn't necessarily know, what happened that we didn't anticipate, what happened that may be different from what we thought should happen for the next 12 years, and now you have two multiple organizations getting together, saying, "What should I do?  Where should I sign?  Who should sign, and how long should it take?"

     And where the information is not necessarily there, it's difficult to do.  There are legal, social, political consequences.  What do we do?  We wait, and waiting costs money.  It costs money because I'm going to have to find out a few months later what I did and how to connect these two things, and that's the finished goods inventory, tracking up.

     And if you were to now say, "What is the finished goods inventory," you will see a huge number.  They are waiting for the decisions.  Not only are you waiting for the decision on that, but you're waiting for the batches soon after that, because you want to be confirmed about why there was an exception.

     So this would be the third technology, and the result of this is a capacity utilization that is extremely low.  I would say that many of the numbers that were discussed in the section before us are really not that far away from reality.

     We have to find a way to remember and agree that we really make two products:  a physical product which is a tablet or a capsule that has great cost benefit to society and is used for patients, and I think that's one of the greatest things that the pharmaceutical industry has ever done.  It's much better than all the other alternatives.

     But we make another product, a so-called documentation, information product, and that second product has its primary customer, we think, for the FDA, but it has a lot of information and a basis to look at exceptions.  And I think working together with the FDA around that technology, I think can fundamentally change pharmaceutical manufacturing as well.

     Coming to the end of my talk, I said we got together, said we want to find a way to win, and we had a large number of vice presidents who decided that there was a way out.  We've looked at technologies for all the aspects that I've told you about.  We've carried out different aspects of these technologies in different places for different products, in different parts of the product life cycle, and we've got some really exciting data, and we've called this initiative Continuous Quality Verification.

     And we say that we have many pieces of the puzzle that we think can become part of a transformation of this industry.  We have got an understanding of the needs.  We have some of the best universities in place.  We have presented this to the Division of Pharmaceutical Sciences and the Advisory Board.  We have got a very favorable response.  We have now talked within CDER to a number of people.  We have gotten really excited.

     If you remember, I showed you that slide that said, "We want to talk to the FDA at some time."  In the last three or four months we've been talking a lot, and we've been very impressed with the openness and the awareness and the good intentions of the people that we've talked to.  This is today.  This is the Science Board.  And as we go past this and we go forward, somewhere along the way we want to be able to also talk to the investigators who might be behind the curve in some of the new technologies.

     And this is where we're headed.  To summarize, we think technology, when you come back to science, understanding of the needs, we have put together a place where it could be a huge win for the industry, the FDA, and society.  But we can only capture this potential if we win together, and we really mean it.  And I think if we don't, we're all going to lose, and it's very, very likely that if we leave any one of the wins-wins out of the three wins, that we will be doing this and saying the same things 50 years from now.  Let's find a way to all win.

     Acknowledgements:  The consortium itself; two colleagues of mine, Professor Charles Cooney and Professor Steven Byrn.  And particularly relevant for a presentation such as this, these are my personal opinions and nobody is liable for them except me.  Thanks.

     DR. WOODCOCK:  Thank you very much.  I think I'll turn it back over to the Chair for a break.

     CHAIRMAN LANGER:  How long a break would you like?  10 minutes, 15?

     DR. WOODCOCK:  Ten minutes.

     CHAIRMAN LANGER:  Why don't we take a 10-minute break, then?

     DR. WOODCOCK:  Thank you.


     CHAIRMAN LANGER:  If everyone would be seated, we'll get started again.

     DR. WOODCOCK:  Our next speakers are from the industrial sector.  Dr. Norman Winskill and Dr. Steve Hammond are going to be talking about quality regulation from the pharmaceutical manufacturer's perspective.

     DR. WINSKILL:  Good morning, everyone, and thank you, Janet.  It's a pleasure to be here this morning to give you an industrial perspective on what I think is a very important and a very timely topic.

     As you can see from the slide and as Janet mentioned, we have a double act from Pfizer.  I'm Norman Winskill.  I'm going to be followed by Steve Hammond.  We're going to share the presentation between us.  One of us is a pharmaceutical technology expert and the other one isn't, and I'm the other one.


     I'm not interested in the technology per se.  I'm interested in what the technology can do for me.  So I'm going to try and explain a little bit of that, and Steve will concentrate on the technology itself, and then we'll come back together and see how we put the two together.

     So, just running through the order of what we'd like to cover--actually we have the wrong presentation up here, I think.  Do you have another presentation that we--

     DR. WOODCOCK:  A shorter one?

     DR. WINSKILL:  The title is right, but there was a long version and a shorter version.  Sorry, you'll have to give us a moment or two.  We have to switch computers.

     CHAIRMAN LANGER:  While this is happening, are there any questions anyone wants to ask?

     DR. WOODCOCK:  David Feigal had some information on recalls in the device sector that might be germane to this.

     DR. FEIGAL:  One of the questions that was asked before is, how many recalls are there?  And in the device area there are about 1,000 recalls a year, so if you figure there's approximately 200 business days in a year, that's about five recalls of products per day.

     There's about 80,000 products on the market, or 80,000 actually types of products on the market, so if you look at the number of products newly approved each year, which is sort of another sort of metric, there are about 7,000 products approved each year.  So it really isn't anything that approaches 6 sigma, if you do the math.

     Now, many of the recalls actually probably have more economic consequences for the company than public health impact.  About half of them are the lowest class recall, where there is something about the packaging or the labeling or some other type of issue that is a significant cost to the manufacturer but there's no health risk associated with the problem.

     But there have been some fairly important recalls that actually happened due to manufacturing problems in the device area this year.  Probably one that's still getting quite a bit of publication is the Salzer hip implant, which actually threatens the viability of that whole division of that company, which I think the company's theory still is that that was a problem with leaving a bit of residual oil on the surface of the hip implant so it didn't seat properly and it would loosen, and that has been a problem.

     But there also has been a worldwide recall of ceramic hips, which fractured when there was a change in the manufacturing method, in the type of firing and heating of the ceramic material.  So, although most of the recalls are in that low-risk category, there are important examples of products that are recalled where there really are not only quality problems but there are health implications for the patients, as well.

     DR. WOODCOCK:  Other questions?

     Are you about ready to go?

     DR. WINSKILL:  Yes, we're ready.  Sorry about the delay.

     So what we'd like to do in the next 25 minutes or so is give you a very brief history of the evolution of process analytic technology--I'll refer to it as PAT quite often--and also our vision for the future.  I'll then hand over to Steve, who will describe some of the specific applications that are of interest to us right now and how we might use those to improve our process knowledge and control of our processes.  I'll then come back and describe how we might introduce some of those technologies or how we might now introduce them, and what sort of environment we could create to make sure that we do introduce them appropriately and use them appropriately, and that's referred to as "the win-win scenario," and I'll describe what that is.

     So first a quick overview of the evolution of this technology within Pfizer.  A lot of the examples I will use are obviously taken from Pfizer.  I decided to use specific examples rather than hypotheticals because I think they illustrate the point.  I don't apologize for using Pfizer examples.  I think it is essential and probably necessary to see specifics, but I don't try to claim that what we are doing is anything different from what a lot of our colleagues in the industry are doing, and I think it's fairly representative.  But rather than hypotheticals, I decided to use specifics.

     We started looking a process analytical technology, particularly near infrared and mass spectroscopy, in the mid-'80s, early to mid-'80s, and we were looking at control of fermentation processes.  That proved very useful, and we quickly developed and applied the techniques to other processes, particularly near infrared.  So in the middle to late '80s we expanded the use of near infrared to synthesis operations, raw materials, packaging operations.

     And at this point the application in drug product manufacture, which is what we'll focus on mostly today, was really for a troubleshooting mode.  However, in using it for troubleshooting, we found it gave us an awful lot of information we didn't previously have and that conventional tests didn't have.

     So at the beginning of '90 we created a dedicated group--and we called it the NIR Group, and it was headed by Steve, who is coming up next--specifically to spread the word and to develop applications and put them into our processes to enhance process knowledge.  And so that dedicated group was formed.

     At that point in time it was very difficult to go and buy instruments off the shelf and apply them to the production plant, so a lot of what the group did was develop--not only work with vendors on the instrumentation, but work on a lot of the engineering solutions like sample presentation, automation, and robotics, and that was essential to enable us to put near infrared and other techniques into the drug product plants.  We did that in quite a big way in the early '90s.  I think I'll show you some of the applications that have ensued from that.

     Later, and probably for the last five or six years, I think, other techniques have emerged.  Near infrared is still important, but as others speak, as they have emphasized, it's not just near infrared.  It's not a panacea.  So LIF, mid-IR, acoustic, and a whole range of other, Raman techniques, are now being studied and they are increasingly being applied.

     So, given that evolution, where are we today?  And this is just a summary of some of the applications that we have in commercial use on our drug product plants around the world, and it's in chronological order, and you can see that we are using it actually in a commercial environment, everywhere from the beginning of the process, raw material testing and release, evaluation of packaging components, blending, tableting, encapsulation, tablet coating, packaged product.  We can actually scan tablets in a blister pack, not just to make sure that the tablets are present, but we run a spectrum on the tablet to make sure it's the right tablet in the right pack.  So quite an extensive use, and then at the end of the process we use different process analytical technologies to help with cleaning verification, to ensure everything is ready for the next step.

     There's a footnote at the bottom I think that has been referred to.  Janet referred to it at the beginning.  Interestingly, I think there are over 30 discreet applications in use around the world, very few in the U.S., less than 15 percent.

     And I think that's not atypical of novel technologies in general.  Process analytical technologies is a model, but I think if you take any of the new technologies we've looked at--microwave drying, automated guided vehicles, you name it--it tends to be evaluated and implemented and shaken down overseas, and it takes a long time before that technology is then brought back into the U.S. or used to make products for the U.S. market.  And I think a key question is, why is that?  Is that the right environment?  And if it's not, change it.  So we'll talk a little bit about that.

     So that was the current state of process analytical technology.  What does the future hold?  Now, this is obviously a personal vision, and the future for me is about 5 to 10 years in this example, but I think we will see a significant increase in the number of applications.  I think we will see a broadening of the type of applications--Raman, light-induced fluorescence, etcetera.  Acoustical is increasingly used to hear what's going on in the processes, gives a lot of information.

     I think what we will see, and what will help spread this technology throughout the industry, is the availability of off-the-shelf solutions from vendors.  Right now a lot of us have to develop our own engineering solutions, and go in and, like G.K. showed, adapt them onto blenders to use them.  I think that within five years we'll see them being offered by the equipment manufacturers as an option, and that will increase the utilization tremendously.

     The other thing I think we'll see, what I described on the previous slide was a lot of individual steps that are being controlled.  I think where we are going to is to see all of those steps integrated so we control the whole process.  Instead of doing conventional control up to one step, and then we have a nice process analytical technology on-line to control, for example, blending, and then we take it back into lab-based testing for the rest of the process, we'll integrate the process from cradle to grave, so it can operate at a fast cycle time with tremendous process knowledge which we don't have today.

     So our vision of the future is--and this is a pictorial representation of what others have described--moving away from discrete unit operations with laboratory-based testing at the end of each step.  And the reason we often wait for that laboratory testing is that if we proceed to the next step--which we can do, there is no regulatory reason why we have to wait for the result to proceed to the next step--but if that laboratory result comes back, and it's our only information today, if it comes back and says there is something wrong with the blend, it's not uniform, if we've taken it through to a tablet, there's a huge cost involved in having to go back and reprocess that, or if there's no rework option, throwing it away.  That's the scrap.  So it's risk management that forces the long cycle times and the discrete unit operations with lab-based testing at the end of each, not regulatory.

     Where we want to get to, and the vision for the future, is what I think G.K. called continuous process verification:  continuous, more frequent, more meaningful on-line analysis at every step of the process, so we can proceed to the next step of the process knowing that what we did before was compliant and of correct quality.  We don't have to wait six hours for a lab-based assay.  And that is the sort of manufacturing paradigm that we're trying to evolve to in the not-too-distant future.

     What are the challenges in getting there?  Some of them of course are technical.  However, I think the progress that we have seen, and Steve will describe a little bit, is sufficient that it really is not a significant barrier at this point in time.  Technical issues can be overcome.

     We have made considerable progress in the areas of chemometrics, robotics, the industrialization of instrumentation.  Yes, there are still some opportunities, and probably more significantly I think in the development of faster, smaller, cheaper instruments, so they can be put in more places more often, and probably still some work to be done on the sample interface, how the instrument interfaces with the sample, and how that can be an off-the-shelf solution.

     But I think there are solutions to those, and I think that's not a hinderance right now to the widespread application and moving towards the paradigm I described on the previous slide.  And maybe the major hurdle for the U.S. right now is the real or perceived regulatory hurdle, and maybe it is more perceived than real, and we'll come back to that at the end.

     At this point I'd like to hand over to Steve, who will describe some of the particular applications of interest, or the ones that we are particularly interested in, and then we will come back and talk about implementation.

     MR. HAMMOND:  Thank you, Norman.

     I just briefly want to go through three examples of where installing PAT, this is being driven, the latest advances in this are being driven by a new potent API that we're dealing with, and we've had to look a systems that are totally automated and work in a containment facility where just can't have plant operators even sampling blenders or even sampling off the tablet presses.  So we developed a system, and I'm going to start with on-line blending, we developed a system that uses a battery-powered radio communicating spectrometer.  It's very small, fast diode array instrument.  We actually mount this on the moving blender.  We control it and collect data from it remotely in another room.

     This is a schematic of the installation that we've just finished performing in our plant in Brooklyn, in New York.  The blender is contained in a separate room.  There are two containment barriers you have to go through to get into that room.  So we have the NIR mounted actually on the blender in a separate room, and our PC controlling that system is actually out in the corridor in this instance.  When this gets to a full manufacturing plant, there will actually be a containment area again for the blender, and the control of it will be in a specialist control room adjacent to that particular room.

     For this example I'm going to show you now, the point is that the PC driving the spectrometer and where the data processing is done is some 25 feet away from the blender in another room.  This is what the full GMP installation looks like, and you can see that there are two blue boxes actually mounted on the blender.  So everything that's back to the right-hand side of those two blue boxes is stationary.  What's to the left of the two blue boxes all rotates.

     The top box is actually the box that contains the battery and radio-communicating modems.  They are what is sending the spectra, once we have collected them, back out to the PC which is outside the room.  The bottom box, the bottom blue box, the smaller of the two, is actually the spectrometer.  It's a solid state instrument, so it can be put up with being spun round as the blender moves.

     The business end of this is actually the thing that looks like a black cylinder on the bottom of the bin, that's actually shining the infrared light through a window we put into the lid of the IBC, and it's collecting spectra when the bin is inverted.  We have some gravity switches that only fire the spectrometer when we know the blend has fallen down against the sapphire window mounted in the lid.  The spectrum is collected with the fiberoptic that goes from that reading head on the bottom of the bed back up to the spectrometer.

     Now, Norman talked a little bit about the design of the sample interfaces, and with this particular application it's very important, because what we need to do is to collect the spectrum from a known amount of material, and that amount of material must be something that is, in terms of unit dose, reasonable.  So we've done a lot of work in designing this reading head, that we collect the spectrum of between 200 and 300 milligrams of sample.  We've done a lot of work in looking at depth of penetration, density of the blends, and how much sample actually contributes to the spectrum.

     I've seen a lot of publications recently on doing on-line blend analysis using near infrared, but this fundamental thing of how much sample actually contributes to the spectrum is critical in getting these systems to work and give you realistic answers that you can match to off-line HPLC, and the design of this head allows us to do that.  We illuminate an area of some 3 centimeters, a circle 3 centimeters across, with the right intensity to get depth of penetration of about half a millimeter, and we know we collect information from the whole of that sample.  So it's very controlled in how many unit dose weights are we seeing.

     The sort of information that we're looking to get, the plot on the left shows you the near infrared spectrum of ingredients in a simulated blend that we used to commission this piece of equipment.  We couldn't actually use the active because it is a Class V material, so we substituted that with saccharin, which is innocuous but has the right sort of near infrared spectrum to compare to the active we would have used.

     What you can see here on the left is the spectra of those pure ingredients that we scanned before we started the exercise.  The change in pattern you see on the right is the movement in the spectrum of saccharin at a specific aromatic absorption for that molecule.  That is what we try to do, we find specific absorptions for these molecules and watch the movement at those specific absorptions, so we can track just that one ingredient.

     But we don't just focus on the active, we focus on every ingredient in the blend.  We look for the specific parts of the spectrum where the movements are really reflecting that ingredient in the blend.  So as we run the blender--and this is the first stage of the exercise that we did, this ran for 15 minutes--we can track the change in absorbance for each ingredient.

     And there I'm showing you the change in absorbance of saccharin, which was our active in this case, and lactose and Avicel, two other ingredients in that mixture.  So we can track this.  As the curve comes down to the bottom and we finally flatten out, we know we've reached the end point of blending, but we can watch the end point of blending for the active and for the other two ingredients in that blend.

     Now, to turn that into the normal sort of measurement that we would look to make on a blend, content uniformity, what we do is to take the spectra we collect in groups.  The blender was actually rotating at eight revolutions per minute, so what we've done is collect eight revolutions, or the spectra we collected from eight revolutions together, and then calculate a variance across those eight points.

     And this is mimicking taking eight samples from the blender into a laboratory, doing HPLC analysis on them, and calculating the content uniformity.  So this is a variance measure, so the Y axis is the variance across eight scans.  Along the bottom we're plotting time.  So what we can see is the movement in essentially content uniformity for the ingredients in that blend, but not just the active, all of the ingredients.

     There's one big advantage to this technology.  It is gaining more and more process understanding.  The other things that have been talked about, cycle time, are obviously of value, but one of the big attributes of this is the amounts of process understanding that you can get, and plotting the uniformity of all ingredients in a blend is one of the key gains in this sort of technology.

     And really to illustrate that, I want to show you the second step in our blending.  Once we had blended the main ingredients, we did the normal thing you would do, which is then to add a lubricant, and we blended that.  What you see here is the change in uniformity of the lubricant as it's added to the second stage of the blend.

     So with this system we can, in real time, watch the mixing of all the ingredients, look to see when the blending is done, and for the high potency--the product we have to make in a containment facility, we will develop specifications for the amounts of variation that we will allow in the spectra, and that will be validated against conventional HPLC measurements.

     The value to us in that on-line blending system, where we have a new product that must be made in a containment facility, the major benefit is no operator contact.  Robots will load the bins into the blending area.  the near infrared will be placed onto the bin using robotics.  Measurements of blend uniformity will be performed in that room, but the data will be transmitted into a control room, so we can avoid operator contact with that product altogether.

     There are other benefits.  There is no sampling, there is no sample thief error.  We get real-time information, which can help recycle times.  We get these multi-ingredient uniformity measurements.  We gain a lot in process understanding.  We can actually fingerprint the process.  We know that those curves I showed you, you can actually change them by the order in which you load the bin, so we can fingerprint even the way that you load the bin and what impact that has on blending.  And what this really comes down to in the end is the objective to go to "right first time" manufacturing.

     I just want to now show you the sorts of things that we're doing with tablet core analysis because, as Norman said, we're trying to look at cradle-to-grave control of the manufacturing process, and one of the key steps is obviously monitoring what you're doing when you're making or pressing tablet cores.

     This really started in an at-line situation in our manufacturing plant in Australia.  These people you see there are the plant operators, and they are people that have been using near infrared in that production plant to look at tablet cores and actually at-line, looking at blends as well.

     What I want to focus on is the fact that about once an hour those operators go to the tablet press and they take a handful of tablets.  They go to the near infrared and they test content uniformity and potency of those tablets.  They do that by passing near infrared through the tablet as a bulk measurement, which means that we do capture everything that's in that tablet and we're not subject to variation at the surface, which can be a problem in some measurements.  So we see everything there is to see in that tablet.

     Just to illustrate the information value of that, this is a product that was manufactured in the Australian plant, and the conventional analysis suggested there was a problem with blend segregation, maybe, during the process.  Using near infrared and looking at 300 tablets across that batch, rather than just 10 tablet as we would conventionally test, allowed us to pinpoint exactly at what point in that batch there was a problem, and then it became very simple to cure it because it was just a transfer chute that was causing some segregation in the blend.  But the extra information that you get from using these sorts of technologies to get analysis of 300 tablets a batch rather than just 10 or 20, really allows you to get to grips with that sort of issue very quickly.

