American Enterprise Institute

March 13, 2008

[Edited transcript from audio tapes]

March 13
8:15 a.m.	Registration and Breakfast
8:30 a.m.	Welcome:	John E. Calfee, AEI
		Scott Gottlieb, M.D., AEI
8:45 a.m.		Panel I: FDA Regulation and the R&D Environment
	Panelists:	Richard A. Epstein, University of Chicago Law School
		Richard Miller, M.D., Pharmacyclics
	Moderator:	Tomas Philipson, AEI and University of Chicago
10:15 a.m.		Panel II: Clinical Trial Design
	Panelists:	David Alberts, M.D., Arizona Cancer Center
		James Doroshow, M.D., National Cancer Institute
		Gwen Fyfe, M.D., Genentech
	Moderator:	Bruce Cheson, M.D., Georgetown University Hospital
12:15 p.m.	Luncheon
	Keynote Speaker:	Andrew C. von Eschenbach, M.D., Food and Drug Administration
1:30 p.m.		Panel III: Statistical Issues in Analyzing Clinical Trial Data
	Panelists:	Anup Malani, University of Chicago Law School
		Richard Simon, National Institutes of Health
		Mark van der Laan, University of California, Berkeley
	Moderator:	John E. Calfee, AEI
3:30 p.m.		Panel IV: Puzzles in FDA Standards for Oncology Drug Approval
	Panelists:	John E. Calfee, AEI
		Ken Carson, M.D., Feinberg School of Medicine, Northwestern University
		Robert Ward, George Washington University School of Medicine
	Moderator:	Richard Miller, M.D., Pharmacyclics
5:00 p.m.		Adjournment and Reception
March 14
9:00 a.m.		Panel V: FDA Past, Present, and Future
	Panelists:	Nancy L. Buc, Buc and Beardsley
		Scott Gottlieb, M.D., AEI
		David Kessler, M.D., University of California, San Francisco, School of Medicine
		Mark B. McClellan, M.D., AEI and Brookings Institution
		Douglas Throckmorton, M.D., Food and Drug Administration
	Moderator:	Richard A. Epstein, University of Chicago Law School
11:15 a.m.		Roundtable Discussion among Speakers
	Moderator:	John E. Calfee, AEI
12:30 p.m.	Adjournment

Proceedings:

Day 1

John E. Calfee:  Good morning and welcome to the American Enterprise Institute.  I’m Jack Calfee, I’m an economist here.  I’m one of several AEI people you will see.

We are delighted to have so many of you here, delighted to have such a wonderful lineup of speakers.  I’m just going to say a little bit about them, the background to this event, why we’re holding it and then turn it over to Tom Philipson in the first panel.

I think you could safely say that there is always a debate over the FDA’s review of new drug approvals and how they handle drugs after they are on the market.  I do not think that debate will ever end.  I think it will always be emotional.  I think in the case of cancer, it will be more emotional than it is for most other drugs. 

At the same time, there is scientific progress.  As everyone who has looked very much at pure science knows, scientific progress comes in waves.  And I think you could say that in the last two or three decades, we have seen a series of scientific breakthroughs, both in theory and in applications that have permitted and resulted in the development and the introduction of, at this point, quite a few new, and in some cases, very effective drugs for the treatment of cancer.  Most of these drugs are based upon biotechnology methods, especially monoclonal antibodies such as Herceptin, Avastin, Rituxan, et cetera.

 But at the same time, in the background, there is a science that is associated with how drugs are developed, how they are tested, how are they reviewed by the FDA, et cetera.  And this is, in a sense, a science unto itself.  As we learn more about drugs, as drugs work differently, it is newer drugs that are more targeted to do things that are somewhat different from the older drugs. 

What people have discovered is that these have implications for the way clinical trials are designed and the way they are run and the way that data from those trials are analyzed.  The thinking about these changes has also been percolating through academia, through the FDA and elsewhere.  But I think it is safe to say that the science of drug development, especially the science of clinical trial design and the interpretation of clinical trial data -- that those scientific advances have received far less attention than the advances related specifically to the drugs themselves. 

The purpose of today’s get together is to bring some people in who have been in the middle of what has been going on, in the middle of working on how you deal with oncology drug development, testing and approvals and to talk about what people have been working on, what some of the ideas they come up with, many of which involve either proposed or in some cases actual changes in how the FDA thinks about this entire process.

You will hear some criticisms of the FDA.  You will hear some support for the FDA.  You will hear from the FDA itself tomorrow.  But I would emphasize that this is very different from the usual “FDA is too fast, FDA is too slow” kind of debate.  What we are really talking about is new ways of thinking about some scientific aspects of this entire process and how these new ways can be harnessed towards the ultimate goal which is getting better drugs developed more rapidly, getting them to the patients faster, learning more about how they work and making those drugs more useful.

We have speakers from academia, from the National Institutes of Health, from the FDA - we will have the FDA commissioner for lunch - and tomorrow, we will have several former FDA officials including two former FDA commissioners.  I think it is probably safe to say that the one - how shall I say it -- sort of an interest group that we do not have are the patient groups -- that may have been a mistake, but I know there are people from patient groups here today and will be here tomorrow.  And I expect that we will hear some questions from them as everyone who has organized a conference like this knows it is not easy to get patient groups involved without getting two or three involved because frankly they do not always agree with each other.  But the focus here is on what is going on at the FDA and the academia, the scientists, the statisticians and so on.  And we will hear a lot from them.

A few thanks are in order:  Walton Dumas and Kristy Hawley of AEI staff had a great deal to do with organizing this conference and making sure that it works.  Our conference’s staff, who are, of course, indomitable as usual.  They are also making these things work today and tomorrow.  My colleague, Scott Gottlieb, former FDA commissioner, who is one of the organizers along with Richard Epstein of the University of Chicago Law School who is to be on the next panel. 

And Richard Miller of Pharmacyclics who is also on the next panel, and Richard Miller happens to know almost everyone as far as I can tell.  He is involved in cancer drug development testing and now it is et cetera, and therefore has put his network to good use in helping us to recruit speakers and actually get them here in Washington and to persuade them that it is worth taking the time and trouble to come to Washington D.C. right smack in the middle of silly season when everything gets twisted around.  On some nights you never know exactly what is going to be the result of anything that anyone says about anything remotely connected with politics or with the federal bureaucracy.

I think that with those background remarks, I will turn it over to Tom Philipson who will get us started on the first panel.  Thanks, Tom.

Panel I: FDA Regulation and the R&D Environment

Tomas Philipson:  Thanks, Jack.  My name is Tom Philipson.  I’m at the University of Chicago, an economist there.  I’m also a visiting scholar at AEI.  And as it relates to today’s topic, I used to be an employee of FDA as the senior economic adviser to Mark McClellan when he was commissioner there. 

In terms of the science of FDA, I have always argued that it is remarkable the lack of science in analyzing FDA policy in the sense that if product applications came to the FDA with that kind of science, they would just be thrown out the door in some sense.  There is very little comparability with the scientific merits of certain regulations relative to the evidence needed for products to actually pass muster at FDA.  And I hope conferences like this will start changing that because the understanding of the relative cost and benefits of many FDA policies, particularly in a quantitative sense, is very poorly developed.

Today’s first panel is going to discuss two aspects of that and after long fighting, we decided that Richard goes first and Richard goes second.  So there are two speakers - one is Richard Epstein; he is the James Parker Hall Distinguished Service Professor at the University of Chicago.  He is also the director of the John M. Olin Program in Law and Economics at the Law School as well as a Senior Fellow at the Hoover Institution.  The second speaker is Richard Miller, a clinical oncologist and a biotech entrepreneur.  He is currently the CEO of Pharmacyclics and also a clinical professor at Stanford Medical School.  So the first speaker will be Richard Epstein and then we will follow with Richard Miller.

Richard A. Epstein:  Thank you so much for the introduction.  It is sort of odd that a lawyer would have to be asked to open up a conference, which deals with various questions having to do with clinical oncology.  I will say by way of feeble self-defense at this choice that I have been involved extensively in dealing with clinical medical ethics and other kinds of issues at the University of Chicago for the last 30 years, so even though I’m certainly not of the medical establishment, I have been nonetheless glommed on to it like some unwelcome antibody, I suppose.

That having been said, my job here is I think somewhat different from most people, which is to sort of figure out how it is that we try to put the FDA into perspective.  And the thesis that I’m urging here is one which is rather different and somewhat dissident from the rest of it, which is, at the bottom, I think there is really no role to have a gatekeeper with respect to cancer therapies, and that this is something which is better distributed through the various forms of voluntary agencies that work on the subject.

