Friday, February 16, 2007

Session 5: The Genomic "Revolution" and the Practice of Medicine

Francis S. Collins, M.D., Ph.D.
Director, National Human Genome Research Institute
National Institutes of Health
Bethesda, Maryland

DR. PELLEGRINO: Good morning. Our speaker this morning is Dr. Francis Collins. It's been customary, Dr. Collins, not to go into a long dissertation on people's accomplishments. In your particular case, that would be certainly overdoing it.

We don't have to say very much. No one is better known than Dr. Collins in this field. And so I just want to express our delight and our pleasure and our honor of having you with us. And you take off in the direction that you have indicated in your title.

Following Dr. Collins' presentation, the discussion will be open by two of the members of our Council: Dr. Janet Rowley and Dr. Bill Hurlbut.

Dr. Collins, the floor is yours.

DR. COLLINS: Well, thank you very much, Ed.

DR. COLLINS:  It's a real pleasure to have a chance to address this distinguished group.  And on this rather cold morning, nice to see you all here looking so bright and ready for an interesting discussion.  That's what I'm expecting will happen here because I think the topic is certainly a timely one.

I know that in November at your meeting you heard presentations from Kathy Hudson and from Bob Nussbaum on this topic of where genomics is taking us as far as its implications for medicine.  And I will try not to duplicate very much the materials that they went over, although there will be some unavoidable overlap of a sort.  But I did have a chance to look at their presentations.  So I think I know some of the ground that has already been covered.

So I thought what I would do is particularly this morning to focus on the science, to tell you about things that are happening.  And they're happening quite quickly in the applications of genomics to understanding the hereditary factors and somatic factors that play a role in health and disease that are moving rather quickly in the direction of application in the practice of medicine, maybe not as quickly as some of us would like because we're impatient but certainly more quickly than I think a lot of practitioners are aware.

This is I think, therefore, an appropriate topic for this Council to consider because it also carries with it a lot of ethical, legal, and social issues, which I will try to mention a few of those anyway towards the end of the presentation.  And then I hope we can have a vigorous discussion about all of this.

So we do, after all, have the circumstance now, a rather historic one, of having our own human sequence accurately determined and sitting in an internet-accessible database as a consequence of a Human Genome Project happening.

Interestingly, just about exactly 50 years after Watson and Crick's description of the structure of DNA, now in April of 2003, all of the goals of the Human Genome Project were completed, including this 3.1 billion base pair, human DNA sequence deposited all the way along by more than 2,000 investigators, who worked together to achieve this goal in a really remarkably efficient and selfless way.

But, of course, we are just beginning readers in a certain way when it comes to trying to understand that genome sequence, our ability to look at those A's, C's, G's, and T's and figure out what information is encoded there.  And how it all works is moving forward, but we still have a lot to learn to try to understand how this script gets read and how sometimes glitches in the script result in disease risk.

I think many people, sort of excited as they were about the consequences of having our human genome sequence, now in 2007, 4 years later, are saying, "Where's the beef?  Why haven't we completely transformed everything that we are doing in the medical arena?"  And anybody working in the field would have told you that wasn't a realistic expectation, that this was a foundation.  But now we have to build on it.

And I think this is a pretty good representation of the first law of technology, which is we tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.  And I think that will turn out to be true for genomics and we can decide what is the definition of short run and what is the definition of long run.  But this law, which is played out in many other instances, I think will play out here as well.

So today I am going to talk to you mostly about things that are present now in the research arena, which I think will find their way into the clinical application, but some of them are already there.   And clearly, as we try to anticipate the consequences of all of that, the time is now to pay attention and try to put into place appropriate considerations to be sure that this is all done in a benevolent way, that it helps people avoid or deal with disease, as opposed to putting them at unnecessary risks.

So many of us do have this future dream that medicine will, in fact, move into a phase where individual information can guide the process.  You wouldn't think of buying shoes in a single size.

I'm quite sure that all of you here when you went to the shoe store did not just randomly pick a pair of shoes off any old shelf.  You looked for the section that had your particular size.  And then you applied other criteria to what you might want to pick out.  But you certainly if you wear a size seven didn't start in the size ten section looking for what would fit you.

And, yet, for medicine, we have largely been forced to do that kind of generic approach because it was what we had.  It was the best information we could go on.

We did some tweaking of that, you know, trying to adjust drug dosages by body mass index or body surface area and certainly treating children different than adults most of the time, although not always.  But it has been still very much a one size fits all kind of approach in terms of the diagnosis, the prevention, and the treatment of disease.

And since we know we are all different individuals with different risks, this is not an ideal circumstance.  And one of the promises of genomics is to try to do a better job of this kind of individualizing of the approach.

So let me just mention a few notes from the front lines here for using this metaphor of the genomic revolution, which, you know, is maybe not the greatest metaphor because it sounds all very war-like.

This is certainly not our intention here.  But, in fact, it does seem as if there is a real groundswell of change underway.  And so we may as well call it that.  I guess that was the title I was assigned.  So I'll stick to the word "revolution."

Let me just point out a few notes from the front lines of things that are happening, some of them just in the last year or two, that may not have necessarily reached the ears of the public to the degree that perhaps they should have because I thought this group of distinguished experts might want to consider what some of the consequences would be.

Some people somehow assumed that DNA sequencing was all over once we had gotten that reference sequence of a human genome and some other model organisms, like mouse and rat and dog and horse and so on, that that would sort of be it.

And that is obviously turning out to be a very narrow view.  In fact, DNA sequencing is emerging as a tool for human medicine in ways that I think most people had not quite been prepared for.  And the idea of being able to sequence any of our genomes for an affordable cost, maybe $1,000, could well be no more than 5 or 6 or 7 years away given the way in which the technology for doing this is moving forward so quickly.

Two instruments have found their way into our large-scale sequencing centers in the last year and in many other places as well.  These machines are beginning to be tried out.  And at a meeting in Marco Island, which is one of the gatherings every year where technology gets trotted out and discussed, the production efforts being built upon these instruments were truly impressive.  And this is just two weeks ago.

So one of these is a machine called 454, which is a rather odd name, I suppose, for a company and an odd name for an instrument, but that's what it is.  The idea of both of the two machines I'll tell you about is to get away from the traditional gel-based methods, which is basically the method that Fred Sanger invented back in the late 1970s.

All of the DNA sequencing we have done up until now has largely been to build better and better efficient steps into that gel-based dideoxy method of DNA sequencing.  And it's served us very well.  And we have driven the cost down by orders of magnitude, particularly because of the Genome Project and its need to do so.

But we are now seeing a real quantum change in the approach.  And the two methods that are now finding their way into practice are really built on different principles.  This one is done on beads, where the idea is that you attach DNA to specific beads.  You amplify them on that bead.

And then — and here is the trick — you can then do sequencing at massively parallel platform arrangements so that, instead of in a typical current sequencing machine, where you might run 96 reactions at once, here you can run hundreds of thousands or even millions of reactions at once, simultaneously collecting all the data in a rapid way.

There are still bugs in these systems.  Their quality of the data is not as good as the tried and true method that has been optimized now over more than 20 years, but it's coming along very quickly.  And it's clear this is going to make a major impact on our ability to collect the DNA data.

Perhaps even more of an impact may come from this machine built by a company called Solexa, which again uses massively parallel approaches instead of beads.  Basically the DNA fragments are attached to a surface and then amplified in place using a polymerase chain reaction so that you end up with clusters of molecules that can then be sequenced.  And that's just, in fact, what one can do.

Once those clusters are there, you can actually see them affixed to the solid surface, as in this picture.  And each one of those clusters can then be sequenced using an enzymatic reaction.  These give you short reads but, as you can imagine, just a very high volume of them.

And it looks as if we have already reached the point of having the ability to sequence an individual human genome for something in the neighborhood of $100,000.  And that is two orders of magnitude cheaper than it would have been if we had stayed on our current cost curve.  And the expectation is that that will continue to drop.

Already individual human genomes are beginning to be sequenced by people using these instruments for research purposes.  And you should anticipate that this will increasingly find its way into larger-scale research and ultimately into clinical practice of medicine.

And I think that is where the real growth area for DNA sequencing is: we're already now applying this kind of very high throughput in expensive sequencing to try to find the genes that are still out there unidentified from Mendelian disorders because there are many rare conditions where we have not been able yet to find the responsible gene. And to be able to sequence the entire genome or a segment where you know that the gene must reside, based on linkage evidence, is undoubtedly going to help a lot finishing out this work of identifying those conditions.

We know, for instance, there are probably 50 different types of x-linked mental retardation that have not yet had their genes identified.  And you can imagine a project, in fact, that is now getting underway where you simply get DNA samples from large numbers of affected males and you sequence all of the exons of the genes on the x-chromosome.

And you can do that, actually, for a very affordable cost.  And undoubtedly you will make a lot of discoveries of genes that are involved in brain function and that account for these particular conditions.  And, of course, the hope would be that some of those will lead you in the direction of ideas about treatment.

Common diseases also sequencing can be now applied to candidate genes and as it gets cheaper and cheaper to just sequencing the whole thing.  But perhaps a particularly attractive approach that is being undertaken right now with DNA sequencing is for cancer.  This has been unaffordable until now, but now it is becoming so because this cost curve is coming down so rapidly.

The cancer genome atlas is a joint effort of the Genome Institute and the Cancer Institute currently being conducted as a pilot project on just three cancers, the three that you see here.  And the goal is to sequence something in the neighborhood of two or three thousand genes in four or five hundred individual tumors of each of these types for which DNA is also available from blood so you can compare the DNA sequences and figure out what is a new mutation somatically and what is a hereditary change and not only to do sequencing but also systematically to look across the genome for changes in copy number, where you have delegations or amplifications, as well as to look at gene expression.

And we put all that data set together, together also, I should say, with epigenetic information.  We may finally begin to have a comprehensive view of what goes on in these three tumor types.  If this goes well, we would hope to expand this after the pilot phase to look at many more tumor types and apply the same approach.

Obviously the more rapidly the cost comes down, the more practical that is, especially as NIH is at the present time rather significantly strapped in terms of budgetary restrictions on what we can do.

And TCGA has several components.  I won't dwell upon this.  But basically it is a highly organized effort involving now dozens of laboratories that are working together to try to apply all of these new technologies to the same set of tumors and to put all of the data into a database where any qualified investigator can immediately see the results and there is technology development as well.

