Genetics Research in Individuals, Families, and Communities

DR. SHAPIRO: I have just been informed Dr. Childress has been called to the phone, but he'll be back here in a few moments. But we are going to, again, move some of our agenda forward. Dr. Mary Claire King is here, and she is going to speak to us on genetic research, individuals, families, and communities. And, really, we couldn't have anyone better do that, and a more experienced understanding in this area, a very distinguished professor of medicine and genetics at the University of Washington, and well known from so many things, I won't take time to enumerate them all, if you will forgive me. But it's a real pleasure to have you. We are very honored that you took the time to come down here to share some of your thoughts with us. Thank you very much for being here, and thank you also for all of the work you've done over the years that's so relevant to so many people. But welcome, it's very nice to have you.

DR. KING: Thanks very much, Harold. I have to tell you all something that only Harold and I know, and that's the real reason he's ever heard of me at all is that my brother graduated from Princeton and has been a loyal alum for many years. And Paul was in the last of the classes which were male only. And since he's my younger brother. . . .

DR. SHAPIRO: Don't hold that against him.

DR. KING: I didn't go to Princeton. Had you reformed a bit earlier, I would be an alum myself.

DR. SHAPIRO: All right. This meeting is full of tragedies.

DR. KING: Perhaps, mine as well. It's both a privilege and daunting to speak to the Committee, because I'm very much like one of those children who moves to town when the semester is well underway. You all have addressed every issue I could probably bring up. And in many ways there is no wisdom I, as one individual, can add to what you haven't already discussed. What I thought I would do is tell you from our experiences that go back to 1974, three different studies that we've been involved in all happen to be breast cancer studies. Because, of course, that's a major interest of mine. And in each of these studies, which are quite different from each other, just name some of the issues that have arisen and how we dealt with them, and where we see problems, and where we see opportunities, and so on. But let me start by stating my premise of where we've come from philosophically in just a few sentences. First, that to some degree scientific evidence to the contrary, notwithstanding, the questions of genetics, genotype, and genetic identification are, in my view, different from those of any other medical problem. The parallels with HIV are real, but the differences are real, as well. The parallels with any other concern for privacy are real. The differences are real as well. There are I think two fundamental reasons for it, which are human and not scientific. The first is that everything that is genetic has implications for our children, and nothing is any more important to any of us than that. And, second, that an enormous amount of murder has been committed in the name of genetics, and we would be extremely naive to believe that that's finished. Our human rights work in the former Yugoslavia now is like dramatic evidence that even though we are working with three populations of people, which, when you overlay their genotype are indistinguishable, nevertheless are engaged in patricide against each other of scientific evidence, to the contrary, notwithstanding. So, I take enormously seriously the work that you've set for yourselves. I consider it a blessing that you are doing it. It's not fun. It's not easy, and thank you very much for spending the hundreds and hundreds of hours you're spending on the problem. So let me tell you a little bit of our experience, and then just reflect with you on what it means and what use it might be to you. First, our project with families with multiple cases of breast cancer, a project that led over a couple of decades to identification of two genes which are responsible for inherited predisposition to breast cancer. In this project, we were working—we were and are working with very large families in which breast cancer and ovarian cancer are extremely common. The work began in the era before even DNA analysis could be done. And, of course, proceeded into the era of DNA analysis and was dramatically shot forward by the development of PCR. From the time of the invention of PCR, the availability of sources of DNA from persons who had died of the conditions of concern to us, that is, from people who were in the original generations of these families and had died of breast cancer decades before became critically important. BRCA-1 could not have been mapped, let alone found, without the use of specimens from pathologic materials from individuals who had died of this disease long before our project was ever undertaken. It was also, of course, essential that we know, specifically, who each sample belonged to. It's not a question of an optimizing sample, such as a matter of it being terribly important that we have my grandmother's biopsy specimen, in order to obtain DNA, both from the normal cells and from the tumor cells, and to work with both. The way that we addressed this question was to ask the next of kin of the persons who had died, if we could work with the pathology specimens from their deceased relatives. Sometimes that was a widower; very frequently, it was a child. This offered us no difficulties. We carried out that work. We kept the names attached to the specimens as we were working through them. We talked to people about their relatives directly. It was very open communication. Of course, when we published pedigrees we were and are very careful to publish pedigrees in such a way that people who are still unaffected and young do not have their genotypes identified. They are simply excluded from pedigrees. But there is no question that we worked, and continue to work, with specimens from persons who, themselves, are in no position to be able to give us informed consent, because they have died and are never heard about. And those specimens are critical to gene-finding activities. So this is really an example, this first class of studies, of studies that involve retrospective materials, stored specimens with known people, where we do get permission from next of kin. The second kind of studies is exemplified best by our New York Breast Cancer Project, which is a project that Joan Marks and I are undertaking with now 12 institutions in the Greater New York area. This is a study of breast cancer among women of Jewish ancestry, who are breast cancer patients. It has really two goals. The first is to understand what the actual risks associated with BRCA-1 and BRCA-2 are; and the second is to understand if those risks are in any way modified by other genes, or by environmental exposures, or beneficent life events. The design of the study, to tell you just real briefly, so you'll see where we're coming from, is that every woman who is diagnosed with primary breast cancer in any of these institutions from 1994 to the present, regardless of her age at diagnosis, and regardless of her family history, is asked to participate if she self-identifies as Jewish. It is entirely self-identification. If she says, "Yes, I'm Jewish, and, yes, I had invasive breast cancer." Then she's asked to participate in the study. This is done through a stepwide genetic counseling process, which incorporates genetic counselors, who are the genetic counselors of their institution. So, it is a formal clinical genetic counseling process. She gives a blood sample. We genotype that sample first for three agent BRCA-1 and 2 mutations that are found in Jewish families. And if—I won't get into the question of what about people who are negative, but from high-risk families. That's another project entirely. But if, for this project, if this person—this patient has one of these three mutations, she then becomes the index case for a family, her family, and we work with her to trace the genealogy of the mutation in the family, regardless of the cancer status of the persons with that mutation, in order to obtain rigorous properly ascertained estimates by age of risk of breast cancer and ovarian cancer associated with mutations, and the interaction with potential environmental factors that I've told you about. Any person in this project who wishes genetic information back for themselves receives it and receives it with pretest counseling and posttest counseling, and no one is involved in the project until they have agreed and had pretest counseling. It is about 95 percent counseling, and about a 5 percent genetic analysis, so it's going along. As you might imagine, it's elicited a lot of interest in the New York medical world, and in the New York Jewish world. And, oh, sometime last year, some staff members of Hadassah became particularly concerned about this project, and I think some others. The exact history of what their—the origins of their concern elude me a little. But a consequence of their concern is that a fascinating meeting was held, sponsored by Jewish organizations in New York with senior officials of our NIH. Rick Cosner was there. Francis Collins was there. I suspect others as well. I did not attend. I was not invited, as it happened. But many of our clinical colleagues did attend. And the American Jewish Congress in the person of Lois Waldman, who is a lawyer, who works with American Jewish Congress, prepared a statement about genetic diseases in the Jewish community, which I've just received from Lois, and have consequently distributed to you. A number of us were asked if we were interested in signing the statement, and I very enthusiastically did so. So, my name is checked down here. To make a complex story short, I believe that the denouement of this experience was the following: That stigmatization, the possibility of stigmatization was directly confronted in discussion and was acknowledged as a real and legitimate concern. But that there was passion for not stopping biomedical research as a consequence of that concern. And there was, in particular, great concern that informed consent be individual, that every person have the right to either participate or not participate in a study. And if they chose to participate, to either participate anonymously, or not anonymously, and that the individual's right to make that decision for herself or himself be retained. I'll point out just a couple of the relevant paragraphs to you. But, like I said, I refer you to Lois's statement in general. Her comment at the bottom of page six, "Throughout American history, Jews have insisted that they be dealt with as individuals, not as some corporate body as in medieval Europe," and so on. And then toward the end of her remarks, "Genetic researchers