In order to get to the topic, I think it is important to distinguish between two particular elements of importance in dealing with this issue.  One of the questions that you always have to ask is what is to be decided with respect to the use of a drug, and on that I’m going to make to a few comments.  But the second question on which I think I have the comparative advantage is who are to make those kinds of decisions.  It is in general, I think, a serious mistake to assume that simply because there is a function that has to be discharged, that that function has to be discharged by a government agency as opposed to somebody else. 

So let me talk briefly about what I think to be some of the problems associated with the “what” and then some of the more serious problems associated with the “who” with respect to the kinds of treatments in question.  Well, the first thing I think that one wants to understand is that there are very different perspectives in trying to deal with various kinds of medical treatment.  There is an institutional perspective in which people are trying to figure out how it is that you identify those therapies which on balance promise to do more good than harm.  And for this, most of the people in this room believe, as I do not, that the randomized clinical trial is the gold standard upon which everything ought to turn.  And so, therefore, what they are constantly trying to do is to get the level of scientific validity up there to a point where you can have reasonable confidence in the overall administration of a particular product.

For somebody who is looking at this thing from the point of individual choice, it is a rather different kind of equation.  Your interest at this particular point is to figure out how it is that you maximize the expected utility that you have.  And in those worlds in which there are not randomized clinical trials out there, the question is should you be able to rely on information which is concededly less valid than the information that you might have otherwise gotten with greater expenditures.  Or, do you have to be told that the choice will be made for you on the grounds that after all that there is no trial, the thing does not get on to the market.  If it does not get on the market then you are not going to be allowed to use it at all.  And one of the constant sources of tension in this particular situation is that the private calculations of maximizing individual personal utility under uncertainty often take a very different form than the kinds of calculations that are made by the FDA or by any sort of collective organization on how these things ought to be done.

 Well, what is it that you want to make when you start looking at the individual choices?  Essentially, what happens is given uncertainty, there are always two kinds of error.  And the first of these is that you do not take a treatment that will help you.  And the second one is that you do take a treatment that will hurt you.  And you have to therefore know both probability estimates and severity estimates in both directions. 

One of the reasons why individual choice is more attractive than collective mechanisms over this is that a lot of these variables in terms of the weight of the gain or the loss are private and subjective, and to the extent that one has to use a collective mechanism to deal with this, it is an on-off decision which may be rational for some people and not for others.  So when you start to talk about the “who” question, what you have to recognize is that the downstream decision is always going to be made by people with more information and better incentives to get it right, than people at the upstream who have less information about the particular harms and benefits the particular circumstances of a given case, and in fact, do not have quite the same urgencies associated with making the decision in question.

 Now, how would you want to make this decision?  I think one critical point about this is that for ordinary decision-making, most of us are primitive tort lawyers.  And what I mean by that is that we think that it is very wrong to cause harm to a person, even yourself, by some kind of physical act or agency.  And most of the tort law is designed essentially to protect against various kinds of physical invasions that one person inflicts upon a stranger. 

But you are not dealing with harms to strangers when you are dealing with choices of medical care.  What you are dealing with is harms to self and to simply say that you do not wish to do harm.  And to follow Hippocrates down to the last line of primum non nocere, “First, do no harm,” misses a very important point which is that you may be willing to take the chance of inflicting harm upon yourself if you think that in the alternative, the benefits that you could get will outsee the harm in question.

 Well, to put it otherwise, if you are trying to make an individual decision to maximize your individual choice and you are not subject to any kind of cognitive vices or behavioral anomalies, you will not worry whether you die from a treatment or from a disease.  What you want to do is to minimize the chances that you are going to die from one of the other of those things and you are willing to increase the chances of dying at your own hand, as it were, to the extent that you think you can get a greater reduction in harm by virtue of reducing the probability of natural losses.

This tends not to be the way in which regulators work, so it creates a tension because the stress on warnings, for example, that you put on FDA packets are always talking about the downside of a treatment, never talking about the upsides of a treatment.  And in fact, in my view that constitutes a systematic form of misrepresentation precisely because it directs you to one side of the equation while ignoring or underemphasizing the other part of the equation in question.

So I think, in effect, that when we look at this, you really have this massive amount of uncertainty and then the question is, what is it that you do to solve it?  And on this particular point, one of the common mistakes that people make about how organizations and individuals work is they pass [sounds like] at either a government expert body which is going to watch over us on the one hand, or a series of individual patients and physicians casting about ignorantly to try to figure out what to do.  And what they come up with is the conclusion that the centralized knowledge will tend on balance to trump the rather disorganized knowledge at the periphery while a bunch of people really do not know what they are talking about.

But as I have looked at this cancer stuff and have talked to people about it, one of the things that is striking about it is -- and I’m not the first person to have thought of this -- and that the entire medical profession in fact is organized around voluntary associations - in this particular instance, the National Comprehensive Cancer Network whose major function is to try to collect into organized information about how it is that various therapies work vis-à-vis one another.  And so what happens is if you, in fact, have a voluntary clearing house, the question then comes, why is it that you need to have a powerful government coercive force which limits the ways in which things come?  So this gets us very naturally on to the question of who is going to make the decision.

The way in which things work today, it is striking that most of the cancer treatments that are given, in fact, constitute off-label uses.  In some cases, they go back on-label when people do clinical trials but in many cases, it seems like you are on steady state to say that 70, 80, in some cases 90 percent of particular kinds of treatment modalities are done with respect to material which has not been approved by the FDA through a randomized clinical trial with respect to the indication for which it is done.  And we do not see any real violent move in order to limit the amount of off-label uses.  Rather, what we see is this kind of above the ground gray economy out there.  And what it does, in effect, is it publishes learned articles in medical societies and so forth on the various benefits and costs associated with these kinds of off-label uses.

Now why has this market turned out to be relatively stable?  Well, think of it in this particular way.  What happens when you start making an off-label use is you already have two data out there, which really matter.  One of them is somebody has done a Phase 1 clinical trial and with that establishes that you are not going to kill somebody instantly with high levels of toxicity.  So that is known going in to the particular case.  And so therefore, you are not just picking things out at random.  And of course, you could run toxicity trials, which I gather are relatively cheap in the grand scheme of things, in a voluntary system just as you could in one which is run by the FDA. 

And then in addition, you also have at least some kind of strong documentation that the drug is going to be used effectively - that term is used in quotes in the FDA - with respect to one particular medical indication.  So anybody who is now thinking about an off-label use is not simply writing on a blank slate.  The question that they are asking themselves is, do I know enough from these two sources of information, coupled with whatever else that I can glean, to say that to a particular patient the expected value of taking the treatment is higher than the expected value of turning it down?

And this is, I think then, also turns on another reason why it is I’m dubious about clinical trials is not that I want to ban them; it is just sometimes you cannot participate in them.  Sometimes they are expensive to run.  But they are not the only source of information.

When Tom and Jack talk about the advances that have taken place in various kinds of cancer resource, a lot of the advances are, as we now start to develop a theory about how mechanisms work, organized so that one can combine the field information that you have from the available usage.  Couple that with information you get from other physicians who made desperate maneuvers in hopeless cases. 

Add to that a dose of theory of one form or another and these multiple sources of information all combined may, in fact, work better than simply waiting for a randomized clinical trial.  And then hopefully, as the information starts to gather through voluntary organizations, you can systematize it and have the advantage that large numbers of off-label, perhaps irregular uses, may supply you with more information than a clinical trial which will take longer to run and has a smaller patient base upon which this thing is going to be operated.

So as one starts to go back from this, then the question is, what happens in a world when you have off-label uses where you have to get through the FDA.  Well, I think it creates some very odd questions at the front end because you only can have off-label uses with respect to drugs after you have been able to establish one on-label use, which means that for those things which have not been led on to the market at all, what you do is you can have the secondary voluntary market growing up behind. 

This, in fact, I think, puts real pressure on it because if I were a drug company and I were manufacturing something and I knew that 80 percent of the uses of most drugs turned out to be off-label, I would want to push very hard to get any on-label indication out there so that afterwards, everybody else can experiment and then you are in this funny gray land -- is that they can experiment but you cannot promote.  And this is itself, I think, a very serious cause because we know for example -- I guess it was Genentech which has a bastion on the market -- I do not run the stock market but the prices of stocks went up eight percent when one new indication was approved, which indicates that there is something very imperfect about an off-label gray market in which nobody could promote, namely, in effect, that sales efforts do matter.  What they do is they expand the user base, they get you more information, they probably lower the average cost of the drug and then they probably create greater profits for the companies at hand.

So how does one then think about this now?  I will just end, I guess about two more minutes? 