And ultimately, as I have been saying, I think the next step beyond these more focused efforts will be complete genome sequencing, initially for research and then as part of mainstream medical care.  And that, of course, is some years away but not as many as one might think.

So what are we to do in the meantime for those common diseases like diabetes and asthma and schizophrenia, autism, hypertension, the common cancers, where we know there are hereditary factors?  Some have been found, but a lot have not.  Do we have to wait until we have the ability to do complete genome sequencing to find those variations or could we speed up the process now by taking some sort of shortcut?  And, of course, I wouldn't answer the question if the answer wasn't yes.

So, in fact, there is a very exciting phase getting underway now about how to find those factors.  If there are relatively common variations, we have the ability now to do that without having to look through the entire genome.

So here is the next sort of note from the front lines — the discovery of genes for common disease is happening, particularly in the last few months.  And I think you're going to see in the next few months a virtually outpouring of discoveries about variations in the genome that are relatively common in the population that play a significant role in risks of common diseases.

The idea is to develop methods that can find the ticking time bombs that are residing there in the DNA.  And we all have them, probably dozens of them.  None of them carry a very high risk of going off.  Perhaps if you have one of these, your risk might go up by 10 or 20 percent over the situation if you did not.  And it's the combination of these particular genetic risk factors plus environmental exposures that generally is responsible for whether disease occurs or not.

And we need to uncover what both the environmental and the genetic risk factors are in order to figure out better ways to intervene.  And, frankly, the genetic factors we're not going to be able to change, but if we understood how they occur, who is at high risk, and could modify the environment or the medical surveillance plan, that would be a good strategy for individualized prevention.

So how are we doing this?  Well, basically what you would like to do would be to sample across the entire genome and find variations, here color-coded to make it somewhat simpler to explain the strategy, like variant b.  So in variant b, our ten affected individuals have a higher proportion of the orange spelling than do the unaffected individuals, who have many fewer orange spelling individuals.

Most of the genome will not give you that kind of answer.  If you're looking at a disease like diabetes, most of the genome you would not expect to see a difference between the cases and the controls.  But somewhere out there if you accept the idea that some of the hereditary risk factors in common disease arise from common variants, you would expect if you had a systematic way of surveying the genome, to be able to find examples like this.

Now, my cartoon makes it look easier than it is because most of the risk factors you find are not going to have this kind of dramatic skewing between cases and controls.  You're looking for a relatively modest difference because these are not going to be deterministic findings.  They are going to be risk factors.

And so, instead of 10 cases and 10 controls, in general, you need 1,000 or more if you're going to have sufficient power to be able to assess the situation.  And that has been the daunting factor.

That means you have to identify a large number of individuals with the disease and carefully characterize their phenotypes.  And you have to have well-matched controls so that you don't end up chasing down false positive results that have nothing to do with the disease.

But the most daunting part of this until recently has been how would you actually survey the whole genome to do this kind of comparison?  And so people have largely been forced to look at so-called candidate genes, which is to say you've got to have a hunch and you've got to go and figure out amongst the 20 or 22 thousand human genes what's on your short list that might be involved in the disease.

And the problem is for most diseases that you would like to apply this to, we're not smart enough to know what the short list ought to look like.  We're still pretty ignorant about the genetic factors in a disease like, say, diabetes.

And so most candidate gene studies have been pretty disappointing.  There have been false positives that didn't hold up.  There have been occasional success stories, but mostly it has been one long story of disappointment and frustration.

And, of course, you would like to get beyond that forced hunch, sixth sense situation, to be able to scan the whole genome systematically.  And that's what we now can do, and that's why there is this real dramatic leap forward occurring in discovering these genetic factors because we finally have the tools to do this.

A big part of that comes out of this project which was an international project called the HapMap Project.  I had the privilege of serving as the project manager of this effort, as I did for the Human Genome Project.

This was focused on that .1 percent of the genome where we differ, not the 99.9 percent where we're all the same.  And we wanted to get a really good catalog of what that variation looks like in different parts of the world.  And we were limited in how many different parts of the world we could look at.

So we basically looked in a population with northern European background.  We looked at the Yorba tribe in Ibadan, Nigeria.  And we looked at in individuals living in China and Japan.  That was our sample.  And it actually turns out that that's a pretty decent representation of the variation that you might have found in other places as well.

And what HapMap did was to basically tell you where the variation in the genome is and how it's organized into neighborhoods because it doesn't turn out that these variations in the genome are all independent of each other.  They're actually traveling in a bit of lock-step, reflecting our relatively recent descent from common ancestors.

And the good news about that is that means that you don't have to sample all of the ten million variants that are common in the genome to find the ones that are associated with disease because they are associated with each other in local neighborhoods.

If you pick wisely, you can pick a much smaller set of these variants.  And they basically stand in as surrogates for all the rest.  And if you're smart about how you do the picking and that's what HapMap allows you to do, instead of having to look at ten million things, you may only have to look at a couple of hundred thousand.

That has now become possible with the kind of chips that are being used to do DNA genotyping.  One can look at more than 1,000 cases and more than 1,000 controls with hundreds of thousands of these genetic variants and afford all of that for something in the neighborhood of one to two million dollars, which is a lot less than the cost of having collected all those cases and control in the first place.

And so there is a huge amount of interest and activity going on right now to apply those ideas to common disease.  And the success stories are starting to build up.

The first success story and one of the most dramatic one is this common cause of blindness in the elderly, macular degeneration.  Most people were not that convinced that this was going to be a disease where genetic factors were all that important.  It doesn't come on until your 70s, 80s, or 90s.  There was some familial clustering, but it wasn't all that impressive.

Applying the HapMap, Josephine Hoh and colleagues were able to identify a variant in Complement Factor H, which plays a very strong role in disease risk.  And not too long after that, another major risk locus was identified, HTRA1.  And just these two variations seem to account for about 50 percent of the risk of this disease.  Another significant environmental risk factor is smoking.

So we have gone from knowing very little about the cause of this illness to having a much more refined picture and one that actually suggests the possibility of interventions because both of these genes seem to be involved in the inflammatory pathway.  And it may very well be that prevention could then be mounted by using an approach that depends on anti-inflammatory agents, including ones that are already FDA-approved.  So this is a pretty exciting story to go from essential ignorance about a disease to a really nice view of what is going on in the space of just a few months.

Other early results from HapMap from about a year ago, the identification of a gene, TCF7L2, in Type II diabetes, now confirmed by almost a dozen groups as a major player in risk of this particular common adult-onset disease that causes so much mortality and morbidity; a variant associated with prostate cancer, which, interestingly, this variant is more frequent in African populations and might account for some of the health disparity in prostate cancer between different groups, although not all; and in Science a few months ago, an association identifying IL23R as a gene important in Crohn's disease.  And this one also suggests immediately a therapeutic strategy that I'm sure is being pursued by more than one group at the present time.

And just three days ago — this is from online Nature, not yet out in print, another study of Type II diabetes, which both confirmed that TCF7L2 that we already knew about but also identified three other genes, including this guy, SLC30A8, which is a fascinating discovery and one which my own lab has already independently found, which is a zinc transporter that is only expressed in the islet cells of the pancreas.  Insulin in the pancreatic islet cells is complexed with zinc.

And you can imagine how this kind of transporter, if it's not quite doing its job, might have an effect on the ability of your pancreas to make insulin when it's supposed to, also immediately suggests a dietary form of therapy, which might be a very nice nutrigenomics kind of outcome for this sort of discovery.

So these things are happening.  And I tell you in the next two to three months I know of at least three or four other major papers that are coming out on other conditions, such as schizophrenia, that are going to be quite revealing in terms of finally nailing down some of the well-validated genetic factors that play out in diseases that have been so puzzling.

To try to stimulate this to go even faster, NIH has been engaged in an unprecedented public-private partnership, again trying to speed up this process at a time where NIH funds are somewhat constrained.

We built this public-private partnership using the foundation for NIH as the moderator of the whole thing and obtaining funds from the private sector to support what is basically a completely open access competition for individual investigators who already have collected samples from cases and controls of particular diseases to get access to this kind of high throughput genotyping.

The requirement is that all the data has to be put in a database, where any qualified investigator can see it.  And that has been agreed to.  And we had more than 30 applications for this particular competition.  And when the dust all settled, these are the six studies that are now underway and already well into the pipeline.

Interestingly, four of those six are mental illness conditions.  And we didn't necessarily expect that was going to be the case.  It just reflects the fact that the National Institute of Mental Health was thinking about this five or six years ago and really preparing for the moment where this would become possible.  And so they had already collected very well-characterized cases and controls, dealt with the consent issues in an effective way so that these samples could, in fact, be put into this pipeline and have the data from the genotyping accessible to lots of investigators around the world.  And that's a serious issue for some of these studies where the consents are, in fact, going to be a problem.

So that's one big initiative that's underway.  And then on top of that, just starting in the last couple of months and strongly supported by Secretary Leavitt, we have an NIH genes and environment initiative, which really is now trying to pull together advances in technology to detect environmental exposure in a more rigorous way than we have been able to do in the past.  And by that I also include diet and physical activity.

There's a lot of opportunity here, particularly using cell phone technology, to do a better job of actually recording what people are exposed to, than the questionnaire-based methods, which have been the mainstay in the past.  And this will also support additional genetic analysis.

And this is all being done in a coordinated, integrated way that David Schwartz, who directs NIEHS, and myself are co-leading.  And this will involve the expenditure of $40 million a year for each of the next 4 years to try to speed up this process of identifying factors and coming up with strategies for intervention.

So I think it's fair to say we're going to see these major common variants for common diseases coming out — they already are — in the next two or three years, that we will go from a pretty significant level of ignorance to a much better informed situation about what factors are involved in disease pathogenesis.

And each one of those discoveries provides both an opportunity to suggest who is at risk by using this diagnostically, but in the longer term, I think what we're most excited about is that this points you towards a pathway that could be used to target a therapeutic strategy.

We desperately need new ideas about therapeutics that go beyond our current hunches about what pathways are involved.  And I think in the longest term, this whole business of discovering, using the tools of genomics, hereditary factors, and disease is going to have its biggest impact by revolution in therapeutics.

