Discussion Continues on the Research Use of Human Biological Materials

DR. SHAPIRO: Colleagues. This morning we were looking over the various recommendations/guidances etc., various statements if you wanted to get some discussion on. We got all the way on page 7 of the memo that came along with the human materials draft. We got through recommendation 6; that is, we got some comments/materials on that and we'll recast that and write along the lines that have been suggested this morning. We now have the two next recommendations, recommendations 7 and 8. Really we perhaps ought to consider together whether one is sort of a notion again which we may want to change or shift the focus on which recommends that somebody else do something, namely some scientific community, whatever that is, to agree on a set of practices that would eliminate the need for a complex recontact efforts. And recommendation 8, either the therapeutic or the research, clinical or research contents. And recommendation 8 really deals with a similar matter but that has to do with collection of prospective collection of future samples. Now the, as I see it at least, the key issue here is whether and to what extent we want to encourage comment on, say, something in detail about the nature of the consent that would be obtained when collecting materials in ways that would make things simpler in the future. Obviously this deals with prospective samples, and so the question I want to ask is is this something we want to address and if so in what way do we want to address it? Do we think it's a good idea; in that case how do we want to opine on this?

DR. LO: Yes. I mean to pick up on the theme that we were discussing two breaks ago, I think that as with all our recommendations or guidelines, the more we can flesh it out the better. And for recommendation 7, I think we've already discussed a lot of elements that would need to go into a set of standard practices or best practices, I'm not sure which, and some of them I think include having a community advisory board, having set policies establishing some sort of written procedures for determining both the scientific validity of requests to use the archives and stuff, some mechanism for coding that meets certain standards. And so I think to the extent that we can at least start to point people in the direction and leave it to others to flesh it out, given that practice is evolving and people in the field who are really working on it probably can come up with better ideas in detail than we can, yet still to give some direction I think would be useful.

DR. SHAPIRO: And we do have at least a modest collection of what looked like best practices from our...we got that together and we've got those forms etc. and guidelines, various institutions which I think do a good job here or at least some of the better jobs I've seen. And we can certainly take guidance from that.

DR. LO: And then the other thing which was suggested just a couple of meetings ago was to really give NIH the sort of recommendation that they put some effort into having consensus conferences, training, model training programs, funding people to do research on how to do this better. I mean make it really come alive, put this in as part of center grants and stuff to facilitate further developments in the field.

MS. CHARO: I think in fact we may, depending on how much time we have and how much attention people want to give it, we may be able to use some of those models that we now have and ourselves pull out the elements that seem to be most essential that we would want to make sure are included in whatever becomes a standard model. Especially keeping in mind Allen Buchanan's analysis of this situation as one in which it's not genuinely informed consent that we're obtaining because it's not really possible to consent to that which you do not know. It's really about prior notification, in which case you want to alert people as much as possible to the variety of things that may happen so that it's a no-surprises scene. And some of the elements that might be pulled out could include such things as the distinction between research that plans to come back to you with its results and research that does not plan to but might yet do it because of something surprising, versus research in which it will never go back to you under any circumstances, no matter how useful it might have been; alerting people to the complexity of state law with regard to discrimination and the fact that in a transient society no matter what your current state laws are you may not know where you'll find yourself on the day when this information might possibly be used. And looking through these forms ourselves and maybe pulling these elements out might give this guidance even more robustness.

DR. COX: I certainly agree with what's been said so far. The ... one point, though, at least in terms of the wording of this. If it could be read that saying the only reason one has these complex recontact schemes is because people are just not being clever enough to avoid them. And there are certain types of research that it's incumbent upon the research to have recontact. So that just so we're sure that the wording isn't done in a way. And there are many types where it doesn't have to be done at all. So just so we're sure that the wording is such that we cover that and have the whole system be as simple as possible because otherwise the implication is that the only reason to ever do this recontacting is because one's just not being thoughtful. And I think that my second point is this one about prior notification. I think it falls under this contact of respect for individuals, so you tell them that these are the possibilities that are about to happen and then as we keep hearing, if it's nine percent of people that say yes, you told me, don't bother me anymore, just go ahead and use it, then that's important information. But it still doesn't obviate the need for the respect of telling people ahead of time by my view. But it's pretty cheap to tell them ahead of time; you don't have to go back and send them a million-dollar questionnaire to ask if it's okay later on.

DR. SHAPIRO: Any other comments? I think with this, as with others, some of the other recommendations and so on we dealt with today does need to be fleshed out in ways that Bernie and others have suggested. And I certainly intend to do so. I just wanted to be sure this was something you thought was important for us to address. Any other comments? On this? Well, what about recommendation 9? At the bottom of page 7. Eric, do you have anything with respect to 9?