Male Voice:  [Indiscernible]

Richard A. Epstein:  I do?  Four more minutes -- okay.  Well, then I will continue to expatiate.  [Cross-talking].  No, I mean I try to keep myself within time.

So what does this then start to tell us?  Well, when you start to think about the FDA, what happens is you have to take into account not only the benefits of getting the information but the cost of acquiring.  And these costs are extremely hard.  I mean, for the first point, there is no particular reason to believe that when the FDA requires clinical trials, it is requiring the kinds of clinical trials for the kinds of indications or the kinds of uses that you would necessarily have in a voluntary market. 

In addition to that, one has to count amongst the costs two very important elements.  One of them is while this thing is being tried in effect, what happens is you are not allowed to use it freely in the marketplace so there is the real high cost associated with delay.  And even though patient groups are not here, it is very clear that while folks in the United States Senate may think that the major thing you have to talk about is strengthening the protection that the FDA will give us against crack drugs - you know, lousy products - most of the patient groups tend to argue in exactly the opposite direction and they are trying to figure out how they get exceptions for individual people so as to try something - when you are going to die anyhow, you may as well take a long shot.  And that constant trade-off is I think something, which is systematically underestimated in the way in which the clinical trials work.

And then, I think the next point that one wants to make is that there are all sorts of ways in which drugs are used and evaluated which cannot be captured in a clinical trial.  I know Anup is sitting here and for years he has constantly impressed upon me the importance of heterogeneity in trying to deal with drug evaluation.  That in fact you could have drugs that are going to be relatively powerful for a small portion of the population but not for a large portion of the population. 

And the question is:  What is the fastest way in which you could get it to that group?  The danger that you have with the FDA is when they keep something off the market, no sub-group can do it.  When you let something on the market, then you can get further investigation to figure out whether or not it should or should not be used in particular cases.  So you have to be, in my mind, extraordinarily confident before you keep something off the market, given the fact that there may be ways out there in which people can identify either through trial and error or through, in fact, some kind of theory the sub-groups for which it may be that these products turn out to be relevant.

And the third feature, which is extremely important is that when you are trying to test something through the FDA, you tend to test every drug in isolation.  I’m no cancer expert but I have heard the phrase “chemical cocktail” and what that seems to suggest is that when you are dealing with complex phenomenon, oftentimes the real art in oncology is to figure out how it is that you mix and match various kinds of treatment in combinations which themselves have never undergone rigorous kinds of clinical trials.  And if that is the way in which one is doing it, then it seems to me that one wants to encourage this kind of process.

So as I come to sum up on this, I then ask myself the next question:  Which kinds of organizations are likely to be more nimble in the way in which they adapt to various kinds of responses?  And when one looks at the FDA -- this is the way in which I think -- I regard the FDA as having an indispensable function that private voluntary organization can achieve and that is mainly dealing with the problem of counterfeiting, piracy, fraudulent drugs, whatever it is.  And as we look around today, the food supply and the drug supply have been attacked very heavily in those areas.  I want them working on that thing over time because I do not see how the private responses are going to be sufficient to handle what is essentially a policing function.

But when it comes to the sorts of things that we are talking about here, my own instinct remains that of the good classical liberal, which is government coercion is presumptively a bad, unless it could be shown to be a good.  When you start to deal with areas in which autonomous choices that individuals make with respect to their own welfare, it seems to me that the case for the government power is at a slow ebb. 

When you then add to that fact that it is not going to be individuals making autonomous choices in the blind but rather working through these complicated voluntary networks that are already there and which must be used even after the drugs have been approved, it seems to me that you cannot make out the case for monopoly power in the FDA.  If it wants to issue warnings, I cut off all the black boxes; maybe a government agency would have some modest use in this particular area. 

But in general, I think that the Hippocratic Oath applies -- or applies this much to the FDA as it does to doctors: primum non nocere, “First, do no harm.”  And when you believe in that mechanism, you do not believe very easily in giving any public agency a monopoly palette over something as complex, diffused, disorganized and important as healthcare including health care for treating cancer.  Thank you.

Tomas Philipson:  Thank you, Richard.  I had two comments, one is a suggestion and one is a question for Richard. 

Male Voice:  [Inaudible]

Tomas Philipson:  No, I will go in between.  One is a suggestion, which is essentially -- I think for an economist looking at the cost of R&D, I think Richard’s paper suggests new ways of doing that.  I do not know if most of you are familiar with the top studies that basically claim that R&D costs have gone up to $800 million recently and then over a billion in the most recent study.  I think those numbers are misleading if you look at the social cost of R&D which has to do with the delay of taking medicines.  I think those costs would be peanuts compared to the full social cost of R&D or delays and I think economists should try to incorporate, quantify those aspects of R&D costs due to the delay.

The question I have for Richard was - I am grappling with reading your paper - what is different about cancer and I was wondering what your reaction would be to that in some sense.  My interpretation is that the opportunity costs of treatment are so glim [sounds like] that you are on the verge of dying, so safety concerns are not really an issue because if we do not do anything, you are going to be pretty unsafe to start with.  So I’m wondering what you said in the talk is not particularly unique to cancer.  So I’m wondering what, if anything, is essentially unique to cancer in that area.

Richard A. Epstein:  I mean, I think the answer is and it is implicit if you look at the way in which the [audio glitch] away from cancer to everything else.  I think the problem is essentially not one of cancer in terms of the regulatory structure.  I think the problem is one -- when does state coercion intervene so as to upset individual autonomous choices.  And this would apply for thoracic surgery; it would apply to neonatology where you have family considerations and so forth.

So I do not think that it really is different and, in fact, I mean on this particular issue, I think it is clear that the FDA does not think that there is any different side or it does not take a different attitude towards its basic role when you start moving from one kind of drug to another kind of drug.  So the answer is, there really is none in terms of the way in which the structure is set up. 

In terms of dealing with individual choice, there is always a difference.  For example, trying to figure out whether or not you are going to take something like a statin in order to stabilize a condition like high cholesterol.  Your options may include exercise and less eating or a whole variety of other things, hence the options start to differ.  And so what happens is that the behaviors were different within the sub-areas.  But this is an argument again for weakening FDA control over it, if in fact the kinds of medicines that you regulate would play out so differently in day-to-day interactions with ordinary people.  You constantly meet the issue with, why does one formal regulation fit all types of situations?  And I did mention some of the differences there. 

I gather there is some direct-to-consumer advertising of cancer drugs, although I’m not exactly clear how that works but there is no question to me that a market which is operated mainly through clinical’s professional is going to present different problems of organization than one which deals with a lot of direct to consumers sales and so forth.

So I would think that the deregulatory message is one that carries pretty much across the board.  What will change is the form of private decisions and voluntary associations, and those things can be much more responsive to the differences that one sees in the kinds of risks that are posed by the use of various kinds of treatments and therapies.

Tomas Philipson:  Okay.  And also for everyone’s information, we are all going to have an audience participation after both speakers.  The next speaker - Richard Miller.

 Richard Miller:  Thank you and good morning.  That is a very tough act to follow.  I would like to move now from maybe some of the philosophical or theoretical issues to maybe more of the mundane issues, which are really some of the issues facing developers of drugs, which I have been involved in now for 25 or 30 years.

 Let me start with some of the statistics.  Cancer is a very common problem in our country.  There were 559,000 people who died from cancer in 2007; it is the most common cause of death for people who are under age 85.  Of course, if you are over age 85, you are dying of a heart attack or stroke or something else.  According to the American Cancer Society, mortality rates from cancer have been declining about one percent per year since 1990 but that is a little bit misleading because a lot of it is driven by improved imaging techniques for things like breast cancer - mammography being a big contributor there, colorectal cancer screening with colonoscopy, prostate cancer screening. 

So for many of the very important cancers - lung cancer, gastrointestinal tumors - we really have not been making that much progress.  And I think the final point - cancer has many diseases and this is something that we often forget.  It is not one disease; it is hundreds of diseases and they are not that common.  And they are not that common and within each cancer, you start looking at drug development for a stage or a class of tumor and then you start getting into really pretty small patient populations which make clinical trials and testing even more difficult.

 This is data from Parexel [phonetic] study; this is indexed, inflation-adjusted spending and biomedical research.  As you all know, we have been spending more money in biomedical research, both industry and the government, and in fact, industry and government over the past decade or so has spent about 150 percent more in biomedical research funding during that period of time.  The National Cancer Institute, for example, increased its funding from about $2 billion per year to about $6 billion per year although that has plateaued in the last few years.