     The at-line system I've shown you is fine for most of our products, but with this high potency product that we're going to introduce, we needed to take that further, and we've needed to automate that near infrared testing.  And what we've done now is to design this unit, which actually takes the conventional weight, thickness, and hardness modules that are very often at the side of a tablet press, and then introduces near infrared transmission capability into the unit as well.

     So tablets feed into this box, they are weighed, they are scanned on the near infrared, and then they go back to be measured for thickness and hardness.  And this is actually at the tablet press.  It's fully automated.  We're actually having two companies make one half each of this device.  Bruker are doing the near infrared side of this instrument, and a company called Schleuniger Pharmatron from Switzerland are making the other half of it.  But it is to be a totally integrated device.

     I just want to show you some of the spectra behind using a device like that.  This is actually spectra of this new high potency product that we have.  The black line that you can see there is a placebo tablet, and then the colored spectra are tablets of different strength of that product.  So you can see that we have specific information about the active if it's present in that product, and changes in concentration that we can actually measure from that spectral information.

     We can use that spectral information to compare HPLC values for single tablets against the value that we would get from the near infrared based on the spectral change that we see.  And what I'm showing you there is the correlation between spectral information and HPLC, but what I want you to note is the concentration in that product.  This is from .1 percent to 2 percent, so this measurement is extremely sensitive if it's set up correctly.

     I want to finish by describing some work that we've been doing introducing microscopy to look at pharmaceutical formulations.  What I'm talking about here is to look at a blend, but not as a bulk measurement, but actually to get in there and have a look at the matrix of the blend close up, and to do the same with a tablet, to get in and actually look at how each of the ingredients are lying alongside each other, and how do we actually make a tablet matrix.

     The way we do this is to take an area of a tablet, usually about 2 millimeters by 2 millimeters, we take each of the pure ingredients that we manufacture the tablet from, and we collect their spectrum and we file that into the computer, so we have the spectrum of each pure ingredient.

     And then that 2 millimeter area of the tablet, we divide it into squares, usually around 10 microns by 10 microns, and we use a microscope to collect a spectrum from a minute piece of that data matrix.  So each square is scanned in turn, each square of about 10 microns, and we collect a spectrum of that square using a microscope.  Then we match the spectrum that we get for each square against the spectrum of the pure ingredients.

     So what we can do is to take each square and color it in.  If we find the active, we usually color it red.  If we find Avicel, we'll usually color it blue.  Disintegrants, we usually color them green.  But we can build up a color map of the matrix of the tablet at a microscopic level.

     This is just one illustration of the sorts of information that you can get from doing that.  This is an example of two blends of the same product.  One blend would flow correctly into the encapsulation machine; the other blend would not flow correctly.  The microscopy information revealed that in fact our lubricant was clumped in the bad-flowing blend and nicely distributed in the well-flowing blend.

     In fact, it was interesting, the plant manager when we showed him this information said, "Yes, that's exactly what I thought it was."  But at least you can go back and get good scientific data on exactly what is causing that sort of problem, using microscopy.

     Here is another illustration of a product that occasionally suffers sticking problems on the tablet press.  We analyzed matrix using microscopy.  You can see there is a big difference in the way that that tablet matrix is actually sticking together.  And what I'm showing you here is the mixing of an inorganic diluent with one of the carbohydrates that goes into that formulation.

     In fact, what microscopy has shown us is, if those two ingredients mix together really well, we actually get a slightly weaker tablet that has a tendency to stick to the tablet presses.  In fact, you can track back and explain what that difference is.  It's a difference in the particle size of the sugar, the carbohydrate that's fed into the process.  By controlling that particle size well, you can avoid this problem, but only after you got the information to explain what the problem was could you go back and cure it, and microscopy really has an enormously powerful contribution to make to explaining process problems.

     Up to now, getting that sort of data has taken a long time.  Most of the maps I've shown you, our spectrometer and microscope have to work very hard for up to 24 hours to make those maps, because there are about 8,000 spectra in each of the maps.  But just recently imaging systems have started to appear that can actually collect the same information in about 10 minutes.

     We're hoping within a few years to get these systems so fast that we could take the spectrometer I showed you on the on-line blending system off, and actually put an imaging camera there in place, so we could image the blend as it's mixing.  And the sort of information that we should get from doing that should improve our process knowledge orders of magnitude beyond where it is at the moment.

     I'm now going to hand back to Norman.

     DR. WINSKILL:  I'll finish up very quickly here, but I hope you got a sense from what Steve has described, that we are quite excited by the additional information we can obtain on our manufacturing processes if we can get this technology into the plant routinely.  And we think we can, and we think it can be part of the vision I described earlier.

     Certainly the technical challenges I think we can overcome.  I think what might influence the speed at which it's rolled out and the general acceptability of the technology might be the real or perceived regulatory hurdles.  And history has taught us over these last 10 or 15 years generally about the introduction of technology, that it may not be as smooth as we would like to see it.

     In fact, I'm going to describe three possible scenarios, all real life examples that we've lived through.  One I will call the "don't use" scenario.  The "don't use" scenario is a worst case.

     These technologies are not used or developed during the product development basically because of our fear of delays in the regulatory approval.  We don't want to put novel technologies into an NDA.

     Once we've transferred a process with its controls into production, there may be a tendency not to want to "waste" resources to develop duplicate methodologies and controls when the existing ones work okay.

     And, quite frankly, there may be a concern on our behalf of raising the bar.  The more information that's available on a process could possibly be used inappropriately against us, and that's a genuine concern.

     The problem with that is, if that leads to the technology not being used, I think we all lose.  And there is a whole body of information that's just going to remain unavailable to anyone, and that's not a healthy situation.

     But again, I said I would use real-life examples.  This is a real-life example of "don't use."  It's taken from one of our recent products.  It's an antifungal polymorph. Conformation for this product was key to the product attributes.  During the development, we developed and looked at two different methodologies to conform, to confirm the polymorph, powder x-ray diffraction, obviously a well-established technique but not common on our manufacturing plants to QC labs, didn't exist at the site of manufacture.  So the only way to confirm the right polymorph was to send a sample 3,500 miles and then wait about a week for the result to come back.

     We developed an alternative, near infrared, common in the lab, available at the site.  We could get results within minutes, but it wasn't a standard technique for polymorph conformation.  Our initial draft of the NDA included both methods, but our fears and our conservatism made us take the near infrared out of the NDA because we were fearful of questions and delays, and so right now we have the method on the left, and we send samples across the Atlantic, and we don't use near infrared, not a very healthy situation.

     The second scenario is "don't tell."  Under this, we want the information so much, we use it but we don't register it and we don't openly talk about it.  So we have one set of methodologies that are in the files, and these are used for regulatory approval, and we conform to the specs, we conform to the dossiers.  But in addition to that, and in addition, not instead of, we use all these model techniques in parallel, and really we operate in two parallel universes.  We have a regulatory universe with old-fashioned conventional technologies.  We have another universe that really is the one that counts, but we are afraid to share it.

     Another real-life example, and this goes back to, I started life with Pfizer more than 25 years ago in the fermentation area.  That's where a lot of the near infrared came from.  On the left, you don't need to read that, it's an eye test, but on the left there's three or four registration specifications and control methods which are fairly conventional, lab-based assays, 12-hour turn round time, and that is still the case.

     Today that's the registration method, but over 20 years we've developed a whole set of advanced near infrared, mass spectroscopy, and probes, on-line probes that we really use to control the process.  And basically we, as I say, operate in a parallel universe.

     The conventional methods work.  They give product that will conform and is fit for its intended use.  There's no question about that, and then final end product testing is the gatekeeper to make sure of that.  But it's inefficient, and really the advanced control and the reason we are prepared to duplicate the universe is, we get much better batch-to-batch consistency, less impurities, fewer byproducts, less rework, etcetera, etcetera, all the advantages we talked about earlier.

     We could take this slide from this example and, I think, apply it to today's situation for drug product manufacture.  I think the universe for fermentation control has evolved significantly from a black box art 20 years ago, to a very highly controlled environment with dual networks and advanced computer control, using this information to give us assurance of quality.

     We're nowhere near there on drug product, but we could be.  And we have to find the right environment to get there, and I think if we could, that's the win-win situation we're talking about.  So that's a description of the win-win situation.  I mean, we don't need to do the parallel universe and the duplicate testing.

     What will it take to get there?  I think basically it will take an environment in which the methodology is understood and accepted by regulators and industry alike.  We have the same information, the same concerns.  We see the same opportunities around the use and application of this technology.

     We are certainly not there yet.  I think we're making a lot of progress, and I think today's meeting is a good example of that, but we have a little way to go.  And really what we're trying to do in this is, we're dealing with, we're removing the real or the perceived regulatory hurdles.

     And I think to do that, we need--and these are personal suggestions on how we can create that environment--I think joint forums to openly discuss the technology and openly discuss the issues and concerns and describe the technology, I think goes a long way.  And I know Dr. Hussain, Ajaz Hussain and others, have believed very strongly in this and are starting to do that, and that's encouraging.

     I think we need to create an effective process to evaluate these technologies, for example, PAT.  And part of that, I think, and maybe the root of it, is appropriate guidelines for the development, for the implementation and the validation of these methods, scientific-based guidelines that we can follow and we can understand, and then you can measure us against.  Absent that, it's down to personal interpretation, and that's where our perceived fears some into how it might be interpreted differently by different people.

     How can we do that?  Well, obviously we can sponsor joint forums, I would suggest industry/FDA forums, to work up some guidelines.  I think we have to recognize that process analytical technology is different from lab technology, and you have different expertise that need to be at the table to develop those guidelines, people from the process control, instrumentation side of the industry.

     Another suggestion is to participate in "dummy runs."  We have introduced a lot of these technologies.  We don't do so without appropriate internal controls for development and validation and implementation, and we have them.  We have SOPs for all of that.  Like I say, we don't share them because it's a parallel universe in most cases, but we would be willing to share them, and we would be willing to make some dummy submissions.  We will submit some methods that we've developed to see what you think of them.  We will submit the controls and the methodology and the SOPs we have used, to see what you think of those.

     Quite frankly, I call it "dummy" because we will submit things that are not linked to an NDA approval, so there is less risk for us, and probably that is a way to create a win-win situation.  If that helps to evolve to a set of guidelines that we can all understand quickly, then I think we'll be better off.

     And then finally I think what's important to us, probably to all of us, is consistent use of those guidelines not only by Center but by field investigators, and that will remove an additional concern that we may get approved but we may get additional questions and a different interpretation of the technology on an investigation.  And a set of guidelines that we can all--a bible, if you like--that tells us how to do it, that we can all refer to, and refer to the same chapters in the book, I think will go a long way to remove those perceived concepts.

     So, thank you.

     DR. WOODCOCK:  Thank you very much.  I appreciate Pfizer's willingness to come and talk about these things.

     The next speaker, who will speak fairly briefly, is Dr. Ajaz Hussain from the FDA, and he'll give the FDA's perspective and some ways that we perceive we could move forward on this, and then we'll try to save enough time for discussion and questions.

     DR. HUSSAIN:  Thanks, Janet.  I did have an extensive presentation, but to the time, I'm going to cut back.  But when I sort of put together that presentation, I thought I would have to defend an FDA position:  Why do we require product tests, and so forth?  But in many ways I think the case has been made by others, and I'll use an example to illustrate some of the challenges from and FDA perspective, and then follow up with a set of steps that we have taken and we are planning to take, and then pose the question Dr. Woodcock posed to you at the beginning of the presentation.

     One aspect which I just want to share with you is, why did everybody talk about blending?  It's mixing of powders.  I mean, it's at least a 150-year-old technology.  But for last 10 years we have been debating that, industry, FDA, so extensively, we probably have spent millions of dollars just talking about it in workshops and so forth.  That illustrates in my mind the state of the manufacturing today.  That's not the only unit operation.  There are a number of more complex unit operations that we have to deal with, but we are stuck on blending.  And so that is the situation from my perspective.

     Please pardon me.  I'm going to skip through some of the slides and get to the most important ones which I want to make some points on.  The original outline I had was to just redefine the emerging regulatory issues, share with you my perspective, FDA perspective, look at the problems, and see how we can proceed from here.

     The main issue here is that science and technology is progressing rapidly.  It is, in fact technology is not a problem right now.  I think getting it into practice is.

     Just to reemphasize, I think the discussion topic on process analytical technology, we use that as a model and initial focus point to facilitate discussion on emerging regulatory science issues in manufacturing in general, so that was a model.  People have talked about near infrared and other vibrational spectroscopy methods.  Again, as a case study, there are many different technologies, any different tools, and not discussing those here doesn't mean we are not considering those.

     I think one major issue I think in my mind is why is FDA leading this effort.  But when we started talking about this, the reactions that we received from industries, "You're going to do what?  This is not FDA's role."  But we felt it is, and think we have to take the lead.  If we don't do that, we get blamed for it.  I think the one aspect we keep hearing is, we are the hurdles, and I think our perspective is, we don't need to--we are not, and we don't want to be.  So how do we move forward?

     So just to summarize, we have heard before industry is hesitant to introduce process analytical technology in the U.S.  They have done it in Australia.  They have done it in other places.  It's in practice.  Not in the U.S.  The points that are made is regulatory uncertainty, risk.  That leads to "don't tell" or "don't use" practice.  I translate that into uncertainty or lack of understanding or knowledge of how FDA would assess that, as new technology leads to new questions.  These questions would be in method suitability, chemometrics.  This is status of pattern recognition and validation of that.

     The other concern we hear is, old product plus new technology leads to new regulatory concerns which could be added burden, so how to do you deal with that?  And clearly a mind set:  Why change?  One contributing factor to that is, this, when we bring it to FDA, will become an additional test.  We'll be asked to do the old method and the new method.

     And so those are some of the concerns that we kept hearing, and we said that's not how FDA operates.  We are more open to that.  Why is this perception out there?  And we started talking about this extensively.  Clearly we are approaching this from a public health perspective, and to ensure high efficiency of the U.S. pharmaceutical industry from many different views.

     Also I think I'm going to start skipping the slides.  The point here is, we hesitate to improve or learn about our process during new drug development because we don't have the time.  We don't do it after approval.

     So when is the right time for process improvement?  In some cases, never.  We have product, I'll give you an example from a 1997 warning letter.  This is a narrow therapeutic index drug which is used in a controlled release formulation.  How are we making it?  Just read this.

     XXX, drug XXX, "time release pellets are prepared by hand-coating powder...This manual process results in formation of agglomerates and in an accumulation of ingredients on the sides of the coating pan.  Operators sporadically scrape this undistributed material...manually breaking up agglomerates...and crushing them during processing."  This is, in some cases, the state of the art, not an example that can be generalized, but this is reality.

     Clearly, the point has been made that regulatory risk and uncertainty is a hurdle, and we have been working for last several years to remove those hurdles, and there are significant challenges.  I was going to talk about the guidances that we have already developed, but let me move on.

     The heart of the matter is science.  Where is the science in product development?  And clearly there are trends where we are going from dosage forms to drug delivery systems to more intelligent drug delivery systems.  That is happening, but that has to happen more quickly.

     The molecules that we are developing as drugs are more complex.  They need to be managed more carefully.  And so design of intelligent drug targeting systems and so forth is happening, but we are still stuck in a 100-year-old technology at the same time.  The principles of what we do originated 100 years ago in the art of compounding.  In many ways we are still--a lot of those things remain.  Most dosage forms are complex multi-factorial systems, yet we treat them as univariant or multi-incident systems where we study them one at a time.

     From an FDA perspective, when we have to establish classification, when we have to establish controls, what we face is a high degree of uncertainty on what the impact of independent variables have on performance.  So when you want to change something, we have no clue generally what that impact may be, so the additional tests come in.

     So not to belabor, not to just harp on that point, I just want to move on, but at very fundamental levels, material science, if you look at polymer science, if you look at all other fields, do we understand our materials?  Not necessarily.  In many cases the functional attributes of the materials we use, the ingredients that we mix in a tablet, are not well understood.

     The official monographs that we have are focused on chemical identity and purity, and that's probably what it should be.  Defining the functionality of an excipient in an official monograph is probably very difficult to do and probably not necessary to do, because when you mix powders, you lose their functionality and you have to really deal with the functionality of that powder mix.  So doing things on-line, doing analysis for that mix, is more relevant.

     So just I'm going to quickly focus on the current paradigm is testing to document quality, and predominantly with wet chemistry, and that case has been made.  But that's not what our FDA policies are.  In fact, it's to look at the guidances and, say, the GMP guideline.  These are the words that we use.  Quality cannot be tested into products, it needs to be built in.

     That's what we say, but we do focus on testing.  And the main reason for that is, if you want to build quality in, quality has to be built on knowledge, not data, and the level of sophistication and the details that our data can resolve is either medium to low.  So we are in the bottom of there, where you're looking at historical trial-and-error data to establish specification and so forth.  That is a contributing factor.

     We have talked about blending.  I'm going to quickly skip through this and say, why are we debating this?  What is this debate all about?  For 10 years we have debated this, and from an FDA perspective one could argue it's assuring quality.  From an industry perspective it's simply to document.  There is no quality problem we have to document, and we struggle to document that.

     But it is question of representative samples, and it is an indicator of art versus science debate, and is illustrative of test versus control mentality.  Blending assay that we do in process is actually a test.  You take a sample, test it.  If it's not homogeneous, and if it don't have a protocol for reprocessing, you would throw away that batch.  If it was a control, you would blend until it's homogeneous.

     Just to illustrate, from an FDA perspective, why we raise some issues here is, how do we control the quality of tablets right now?  Suppose you have two steps.  You have blending and making a tablet.  You would blend, take 10 samples or 6 samples, and if the percent RSD or standard deviation is less than 6 percent, it's homogeneous.  Then you would make your tablets.  And how many tablets do we test for content uniformity?  Ten.  If the batch is 10 million, 20 million, that defines what goes up.

     And we recently did some research in collaboration under a consortium that we formed, Product Quality Research Institute, and a major company in the industry who did this work for that thing, sent this data to me.  And this was a commercial, is a commercial product.  They actually did this test to support the research efforts that FDA is having, and found a problem in a commercial product.  And the problem was, those 10 tables were not really, truly indicative.  They had to go back and correct that process.

     I'm not going to get into how blending is done in chemical engineering.  I was planning to do that, but let's skip to certain advantages that we see moving towards PAT.  You are shifting the paradigm towards feedback control.  You are helping to build quality in by improved and more efficient control of raw materials.  You have process data that can be used for scale-up and modeling.  Adequacy of mix with respect to all critical components, and Steve Hammond made that point.

     And just to illustrate that point, content uniformity is one attribute.  Dissolution, drug release, is another attribute.  If we do 10 tablet testing for content uniformity for a 10 million batch, we do six tablets for dissolution.  Dissolution depends on a number of factors.  We do not require content uniformity for critical excipients.  You only do for drug.

     Here is an example.  The person who provided, again, is from a major company.  He is in the audience.  He didn't want to be named, so I'm not naming him here.  This is a situation where we would not even have tested for blend uniformity because the amount of drug is so high, and the tablet was failing, and was failing in dissolution as a function of time.

     So if you have 10 million, what happens early part of the run, late part of the run, you might miss that.  And new technology--this is from Steve Hammond--can address that.

     Something that I just wanted to point out, which is, our experience is slightly different from Steve Hammond.  We have been working in our labs with near infrared imaging, and we can actually do image analysis where you're looking at the chemical image, the grey and white spots, so each pixel of that has the complete spectra.  And actually we acquire that in less than a minute; he said 10 minutes.

     So we can actually look at a tablet, take a visual picture, and each pixel can give you full information, so non-destructive and so forth.  And you could see whether it's uniform or not within a minute, and you can distinguish whether it's not mixed properly and so forth.  But the technology is not an issue.  We can do this step.  That was the only point I was going to make here.  And it is a win-win opportunity from public health as well as industry.

     And one point that was raised was out of specification and recalls.  On the average, I think from a quality reason--not average--last year, the number is in my mind, I don't have an accurate number, we had about 150 recalls due to quality reasons on the drug side.  Now, there were more recalls for other reasons, packaging, labeling, and so forth.  But in my mind, the number 150 is in my mind, so I'm pretty sure, but that's the score.  The large percentage of out of specification recalls are for deviation to the physical attribute changes, and we right now are not focused on physics, we're focused on chemistry.