But we all know how long that takes because that means you have to come up with a hypothesis.  You have to then go down the pathway of identifying a small molecule, of finding out whether it works in cell culture, then in animal models.  Is it toxic?  Does it have appropriate distribution, metabolism, and excretion patterns?  And ultimately can you run a clinical trial, show benefit, and get it approved?  So one should not expect the therapeutic impact of these discoveries that are happening right now in less than a decade or even more.

So let's move, then, to those clinical applications.  Again, as I'm saying, I think the full consequences are going to take a while.  But some of these things are coming along more quickly.

And they will all come sort of helter-skelter.  It won't be every disease is moving through this pathway of gene discoveries and therapeutics at the same pace.  A lot of that will depend upon how much energy is being put into it, how many resources, and also serendipity.

You know, you run into a gene, like that zinc transporter that immediately suggests a simple treatment.  Well, you're much further along than if you encounter a gene than nobody has ever studied before and you have no clue what it does.  First you have to figure out a lot of basic science about its function before you can begin to apply your ideas about treatment.

So let's consider, then, this flow of information that's coming out of this revolution in genomics.  And I think what's basically happening now in a big way is the diseases with a genetic component, which is virtually all diseases if you look hard enough, are going to have their genetic defects identified because of these increasingly available and efficient tools.

Then the clinical implications will kick in.  And probably first of all diagnostics because it is, at least in concept, rather simple once you have identified a variant that you are quite confident increases the risk of disease, then you could begin to offer that prospectively to people who might want to know.

And there are all kinds of interesting issues about why would people want to know.  And a lot of that relates to whether there is, in fact, a preventive strategy associated with knowing you are at high risk.

After all, we have known for quite a long time, about 15 years now, that a major risk factor for Alzheimer's disease is having the e4 allele at the APOE locus.  And, yet, that test is not very much used and is not all that interesting to most people because we have nothing to offer the people who are found to carry that allele other than to say, "Well, you might want to plan your future a little more carefully."  And most people don't find that a sufficient reason to take the chance of getting information that could actually be quite distressing.

So I think perhaps what has been underestimated is just how important and how complicated it is to go from having a hypothetical diagnostic to being able to demonstrate that you do have an intervention that only sort of makes good common sense, but it actually works.  You want to have that data, I think, before you begin to advocate for a change in the practice of medicine.

In some instances, we are there.  Perhaps the best example in my mind is in cancer because familial cancer syndromes have as their mainstay early detection if you want to try to improve the outcome.

And certainly colon cancer is a wonderful example of this, families such as this.  And this is a real family from Baltimore.  Not too many years ago, if this person marked with the arrow here had come in asking what her risk is and whether there is something she should be doing about that risk, she would have gotten sort of a general recommendation that yes, you're probably at higher risk because of your family history.  But there's not a whole lot more we can say than that.

That's all changed now with the discovery that families like this are extremely likely to have mutations in one of a couple of DNA mismatch repair genes, which can be tested for in this family.  That was, in fact, done and showed that the affected individuals had a mutation in the MLH1 gene, which is a mismatch repair gene.  And then that made it possible for the unaffected individuals who are at 50 percent risk.

And there are in this picture five of them to find out whether they also carry that mutation.  And in this instance, that was done.  And it turns out two of them do have that additional risk and are at about a 60 percent likelihood of developing colon cancer during their lifetime.

Now, here is a case where we have very good data to say that if you're in that situation, you should begin colonoscopy probably at age 30 or 35 and you should do that every year, not every 5 years.  And with that kind of rigorously adhered to program, the likelihood of finding polyps while they're still small and easily removed before they become invasive cancers is extremely good.

And so here is a case where this kind of family history taking, which was an important start point — you notice here family history is not going to go out of date just because we have fancy ways of assessing DNA sequence.  Family history taking followed by a thoughtful analysis of the potential here of a specific mutation being tested, followed by careful recommendations and implementation of those in medical surveillance is life saving and actually saves money as well.

Now, not all diagnostics are going to be on the germ line DNA.  And, in fact, I think the area of greatest growth in diagnostics right now is actually in cancer and relates to somatic changes and particularly the effort to try to predict from what you find in a tumor, what's the likely future course for that individual.

And perhaps the most widely applied example of this is the oncotype DX approach, which genomic health has marketed for prediction about whether a particular breast cancer is likely to recur or whether it is very unlikely to and, therefore, whether adjuvant chemotherapy might be passed over.

And this is data from a paper now almost three years old showing that, in fact, the score, which is basically built upon an assessment of gene expression for a rather short list of about 17 genes, correlates reasonably well in a prospective way with the likelihood of distant metastases ten years later.

So this test, in fact, has been adopted by many oncologists in the practice of taking care of women with breast cancer, although the FDA is still questioning whether, in fact, this test is appropriate for them to intervene and try to regulate.

And, interestingly, FDA did approve a competing test just a couple of weeks ago based on a similar strategy but which requires frozen tissue, cannot be done on a paraffin-imbedded formalin-fixed section, which is where most breast tumors currently end up.

So it would be, in fact, unfortunate from the perspective of applicability of this test if only the frozen sections were useable.  Many women would not be able to have the chance to learn about this.  So that's an interesting debate that is going on in terms of how much FDA regulation and this kind of test is appropriate.

So that is the diagnostic part.  A big area of current interest, which has grown rather quickly and which is I think coming closer every day to becoming standard of care but not quite there except in a few uncommon circumstances, is to apply this information about genetic variation to predict drug response.

And already now for almost two years on the market has been a chip that will determine an individual's variations in the p450 genes, which are responsible for metabolizing a very large number of drugs.  And certainly knowing your genotype for p450 does have predictive value in terms of how you might respond to antidepressants and anticonvulsants and anticoagulants and a number of other drugs.

Interestingly, while this was marketed two years ago, the uptake has been very limited.  And I think that's because most physicians are really not quite sure what to do with the information that comes out of this.  There are really no good guidelines about how to take that information about somebody who is a poor metabolizer or an ultra rapid metabolizer and change the dose that you're prescribing of a particular drug.  And I think that is an area where we need more data and certainly more education.

A drug where this particular approach together with other kinds of pharmacogenomic analyses might very well find its way into practice relatively soon is the drug warfarin or Coumadin, which is one of the most commonly prescribed drugs and the one which is used for anyone who has had atrial fibrillation with a risk of a stroke or a deep vein thrombosis.  So this is an anticoagulant.

Many individuals, including my mother, are on this drug.  It is a very difficult drug to manage.  The window between an adequate dose and an overdose is extremely limited.  And many people do, in fact, suffer consequences, particularly at the initiation of therapy, because the drug dose turns out to be too high or too low.

And it's very clear in retrospective studies that you can make a pretty good prediction about what the maintenance and loading dose ought to be.  If you know the individual's genotype at both p450 and at a gene called VKORC1, something like 55 percent of individual variation in the dose requirement comes from those 2 sets of genes.  And we can now do that prospectively.

The question is, do we have enough data to begin to make that recommendation?  Should the FDA now put this on the label, saying, "If you're going to prescribe Coumadin, then you should determine the genotype first?"

Well, there is an issue here.  How do you get the data quickly enough?  You've got somebody with a deep vein thrombosis.  You need to start them on the drug.  If it's going to take two weeks to get the genetic tests back, it's pretty irrelevant.

So we need to figure out how, in fact, logistically to actually implement this test.  And most of us think if this is going to be the first one that we really try to move into practice and it will be a real sea change in the mindset of many practitioners, you want to be darn sure that this is actually going to help.

And so NHLBI and ourselves are organizing a prospective trial to compare the outcome using the genetic test versus using a dosing algorithm that incorporates everything else that we know is a variable, like age and gender, which are also important variables for predicting the dose.  And we'll have to see.

And if, in fact, the genetic test adds only minimally to that and adds additional cost and logistical challenges, then it won't be appropriate to do this.  If, on the other hand, you can show that including the genetic information reduces the risk of somebody getting way out of range or having a bleeding complication, well, then I think that will be a compelling case.

Maybe the first drug, though, that is ready for pharmacogenomics is not Coumadin, but it's a drug probably many of you haven't heard about, Abacavir, which is a very effective drug against HIV/AIDS.  And, yet, about seven or eight percent of people who are given this drug get a rather severe hypersensitivity reaction, which can require hospitalization and can even be life-threatening.

We now know a lot about what that is about.  And this particular hypersensitivity reaction comes about because of a variation in a gene in the HLA complex.  If you happen to be one who carries the HLAB5701 allele, your likelihood of getting a hypersensitivity reaction to Abacavir is pretty high.  And if you don't have that, it's essentially zero.

So this is a very attractive opportunity to try this out.  And the Australians have done this, not just retrospectively but prospectively, and were able to show that when they introduced this prospective genetic screen, that they dropped the incidence of Abacavir hypersensitivity almost to zero.  And this looks like, then, a good poster child for being able to implement a pharmacogenomic approach to a drug.

And we're talking intensely right now with FDA and NIAID about whether there is any need for additional data here or whether this is one where we already have enough information to take that step.  And there are labs that will do this test and will do it in the space of 24 to 48 hours.  And the need to prescribe this drug immediately, as opposed to waiting a couple of days, is not as compelling.

Ultimately, though, where you really want to get to in our diagram — and I mentioned this earlier — is the therapeutics.  And that is going to be the longest lead time.  And it's, therefore, going to be frustrating to us.

We have to, I think, take some comfort, however, in the fact that the strategy is a very appealing one.  And in a few instances, it is beginning to play out quite nicely.  And I can't come to this group and talk about therapeutic advances based on the genome without referring to the most compelling example, which comes out of the work of Janet Rowley but, of course, based upon work that was done a long time ago.

So one of the reasons this is now such a success story is that the basic work to get us here was started a long time ago.  It wasn't a direct consequence of the Genome Project.

And this is, of course, Gleevec, a drug which turns out to block the active site of the kinase that is generated by this Philadelphia chromosome that Janet described, which is a fusion of chromosomes 9 and 22, seen in most patients with chronic myeloid leukemia.  That particular kinase transforms well-behaved white cells into leukemic cells.  And Gleevec, the drug that Brian Drucker and Novartis collaborated on, ends up blocking that active site.