 Now, unfortunately, we are not seeing an increase in new molecular entity submissions to FDA.  Those are novel breakthrough drugs, not “me too” drugs but drugs that have different structures, different mechanisms of action.  This is data from, actually, the Government Accountability Office -- submissions for new molecular entities, biologic applications have been decreasing.  And not only have these submissions been decreasing but approvals have also been decreasing.  And in fact, last year there were 16 new molecular entities approved by FDA for all areas of medicine and that represented the lowest approval since 1983.

 Now this is a paper - a very good paper by Moses, not the original Moses but just as smart maybe.  In JAMA [sounds like] in 2005, he looked at the anatomy of financing of biomedical research and very interestingly found is that from 1994 to 2003 for industry, research funding went from $26 billion to $54 billion.  And the amount of money spent in Phase 1 to 3 clinical trials went up from 28 percent as a percent of that total to 41 percent. 

So clearly, industry is spending way more money on clinical research.  NIH, for example, during that period of time when funding one from $13 billion to $26 billion -- clinical research was actually kind of flat.  I know most academic people who have been involved in clinical research are quite upset about that point and actually it gets to the point I think Tom made earlier or Jack made which is that we are spending a lot more on the basic science.  We are generating a lot of knowledge but we have not done a very good job of fixing the other end of that pipeline, which is how do we test these things in the clinic?

 The good news is that the research is paying off because if you look at new cancer therapeutic center in clinical trials, according to this Tufts study in late ’07, it is actually progressively been going up since 1990 in these four-year periods.  So you have a nice increase in the number of new molecules that are going into Phase 1 clinical trials for cancer.  Now since it takes about an average of eight years to go through a clinical development and get a drug approved, one would expect that you would see a displacement of this curve and you would be seeing more drugs approved if they are getting through the development and through the pipeline.

 Unfortunately, that does not really seem to be the case.  This is approved new molecules, new drugs for cancer and it is actually bounced around and pretty flat except for the decade 1990 to ’99 when everybody knows were the golden years for cancer drugs.  We have got a lot of drugs, a lot of new drugs that came about in the 1990s and I know this quite well because when I was in training before that period of time, we did not really have very many drugs and after the 1990s, we suddenly had a lot of drugs, so many drugs that I could not really even understand them all. 

Now, interestingly, in the 1990s, what happened is you had AIDS activism, you had accelerated approval.  Half of these drugs or almost half are accelerated approval, which we will talk a little bit about later.

 Now in this decade, we have actually slipped back down and in fact, in the last few years, there have been a decrease in the number of new cancer drugs approved.  So, at best, many people would argue that the 1990s -- maybe that is a fluke, a statistical fluke -- but even if you buy that argument, given that we have more science and more drugs entering the clinic, we should be seeing an increase in new cancer drugs, not a flattening.

 So let’s talk about some of the issues that are facing oncology drug development.  First point - and Professor Epstein mentioned some of these - oncology drugs are different.  Obviously, they are life-threatening diseases.  But even the FDA and others, of course, consider this because more often than other classes of drugs, oncology products do get priority or expedited review.  That is a faster review time; on the average, six months shorter, and that is of course because these are addressing life-threatening diseases.  Half the time, oncology drugs receive orphan drug designation getting to the point I made earlier that the individual indications for these diseases are frequently not that common.  So there is no question oncology drugs are different and they are regulated different and they are treated differently by the FDA, at least, with respect to their review times and orphan drug designation, et cetera.

 Now according to the Tufts Study, which came out last year, oncology drugs spend an average of 1.8 years longer in clinical trials than other drugs - 7.8 years versus 6.3 years.  So there is a little bit of a disconnect there.  We are reviewing them faster; we have laws to review them faster but yet, the average clinical time is actually longer.  Now the most frightening part of the statistic, however, is the next line, which again is the Tufts Study -- is that only 8 percent of oncology drugs that enter Phase 1 actually come out the other end as approvals - 1 out of 12.  And that compares to 20 percent for other classes of drugs.  Now why this is even worse than it looks here is that most of these failures occur in late clinical development as a opposed to early clinical development.  And of course, that comes after you have spent a lot of money and time and effort on that.

 So we will examine some of these issues but there is no question that there is increased bureaucracy involved in getting cancer drugs approved, not just cancer drugs but other drugs as well, but particularly in cancer.  And it is not just FDA issues.  Institutional review boards, university contracting grants offices are playing, I think, a big role in delaying some of these efforts.  Another point, which I think other speakers will talk about later, is that we are forced now because of what I have told you about these -- these indications are actually quite small.  We more often have to go to XUS centers to get clinical trials done and that raises a degree of complexity that is enormous.

 So what about the cost of drugs?  I think Tom mentioned this:  DeMasi Study several years ago from Tufts showed that it is about $800 million to get a drug approved, and Adams [phonetic] had a paper in Health Affairs last year -- actually DeMasi was somewhat off and Adams used a different data set, different techniques -- he comes out with a little bit over $1 billion and oncology is amongst the leaders in terms of cost and it is hard to see but cancer drug, on the average, he finds is about a $1 billion -- slightly above $1 billion. 

Neurological diseases, interestingly -- so getting drugs approved for things like Alzheimer’s and Parkinson’s is also very expensive -- is about $1 billion.  Respiratory diseases, about $1.1 billion but respiratory is a little bit misleading because those are -- you have a lot of drugs that go into testing but they usually fail in Phase 1.  You know in Phase 1 whether they are any good or not, whereas in the oncology area and neurological diseases, we do not find out until later.

 Now, what about institutional review boards?  There are a couple of papers by Dilts, who is in the business school at Vanderbilt who studied using these typical -- I guess they are typical kind of business school diagrams.  So he studied Phase 3 trials from the CALGB, that is a large cooperative cancer group -- these are Phase 3 trials.  He found out, on the average, from protocol concept, that means you already figured out what you want to do, to the time that you can enroll your first patient in the trial was 580 days median.  And it ranged up to over 700 days on the high end.  So it takes a year just to get these studies open. 

What Dilts found is that there are over 300 steps involved in getting a protocol through the IRBs [phonetic].  There are over 70 signatures required, there are 30 different groups of people who have to review it and not a single protocol, not one protocol makes it through the whole process without having to go back over and start again.  Now the sad thing, even sadder is that about 20 percent of the studies, even after they get approval, they never enroll any patients.  So all that work for naught.

 On the average, if you throw in Phase 1 and Phase 2 trials, which are generally faster, it takes about a year and a half from protocol to get it open.  And again, a large percentage of the time these trials do not enroll enough patients to learn anything.  So you can think about the drug development time when you think about the clinical trial period, keep in mind that you are probably talking about two years or so, which is just waiting around for higher beings to approve things where you are really not making any progress.

 Now let’s talk about testing outside the United States.  Because of the fact that these are small indications, because private practice, frankly, has siphoned off a lot of patients from academic centers, and because clinical trials are -- there is a need to make them larger now -- one must go overseas to conduct clinical trials.  And here is data from Parexel:  Based on 1572 - those are forms that you file with the FDA -- the clinical investigators file so they can get the information where these centers are. 

So you can see that there are a lot of other countries now participating in clinical trials that are FDA-regulated.  And unfortunately, it is a small percentage of a lot of different countries.  So you find that when you open a study now, you have to go get approval in Canada, in many European countries, Australia is not uncommon and this just increases the labor, increases the work.  But more importantly, raises the question of when you get this data, is it really representative of what the standard of care is in our country?  Is this really good for American citizens if we are basing clinical data based on clinical trial information from Poland, for example?

 By the way, it is estimated by 2012, that 25 percent of patients will be enrolled in India and China on trials.  There are 100,000 Chinese post-docs in the United States now in biotechnology and it is thought that this training et cetera will be transferred back to China and that would become a very good place to do clinical trials.  And it probably will be because in China and India, you have cheaper, you can do trials cheaper, but equally important is you have naïve treatment populations where it is hard to get more naïve treatment populations in the United States.

 So let’s talk about the regulatory system and how it is hampering oncology drug development and potentially stifling innovation.  There are a lot of issues here but I would break this down into three points.  And that is a one-size-fits-all approach; secondly, use of outdated statistical methods and clinical trial endpoints; and then finally and a very important one is the erosion of accelerated approval techniques or regulations.

 So “one size fits all,” what do I mean by that?  Well, I mean it is failure to put things in context.  Recent safety problems starting with Fen-Phen and Baycol in 1999, 2000, more recently Vioxx, Avandia, really have made regulators more risk-averse and they really had an impact on their behavior.  And while that may be suitable for drugs that are intended for benign conditions, it is not so suitable for diseases which are life-threatening and where treatment options are more limited.  During this period of time, in the last several years, there has not only been an increasing demand for pre-clinical experimentation with new agents, but also more lengthy and more numerous clinical trials. 