DR. MESLIN: This is an issue I think that Bernie has already alluded to with respect to additional feasibility of doing any of these things. So as a guidance it's one thing to offer the idea about the termination for need to reconsent, etc. is important, but that's...that may be more of a judgment or a guidance or educational issue rather than a particular type of recommendation. So I'd be delighted to hear Commissioners' views as to whether that responsibility can be operationalized in any particular way rather than just say we recommend that it is their responsibility. If it's self-evident, we should probably say either why it is, which I think is easy to do.

DR. COX: But I have a comment on this. If we don't make clear and crystal clear what the sort of priorities and goals are, then the investigators are not going to know whether they should come back and worry about it. And Mary Claire said that several times. Even she, as sophisticated as she is, you know, was still trying to go back and forth about really what are the principles and what are the rules. So that the ... I think we should strive as much as possible not to be able to have anyone say that, that they don't understand what the rules are. Because that's a double-edged sword. Some people may not understand what the rules are, but I play basketball with my four-year old, and he doesn't understand the rules for a different reason because he wants to win. So the...and I think it's the same thing here.

DR. MIIKE: I think we have to be careful about these because they seem to be in dark opposition to 7 and 8, because 7 and 8 say let's re-obviate the need for recontact, and yet the following one says they should be monitoring when they should reconsent. So I think....

DR. COX: I think the issue, too, I think that's right, actually. I think the issue here is for us to clarify when the reconsent process gets initiated and how one thinks about that. And when that's not necessary. And 7, 8, and 9 are dealing with it in various different ways. I really think it's that issue and I think your point is well taken, that as just written down here. That's not clear and I just think we have to put these together. One more suggestion, that is, Mary Claire's example with the tomoxofin will be an extremely powerful example to put in a report, not only the way she handled but alternative ways that one could have handled it.

MS. CHARO: In some ways I'm sorry that Alex had to leave early because he once wrote an article about informed consent that held some lessons that I think are valuable here. If we go back to Mary Claire's protocol in which people have consented to research looking for markers but now in fact we're looking for the actual gene, it was a genuine interpretive question here about whether or not the consent they had given originally was sufficient. And in the article on informed consent, Alex walked through the different points of view from which one could answer the question. Whether a professional would say that this is in fact substantially different is one way of saying has this changed enough. The second way is a more subjective view from the point of view of the patient to hear the subject and ask, would the typical subject feel that this consent had covered this new scenario or would they feel that this was new. And in the law, we've got a very confused set of rules about which point of view ought to dominate, but I do think that in a research context it would be appropriate to conclude as a principle that the point of view that ought to predominate is that of the subject. You ask would a typical subject feel that they had in fact consented to this particular kind of research and use that as your benchmark, rather than the IRB's view or that of the professional. It's different in a clinical setting, but in a research setting where essentially the subjects are doing you the favor of being volunteers, one tends to be more considerate of their subjective viewpoints.

DR. LO: Could I follow up on Alta's comment which I think is very, very helpful. Is the implication then that we need to have a community advisory board and somehow that board is given the choice?

MS. CHARO: IRBs are the role of trying to figure out what a typical subject or potential subject would think, feel, or say. They do it all the time when they look at consent forms and ask, "Is this comprehensible and is this sufficiently informative?" They always do that with the subjective viewpoint of the potential subject in mind. It strikes me that good IRBs frequently ask for a consult when they feel like they're out of their depth, and this is certainly another example of a case where they might want to. But mostly I think it's to remind people there that the shoes they should be wearing when they say, "Is this really any different?" are not the shoes of a researcher or a trained medical professional. They are the shoes of somebody who is a layperson who is reacting to this, not from the point of view of technical matters but from the point of view of emotional matters. And that's often enough. Hopefully it's most of the time enough. But they do their best. The obligation is for them to do their best to figure out what the typical subject would want.

DR. LO: Again, if I can just take a second to try and flesh this out. Would we then, having said the IRBs, the proper locus, say the IRBs ought to have someone on their panel who's an expert in this kind of genetics research and adequate representative of community or representatives of the subjects of the study?

MS. CHARO: They already have a layperson on there by virtue of the statute.

DR. LO: Right. But we were concerned before about whether one layperson gets drowned by the 33 other professionals working in the institution.

MS. CHARO: A very valid concern, and yet another one for the list of things to worry about in a kind of generic look at the whole system and whether it has the capacity to carry the burdens that have been placed on it.

DR. LO: And again, are we asking IRBs to take on yet another tough issue?