Just recently, a couple of months ago, a five-year follow-up of this drug published in the New England Journal showing that, in fact, the long-term success here is really quite startlingly wonderful that there are individuals who become resistant to the drug.  But, as you can see from this curve, all told here, the number of deaths related to CML after initiating therapy is actually quite small in a five-year period.

So we would love to see that replicated over and over again.  I think that is the dream that we all have as we begin to discover these genetic factors involved in lots and lots of diseases, but recognize the pipeline is a long one to get to this kind of outcome.

I can't help but point out a couple of other examples where smart researchers aided by some serendipity may be able to short-circuit some of those very long steps.

This is the work of Hal Dietz at Hopkins working on Marfan syndrome.  Hal was involved 15 years ago in the discovery of the gene, which is a gene called fibrillin.  And everybody assumed this is a structural protein and that's why they have the heart disease and there's probably not much you can do about it because it's a structural protein.

Well, it turns out fibrillin has another function.  It's also an inhibitor of TGF-beta.  And it turns out if you have a mutation in fibrillin, you don't inhibit TGF-beta well enough and so you have over-activity of that particular factor.  And that contributes substantially to the phenotype.

And in this paper, Hal was able to show that this drug, losartan, which is already FDA-approved for the treatment of hypertension, is also a TGF-beta antagonist and in this mouse model of Marfan syndrome essentially prevented the aortic dilatation, which is the cause of death in many patients with this disease.

And I've seen recent data that Hal has presented at meetings, showing that this drug, now given to children who have particularly a rapidly advancing course of Marfan syndrome, seems to stop the dilatation of their aorta as soon as the drug is started.

And there is now a big trial underway to apply this in a much larger group and compare it to the standard beta blocker approach.  It looks very promising that, in fact, by banging away at trying to understand the biology of this disease based upon the gene discovery but not assuming that it is just as obvious as what it initially seemed, that Hal has happened into something here that might be an incredibly valuable discovery for the treatment of this disease.

My own lab works on this disease, progeria, the most rapidly progressive form of premature aging shown here in three pictures over the course of about 12 years of a young man with this disease, whose hands you see holding the photo.

We discovered the gene four years ago, found that it's a sporadic mutation in a single base pair of the Lamin-A gene that creates a dominant negative protein that results in the premature aging phenomenon.

But, again, because we fell into a pathway that other people had worked on for 20 or 30 years, Lamin-A has been the subject of much interest among cell biologists for a long time.  And we know a lot about that protein.  We are now about to initiate a clinical trial using farnesyl transferase inhibitors, which if I had more time I would explain the logic for.

Here is a girl with progeria on the treadmill getting ready for her to establish a baseline so that when the treatment is started, one can see whether it's working.

And, again, this is putting together a number of previous observations with gene discovery, with the fact that there are a lot of drugs out there that have been developed for other purposes that may potentially turn out to have unexpected uses.

And even cystic fibrosis, a disease that had certainly hoped to see advances now for quite a number of years given that the gene was published, the discovery was published, in 1989, we are now seeing because of this new approach of applying gene discovery to drug development drugs coming into clinical trials.  And so the first gene-based drug has now entered clinical trials.

Just out of curiosity, this is the same person.  This is Danny, who is on the cover of Science.  This is a couple of months ago, when I ran into Danny at a cystic fibrosis event.  And he's obviously doing very well.  But we need to come up with a better strategy as soon as possible for this disease.

This is an interesting cover.  This is Drug Discovery World, one of these magazines that I don't generally read, but somebody gave it to me.  And what are they saying the future of drug discovery is?  GWAS.  Well, what is that?  Genome-Wide Association Studies.

This is HapMap-based approaches to understanding genetics of common disease, mapping the future of genetics.  They're concluding it is going to be the best engine for therapeutic discovery that we have had in a long time.  And I certainly agree.  And it's kind of a cool cover, too.

So, finally, to bring this back more perhaps to this Bioethics Council in terms of the issues to worry about and to think about, certainly the attention to ethical, legal, and social issues, which have been part and parcel of the Human Genome Project from the very beginning, are more important than ever as we see this accelerating pace.  And I'll just mention a few.  And I'm sure others may come up in the discussion.

The number one issue for us at the Genome Institute from the very beginning from the 14 years I've been there, if you want to say what is the most important policy issue that we need to attend to, it's to prevent an outcome where people who find out information about their DNA have that used against them and particularly in health insurance and in the workplace.

Here we are, 2007.  We still do not have effective federal legislation, but we are a lot closer than we have ever been.  Going back a couple of Congresses, in the 108th Congress, a good bill that covered both health insurance and employment passed the Senate unanimously.

The House failed to act.  So we had to go on to the 109th.  In that case, the House did actually have a bill introduced, but, again, no action was taken upon it.  And, as you know, the 109th Congress came to an end about two months ago.  So now we're at the 110th.

And I'm happy to say that the momentum is certainly better than it's ever been.  Both the Senate and the House have bills that were introduced early in the session.  Both have now been marked up, the Senate bill a couple of weeks ago, the House bill just this week.  And there is a growing enthusiasm for the idea that this might actually come to a vote on the floor of the Senate and the floor of the House.

I was very gratified when President Bush visited NIH a month ago and I had a chance to spend 45 minutes with him, he used a good chunk of his comments at the beginning of this session to talk about this issue of genetic discrimination and again to underline his perspective that this is a issue that needs a legislative solution and that he would hope to see this particular legislation passed.

And he would be prepared to sign it right away.  So maybe we might actually get there this year.  After many, many years of false starts and disappointments, I'm trying not to get excited because it is such a letdown when something falls through, but considering all of the tea leaves at the moment, this looks more promising than it has ever been.

The opposition here is largely coming from the Chamber of Commerce, who feel that, in fact, employers may be subjected to frivolous lawsuits if people who have been fired for good cause then come back and say, "Well, they did it because of my DNA."

And the bill is actually written rather carefully to try to discourage those kinds of frivolous lawsuits.  And it's also I think important to note that more than 40 states now have legislation of this sort and there has not been a single such lawsuit in more than ten years.

Oversight.  I know when Kathy Hudson came and talked to you in November, she spent a lot of her time focused on this and appropriately so.  We want to find the right balance, but   squashing an area of genetic testing between not squashing an area of genetic testing that is potentially growing rapidly and has a lot of promise but also doing something to put the brakes on tests that are actually not well-validated.  And then it might actually do harm.

The FDA up until now has largely taken a hands-off approach.  I mentioned that they have now started to look at these multiplex tests for breast cancer.  Whether they also begin to look at other tests is an open question.  And what other kind of oversight ought to be there, especially when you look at the wide, wide work of direct-to-consumer testing that is going on out there, some of which is really pretty bizarre.

U.S. News and World Report just a couple of months ago had a whole discussion about this, pointing out in a pretty good story that these tests can show you risk, but how good are they?  And many of them are not yet I think at the point where you would want to fully trust the results because most of the discoveries upon which such tests would be based are just now happening when you're talking about common disease.

I am actually concerned that this topic is still not getting as much attention as it should.  It is easy to say, "Well, the 99.9 percent identity of human DNA ought to be an argument for saying how much alike we are and ought to be an argument for reducing bias and prejudice."  And I think that has occasionally been an argument that has gotten attention.

But also as we look more carefully at DNA, it's clear that if you give me a sample of DNA and ask me, "Did this person's ancestors come from Japan, West Africa, or northern Europe?" I could answer your question given a few days in the laboratory because there are variations that are differentially distributed around the world, reflecting all kinds of migration patterns over the last 100,000 years.

It is even clear that there are regions of the genome that have been under recent selection.  And you can see the signature of that.  And for a very small number of those, we understand the reasons why.

So if you look at the beta globin gene, for instance, in the malaria belt, you will see evidence for selection.  Well, okay.  That's because that has been protective.

If you look at the lactase gene in people from northern Europe, you'll see evidence of recent selection that allows them to digest milk as adults; whereas, others in other parts of the world may not be able to do this.

But there are hundreds of these segments of the gene for which we do not know what the function is but clearly show differential selection depending on which part of the world you're looking at.  And it would be hard to imagine that some of these aren't going to turn out to be controversial in certain ways.  And it's not clear to me that we're fully ready to face up to that.

Again, I think the health disparity question is one of our most important issues as researchers, but it's a very complicated one.  What really is the relationship between self-identified race and ethnicity?  There are a lot of connections here.  And people tend I think to blur through them rather quickly and maybe oversimplify.

Maybe in my limited view here, it's good to keep in mind that an awful lot of health disparities have nothing to do with genetics or heredity, but it related to differential environmental exposures, including education, access to health care, culture, socioeconomic status, and even things like stress.  And those play a major role in health and disease.

But it's clear — and I mentioned the example of prostate cancer — that at least in some instances, there are genetic connections as well that reflect ancestral geographic origins.  And those, in turn, are reflected in genome variation.  And that, in turn, may then result in variance in specific genes that play a role in disease risk.

So self-identified race or ethnicity is a proxy for a lot of other information.  The sooner we can get closer to the action and really identify what the proximate factors are that are involved in disease risk and stop using race or ethnicity as a surrogate, the better off we'll be, both in terms of the precision of our information and the lower likelihood that this is seen as a means of reidentifying the concept of race in a way that it doesn't deserve.  We have got a lot of hard work to do, I think, to get through to that outcome.

Access is a big issue.  As we contemplate the $1,000 genome, for instance, okay.  Who is going to have access to that if it turns out to be valuable?  It probably won't be everybody at once, at least not in this country.

I would refer you to what I think is a very thoughtful piece coming from the Secretary's Advisory Committee on Genetics Health in Society, a group which I'm sure this Council has been watching closely because they have struggled with many of the same issues that you all are dealing with in your discussions, particularly the ones about genetics.

And they, I think, pointed out all kinds of things that potentially could be done but require a lot of national will and political will to achieve because at the present time we're clearly not in a circumstance where coverage and reimbursement of tests and services based on genetics are on a trajectory to make access even and relatively straightforward.

When I spoke to this Council five years ago — and I looked back, it was five years ago — the topic was enhancement.  I think at that point the conclusion, and justifiably so, was that a lot of the scenarios that were being put forward about enhancement, like the Gattaca movie, which we watched a clip of that day, — right, Leon? — were rather fanciful.  And we should spend our time worrying more about things that are realistic in the sort of ten-year time frame than things that won't happen during that interval, if at all.  And I think probably that was a wise recommendation.  And I wouldn't really stray away from it.