and members of the Jewish community should maintain an ongoing dialogue, so they each may be apprised of the others concerns about genetic research involving Jews. The Jewish community should support passage of legislation to correct loopholes in existing law banning genetic discrimination in health insurance, and also support bills to prohibit genetic discrimination in employment, and to protect the privacy of all medical records," and so on. Development of this project in New York passed through IRBs at all 12 of the institutions, with whom we work, and, of course, at the University of Washington. And the concerns raised there are very much the ones that your Committee has been dealing with; that is, the ability of the individual to make informed consent. Since this is a prospective study, many of the issues that we're dealing with today don't arise. However, the same question that I just brought up, namely, the working with specimens from, for example, mothers of persons who have died, may come up in the future. So, it's not necessarily buffered from that entirely. Because of the unique issue of working with one of America's remarkable populations with a still coherent genetic ancestry, I have realized for myself early on that I needed to know a lot more about Jewish history in America. I am just parenthetically one of these persons who learned only as an adult, that she is in fact partly Jewish. Madeline Albright is not alone. It turns out my mother was born a Cohen, and I learned that after my daughter was born. American history is full of these stories as well. What we decided to do, in order to educate ourselves was to establish before we began the project an advisory committee, not an IRB, but an advisory committee to educate us about what we—and by we I mean Joan Marks, and the oncologists, and the clinical geneticists, and the counselors, and I thought of as the critical kinds of issues that were likely to arise from our participants. Stated socially, what does it mean to be a Jew in America? What does it mean to be a child of persons who were survivors of the Holocaust? What does it mean to be both an Israeli citizen and an American citizen? What about our collaboration with our Israeli colleagues? I haven't discussed that, but we have an exactly parallel study going on in Israel by a collaborative group there. And this committee meets about once a year. And, importantly, has undertaken education of genetic counselors in New York about those issues. So, it's been for me an enormously enriching experience to have that committee involved. They have not been a hoop through which we needed to jump. They haven't had any veto power over the study at all. They never asked for that, and that was not the purpose. But they have been enormously helpful in suggesting how to do it right. There has been mutual respect and we're learning a great deal. The third project I want to discuss is, yet, again, different. This is a project which is just now getting under way. It's our collaboration with the NSABP, and their tamoxifen prevention trial. As probably most of you know, the tamoxifen prevention study of the NSABP has just released its first results. The nature of this project was that women who were at high risk of breast cancer by dint of family history, or age, or personal threatening history, but not actually having ever been diagnosed with breast cancer were asked to volunteer for the project. And if they volunteered for the project anywhere between five and eight years ago, were randomized to either tamoxifen or placebo, were asked to remain on the regimen to which they were randomized for five years. And at the beginning of this study no one had breast cancer. Thirteen-thousand women volunteered for it, were randomized 50/50 into two groups, and the study proceeding forward triple-blind, with neither the women, nor the physicians, nor the investigators knowing who was in which category until one committee, which consisted of only three statisticians, in evaluating the results reached a stopping rule, which had been previously defined on the basis of statistical grounds. So when the stopping rule was reached, that is, the stopping rule was either the tamoxifen would have a benefit that was significant at a preordained statistical level, or would have a detriment that was preordained at a statistical level. When that stopping rule was reached, the results were revealed. That happened, it happened earlier than they had anticipated. The results of the trial are that tamoxifen—that the women who were randomized to tamoxifen have 40 percent fewer breast cancers than the women randomized to placebo. Concomitant with that, the women randomized to tamoxifen have higher risks of endometrial cancer, so this is not one of these entirely beneficent situations, and each woman needs to decide for herself. And that's a whole set of issues in itself. My interest in this for the last eight years has been whether women who would turn out to have inherited alleles predisposing them to breast cancer, would be particularly benefitted, or particularly not benefitted by taking tamoxifen. One could make the argument either way, depending on how one interprets biological roles with BRCA-1 and 2, which is hardly a settled matter either. So what—the study that my lab designed with the NSABP and the NCI is to—well, I need to also tell you parenthetically that each woman at the time that she entered this study gave a blood sample and signed a consent saying that this sample could be used for investigation of biological markers. There was no specification to BRCA-1 and 2, because they weren't known yet. They weren't named by name. So, the study we designed from the purely design point-of-view, was to take the blood samples from all of the women among the 13,000 who had developed breast cancer. And about three women from the cohort, who had not developed breast cancer for each woman who had, so about a thousand samples in all. To genotype in my lab, all thousand of the DNA extracted from each of those blood samples all the way through BRCA-1 and all the way though BRCA-2, so as to determine who had mutations in those genes, and then to answer to questions. What is the absolute benefit or detriment associated with using tamoxifen if you are a mutation carrier? And what's the relative benefit or detriment? There, clearly, is not going to be a way to say this with any level of significance, but we will at least get some point estimates. The question, of course, arises, how does one do this? To what degree does one anonymize? I used to think anonymity was like pregnancy, but I now realize it is not like pregnancy, that there are degrees of anonymity. Right. So, we turned to the Participant Advisory Board of the tamoxifen or of the NSABP, and, in particular, of the tamoxifen trial, who are women who are on the trial, and discussed it with them. To take a couple of extremes: at one extreme, we would have attempted to go back to every one of 13,000 women and asked if we might put their name into a hopper to be randomized to being in the trial or not, and not have put their name into that hopper, unless we could obtain that consent. That was not feasible. As in any other large randomized trial, large numbers of people drop out. It was possible to follow people's ultimate cancer status through publicly available records. But something like 15 percent of people from each arm had dropped out, just as people do, and we would—expense aside, we would have had a highly incomplete follow-up had we tried to go back to every woman and get explicit permission to use her sample, and we would have had a serious bias to the trial. So, I said I would rather—I said to myself, "this is a deal breaker." I would rather not do the trial after that. Also to take the opposite extreme, we would have tried to contact every woman and give back to her, her individual results. I also oppose that, as did the Participant Advisory Board, because the samples, although they were collected by sincere, extremely well-meaning investigators, were collected at hundreds of different hospitals. They weren't collected using clear guidelines. There is always the possibility of sample mix up. And while that's a detriment to a statistical analysis, it's devastating to an individual. So, if someone needs an individual result, they really have to do it properly with prospective collection of sample. So, we were really left with some intermediate possibilities. One was to—I'll tell you the two that we've struggled over. They start the same way. That is to put all names of all—well, no, I'm leaving out a step. The women in the NSABP tamoxifen trial receive about four-times-a-year a newsletter. And we've been describing the intent of doing this genetic study for years now. And in the most recent newsletter, the study is described, and so that it will get under way. And in this most recent newsletter it says if you do not want the sample that you gave years ago to be used, send back the send-back card and we'll remove you from the randomization, and the time to send back that card isn't over yet. So, I don't know how many will get that. But, barring that, everybody's name goes into the randomization, and about a thousand names are pulled, including virtually all of the cases. Then the samples are taken from the freezers, and are taken to an independent laboratory, a clinical laboratory, where DNA is extracted, and the NSABP identification number is removed and a consecutive number, one to a thousand is put on the DNA. The DNA is given to me. Any extra DNA is returned with its original NSABP identification number to the original freezers. I never know who these individuals are. At the same time, the NSABP statisticians who have assigned these one to one thousand numbers extract from the enormous record of information on each person the critical pieces of information that they and we perceive that they're going to need. So, this is all material that's been obtained up to this moment. For example, do they have cancer or not? What year were they born, and so on? And that's put into a separate file with the consecutive number. We do the genotyping. We give the genotypic information back to the statisticians at NSABP, and together we do the analysis and we get an answer. The critical question then is, is the link between the identifier in NASABP, it's original identifier—which, of course, could then be traced back to the woman herself- and the consecutive number kept or destroyed. The way that this study now stands, and the way it's been passed by all of the IRBs, that link is destroyed. The reason we decided on that is that this committee hasn't finished its deliberations yet, and we don't know what your wisdom will be and we want to be conservative. The downside of this is that we will never be able to do—