On the pre-clinical side, some of these things are actually devastating and there is no way to calculate the cost.  For example, drug developers now need to do in vitro test of drug interactions - actually they need to do this in the clinic as well - small studies, effects on liver metabolism.  One recent one is effect on cardiac physiology, the electrical conduction of the cardiac cells called the hERG assay.  Everyone knows this test is poorly predictive of toxicity in humans, but yet we are forced to do this test and we are excluding potentially useful drugs.  We just drop them before we even go to the clinic because they score in these assays.

 Clinical trial endpoints - you are going to hear a lot more of them; we have great speakers on this later.  But survival is a problem.  It is a great endpoint, of course; it is unequivocal; it is easy to measure.  It is obviously important but it is too hard to measure now and the main problem here is cross-over.  There are more drugs available to patients -- they get treated on newer study drugs and then they go off and get other drugs.  You have no control over that and it obliterates any potential difference that you would have in survival.  Not to mention we are getting patients on trials now who had been through 2, 3, 4 other therapies or more and it is almost impossible to think that you are going to make that kind of impact in that situation.

 And the final point here is that in some of these newer drugs, Avastin - you will hear about more later - Nexavar (sorafenib), they do not work by the classical mechanisms and some of the traditional oncology endpoints are problematic.  What about the statistics?  We have far more capable people to talk about this later but this is a really great paper by Kent and Hayward - a shorter version of this is in JAMA a few months ago. 

You will notice this is a picture that has 12 different individuals.  There is an old man and an old lady and an African-American and a Caucasian and in the middle, there is a computer simulation, a composite of all 12 people.  The clinical trials that we do now really test and express the data for an average person - that is the person in the middle.  That person is an abstract; he does not really exist.  Really, what we want to know is how are the people, the individual people, going to do and really the techniques that we use now largely ignore those subsets.  And that goes both directions.  There could be subsets that are benefited and there could be subsets that are maybe harmed by a treatment.

 Let me just talk about accelerated approval very quickly.  This came about in the mid-90s as a result of AIDS activism.  It allowed for approval based on more limited or surrogate endpoints.  We have 26 oncology drugs approved for 30 indications; many of these drugs are among the workhorses now of our therapeutic armamentarium.  In the years ’05, ’06, ’07, we had only two drugs receive accelerated approval.  Avastin just received accelerated approval in 2008 but really basically a contorted use of that regulation since Avastin was available -- it was on the market for patients approved for other indications. 

What has happened with accelerated approval is really a tragic event in that now before accelerated approval can be granted, you not only have to have a plan for your confirmatory trial, it needs to be started.  It needs to have finished enrolment and based on last year’s experience with four or five drugs, given that the trials were already enrolled, FDA wanted to wait for the results of some of those trials before granting the approval.  So, basically, accelerated approval has been completely dismantled.

 Now the FDA realizes this and in 2004, they issued their Critical Path Initiative, which was a great idea.  They recognize that the development path is inefficient and costly.  We need new tools to increase efficiency, biomarkers, assays, clinical endpoints, statistical methods.  The problem is, at the same time, we are reading that from the science subcommittee report that the scientific infrastructure at FDA is failing.  So at the same time, we are asking for new things and innovation and more sophisticated analysis and things.  We are really wondering in industry if you come in with these ideas will they really have the capacity and flexibility to deal with these issues.

 So let me conclude with just some recommendations.  We need a rational and flexible policy that takes into account the fact that diseases are different.  Cancer is a different disease than other things.  Clearly maybe the risk-averse behavior needs to be varied for different categories of drugs, kinds of treatments.  We clearly need new clinical trial methods.  And I think subsequent speakers are going to cover that in detail. 

We need to return back to the way that accelerated approval was originally intended for.  We need more input from academia and industry in the regulatory process.  It is impossible for a government agency to keep track of every new advance in medicine, statistics and science.  And if there is ever a place where you can apply these new methods, it is going to be oncology because of the life-threatening nature.

And finally, I think if FDA were to provide incentives for this innovation, for example, like, “Hey, we want you to work on pancreatic cancer and we are going to work with you and we are going to be more flexible about how we regulate these things,” I think that would be a tremendous stimulus to working on these cancers where we have really no good treatments at all.

Thanks a lot.

Tomas Philipson:  I have a question for the second Richard which is -- I wanted to play devil’s advocate and see what the response would be from him or others taking that position.  You showed that basically new cancer treatments have doubled going from 150 to 300 coming in and being investigated.  And the way I think of clinical trials is essentially a labor market for subjects where subjects supply observations and investigators demand them essentially.  So it is a little strange labor market because there is a maximum wage of zero imposed by regulation even though these subjects provide what economists call positive external effects in the sense that we learn from their consumption, so you should actually stimulate that consumption. 

In that labor market, you have an increase of -- a doubling of the demand for observations, in some sense, coming in with these new treatments.  If we assume prevalence have not changed as much, let’s say the supply of observations has remained constant, would not a natural consequence be that the speed of doing a trial when you cannot pay people to participate would dramatically be reduced?  It looks like that is what is going on because we are going abroad for subjects as well because the domestic supply is essentially limited. 

And how do you separate that reduction in speed by the mere fact that you have more demand for observations than supply now from increased sub-group analysis -- whatever regulatory changes that you argue taking placing slowing down the process?  And could that be done in a quantitative fashion, separate those two forces out?

Richard Miller:  Are you done with that question?  Okay.  Well, that is a kind of a typical question you might expect from the economist.  So I’ll try to answer that.

I think you are quite right.  One of the problems with enrolling patients is that there is more competition for patients now.  And some of these diseases which are quite popular - and maybe this is a good thing -- so when a drug gets approved for a disease -- like take a Rituxan for lymphoma.  Before Rituxan, nobody really wanted to work on lymphoma and I know that because I was involved in that.  No company wanted to work on it; everybody thought that we have plenty of drugs for lymphoma.  Finally, Rituxan is approved; it does well in the marketplace and then everybody wants to study lymphoma.  You have dozens of companies working on lymphoma treatments and then it starts to get competitive to enroll patients on these trials.  So I think you are quite right.  That is one of the problems.

But at the same time, the number of patients required for NDAs now has almost tripled if you look over the past 20 years.  The number of patients on trials has tripled; the number of trials has about doubled like 30 to 70 for an NDA.  So I think all those things are working together.  And in terms of going overseas, I think part of it is driven by cost as well.  It is cheaper now to do some of those trials overseas.  Unfortunately, that is one of the motivating factors, but part of the cost there is also the enrolment is so much faster.  I mean, for a company, time is obviously money and if you can enroll the trial in half the time; that is important.

And then the final point about that is I think the treatment of naïve subjects is a big factor.  For some of these indications, it is really hard to find patients that are suitable for the question you are asking.  And at some of these Eastern European countries, for example, it is easier to find patients.  A good example is second-line treatment of lung cancer.  There are more treatments for lung cancer now and if you wanted to study a new drug for second-line lung cancer, it is actually hard to do that in the United States because there are two or three other drugs that usually people are getting and it is just a lot harder.  But if you go to Eastern Europe, Poland or something or Czechoslovakia, these people cannot get access to some of the second-line therapies.  And if the company is willing to pay, these patients sign up quite readily.

So there are a lot of factors at work here.  It is not just regulatory.  But what I would say, though, is that we have to take this into account as we try to determine what is an acceptable clinical trial.  How do we do that given the complexities?  So, you know, I’m like Richard Epstein.  I would say randomized control of clinical trials is a good thing.  I was taught that way when I was a medical student.  But sometimes and more often now, it is getting very, very difficult to do that.

Richard A. Epstein:  One simple observation.

Tomas Philipson:  Sure.

Richard A. Epstein:  One of the problems that you get with randomized clinical trials is it reminds me like a Ford Motor plant assembly line in 1932.  You have thousands of people doing the same thing and relatively little variation.  But two things have happened: one is that there are many more efforts to try to get drugs in the market place so there is more competition for the subjects.  And secondly, it turns out that as specificity becomes a dominant mode of biological research, the heterogeneity of a population really matters because there are lots of people who are no longer suitable for it. 

So if you are demanding larger ends when it turns out that each condition turns out to be more discreet, you are not going to be able to find them because instead of being able to pick one person, you get 10,000 people who are suitable if you really have to find your biomarkers with sufficient specificity.  Out of that group there may only be 300 or 400 that are usable. 

And you have to think of other things and the way I would put the point is this:  As knowledge of mechanism starts to increase, we have to treat that as a substitute for clinical trials and validation.  And if we continue to hold stricter demands on the ends at the time that we have knowledge, which means that we have smaller ends that are suitable for population, the whole system is going to crash.