I think in terms of enhancement, the practical applications at the present time really fall much more in the direction of PGD, where you have an increasing ability to apply more and more kinds of genetic tests to pre-implantation genetic diagnosis.  And the question is, where should those limits be set?

It's one thing to tests for Tay-sachs disease.  It's another to test for gene variant for, say, obesity.  And I know that there is a gene variant for obesity that will be published soon that is clearly highly validated.  So is there going to be an application there that enterprising marketers to couples who determine to optimize everything will see as something they want to begin to offer?

And then there is this more philosophical question and one that I am very interested in and have been from the beginning.  Are we running the risk with all of this excitement about genetics — and, believe me, I'm excited — that we overemphasize the role that DNA plays in humanity and undervalue other things, such as the environment, free will, and the human spirit?

And for me as somebody who very much enjoys the opportunity to seek the truth, not only scientifically but spiritually, I do worry that one of the contributors, one of many, to the increasing view of humans as more machine-like than spirit-like, may be our own field of genomics.  And we really should work very hard to explain what this can tell us and what it cannot.

And in that regard, I seem to have had the opportunity recently to get engaged in conversations like this one in Time magazine, a debate with Richard Dawkins, who obviously is putting forward the view that what we're learning about genes and evolution means that we should sort of get over the idea that there is any need to consider things beyond that.  My view is rather different.  And this was an interesting experience, I can assure you.

So let me not go on any longer because I went on a little longer than I intended to.  Basically I do think it is an appropriate and ethical stance to try to apply the tools of genetics and genomics to alleviate human suffering.

I think one of the least ethical things we could do would be to say we should slow this down.  I would have a very hard time explaining that to the parents of a child who has an illness that desperately needs some new intervention.

At the same time, I think we would have to have our eyes wide open to the ways in which these kinds of advances may lead to misuses that society will be comfortable with.  And that's why I'm glad all of you are spending as much time as you are in rooms like this thinking about those issues and trying to sound some sort of warning signals when there are things that need more attention than they are currently getting.  And I am glad to be part of your process.

Thank you very much.

(Applause.)

DR. COLLINS:  Shall I come and sit down over there?  Would that be —

DR. PELLEGRINO:  Yes, please.  Thank you very much, Dr. Collins, for an absolutely superb, breathtaking, exciting review of the field of genetic medicine.  You have really taken us from the molecule to the ethical and the social and even the spiritual.  And it's been a great privilege to hear your presentation.

What we plan is to have two members of our Council open the discussion of your paper.  And I will ask them to make their comments and then open the discussion to the members of the Council generally.

Our first discussant will be Dr. Janet Rowley.

DR. ROWLEY:  Well, firstly, Francis, I want to thank you very much for your comments related to my own research in Gleevec and point out to members of the Council that Dr. Collins has been far too modest himself to point out that he has contributed to many of the things he described, including helping to clone the gene for cystic fibrosis.  So that he certainly merits the thanks of all of us and also I think the acclaim that he has received.

I can only echo Dr. Pellegrino in saying that, as always, Francis has presented a very dramatic and exciting talk.  And it has been wide-ranging.  So that when you think of a President's Council on Bioethics and think of the ethical issues that may be raised by the discoveries of the Genome Project, that, in fact, you covered pretty much what the ethical issues are and some of the status of trying to deal with those.

And I think that the question of trying to make sure that genetic information is not used against an individual or a family is certainly one of the major challenges that we face.

And I can remember when Clinton was president and I was president of the American Society for Human Genetics writing a letter on behalf of the society urging that individuals not be discriminated against.

And I think, unfortunately, many members, who are sort of similar to the pedigree that Francis illustrated of a family with colon cancer, in this case with the gene MLH1, but women will not get tested for VRCA1 because they don't want that information available to insurance companies or to employers, but I think in one sense women or families are more concerned about insurance and that this country has not protected those individuals yet, as I think a disgrace and an ethical, moral disgrace.

So that is an area I would guess with bills in Congress, that there isn't a whole lot that the Council can do except possibly to encourage the consideration and passage of that bill.

I think another issue that you raised, Francis, is that as a society and the physicians in that society are not really prepared to deal with considering diseases on this kind of genetic basis or probably more appropriate is how to evaluate information that is coming forward.

And, again, this is not an ethical issue.  This is more one of education in saying that medical schools have got to do a better job of teaching medical students about the medicine of the future and about societies and other organizations to which physicians and nurses and other health professionals, the societies to which they belong have got to be important.

The question that I had thought of before we started was for you was going to be what the ethical issues were for the President's Council.  And, yet, you have gone through most of them.  I guess I would still come back to that question.

Are there issues that you didn't in your presentation have an opportunity to discuss as fully as you might have liked that the Council could deal with or are there additional issues that for constraints of time you didn't have an opportunity to speak about?

DR. PELLEGRINO:  Thank you, Janet.

Dr. Hurlbut, you are the next commentator.

DR. HURLBUT:  Actually, I would like to give as much time to Francis to answer the last question, but just let me make a couple of comments to frame this.  It's obvious that we're talking here about a subject, right or wrong, that touches on the very meaning of origins and ultimate ends when humans think about their genetics.

And that clearly is at the foundation of individual and social identity and the grounding of morals itself.  We tend to think of this in a very special way with regard to the causal circle of being, right or wrongly.

But it does strike me that there are some interesting things we ought to put into the mix of our discussion.  You mentioned the issue of race and the parent exaggeration of determinism and so forth.

You [mentioned the 99.9% similarity between individuals].   I just want to point out that percentage has relatively little to do with biology in the sense that we have dynamic systems and we are percentage-wise very similar to chimpanzees, but there is quite a dramatic leap between us, at least in some characteristics. [So the basis of our moral equality must be something other than the percentage of our shared genetic sequence.]

Also, given the nature of gene-gene interactions and environmental interactions and so forth, most of our data on cause is going to be placed in terms of statistical probabilities.  And there will even be stochastic events that trigger certain dramatic changes in the outcomes of events, even if most things seem very similarly aligned.

So that strikes me as important.  And, likewise, because of these confusing questions of determinism and the reductionistic model that genetics seems to promote in the common mind anyway, there is a special fear of genetics, both excitement and fear essentially, promise and peril, as it has been said 1,000 times.

And here is an issue I really appreciate some comment on, Francis.  Apparently people are quite reluctant to go into studies that involve the use of their genetics.  They're more willing to do studies that involve their RNA than their DNA.

And obviously if we're going to exploit the full possibilities of this field, we're going to use huge databases with demographic and epidemiological data.  And you might comment on that a little bit.

I think, reflecting on what our Council might do, I think I would like some guidance because this is a subject that has been thought about a great deal, funded a great deal.  And if you look at what has happened over the last 15 years of discussion, it actually hasn't moved that far in terms of conclusions.

So I would like your thoughts about what issues remain to be talked about deeply and what some avenues in to those might be.  And, just to mention a couple that strike me, the issue of discrimination that you mentioned in the bill, some people have raised a question about whether that bill and our general social attitude sufficiently address the question of pre-implantation genetic diagnosed and whether families might be vulnerable to not having their children covered by insurance or so forth.  Would you address that question for us, whether that's a subject that's been adequately dealt with?

Also, what I mentioned in the way of DNA records, recently there was an op ed. by Michael Crichton concerning patenting of genes.  Would you mention something on that?

And, finally, two final things.  One, could you say something about the advances in synthetic DNA synthesis and where that might lead us?  And one thing that you only implicitly touched on that I would be interested in hearing about is you brought us back down to the notion of interventions that are preemptive or preventative medical interventions.  If we could only know what was influencing disease development, we could perhaps go down to the bottom and prevent the expression of these diseases.

It does strike me that that presents a rather interesting dilemma for medicine.  We don't want to intervene in a way that is damaging to an individual.  A lot of disease expression is secondary physiological response to unfolding disease conditions.  It might be much more subtle to intervene in these diseases than we think.

Do you get what I'm getting at there?

DR. COLLINS:  No.

DR. HURLBUT:  Not entirely.  Well, quite a lot of disease symptomatology is actually the result of the body's response to a disorder.  So, for example, tuberculosis is an immune reaction largely, the body's compensation mechanisms.  And we could very easily target the wrong response.

Well, just go to the bottom of how we might intervene in genetic disease early in the special challenges in watching how we ramp up to that.

Do you understand what I'm saying?  So those are the main things that strike me.  And what I most of all would like to hear about is what you think we could do that others have not done.

DR. PELLEGRINO:  Dr. Collins?

DR. COLLINS:  Well, a very challenging pair of commentaries.  And I'm not sure that I'm well-positioned to tell this distinguished Council exactly what direction to go in, but I will reflect a little bit on the number of issues that you all have raised.  And thanks to both of you.

Maybe we'll go straight to this genetic discrimination question because both of you brought this up.  Certainly there's no, I think, real disagreement from an ethical and moral perspective on this topic, that this is an issue of justice, that if, in fact, you don't get to pick your DNA — and we don't — that that should not be used to deny you access to health care or to a job that you would otherwise be qualified for.

The recent discussion about the details of the bill that you mentioned, Bill, in terms of whether, in fact, it's written appropriately to cover the circumstance of PGD has, in fact, represented the latest hiccup in trying to actually get this done.

I must say from my perspective, I think the bill is written in a way that covers those circumstances, but it quickly gets you into some pretty deep weeds as far as legal language of particular provisions.

And the fact that it has been raised — and you probably saw a letter from a Catholic bishop specifically pointing out the need to address this.  And there was an amendment proposed in the House markup that would have changed the language in the existing bill, which did not, I think, pass — in fact, I know it didn't pass — and which actually worried a lot of the people in the Senate that the whole process might be falling apart on these grounds.

I would say this is an important issue, but I think those who are proposing changing the language of the bill to try to accommodate it really need to sit down very carefully and make sure that that is necessary because I think the coalition to get this bill passed still remains somewhat fragile.  And there are lots of opposing forces out there that would love to see an issue raised that would slow down or stop the momentum.