     One win-win situation from a public health perspective is, when somebody wants to go on-line to save money, time, and so forth, to do that you really have to understand your process well.  And that is a win-win.  You cannot just put something on-line and be happy with it.  So it does support development of more robust processes and a high level of process understanding is needed, and that's one win-win that we are.

     Let me just quickly go to what we are doing.  What should FDA do to facilitate introduction of PAT?  Clearly in my mind, in our mind right now is, eliminate regulatory uncertainty.  We have stated repeatedly the official FDA position:  FDA is a science-based organization.  FDA will accept new technology that is based on good science.  We have done that repeatedly.

     What we don't have right now are standards for PAT, for its suitability, validation, and a whole host of things.  In fact, you will have to approach this broadly, and every aspect of the process, including all of specification, how to deal with, has to be developed.  We don't have that.  In this regard we probably are lagging behind Australia and other countries, which is a bit unusual, for FDA to lag behind.

     What should we do?  Just to continue on that.  Define a clear, science-based regulatory process.  That we feel is important.  Current system is "adequate for intended use" would be one part of that.  We will have to think about a win-win scenario, and to do that, defining that the current system is adequate, it may not be as efficient as it can be.

     So if it allows introduction of new technology without becoming a requirement, we have to think about that.  So introduction of PAT, at least for some time, should not be a requirement, would be one approach.

     Define conditions under which PAT may replace current "regulatory release testing" is important.  Don't simply keep adding the number of tests and hope that helps.  You have to give something up, so you have to balance, based on what is needed, based on the redundancy that is required, balance the number of tests that are required.

     We have to develop a clear understanding of how to deal with invisible problems that are not visible today, but will become visible when you have process analytical technology.  We will have to have science-based review and inspection practices, and we will have to work towards international harmonization.

     Again, the point I'm making here is, generally FDA has led the way in those things.  Here, we might be following, and we need to catch up.

     So the challenges we have, limited institutional knowledge and experience, we have to work towards building that.  Seek input and collaboration, we feel that is the only way right now.

     What we have already accomplished is, we discussed this at the Advisory Committee for Pharmaceutical Science and got strong endorsement from that, and the committee actually recommended that we form a Subcommittee on Process Analytical Technology.  The Federal Register notice has been out.  End of November is the deadline to apply, and we encourage all of you who are here from industry to consider being part of that subcommittee.  It's an open process.  And we are going to define the objectives of that committee in terms of defining what the questions are for FDA.

     We also think we have to partner with industry, maybe with individual companies through a creator mechanism.  Clearly we are already linked to academic pharmaceutical engineering programs and process analytical chemistry programs.  We already have a consortium, PQRI, that we are using for that.

     So I'll leave it, stop my presentation with the questions Dr. Woodcock raised:  Are you able to support what we are trying to do?  What resources do you suggest FDA draw on?  And are there additional aspects to regulation of product quality that we should focus on?  Thank you.

     DR. WOODCOCK:  Thank you, and we'll open it, if that's all right with the Chair, we'll open this for discussion now.

     CHAIRMAN LANGER:  Yes.  I would love to get comments from, questions both from our Board as well as anyone in the audience, and particularly, you know, along these lines.  Bob?

     DR. NEREM:  I mean, number one, it seems almost like a no-brainer that this ought to move forward, because it seems like FDA has a mission.  Part of its mission is in fact to facilitate the use of advanced technology for the benefit of the American public.

     Having said that, what is it in the regulatory process of Australia or of other countries where they have been able to bring this on board, that makes it easier to bring it on board there than here?

     DR. WOODCOCK:  Yes.  Do you want to ask Pfizer?

     DR. NEREM:  Right.

     DR. WOODCOCK:  While you're coming up to the mike, let me give a stab at it.  I think many of the other countries have less extensively developed regulation in the manufacturing sector, frankly.  That makes it more difficult for us to change.

     DR. NEREM:  Does that suggest that we have too many regulations in the manufacturing sector?


     MR. HAMMOND:  The difference in Australia really was the attitude of the TGA, the regulatory body there.  They had an instant interest in the technology, to the point that they didn't just want to hear about it, they actually wanted to touch it.

     They came into the Pfizer plant, they brought staff, it was actually other companies, and they played with the equipment.  They had heard a lot about it, but actually wanted to see really what it could do, play with it themselves, and went away with their own conclusions about what it could do.

     And I think that was the difference, such an interest, and I have to say Ajaz is treading down the same path.  But that was it.  It was a real, "Let's get to know this, let's touch it, feel it, play with it."

     DR. NEREM:  Is Australia the only place where this has happened, or has it also happened in Europe?

     MR. HAMMOND:  It's happening very quickly now in the U.K.  The MCA, with a meeting we had with them in March, they basically said to us, "Well, what's your problem?  Why haven't you brought this to us?  What are you waiting for?  We like it."  So it's happening in a number of countries now.


     DR. DOYLE:  Are these technologies that you have developed in-house, are they proprietaries, so you don't want to share them with the rest of the industry?

     MR. HAMMOND:  No, it's the exact opposite.  In fact, we have developed these with commercial instrument companies, and the only way we can get those companies to develop these systems with us is to agree that they are available to anybody.  That system I have shown you is a part number for Zeiss.  If you go to Zeiss and say, "I want," I can't remember what the part number is, but that's what you'll get.  It's commercially available.

     DR. DOYLE:  Well, in the microbiology arena we use the, what, AOAC, the Association of Official Analytical Chemists, to run these, I guess you would say validation studies, to compare to the gold marker.  Couldn't something like this be developed?

     MR. HAMMOND:  Yes, I think it could, and in the U.K. we're doing-we're running a program with the London School of Pharmacy to take these technologies, particularly near infrared, and develop gold standard guidelines on how you would actually set them up and use them.  So I think that's a very good idea, yes.

     DR. WOODCOCK:  The Product Quality Research Institute, which is a foundation, a separate foundation, was set up partly to do this type of work, which is to collaborate amongst industrial, academic, and regulatory sectors in doing the scientific work to, you know, develop scientific understanding, partly for introduction of new technologies.  So there is already an existing mechanism, as Ajaz said, wherein if general kind of work needed to be done, that it could be, that would be generalizable.

     CHAIRMAN LANGER:  Other questions or comments?

     DR. PICKETT:  Yes, I just had a question.  You know, this is, I would agree it's almost a no-brainer to really try to get this implemented, and one of the issues that I was wondering if it is an issue, is whether or not within the agency, if there is the appropriate scientific expertise in the agency to really begin to address some of these newer technologies as they come on-line.

     DR. WOODCOCK:  Yes, that's one of the things we wanted to talk to the Board about, actually, because as Ajaz pointed out, much of the emphasis within pharmaceuticals for the last 100 years has been in the wet chemistry laboratory.  Much of this is in chemical engineering and mechanical type of sciences and technologies that need to be brought in.  And no, we don't have the range of expertise, neither in the field nor within the Center for Drugs, right now.

     CHAIRMAN LANGER:  Any comments from, we have lots of people in the audience?  Yes?

     DR. WOLD:  I am Svante Wold from Umetrics, Incorporated.  We are going to give a brief comment after lunch, but right now, I think that one way to get things rolling, we represent a technology that exists since many years, and one thing I wanted to say is that this technology, the pharmaceutical industry, interestingly, are far behind.

     Like the semiconductor industry that was mentioned, they applied this, exactly the same technology, and the semiconductor industry is very much a chemical process industry, which we don't understand what it is.  All steps of making chips and wafers and so forth are chemical.  So it's very much the same technology, the same instruments, near infrared is coming, and so forth, and it can just be lifted over.

     Now to be lifted over, I think one very interesting initiative would be if FDA and some drug industries agreed, let's set up a feasibility study on some existing processes where the traditional quality control works in a certain way.  What will happen if we now put on proper process analysis chemistry and proper multi-variate evaluation and see what happens?  And then we would all learn, and hopefully one would see that this is a win-win situation.

     CHAIRMAN LANGER:  Any other comments?  Maybe we could put those three questions up on the board for a second.  But I mean the first one basically is, should you go forward, so maybe just to get a consensus.  I think we sort of heard it's a no-brainer, but is that a fair sort of preamble to the questions?

     "Are we able to support the approach?"  So does anybody have any disagreement with that?  That's a no-brainer.  I think several people said that.

     How about the other two questions?  Are there any comments?  I mean I guess there were comments kind of made.  Are there any particular questions, Janet, that you want people to focus in on?

     DR. WOODCOCK:  Well, I would appreciate any ideas the members of the Board have about academic, other resources that you know of.  Obviously, we are prepared, as we said, to collaborate with the industrial sector on this, as well as the academic sector that we know about, but it strikes me there are many broad areas of expertise that need to be brought into this, as well as we need to hire some broader skill sets within the agency.

     And the other part is the additional aspects of regulation.  I mean, I think the question that was asked earlier about why haven't we adopted this and so forth, it's hard to recognize, I think, unless you are actually involved in this, what a large paradigm shift this will be for the method of regulation of product quality, the way it has been.

     And we plan to go about this by taking some examples, as the person who just spoke said, taking some pilots and so forth and moving forward on small pieces, but moving to this approach really does pose a lot of challenges for the FDA.  I don't want to underestimate that.  And I guess those are--

     DR. NEREM:  Challenges--

     DR. WOODCOCK:  Pardon me?

     DR. NEREM:  Challenges because of your mind set or what?

     DR. WOODCOCK:  Yes, I think that's a fair--well, it's really changing, yes, it's changing the philosophy or the paradigm, okay, from a testing paradigm to a reliance upon physical, chemical, on-line, and other types of trend methodologies, pattern recognition and so forth.  It's a very different paradigm.

     It's going to cause some disruption to the industry, too, because we're going to find out stuff, as Ajaz was saying, we're going to find out things about existing products.  There are existing products out there in market.  We know they have problems.  We know they fail their specifications intermittently.  We don't know why.  Now we're going to find out why.

     And so we're going to have a large range of issues that we're going to have to deal with as we go forward on this.  But if you all feel, and I see you have some thoughts on this, you ought to share them with us.

     CHAIRMAN LANGER:  Bob, did you want to share something?  Do you want to say something?

     DR. NEREM:  I want to let Alexa speak.  Then I'll share something.

     DR. CANADY:  As I listen to you, if I were an industry person, I'd be terrified by your attitude.  You know, I mean, in a sense of the concept that there's going to be tremendous dislocation.  And I guess to me the idea of a successful transition is the avoidance of that dislocation rather than the acceptance of it.

     DR. WOODCOCK:  Yes, that's a good additional aspect to keep in mind.  Obviously, to make this a win-win, we're going to have to avoid those consequences.

     DR. NEREM:  Yes.  I guess, you know, that last question--and obviously I'm not speaking as an industry person--but the word "additional" seems to me not to be the right modifier.  Because presumably, you know, if one does a zero base analysis of the process with new technology now in place, you will come up with different regulatory aspects which won't necessarily be additional regulatory aspects.

     DR. WOODCOCK:  Okay.  Well, I wasn't talking about adding regulatory aspects.  I was talking about what Dr. Canady was talking about.  What are the implications of this are we going to have to be careful about as we move forward?

     Obviously, if this is set up in a way that people perceive severe negative consequences from this, that's an additional aspect that we need to keep in mind.  I wasn't talking about should we add more testing.  That wasn't the meaning of the question.

     CHAIRMAN LANGER:  Owen, you wanted to say something?

     DR. FENNEMA:  Well, I'm a little puzzled about why there's so much concern about the difficulty of executing this kind of an advance.  It doesn't seem that difficult to me.  Maybe that's--maybe I'm naive about this.  I don't know.  But it doesn't seem that difficult to me, from FDA's standpoint, to adopt these kinds of new methodologies.

     What is needed, I would suggest, is simply a rather short document describing what FDA's expectations are when somebody comes forward with a petition proposing a new methodology.  You know, what kind of validation procedures they use, some data they have collected to show that this is effective and accurate and repetitive.  That is, to me, not a very difficult thing to do, and it should be done.

     CHAIRMAN LANGER:  Any other?  I'm just going to make--you know, when you mentioned on the academic thing, one thought that occurred to me was, you know, maybe to give some seminars at universities and chemical engineering schools, and certainly at MIT.  Maybe Dr. Raju and I could work on that.

     But there's also other schools that we might be able to do that.  I mean give, you know, a lecture and a seminar series, I think might, you know, get departments realizing that that would be useful, and maybe some students and post-docs seeing that.

     So maybe if you--I'll be happy to help--you could take some initiative to do that at MIT, and maybe Georgia Tech might, you know, and just different schools, there's a lot of chemical engineering departments that are around that might, I think, benefit from that.  So that would be a useful way of, you know, maybe trying to sow some seeds.

     DR. WOODCOCK:  Ajaz?

     DR. HUSSAIN:  No, I think we are very cognizant and we are actually working towards that right now, and--


     DR. HUSSAIN:  Well, at present, for example, right now I have a faculty appointment at Michigan and Purdue, and we are sort of downlinking University of Michigan pharmaceutical engineering seminars to FDA.  We are making presentations on this quite often now.

     CHAIRMAN LANGER:  Is this just pharmaceutical or--

     DR. HUSSAIN:  No, the School of Engineering has a pharmaceutical engineering program now.

     The point I want to make is, there is a transition.  Pharmacy schools have lost the focus in this area, because they--I came from pharmacy school, I was a teacher there--they moved towards clinical, and a hole got left behind.  And Rutgers, Michigan, have now a pharmaceutical engineering program in their engineering school.  So we are working with them to get our ideas and our needs expressed, so that as their curriculum develops, as their research programs develop, they keep that in mind.  So yes.

     CHAIRMAN LANGER:  I think that's good, but I think that some of the comments that people in the audience made, as well as here, is that there is a lot of work going on in, say, materials, you know, semiconductors, material science, chemical engineering.  And so somehow, again, I think people look for good problems, and this is a good problem.  So that may be a very different set of people that you also want to, you know, get acquainted with.

     I'm sorry, Bob.  Did you want to say something?

     DR. NEREM:  No, I would just simply, and you probably know chemical engineering better than I do, Bob, but it seems like a number of chemical engineering departments have initiated efforts in industrial bioprocessing.  I think that's becoming wider spread than many would think.

     CHAIRMAN LANGER:  That's right, yes, but I think that some of these particular things, I don't know that they are necessarily focusing on.  So I think it's actually a good point.  I think it's actually a natural thing that they would probably be quite interested in this.  Yes?

     DR. PICKETT:  Bob, just another question.  I mean, we haven't heard from some of the other division directors, but I would be curious whether or not there's any lessons to be learned here, because some of the other divisions like CBER certainly have receive innovative new products, have had to rapidly accommodate new technologies in order to release those products, and are there things that can be learned from other divisions that would be applicable here?


     DR. ZOON:  Again, I think a number of comments were made on the importance of an adequate science base for the agency and supporting the scientific underpinnings, to understand both from a process point of view and an analytical point of view the implementation of those processes into the biopharmaceutical field, for instance, which CBER primarily deals with.

     And one of the aspects is really trying to understand the technology early enough, and right now I think some of the areas that we're trying to focus on really deal with microarray and proteomics and looking at their eventual adaptation to processing of biopharmaceuticals in a way that has a quicker turn-around time and reliability and quantitativeness that will be utilized in the future.  So I think having the scientific underpinning within the agency is extremely important no matter what discipline in whatever area we have, and each of the different Centers can lead the way for their particular areas of expertise that they may have.

     One of the interesting thoughts in hearing the presentations this morning, though, which I see as maybe not so much an FDA willingness to deal with the change, it's actually how willing the broad cross-section of the pharmaceutical industry is in accepting this change, because I think people are in different places.  And certainly some of the older conventional products that have been around a long time, it will not be as easy for people to adapt the technology to those processes, or they may not want to make the investment.  And then it becomes, if it becomes the state of the art, then that becomes CGMP, and then how does that relate then to the standard for the industry across different product areas?

     I think it's an important discussion, and where we have the capability, I don't think FDA should be the stumbling block for this at all, but I think we do need to investigate in the broader cross-section of the industry where people are in this, and have an understanding.  And then how can we help get people into this field, to have better and more consistent products using the technology?

     DR. WOODCOCK:  Ajaz said this so quickly, I think, that people may have missed it, but I really think that one of the things we will have to do with introduction of this new technology is that it cannot become the new standard against which all else is judged for a very long time, to be fair, that that be a stumbling block.  If three people, three different firms or lines, production lines, have this technology, then that could not be considered the state of the art.

     And many of you are not aware of how this usually works, but there is a current Good Manufacturing Practices regulation, and one of those has to do with sort of continuous improvement of the basic standards for manufacturing, and we don't think that should be part of this early implementation.

     CHAIRMAN LANGER:  Any other comments or questions?

     DR. FEIGAL:  Could I comment about devices, just very quickly?


     DR. FEIGAL:  One of the interesting things to consider about devices is that some of the things you can do with pharmaceuticals, such as rely on pharmacology and pharmacokinetics because they are all drugs, you can't do that for devices.  They are such a heterogeneous group of products.  And so it's interesting to look at how the structure of the consumer protections were built around devices.

     As you are probably aware, the majority of devices are approved with the 510(k) mechanism, by which they show that they are substantially equivalent to another product.  There is no manufacturing section in a 510(k) application.  The kinds of manufacturing controls that they have to put in place do not get any type of pre-market clearance in that process.  There are still manufacturing standards and controls that are required, but they are to be established by the field at the time of the inspection.

     So it creates a very different environment that actually allows rapid changes in that kind of a sector, because there is no pre-market clearance, there is no manufacturing supplement, none of those types of features.  Now, there are times when that creates a problem, and we have a real concern about the products.

     There have been implants, for example, that have had sort of constantly changing design features in terms of thinness of material, method of casting, which plastics were used as cushions for--it was a weight-bearing implant.  And it was very hard to know, as we looked at failures of that implant, what we were dealing with, because there was no requirement for us to be told when all of the different kinds of changes were taking place.  And that is, it's actually one of the nuances of the device regulations, is when do you change it enough that you actually owe us another application because now it's a new device?

     The 60 or so devices that are novel enough to be approved under the PMA process have similar types of manufacturing requirements, but again because of the fact that devices are so different from each other, I think there probably is a climate where we are much more used to change.

     And like drugs, one of the things that Janet mentioned earlier--or I think you did, I can't remember if you did or if this was a discussion on the break with someone else--but one of the things that has happened is that many more things have moved towards no longer requiring pre-approval from us, but being things that they notify us change is being effected, or things moving into annual reports, or in our case we have something called real-time review which is used in a lot of the manufacturing, manufacturing changes.

     But it's actually one of the hardest things for us to know, is when a change enough that you actually should go back and learn something about the product again?  It's a big issue for biologics.  There are times when a subtle change actually has an unintended disastrous sort of effect.  And the hard judgment in science-based regulation is to say which of those make enough difference that you want to see those in advance, want to stop and think about those, versus what happens with many things, including most recalls, which is you discover a problem, you go back and figure out what caused it, and see if you can prevent it the next time around.

     CHAIRMAN LANGER:  Any others?

     DR. FENNEMA:  There are some questions in the back.

     MR. PARSONS:  John Parsons, and I represent Umetrics, but my background is 25 years in the industry from the commercial side.  And I would just like to, now that I'm not in the industry, make a comment I think that isn't addressed here.

     I think, as I listen to what Ajaz and the group here have presented, it's invaluable to the industry and to patient.  I think that's the key here, that we deliver the quality as a commitment to the patients, and obviously that's what the agency is all about.

     But from the commercial side I can tell you, as a member of an executive board, this kind of discussion from an investment standpoint and a risk standpoint just turns my stomach, because of the concerns that were expressed before.  It's a reengineering effort that has been described by Pfizer, by Norman and Steve, that has to be done through the process change and also all of the investment in terms of the equipment.  And it's also a risk I think that has been identified in terms of what will we find that we didn't know about the product before, particularly for the older products.

     I would just say this.  I would encourage the agency, as you move forward, that you do this as a cooperative effort with industry, and I am sure that you will do that, so that there is a transition period with the enforcement necessary to bring this to fruition, because it's absolutely necessary, but also where there is a cooperation, so the industry doesn't rise up and with the powers that are there, perhaps interfere with something that's necessary and that absolutely will benefit the patient in the long run.  Thank you.

     CHAIRMAN LANGER:  A comment back there?

     MR. ROY:  Suva Roy, Otsuka Maryland Research Institute.