I would just say one thing by way of comment on my friend, Richard here.  I think he is a hopeless optimist.  There is no way that you can get the level of flexibility that you want in any regulatory system that has veto power, period.

Tomas Philipson:  Okay, we want to bring in the audience here and so if anyone has questions, we like you to state your name and institution and also wait for the microphone, which should be somewhere around here.  Okay, there is the microphone and there is a question over there.

Diana Furchtgott-Roth:  Thanks very much for that very, very informative set of presentations.  I have a question for Richard Epstein --

Tomas Philipson:  Can you identify yourself?

Richard A. Epstein:  We want to know who you are.

Diana Furchtgott-Roth:  Oh, I’m Diana Furchtgott-Roth from the Hudson Institute.  And I have a question for Richard Epstein. 

I was just wondering -- I’m very sympathetic to your view, having many friends who have cancer with all the barriers that FDA is putting up for approval of the drugs.  But on the other hand, there is such emotion that would you not get if there were no barriers at all?  What would you do about the problem of people taking therapies that were just completely useless such as apricot pits and things like that?  Because you clutch at straws when you are in those last days and how would you stop people exploiting these kinds of patients.

Richard A. Epstein:  Miss Diana, the problem is not solved even by a system of FDA approval.  You can get all sorts of things out there and people will do hopelessly stupid things with approved drugs when they are counter-indicated.  So keeping drugs off the market does not solve the problem of patient irrationality.  The only way that you can solve that is have other institutional mechanism.  And the reason why the removal of the FDA is more attractive for cancer than it is, say, for consumer-driven products, is that there is always a professional intermediary in the United States. 

And what has to happen is that the patient has to persuade the physician that the treatment that is proposed is going to make some kind of sense.  And if in fact, you as a physician can say, “Look, it is not just some lumbering government bureaucracy which decides if this is or not appropriate, but I have managed to get -- and this is important information from everywhere on the globe, which is these networks are global in many cases whereas the FDA is not.  And everybody says, “You are going to do is kill yourself faster with this treatment; do not do it,” that is going to have a lot more power, I think.

One of the problems that you get with the FDA is it has a peculiar kind of bi-modal sentimentality out there.  There are some people who think it is like a Good Housekeeping seal of approval and so they are going to, “Oh my God, they give this waiting to it.  We will take it very seriously.”  And there are other people who say, “Well, this is the gang that cannot shoot straight and if they want to recommend something, I’m going to do the opposite thing.”  And my view about it is that you just want to put them to one side in the whole patient-management function and deal with that through doctors, social workers, family members and so forth. 

And to try and have upstream regulation of what is a quintessential downstream problem, it is impossible.  And if you were correct, you would not allow off-label uses, if this were a concern.  And you cannot believe that -- I mean, is there anybody in this room who wants to say that off-label uses on aggregate is such a mistake that they were to be banned?  And when I start hearing that, then I will start thinking about your objections being one for strengthening the FDA as opposed to dealing with other mechanisms.

Finley Austin:  Finley Austin with Roche.  The Cooksie [phonetic] review in the U.K. last year posited a sort of -- accelerated into the market with post-market surveillance approval.  Your thoughts on that mechanism, both Richards, and is that something that is maybe viable or is that just to differentiate a lipstick on the [indiscernible].

Richard Miller:  I think the Europeans are ahead of us on that.  So we had accelerated approval, of course, but one of the problems - and I did not have time to go into this - is that some sponsors were delinquent in performing their confirmatory trials.  Now there are many good reasons for some of that and I think one of the speakers later is going to talk about that.

So the European system is you can get accelerated approval, which just started about a year ago, maybe one year ago or two years ago.  But it is automatically taken off the market in one year unless you actively file an application to keep it on the market; that is, supply some data.  So the default position there is if you do not do it, you have to do it as a sponsor that is coming off the market where, of course, the problem, many people argue with our accelerated approval system here is you get it on there and then, really, you do not have to do very much and it stays out there.  In fact, people argue there is an incentive not to do anything because you might get the wrong data.

So I think the European system is actually good and it is ahead of us in some ways.  But it is really too early to say, because I think there is only one or two drugs approved now by Accelerated Approval in Europe.

Richard A. Epstein:  I have a different reaction, which is not to oppose what is going on in Europe if it will break the law.  But to ask the following question of you, which is, why is it when you want to do post-clinical studies of observed drugs that the drug company is the appropriate entity to do it?  I mean for me, I would much rather have an independent board of physicians put together from all walks of life collecting the information than the drug company.  And in fact, much of the work that is done on off-label drugs is, in fact, has to be done in some cases by non-parties.  And so I would encourage again the use of independent mechanisms even within a regulatory framework and certainly without it.

I think it is very difficult to get unbiased information by a company, which has a very strong financial stake.  I think it is almost humanly impossible to separate it and so what is so ironic about this, instead of lifting the function from parties who have the built-in conflict of interest, we mandate that they do it and then we oversee them and then we get really angry at them if it turns out that they take some of these data and they shade it in a way which we disagree with. 

I mean, one of the things that is striking about this - and this is the second half - after you get this information, there is nothing which says it is going to be a unanimous sentiment on the way in which it ought to be used.  All of these ODAC [phonetic] committees -- in fact, they always wrote 7, 8, 9, 4 or whatever the hell it is.  And under these circumstances, if you are going to get that kind of divided sentiment, it seems to me that all you were to do is to collect and organize the information and let the dissenters do with it what they will.

So I do not think that the coercive mechanism should be imposed even in this post-regulatory framework, although I do agree it would be better to let the thing on the market first, get some data than it is to keep it off until you try to meet these impossibly high standards for clinical trials.

Warren Greenberg:  I’m Warren Greenberg from George Washington University.  And I just wanted to focus on Richard Epstein’s stimulating talk.  You have an interesting way of using euphemisms like voluntary organizations or institutional mechanisms.  Why use these euphemisms when we have for-profit health insurers out there?  To what extent do you think that for-profit health insurers can help us with this information if they are indeed out there to provide lower cost?  And if we provide incentives for them to provide higher quality, why can we not rely on these health insurers to give incentives for lower cost and providing incentives for higher quality to do the job that you are talking about?

Richard A. Epstein:  Well, I’m certainly not -- let me put it to you this way:  I absolutely agree with that market -- the question that one is going to have to do is the following way.  If you are going to get rid of structural impediments from regulation, you are going to have some voluntary thing.  Your argument is do it through the insurance companies.  My answer to that would be if I were an insurance company doing this, I would then have to figure out where it is that I’m going to get this expertise and, in many cases, one of the things that I would do will be try to rely on the way in which the various organizations in the professional societies get in to organize this information.

But the second half of the argument is this is already taking place again.  Anytime somebody wants to sell a drug to an HMO, almost invariably the drug companies present to them data which is different from that which is required to be presented by the FDA.  So that voluntary market is out there.

And when I was talking about decentralization, I was giving this as an instance as to why it is that.  I think these organizations have a useful role to play.  But the whole point about decentralized authority is that anybody who is going to be a buyer to put up whatever obstacles they want to the purchase so long as they do not impose the obstacles on others.  So the more, the merrier.  If it turns that these companies can, in fact, organize other devices for the collection and organization of information, that is fine.

And let me just mention because it is a nice theoretical point here as well.  The usual model that people have with respect to the disclosure and the discussion of medical information is to treat it as a public good and then to try to figure out ways in which you can collect it and to broadly disseminate it.  And for that particular function, something like the NCCN is, I think, pretty well suited.

An HMO in effect -- when it gets information about a protocol and how it ought to treat and develop it, its tendency may well be to treat this thing as a trade secret.  To say, “Look, I have got a comparative advantage over somebody else because I can now supply services at a lower price.”  And I think that is actually a good thing as well because what happens is you will not get the information out of a profit-making institution on the same terms that you will get it out of one of these voluntary professional organizations. 

And it may well be that you are going to have the usual problem, which is by allowing trade secrets, you will have a large stimulation for the production of information but a narrow dissemination of the information once it is collected, which will force you to the additional cost of duplicate information.  You are going to have to get every HMO, which is going to be working its own protocols.

And my own view about it is that is the way information markets work.  It is that sometimes public disclosures are optimal, sometimes trade secrets are optimal and as a lawyer, the simple observation is patents have never displaced trade secrets and trade secrets have never displaced patents in the operation.  So my view about it is, join the party.  I mean, it is just another place at the table and if you are right - and I think you are - that is just a further reason for saying the last that we want to do when we have multiple mechanisms to review this information is to give any government agency, any bureaucracy a veto power over the ability to get stuff into the field.