This is the latest issue that has been raised.  And I'm not sure that it is one that is understood well enough by those who are suggesting amendments to make those changes.  And it could potentially derail the whole process.  So I think we have to be very careful in making those kinds of proposals.

In terms of a related question, you asked about people's reluctance to participate in research that involves DNA testing, as opposed to RNA.  We have done actually studies on that phenomenon by asking people who are otherwise qualified for NIH research projects and who decide not to participate, "Why was it you decided not to participate?"  And it is this concern about discrimination.

And that's why they're worried about the DNA part.  They get it that RNA expression is probably less likely to get them in trouble because it comes and it goes, but they understand that DNA is with you for life if you're talking about the germline part of it.

And once that information has been determined, you may not be able to pretend it wasn't.  And it could in the current circumstance come back to haunt you, especially if you have to apply for an individual health insurance policy down the road, which is the part that is currently not protected.

Fully one-third of people who were in a situation similar to that family I showed you with colon cancer, who were given the chance to participate in a research study at NIH to go through the process of genetic testing and counseling and particularly to monitor what do they do with that information, do people actually change their health behaviors, do they enroll in colonoscopy programs, do they follow them rigorously, we need to know all the steps here.

A third of the people who were in the same high-risk situation as the woman I showed you in that pedigree ultimately decided not to participate because of this concern about discrimination.

So this is a very real and present issue.  And the legislative solution, it seems to me, would have a huge impact on that.  Whether it would reduce that risk completely to zero or whether people would still have anxieties about what might happen in other areas, like long-term disability or life insurance, which are not currently covered, I don't know.

By the way, those are topics which we have essentially avoided getting into in this country.  And that might be an area that is worth some consideration.  I think people have been anxious not to distract the conversation from the highest priority of discriminatory circumstances, which is health insurance and the workplace, by raising these other issues, but they are going to be there.

In the U.K., there has been a pretty productive discussion about life insurance, for instance, with the conclusion being that also there needs to be some ability for individuals to be able to at least get a basic level of life insurance, which you need in many instances to get a home mortgage, without having questions asked about your DNA.  In this country, we really haven't had that conversation.

Long-term disability will be a very difficult one because I think you can argue there that adverse selection could come into play and that if you know you're at high risk for Alzheimer's disease but the company issuing the disability policy doesn't know that, you're really going to destabilize the economics of long-term disability insurance.  And that will be an interesting and difficult area to get into.  And, as I say, presently we really haven't had much progress in that direction.

So, again, I guess maybe because it is the big elephant in every room that talks about ethical, legal, and social issues coming out of the Genome Project, genetic discrimination comes immediately to my mind in a response to both of your questions but other areas.

You mentioned intellectual property in the sense of asking about Michael Crichton's op ed.  I don't know how many of you read Crichton's book called Next.  Well, I have.

(Laughter.)

DR. COLLINS:  It's rather fanciful.  It includes a cloning experiment gone awry that results in a sort of chimeric chimpanzee human.  It includes a parrot who seems to have remarkable abilities to speak and think.  And they all get mixed together in this sort of "biotechnology gone wild" soup and a great deal of additional sort of political commentary as folded in there by Crichton about how intellectual property is really the evil that has resulted in all of this.  And biotechnology doesn't look good in Michael Crichton's book, as you might imagine.

In the appendix, he sort of goes on a bit of a tear about what he has learned about this and what should be done.  I think there are a number of aspects that he has got right and a number of aspects that I don't think are quite properly presented in terms of the facts of the matter in that op ed.

I am certainly one who has for the last 14 years, since I have been at NIH, tried in every way to discourage the idea of patenting information that in the past would have been considered foundational and pre-competitive and ought to be in the public domain.

You have seen NIH move successively in the direction of not only saying that but enforcing that by issuing guidances about intellectual property claims in genomics and by putting as conditions of grant awards those kinds of statements that "If you're going to do this, your data has to go into the public domain immediately."

I think we have had a pretty positive impact there.  And I think the landscape really has changed.  Crichton's complaints in the op ed. would have resounded more ten years ago than they do now.

But we do have this legacy of a big mess that got created during the 1990s, when there was, frankly, a bit of a gold rush going on by people claiming intellectual property on snippets of DNA sequence whose function was really not known and getting in some instances the Patent Office to issue very broad coverage of those claims in a way that now ties things a bit in knots for people who are trying to do follow-on experiments that might actually help the public.  So I think the big problem we really have to deal with is more what, the mistakes we made in the past than quite so much the present.

We are still pushing that envelope as hard as we can, not just for DNA sequences.  We have a new program at NIH to make it possible for academic investigators to get access to this high throughput screening process to find small organic compounds that have activity in interesting biological essays and that might ultimately lead to the development of a therapeutic.

And we have insisted that all of the participants in that project put all of their screening results up on the database called PubChem immediately without claiming intellectual property.

There was a big squawk about that, particularly from universities, who I think kind of overvalued what it is that you get out of a very early stage screen of this sort and thought that they ought to be able to hang on to this information and patent it and maybe not publish it for a while.

I, frankly, think that is not the way you are going to lead to the kind of public benefit that the program was set up to do.  So we have held the line on that and insisted that if you are part of that program, all the data has got to go into PubChem and no IP can be filed.  I think people are now getting used to it.  Every time you come into a new domain here, there is one of those issues.

With all of these genetic association studies that NIH is funding, there is a discussion going on right now and lots of public input has been sought about whether, in fact, the discoveries that come out of that ought to be placed in the public domain or whether if you have found a variation in a gene like that zinc transporter I told you about that's associated with diabetes, that you ought to be able to patent that as a therapeutic.

Our arguments are that the patenting really ought to apply to further downstream efforts once you actually have shown something that's on the pathway towards public benefit.  And so our insistence is that all that data also goes into a public database immediately.  And, therefore, patents cannot be applied to it.

We are encountering some resistance on that point, but I do think just looking at the landscape over the last 15 years this pendulum has really swung from where it was in a real land grab environment to a much more sanguine realization that if our goal here is to advance science and to benefit the public, you don't want to put a lot of toll booths in the way early in this road towards discovery.  You want to let people travel on it freely and quickly.

You asked about DNA synthesis and the ethical issues there.  And I am glad you brought that up.  I didn't have time to mention it, the abilities to basically make any DNA sequence at will.  While they are not moving forward orders of magnitudes every year, they are certainly improving substantially.

For instance, in a program we had been leading for the last four or five years to try to get all of the human full-length coding regions of the genes into a public database that everybody could have access to and other means of sort of spurring research, we realized about a year ago that, instead of trying to identify those coding sequences from some library, that it's cheaper now to just make them.

And so we are now synthesizing those genes from, you know, organic chemicals at a more affordable cost.  And that tells you we really crossed a line somewhere.  And I don't think it's out of the range of possibilities that many laboratories in five or six years won't bother to store all of these clones that we keep around and fill up our freezers with stuff we can't even quite remember what it was.  You'll just, instead, make it and say, "Oh, well, I need that vector with that sequence with that particular mutation," punch it in the computer.  Tomorrow morning there it will be because the synthesis will be just that good.

But, of course, that raises the specter of how this might play out if falling into the wrong hands and people begin to make pathogens that are really horrible.

And I think the group at NIH that is considering all of the consequences that may come out of advances in biotechnology that could be used for bioterrorist purposes are quite concerned about this issue of DNA synthesis and is there any way to try to keep track of that.

So that, for instance, companies that are providing that service monitor what it is that people are asking them to make.  Instead of just saying, "Oh, it's just all A, C, G, and T.  It doesn't matter," we need to have some kind of system of keeping track of that.

That will work for a while as long as most of the synthesis can only be done in a large facility, but it won't work forever if it becomes increasingly portable.  And that's an area to keep track of.

I need to quickly sort of get to the meat of your question, Janet and Bill, what other issues and where do you think that the focus might be for this Council.

I have to say I come back again to this concern about genetics and the study of genetics and how that impacts our concepts of race and ethnicity and what it means to be human.

All men and women are created equal, endowed by their creator with certain inalienable rights.  How is that affected by our increasing ability to discover that where your particular ancestors came from is going to be reflected by certain things that have been selected for in your genome that might not be there were it not for that history?

Those hundreds of places in the genome that show those fingerprints or those footprints of that kind of selection are ultimately going to bit by bit get figured out.

And some of them will be simply understood and noncontroversial, like your ability to drink milk as an adult, but others of them will be more controversial in terms of your abilities in certain intellectual ways.  It's possible.  It's going to play out there.

Behavior traits.  We don't know what is going on there, what kind of influences down through the last 100,000 years have shaped the genome in different ways depending on where you are.

As we apply the tools of HapMap and the ability to scan across the genome to find subtle variations that play a role in disease risk, people are also applying those to look for things that aren't diseases.

Intelligence is clearly going to in the next couple of years have variations discovered that are associated with how you perform on an IQ test.  That's inevitable.

Similarly, behavioral traits that you measure on a personality test, some of which have already been discovered, although some of them haven't held up very well, we're going to have a big outpouring of that as well.  And if those are differentially distributed across different geographic groups, people are going to draw conclusions about that.

And people will use that for demagogic political purposes in certain ways that we will all find offensive.  But are we prepared to come up with an appropriate counter response or are we kind of headed into a future and a not terribly distant future where the wonderful idea that genetics is going to bring us all together will, instead, be used to drive us apart?  I am deeply concerned about that.

And I don't think at the present time we have a particularly well-coordinated strategy.  It's one of the things when I meet with my staff that works on ethical, legal, and social issues that's always at the top of the agenda.

There are many perspectives about this that are I think not fully cognizant of what is happening in the field of genetics yet because it's come along so quickly.  We need to have a strategy here.  And it needs to be a strategy that reflects our shared principles of equity and justice.  And I think there's a lot of work to be done there.

DR. PELLEGRINO:  Thank you very much, Dr. Collins.

I would like to now throw the discussion open to the Council members.  We are close in time.  So I will take the Chairman's privilege of extending this discussion to 10:15, rather than 10:00 o'clock.  So if you indicate your desire to speak, I will recognize you in some sort of order.  Dr. George and Dr. Meilaender in that order.

PROF. GEORGE:Thank you, Dr. Pellegrino.  And thank you, Dr. Collins, for that wonderful presentation.