DR. SHAPIRO: Say that one more time.

DR. KING: I said we don't know what your deliberations will yield. It's perfectly possible that your deliberations will lead to a recommendation that studies of this type be completely anonymized with no possibility of anyone ever getting back to the individual, which we will not now be able to do. And we don't want to put ourselves in conflict with this Committee at any time in the future. The downside is we will not ever be able to do the survival studies, and I am very concerned about that. The Participant Advisory Board is, if anything, more concerned than I, but it is the price we've paid. So those are three of our experiences, and those are where we stand. And I have no special wisdom that’s what's up. I'd be happy to talk with you.

DR. SHAPIRO: Thank you very much for describing these studies, and this is not my area of expertise. But I really, and, nevertheless, must say I'm very impressed with the care that you've taken in proceeding in studying diseases in this particular community, and it sounds very impressive—I'm sure to my colleagues as well. I have a bunch of questions, but let me see if other members of the Commission want to speak. Yes, Diane, then Bernie.

DR. SCOTT-JONES: Thank you very much. And I have a lot of questions, but I'll limit myself because I'm sure everyone else has questions for you as well. I was very impressed that it seemed that your learning about the Jewish community in your work was something you saw not as a burden, but as an interest in the population that you study. And I suppose that the ideal, and how we promote that through the variety of mechanisms that become available to us for promoting scientific standards and ethical standards probably promote just what we did.

DR. KING: I think it's terribly important that the role of the citizen advisor, as educator, and the role of the IRB, as regulator, be separated. We need citizens on both, but I think they are different roles.

DR. SCOTT-JONES: Could you say more?