DR. SHAPIRO: I think in the terms of that latter point, Bernie, the statistics, someone mentioned before that 10 IRBs have 48 percent of the protocols, obviously some very large workloads around. It seems to me that there are—putting the resource issue aside—a number of solutions to this. That is, that IRBs that are dealing with that many protocols probably have a lot, or regularly get protocols that have or are dealing with genetic issues or regularly get protocols that deal with human capacity issues in a sense. And therefore there's plenty of room to mobilize your IRBs in a somewhat more specialized way than we often do. Now that's a resource issue; I understand that part. That just leads you back to the resource issue, but it seems to me there's no lack of potential solutions if one feels this is all important enough to put some resources behind it. We have to think of how to do it and so on. And my view on this is if we think it's important enough, my test is do I want, well I suggest we spend some resources; that is, do one less NIH grant because we have to put the resources here or something of that nature, not institution by institution, but it comes out somewhere. You know, that's a test I use in determining myself. Am I willing, is this important enough that I'm willing to forego some other thing we could do with this money? I don't mean institution by institution but as a whole social policy. And if it doesn't pass that test, then it doesn't pass the test that it's important in my view. Except for once in a while, once in a great while. Okay. There are recommendations 10 and 11. I understand 11 better than I understand 10, frankly. But let's see. Maybe Eric can help us make sense of what we're talking about in recommendation 10. There are two parts of it, and I presume this has to do with, Eric, what happens when using this material for some new project. That's the context, I presume. I can't really quite figure it out.

MS. CHARO: You know, the way I'm reading 10, it actually is not terribly consistent with what proceeded, because of course these were not meant to be consistent with one another. They are truly alternative recommendations. The way I'm reading 10 I understand it this way: Prospective consents can be used only with regard to protocols that are going to involve unidentifiable samples. In other words, you can allow your sample to be put into the great pool of unidentifiable or not. Prospective consents can't be used for future research that involves identifiable samples because we are not going to waive the usual individualized consent requirements. We're not going to let prospective consent trump it. That's how I'm reading 10. Am I understanding it accurately? Can I say that I don't like it?


MS. CHARO: I don't like it. I think that prospective consents are not a bad way to go. It may be that we want to give some more thought to the idea of prospective consents being used not to eliminate the need for recontact, but to ameliorate it—allow for an opt-out as opposed to an opt-in consent in the future. For example, I contract with you to agree that an opt-out is sufficient. But I think the prospective consents are tremendously useful and in the long run, can make a research endeavor far more efficient. And I hate to hamstring them from the very beginning.

DR. MESLIN: I was going to say, at the risk of trying to move you one way or the other, one of the reasons why we presented that type of recommendation, somewhere after recommendation 8, which was the recommendation that says there are a number of nice consent forms and processes out there that we might want to recommend to people, is just to juxtapose them. We heard from the NHLBI. We heard from NAPBC. We heard from a number of groups whose general and/or tiered consent activities relate precisely to what you I think would like to see. So you might just simply want to reject the concept of recommendation 10, but I would encourage you to think about adopting what is in recommendation 8 if you want to reject that.

MS. CHARO: But this would be yielding to manipulation by the drafters of this memo.

DR. MESLIN: Well that's why I waited until 10 before.

DR. SHAPIRO: The issue of this consent is what consent is valid for future—not completely known at the time—procedures or uses of a material is not that different from what we were discussing just a few moments ago in the advance consent. It was a very similar kind of problem. And it's a generic problem and it's worth thinking about as to whether, just how narrow Bernie's example before was, that if we interpret this consent very narrowly, he has a lot of trouble. He may have trouble with it anyway, but certainly if it's interpreted very narrowly—let's say MRI versus image studies—it makes a big difference, probably, to how he would feel. He may not like either, but at least he'd feel differently about one versus the other. And really that's the same; it's in some sense the same issue here in different guise. Just how much can we say that advance still feels acceptable to us? So for example, in the case of Alta, or the proposition Alta just advanced, do we feel that some kind of advance consent which is general, together with an opt-out or something, is really informed consent or at least sufficiently close to informed consent that we would be happy to deal with? I think that's a very interesting issue. Larry?

DR. MIIKE: You know, I want to reintroduce a notion I talked about a while ago, which is that I don't have a problem with giving advance consent or use of my tissues and they know who I am so long as the information that they have is limited in time. I would be very offended if they never contacted me but they were delving into my medical record subsequent to the giving of the tissue. So I'd at least consider a notion of the advance consent about the timing, the time framing, which I understand what I'm consenting to. I can't consent to information that may be generated in the future about me of which I have no idea at the time that I'm giving consent. But I can certainly accept the notion that, if I went in for a cancer biopsy, anything they wanted to know about me in that episode, if they asked me and if I said yes, I would feel comfortable with it. But I would not want them to continually delve into my private life without my knowledge.

DR. SHAPIRO: Regarding future access to your medical records. Bernie?

DR. LO: There were others before; I was just trying to get in line here.