I would like to explore with you for a few minutes those philosophical questions that you touched on at the end and then again at the end of your responses to Janet and Bill.

As a way in, I would like to just quote to you from a paper that was included in our materials, though not for this session, for our next session, in which an eminent philosopher, Daniel Dennett, says the following, "Science has vanished the soul as firmly as it has vanished mermaids, unicorns, and perpetual motion machines.  There are no such things.  There is no more scientific justification for believing in an immaterial and immortal soul than there is for believing that each of your kidneys has a tap dancing poltergeist living in it."  Is that so?

(Laughter.)

DR. COLLINS:  That's quite an interesting way to phrase the question.  So I know Dan Dennett.

DR. PELLEGRINO:  Yes or no?

DR. COLLINS:  He and I have occasionally debated this issue about science and faith.  And, of course, his book Breaking the Spell got a lot of attention about a year ago when it came out and expressed many of the same perspectives that are included in that quote, which is really quite a zinger.

I think the problem with what he is saying there is that it basically applies a scientific approach to a nonscientific issue.  And so it's guaranteed to get you nowhere.

If there is such a thing as a soul, science is not going to discover it.  If there is such a thing as God who is outside of nature, science is the wrong way to discover him or her as well.

And so in the same way that I think this argument coming from the sort of extreme atheistic wing of the scientific community doesn't get you very far, it basically takes the tools of science and applies them in a place where they don't work.

So obviously science is the way, the tried and true, the dependable way to find out the truth about nature and how it works.  And one should not settle for anything less than scientific proof if you're trying to ask a how question about how something in nature actually operates.

But if you're trying to ask questions like "Why are we all here or is there a soul or is there a God?  And if so, does God care about me?" it's immediately apparent, I think, as soon as you step back for just a minute from that question to say, "Science is the wrong way to try to approach that answer."

From my perspective — and I wrote much more extensively about this in this book called The Language of God that came out last summer — it is truly unfortunate that we have arrived at this point in our culture, where there seems to be a battle going on between the extreme views, where we have atheists, some of them from my scientific community, arguing that we ought to just get over this idea of faith, that it's a throwback to the past.

On the other hand, we have people of strong fundamental faith who say that science can't be trusted because it contradicts certain views that they hold about the origins of the Earth and human life.

Most of us live in the middle.  Most of us are pretty happy there.  Forty percent of scientists are believers in a personal God.  It really is unfortunate that there is so much noise coming from the extreme perspectives that people are beginning to wonder if that is all there is.

I find that particularly difficult for young people, particularly young scientists, who are being told that you have to make a choice between these world views.  That choice is both unnecessary and unfortunate.

PROF. GEORGE:If 40 percent are believers, that would suggest that 60 percent are either nonbelievers or agnostic on the question.  Would it be fair to infer from that or perhaps just from other information that you have available that a substantial percentage of scientists, perhaps not a majority, agree that science has vanished the soul?

DR. COLLINS:  I think that is a very strong statement.  And I don't think most scientists would probably acknowledge the truth of that statement.  A few would.

I think most of that 60 percent, although the data is not particularly strong here, are not scientists who are in the strict atheist community in the way that Dan Dennett and Richard Dawkins and Sam Harris are.  I think they are mostly people who are agnostic who basically say there is no way to know.  They are not particularly hostile to religion, but they are just not themselves participants in that world view.

Frankly, I think many of them haven't given it much thought.  And I say this because I used to be one of those.

PROF. GEORGE:So you would say that most scientists don't believe science establishes what these philosophers claim science — that these scientists have, in fact, shown?

DR. COLLINS:  I think in the phrasing that Dennett puts in there about science having disproved the soul, most scientists would have trouble agreeing with that.

PROF. GEORGE:My way of framing it would be that science is concerned with understanding material and efficient causes if I can use those old categories.  And then the effort to suppose that since science is concerned with studying efficient and material causes, there can't be any other cause.  In other words, there can't be formal and final causes.

DR. COLLINS:  Sure.  If you imagine a circle that contains all the truth that could possibly exist now or forever and then you ask what is the scientific part of that circle, it's not the whole thing.  And so if an answer to a question you're looking for happens to be outside the scientific circle, that doesn't mean it's an inappropriate question.  It just means you need another way to approach it.

PROF. GEORGE:Now, the question that you raised at the very end of the response to Janet and Bill about how they relate to each other, does that mean that the answer to the important question about how we will, whether we should, sustain the belief in the fundamental equality and dignity of human beings is something, a challenge that has to be met, not with the resources of science itself since efficient and material causes are not going to even in a deep understanding at the moment explain the basis of human equality and dignity, that science itself will have to be subject to judgments brought from beyond science, where we can explore realms of knowledge that would enable us to judge, in fact, that all human beings are created equal, endowed by their creator with unalienable rights.

DR. COLLINS:  I certainly would agree with that.  In my view, morality is not sufficiently explainable by scientific or even evolutionary arguments.  Otherwise you're forced into the conclusion that right and wrong are simply an evolutionary contrivance and artifactual sort of representation of something that has no absolute truth associated with it.

And when it comes to the principles of justice and equity, I think those are principles where science needs to be brought into the conversation so that we know what the facts of the matter are about how humans are similar or how they're different, but I don't think we're going to resolve the issue about equality and justice based upon purely scientific arguments.

DR. PELLEGRINO:  Thank you.  We have four members of the Council who wish to speak.  The time is short.  And I would like to please ask that you be as direct as possible out of deference to your colleagues, who would like to also participate.  Dr. Meilaender?

PROF. MEILAENDER:  This is very brief, actually, but you puzzled me near the end, when you were talking through the various ethical questions that you saw arising.  And you came to enhancement.  And you recalled being here five years ago and you said when you were here five years ago and talked about it, sort of general agreement that certain things which were interesting to think about were not really likely to happen in the near future.

And the way that sentence was moving, it sounded as if what it was about to say was "But it's a little different now.  It looks a little different now than it did five years ago, but it sort of just petered out at that point and didn't conclude anything."

Do you hold the same view?  Do you hold a different view?  What is your view on that question?

DR. COLLINS:  Well, Janet will remember this because we certainly tried to struggle in that discussion about differentiating between the designer baby scenarios that make good Hollywood movies but really don't have much scientific legitimacy to them and to things that might actually happen.

I think some of the things we talked about that might happen in a ten-year horizon, things such as using human artificial chromosomes, for instance, to introduce a new set of genes that weren't already there haven't actually moved very quickly over the course of those five years.

The one thing — and I was at that point aware that I was running out of time.  So I probably did hurdle forward rather quickly at this point in the presentation.

The one thing that we did identify, which I do think has moved forward, will continue to move forward, and is deserving of a lot of attention, is the use of pre-implantation genetic diagnosis, not that you're going to use that as a circumstance to insert a new gene, not that kind of gene intervention, not that kind of enhancement, but that you're going to try to skew the odds basically by having a number of embryos and picking the ones that you're going to re-implant based upon a DNA test that you perform, a biopsy, if you will, on those embryos.

And, again, there is increasing availability of that technology over the last five years.  And there is certainly increasing knowledge about genetic factors in conditions that are increasingly more like traits than there was five years ago.

And I mentioned obesity in my remarks as an example that might be seen as appealing to some enterprising provider of IVF services.  I don't know whether that will happen, but that is the sort of scenario I think to think about, which five years ago we could think about, but now it's getting a little closer to reality.

So I guess if I had to focus on an area of enhancement that deserves another look, that would be it.

DR. PELLEGRINO:  Dr. Kass?

DR. KASS:  Thank you.

I could listen to you all day.  Thank you for this presentation and for your absolutely spectacular leadership, not just in terms of science but in bringing this conversation to the public.  You've done it here.  You've done it everywhere.  And everybody should acknowledge with gratitude your service.

I want to leave these large philosophical questions alone.  I want to come to the practical questions of the translation of this genomic knowledge into clinical medicine.

In the paper that you submitted, there is a nice paragraph which says, "Health professionals will need to become genomically literate.  New curricula of educational models must be developed.  Behavioral science research will need to establish how best to use genomic information to affect health behaviors and outcomes," all that sort of stuff.

There's a lot packed in there.  And there's a certain irony to me in the little slide which was shown twice, "Personalized Medicine:  A Future Dream."  I mean, there are some of us who would say leaving genomics out of it, personal medicine of the sort you and I were trained to practice seems to be a vanishing species.

I guess there are two issues.  One is given the doctor-patient relationship and its dwindling character, given the amount of education that would actually be required to discuss risk management when you're dealing with statistical evidence here, statistical factors at best, doesn't that particular part of this — it's not just a matter of information.  It's a matter of the institutions that could act on that knowledge and do the kind of education with stuff which is really not intuitive for the average patient.  That would be one.

And yesterday we had long discussions, several discussions, about the efficacy of preventive medicine, which involves people's changing behavior for things for which we have 100 percent correlations, not 100 but very, very high correlations, of behavior and misery.  And Dan Foster took the very dour view that we just can't do very much about making people change, even when the evidence is staring them in the face.

So, I mean, the slides are marvelous.  Some of the innovations are marvelous when it comes to these targeted things.  But in terms of risk management, the first side, shouldn't we be much more modest about our ability to use this knowledge efficaciously until we have either a different way of doctors relating to patients and the other matter about actually affecting people's behavior based upon this knowledge, which is, at best, merely statistical and from the patient's point of view like astrology?

DR. COLLINS:  Those are great questions, Leon, as usual.  I think you have cut very much to the heart of the question about the applicability of all of these discoveries, when it comes particularly to the prevention arena.

You could argue that the therapeutic arena is more likely in the long run to have a bigger impact because basically what you're talking about there is a discovery engine for new approaches to developing therapies.  As long as we at least have a medical care system that's capable of diagnosing and treating, having treatments that work better would be a good thing and would probably advance things.

How realistic is this individualized prevention strategy going to be?  There are a lot of roadblocks.  And you have mentioned most of them.  Certainly this circumstance, where at the very time you would hope physicians would have the opportunity to learn about this and to spend the time necessary interacting with the patient to walk through this complicated information is the very time where they're being taken increasingly to task for spending more than four minutes per patient.  It doesn't look like a very promising scenario.

Actually, ten years ago we founded with the AMA and the American Nurses Association an organization called the National Coalition for Health Professional Education in Genetics, NCHPEG, which now has more than 100 member societies.