     Having lived on both sides of the fence, so to speak, being in FDA, being now in industry, I don't think the regulatory hurdle is as big as people think it is.  There is the process of alternate controls that can be applied, and FDA doesn't even need to approve those things.  Perhaps what the FDA can do to change that, to make it a formal process, is allow the companies to submit supplement to the application, and allow that for approval which is not a current process that is entertained or used.

     And secondly, the other comment I wanted to make is that it is very heartening to see that now, after probably about 25 years after I had played with something, that something else is coming to fruition.  Back when I was working in industry, I had worked with, played with, literally with tablet compacts and acoustic vibrations to see if that tablet fractures.  However, back in the early '80s there was not enough computer power to do that quickly.  As a result, it was just an academic thing, but it is very interesting to see it coming through, and I really, really like to see this develop, and I commend Ajaz for bringing it to the attention.  Thank you.

     CHAIRMAN LANGER:  Any other comments?  Yes?

     MR. TURJAC:  I had to say something.  Emil Turjac.  I'm with Purdue Farmer right now, but I've taught and I've been a consultant, and I've been around even longer than the last gentleman, about 30 years.

     To our academic friends, I was there early enough, when we tried to introduce HPLC, and the FDA did not have instruments or people who knew how to run it.  And as a time-consuming thing, to keep stalling, it was "What's wrong with titrations?  They've worked for 40 years," until they could hire the people and get the material.

     Having done that, and most of the people of my genre are now directors or the like, and they're sitting back saying, "If we put something new in, it's going to delay our NDA, so let's just go to the USP.  We know it's better, and for alternatives we would have like laser light scattering for particle size as our alternate method, and for thermal analysis for melting rings as an alternate method, because God forbid we hold up, because we've got 15 years of our 17-year patent already shot.  We can't have that come back to us."

     So I think it's gun-shy.  The younger chemists and the younger FDA people are going, "What's wrong?"  But the people who make the decisions have been burned and they don't want to do it again.

     CHAIRMAN LANGER:  Any other comments?  Yes?

     DR. HUSSAIN:  Just to comment on the alternate approach, the alternate approach is fine.  I think that leads to the two parallel universes that Pfizer talked about.  You still have the old method that you have to do for regulatory compliance, and then you can have an alternate.  It doesn't solve the problem.  I think we really have to bring the two universes together.

     CHAIRMAN LANGER:  Any final comments before we eat?

     Well, I think that the consensus is certainly that everybody seems to think that you should go ahead with this

--oh, is there another, another one?

     MR. ROY:  I just wanted to add to Ajaz's comment, that if FDA has got a process of approving the alternate methods once they are mature enough and the company wants to do so, that solves the problem.  Then it solves the problem of parallel universes.  Unfortunately, that's not a process that is right now in place or is actively entertained.

     CHAIRMAN LANGER:  I think on that note we'll probably adjourn the session and meet back at 1:30, but I think hopefully the dialogue can obviously being continued, and people should feel free to give you feedback, and it will be great to hear in future sessions how this is going, but obviously it's very positive.

     DR. WOODCOCK:  We thank the Board for their advice.

     [Whereupon, at 12:45 p.m., the meeting recessed, to reconvene at 1:30 p.m. the same day.]


     CHAIRMAN LANGER:  If people could take their seats, we will get started.

     A number of people have requested to make comments, so the first one is Dr. Wold and Dr. Parsons and Dr. Josephson from Umetrics.  What I was going to do is ask each group to hold their comments to 5 or 10 minutes maximum, but if the first group would like to get started.

     DR. KETTANEH-WOLD:  I would like to talk a little bit about real quality control of batches as they are evolving, rather than doing what we have talked about, was one waits until a batch is finished, do some quality control, find that it's not up-to-date.  You cannot--the accepts are not found, no correction can be done, and you get only scrap.

     Instead, one can have real-time quality control.  How do we do that?  Well, first of all you have to have some infrastructure.  That is, on your batch you should be measuring on-line some adequate variable, adequate parameters, like temperature, pressure, whatever.

     We summarize that multivariately in a good way, and model the evolution of the batch, and once we have that, we have good representative set of batches.  You can make a fingerprint.  That is, you can have the average trace of good batches within three sigma limits.

     Once you have that, which is based on modeling the evolution and having this control chart, new batches as they are evolving in the real time are displayed inside this fingerprint, and you can see it.  If they go out of the limits, all you have to do is find out which variable is causing that.  You can just double click on the software and say, "Why is my batch going out?"  And you can make immediate correction.

     And not only that, but when the batch has reached 50 percent of the evolution, we can predict what the whole quality will be, and you can see this prediction changing as the batch is evolving.  And this is just based on multi-variate analysis and then taking the average and making control charts.

     And here I will show you a blending.  This was a pharmaceutical process of mixing, and you can see here the trace, the fingerprint.  This is the fingerprint.  All good batches should evolve right in this little, little interval.  The red line are the three sigma; the green line is the average trace, the golden batch, the average trace of the good batch, and these are two summaries of it.

     And you see now a new batch as it is evolving, and you see that there is first a starting phase, then there are levels that are changes of the variables, and then it showed evolving here.  This batch has started increasing the level much too early, and if we just double click on that, it tells us which variable has been increased way too fast, and then you can immediately correct and bring this batch back to make it evolve within the limits.

     And if you want to see what is the control chart of this variable, you can just double click on the variable.  You can see that for this variable it shows first there at 3,000 something here, and then raised the level, and they have raised the level way too early.  And this allows for correction immediately rather than when the batch is finished.

     And if you have a lot of these and you follow them, and you know that these batches stay within the limits, you are almost sure to have a good batch.  And this is very simple, it's visual, it's based on good science.  It uses multi-variate analysis to take all the variables in account, including their correlation.  It's like your Dow Jones that's a summary of our stock market.  It tries to do the best possible summary of the evolution, and it also takes in account all the raw material and all the initial conditions.

     So the benefits are enormous.  It brings the analysis of three-batch data to a simple framework.  It allows interface as the batch is evolving.  You can predict final quality.  And it applies to both the evolving batch and the whole batch.  The results are very easy to interpret, and it can facilitate compliance with regulation, because instead of sending numbers, just send the fingerprint.

     And one last comment on everything.  As somebody once said, that change is the practice complicated and frightening, but not changing is worse.  It's just that one has to manage the change with a transition.  Thank you.

     CHAIRMAN LANGER:  Are your colleagues talking, or just yourself?

     DR. KETTANEH-WOLD:  No, that is--

     CHAIRMAN LANGER:  Okay, great.  Any comments or questions?

     Okay, then we'll go on.  The next statement and comment will be by Gideon Kantor.

     DR. KANTOR:  The purpose of this talk is not for you to find out whether I'm ambidextrous.  By the way, you need to change gears because you are now going--excuse me, you have the page on that, what I'm going to cover, and it's a little bit of a different topic.  And what I'm talking about is enhanced regulation, regulatory science, for animal research.

     First I think I should kind of give you a little bit of a sketch of my qualifications.  Okay, what I'm going to talk about is first my qualifications, then I'll talk about the rationale for my proposal, then I'll talk about the enhancement considerations, and then finally one thing I always end up with is final comments.

     Okay, my qualifications.  I am a past president of the FDA Sigma Xi Chapter.  I would like to brag a little bit here.  I was the first CDRH president of the FDA Sigma Xi chapter many years before it reached popularity at CDRH.  And I am a member of the chapter now.

     In 1995 I retired as a research physicist from CDRN/OST, and since then I have regularly taught as Adjunct Associate Professor, Biomedical Engineering Department of Catholic University of America.  I have taught a course in neural stimulation in rehabilitation, and I also give some lectures on regulatory aspects.  I am trying to teach to the students that really want to go into biomedical engineering in a practical manner, if they don't like to get involved in regulatory aspects coming up as a part of it, they better change their field.

     And I am presently a member of the Institutional Animal Care and Use Committee, and I want to emphasize that my statement below is strictly my own.  I would like to reiterate that.  My statement is strictly my own.  Any members of the committee, in this particular case guilt by association does not apply.  But I am mentioning that to the membership of this committee to explain how I developed an interest in the regulatory science issue of enhanced animal research.

     Now let me talk about the rationale.  The unjustified death of a volunteer at Johns Hopkins Medical Center, and previously the unjustified death of a volunteer at the University of Pennsylvania Medical Center, are of great concern to me.  I would like to say why this is of concern to me.  I am a product of Hitler Germany, and I do know I exaggerate, but whenever I see what I consider clinical experimentation which is not carefully regulated, I get the jitters.  But this is strictly a personal problem and it's not meant to convince anybody else about the point of view that I have.

     And there are more cases that were discussed in the August 6, 2001 Newsweek issue on page 36-42, thanks to my wife pointing it out to me.  These are the reasons that I would like to propose here that possibly enhancing the role of regulatory science, animal research, might result in minimizing these types of fatalities.

     Before I become more specific, I would like to first refer to the FDA 2001 Science Forum that was concerned with establishing linkages between various scientific disciplines.  On the sideline, I would like to say I am very much interested in this whole question of linkages, and it is my opinion that a meta search engine that links different type of data banks with their keywords is essential for us to move into the 21st century and not feel inundated with information.

     Okay, excuse me.  I believe that there is another opportunity to establish new linkages between animal and human research, and by doing this possibly increasing safety of volunteers in clinical trials.  That is one thing that I noticed, I am very much clinically oriented in terms of a biomedical engineer, and one thing I noticed when I joined the IACUC committee is that that linkage was really not there, at least the way I saw it.  I did not notice any linkage.  People did their animal research but they didn't realize that it has some links with human research.  Again, this is my personal point of view.

     Based on my own research, I have gotten involved in discussion of global-local thinking and the information explosion, so I really felt pretty good.  Sometimes when you go on your own, you feel that you are really kind of crazy.  I still am probably kind of crazy, but at least here was an instance where I was not.

     Enhancement considerations.   Since I do not have specific data available pertaining to results in regulatory science animal research, all I can do is ask a few intuitive questions and then hope that the Board will have time to respond to them, and I'll be covering three topics:  IRBs, FDA reviews, and in-house animal research.

     Institutional Research Boards.  Could the Institutional Research Boards enhance their awareness of considering animal research in terms of trying to minimize fatalities in clinical trials?  And as I was re-reading this paper here, what came to my mind, maybe my immediate training in this respect might be purposeful.

     Number two, FDA reviews.  Could the awareness of FDA reviewers be enhanced in terms of assessing whether animal research is needed before human trials are approved?  Since I worked in FDA until 1995, I realize that the FDA reviewers are under terrific pressure to get things done, so this is here kind of a rhetorical question.  It seems to me that if this were done at all, then some more money must become available to project for this particular purpose.

     In-house animal research.  Could the FDA focus on doing animal research in support of clinical trials be enhanced.  Now, that I think is something that you can subconsciously develop, because you can always have ideas at night where you don't get paid for them and bring them in in the morning.  But it seems to me that if this is in the background of animal researchers, this idea that yes, that thing of those clinical trials, that kind of self-consciously it seems to me eventually they would go in that direction, but again this might mean some training.

     Final comments.  In support of my proposal of enhancing regulatory science animal research, I would like to refer to the statement of Dr. Schwetz in his introduction to the 2001 Forum:  "As we continue to enhance the science foundation of FDA, the effective training and retraining of our scientific and our medical personnel is among our highest priorities."

     And this really brings out not only the training at FDA which I greatly enjoyed while I was there, but also the outside world like hospitals and so forth.  Can there be maybe a foundation that has too much money, could spend, could start a grant in this direction?

     Accordingly, it might be useful--I kind of interrupted this, but based on the last Forum and the statement of Dr. Schwetz--it might be useful to have a session on enhancing regulatory science animal research at the 2002 FDA Science Forum, if possible, or instead later on at the 2003 Forum.

     I like to thank the committee for giving me the opportunity to present my very, very, very personal views.  Thank you very much.

     CHAIRMAN LANGER:  Thank you.  Any comments, questions?  Thank you very much.

     The next person to speak is Scott Ratzan, who is the editor of the Journal of Health Communication.

     DR. RATZAN:  Good afternoon.  Thank you, Mr. Chairman, committee, and everyone here for the opportunity to speak today.

     I wanted to sort of switch the discussion, which I think presages what Dr. Nerem is going to talk about in terms of some of the challenges subsequent to the public comment period on the need of how we communicate risk and how we make regulatory decisions, and the impact that it has upon the public.

     As you can see, I edit a journal, the peer review Journal of Health Communication, and I'm on the faculty at three different universities where I basically teach health communication, one in the School of Epidemiology and Public Health at Yale; at Tufts in the Department of Family Medicine and Community Health; and at George Washington in Public Health and Health Services.

     As I'm told here, I'm supposed to also disclose other conflicts.  I consult on a variety of areas, with the common denominator being communication.  I consult with about five different pharmaceutical companies, on and off, dealing with communication issues.  I'm on the study section for the Agency for Health Care Research and Quality, for communication.  I consult with about six other Federal Agencies, NIH, CDC.  I've done consultations with WHO and a variety of others.  I also have sat on two IOM committees, most recently in terms of communication with quality issues and health indicators for the country.

     With all that being said, nonetheless I'm representing myself today.  I'm not representing any of these other groups that I've spoken about.  And I'm trying to raise the debate to deal with ethical communication.

     There were busy Brueghels in the 15th, 16th century.  Our life is just as complicated today.  However, we try to explain a lot of things in two dimensions.  We try to explain things with one versus the other.

     So what I'm going to try to do is a 10-minute challenge for me, as Celeste had given me.  Background on communication.  Lessons that we have learned from BSE.  Thimerosal and vaccine risk.  And recent challenges that are very fresh in our mind, dealing with anthrax.  And some ideas that I think this committee could well consider and advance the public health.

     Our goal is really similar to what the World Health Organization has presaged many years ago:  "Informed opinion and active cooperation on the part of the public are of utmost importance in the improvement of the health of the people."  Clearly we embody that in the open process, and again, thanks for even speaking today.

     But how do many of us make decisions, whether we're on advisory committees, different committees, and so forth.  Sometimes we say the data speak for themselves.  Of course we know, we've already heard data don't speak.  There is data, information, knowledge, and wisdom.  We can explain the issue with statistical significance.  Can science always explain the areas?  We have progress that is incremental with evidence-based hypothesis testing, the 21st century approach.  And we still believe that scientific method can solve most dilemmas.

     But how does the public make decisions?  Very differently.  Many think the mouse is a little human, or even worse, the plural form of the word "anecdote" is evidence.  And this goes back to what I think James Fennimore Cooper first said in 1831:  "They say" is the monarch of the country.  It doesn't matter who says it, as long as the public believes "they" say it.  We have to think, who are they "they"?  Is it the media?  Is it us?

     And how ought we make decisions?  George Campbell reminded us, "Passion is the mover to action, but reason is the guide."  And we're all here doing reasoned, evidence-based, but I believe that evidence also means hard--and I'm not saying soft sciences, but the social sciences.  How we measure public opinion, as Walter Lippman had mentioned last century, in a whole variety of other ways.   That we're goal-driven, thinking of the public health and the people that we basically serve.  And then finally, we're using credible, trustworthy, understandable, and emotional and cultural sensitivity.

     So what does all this mean?  It means what we do here today, sound science and evidence.  We add value with deliberation, debate, and dialogue.  And, finally, we're involving a variety of different people in the process, ideally a partnership that I think other speakers have already mentioned today.

     But nonetheless, when committees often sit down, we have a sound science and evidence approach.  What is the strength of the evidence, and what are the scientific criteria for regulation?  Or, on the other axis, how high is the risk?   And if risk is so great, we have to regulate or do something specific.

     But there's been some changes, as you might know, over the last 47 years or so.  It was really just a randomized clinical trial that started in 1947.  I mean, it's not like that's been the gold standard for centuries.  And now we have new standards, a precautionary principle, which is the axis going down, of where we basically lower and have changed some of the burden of proof.

     And more recently, something that is very much of a concern to me is the Thimerosal issue, and we'll deal a little bit more with that in the presentation.  Because Thimerosal, a recent Institute of Medicine committee has issued a different measure of biological plausibility, and whether you agree with where I actually put that arrow, I think we need to think where is that red line and how are we going to be able to continue to do that committee after committee after committee that has different views.

     So how do we deal with uncertainty?  The precautionary principle has a variety of different definitions.  Some people take it down to "It's better to be safe than sorry."  Others are saying no, in advance of having complete scientific knowledge, theory testing, basically we're using foresight.

     But the changing paradigm has really challenged us.  It has challenged us that the burden of proof is shifting.  It used to be to make a change you would have to say why something is better than the status quo.  Now we have to, for the Thimerosal issue specifically, why is the status quo specifically bad?  Scientific uncertainly, and hence subjective evidence, is in the equation.  And, finally, scientists and regulators are requiring evidence now that still needs experimental design.

     What can we do?  I've been looking at the evidence based on this, not only with my journal, but also I edited a complete current bibliography of medicine with the National Library of Medicine on health risk communication, and there's 847 different articles we found.  Institute of Medicine has done about three reports already.  But still, nonetheless, we often don't follow the same science base that has been presaged by them, of how we deal with value judgments and how people make decisions about risk information.

     Vince Covello has done a lot of this work at Columbia in terms of his Center for Risk Communication, and what he shows us in a variety of different studies with other people is that trust and benefits are looked at as the most important elements.  These are not scientific elements of what the actual risk is, or probability based.  This is trust in the messenger.  Who is saying it, and what does it mean?  And we'll think about that just in terms of the anthrax issue that is all probably still fresh in many of our minds.  Myself, living in Washington, I still haven't got my mail that's quarantined.

     This is the process, what happens.  Policy-makers, experts, opinion leaders and the public are in partnership.  And it says here, "Upon further consideration, the evidence of fire is not as strong as first as it appeared.  We regret any confusion that we may have had, and will continue to," and you can see everybody stormed out of the theater.  This is my segue into BSE, vaccines, and anthrax, and some of the challenges that we face.

     I will very quickly go through the BSE issue, because I know I only have 10 minutes.  And on the back cover of a book that I edited, "The Mad Cow Crisis:  Health and the Public Good," I have these three quotes.

     "The biggest crisis the European Union ever had," according to Franz Fischler.

     "The worst crisis the British Government has had since the Falklands," said John Major.

     And "If one wanted to study the perils of imperfect policy-making, this case provides them all."

     I could probably update.  The book came out in 1998.  Last year, in 2000, the Frankfurter Allgemeine said BSE was going to be the Black Plague that had not hit Europe yet, and some people still say it's the AIDS crisis that the British never had.

     This is, however, a back report looking at it:  "There still is no scientific proof that BSE can be transmitted to man by beef, but this is seen by SEAC"--which is the Spongiform Encephalopathy Advisory Committee--"as the most likely explanation, and all our control measures are based on the assumption that it is."

     We have continued, and we could explore this everywhere from our TSE Blood Advisory Committees to a lot of different policies that have changed around the world, this has been a problem of where the scientific proof lies.  Above the line here the trade union officials and others are who people trust, and below the line are who people don't trust.  And as you can see, the government scientists, the business leaders, the politicians, the government ministers, and the journalists all have lost as part of this BSE issue.  This is in the U.K.  I don't have data in the United States.  We haven't measured it as such, like this.  It would be very interesting to see what this all means.

     Last, in terms of what the House of Lords report that came out last year reminded us, the government did not lie to the public about the BSE.  Government was doing their job.  They believed the risks posed were remote.  Confidence in government pronouncements about risk was a further casualty.  And of course it then says that this is actually affecting everything in the areas of science, including biotechnology and information technology.  And the British, later in the report they actually recommend a committee for the public understanding of science be integrated at the federal level, at the U.K. level, and then in different universities.

     I'm going to switch a little bit to Thimerosal, but I'm going to do this quickly because this is very data-rich and it is still, the jury is still out, and part of it is early.  You may recall that almost over two and a half years ago, Thimerosal, by part of an act it was released that it was a vaccine preservative that could have exceeded the actual mercury dosage.

     Needless to say, a lot of policy happened very quickly, our public health groups, our academies of pediatrics and so forth.  And the long and short of it, if you read through the basic pieces, are that Hepatitis B vaccinations, despite that there was vaccine on the market, the real experiment, shall we say, once policy was made, caused thousands of people, of children in this country, not to get vaccinated, and there's already some documented cases.  Particularly CDC has looked at four states, one state where there is a definite death and others that are still there.  And this is how it is still ongoing, and right now because of the uncertainty, each state by state is only slowly increasing to levels of Hepatitis vaccination that's necessary.  And as you see at the top, it could prevent as many as 5,000 deaths a year.