DR. COX: I really agree with that and I really support the idea of this prospective information. It's not really a prospective consent, but it's telling people sort of what the rules of the game are again, that we don't know exactly how it's going to come out but here are the things that are likely to take place. And that's a situation where I can really go for an opt-out, because if people say, "Listen, I'm one of these people who—basically I've signed up for research—that I'd like to know what you're doing but if you don't hear back from me, go ahead and do it." Now if somebody's already told me that, then that's how they feel comfortable doing business and that they will get information from me so that when I know more, they know more. But it's two different ways of doing it, either getting assent from them or getting them to opt out. And not only is it respect, but it actually has merit to it from the point of view of having better understanding of the people. It gives people options, real options. But my biggest concern is exactly what Larry brought up, because this isn't a carte blanche to go into somebody's medical record whenever somebody else wants to do it. So the original research that you get involved with has to be time limited and/or content limited in terms of medical record entries, which are often enough I believe.

DR. LO: I want to try and see if I can draw together some of the comments which I agree with and I think are very helpful. I like option 8 rather than option 10. And then looking at option 8, I would like us to encourage tiered consent and to discourage general consent—you can use it for whatever you want—because I think that's so vague as to not really be consent at all. And I just want to raise a couple points for discussion or clarification. One, tiered consent plus an opt-out sounds good, but we need to keep in mind that using the opt-out too often can be an invasion of privacy if every time they want to do a new study I get another postcard. I'm moving around the country trying to escape from my past and they keep tracking me down, I may not like that. And then I want to pick up on Larry's point. I think it's really important and I think there is a big difference, as he pointed out. But I want to test what the boundaries of that are. So on the one hand you use the sample but not only use the clinical information existing at the time of the sample, but 10 years later using publicly available information from death certificates, look at overall mortality and whether a given gene predicts that. So presumably that would be okay if it's publicly available information. Okay. How about going to a little harder case, cancer registries, and saying that I've given consent in 1998 and in 2000-whatever, 10 years from now, 2008, the investigator says, "Well, I'll go to the local Northern California cancer registry, which has all the cancers of interest in their registry, not going rifling through my entire medical record, but finding out something about me. Does that fall under your invasion? Would you be offended by that?

DR. MIIKE: If I'm dead, I wouldn't care.

DR. LO: But you wouldn't want a researcher to look in a cancer registry for a 10-year prediction of the . . .

DR. MIIKE: That's a hard case. But if they're trying to find me in this registry, then to me they're delving into my personal life, subsequent to the time I gave permission.

MALE VOICE: So you would have wanted them to have said up front, "We're planning . . ."

DR. MIIKE: But obviously there's another issue sitting over here. Do cancer registries, is that my personal information? I don't know.

DR. LO: Well, it's linked to you. It would have to be linked to you.

MS. CHARO: Mary Claire King's research, in which she talked about how ideally she would like to have had survival data, is exactly the kind of research that might require repeated trips to your abstracted medical records over a period of two decades. And she was doing that research with a single opt-out moment, in which people opted out at the very beginning but were not subsequently recontacted to say, "Well, it's been 5 years and every 5 years we always go and check and see if you're still alive. Do you mind if we go and check and see if you're still alive?" Do I understand you correctly saying that you would like her to have had . . .

DR. MIIKE: How is she gaining access to my medical records?

MS. CHARO: The medical records are being abstracted by somebody who's on somebody's payroll to do just this. They're told which medical records to pull.

DR. MIIKE: I know. But how are they pulling my medical records? How are they getting access?

MS. CHARO: Because you were initially in the study because you did not opt out.

DR. MIIKE: Yes. I understand that. But how are they getting my clinical records?

MS. CHARO: In the first place? How are they getting them? Well, you know, that's a practical problem that varies from research protocol to research protocol. If there is a registry, probably this information is being sent to the registry on a regular basis by all of your various attending physicians because they are all part of the process of collaborating with the registry and you've been collaborating actively because somewhere along the way you said it's okay for you to continue to collect data while you treat me and send abstracted versions to the registry.

DR. MIIKE: Well, that's a different situation where I was actually told that and I said yes. So I would feel comfortable with that. But I would have had the opportunity to say it's okay. It's just the situation where they're accessing information subsequent to my saying yes.

MS. CHARO: Like, for example, recontacting your attending, assuming that you're one of the rare people who has the same doctor for two decades.

DR. MIIKE: Yes. I wouldn't want my doctor to be providing information without their ever telling me that my personal doctor is providing information.

DR. COX: This is a recurring thing that happens right now. When researchers hook up with Kaiser or other situations and Kaiser says, "Sure. We've got a contract with you. You can look at our patients." When the researcher calls up and says, "You know, your doctor said it's okay to call you," not very many patients object, but it only takes one to shut the study down. And that patient then says, "What the hell's my doctor giving you my records for?" On the other hand, if you do this and you say, "I'm calling from Kaiser. I work for Kaiser and we're asking, is it okay to use your records in this study." And people say "Yes. It's fine." And then you go and you ask them questions. So you say, well that's a trivial point. But it's the difference between being able to conduct the research and not, because it just takes one person; that's all it takes. And these aren't theoretical things. These are real things that happen every day right now. And it shuts down—just 2 days ago I had an example of this in Hawaii. It shut down the study for 6 months.