DR. KING: On our committee that Joan and I put together—anyone who knows Joan Marks, knows that for Joan Marks to put together a committee like this in New York, where there are people pounding on her door to be on it before it was over includes a psychiatrist, a historian of Holocaust studies, a journalist, no one medical. The other people are women who are philanthropists, and I don't know their original professional training. But everybody in the committee is Jewish, except me, then it turns out I am, too. I didn't know when we started all of this. So, anyway, what that committee does is educate us and help Joan, and to a lesser extent me, establish lectures, and seminars, and courses, which are required for the counselors who are involved in the project. That's teaching. That's not the same thing as our going to them as an IRB and saying we need your approval for this. It's a mutual teaching process. And I think it—I think it allows us, as scientists, to let our hair down, and to express our insecurities over these issues. I mean to take one very dramatic example. I said we have—we encounter almost no one who has inherited mutations who is elderly and has not yet had breast cancer, but there are a few. And one is a woman who is a Holocaust survivor. She since died, but she was alive last year. She's in her eighties. I said how is anybody going to talk to this woman about her experiences as a young woman? I mean this is beyond—completely beyond the experience of anyone, and we worked out a way to have this person interviewed with the help of a person for whom this was not beyond their experience. And I was able to admit completely freely my need for help in that area. That's very different than a regulatory process. So, teaching and regulation are both essential, but should be independent I think.

DR. LO: I, first, just want to thank you for a wonderful presentation and a lot of food for thought. Two questions for you about the tamoxifen study. The first has to do with deciding whether you can use someone's sample. Others have suggested that in a study like that where you really have a well-defined population and a participant advisory board that there be some sort of surrogates for individual consent since you have to have 95 percent of the people who you can contact that would agree their samples to be used. That gives you perhaps some ethical warrant to do more than the opt-out. So my first question is about your thoughts on the other options you could have employed. My second question, which I think really is directly in line with some of the issues we've debated, is how you handle the double coding at the NCI and then the code that's given to you with your thousand samples. At one point in our deliberations, we had talked about the status of the sample as the researcher has it which may be different from the status of the sample in the original repository. So as I understand it, the NCI can link everything...No, but it passes its use in a way that you can't link back to the original but they can feed you certain information. And what your case really illustrates is several things, but one is certainly the cost that you pay when you give up not being able to retrace that linkage. And let me elaborate what I think is the situation. I mean in breast cancer disease, you know it's a 20-year disease, and so whatever trends that were strong enough to prompt the stopping of tamoxifen are of marrying up the totally predictive of long-term trends. And so this is a disease where if there ever was a disease where you'd want to be able to walk back and get feedback at 10 years, 20 years, on the incidence of the cancer, this is one of those. So that's one thing, just that we keep in mind the cost of policies that are intended to enhance individual autonomy and things like that. But the second point is that we have also considered what may be in fact computer science fiction: but encryption schemes, where you continue to be blinded to the ultimate identity of the patients but you're able to get updates on the breast cancer status of the sample through the central registry as they get it from whatever cancer registry direct. So it seems to me it would be important to try and develop both the technology to do that—what we've called sort of one-way communication—both to really look seriously at the technical ways of doing that and still protect confidentiality, but also to carve out the ethics of that because it seems to me your case is an example where my initial reaction is, let's try a way to protect those people who really don't want to know of, don't want to be bothered by recontacting, but somehow to also answer the very pressing scientific questions that you played out, which ultimately, I think, are really going to provide tremendous clinical benefit to people at risk for the devastating condition. So I think it is really going to help us to try and puzzle through this third case with you because it really touches on issues that we have struggled with. I guess my final point is that I guess it bothers me that because we haven't said whatever it is we're going to say, people are rather trusting their best judgment, it seems to me that everything we've heard and everything I know about your work is that I would trust you as an investigator with your colleagues and your advisory boards to do the right thing in ethically perilous territory. I think there's a cost to the perception that until we've made a decision, investigators take the most "conservative approach," but it's conservative from a certain point of view, it's very wasteful from another point of view. And I think that's something we have to...I mean colleagues of mine have expressed their concern at our not making some sort of statement, which then may or may not get picked up as policy, but I think your case really drives it home that our not acting in a timely way is having a real impact that's not going to be reversible on studies that I think anyone would say are really important, significant vital signs that need to be touched.

DR. KING: That's exactly right.'s not scientifically or medically justifiable to do what we are doing because we will not be able to answer the question if a woman develops breast cancer and had an inherited predisposition and was on tamoxifen, is she best advised to have another round of tamoxifen, to stay off it, is her chance of recurrence less because she's already been on it. We won't get any of the that information about people with inherited predisposition because we won't be able to follow them up. The way our coding scheme works is that—the way it's set up to work right now—is that the NSAPB statisticians who are the people who actually do the analysis—who, incidentally, are superb—will make the file of the basic clinical outcomes as they now stand and the basic demographics that they need. They'll put that file in place. Then that file will have numbers attached to it, which will be the same numbers that will be given to the clinical laboratory who is doing the DNA extraction, and then the link will be broken. I will get the DNA samples with only numbers one to a thousand on them, and by that time that's all that will be on the file that the statisticians will have, so none of us will have the linkers. The material from the clinical laboratory will go back to the repository with the original numbers on it, but those numbers will no longer have been physically on the same tubes with the one to one thousand on them so nobody, short of reverse genetics and sequencing the entire gene of the sample, nobody would ever be able to go back. There are ways-and the statisticians developed one when we were more optimistic about being able to do follow-up for clinical outcome—of doing encryption so that I, as the geneticist, don't ever learn who these people are. I don't need to know that. However, there is no way you can have clinical outcome and not have some human being know what all the information is on that person, someone has to be able to know. Now, bearing in mind that these are people who are caring for these patients, it is an extreme position to take to say they don't have the right to have any of this information, but that's what we've said. To go back to the surrogates versus the opting-out strategy, again, we took the most constrained route. It has added, oh probably, eight months to getting the study started, and it's added a great deal of cost. In this case, the NCI was prepared to sponsor that cost so we were all right. It has had, I think, real costs in public health costs because when we decided on the opt-out strategy rather than the surrogate strategy, none of us thought that the results of the trial were going to be coming online so quickly. And consequently, we thought we had a little bit more time to play with. The reality is I still don't have those samples in hand. It's going to take me a year to 18 months to do all this genotyping and I haven't even been able to start yet, and I've been talking like this since 1990. So this is not a good example of getting something done in an efficient way, and yet, everybody involved knows what they're doing. So there's got to be a better way, right?