     And I'm rushing through this, unfortunately.

     CHAIRMAN LANGER:  It's about 10 minutes, if you could try to finish up.

     DR. RATZAN:  Okay.  I'll skip over the IOM stuff, and I just have three slides on anthrax.  Who does the public believe in health, in issues of health?  And if you  see, on the bottom are the people we usually hear from; and if you see at the top, people who people trust, remember that's science-based trust.

     And these were the messages that came out in the last three weeks.  Most of the messages that came out from politicians and government leaders are below the line in the mistrust area.

     This was what Scott Lillibridge said last week:  "We knew that communications would be important, but I don't think we knew it would be this dominant in the response."

     So what's the common denominator?  And thank you, Mr. Chairman, for another minute or so here.  Public health values need to be integrated.  How do we do that with societal risk quotient?  When we make decisions, how does it play out in society?  And if we don't consider communication, we're going to have a fulminating, unintended effect by our policy efforts.

     So these are the questions I have and things that we should think about.   If we embody on evidence-based approach by providing stringent scientific reassurance related to regulatory issues, we're not reassuring the public(s) that we serve.   That's really not scientific.  What we really need to do is think about value and trust driven, so we have a common rubric.  And I think that this, the field that I am in, is the most humane of the sciences and purest of the arts, and we try to balance both of those.

     So these are the five questions:  One, could every committee have an expert or someone who thinks about communication or public health, who could add the science of how this most likely will play out?  If we make this decision, what will happen?  And think about perceptions and practice.  There is a scientific approach to this.

     Secondly, does the status quo consumer representative that we have on the committees really represent the public, or are they more advocates representing a subgroup of that public?  That's an important question to think about.

     Can we develop a common metric?  And this is something I think is one of the biggest challenges, particularly with all the different advisory groups, not only here in Washington at FDA but all around, and CDC committees and so forth.  Can we try to have a rubric for biological plausibility and safety and precautionary principle?

     And the final questions:  Can we integrate societal risk, that risk is not just for the molecule, it's not just in the Petri dish, it's not just in the laboratory, it's where we all live.  And that's just not with vaccines, that's if we're thinking about how we may recall or limit certain drug use and then other potential morbidity and mortality to other drug use, on common over-the-counter drugs, for example.

     And then finally, could the FDA establish a coordinated communication program to inform the public and their intermediaries of how decisions are made under uncertainty, as well as the real risk of products?  And that, I think, is the more opportunity to develop what my research is in, is in health literacy and how we can develop a health literate public.

     And unfortunately 90 million Americans can't read or write on an ongoing basis, to understand the level of what we're putting out in terms of informed consent and a variety of different scientific pronouncements, and I think it's our job as well as the public's and others' job to really increase that communication and understanding.

     And I know it's tough to do all this in 10 minutes, and I appreciate the chairman and the committee and the public for hearing me.  Thank you.

     CHAIRMAN LANGER:  Thank you.  Any comments or questions?  Thank you.

     The next speaker will be Dr. Levin from Brimrose Corporation.

     DR. LEVIN:  Thank you very much for an opportunity.  As someone who spent like almost 16 years in a different industry, I wanted to bring a few things from my experience in another industry that can be useful, and then share with you some other information.

     My personal belief is that from our history, we will move in this direction for more real-time process control, more real-time nondestructive product testing, and biometric release.  I believe it's going to go there because I think it doesn't have an alternative.  The expected results will be high product quality and security, substantial cost reduction, benefits to shareholders, and benefits to the public, universal drivers.

     Can we learn from other industries?  I worked 16 years in the aircraft engine industry.  Both industries hold people's lives in their hands.  You fly up in the air, you are in their hands.  You take a drug, you are in their hands.  Both are federally regulated.  They are very competitive.  They drive their products with strict performance limits.  They are being chased by generic products behind them.  They are also driven to lower prices by government and by the users.

     Okay, very quickly, aircraft engine industry in the '70s.  Product testing was the major quality thrust, backed the product.  Pratt & Whitney, incoming house testing lab, 600 people.  Ratio of inspectors to operators, two to three, about.  Operators not responsible for the quality; the inspectors are.  Statistical process control, nonexistent.  Scrap factory, 20 to 30 percent of the total volume production.

     In the early '80s they made a change to active and statistical process control, became a major Total Quality Management.  Pratt & Whitney incoming, in-house testing, down to 100 people worldwide.  Ratio of inspectors, about one to eight.  This is a huge savings, cost reduction.  Operators now become responsible for the quality, not the inspector who is supposed to catch something.  Scrap factory down to 5 to 8 percent, it's going down.  Quality increased, engine failures became rare.  This week I really got scared, because what if this American Airlines was downed by an engine, but it was not, so I kept this slide in.


     But I really got scared.  Cost, dramatic reduction.  Typical cast blade, I was involved in this technology, is now 3 to 4 times cheaper for the cost but the quality is much higher.

     Where is the industry today?  I think I can skip this one because you were already told where we are today, so I don't need to say much about it.  Still product testing, on-line active process control is still minimal.  I don't know what the range of QC to operators are, maybe one to two, I don't know.  Statistical process control, I was told practically nonexistent.  I don't know what the scrap factory is, but you were told today it was about 20 percent, something like that.  Still too many recalls, 200 about a year.  Product uniformity still an unresolved issue.

     Can we learn?  Yes, we can, because the drivers are the same drivers for the change, the government and customer, pressure to reduce cost to operate, competition, performance, pressure to maintain profitability for shareholders, sense of "dead end" in doing more of the same.  This is a huge driver.  And the thing that today we see that there is a sense that if we keep on doing the same thing, more and more and more, it's a dead end.  And the FAA provided strong support, anticipating the benefits to the public.  Being a Federal agency, they thought about the public, and it's their duty.

     The conclusion that I draw from that is that the pharmaceutical industry will follow the same path taken by the aircraft engine industry in its bid to be more cost effective and, above all, more profitable.  After all, you got to make money; otherwise, nothing happens, and we know that.  So the facts are, same drivers, the FDA is providing now increasing support, realizing the benefits to the public, and we have Ajaz and other people, and the industry is getting again the sense of "dead end."

     How?  The question is, how we going to do that?  And there is more than one tool, and we have seen some of them, but I still think that NIR is probably the most significant, and as success will create new and improved tools, soon we will not be able to understand how we could do it otherwise.

     Why is NIR so important?  I think it is the only tool that really provides significant chemical and physical information on the bulk of the product, not just the surface.  It is the only method that penetrates tablets and capsules for complete characterization.  And it provides information on all ingredients, not just the active.  Is a tablet good if some excipient missing?  I think it's not.  If it was there to begin with, it's supposed to be there when I take it, so I like it to be there.

     Sorry, forgot to turn mine off.

     DR. NEREM:  That's your timer.


     DR. LEVIN:  No, no.  Sorry.  Forgot to turn mine off.

     Why is NIR, once very quickly, why is NIR so good?  Because it penetrates.  The light that goes into, either into reflectors, into powders made in a blender or any other operation, it penetrates to a depth between 2 to 4 millimeter.  It doesn't just scrape the surface.  That's why it is so important, being a tool that tells us enough information on the bulk of the product we are processing.

     Now I'm going to represent my company.  I don't own it.  Why then is NIR so important?  It is very fast, and people say today we need speed.  It is simply rugged, and it was tested dropping on the floor, kept on working.  It's a dual beam for real-time ratio, so we don't have to stop anything for taking some reference spectra from some reference sample to do some adjustment.  It is always dual beam, giving real-time ratio.

     It is full scanning, to provide information on all the ingredients, not just the active.  It comes with a full computer on board, so you can process as many algorithms as you may want.  And it's miniaturized to do blenders, battery operated, you will see immediately.  It is insensitive to ambient light variations, so it can operate in any ambience without any consideration to what's happening around.  And it is backed by 21 C.F.C. 11 tested from Brimrose.

     It's important because I think it's the only analyzer that can do them all from one company.  We can do incoming raw materials, fluidized beds, rotating blenders, tablets and capsules, various lyophilizers, spray dryers, blister pads, transdermal patches.  So you have a company that has one source for every possible application that you may want.  Why would you want to go and use application A from manufacturer A, application B from manufacturer C?  We will provide complete solutions for every need.

     I turned it off, I swear I did.

     CHAIRMAN LANGER:  Now I think it's getting to be the time.

     DR. LEVIN:  I turned it off, I swear.

     CHAIRMAN LANGER:  Dr. Nerem reactivated it.

     DR. LEVIN:  This is all the material I need, but we also have ability to connect it to a multiplexer, so you can have a multiplexer for doing more than one location with one analyzer.

     This is a typical installation in a fluidized bed.  You can see the globe, but we have done these fluidized beds not only with globes, we have done fluidized beds with what we call the free space, and we have an installation running now in New Jersey on that.

     This is our miniaturized spectrometer.  It has two batteries.  It's only 18 inches by 16 inches by about 4 inches.  It has got a complete computer on board, so you can process more than one algorithm.  It can be mounted on blender, it could be mounted on fluid bed dryers, it can be mounted against a bio line if you want to test bios, if you want to do reflectance on tablets.

     CHAIRMAN LANGER:  We're at 10 minutes.  Can you wrap it up in a minute?

     DR. LEVIN:  Yes, just one more slide.  And this has radio transmitter, this one has a radio transmitter to stop the blender.  So all the processing of the data is done on the spectrometer.  We don't need to transmit data.

     To the last one, I think.  This is a tablet analyzer doing final testing before shipping.  You can see again the whole spectrometry is in this case.  It is operating on 24 volts coming from this cable, and has internal cable to connect to a site computer, but during operations you don't need the site computer, because actually during operation you have a model, an algorithm that you use on the spectrometer for decision-making, and it links the, connects the decision to some other.  This connects automatically to a tablet press, and it sees the tablets from the press without the contact of hands, so it's automated and  can be stored in a remote room or anywhere else.

     I think that's the last one.

     CHAIRMAN LANGER:  Well, it's the next to the last one, but we're at 11 minutes.  Why don't we wrap it up?

     DR. LEVIN:  This is a typical multiplexer.  That's it.  Thank you.

     CHAIRMAN LANGER:  Any comments or questions?  Thank you.

     The next talk is by Robert Chisholm from AstraZeneca.

     MR. CHISHOLM:  Good afternoon, everyone.  My name is Bob Chisholm, International Technology Manager for Engineering Science and Technology for AstraZeneca, and I am based in the U.K.  First of all, I'd like to say how pleased I am to be back in the U.S.A., in Washington.  Unfortunately, it's only for one day, but it's very, very nice to be back here.

     What I would like to talk about I hope will supplement and complement some of the excellent presentations that we had this morning, and maybe help you answer some of the questions that I've heard posed.  I'll keep this down to 10 minutes, so I may speak very quickly in my Scottish language, so you may find that incredibly difficult to understand.  So if I need to slow down, tell me to slow down.

     What I want to talk about is TQMS, which is the AstraZeneca Total Quality Management Strategy in our facilities, and it's a very statistically based end process control with real-time quality assurance, and it's about a plant that we have designed and built in Germany, and I'll tell you all about that.

     What do we do just now in pharmaceuticals?  Basically, traditional QA means that we validate processes, the usual three batches, etcetera, etcetera, and then we run them under standard operating procedures, virtually no end process control for a long number of years.  We supplement this, of course, by testing a very small number of samples at the end of each batch, and that's our QA assurance, and typically that could be 10 samples out of a million, two million.  Taken in isolation, clearly that is not statistically significant.

     The way forward, I think, for future products, the way we would like to go, is TQMS, which you have heard a lot about, I think, already.  It's real-time end process monitoring and control which is being made continuous, and its real-time quality assurance which is statistically based throughout the batch.

     In our particular case that would be done automatically, but you could also do it at-line, so you're taking samples all the time.  So you actually have an increased testing frequency, and that increased testing frequency which is statistically based, provides you the platform to discuss with regulatory authorities so-called parametric release.  A term which I don't like, by the way, because we actually increase the testing, not decrease it.

     Okay, how do we do this?  Well, I think first of all in any pharmaceutical company you've got to have the sponsorship of your senior executive team or you won't succeed, because it is often harder to change your own company than it is to talk to any regulatory agency, believe me.  There's a lot vested interests in the companies.  This is a paradigm shift, and if you don't have the cooperation of your own board, you can forget it in the industry.

     We have created a process on what we call Technology Center of Excellence, which is based in Sweden, and they are looking at the whole range of methods, not just near infrared, and the product development using these methods.  But the thing I want to talk about today is related to the Center practice.  We have built a plant in Germany, in Plankstadt, which we sanctioned early '99, and we have integrated TQMS at this plant, and I'd just like to show you that.  So this does actually exist and goes live on December the 1st.

     Okay, what have we done?  Key process operations are now statistically in process controlled and monitored, so that identification of all raw materials in the dispensaries, things like control of fluid bed driers on line, continuous end line monitoring of blending similar to what Steve showed you earlier on--okay, and that's end point control of blending.  What that does, that's in process control.  That ensures that everything that you put into that tablet press is in spec and is the way you want, and the blend has been correct every time.

     We then have the tablet analyzer automatically within the tablet press, and again it's looking at the tablets which are going though check, so it is monitoring tablet quality throughout the batch.  That in itself is a big paradigm change for the industry.  We've also designed a 21 C.F.R. 11 data management system to go along with it, because with all this data on all these spectra, compliant data management is clearly of the essence.  And through that, we believe that we have real-time continuous quality assurance.

     This is the actual architecture, and it's so complicated on this small screen I can't see it, so I'm going to have to just talk to it up here.  If you look at this, you'll actually see that the analyzers we've used are by Brimrose, because of the OTF.  You'll see there are four Brimrose analyzers which run the plant.

     If you then look on each analyzer, you will see a panel PC and you'll see actually a bar code reader also on there, which is attached to each measurement. And these are managed by an NIR server on the top there, and all the data is reflected to what we call the PacMan server, which is a system for storing this data in a correct forum which was developed by our colleagues in Astra.

     So to give you an example, in the dispensary for instance, the operator would come along.  He would log onto his panel PC, using his password, because it's 21 C.F.R. 11. Having done that, he would then take the bar code reading, so the batch attributes, hence the panel PC, so it has the operator's name, all the batch attributes.

     The panel PC then contacts the NIR server, which enables the analyzer to do the correct measurement and puts any models, etcetera, down into the analyzer.  He takes the measurement, the panel PC gives him the result, and then all the data is automatically transferred from the analyzer via the NIR server into the PacMan system, so it's in there for inspection by regulatory authorities or anyone else, for that matter.

     The way we have designed this system, by the way, it's also capable of being inspected remotely through modem.  So someone could sit in Washington, connect to the modem, with our permission, of course, this is complete openness, and actually look at that plant as it's running to check the compliance, which could be a thing for the future.

     Okay, just moving along, you'll see we have our fluid bed drive analyzer there, which is a dryer end point control.  That's actually the one multiplexed analyzer.  We then have the blender analyzer, which sits in a base station, comes off the base station, is mounted onto an IBC which is spun, and looks through the sapphire window, tells you when the blend is finished, stops the blender.  When you put it back on the base station, the data, the spectra, automatically again transfer through the system into the PacMan system.  That's all for regulatory authorities or anyone else to inspect.

     Okay.  And, moving on to the tablet analyzers--I'm trying to watch the time here and keep this down--there are two tablet analyzers, two tablet presses.  We have only put one in at the moment because nobody really believes all this will work, and that's the problem with the industry, I think.  But I come from ICI Petrochemicals, and was very, very used to doing all this sort of work, so it's second nature to me and my team.

     Again, the tablet analyzer, it's using transmission and reflectance, and all the data is transferred up and stored in PacMan, the same as with the other systems.

     Okay, now I'm not too sure what other slides I've got in here.  I'll just try them.  Oh, yes, that's just a schematic, obviously, of the plant, and that's a solid dosage facility, which you'll all be aware of.  That's an example of an IBC with the blend monitoring unit actually mounted on it.  This only 120 kilograms.  That's why the IBC looks so small.  The whole system just spins together there.

     A tablet analyzer.  That particular one I think is the one in the lab.  You need to have them in the lab also, because you need to model, and it's much easier to model with it in the lab.  But it's connected up to use on that system just the same, and is the analyzer that actually sits in the tablet press.  And that's it actually in the tablet press, I bit difficult to see, I think, but the box here is actually the analyzer, the tablets coming off and going down to be analyzed as they pass.

     Okay, that's what I wanted to say to you.  As you see, that plant does exist, so the industry is moving forward.  In terms of the actual way forward, our intention would be, listening to this morning's presentations, our own intention would be to obviously start talking to regulatory authorities.  I'm a bit premature here.  The opportunity arose.

     We would chose one of our existing products that we make, which we have five years say operating experience of, so we have lots of means of comparison.  We would model with that.  It would be our intention then to run parallel dossiers, because we can run the plant with or without the system.  So we would run near infrared and run all the existing registered systems, and compile parallel dossiers, and then bring that to regulatory authorities who we should have been talking to, will talk to all the time, as a means of comparison.  So if you like that as the sort of pilot project that I think someone mentioned this morning, then that would be our intention.

     So that's all I have to say.  I've cut it down, obviously.  That's normally a one-hour presentation.  But hopefully I got enough over to let you know what we're doing.  Okay?  Thank you.

     CHAIRMAN LANGER:  Thank you.  Are there any questions or comments?  Any other comments from anyone here?  Yes?

     MR. WOLD:  Yes, Svante Wold from Umetrics.  I just want to emphasize one thing, and that is that we see a lot of very nice technologies, and that is essential.  You have to measure the right data.  But one also has to tie everything together.  You have to have, as in the last talk we heard, a data-based system, and you have to have tools to follow all these data.  Otherwise, it just becomes a data cemetery.  And there exists technology for that, too, and that is what we presented.  Thank you.

     CHAIRMAN LANGER:  Any other comments?

     Okay.  Well, thank you all very much.  We'll move on now, and we're going to get an update on the CDRH External Science Review.  At the end of this, we need to vote on acceptance of the report at the end of the presentation, and Dr. Nerem, who was Chair of this, will lead the discussion.  Bob?  This will be about a two-minute thing?


     DR. NEREM:  Actually I was going to use this time to sell coming to Georgia Tech to do your educational program.


     CHAIRMAN LANGER:  We'll make sure that's on the next Science Board agenda.

     DR. NEREM:  Well, it's a pleasure to be here and to represent the committee which produced this report, which we ended up titling "Science at Work in CDRH:  The Role of Science in the Regulatory Process."  The next slide actually shows the members of this committee.  I'm not going to go over it in all the detail, but I do want to recognize and introduce Alexa Canady, who was my Co-Chair and who is here to make sure that I'm honest, I think.  But it was an amazing committee, and I believe FDA and CDRH really owe a vote of thanks to all of these people who just performed in a marvelous way.

     In addition to a note of thanks to my committee members, I want to thank the CDRH staff members who worked closely with us.  Of course that was at a variety of levels.  There were people that really made it happen, and that includes Toni Marie Nearing who is sitting over here.  Toni, thanks for all you did.  I see Sandy Weininger out there, who helped--I was going to say helped write the report, in fact should be listed probably as a co-author, but he really was very neutral in his approach to what we did.  I see Mitch Shein back there.  And there's another person, I don't know if he is in the room.  Is Heini in the room somewhere?  I don't think Heini is here, but he was also extremely helpful through the entire process.

     And of course we want to thank the entire management of CDRH and the staff who participated in the internal/external reviews.  I think all told there must have been somewhere on the order of at least 150 people, and maybe there was more than that, David.  I don't know.

     Our objective was really to assess the quality of science across the organization and its relevance to the organization's regulatory mission.  We were not put together as a committee to evaluate the research going on within CDRH, but really to look at how science was playing a role and how it could play an even more effective role in regulatory decision-making in CDRH.

     This is an outline of the report.  I won't go over this in any detail.  I'll talk a little about process.  I'm going to come back to process at the end; talk about the findings under these three categories; our recommendations; and then some concluding comments.