DR. MIIKE: I wasn't the patient.

MS. BACKLAR: We wouldn't do it any differently in a research protocol with mentally ill people where I'm getting names from the information system and then I have to get somebody from the state to contact them first to see if I can contact them. So it's exactly the same situation.

MS. CHARO: It's a very commonly misunderstood. People think that review of medical records in order to determine who should be recruited is not subject to these requirements. They're wrong, but this is a frequent misunderstanding I've discovered. But, Larry, your situation's not about recruitment. Yours is somewhat different. I mean, basically, if I understand you correctly, you are advocating that a single moment of giving permission to be a research subject in some fashion, a single moment in time, is insufficient if being a research subject is going to involve multiple requests to your physician to go into your records and abstract them for the researcher.

DR. MIIKE: Let me put it at a gut level. I can distinguish my personal self apart from a piece of my tissue that was taken from me a long time ago. If they continue to look into my medical record on my living self, that's a little different from giving permission. But it's that level that I'm talking about.

DR. COX: Larry, can we ask you a question? What if somebody said, "Listen, what we want to do is we want to look over 20 years for this piece of information."

DR. MIIKE: Well, that's different. Then they've asked me that specific question.

DR. COX: I don't think it's actually the multiple times going in; it's up-front telling somebody what you have in mind of doing. So that's exactly what happens with a lot of these things. You get tapes. They become public tapes of these longitudinal studies, but it's clear what's going to go into those tapes up front. And you don't come back halfway through and say, "Oh, by the way, now I want to actually start looking at an additional X, Y, and Z." I think that's what you're objecting to, Larry, and that's what most people object to. But if up front you could basically say what you're going to be assaying or sampling from the chart, then I think most people . . .

MS. CHARO: Right. But they could have told me when I gave permission like this that we're going to spend the next 15 years going into your chart once a year, looking at your mammogram results, right? And they're doing that, and then along the way somebody suggests an association between breast cancer and abortions, or miscarriages, or years on the birth control pill. And so now, 7 years into this, what they're going to do is they're going to go in and they're going to pull that information. And that is distinctly different, and yet it's part of the same study when they suddenly are thinking, "You know, we might be finding an association here."

DR. COX: But that's not kosher because basically what you're doing is that you're doing something different in terms of the body of information than you set to begin with.

MS. CHARO: Right. But these prospective consents are written very generally. And the point is do we want to permit these to be written generally enough that people can say, "Yes. You can go in and you can use whatever you want over the long term." Or are we going to prohibit people from doing that because we think it's so far away from . . .

DR. SHAPIRO: The big difference here, whether we're dealing with identifiable or nonidentifiable cases, nonidentifiable are not a hard, nothing very hard for me. We don't have to spend any time right now. It seems that we're dealing with the identifiable cases. And my own view is there are limits to this prospective consent, and they're pretty stringent. And that is that you can't just—I mean I can't quite imagine—there's a certain amount of tiered consent you can do at various times I'm sure—and we'll look at what some of the people attempt to do—but there are limits there because if you're going to have the kind of permission that lets you go back into the medical record, that's a serious matter, a very serious matter. Go back without reconsenting. Let's see here—Bernie? Bette? Bernie's had some chances. Let's start with Bette.

MS. KRAMER: I'm confused. I had thought that the proposal was a prospective consent for continuing use of the sample, but that that did not include consent to continue to access the medical record.

DR. SHAPIRO: I think that's an open issue. That's a question of how we want to, how that original consent is structured. It could be structured either way in principle.

DR. COX: And that's what we're saying, Bette, but then there's a coda on that that if you tell people up front that you're going to access the record and for what, then it's a bounded thing in the beginning, and people get a chance to consent to it or not consent.

MS. KRAMER: Yes. But they're two very different cases. And I think we need to distinguish them.

DR. MIIKE: You know, Bette, I would say, "Yes" if they said, "We want to follow your case and the medical record for this specific issue." But if they say, "We want access to your medical record indefinitely to do research," I'd say no. I'd say you have to come back and tell me for what reason.

MS. KRAMER: That's the third distinguishing . . .

DR. SHAPIRO: Right. That's not consent in any way. It wouldn't qualify in my view. Bernie?

DR. LO: Realistically, I think what we're going to do is reinforce the tendency to develop large cohorts specifically for the purpose of doing research, so this is a take-off on Framingham—Heart-Lung, which is doing this. There are other large prospective studies because, in point of fact, the information I really want to get is not in your medical record. I want to come back and ask you questions about various genetic conditions, family history that probably I can't get at through the computer. And realistically, I think what we're going to be seeing is these ongoing studies where you agree to be contacted on a yearly basis and they ask you, "Is it okay to ask you these additional questions?" or to do this or that. Sort of the prospective, ongoing nature of the followup is built in. I think ethically, that's much more acceptable because there's that ongoing interaction that means you're not doing things that people didn't understand and didn't consent to. And frankly I think for the researchers, it's a better design because you get the chance to ask the questions you want to ask rather than having to hope that it's in the chart that was there for other purposes—but that's more expensive. And you've got to realize that. To put together these cohorts costs money.