MS. CHARO: I always hoped this Commission would have an impact, this is not the one I'd hoped it would have. Mary Claire, if I understood the protocol correctly with regard to situation number three, the tamoxifen study, the only thing that you would have had to do differently in order to be able to retain those code links and get all the follow-up that you want, and still be in compliance with the federal regulations, the only thing you would have had to do is to have sought consent as opposed to simply having an opt-out provision. You were already mailing to people at known addresses, and you chose to mail out something saying if you don't want to be in here, mail this back. If all you had done was to mail something out that said, Would you be willing to participate, you could in fact have kept all those links under the federal regs, no problem. You would have met every requirement. And since up until now, as of now, the discussion on this Commission has been really about how to enhance enforcement, appropriate enforcement of those existing regs, it seems particularly tragic you would have destroyed those links. I'm asking how much more expense, how much more delay, how much more sample size reduction do you think you would have had if you had taken that last step in order to uncomplicate your lives and allow yourself to get the updated info?

DR. KING: It's a scientific problem, and naturally we discussed it. The difficultly is that the people that you lose by an opt-in strategy are not a random sample of the group as a whole. For example, you lose all the women who died of breast cancer. And you also lose all the women were lost to follow-up for the trial. And that's a lot of people and they are not random. If they were random, it wouldn't change your point estimate, but we did simulations on how much nonrandomness/randomness in order for us to follow up, rather, for death or other reasons we could sustain and still have a meaningful response, and decided it was potentially devastating to the results, that we couldn't maintain the statistical rigor.

MS. CHARO: Just a clarification on this, if I may? For those who died, their samples could be used without any kind of consent because they're not considered to be subjects under the regs and you're allowed to use their stuff with impunity. The question being, could you know who those people were so you'd be able to pull the samples out of those people who are now dead and say, fine, these are okay, and add them to the samples of the people who'd opted in. So, in fact, it would have might have been possible if you can identify them to have gotten at least that subset of nonconsenters back into your sample, leaving only those who actively, who failed to consent and are still living. Either lost to follow-up or genuinely failing now.

DR. KING: Right.

MS. CHARO: Earlier today, Diane was talking about some of the challenges of statistical validity that always exist in study design and I'm now getting interested in understanding exactly how significant this problem is as compared to the general problem of validity that Diane talked about because the current regs, if the current regs have this high a price, we should be very much aware of it.

DR. KING: This, of course, is the world I come from, and it doesn't take very much ascertainment by us to throw point estimates off drastically. I think had it been only a matter of the decedents we could have dealt with it. The greater difficulty is that when everyone asks 13,000 people to send back a card saying they're opting in, your response is very, very low. And the cost of going back and back and back and back and getting that up to 80 percent, let alone higher than that, is extremely high. Then you have to prove to your own satisfaction that the people who haven't responded are not biased in one way or another. It's ... it was the statisticians that basically drew the line. We did a little pilot and the response in sending back the card to the pilot was something like 30 percent, 70 percent nonresponse.

MS. CHARO: And this was with highly motivated, educated, newslettered people.

DR. KING: Oh, terrific people, newslettered people. Basically they're probably newslettered to death, but what can I say? And it wasn't that they weren't interested. I mean you follow up with this little pilot and you say why didn't you do this, and they say, oh, but you already have my sample, of course you can use it, oh geez, did I not send it back? It's a big study. And it's a sort of a specimen of these large trials that one will need to confront.

DR. SHAPIRO: Other questions? Let me go back to something you said right at the very beginning in dealing with genetic information, and your own view as to why it's different from other kinds of information. At least as I recall, one thing you mentioned was that it had implications for one's children and so on downstream. And the other I wasn't quite sure that I understood what you said. It sounded to me like you were saying this kind of information's been misused in a serious way, very often in a historical sense, or be used for...I didn't quite understand. I mean I understand what's going on in the former Yugoslavia, but I didn't quite understand how that related to genetic information.

DR. KING: I think the point I was trying to make is not that legitimate genetics has been used that way, but that all of our concerns, just as citizens, not as scientists, about the way that a stigmatization, racist arguments, and eugenic arguments have been used against people are real. No question about it. The fact that it has, the fact that one can use population genetics correctly to debunk those arguments is important. But it is not sufficient. It is not sufficient for me as a geneticist to be able to say look, there's nothing new here, this is all the same as it was before. There are too many historic counter-examples. So what I'm saying is that the problem is a historical one and not a problem with genetics itself. Does that help?

DR. SHAPIRO: Okay, no it's very helpful. Let me ask you about what I think was the second experiment, or second...well, it had to do in any case with the need when someone is deceased. You went to their next of kin. And I was trying to think here how you figured out who that was. They might have had two children, they might have ten children, they might have I don't know what. How did you figure...and what if they disagreed? Let's say there were just two children and they disagreed. How do you deal with issues like that?

DR. KING: Right. Typically the way families come to us is by self-referral or referral of a physician of one case. And we quickly become bonded with the family, with lots of people in the family. So very frequently next of kin—for example, if a woman's mother has died and a father's long since deceased also we ask that woman for permission. We have not had the worst possible scenario in that, a situation in which there were, let's say, two daughters, and one said don't you dare look at my mother's sample and the other one said I insist that you do. That hasn't happened. We have had a number of times a situation in which one daughter has said I want the study to proceed, I want our mother's sample to be used, and the other daughter has said I don't want to be part of this. And then, as you can see, we proceed. And the issue of intrafamily confidentiality is what becomes enormously important: the daughter that doesn't want to be part isn't part. I think what we would do...the scenario you bring up is analogous, I believe, to a scenario that has happened, a number of times, in which we know a woman's genotype, we know that she has a predisposing allele, and she doesn't want us to tell anyone in her family. Or in particular, she doesn't want us to tell a younger sister who is at risk, 50 percent risk, of inheriting the allele. And we are bound not to tell that younger sister. We work on it over months. That's why genetic counselors are trained the way they're trained. And we simply discuss the issue with the person and hope that in time the person will come around. I think, and so far—knock on wood—each time that's happened they eventually have. There are all kinds of personal things that come into it; eventually each time it's worked. I think if we did confront the scenario you present we would do the same thing. We probably wouldn't use that sample until we had gotten past the unhappiness of the other sister. But we'd just keep talking to them. We're talking about real people here, and we have real relationships with them, and the relationship doesn't stop when we know their genotype.