     There was an internal review process.  I think one of the key issues was that it was decided early on by CDRH that electrostimulation devices would be chosen as the representative technology.  That is really where we focused most of our work, although we believe that in fact much of what we found can be generalized to other parts of CDRH, in fact can perhaps be generalized to all of FDA.

     The ground rules are indicated here, and I'm not going to go over that in any detail, but I do want to take this opportunity to commend CDRH for the substantive nature of the internal review and the spirit in which it was conducted.  I really believe that internal review, which was a time-consuming process for CDRH, provided the foundation for them to move forward as an organization and for us to come in as an outside group and get some insight into what could be done in the future.

     In terms of the external review process, I have already indicated that built on the knowledge provided by the internal review.  There were really three different meetings, all of which were important, in my opinion.  The first was a preparatory meeting held in Atlanta, where the External Review Committee came together and where we really sort of sorted out what we were about, and in fact we made some assignments at that time in terms of some of the case studies that we would be involved in.

     There was then the three-day review held in Rockville, July 24, 25, 26, and then finally a good part of the committee came together for a one-day report writing session on August 8th.  At the three-day meeting, we began to put together the outline of a report.  We made assignments for different sections of the report, but the report really came together at that August 8th meeting, and then we had a final draft this past month.

     In that July 24-26 review, there were the case studies; there were role-playing session, both for pre-IDE and post-IDE; there were omsbud reviews; industry interviews; and in fact international interviews.  The international interviews being, number one, we had Beth Pieterson from Health Canada on our team, so we could talk to her directly.  But we also were hooked up by video conference with David Jefferies in the U.K. to get a perspective on what was going on in Europe.

     Now, under scientific expertise we break down our findings into these different areas, and I'm just going to highlight some of the findings.  The complete report is available, as well as a copy of this presentation.  So let's move to the next slide, John.

     And in terms of the findings, I mean, to start with we certainly wanted to go on record as reaffirming that good science is critical to good regulatory decision-making.  Furthermore, as I think we all recognize, the complexity of applications requiring review has increased and will continue to do so.

     What was evident to us was, in general, the overall high quality of reviewers, medical officers, scientists, engineers.  Even so, the expertise across fields is uneven, and that's something I'll come back to.  We also felt that perhaps the level of expertise among staff about the clinical environment, at least in some cases, was limited.

     Continuing with the section on scientific expertise, we felt that there was not enough emphasis placed on the quality of decision-making as compared to the timeliness and volume of review, and I'll come back to that in the recommendations.  Furthermore, there appeared to be a strong tendency for the Office of Device Evaluation to operate primarily in-house, and as indicated there, we felt that was at least what was happening in fact, whether it was not by plan, but certainly that was our perception of the way day-to-day business was being conducted.

     We were very interested in learning about the use of third parties in other countries, for example, the notified bodies in Europe.  And as we look to the future, we have a concern as a committee whether CDRH or even FDA as a whole has the right expertise for the evaluation of combination products, those products that will be a combination of a device and a drug, a combination of a device and a biologic, or whatever.

     Moving to the next section of the report, which is the human resource issues, it's organized by these different categories, and let's go to the first slide of findings, John.

     Again, I want to note that we were impressed with the quality, professionalism, and dedication of the staff we encountered.  However, it's clear that there is a gap between the scientific expertise needed and the competencies of the current staff.  There also is a woefully inadequate investment of resources and providing of opportunities for staff training and development.  There are clearly too few staff to carry out the necessary activities as CDRH now functions.

     And for CDRH scientists, people who not take the track of management, it seemed to us that there was a lack of promotion opportunities, at least opportunities that could be somewhat rapidly taken advantage of.  The process apparently to be promoted as a scientist was a rather long, extensive one.

     Moving to the organizational and process issues, again the report is structured along these lines, and let me just say a few words about our findings.

     To start with, we characterized CDRH as an organization that was basically "semi-porous silos."  I suppose the good news is, they're semi-porous.  We're not quite sure how large the pores are.  But there needs to be attention paid to that.

     There also needs to be attention paid to metrics about quality, for as far as we could tell, there appeared to be no quality metrics about CDRH as an organization or even necessarily the decision-making process.  Certainly there seemed to be no system of retrospective measurement and analysis of specific CDRH decisions.

     Now, in the case study that took place as part of the internal review, in fact that's one of the things that took place.  There was some reflective looking at things, but that does not appear to happen on a regular basis.

     Also consistent with the semi-porous silos is the fact that there is no effective interoffice communication and coordination.  Furthermore, external experts are seldom used beyond those who sit on existing FDA advisory panels.

     And in the case of combination products, there is no clear pathway or guidelines for the regulation of these products.  There is really no single entry point for these products.

     So moving to the recommendations, recommendation number one is that CDRH needs to communicate, both internally and externally, a clear vision of the fundamental role of science in the regulatory process.

     Secondly, it really needs to rethink, in our opinion, how it carries out its mission, prioritizing its activities, outsourcing those functions it can, while still maintaining oversight, and reallocating its resources so as to expand its investment in science.  And as part of this, CDRH should examine its existing organizational structure as well as other regulatory models.

     As part of its restructuring of activities, recommendation three is that to enhance the fundamental role of science, CDRH should assess and reconsider the structure of the Office of Science and Technology, to focus that office on emerging science and technology.  This will probably require a separate review of OST, but we believe in fact that OST should be that part of CDRH that is really leading CDRH into the technologies of the 21st century, and that we believe requires some restructuring, but since we did not have a chance to look in depth at OST, we feel a separate review is in order.

     CDRH should develop a plan for enhancing cross-office and interagency communication and collaboration.

     The next two recommendations, five and six, really come to information technology, and that may be a problem for FDA as a whole.  Certainly there should be an electronic database for liaison functions and an internal and external expertise inventory.  Furthermore, we believe that CDRH should develop and implement a formal process for capturing institutional knowledge, so that when a decision is reached it does not remain in the mind of the reviewer.

     I think an important recommendation is that with the large staff turnover anticipated in the next five years, and in order to fill gaps in scientific expertise, CDRH should expeditiously perform an assessment of the current level and breadth of expertise so as to develop a long-term strategic staffing and recruitment plan.  As an organization, it really should be looking at where it needs to be five years from now, what kind of expertise is going to be required, and develop a staffing plan that is going to allow that to take place.

     There also needs to be the development of procedures and staff development opportunities to ensure that reviewer mandates for such issues as sample size or randomized trials are shaped by realistic clinical perspectives and relevant ethical considerations.

     Recommendation nine goes back to a comment I made earlier, but that is that CDRH needs to streamline processes that encourage scientific growth within the staff and provide for a more inviting career path and reward structure for scientific personnel, people who are not moving into management but are valuable as scientists within the organization.

     There also should be an encouragement and the facilitation of ODE using internal but non-ODE expertise, and also external expertise, including the development of policies that promote a more liberal use of external experts.

     As part of this, CDRH should expand its outreach to and scientific interactions with both industry and universities.

     The final three recommendations are that CDRH should develop a plan in collaboration with other Centers for the evaluation of combination products.  This plan in fact may require changes in organizational structure and operational procedures.

     Number thirteen and fourteen really go together, really relate to quality improvement.  Thirteen is really more at the regulatory decision-making level.  CDRH should implement a quality evaluation improvement program, and as part of this develop metrics for the assessment of quality as well as the timeliness of results.

     Fourteen is, at the organizational level CDRH  should implement a quality system with a focus on CDRH as an organization, and on development of activities that contribute to high quality decisions and the most productive use of resources.

     With this, let me say a few words about the process itself.  These may have a bias of my own.  David Feigal and his office actually sent out a survey form, and I think they are reasonably consistent, but this is my take on the process.

     I believe that the review, in focusing on the role of science in regulatory decision-making and not on scientific laboratory research, that that was the right focus, and we recommend it to the Science Board for use in future reviews.  The deliverable of an organization like this is not good research; the deliverable is good regulatory decision-making, and I think that needs to be the focus of these reviews.

     The internal self-study not only provided a foundation for the external review, but was a significant learning experience in its own right.  The external review, as I noted earlier, had three separate meetings, and I believe that each of these meetings was critical.  The pre-meeting in Atlanta really allowed the review team to get organized into how they were really going to conduct their work in a three-day period.  The three-day meeting in Rockville allowed us to carry out the review, and the final meeting on August 8th allowed us to complete a reasonable first draft of a report.

     In terms of the components of the process, I thought the case studies were important to our success, and also the fact that we assigned at the initial preparatory meeting small teams to investigate each case prior to the three-day review.

     The role-playing, my own view was that the role-playing was not as effective as it might have been.  I don't really feel like the committee ever got into the role-playing.  I don't know how Alexa feels, but I just didn't feel that we really put ourself in the roles we were supposed to.

     The on-the-spot reviews, that was basically where we could get any information on any kind of a review decision that had been made, and it was difficult in a way to intervene in that and get something.  At the same time, CDRH was offering us everything and anything, and just that gesture by itself was a very clear signal that they were open to us looking at any aspect of the operation, and I think that was an important signal.

     The industry interviews were important.  Unfortunately, they weren't all face-to-face, and I think in the future if this kind of model is used, it needs to be clear that these need to be face-to-face, these discussions with industry people.

     The international interviews, we thought that was quite useful, both having Beth Pieterson on the committee as well as the teleconference with David Jefferies.  The CDRH management and staff meetings are equally important, and we did, particularly in the case of the case studies, during each of those meetings we asked senior management to leave so we could meet with working staff without management in the room, and we believe that was important in terms of creating an environment where there would be a totally honest conversation.

     We also had a meeting with the union management.  I think that meeting could have been much more useful if it had been organized well in advance.  The fact of the matter is, it was only at the last minute that we asked for the meeting, and there wasn't the same preparation, both on our side as well as on the union management side, and so therefore I don't think it was as useful as it could have been.

     Some concluding comments.  I again want to commend CDRH for the dedication, integrity, and commitment to excellence exhibited by this effort.  In many ways CDRH is doing an excellent job.  Even so, with new products arising out of the biological revolution, with breakthrough technologies which will be increasingly complex, CDRH is facing a significant challenge.

     We felt that this review was conducted in the spirit of trying to be of constructive help to CDRH as it faced up to these challenges.  From the viewpoint of the committee, there clearly are changes necessary if CDRH is to significantly increase the role of science in regulatory decision-making.

     This slide really has what I think are, of all the recommendations, what I think are the three key things.  First, I really think there has to be a rethinking as to how the business is conducted.  Again, what do you do in-house, what do you farm out, what are the priorities, how do you get your hands around science and technology, which every day expands further and further.

     Secondly, as part of this, as part of this reinventing of CDRH is a reinventing of the staff through strategic recruitment, the continuous professional growth of existing staff, and policies that reward staff for the quality of scientific expertise.  And that goes back to really creating a long-term strategy for recruitment over the next five years.

     CDRH must reach out to external resources to create partnerships that will accelerate making new technologies available that are both safe and effective, and so as to enhance patient benefit in America.  No organization can have all the expertise, and I think CDRH needs to more and more use external expertise.

     Finally, the subcommittee review team appreciates the fact that these recommendations, even if accepted, cannot be put into place overnight, and certainly the way to go would be to incorporate these in some active way into the strategic plan of CDRH.

     I think that may be it.  Is there another slide?

     Okay.  Thank you for this opportunity to present the report, and again, thanks to everybody who worked with us.  And I don't know if you want to open it to questions or whether you want David to have a chance to--

     CHAIRMAN LANGER:  What do you prefer?

     DR. NEREM:  I'm easy.  But I'm glad you asked about Georgia Tech, Bob.


     CHAIRMAN LANGER:  Maybe I'll let David present.  There will be many questions about Georgia Tech later.

     DR. NEREM:  Well, particularly since we're trying to get David's son to come down to Georgia Tech and be a student.  Right, David?

     DR. FEIGAL:  Well, I need to begin by thanking Bob and Alexa and the other 10 members of the committee that joined them for the tremendous amount of time and the thoughtfulness of the effort that they put in, and I think you probably all appreciate how busy Bob is likely to be, but part of the reason we met in Atlanta for the kick-off meeting was, that seemed to be the only way to accommodate Bob's schedule, and we were interested enough in getting this to move along and get things, that we were happy to travel down there and begin with the orientation.

     This report comes at a very important time for us, because it's coming at a time when we have been working on a strategic plan to ask how do we meet the challenges of the future, and I have presented bits of that to the Science Board before.  Part of that is a vision that the Center has, that medical devices have a life cycle; that the whole life cycle is informative in the scientific decisions we made; that in fact it's a pipeline of multiple generations of products.

     There is a regulatory structure that surrounds that life cycle, but what we really were asking the team to do was to come in and look at this.  And what this is, it's the underlying science that we think is necessary to do science-based regulation at the different parts of the life cycle, for all the different regulatory tasks that we have. And our focus and our interest was not as much about asking how we got here or why we were the way we were, but really looking forward and saying how do we need to go from where we are now into the future, and I think that we appreciated the very constructive approach that the committee took in helping us think about that.

     Let me pause just for a second to show you, somebody had said something about the Center.  The room was pretty full this morning but it wasn't the same people.  Would everybody who works for the Center stand up?  This is a group that--you can sit down now--we are very, very interested in where we're going and the help we have getting there and the comments, and it's a process I think that, Bob and Alexa, you realize we take very seriously.

     As you pointed out, this began with a planning process that began in November of '99.  It was helpful to me as a new Center Director.  I had started about six months earlier than that.  Unfortunately, now I've been there long enough that many things are my fault.  If we had had these recommendations just as I arrived, I would have felt even better, but that's all right.  We'll move forward.

     And where we are today is near the bottom of this chart, and the important thing, part of what I want to show to you is our strategy for implementation.  But before I do that, I actually want to share some of the things that we did, that you alluded to, with our internal report process.  It will give you an opportunity to see whether or not we were on sort of some of the same tracks that you were, because I've taken the documents, unedited, that we provided to the committee when it first began to meet.

     As you pointed out, we really were interested in the scientific decision-making process.  We think that is our most fundamental product, our decisions.  In fact, if you hear complaints about us, it's that we haven't decided something, and then the second complaint is what we did decide.  But we're in the decision-making business, and it's important that we understand the impact, the resources that are required for this to be a science-based process, how well those decisions are integrated with all the different processes that we're responsible for, and how the organization learns from the way that it does its work, and our preparedness for future issues.

     I'm a little sorry that Dr. Skulnick isn't here, because you remember at the last meeting he said, "Why don't you give us the top 10 list of the best things about the Center and the worst things about the Center?"  We in fact actually provided a slight modification of that to the committee as a product of the internal review, and I'm presenting this as one way of summarizing some of the work of the internal review.

     We presented a top 10 list of the greatest challenges and problems for science-based regulation at CDRH, and then we also made recommendations of what we thought we had to do to address some of these.  And it may be interesting for you, having heard the External Committee's recommendation, to take a look and see how self-aware the Center was or was not about some of these issues.  In the notebooks I have presented these just on one page, and I've broken these out and organized them so that the challenge is met with our own recommendation.

     And so our first observation was that we're not always recognized as a science-based organization, sometimes not even by parts of our own structure in FDA.  It certainly is not a novel experience to have Health and Human Services organize a scientific group and leave us off.  When they organized the task force for the Biomedical Engineering Institute at the NIH, they put together a Public Health Service Advisory Board, and FDA, CDRH was not included in the PHS group that was to advise the NIH on the scientific needs.  Congress at times really is very--well, they're always very aware of the freight that we need to move, but they're not always as aware of the scientific basis of that, and at times that's true of industry as well.

     So where we began with our first challenge--and these are in rough priority order--was that we need to communicate our scientific vision and the scientific business for our regulatory actions.  It isn't adequate to simply say we're doing something because of precedent or level playing field or because we said so.  We need to make it clear that these are science-based.  We also need, and this is a request from you, we need advocates for our scientific role in medical devices and radiological health, and there are ways that you have been doing that.

     The second comment that we made when we were being self-critical of ourselves is that the Center leadership, meaning me and the senior people in the Center, do not always communicate science as a priority.  I think we're always quite clear about meeting performance deadlines in some of the goals, particularly the ones we report to Congress or with a trade press or an industry track.

     But we miss opportunities to create the resources and time for our scientists to have the training.  We don't create the expectation in our own staff that part of their job is to stay at the top of their game and stay current.  And the budget in our resource planning has often been reactive and short-term, and we need to walk the talk and show that science is really a priority to us.  Our recommendation sort of is the mirror of the observation.

     We are also very aware of the fact that the CDRH's scientific staff is graying.  This year we actually saw the retirement of the employee who was the longest working employee for the agency.  He has worked for the agency for 62 years, and I hope he's enjoying his retirement in Florida.  But one of the real challenges for us is that there are time when we go through waves of hiring and long periods without hiring, and that gives us waves of retirement, and this is both a challenge and an opportunity.

     And one of the elements of our strategic plan is one that we call Magnet for Excellence.  We borrowed that concept of being a magnet from the magnet school system.  We really want to be able to attract the type of employees that want to help us accomplish our public health mission.  And I really resonate very well, Bob, with your phrase "strategic recruitment."  We really need to not just think when we lose someone, even though that person was doing valuable work and had built up an in-box that now needs to be taken over and a specific area of expertise, we really need to look and say "What do we need now?"  That person was hired at a time when we needed that.  And we need to think about what the process is, because if we just backfill position by position by position, we will be configured the same way in five years that we are now, so we need to think about how we're going to do that.

     The budget policies of the last eight years markedly reduced our operating dollars, as we were absorbing the salary increases, and the good news that you heard last night from Jeff Weber is that in this year's budget in fact we don't have to absorb 4.6 percent of our staff in order to pay for the appreciated pay raise, but it's even more appreciated when they give us the money for it.

     But I think the concept for us is that it really doesn't matter if we're rich or poor, we have to have the same scientific values and the same approach to scientific problems, whether it's a year where we have some budget flexibility or some budget challenges.  And we need to really look at how to take care of our existing employees to make them as effective as they can.

     There was a very nice comment by one of the members of the Science Board who couldn't be here today.  Earlier this week there was a meeting at the University of Maryland.  I think they beat Georgia Tech, didn't they this year, Bob?  But there's always the basketball season.  We'll see how this goes.

     DR. NEREM:  We're even better in basketball.

     DR. FEIGAL:  But the comment that was made is that art is "I" and science is "we", and we really train people almost as artisans, as apprentices.  They work with a small team.  They learn what that team does, how it works.  We really need to take the strength of the scientific method, which is really a group process, a process where everyone learns from each other, we need to identify much more systematically, particularly with employees whose jobs are changing, to identify the core competencies and the type of experiences that will develop them as scientists and create flexibility in our scientific work force to meet future challenges.

     Our current system actually tends to have a system where people almost need to burrow in to get promoted.  If you're not going to be a supervisor, then you need to become an expert, and an expert often is someone who--it's more of that "I" model, where you are the expert.  You are the one that has the knowledge and doesn't share it.  And one of the things that has happened as part of the strategic plan and part of our grappling with this, is that we have actually created and gotten approved a program called the Master Reviewer that supplements the expert path, that rewards breadth, and a different type of experience for promotion.  It's a program that Janet Woodcock had in CDER, and then we have actually crafted our own version of it which is just now being launched.

     There is the very real fact that premarket deadlines, acute problems, squeaky wheels, meaning any type of, sort of contentious situation, often dominate resource allocation in a way that can leave programs disconnected and sometimes out of balance.  One of the hardest things for us to figure out is, what's the right size for different parts of the unit, because everyone is busy and everyone could do more with more resources.  And are we just putting it where Congress squeaks or where a group of manufacturers create a lot of public attention?

     We need to have our own vision of sort of the public health mission, and be able to balance and prioritize even through that.  Even though we must meet these deadlines and must deal with these problems as they come up, we need to more deliberately prioritize our work proactively, rather than just being reactive.

     Scientific communication opportunities are under-utilized, whether this is with our scientific peers, whether it's medical device users or the general public, and this hides what we know.  It hides the knowledge that we in fact manage, and limits our mission effectiveness.  And so one of our real goals is to understand better.  I really appreciated the earlier public presentation on risk management and risk communication.  That's something we think a lot about.  And the public's hunger for knowledge is illustrated by the fact that nearly a million people will read the Lasik web site this year.

     We solve many problems too slowly in a rapidly changing world.  Some of our decision-making is timely, particularly the ones where the rules are set out in advance that say, "You send us this kind of application and we'll review it in that many days."  But there are other kind of problems that are much more difficult, and we need to really be able to set goals, choose important problems, assess how to measure the impact in those areas, create the team needed, and then be accountable for timely results of the efforts.  And we're going to need to learn to prioritize and do that.