MS. KRAMER: But that's like I said before. We do longitudinal studies like that where you go back every 8 months, every year to follow a subject over time. It's part of your research plan.

DR. SHAPIRO: Okay. There have been some helpful comments, but let's move on here to look at recommendation 11 here. Eric, do you have any comment on that?

DR. MESLIN: I think this is a bit more straightforward than 10. We're really speaking about prospective collection and the requirement to obtain an informed consent in order to use that sample. I may just say in passing that, having heard the conversation about essentially recommendations 7, 8, 9 and 10, one way of getting to Bernie's helpful question about tiered consents, is that we could summarize these concepts in one or two recommendations—one relating to what we are expecting in the prospective consent process, both in terms of the forms that may be developed and how specific we would expect those forms to still qualify as consent documents and how general they would be to still permit use of those samples without an opt-out. The Commissioners have mentioned that there are examples where individuals have indicated that they would be prepared to have their samples used, but the more uncertain you are about what is going to be done with them, the less that should count as prospective consent. And we haven't set that boundary.

Recommendation 11, which could have occurred earlier in the list because it's supposed to be a bit more straightforward, is simply meant to indicate that this is the best-case scenario, the prospective collection of information from a sample, use of a sample whether stored or collected, where the information can be linked back to the individual requires what we're calling full informed consent, and that's the informed consent we're all familiar with, a listing of risks and benefits and the like from the regulations.

MS. CHARO: I do have some difficulty with some of the language that's being used in this to express its ideas. I find myself uncomfortable with the parallelism between a sample that's linked to an individual being made parallel; that is, his identity is not concealed from the researcher. The identity probably is concealed from the researcher. It's probably coded. And so I think that that particular sentence actually offers the possibility of a fresh round of confusion. I'm also uncomfortable with the phrase "full informed consent." I found myself scribbling, "as opposed to partial." Now I can, however, understand that informed consent as opposed to a mere opt-out might be what you have in mind but would like to again suggest that these phrases might be confusing.

DR. MESLIN: Well, your point, Alta, is that it's largely editorial. It's not that there's a conceptual confusion. The language is saying "full," which I believe we understood to be....

MS. CHARO: Well to be honest, I wasn't sure what this meant until you just explained it because the language was so confusing to me I really wasn't sure what the agenda was behind that particular recommendation. I read it four times and then just left it in red.

DR. SHAPIRO: Are there any other comments beside with respect to what is marked here "Recommendation 11"? The other ones that are left here are really—I don't know if they require any discussion right now. We'll have to wait. They're really just . . .

DR. LO: Before we move on, recommendation 11 seems to restate what's the current situation, that if it's identifiable, right, not linked, you need informed consent. Conversely, I guess, do we want someplace to affirm that, if you have a sample that's already been collected in the past and you're going to be using it . . . I guess I'm asking a question. For all these stored pathology samples that are currently in hospital pathology departments that now are linked to somebody's medical record number, are we going to allow those samples to be used if they're given to the researcher in an unlinked fashion? If the pathologist just says, "I'm going to close my eyes and take 150 samples of Cancer X and I'm not going to know whose they are and send them off," we're going to allow that. Bette says no pathologist would ever do that because they're such compulsive recordkeepers. But if they were, are we going to . . . The question keeps coming to me. We heard a lot earlier about this is a valuable resource, there's a lot of good research that is done on that, we shouldn't sort of throw that away. I want to try and be as clear as we can about what are we going to allow to be done with those archived samples?

MS. CHARO: If they're given without identifiers or if, as Mary Claire King did, the links in those codes are severed irrevocably, it's my understanding that we were comfortable with the current rule which says they can be used with impunity.

DR. LO: Okay. So you were saying it's not how they're stored in the archive, it's how they're given to the investigator.

MS. CHARO: It's whether or not the individual is identifiable. It's neither how they're stored nor how they're given. It's whether, in the end, the individual is identifiable. That's the real question, is whether or not the donor of the tissue could ever be identified. And a severing of the link is one way to make that individual unidentifiable.

DR. LO: Okay. For Bette's pathologist, to make this work, the pathologist at some point has to delete the code.

MS. CHARO: That's right. That's another way of making them unidentifiable. By the way, since Bernie made us go back to 11, let me just say that it's actually written a little over-broadly the way it is, because it does not include the exemptions and exceptions that currently exist. Not the exemptions, sorry, the exceptions that currently exist that permit a waiver of consent.