DR. SHAPIRO: If you were—and I apologize if I don't ask the next question in a coherent way, I'm not sure I have a coherent question, but there may be something here—if you were trying to design a study which required genetic material from lots of members of the family, some of whom were deceased, and then going on a few more degrees in one way or another, and the question is for the deceased relatives, the deceased kin this case. There's some misunderstanding...there's some kind of ambiguity about current regulations, whether they are human subjects or not. On one hand they're not; on the other hand, they have information which impacts other people who are living and one could argue that they are. Our understanding of current practice is deceased is deceased, and this other issue is they're just not treated as human subjects for the purpose of the regulations. But the kinds of studies that you're concerned with and deal with, would it be a huge and super—not be huge—would it be a significant barrier if you were told that if you want to use the material from deceased kin that you have to get permission, as you've mentioned next of kin, whoever the first living relative is with the oldest living relative, would that be a significant barrier, an irrelevant barrier, a big problem, small problem, if we began treating these deceased individuals in this sense as kind of human subjects? When you consent of one form or another from their living relatives.

DR. KING: Are you referring to a situation in which I, as the investigator, need to retain the identification of the person?

DR. SHAPIRO: Yes, right.

DR. KING: Right, because for the anonymous case it doesn't matter.

DR. SHAPIRO: That doesn't matter.

DR. KING: Right. I think any, whenever anyone is working with real families, any regulation that sets any one policy in stone is probably asking for trouble. I think that...I have no problem at all in asking for a next-of-kin agreement. We've always done it, and I can't imagine that we wouldn't always continue to do it. It's simply a matter of showing respect. What I would find very burdensome and would probably—I mean it would probably be a deal-breaker for me, I probably wouldn't do this work any longer—is if I really had to go to everyone in the family because that forces people into dynamics with each other which they may choose to avoid. And it destroys the privacy of the individual. And so I think this becomes one of those situations in which it's necessary that individual investigators and individual IRBs and individual genetic counselors—who are after all trained in this kind of thing—see cases one-by-one, and when it makes sense to work with this daughter rather than this daughter that they be allowed to exercise that judgment. But that families be respected and that there be "a" next-of-kin permission granted. That in itself I don't find and anathema at all.

DR. COX: So, Mary Claire, I, too, would like to thank you very much for coming because you give a perspective of a person in the trenches who's thoughtful about this, that I think that oftentimes we don't get before the Commission. Having said that, though, I'd like to come back to this issue with respect to the bias in the sample from a scientific point of view because I'm having trouble with this. I'm not an epidemiologist and so I really am seeking clarification on this. Certainly I can envision that if you had only 30 percent of the sample it could be biased, but it might not be. And I don't have any idea for figuring out whether it is or not. On the other hand, doing an opt-out, even if only ten percent of the people that opt out, how does one know that those don't bias the sample? So I'm having trouble with the logic here.

DR. KING: Right. There's a huge statistical field devoted to this problem, right? The way that one can determine whether one's opters-out are biased is that a person—this all happens before there's any randomization at all into my study—one's opters-out send back their cards and the statisticians can say, "Right, the people who are opting out are people who are over age 70." Right? And if it's something that simple, one says right, this project, therefore, is generalizable to people under age 70. If the people opting out are people with a family history, that poses something different in terms of genetic analysis. The relationship between multiple, independent variables influencing a trait and the degree of disproportionate opting out that one can sustain, and the consequences of that for point estimates are, as I said, a matter of statistical analysis for which there are protocols that one follows.

DR. COX: But that approach could really screw you, too, right? Because the people that choose to opt out could be very nonrandom and it would make it very difficult to use the study.

DR. KING: Right.

DR. COX: So I see that you could get screwed in either direction.

DR. KING: You could, in principle. However, in or experience with lots of different opting-out and opting-in pilot studies and real studies, asking people to opt in when the sample is very, very large is enormously expensive and fundamentally unwieldy. Opting out simply works better because there are far fewer people...people, if they don't want to be in a project, are much more willing to send in a card saying I don't want to be in a project than are people who are willing to opt in. I mean, I myself, as a subject, without realizing it when it was at Kaiser, of one of these opt-in studies, I forgot to send the card back. It's just something you have to remember to do.

DR. COX: So I understand that point. It's the point that you just made was an unwieldy one, but not a scientific one.

DR. KING: But it becomes scientific, David, because if only 30 percent of the people send a card back saying yes, I'll be part of the project, it would be remarkable. It would be testable and remarkable if those 30 percent were a random sample of the overall group. And chances are they won't be and you will lose your ability to work.

DR. COX: Because I just wanted to be very specific about this because there are several reasons not to do it. But you're saying in addition to the unwieldiness with very large samples, there's a scientific reason. So my final question on this is what's a very large sample?

DR. KING: The interfamily studies, clearly we have an opt-in strategy, right? In our New York breast cancer study, we have an opt-in strategy. I think draws the line depending basically on how much time one has before the issue becomes either devastating to the world, as in the case of AIDS, or moot as in the case of tomoxofin and breast cancer. Any amount of money that one has to do triple follow-up through the various ways—through mail, telephone, and personal visits of the people who don't send back cards. It is a very expensive thing to do a full-scale follow-up study. I mean it would be millions of dollars for a sample this large. So I can't give you a number that would apply to all studies. But I can tell you for the New York breast cancer study, numbers don't work, David, because it depends on where you are. The New York breast cancer study, because that involves so much activity on the part of the participant, that is, if she wishes to know her results she has to go to pretest counseling and genotyping and post-test counseling. I am sure we are getting much, well we're getting a 60 to 70 percent assent to be in that study. I am concerned that the people we're losing are not random. I'm concerned that we're losing the elderly women, I know we are. I'm concerned that we may be losing people who have less family history. But I don't know any other way around it. So you have to use your brains; it's a balance of the scientific questions you're asking, the money you have to spend on it, the time you have to spend on it, and the actual risk to the real person that's involved.