     We agree with your comments about the way that people's work is reviewed, whether it's the quality or the impact of the decision.  Peer review is under-utilized as a method for prioritizing our efforts for evaluation.  And usually when we do evaluation, it's through the usual hierarchical supervisory structure, and I think this actually misses an opportunity for people to be reviewed by their peer, to look at the incorporation of science into the decision-making.

     I think that this is all the more important as we make the results of our decision, not just the decision itself but also the logic behind it, as we start publishing our summary basis of decisions.  You know, the science is laid out there bold for everybody to see, and we need to take advantage of that.

     And then finally, and again I think is very concordant with one of your recommendations, scientific partnerships with the NIH, the National Academy of Science, universities, professional societies.  Many of these exist, but they are under-developed.  We could do much more with them than we currently do.

     And so that was our top 10 list and our 10 recommendations that we gave last spring.  We put together sort of a different structure for this review, and we were sort of making it up as we went along.  And so one of the things I would be happy to show you is the survey that Dr. Nerem alluded to.  All 12 members would have responded, but one was on travel and couldn't be reached.  And this is in your packet, in a handout we gave at lunch time in a tabular form that I've reformatted for the slides.

     One question was, was it the best thing to open the scope of the review to be the entire Center and not, for example, just to limit it to the research programs.  And that, after the fact, after the review was over--this was a five-point scale where the green at the end is a five and blue is a four and yellow in the middle is a three and so forth--and so you can see that actually that was a concept that resonated well with the committee.  They agreed with you that the meeting in Atlanta was useful, and a complement to many of the people here in the room, that the background materials on the mission and organization that helped jump start that process were useful.

     Case studies.  We asked separately about the concept because we weren't--we also, particularly as we looked at the different ones, there were different levels of execution.  I think the committee got a little better into the pre-IDE one than the post-marketing one.  And the most useful thing, and you'll see this theme again, was being able to have access to interview them about the process.  Materials, they are, I mean these are complementary marks, but clearly the staff interviews were the most valued.  And the concept by and large seemed to work, whether or not--you know, I think we could have improved the execution, and some of the problems at times was trying to get it all crammed into three days.

     On-the-spot concept didn't work as well, and the committee agreed with you.  And again, the thing they liked the best was having access to staff, to talk to them about specific decisions that came up.  The role-playing didn't score as high as some things, but still complementary.   The interviews, and the importance of having a session to come back and collect your thoughts about a month after the three intensive days, and not try and do the writing in that same session, I think was a strategy that the group liked.

     We asked four open-ended questions, and we've given you all the responses to that in the handout, and I won't--they are on slides but I'm actually going to skip them, partly because the slides are unreadable, but also so we can have some discussion.

     DR. NEREM:  Your time is about up.

     DR. FEIGAL:  Oh, is my timer going to go off?  Okay.

     So what are the next steps for us?  One of the things that we did is, we established a CDRH Recommendations Committee, a committee to go over the Science Review Board recommendations and to make recommendations to the senior management and the team that's developing and continuing to develop and implement the strategic plan, to really look at how we incorporate these recommendations into our other activities.  And let me just acknowledge this group.

     You will notice this group has, if you know our alphabet soup, which Bob and Alexa now have memorized, there is someone from all six of our offices.  And because of the emphasis on quality and quality systems and peer review, we actually have quality systems experts because it's one of the things we inspect industry on, and we've actually asked them to take a look at us.  One of our one-liner goals for ourselves sometimes is, "Gee, we'd like to be good enough to pass an FDA inspection."

     We have scheduled a go-away for our division directors in December, and one of the real focuses there is going to be to particularly look at the human resource issues and the kinds of things that we can do already at a local level.  And the focus of that day will really be to ask the division directors, what can I do in my own shop now?  If I were to say things that I could take and implement all by myself for the family of staff that I'm responsible for, what is it that I think I can accomplish for the next year?  And then, of course, they can also lean on the rest of us, to let us know what kinds of resources and support are needed for that.

     And as of this afternoon, we're posting this report.  We really welcome the report, and we even had a preview, you know, because Bob was kind enough to come up and present this to the office directors two weeks ago.  We knew what we were posting.  And we have arranged to have videotapes of your presentation and this afternoon's comments replayed in the Center, and will be available for people.  What we will ask the CDRH Recommendations Committee to do is to prioritize the recommendations, to identify which are things that are short-term and which of them are longer term goals, as I mentioned, to make these recommendations on how to merge into the strategic plan.

     I think one of the themes that you developed, I would just like to sort of comment a little bit about at a high level, sort of how I see the strategy.  And I think in some ways it begins to change the way that we think about doing the business, even though many of the elements are there, and many of the things that you ask us to do are not things that we don't do at all, but we do them as you describe, in groups that sort of work more or less autonomously.  They actually pull together quite nicely in a crisis, and I think you saw evidence of that when we presented some of the cases, and there is quite a bit of interaction, and when they need to get together, they know, the staff knows how to do that.

     But I think one of the things that your recommendations make clear is that we would be more effective and much more powerful as a scientific group if we could knit it all together.  And one of the issues is how do we deal with new technology.  We've had several on the horizon.  One that gets mentioned sometimes, very often, is the revolution that's going to occur for genomics, and our part of that will at least be diagnostic testing.  Over 1,000 diagnostic tests for genetic diseases are under investigation and are available now that weren't available five years ago, so it's really an area that is exploding.  And ask us if we could do 1,000 PMAs in a year.

     Well, part of the process, this is sort of--you know, I think what your challenge to us would be is to get down to the nitty gritty to do this, is to have a process that scans the horizon.  We have done that, and some of the responsibility for that actually has been OST, to identify the kinds of technologies that are coming along.

     One of the first things industry always wants to know from us, though, is what's the regulatory path, and that's sort of one of the complaints often, is that there is this down time while they try and figure out how they're going to get this product to market, and whose product is it?  It is going to be Kathy?  Is it going to be us?  Heaven forbid, both of us?  And they probably haven't come up with four-Center combinations yet, but I think we've had a couple of three-Center combinations.

     And this is the time, shortly after that, to begin identifying the external expertise and using external expertise.  That is the best way to actually build that expertise into our own staff and our own workings, at a time before we have much action.  So while I mentioned that we actually are at the point where we have work for geneticists on the staff, but for a long time it was on the horizon, it was coming, but there weren't any products, people weren't talking to us.

     But I think this is something we often don't do, and we haven't done aggressively, is to really identify how do we find the external experts and really build them and make them part of the team, and then at a point when a product area becomes busy and begins to pan out, to build internal capacity and develop our own staff.  Some of that initially would logically be retraining of people that have done similar types of things.  That is how we have had to deal with many of the issues of bioterrorism, is to look at the skills mix of existing staff who really weren't doing bioterrorism before and say, how do we now turn you in this way?

     And then, finally, I think we need to begin to consolidate a team that works across the whole life cycle.  We're often actually not too bad at taking a new technology and getting it to market.  But if it's a brand new technology, we may not have thought yet about the human factors, about the post-marketing problems, about how rapidly the generations of that product are going to change, what the issues are in terms of risk communication and communication about the products.  And I think that we are beginning to pull together a conceptualization of sort of how we need to do business in a way that explicitly takes on new technology, and doesn't just put us in the reactive mode of waiting to see what comes in and what gets filed.

     As you know, we have presented before in a brief form the theme areas that we developed about 15 months ago for the strategic plan, and one of the good questions you can always ask of strategic goal areas is whether they serve you when you get a new challenge and have something that you need to do.  And actually I think that many if not most of your recommendations fit well into groups we have organized to work on these issues.

     The semi-porous silo issue is a quote I liked enough that the staff have already heard me repeat it as the characterization of the Center.  It's something that we are aware of, as we look at the need to work the Total Product Life Cycle, and are aware of the fact we don't quite do it yet.  We do it with handoffs now, more like a relay team than a team all pulling together.

     The Magnet for Excellence to really develop our staff and human resources is an issue not just for recruiting new people but developing the very talented and dedicated people that we have.

     Knowledge management, and where these two things meet or how to develop expertise databases, there are some very interesting tools and technologies out there now that may make this more helpful.

     And then the final area is meaningful metrics, which is that we really want to be able to measure the impact that we have when we take something on, not just do it because we think it would be a good idea, but actually to be able to say, what is it that we're hoping accomplish with this?  So, for example, if we have a mentoring program, it's good to have mentoring programs, people in them like being in them, but I would like to then ask the question a step further:  What are we trying to accomplish with that, and how do we know if we have a successful mentoring program?  In my mind, one of the things it needs to help us do is transfer some of that institutional knowledge and help us with succession planning, because a lot of our retirement will be with our more senior people in the Center.

     So, you know, in short I think the other way to come back and ask whether or not our efforts to do the internal review and the advice we got with you really stayed focused on the core thing about the Center, which is our mission, which is really pretty straightforward:  promote and protect the health of the public; safe and effective medical devices; safe radiological health products.

     And I think both of us are actually on that, on that mark.  I think that there are things that, as you said in one of your slides, were not things we would accomplish overnight.  There actually I think are some things that we should be able to start on very quickly, and I think one of the things that we need to do is to really identify where we want to go and what we think the challenges will be to do that.

     And we need to do that without saying, "Well, we'll do it if we get a good budget," or "We'll do it if we get these new things funded," or whatever.  These are things that are so fundamental to the way we do business, we have to take a look at our resources and say, "Hey, we've got 1,000 people in Rockville, 1,400 people nationwide, a budget of about $140 million.  We ought to be able to do something with that."

     Let me just close with a thank-you for all your time.  I hope you will expect a progress report from time to time, and we will--we are starting on this activity already.  So thanks very much.

     CHAIRMAN LANGER:  So, comments or questions for David?

     DR. NEREM:  Alexa, did you want to make any comments?

     DR. CANADY:  I just think that, sitting here this morning, many of the issues that we discussed this morning are directly applicable here.  So I think it--but the key issue I see is the need to support with education and training the existing staff, as well as the new staff.  That, like in most places, loses out to number counting and just budget deteriorations, and I think it is critical in a time of technological advance like we're in now.

     CHAIRMAN LANGER:  Comments from people on the Science Board, or from the audience?

     DR. KANTOR:  Can I make a comment?  I am Gideon Kantor, Adjunct Associate Professor of Biomedical Engineering, Catholic University, and as I said before, until '95 part of FDA.

     These are excellent reports, but I would like to draw attention to a point that I think was part of the report but maybe not as emphasized as I like to do it, and that is linkages.  Linkages are extremely important.  Now, for that purpose I have mentioned briefly before that you need a meta search engine, and what I'm talking about is that the different offices are looked at from the keyword point of view.  What are the keywords that identify the differences in their subdivisions?  Then all the computer does is link those keywords together.

     Now, let me give you an example.  For example, medical device panels, they are the categories of medical devices in terms of clinical applications.  At the Office of Science and Technology there are breakdowns in terms of science and technology.  Obviously the medical devices contain some of the components of the Office of Science and Technology, so that's just one example.

     For example, you have an implanted defibrillator.  You are concerned about particular technical issues, say interpretation of signals, say safety of equipment.  If you have these keywords, you could easily link them.

     So what I'm proposing is the following:  that each office and subdivision looks for some keywords that are common to other entities of the organization, and then when you have established these linkages, establish a panel of experts that validates these linkages.

     I know it is a new idea, and a lot of people are opposed to it for many reasons.  I think many of them are good reasons, but when we look into the 21st century with the information explosion, we need to put some order in the complexity that surrounds us.  And just looking individually at issues is a very good idea, but it is beyond sometimes our capability of our brain, that which decision scope has not been increased.

     So I very strongly believe in this, and I hope that people tell me if they do not believe in it, because at least I can learn from these.  Thank you very much.

     MR. BENSON:  I'm Jim Benson with Avimed, and I think this is not on but that's probably okay if you can hear me.

     DR. NEREM:  You have to lower yourself, Jim.

     MR. BENSON:  I have to lower myself?  I thought I was raising myself when I came to this meeting.


     Just for the record, I went to Georgia Tech and the University of Maryland.


     Also, I didn't know whether to stand up when David asked for people in the Center to stand.  I was a little torn there.

     I want to just say a couple of things.  One is compliments to this Board for establishing this look, and specifically to the subcommittee which Bob chaired, and also very much to the center, to David and the folks in the Center, because from my own experience at FDA, introspection, organizational introspection isn't always a fun process, and so I think everybody should be complimented for that.

     A couple of specific comments, if I may.  One, I think the emphasis on looking at new technology, the exploding, I think that's a critical step.  The concept of OST looking at, paying maybe perhaps more attention to new technology, perhaps along with that letting go some of the ongoing projects, I think can be a challenge and a very exciting one.  I think looking to the outside for help for the existing expertise in the Center is terrific.  I think that budget is a problem there, but I think there are ways that the budget can get increased, as well as under existing budgets to be able to enhance that.

     I noticed in Bob's report, you included industry, scientists in industry as part of that.  I would encourage David to include that in his outreach slide.  And I'm not nit-picking here, I just think the concept is important.

     If the agency wants, for example, to really take a look at new technology and some of the problems associated with that, probably if you look to the industry as well as to academia and other institutions, that combination I think can be enormously helpful, and we really need to figure out a way to accomplish that.  Thank you.

     CHAIRMAN LANGER:  Other comments?  Anybody on the Science Board, any questions?

     Well, then we are supposed to decide whether we want to accept this report, I guess.

     DR. FENNEMA:  I would move acceptance of the report.

     DR. PRINCIPE:  Second.

     CHAIRMAN LANGER:  So why don't we vote?  It has been moved and seconded, so we're going to vote on it.

     DR. SCHWETZ:  I did have a question, but Bob and Alexa, I want to thank you very much for all of the work that you obviously put in, and you took this very seriously.  You were innovative in responding to what David came up with as an innovative approach.  So on behalf of the agency, if we have a good review of a component, it helps the agency broadly.  So I thank you for all of the work you did as well as the rest of your team, and we will get thanks out to the rest of the team as well.

     My question is this:  These reviews were not meant to be the end, they were meant to be the beginning of a process.  Having taken this innovative approach of taking one product line from top to bottom, as opposed to the horizontal approach that we've taken in other Centers, what do you recommend as a follow-up?  Do we go in greater depth to this, or do we go horizontal, or do we pick another piece and go vertical?

     DR. NEREM:  My own view, and Alexa may wish to comment also, but I think, I really believe that the most important part of what happened was the internal review that took place.  And I think that somehow needs to be brought into, if you wish, business as usual.  It's part of one of our recommendations of taking a reflective look, not to criticize people for decisions that may or may not have been made but simply to look at what has been done and use that as a learning experience so as to do things better in the future.

     And certainly there are other areas of CDRH where the report card might not be exactly the same as in the electrostimulation device area, so it's important to look at these other areas.  I don't think that requires an external team to come in.  I think that's something that can be built into a regular internal review process, and maybe each year look at a part of what's going on.

     DR. CANADY:  I really agree with everything Bob said.  I would add one piece.  I think that it was valuable for the members of the CDRH to have an opportunity to talk without management present and make arguments, and that you can't really get with the internal review process, but I agree that the internal review process was critical.  We could never have gotten this far.  We would have spent a lot more time trying to understand things if we did not have the internal review process.

     DR. SCHWETZ:  I just want to give credit where credit is due.  Dr. Fennema was the one who had the idea that this was an essential and critical part of the review process, and we have used it since he provided leadership in a review team of CFSAN.  And when that was brought back to the Science Board, it was approved as an institutionalized process.

     So I have always felt that the internal review is the most beneficial part of any of these reviews, and what you get from the external reviewers is additive to that.  But the reality is getting your own people internally to recognize what they're doing and what they haven't done and to deal with it.   And I would say that we are going to continue to have these internal reviews as we go through the rest of the Centers, and I would encourage that, to the extent that they are useful internally in the absence of an external review, that we would look at that as well.

     So, Owen, thanks for an idea that has become very beneficial to us.

     DR. NEREM:  But in addition to this being used as  part of the Science Board review structure, it also could be built into the business as usual of the Center.

     CHAIRMAN LANGER:  One question I had--go ahead.

     DR. FEIGAL:  I was just going to comment on some of the processes that we plan to take forward.  We actually appreciated the suggestion to actually now do a review of OST.  We think actually this larger view now actually gives a context to look at the research efforts both in OST and in some of the other areas like epidemiology where we have specific research projects, and we can actually look at them.

     Another effort which has been going on, and actually has an external component as well, has been for us to look at the Radiological Health Program, since this Center is the merger of two different programs with different authorities and different laws, and we didn't really try and do the Rad Health Program, except there was some overlap.  And so those are two activities that will be extensions or continuations of that, in addition to tackling some of the explicit issues from this report.

     The internal review was definitely something that we did get from Dr. Fennema.  We actually hadn't initially planned to do it.  We had an internal group preparing for the external review, but we actually pulled a separate team together that was more senior than that group and asked them to actually prepare the internal review report, and it was the right decision to do that.

     CHAIRMAN LANGER:  Any other comments or--yes?

     DR. ROY:  Suva Roy, Otsuka Maryland Research Institute.  I have never worked at the CDRH, but I have worked in the Center for Devices.  One of the things I would like to see perhaps this meeting address, and that is the qualification and experience requirements for reviewers.  Many of the job descriptions, at least I can speak from the Center for Devices, many of the job descriptions were written 30 years ago and it has really not changed, or the experience or qualification requirements.  But if those things are not updated, the Centers may not be able to attract or retain the best possible people.  So you can retrain people, but it's like somebody said, you cannot buy a Volkswagen Beetle and expect to run it like a Porsche.  You have to have something underneath to work from.  Thank you.


     DR. NEREM:  Yes.  On that comment, I mean, I think as part of really doing an in depth look at the kind of staffing needed five years from now and strategic recruiting, you really have to look at how different positions are defined.

     The other thing I wanted to comment on was Jim Benson's comment.  I mean, we specifically included industry, and I noted it was left out of your slide, David.  I realize that industry on the one hand is what's being regulated.  On the other hand, for certain technologies, the expertise is actually out there in the industry, and if you're going to learn about those technologies you have to take advantage of that, and you have to somehow walk that balancing act.

     CHAIRMAN LANGER:  Okay.  Well, thank you very much.  I think that's an excellent review.

     The next thing is emerging issues and FDA's oversight of clinical research, and David Lepay will discuss that.

     DR. LEPAY:  I'm going to change the focus of attention a little bit here.  What we're going to talk about for the next 20, 25 minutes are clinical trials, the conduct of clinical trials and the oversight of clinical trials, and some of the events that are taking place here within the agency and in our interactions with the department.

     I think first and foremost, though, when we start off here we have to give ourselves a very large round of applause for the progress that has been made over maybe the past 25, 26 years in the area of clinical research.  If you look back, it really is very much since the mid-'70s that certainly most of the infrastructure that exists today for the oversight of clinical trials, for IRBs, has been put in place.

     The Belmont Report, which specified many of the ethical underpinnings of our current models, is less than 25 years old.  The true implementation of evidence-based decision-making at the agency probably has taken place within the last 25 years, even though it was put in place with the Kefauver-Harris amendments to the FD&C Act.  And even our standards for research conduct, 10 years ago we hadn't really defined what good clinical practice was in the agency, and it was only when we began moving internationally into harmonization that we began to move forward in that direction as well.

     And, similarly, we have seen a lot of attention to quality improvement, quality assurance systems in this period of time, and ultimately a very significant improvement in the quality of clinical research as FDA has viewed this in the course of our own oversight, in the course of our own inspection system.

     In 1977, the first year that we began looking at clinical investigators and clinical trials, we were seeing certainly quite a number of problems, and the percentage at least--granted, we didn't do very many inspections back in the '70s--the percentage of what we are seeing now in terms of major problems, the red there on the graph, is down on the order of about 2 to 3 percent where FDA has to come in and take official action in clinical trials.  But that is a percentage change.

     We certainly know that the clinical trial landscape has changed markedly over the past 25, 27 years.  We may see fewer percentage problems, but we know there are more sites, there are more special investigators.  We know that from the standpoint of trying to get good clinical data on populations that will be using our products, we have needed to go forward and encourage the enrollment and recruitment of subjects from more vulnerable populations than we ever have:  children, the elderly, ethnic groups.