 Paul Howard:  Paul Howard, Manhattan Institute for Policy Research.  Is this an issue at which the economists are going to have to step forward and quantify what is happening off-label then try and quantify what is happening off-label on Phase 4 and put some value on it and understand it the way that they can go and turn around the FDA and say, “Here is what we see once the drug is on market.  Here are the benefits you are not capturing in your Phase 3 testing.  Perhaps you should scale that back and look at getting these to market faster because we absolutely know there is much more value in Phase 4 versus a Phase 3.”

 Richard Miller:  I think that is a great comment.  As Richard Epstein mentioned, there is a lot more to be learned after the drug is approved.  And I think that if you look at the history of oncology drugs, we get these things on the market with a relatively -- with not that much information behind it.  And then when it is out there, various government-funded groups like National Cancer Institute Cooperative Groups and other -- and universities who write grants on things get a lot more data and generate really the useful information about the drug and how it is performing.  So these initial indications, of course, are often quite limited but then when the drug gets out there, there is a lot more work done and we really figure out how to use these things.  So to the extent that the economist can figure out the most efficient way through that would be good.

One could argue that right now maybe the incentives, the initial incentive to do the smallest trial possible, get it out -- you know that companies have to get it out there the fastest maybe for the most unusual or bizarre indication is really counter-productive.  And really there is the idea how we get people to tackle the serious problem like metastatic pancreatic cancer, which nobody wants to study because we do not have any good treatments and it is too hard to make an impact. 

So I think that oncologists, in particular, tend to be very narrow in their viewpoint on these things, were taught that randomized controlled trials, highly significant p-values, et cetera is the only proof you ever have.  And I think that plays out for a lot of us, and worse, oftentimes not thinking about the other parts of this puzzle that have a big impact -- almost a bigger impact on patient outcome.

I know the statisticians are going to talk somewhat about this later, but I think that is a great idea.  We need more multi-disciplinary approaches to how we use clinical trials and how we use these drugs out there.

Richard A. Epstein:  Paul, I can even go one step further.  One of the things I keep asking people is you have this huge semi-underground economy of off-label uses.  And has anybody actually compared the efficiency of drugs on off-label users as relative to those that are on-label users?  Now we do know we have repeated usage of very high volumes.  We know that this is done independent of drug company promotion because that gets you put into jail.  What one would like to do is to see whether or not adverse incidents, survival rates or anything else that you are talking about comes out better or worse with respect to these drugs than the other.

As far as I can tell, forget about the economics of this.  I do not think anybody has actually studied the medicine on these kinds of things.  I do not believe that drug companies would feel comfortable going out and trying to organize this information themselves if they are flirting with felony every time they want to engage in that stuff.  I think the learned societies do some of this information but as best I can tell they are limited to the extent that they can only collect information, refer that as a clearing house for the most part -- and maybe as an evaluative [sounds like] agency, but I do not think they do very heavy clinical lifting.  I do not think they have to budge et cetera after that.

And my sense is that that is where all the game takes place.  Obviously, I will make the point again: you send the drug out there; you test it for A; you put it out there in the field, then you realize that standing alone is not very effective.  But if you combine it with B, C, and D in a certain class of patients, all of a sudden, this stuff takes off.  The word spreads underground, everybody starts using it in this particular fashion.  The FDA never gets involved in this.

If that is the way knowledge is disseminated, then the key thing is after Phase 1 trials, which I think are trivial for the most part, you want everything to go out into the market.  And what you want is a system of certification, which says, “This thing has not been approved in clinical trials.  Abandon hope all ye who use it if you want.”  And my guess is you will get the information out there in usable form. 

So yes, I think, in effect, that you solve all of the clinical problems if you do not have to worry about getting naïve subject, whatever the world they are.  You do not have to worry about enrolment; you do not have to worry about IRBs.  Essentially you get patient demand driving and that gets you the volume.  And I would rather have a large volume of imperfect information than a very tiny volume of pristine people because the small end is going to kill the validity of any study that you have out there.

So I think this whole system is so nutty.  I really can understand how we talked into it so I will give you what I think the explanation is.  If you are high [indiscernible] in this point, you believe decentralized knowledge through accretion and hunch actually makes a difference.  If you are a French bureaucrat, you think everything has to come from necessary schools with the various kinds of decal [sounds like].

The FDA works on this progressive, centralized, bureaucratized approach.  And what happens is that the heterogeneity and the variation in the profession moves so fast that they cannot possibly keep up with it.  So every one of their failures leads not to a call for deregulation but for more regulation.  You are feeding sugar to a diabetic, that is what you are doing I think in this situation.

Jean Montgomery [phonetic]:  Jean Montgomery.  I’m fascinated here with the thought of substituting private market for the FDA.  I’m wondering from Richard Epstein’s point of view if there is anything that you could see now looking out on the landscape given that we have Internet availability of information.  Is there anything you see here that would promote the effectiveness of this channel of dealing with these kinds of drugs?  One of the prime things I’m thinking of here would be mandated follow-up on people and more consistent health information systems to track what happens with various drugs and under all the other circumstances that are affecting a given person.

Tom Philipson:  Could you also identify yourself for the Webcast that we are doing.

Jean Montgomery:  Jean Montgomery.

Richard A. Epstein:  I’ll answer that question.  My view is information becomes essentially cheaper, becomes global.  The case for centralized authority, particularly on a mandatory basis, becomes weaker.  And I agree with everything you said with the exception of one word, which was the word “mandate.”  I do not see how that fits into the particular situation as I argued in this paper.  You just take any standard estimate of what the value of a life is, and then you try to figure out how valuable the information is if it is going to change success or failure rates by a month or by a percentage point and so forth and you get big numbers up and down the line.  There is a lot of incentive for people to kind of put this information together.  And we encourage both the open and the private kinds of systems and let the whole thing roll.

What the FDA does is it ends as a kind of front-end barrier, which means that all these other processes can take place until they get their act together.  And I think in effect that the destruction and information is the key variable here.  Tom, I think, agrees with this, rather than the actual varied kind of monolithic estimation of cost, which says, “Look how much it cost to get a drug through a drug company.”  Those are real numbers.  But my guess is that the social cost is probably ten times that amount if you take into account lost opportunities, lost information, inability --

For example, I’ll give you another cost.  You got a drug that is already out there in the public domain and it is going to be twice as useful if some other drug can come out on the market.  When you keep the first drug from setting up and coming out on the market, you reduce the value of the drugs that are in the existing open stock.  These are real costs.  What you will have to do is to get people who think about this in a holistic way.  And as best I can tell, whenever you work with a bureaucratic situation, you got this built-in conflict of interest in which Senator Grassley is only going to ask you about certain kinds of failures.  He would not even understand what the other kinds of failures are and he is certainly not going to berate when they happen to occur.

Tomas Philipson:  One last question and then we will -- we are ending a little early at the request of Dory Aisen [phonetic].  I do not know.

Bob Gibson:  Thank you.  I’m Bob Gibson from the National Academy of Sciences.  I wanted to posit an idea particularly to Mr. Epstein and have him respond.  And I think the idea is that there is a fallacy in his argument in that this is not just about individual action.  This is not motorcycle helmets or something like that.

What we are talking about in this field is people potentially being exploited by thousands, hundreds of thousands of charlatans and sham organizations, not just in this country but in China, Mexico, whatever who are taking advantage of people who are in their most extreme form of vulnerability that people can be in.  And so this is not about just individuals taking control of their own situation but -- so I would argue that we have two choices to regulate this type of exploitative behavior.  One is to control it at the spigot - sort of like the FDA is doing - or establish a huge bureaucracy to fight the fraud and to fight the illegal and other behavior, in that it is taking place in the market place after potentially great harm is done.  And so I for one would favor fix the FDA rather than set up this whole new bureaucracy to fight the fraud and abuse.

Richard Epstein:  I think the question is completely misguided, but let me see if I can give you what I think is the answer.  In part, it is the same question that was asked by Diana at the beginning.  You have the FDA; it does not stop the fraud problem.  There will be charlatans out there up and down the line with respect to approved therapies, different combinations, going overseas, whatever it is.  The reason why it is that the abolition of FDA approval in this thing matters is that you cannot get -- what is best I can tell – over-the-counter cancer treatments.  You have to go in the United States to an oncologist.  And if that is not sufficient protection, then I do not know what is.  That is the first one.

The second point is that you do not need a bureaucracy to deal with the fraud question assuming that you have individuals who suffer major losses out of really heinous behavior.  These are libertarians; this is not something very exotic and if you have somebody who has suffered major personal losses, a private right of action against those kinds of characters will work.  If they are so evanescent that you cannot ever find them, then it is not going to be the kind of large institutional problem that you are talking about. 