DR. SHAPIRO: Any other comments on this? With respect to the remaining ones, I'm . . . It doesn't seem to me that we need any real discussion right now. These are kind of—I hate to use this word—but they're little sort-of "mop-up" suggestions here, which we will get to, but there's more important issues to deal with. It's not that they're unimportant. So I'm just open for any observations, suggestions you might give.

MS. KRAMER: Harold, a question. Back up a minute to the last issue. Wasn't that one of the cases that Dr. Hook talked about at the last meeting, where the cohort that was given—I'm trying to remember now—where the cohort that was given was small enough . . . I guess it would depend on whether or not the repository or the pathologist in question was willing to destroy any record of the samples that they were given. Do we know practically whether or not that would happen? Alta, is there any legal liability on the part of a repository to keep a record?

MS. CHARO: Well, I don't know the answer to that. But I can tell you as a lawyer that if they hadn't kept a record and there was a reason why it would have been helpful and it turned out somebody was harmed, I'd make a good argument that they should have kept it.

MS. KRAMER: Yes. I'm just thinking about that. I wonder if any pathologist or repository, in fact, is going to destroy a record of whose samples.

MS. CHARO: I think more realistically what will happen is what Mary Claire King described. The repository has the samples listed under a set of codes. There is Code A, B, C, D, and E. And they will take those and they will give them a new code, 1 through 26, and they keep a little record that says, "A equals 1, B equals 2," and they give 26 samples to the researcher that are labeled 1 through 26 and the researcher uses them. Would this repository ever destroy the record that says, "We gave away samples A through Z"? Probably not. Might they destroy the record that says, "A equals 1, B equals 2"? Yes. That's what she did. And by doing that, they know as a group what was given out but they cannot do a one-to-one correspondence between a specific sample and the results the researcher found and the particular donor of that sample. That they would do, and they'll do with some hesitation because of the possibility there would be some day some valuable reason to link back to the donor. But they will do it.

DR. SHAPIRO: And if they don't, it just means another set of protections gets initiated. It doesn't mean you don't go ahead. It doesn't mean anything of that kind. It just means another set of protections comes in. They're now identifiable and you go down that road, that's all.

DR. COX: But you see, one of the things that we haven't stated here yet that certainly isn't in the common lexicon of most people is the possibility that things are identifiable. But, it's that when it goes to the IRB, you have approaches in place that say that that information won't go to the researcher. And what that does is lower risk than if you don't have those things in place and you consider it differently. So I think, just because it's identifiable, the research community does now, it's identifiable. Then everyone has that stroke. So it could be under different . . .

DR. SHAPIRO: Yet another harm.

DR. COX: Yes. Exactly. Stroking our researchers. But I think that . . . So then they go to these extraordinary efforts to destroy all the links. Well, I don't think we necessarily have to do that.

DR. EISEMAN: I'd like to address that point that Bette asked because it was a point that Tom Murray asked me to address after the last meeting. And I did do a bit of research, not comprehensive research, but did talk to a few repositories about their practices for dealing with unlinked and coded samples. And I think David Cox yesterday made the very good point that it does appear that the vast majority of samples are used in a coded manner, not in an unlinked manner. And that is actually more rare than I even expected it to be. A lot of repositories don't destroy that link and they know who the samples come from, but they do also have a . . . some of them also do have a statute that, when they send out samples to researchers, the researcher can't come back and get any more information. So even though the repository knows that Sample A is John Smith, when they send it out to the researcher, the researcher can't come back and get any more information than was originally sent. So whether that's considered coded or linked is one question. The other question is about unlinked samples—that in most cases there is a list kept of who those samples came from. So if it's 50 people, there is a list of those 50 people that the sample came from for a number of reasons. The main one that was cited was quality control. They need to know who those samples came from because, if the researcher came back and said, "I need 50 additional samples completely different than what you first sent me," if they don't know who they came from, then that repository can't function. So those are a few points I wanted to make.

MS. KRAMER: Then what does that do with the definition that we accepted, that if, no matter how far-fetched it is, if there's anybody who can link, anybody who can identify who the samples came from, then we're calling that identifiable?

MS. CHARO: It is. Because it can be done and a circumstance will arise where somebody wants to take advantage of that. You do the breast cancer studies in one of these settings where you've maintained the links but everybody understands we're never going to let the information go back to the donors and we don't care who they are. And suddenly you discover that people that have the breast cancer gene seem to be at high risk of ovarian cancer, or the next thing it's going to be is that they're at high risk of developing a third eye, whatever it's going to be that seems to be linked. And suddenly somebody says, "You know, it's absolutely imperative that we send this information back up to the donors because if they get it now, they might be able to get some kind of preventive strategy going." That is why you have to understand that these are identifiable, because as soon as that temptation arises, you're going to have a discussion about whether or not the information is sufficient, is the cure worse than the disease, is the information too ambiguous? That's why you have to understand it being identifiable.