     We have had to face changes in the whole clinical trial system, the way it's conducted, the outsourcing of trials, new technologies, not only from the standpoint of how information is acquired and transmitted by electronic means, for example, but certainly also new technologies, as science has advanced to the point where basic science moved into clinical research, and we have certainly a more sophisticated level of applications that we're seeing now for review at both the research permit, the IND stage, and at the NDA stage.  This is very much an area of global expansion.

     Well, we certainly know over the past few years we've seen a large number of calls to action in the area of clinical research, a great deal of press attention.  I can't say that the IG report back in June of '98 was by any means the first, nor had we not recognized some of the problems that the IG reported back in '98, but it began to certainly consolidate thinking and bring a great deal of public attention to this area, to the concept that institutions were having difficulties in their oversight of clinical research.  Resources weren't adequate at the institutional level.  IRBs were overworked.

     In reports in the New York Times, the concept that we still have cases of very, very marked clinical investigator fraud.  We still have deaths in clinical trials.  Some we would like to think perhaps are preventable, and indeed a death in a gene therapy trial in September of '99 certainly got a great deal of attention.  Problems that we've seen abroad have gotten a great deal of attention in the press, the Washington Post series this past December, and even very recently, a death in what one would consider an academic research trial, certainly a trial that was not being forwarded for purposes of commercialization, but nonetheless a trial that used an investigational product, a drug by our definition, in a clinical investigation.

     We are certainly trying to answer these calls, and I'll talk a little bit about how we're going about doing this in just a moment.  Ultimately, in answering these calls we really are very much based on our mission, and it is certainly a very broad public mission.  We have to be out there ensuring the safe use of all of the products we regulate, 25 percent of the U.S. economy, or close to it, and make sure that those products themselves are safe and efficacious.

     And to do this we need the information that comes from clinical research.  We need the science that's underlying our decision-making, much of what has been talked about all day today.  We need the accuracy, the completeness of information from good clinical trials.

     Our mission is very broad, even from the standpoint of what we have to oversee in the clinical trial arena.  We call it Good Clinical Practice, but it is very all-encompassing, very all-embracing.  It's not simply domestic, it's very much a global issue.

     But fortunately we have colleagues in this process.  GCP is, as it has been developed, very much a system of shared responsibilities, and that is the system that in fact we want to keep as vital as possible.  We think it has done a very good job of protecting human research subjects over the past 25 years.

     And in fact when we look at those few cases that I mentioned in the earlier slide where we've had deaths, we can actually look at this list of responsibilities and say in those cases we've lost some of the control points in each of those cases that would otherwise ensure the appropriate oversight of clinical trials, either by the investigator and the sponsor becoming one, the IRB simultaneously failing, the failure to bring these to the attention of government regulators to be able to interact in these clinical trials.

     We have been very lucky, as well.  Not only do we have the built-in colleagues that come from the shared system of responsibilities in GCP, we also have a number of colleagues in government and a number of new government entities, if you will, or new people coming into government entities to revitalize this process.  We're working very closely right now with the Office for Human Research Protections in the department to try to develop human subject protection as a unified entity, working as a single voice across government.

     We're even extending that beyond the department, working with the National Science and Technology Committee's Subcommittee on Human Subject Research.  This is a group of representatives from all the agencies that are involved in human research, behavioral, social, as well as biomedical.  And we've seen a lot of new infrastructures coming into play, the VA putting in its own series of systems, again for research compliance, for research assurance, for furthering the oversight of clinical trials.  New advisory committees coming down the line, as well as advisory committees to our own department, the National Human Research Protection Advisory Committee.

     But at the end of the day, though, we have all of these colleagues to work with, but we are still in some measure left on our own because FDA certainly has very unique responsibilities, responsibilities that ultimately play on decision-making in applications.

     So what are we doing to address the issues that we've seen, the problems that have come to public attention, the problems that are threatening the ability in fact to even conduct clinical research because they're scaring off subjects from enrolling in clinical trials?  We obviously have to address these concerns and make sure, in fact, that FDA is out at the forefront assuring that these protections are in place.

     So we're looking at this from the standpoint of broad initiatives, initiatives to further protection of human research subjects, initiatives directed at defining and improving the responsibility of those who are involved in the clinical research process, looking at reporting to FDA, trying to pull in and trying to enhance the reporting of problems to FDA so that in fact we can help work toward creative solutions in dealing with them.  Ultimately, as well, issues in education and outreach.

     And I'll talk also a little bit about how we're moving in these directions through an infrastructure that we put in place, the Office for Good Clinical Practices, and the collaborations out of there.

     Well, first of all, protection.  We know that there are a number of areas that we certainly have to strengthen and that we are working to strengthen.  These aren't the level of the IRBs and institutions, looking at issues in real-time oversight of safety.  It's not simply enough to come in after the fact, after the damage has been done.   Looking at effective sponsor monitoring, not simply rote sponsor monitoring but mechanisms that will make the clinical trial process better.  Strengthening our system with regard to clinical investigators and site staff, as well as our own responsiveness to subject concerns and complaints.

     Well, let's talk a little bit again about how we're moving to strengthen the IRB system.  One very fundamental criticism that was brought our way by the Inspector General was that we don't even know the entire spectrum of IRBs that are out there, involved in FDA regulated research.  So we have taken to heart the concept of moving toward IRB registration, not simply just to define an inventory but because now we have information technology capabilities available that, as we do put in place an IRB registration, it can be a two-way system that gives us the ability not only to know who is out there doing our work, but to be able to tell them what is new in that area, what we can come to expect in this area.

     We're working with the Institute of Medicine and with OHRP, the Human Subject Research Subcommittee, toward at least piloting accreditation of IRBs.  The goal is to raise the floor above our minimal regulatory requirements, and we think the best way to do that is to move to an accreditation system that will largely be outside of government, where the standards in fact will help to promote improvements in the process.

     And I think very much the third point is one we have to take into account, and that is to start reducing unnecessary burdens on IRBs and institutions where these are adding little to human subject protection, or indeed where they are already covered, covered by industry, covered by us, or otherwise better covered by systems that we perhaps have to still think about but put in play.

     So, again, the challenges for us are still, one, we have a whole series of human subject protection functions that we need to cover.  We know that this includes a review of the ethics, a review of consent, scientific review, general monitoring of studies, specific safety monitoring of studies and real-time safety monitoring.  We have to make sure conflict of interest isn't at issue, and we still have concerns about maintaining privacy and confidentiality.

     All of these functions are necessary to keep the process vital, but we have to now take a look back and say the IRB cannot do all of these.  We have to look again at the system.  Say these all need to be covered, but who best to do these?  We have to sit down and have very clear dialogue about how to define roles, how to redefine roles, and how to clarify roles.

     GCP works because there is some reasonable level of redundancy.  We have to keep some reasonable level of redundancy, but too much becomes ineffective and bogs the entire process down, and this is what we have to work to avoid.  Ultimately it's not simply a question of putting in place structures, putting in place systems.  We all know that the real key is ensuring that these systems are performing, that they are providing protection for subjects, that they are providing the quality of science that's going to be important to FDA decision-making.  And for us to do this, of course, we need to ensure through education, and through an emphasis on quality assurance and continual quality improvement, that we are making strides in this whole process.

     We are working as well in the area of real-time oversight of safety.  Part of this we know.  The only group that can effectively, or the only individual that can effectively ensure the safety of a subject, outside of the subject his or herself, is going to be the person with direct contact, and that's the clinical investigator or site staff.  I think we're all fooling ourselves if we think that parties very remote to the subject and the investigator are going to accomplish this.

     So we have to spend time on education and on changing institutional culture to ensure, if we have cases where an investigator is also the sponsor of a study, they understand that that doesn't mean they have to do less because no one is monitoring them.  Rather, it means they have twice the responsibility and twice the need for education in what they need to do properly.

     We're moving from the standpoint of oversight of safety in looking at other structures, though.  I mentioned this earlier, and one of these is certainly the Data Monitoring Committee.  And I am actually very pleased to say FDA's guidance, our draft guidance on data monitoring, issued on our web sites this morning.  It is a fortuitous time because in fact we had long ago planned to have a public workshop on data monitoring committees scheduled for the 27th of November here in Bethesda, and we are pleased to see that the guidance document is available, such that it can be discussed in time for that meeting.

     And we're working as well with NIH, with OHRP, in discussion of safety databases.  We're also working in the area of protecting vulnerable populations.  As I said, part of the goal is for FDA to speak with one voice across government, and one of the ways we have done this is by starting to look at those regulations, those areas that we can promote protection, that are otherwise part of government regulation, such as the PHS regulations for children, and have moved to adopt these.  And so Subpart B of the interim rule, which looks at safeguards at the level of the IRB, is now part of FDA regulation as an interim rule we put into place in April of this year.

     We are starting to look down the line in other ways to look at consistency, looking at Subpart B, for example, another PHS subpart dealing with protections for pregnant women and neonates.  We are looking also to enhance our bio research monitoring program.  It's always difficult.  Our Office for Regulatory Affairs, our field force, has always had to suffer from continual declines in resources, from continual pulling away of resources from the bio research monitoring arena, but I think now we can say we may have some success in moving toward more resources in this area.

     But again, like with our other strategies, we have to use these resources very wisely.  We have to use them in a strategic planned fashion, and part of this means we need to take on a balance between how we look at clinical trials.  We need of course to keep looking at trials as they are submitted for purposes of FDA decision-making, to ensure the integrity of the information that we get, but we also need to use those resources to be responsive to the public, to the community, by following up on real-time complaints, and also to promote the science of what we are doing in our bio research monitoring program.

     It's important to know at some level the state of affairs in areas that are coming to the forefront scientifically or that are involving vulnerable populations.  We started this with gene therapy, doing a more systematic statistical look.  Certainly we're going to move down this row in other areas.  I put pediatric trials up here because that's one area I certainly would like to see focused upon.  I'm not sure we have yet made our decisions entirely where we are going to next prioritize.

     But in any case, as we are moving forward, we are moving forward with other groups as well.  We have started information sharing with OHRP, with the VA, with others, to make sure in fact that we are not simply preventing compliance information that's important to all of us from going forward.

     And you've heard a lot about quality.  Quality assurance is going to be part of our system.  We know we need to look not only externally and promote to industry that they should have quality assurance programs and quality improvement programs.  That's kind of a hollow advice to industry, if we ourselves are not doing the same with FDA, such that we can assure industry and the community are getting consistent advice from us.

     Well, I said before, part of this also is responsibility.  We have to be out there ensuring understanding.  Right now, of course, based on issues I had mentioned with recent deaths in an academic setting, we have to reiterate what in fact our law, what our regulations currently state.  Our law defines "drug" very broadly, and in fact it is, as indicated here, one of the definitions is "articles (other than food) to affect the structure or any function of the body of man."

     So you can see, I mean, our coverage, what we are responsible for in regulating these products under the law is very broad, and similarly clinical investigation in our regulations is defined very broadly.  It's essentially, for an approved product, any use in medical practice, any experiment that involves a drug or a test article.  And indeed, under our regulations, when an unapproved drug or product is used in a clinical investigation, there is a requirement for FDA to be part of the regulatory scheme.  That's how it currently exists.

     So the challenge for us, of course, is now we have to reiterate that message, that challenge studies, physiology studies of unapproved drugs, biologics, and significant risk devices certainly meet the definitions for FDA jurisdiction.  But we also need to be cognizant of how that's going to impact science and how that's going to impact the community, by understanding the nature and scope of these activities in the community, and ultimately coming to work with the community to define how we can balance our level of oversight with the level of risk in these studies.

     We also know that for a long period of time we have had a lot of ambiguity in what the FDA definitions read, such that this has created loopholes and confusion in the clinical trial arena.  I don't think any of us in FDA ever thought of the concept of a subinvestigator or, excuse me, an investigator being 3,000 miles removed from any physical or verbal contact with a patient, but simply an administrator in an office, while everyone else is a subinvestigator.  That's certainly not what ICH, the International Conference on Harmonization, has put forward, and I don't think that that's what any of our thinking would be, but that's not clear from our definitions.

     We have those intrinsic problems, as I said, with sponsor/investigators.  If GCP is a system of controls, what happens when the sponsor and the investigator are the same?  They are silly questions in some ways.  I mean, the sponsor is responsible to provide the investigator with an investigator brochure.  What does this mean, that the person gives it to themselves?  The sponsor is responsible for monitoring the investigator.  What does that mean in a sponsor/investigator study?  The person is supposed to monitor themselves without any additional oversight?

     These are real problems for us right now.  And similarly there are the problems of what are the responsibilities of institutions?  We said IRBs can't do it all.  FDA certainly regulates institutions.  We have to talk a little bit with the community about what institutions can and should do.

     It's certainly an issue as well of conflict of interest.  We are working as part of a broader group with OHRP, with the National Human Research Protection Advisory Committee, the Human Research Subject Subcommittee, to develop guidance that will help promote minimization and managing of conflicts of interest, not simply payments but any kind of potential conflict of interest, and not simply for the investigator but also at the institutional level.

     We also have to look again at non-U.S. trials.  The Inspector General has just told us that we've seen a 16-fold increase in the number of non-U.S. clinical investigators submitting data to FDA applications over the past 9 to 10 years.  That's certainly a very large increase, but one that you would expect with the globalization of the industry.

     We have to look back at our criteria for accepting these studies.  We didn't have standards, we didn't have GCP defined when we put most of these regulations in place, and we rooted our acceptance in very vague ethical principles such as those in the early versions of the Declaration of Helsinki.  We've moved a long ways in that area on an international level.  Certainly much of the world, even China, is now adopting GCP standards, ICH GCP standards.  I think it's time we update our own expectation.  You know, we expect more than just vague ethical conformance.  We want good science, we want good quality, and we also want good human subject protection.

     And we've made a great deal of progress in GCP harmonization.  We're looking toward GCP as a more concrete standard, and as we do that, we're looking also toward venues in which we can further globalize these standards.  The World Health Organization is interested in this.  They have recently convened a consultation that we were part of.  Pan American Health Organization as well, in the device area they're moving forward through ISO.  And as well we're looking at mechanisms for supporting capacity building.

     We also need to deal with issues of misconduct.  We've talked with various of the trade organizations about this, because in fact we worry that there are loopholes in the current regulatory schema.  The concept that in fact only the clinical investigator is subject to termination with regard to falsification, this is a matter that takes place only if the sponsor can't correct.  And if the trial is over, what does that mean?  You obviously can't correct when the trial is over, but yet that is a loophole in terms of reporting.

     So we're looking back at what we expect to keep scientific validity in an appropriate position, and we're looking at this even from the standpoint of regulation.

     Well, there is no question education is ultimately going to be the key.  We need to target all who participate.  We're looking at technology.  We've moved into web sites that are going to provide access to GCP information in consolidated spots.

     This is our new office's web site, and I'll say a little bit more about that very briefly.  One of the things that certainly we have decided to do within the agency, and I certainly thank our leadership council for that decision, is to establish an umbrella office, a new office to coordinate GCP across the agency as well as beyond the agency.

     We had a few abortive attempts at  naming this office.  Those of you who have followed this perhaps in the trade press, we started off as an Office of Clinical Science.  I think we all agreed that this embraces perhaps more than clinical science, embraces all aspects of Good Clinical Practice.  We tried Office for Human Research Trials, with the hope that this would give the impression of our coordinated work with Office for Human Research Protection at the department.  Unfortunately it created a great deal of confusion instead, so we settled on the Office for Good Clinical Practice.  It is a very small office.  We're going to remain it as a very small office, but it is strategically located within the Commissioner's Office and our Office of Science.

     The key positions are the directorship and two other positions that we certainly moved forward with.  Stan Woollen's position is Associate Director for Bio Research  Monitoring.  Stan has had some 25 years of experience or 23 years of experience in the bio research monitoring arena, at all levels from field to ORA management to Center level and now to our umbrella office.  And Bonnie Lee, who has been involved with FDA in the area of ethics issues from the time of the National Commission and the Tuskeegee studies, again a very long-experienced person in the area of human subject protection policy.

     So our role here, we're taking on a centralized role as a small office, to largely try to bridge the Centers and ORA in the development of GCP policy, in developing those quality systems for our own bio research monitoring program and promoting quality in clinical trials externally, in many of the initiatives I've indicated earlier, in our agency international harmonization efforts in the GCP arena, and coordinating GCP education and outreach.

     In remaining small, we have to rely on leveraging, and so this affects--again, I am very thankful for the cooperation of our senior management, our leadership council, by providing our group with the resources of the key medical policy-makers and key compliance policy-makers within the agency to come together in steering committees, roundtables, and working groups to deal with these problems.  So we don't have to manage them from a top-down perspective, but we can actually pull in operational knowledge from the strongest people in all of the Centers.  We're leveraging with OHRP and ultimately leveraging with our stakeholders.

     So in conclusion I would like to say we are moving forward, but there are certainly a lot of opportunities where we need to work together as an agency and we need to work together with all of our constituency groups.  The reforms are underway here, but the only way we are going to make these systems improve and get the best possible systems is to get our broadest possible participation.

     So I thank you very much for your time and attention, and I will be taking questions.

     CHAIRMAN LANGER:  Questions or comments from anyone?  Okay, thank you very much.

     Comments in general from the Science Board?

     Another comment?  Yes.

     MS. MOENCH:  Yes, I would like to ask a question if you don't mind.  Liz Moench from MediciGroup, and I really want to thank Dr. Lepay for an excellent presentation.  I think there are a couple of things, as I listened to this presentation, that I'd like to ask for consideration on.

     Number one is, I think that we have to give very careful consideration to the role of the PI.  The PI is, I think needs to be redefined, not as a physician investigator but actually as practically invisible or partially involved.  In a study that we just completed, I can tell you that out of 100 PIs, only two were actually actively seeing patients.  So what we're seeing is a very concerning trend where it is the study coordinator who is playing a much more involved role in clinical research today, and not the PI.

     So I think that that is an issue that is certainly going to rear its head in clinical litigation, because we're certainly seeing more and more law firms getting into this, and maybe then sponsors will play a more active role in setting performance standards.  But I think that maybe FDA, in collaboration with the AMA, could play a role in more certification.  I like the idea of IRB certification, but I would certainly like to see greater certification of physician investigators who actually really realize what their role should be and what the consequences are.

     I think the other point I would like to raise also is that we really need to look at overhauling the informed consent process.  I can tell you when we do comprehension testing, many patients have no idea really what they're signing.  And there is some marvelous literature now out there that actually shows how few patients really comprehend in fact that they are even participating in a clinical trial.  So I would really like to see that and some more comprehension testing, that sponsors in fact have to demonstrate that the patients really do understand.  I know Lou Morris did so much work looking at comprehension testing of labeling of OTC products.  Really this is comparable, I think, to that process.

     And finally, I would really like to see that we get more in terms of quality check, in terms of patient feedback.  I know that we have played a role with some sponsors in actually doing satisfaction testing, where we actually get feedback directly from patients and their experience in clinical trials, and I can tell you it has been an eye opener to some of the clinical teams.

     So those are some points that I would like to raise, and I thank you very much.

     DR. LEPAY:  Just briefly.

     CHAIRMAN LANGER:  Very briefly.

     DR. LEPAY:  I was just going to say those are three certainly very good points, and from the perspective of clinical investigators, this is something that we have targeted as an issue for the past two or three years.  We're working with the AAMC.  We've worked with a number of medical schools and medical colleges that have come forth and within their own institutions have put into place, or tried to put into place, certification programs or certification requirements.  We thought that that was a good use of FDA resources in contributing to educational programs.

     On the realm of IRBs and informed consent, I think you're absolutely right, and I think the only way we are going to achieve a better informed consent process is to pull IRBs back to basics, to say in fact the informed consent is one of the leading roles of the IRB, and we would much rather focus some level of time and attention on quality assuring and quality improving the process of informed consent, and maybe less on some of the less important, less protective issues that they have come to acquire over the years, and we have to move forward in restructuring that aspect of the system.

     DR. LEPAY:  Thank you.  So just briefly, a summary, I think what we concluded today is that the pharmaceutical manufacturing issues that were raised this morning were very important.  I want to encourage the FDA to move forward on that and to keep us informed.  And we certainly accept the CDRH's External Science Review.  It's very positive.  So thank you all very much, and we'll look forward to the next meeting.

     [Whereupon, at 4:00 p.m., the meeting was adjourned.]