And it's increasingly clear to me it's not the physicians that are probably going to carry the load of this.  If we do see this kind of genetic information finding its way into a discussion with an individual about their own prevention strategies, it's probably going to be someone like a nurse or a nurse practitioner or a physician assistant who is able in our current cost system to spend a little bit more time.

And we will also have to use heavily teaching aids that are based upon computer learning approaches, as opposed to having everything done in a one-on-one fashion by a busy and expensive health care professional.

Let me also address, though, your question about will people actually use the information or it will just be like "Okay.  That was nice.  Let me go back to doing exactly what I was doing."  We don't have good data on that.

A lot of the intervention about health care behaviors is pretty discouraging when presented with information that ought to be motivating.  People don't always get motivated.

Now, a lot of that comes out of cigarettes.  And we have to acknowledge there is an addiction issue there, which makes that a lot more complicated potentially than some of the things that we will be able to offer up that come out of genetics.  But one shouldn't minimize the fact that we really don't have a lot of data.

We just started a project at NIH where we're going to offer people, hundreds of them, the chance to come and be tested for all of the things that we know currently are validated as predicting future risk of illness.

We'll let them decide whether they want all of the information or just the ones where an intervention is available.  We will then give them the kind of counseling about what kind of change in health care behavior or medical surveillance would be appropriate for their circumstance.  And we'll watch and see what happens.

We need data because right now it's really hard to know what the answer is going to be.  You can look at a little bit of encouragement perhaps from our cholesterol experience.  I mean, the idea that we should all sort of know what our cholesterol is and if it's too high, we should do something, that has found its way into many people's attitude towards their medical management.

And the evidence is pretty good it's actually doing some substantial good in terms of the long run.  Not everybody will agree with that I would judge by your raised eyebrows, but it does seem —

DR. KASS:  Dan thinks it's statins.

DR. COLLINS:  Right.  Well, why do we have so many people taking statin?  Some of them because they have their cholesterol measured and so they figured that they needed to do this.  So if we had a statin equivalent for a lot of the genetic risks that are coming out of current discoveries, you can imagine an improved outcome.

But you are so right to point out all of the uncertainties here.  And I didn't mean by putting this diagram forward to say that we're confident exactly what the consequences will be.

The pharmacogenomics part maybe is going to be a more clear success story, but we don't even know that yet.  The therapeutics I'm sure will be, but it's a long lead time.

The individualized preventive medicine strategy, where we alter our health behaviors based on genetic information, it will be interesting to see what happens.

There will certainly be people who are sort of information seekers who embrace this and run with it, but is that going to be one percent or is that going to be 50 percent of the people who have the chance to get the information?  I don't know.  I just wanted them to have a chance to get the information without being fearful of it.

DR. PELLEGRINO:  Dr. Dresser?

PROF. DRESSER:  Thank you.

I have a general question and a specific question.  The general one is you haven't said anything about proteomics.  And after the Genome Project was done, everyone said that is the next big thing.

So are there any ethical implications, developments we should know about?  And the specific question is — I'll try to get it right — a critique of the anti-discrimination approach to health insurance and genetics, genetic exceptionalism Mark Rothstein has raised.

So if it's unfair to use underwriting, to use a genetic predisposition in insurance understanding, then why is it fair to use an actual genetic disease, like breast cancer related to the bracketing in underwriting as a preexisting condition?

And if that's unfair, why is it unfair to use another kind of breast, somebody who has another kind of breast cancer that we haven't found the gene that it's related to, to exclude them in underwriting?  And if that's unfair, why is it unfair to exclude someone who was injured in a car accident?

That is, if you consider the unfairness in this narrow genetic predisposition area, that's a drop in the bucket and doesn't address the larger unfairnesses in individual underwriting.

DR. COLLINS:  Another set of good questions.  Proteomics quickly I think is moving forward.  I was one of those who was considering that the buzz about proteomics might be a little premature.  It's very difficult to do this.  You have to deal with five or six or seven orders of magnitude in abundance if you really want to sample all the proteins in a particular body fluid or in a cell.

And we don't have the technologies yet to do that reliably and effectively and reproducibly.  So a lot of this has just been technology development.  We're not there yet.

There have been some advances in terms of the ability to look at proteins that are in circulating plasma or serum and make a guess about what's going on internally in the rest of the body, but those have been hard won and not always well validated.  So I think the promise of proteomics mostly still lies ahead.

In terms of the discriminatory or ethical issues, I think the fact that proteins are, after all, synthesized as a result of instructions provided in DNA.  That means the connection is pretty tight.  And most of the same issues apply.  But certainly if you're talking about individual variations that pose a risk of future illness, yes, it's the same kind of story.

Now, in terms of exceptionalism, — and I know this has been an area of considerable discussion and rightfully so — it is probably the case that the fact that there is so much public concern about the use of predictive genetic information in health insurance is in a certain way an indictment of our whole health care plan and the fact that we in this country have more than 40 million people who don't have health insurance and that we are able to somehow live with that and not find that already to be a violation of certain principles of justice.  It does, I think, point you to the — the fundamental problem may be much deeper than whether or not genetics is covered or not covered.

But genetics does in the public mind and with a certain legitimacy occupy a special place.  It does, after all, carry predictive information.  It is permanent.  It is not like a blood test that's here today and gone tomorrow.  Once it's there, it's there.

It is family-relevant.  So the test you have has implications for your relatives, and other tests don't carry that.  And it's personal.  People feel somehow that this is private information in a fundamental way that may not apply to other kinds of laboratory tests that are conducted upon them.

So I do think there is a legitimacy towards carving this out, but in a certain way maybe it is also a way of saying we have a fundamental problem.  We're going to try to fix this one now, but let's pay attention to the fact that this is not the only problem that we have.

Again, your examples about why genetic information should be given a special place here and not used in a discriminatory way; whereas, others can, again, it's not being argued that if you have the diagnosis of a disease that happens to be a genetic disease, that you no longer can be subject to underwriting.

It's the predictive information in a currently undiagnosed healthy person that is off the table.  So that's different than the person who already has cancer or the person who already has a broken ankle.  It's this predictive stuff.

So you might say then, "Well, you shouldn't be able to use your cholesterol either for underwriting in a health insurance plan."  And there is some argument about how much of cholesterol is genetic anyway and where the boundary is there.

But I think the line that is trying to be drawn is saying genetics is special for the reasons I outlined and we're just talking about the predictive part.  And it is a serious problem now in research, and it will be soon in terms of clinical practice.

DR. PELLEGRINO:  Professor Gómez-Lobo and Dr. Hurlbut?  And then I'm afraid that will have to be the last question since we'll be extending the period until 10:30.

DR. GÓMEZ-LOBO:  Just a brief information question.  Could you explain to us the contents of that amendment in the bill that has been submitted to Congress with regard to the discrimination issue?

DR. COLLINS:  I don't have the language in front of me, but it basically extends the definition of a covered individual to include an unborn child.  And, again, the arguments are that the way in which the legislation is written, DNA information obtained on the unborn child in the medical system effectively are considered relevant to the mother.  And as soon as that child is born, then the legislation would apply already.  So that the argument is that there is no window of non-coverage given the way that the bill is already written.

And this addition is unnecessary.  And, of course, it raises all kinds of other issues immediately because it seems to bring another very controversial and difficult public discussion into the middle of this bill that was never intended to be a place for that debate to occur.

DR. PELLEGRINO:  Dr. Hurlbut?

DR. HURLBUT:  My last comment was inarticulate.  So I just want to return to what I think is the heart of the question.

DR. COLLINS:  All of mine were.  So go ahead.

DR. HURLBUT:  And I actually would like one more thing, too, and that is for you to sum up what you think we should do.  But first I was trying to make a distinction between interventions very early, when we have a genetic proclivity and the later kinds of things we are used to, therapeutic interventions for existing disease.

What strikes me is that personalized medicine, so conceived as you have laid it out, could end up bordering on enhancement conceptually.  In other words, sure, if you're going to intervene in something that's clearly leading linearally to a disease, that might be uncontroversial.

But then there will be other things where there will be genetic proclivities towards certain behaviors or certain dispositions of being certain combinations of your possible genomic expression.  And that borders on precluding or promoting certain possibilities within your genome.

And this would come down to small molecule interventions.   You know, that is where this will all go.  The revolution of biology is likely to be a revolution in pharmaceuticals, not gene therapies, right?

And so what I guess I am asking you is in the broad sense, are you worried about where this kind of knowledge is taking us and how we might use it to intervene in human lives to "educate" the genome to do certain things versus others?

DR. COLLINS:  That would be a very interesting discussion to pursue if we had more time to dig into it.  I think we already do this in a less high tech way.

I think parents routinely are assessing the characteristics of their children and trying to figure out, oh, does this child have a particular musical ability?  And if so, should we nurture that?

So they're observing the consequences of some genetic inheritance and trying to adjust what they do in terms of enhancing that child's ability to carry out certain valuable life activities.

Is it categorically different if those kinds of judgments are assisted by some kind of DNA analysis?  I'm not sure it's categorically different.  I'd worry that it might be seen as more reliable and that you might get into the self-fulfilling prophecy kind of outcome, where kids got early on slotted into a particular pathway of what kind of training and experiences they're going to get based upon DNA information that was actually pretty shaky.  That would be unfortunate.

I mean, I guess the thing to fear here is limitation on opportunity based upon predeterministic kinds of conclusions based upon DNA analyses that should have never been taken with that much weight but somebody might try to do.

I do think that there is a risk there that we ought to think about in the longer term in terms of society, but I think we're a ways off from getting to it.  And it would not be one of those where you could draw a bright line.  And those are sometimes the most interesting and the most difficult.

DR. PELLEGRINO:  Dr. Collins, I think I expressed the feelings of all the members of the Council, expressed our gratitude to you for being so generous in your time, responding to all of the questions in a very definitive way, and especially for opening up the whole field to us and particularly also for some of the suggestions you have made about the ideas this Council might deal with in the ethical realm.  Thank you very, very much.

(Applause.)

DR. PELLEGRINO:  We will now break to 10:45.

(Whereupon, the foregoing matter went off the record at 10:27 a.m. and went back on the record at 10:46 a.m.)