DR. COX: That's very helpful, thank you.

DR. SHAPIRO: We have three or four commissioners who want to speak, and then we're going to have to move on to our next subject I’m afraid. But the four I have are Bette, Larry, Alta, and Bernie. And Diane, you want a last question also?

MS. KRAMER: Thank you very much for your presentation. I sit here as a public member and I've spoken with a couple of the other Commissioners informally earlier today because as I sat here this morning and I listened to the conversation, I have this very, very real concern that we are working so hard to eliminate risk from the research activity that we may be inviting an enormous cost to research activity and to what it promises for the public. And when we had mini-hearings back several months ago, we heard consistently from the people who participated that they, insofar as they represented the general public, that they were not really concerned about this. They were only concerned to the extent that they didn't want their insurance companies to get that information. But aside from that, they had a commitment to the idea of research, they had a commitment to the notion that there's a public responsibility to participate since the public was going to be the beneficiary of the research. I'm curious, I have a couple of questions. In working with either the people who participated in your study or the advisory group, what kind of feedback did you get from them? Did you find that they, in fact, were worried about this? Or did you find that in fact that you who were doing the study were more concerned than they were? And do you, in the last analysis, do you have anything to suggest to us that could help us as we try to strike a proper balance between protection and permission for the research to go forward?

DR. KING: My experience in all of my work, across all studies, not just these three, has been exactly the same as yours. This is all my breast cancer work. Participants have been very committed to the work, and participant advisory boards have been very committed to the work. They make remarks that are anywhere from slightly patronizing to quite caustic about the degree that we obsess over these things. I mean someone said, "Our tax dollars at work again?" Someone else said, "How many angels do you want to put on the heads on pins?" Somebody else said, "Wake me when it's over." However it's true that they see only one part of it, and they don't see the whole picture as you do. Surrogacy instead of opting out, let alone opting in, would be an enormous help. I honestly think opting in would kill the field. Opting out we can tolerate but it has costs, both financial and otherwise. Losing the ability to follow up has, as several of us have now said, has enormous medical and scientific costs. And if there is a way that we can maintain our ability to obtain critical follow-up information on people in large cohorts, it would be a tremendous help.

MS. CHARO: I find myself wondering a couple of things here. First, by destroying the links you actually eliminated the requirement to obtain even an opt-out. At that point you could have used the stuff with impunity, saved yourself the statistical challenges that David has described, and yet you chose not to, and I'm curious as to why. Second is, in light of your experience and your certainty that opting out is far... to allow you to use their stuff in the future, either with no further consenting process or with an opting out process, in other words allowing people to prospectively contract with you for a less onerous consenting process, as a way around this problem, at least prospectively.

DR. KING: Right. The prospective studies I'm involved with use very much what you described there at the end, a process in which the ... for example, we're working with the cohort from Kaiser of young adults with inherited...with a hearing loss, and we want to find out if it's inherited or not, and at this point we have no idea whether it will be or not, so it's going to be years in the future before we know. And it has this kind of tiered system.

To be blunt, the reason why we went through more hoops than seem reasonable is because there's a lot of confusion out there, there's a lot of concern out there that if we had not gone through all these hoops that the study would have to be shut down at some time in the future. The concern is in part from our colleagues who work inside the government and who are trying to do the best they can in a very uncertain situation. It's...I can't defend the logic of what we did. I can simply tell you that our goal is to try to get to women good information and not to be completely paralyzed in getting good information to them. And there were enough people, both from the intramural NCI and among us who were concerned that if we didn't do this in the most constrained way possible that we might have to shut the whole thing down later on. Your logic is indisputable.

DR. LO: Well, I'm still trying to understand this enormously complicated dilemma that you're facing. It seems to me there are a couple issues we haven't highlighted yet. One is the sort of rapidly changing HMO field. So when consent was first obtained in the Boston study, you asked for permission to use biological markers because that' didn't really know what they know. And one of the things we're going to have to deal with is samples that some consent was gotten in a meaningful way but you couldn't have anticipated what you were going to use it for. And somehow, is that different than samples where no consent was obtained or only general consent to sort of use it for research? I think we need to think about the grandparenting clause for samples where, you know, an honest attempt was made to get 1994-level detailed consent.

DR. KING: If you write in Shakespearean English, you shouldn't be precluded from now speaking about it in modern English.

DR. LO: Right. The second issue has to do with an appeals process, that you were faced with a situation, it seems to me, that what is behind the original CFR regs wasn't going to anticipate it. And rather than be forced into the situation of having as a matter of prudence to take the most conservative interpretation, it would have been desirable to have some working body that could have said, let's look at your situation as a special case and see if, notwithstanding what the regulations say how they were interpreted, they really shouldn't apply to this situation because it's a different ethical situation. And I guess I'm trying to find ways, because this is going to come up again no matter what this Commission recommends or what gets adopted as policy, whatever is 1998 standard consent for future uses, someone's going to come up with something next year or the year after that we really didn't specifically ask for. And are we going to face this problem again later where people are going to get a new type of study that has different risks and benefits and people will say, well you didn't really ask about that, you asked in such a vague way that who knows what this means.

DR. KING: I like your idea of an appeals process, and the woman who spoke just before lunch representing IRBs the world around might well have ... she and her organization might well have something to contribute to how to set up an ongoing appeals panel. It would have been a big help for us in this situation had something like that been in place.