So I just do not see why it is that you need to stop stuff at the spigot because it seems to me that self-protective actions on the downstream side are going to be better and that what you really want to do is to make sure that families get involved, information gets put on the Web, oncologists are consulted before you get treatment.  I am just not aware of any evidence of the kind that you are talking about which indicates that when people go in for experimental and compassion and exemptions, that they are always thoroughly misguided and completely mislead and goaded into some kind of relief.

I think the consumer protection rationale is the weakest rationale for running the FDA.  What you really have to do is to show that there is better information at the center than there is at the periphery.  And if fragmentation and specificity becomes the new watch words of medicine, the chances that centralized control will do well with a thousand different conditions as opposed to a hundred specialized agencies, each of which deals with them, strikes me as being mythic in the highest level of proportion.  This is basically a try to run department stores in an age when you have malls and boutiques and the model just does not work anymore.

Tomas Philipson:  Okay, with that I thank the speakers and we will go on to the next panel.  Thank you very much.

Panel II:  Clinical Trial Design

Bruce Cheson:  Everybody ready?  At least in this session, we will try and be controversial unlike the last one.

I’m Bruce Cheson.  I am from Georgetown University Hospital, and I have had a somewhat checkered career.  Yes, I spent 18 years at the National Cancer Institute where the branch I was involved with was responsible for coordinating all of the NCI’s sponsored clinical trials around the country and the world.  So I have seen the problems at that end.  I have been on the Oncologic Drug Advisory Committee as has my colleague, Dave Alberts, seen the issues at that end.

What we really need to remember here, and several people have reminded me this, is that what we are here for is really for the patients.  I have several in the audience who came up to me and reminded me of that, which I hopefully do not really need reminding of, and how to get good drugs to the patients as quickly as possible.

There are a number of issues with clinical trials, which this group has lots of experience with.  You will hear things that will, hopefully, help you distinguish studies, which are clinically significant from those, which are clinically important.  New concepts on how to design studies who might get drugs out there more quickly and, personally, I’m in favor of not getting drugs out there that do not work and, I think, this should be the father for a very lively discussion and rather listen to me baffle - I’m going to turn the podium or the chair, where everyone wants to be, over to, my old friend, Dave Alberts, who is the director of the University of Arizona Cancer Center; chaired the Oncologic Drug Advisory Committee, a number of years ago; has had a long and distinguished career in new drug development cancer prevention and control, and is considered one of the leaders in our field.

Dave, take it.

David S. Alberts:  -- but in terms of advancing slides – oh, I guess that is -- that is what I got.  Okay [cross-talking]

I was reminded about how wet I am when I opened the bottle up and it went all over me so I'm starting from a baseline of being all wet.

I, actually, was smiling during Mr. Epstein’s presentation because when I was the chair of the Oncologic Drug Advisory Committee between 1982 and 1984 and we were not getting drugs approved at that time.  I actually would go onto public session after we had talked about our drugs and make the comment that should we not be considering approving drugs that had proven safety in Phase One and early Phase Two and showed some biologic activity, i.e., at that time, biological activity was subjective partial response or complete response measured in two dimensions, and then, have a responsibility of oncologists to collect clinical data and report clinical data in orderly fashion so that we knew what the denominator was for off-label use and we knew we had an understanding of what actually was going on with drug usage and, of course, the FDA as soon as -- frankly, I was expunged in 1984 - never to return again.

Does not seem to be -- there, I have to turn it on.  It is a good idea.

I would like to tackle a little bit about the cancer problem - very little bit about that because it has been gone over very well by Dr. Miller.

I want to talk about a typical cancer patient that exemplifies a lot of what has already been said but, I think, relates to the problems we have with drug approval.  Levels of evidence required for drug approval - I would not spend a lot of time on facts about cancer drug approvals because we have heard them.  The classic drug development process and do we need new clinical trial designs and what are the next steps.

Again, from the perspective, when you say the perspective of a patient – frankly, we are all patients.  Basically, one and two to one and three of us is going to have a invasive cancer and there is in the family that is not touched, as you well know, but there is someone we are talking -- this is a very personal problem - very, very personal problem for everyone of us in this room and that is, I think, why we are here.

The facts are the facts; we lose, basically, the state of Rhode Island every year in terms of population - one death per minute.  Even though we have been saying evidence of a reduction in mortality, that is not going to be long lasting because we have an aging population with higher rates of cancer.

We have hardly made a dent and, in fact, one of the reasons that I have been so focused on cancer prevention is that a lot of the reductions that we are saying are because of early detection - better screening methods.  It is not that I’m not a drug developer, I am, but I think we need to be focusing not just on approval of cancer drugs for treatment but for cancer drugs for prevention.

This is a typical patient history, I happen to have a practice of gynecologic cancer, and this is a typical patient that I’ll see in the clinic after they have been through six or seven different treatments because we are a tertiary referral center.

This is a 61-year-old woman with Stage IIIC ovarian cancer diagnosed at age 61 and she underwent to an exploratory laparotomy - a debulking surgical procedure, and she went on to first line chemotherapy in the community with carboplatin/paclitaxel.  She should have been treated with intraperitoneal drugs, which, by the way, not a single one is approved, but I’ll show you data later to suggest why every woman who is eligible should get intraperitoneal drugs instead of intravenous but that is not happening in the community.  Why?  Because we have not taught appropriately how to do this.

She is in a complete remission and then, she gets consolidated with intraperitoneal fluorodeoxyuridine, which is total off-label use of the drug; it is normally used for intrahepatic infusions for colorectal cancer, metastatic to liver.  After an 18-month period of remission, she recurs and she gets re-induced with another intravenous regimen of carboplatin and paclitaxel.  She gets a four-month remission, which by definition is represented to what we call “platinum resistance,” i.e., recurring within six months of finishing complete treatment with platinum-based regimen.  And then, she is treated with gemcitabine, which does have a label now for ovarian cancer but not as a single agent, so it is an off label use.

After eight months of treatment, she goes on a clinical trial of bevacizumab plus erlotinib – it happens to be a trial.  We just completed Phase II Trial with Genentech as an investigator-initiated trial at their cancer center.  But she, in fact, was VEGF negative and the EGFR negative and she did not respond.  She then has a clinical partial remission lasting 18 months on pegylated liposomal doxorubicin - that is an indication for use - and then, stable disease on capecitabine for 10 months - an off-label use - and then a clinical complete response on microalbuminated [sounds like] paclitaxel that is ongoing at 14 months - again, another off-label use.

So most of the treatment program - and this is now 62 months if you add it up - is off-label use of cancer drugs dictated by what?  Dictated by market forces, dictated by small Phase II publications but, in fact, what we were talking about is happening.  Most of this is not based on randomized clinical trials - the usage.  For me, as an investigator, the real problem is not collecting the data - we are not collecting the data.  We need to have an information database that informs our decisions.  So what is next?

Well, let me just go over of some of the clinical lessons learned in the management of ovarian cancer.

Number one, clinical response and progression-free survival correlate with overall survival – absolutely - and I’ll show you some data.  So if you just focus on the bottom two Gynecologic Oncology Group Studies, these were both Phase III trials - the first one, GOG114 compared in intravenous platinum-containing regimen to an intravenous and intraperitoneal regimen and there was a progression-free survival advantage - 22.5 months versus 27.6 months, and an overall survival advantage, which was at the P 0.05 level.  It was just barely reaching a statistical significance but there was an 11-month difference in survival.

But here, progression-free survival correlates with survival advantage.  GOG172, which was published in 2006 in the New England Journal of Medicine compared to, again, another intravenous regimen versus a combination of intravenous and intraperitoneal drug.  THE progression-free survival - there was an advantage - as you can see - 19 months versus 24.3 months and that a very major overall survival advantage of 16 months.

Here is a disease with - there is no question - every one of these clinical trials has shown an improvement for the experimental regimen has shown progression-free survival is, in fact, correlated with survival, and, in fact, responses correlated with survival.

For the majority of cancer drugs, clinical response and prolonged progression-free survival also correlates with quality of life improvement.  Of course, if you are going to be treating a patient and she goes into a remission for two years, you can expect that there is going to be a quality of life improvement.  My point would just simply be that progression-free survival is a very reasonable endpoint for our clinical trials - not waiting the six years you would have to wait for a survival end point.

Furthermore, survival endpoints are really problematic as Richard Miller pointed out.  If, in fact, the average ovarian cancer patient is going to be treated with six to 12 different regimens, you are, basically, saying that the first line treatment has to overcome all of the crossover that occurs during that time, and it basically says that second-line therapy has no impact on survival.  So survival as an endpoint in ovarian