DR. COX: But, Bette, let's go one step further from what you're saying. What are the implications of it? This comes in my view what Harold was saying earlier about needs to go before an IRB. By basically saying these are identifiable, it says that they go before an IRB.

That doesn't say what kind of hoops people have to go through, so that they go before the IRB but the IRB says, "You say you're not going to ever give information back to the researcher? Sign in blood. Sign right here." And then when the researcher comes back and says, "Well, but I really have to go back," you say, "You see your blood right here. You can't do it."

DR. LO: I understand the rationale for keeping a list of who the 50 samples were so you can send them a fresh batch. What's the rationale for keeping the code that A equals John Smith if David Cox has made people sign in blood they're never going to backtrack? Why have it? What's the rationale for keeping that?

MS. EISMAN: That's a good question, and I didn't get that in the answer when I talked to the people at the repository. The answer was that that's how they do it. I'd be happy . . .

MS. CHARO: It is so that people can go back or so that they can do subsets. For example, you sent me 50 and I looked at them and there were 10 there that have a characteristic that's interesting to me and I want you to send me 10 more of the same. Well, they've got to know which those 10 are. You know them as numbers 17, 29, 33, etc., and they've got to know which those are so that they can go back to the blocks and figure out where they've got to get another section.

DR. LO: Right. But then that violates the signature in blood they gave David Cox because the agreement was I was never going to go back about the subjects.

MS. CHARO: And indeed, David, you and I are talking about different kinds of risks. You're talking about the risk of invading the medical record or the tissue for more information. I'm talking about the risk, quite the converse, of sending information back down to the donor, which are two very distinct kinds of risks.

DR. COX: In both cases you're going back. Bernie's absolutely right that they're different risks. And that's what we need to do, is talk about what these different risks are in different scenarios. Not to try to make them go away by taking something that's identifiable and pretending it's not.

MS. CHARO: This is where, to coin the horrible phrase, how it's used by the researcher begins to get valuable, get very important, because what I want them to sign in blood is that I have absolutely no intention of sending information back down the line to the donor. Now we all know that situations can arise that cannot be foreseen now. By definition, unforeseeable things can't be foreseen. So, hey, there might come a circumstance. But I have no current intention of going back down the line. And so long as that link is maintained, that possibility is preserved. Now if you put into place some kind of system on which we have a filter so that when the investigator says, "I think I've found something that requires me to go back down the line and tell the donor," there's another body of people who say, "Well let's talk about this all together." Whether it's the laboratory's own group or it's the IRB or it's some neutral body, that acts as a check on that particular kind of risk being realized, and is exactly what may allow something to be a minimal-risk procedure even if it involves looking at stuff that is potentially stigmatizing or embarrassing, because you've put into place a series of protections against that information ever getting out.

DR. SHAPIRO: I think it's really quite important to keep on reminding ourselves that, just because it's identifiable, doesn't mean a ton of bricks comes down on the study. That's not what it means. It just means now there are other things to think about and you have to start thinking about them. They may be very small, in which case they'll be dealt with in a very simple way. They may be very difficult, in which case you will have to go through them.

DR. COX: But when you have the research community being involved with thinking about these things, that's going to be very useful because no one has more at stake than they do, at least from the point of view of carrying on business. Patients have a lot at stake. But what you'd like is for the research community to have something at stake here, too.

DR. SHAPIRO: Any other comments? Are there any comments, since we are going to adjourn in a few minutes, are there any . . .

DR. MIIKE: Yes. What does recommendation 11 mean? Do they have to get informed consent or not?

MS. CHARO: Unless they meet the conditions for a waiver.

DR. MIIKE: But for that example, where a repository is the only one that has the code, that the researcher doesn't, we're calling that identifiable and we still have informed consent?

MS. CHARO: Unless they meet the conditions for a waiver.

DR. LO: So if it's not greater than minimal risk and it won't adversely affect the welfare of the subject, you don't have to get consent.

MS. CHARO: And it's not practical to get consent.

DR. MIIKE: Okay.

DR. SHAPIRO: Then we agree that this is one or something or at least further up the ladder. Okay. Before we adjourn, is there anything you'd like to say now? We will have to come back to 12, 13, and 14. They have their own importance. I don't want to minimize them, but I don't know if we can productively say much about them today. Let's see what comments there are so it would be helpful. Bernie?

DR. LO: Yes. I'd just like to say this was really a very productive meeting. I want to thank you, Harold and Jim, for sort of leading us through this. And to Eric and the staff for really sort of allowing this to happen. I think we've made a lot of good progress, had a really good discussion. I think we're getting closer to our final product.

DR. SHAPIRO: Any other comments? Well, thank you very much. It's been a long day. We almost began at 7:30; it was actually quarter to eight. It's now 4:30 almost, 4:25. So thank you all very much and let me extend my thanks also to staff for mobilizing this meeting and, Trish, thank you for encouraging us to meet here. It's very nice to be out here in Portland. We're adjourned.