DR. SCOTT-JONES: I would like to have you say a little bit more about what you're calling "opting in" as a process in your, what I thought was a very strong statement that opting in will kill the field. I was very impressed by your concern about the Jewish community and efforts that you're making to have ties with them, which I think is wonderful to have an ongoing relationship with the population, the community, that you're studying. But it seems that opting in, well what you're calling "opting out" is like what's called passive consent in some other research areas. And that is that you don't actually consent, you just ask people to respond if they don't want to participate in the research. And, of course, it isn't actually consent because you never document that the persons are giving you their agreement ahead of time.

DR. KING: Bear in mind that my discussion pertained to a situation in which a person had given a sample for whose use she had already consented. Right, specifically to that situation. When we begin new studies, they are all opt-in. Right. We are talking about in the tomoxofin case, a situation in which all participants, or virtually all participants, gave a blood sample five to eight years ago-it has to be at least five because four years ago VSRA-1 was cloned-and they said, "Yes, you may use this blood sample for biological markers." But because the word genes was not used, BRCA-1 and 2, this ... all this other concern arose. So these people have already consented, the materials are stored, so it's, as Alta would say, the second step in a process where original consent was obtained. But the question was how does that consent apply to this second situation.

DR. SCOTT-JONES: Okay, so your statement, "opting in will kill the field" was not meant more generally?

DR. KING: It was meant to apply to the situation in which there are stored materials, which were...which are going to be used anonymously, and which were obtained for reasons which in the context of the historic time made perfectly good sense.

MS. CHARO: I'm now...I just want to make sure I understand your answer. The...let me just run a couple of scenarios by you and ask you which ones you're referring to -this may be the easiest way for me to get it straight. You've got a lot of samples that were originally collected in an explicitly-research context in which people said, yes you may do research on my samples. And the problem is that the research that was envisioned at that time does not encompass today's research, and so you're faced with what should I do to go back for a new stage of research: opt in or opt out or nothing. And I'm assuming that ideally you would like to be able to retain the linkages, get the best scientific work out of this. And here your claim would be that an opt-in strategy would kill this area of research.

DR. KING: Thinking in terms of large cohort studies, yes. Assuming that we want to retain the links, if we are ...

MS. CHARO: Right, right, of course. If you don't retain the links actually you don't need to be doing any of this stuff.

DR. KING: I need to take that on a little sign home with me.

DR. COX: We're going to have it on a plaque.

DR. SHAPIRO: David's been busy tattooing things on people's bodies all day long here.

MS. CHARO: You've got a bunch of samples that were collected-not in a research context, they were collected in a clinical context. They are sitting in all sorts of path labs, and you would like to access these. It will be the first time these people are ever being contacted about the possibility of being research subjects. You would like to retain the links in order to maintain highest degree of scientific importance for your work. Are you saying that an opt-in strategy cannot be tolerated here either, because of its cost to science?

DR. KING: Those weren't the questions I was addressing because it doesn't happen to be the one I'm up against. But I believe it is true. Let me say it in a slightly different way, and that is that I think that the number of important studies that could be done will be drastically reduced if that scenario is precluded, if everyone whose biopsy or pathology specimen now sits at UCSF must be contacted individually before that specimen with concomitant follow-up data for the future can be used for a study. How, what would I personally do in a real situation? I would have to confront the real situation and decide, because typically we use these anonymously and we don't try to follow up. I've belabored this already. But there will be enormous medical costs if we cannot use follow-up information in those scenarios.

MS. CHARO: And one last thing, because we were talking earlier about economics and practicability, etc., and now, I don't know, many of you may have had the experience I've had recently, I changed phone companies. And that means that I have now seen exactly what money can buy in terms of constant re-contact. And I'm wondering if—you now, anybody who's been through this knows what I'm talking about, right? If this were considered a serious concern, if the kind of effort was put into going back to the people whose samples are currently in collections around the country and getting prospective, general, or layered consent that would obviate these problems in the future with appropriate incentive, like the ones that ATT, MCI, and Sprint keep offering me, keep upping the ante on, right, would you think it's money well spent? Or do you think that this is really just not important enough that the risks are not great enough that this is an overly protective stance that we are taking?

DR. KING: It think that would be an enormous waste of money. And I think it would even be more negative than that. I think it would carry out the specter of Big Brother to an extent that you wouldn't have anticipated, that people would have said, how on earth did you find me, what are you doing following me around? You've got my specimen, goodness sakes, why don't you just use it, why are you haunting me, why are following me? I think we have to recognize that there are downsides to invading people's privacy, and I would consider it, from my case, I had a lot of surgery when I was a little kid, if somebody got back in touch with me and said, we now know where you have been all these years, we have found you and we want to know about this specimen and can we use it, my first reaction would not be, sure you could use my specimen, although that doesn't bother me if they use my specimen. But it would be what are you doing tracking me down? So I mean Big Brother cuts both ways. Big Brother can be paternalistic and it can still be aggravating to be followed by Big Brother. And I'm speaking now not as a geneticist, just as a person who was a kid with a lot of surgery, you know.

DR. SHAPIRO: Okay, well thank you very much. I really appreciate very much your coming here, it's been very fascinating to have this discussion with you; I wish we could spend more time. Unfortunately we have to get on with parts of our agenda. Thank you very much for coming. Thank you. Let me tell you the good news and the bad news. The good news is we're going to take a five-minute break; the bad news is by the time you come back you should have read page 3 and 4 of this memo so we can help Jim get that discussion started, which is really a critical... It's the, excuse me, it's the backbone of the capacity report. But the items on the next two pages will be quite important, and we talked around them for a long time, so if we can come back in five or ten minutes having read, or reread as the case may be, that part would be very helpful. Is that all right, Jim?

DR. CHILDRESS: That’s right, and we will pick up where we left off because there are still some comments on page 2 at the bottom. And we'll also have to pick up an earlier one we omitted. I don't know whether Laurie's going to be joining us or not.

DR. SHAPIRO: Okay, let's get going. There are still one or two Commissioners who will join us in a few moments, but why don't we begin. Let me turn the chair over to Jim to pick up our conversation of late morning.