Four scholars spoke to GeneWatch in three separate interviews about research biobanks. George Annas, JD, MPH is Chair, Health Law, Bioethics & Human Rights at Boston University School of Public Health; Robert Green, MD, MPH is Associate Director of the Harvard- Partners Center for Personalized Genetic Medicine and a geneticist at Brigham and Women's Hospital;Patricia Roche, JD, MEd is Associate Professor, Health Law, Bioethics & Human Rights at Boston University School of Public Health; and Susan Wolf, JD, is McKnight Presidential Professor of Law, Medicine & Public Policy and the Faegre & Benson Professor of Law at the University of Minnesota.
Susan Wolf: Part of the challenge is that biobanks come in all shapes and sizes. There is a tremendous variety.
Robert Green: There are thousands of different entities that you might call a biobank. So it may be helpful to consider whether we are talking about blood banks, tissue banks, clinic or research banks, tumor banks or actual DNA banks.
GeneWatch: Where do all of these samples come from?
Patricia Roche: A lot of projects have projected numbers, so in order to answer the questions they want to answer, they need a sizeable group or the power of their results is not going to be there. Getting that number of samples is a challenge. If you stop me on the street and say, "I want you to give me a biological sample," I'm just going to say, "Go away." You need the opportunity to engage with someone to get the sample, or you have to get the samples from someone who has already collected them.
RG: Every time a hospital pathology department takes a surgical sample in the course of a patient's care, they maintain that sample somewhere in the pathology department of the hospital. That's been going on for decades in the name of good clinical care. If pathology decides you have a certain kind of tumor, and you get treated on the basis of that decision, it's important to be able to come back years later and look at those slides. Holding onto stored tissue in hospitals is considered standard of care, at least in certain circumstances, and there is a good reason for it.
George Annas: Last year researchers collected samples at the Minnesota State Fair from kids. The alternative was to collect them in a hospital, and they thought that might not be representative, and they wanted to study healthy kids. They plan to follow the kids at the fair each year; and we'll have to wait to see whether this strategy is successful. Of course, when the children turn 18 they should have the right to either continue or have their samples and information destroyed.
RG: You might look at drug studies, too. Every drug that's on the market had to go through clinical trials, and every clinical trial had to take samples for blood levels of the blood, monitoring for ill effects. I don't know how long drug companies keep those samples, but they certainly store them while the study is ongoing, and they may analyze the genetics of those samples for pharmacogenetic features of the new drug as part of their study analysis.
GW: Once a sample is being stored, how long might it be kept there?
PR: One thing that I think is overlooked is any kind of a standard for destruction of samples. It seems to me that rules for destruction of samples are often not built in. Sometimes this might be done intentionally, particularly in studies where they are building fishing expeditions to look for diseases; they don't know how long it might take them to find something. But there are other instances where you could identify an end point at which the project will be completed, and without the endpoints being built in, what happens to a particular collection when whoever created it is done with it? It's up for grabs.
GA: There are no good rules on transferring samples if, say, the collection's owner goes to bankruptcy court. It's a big deal. People might be happy to give their DNA to Joe Shmoe's DNA bank for research on prostate cancer; but not to let it be sold to the highest bidder in a liquidation sale.
PR: And what happens when he's done? Who's responsibility is it to make sure that either the samples are safeguarded and preserved somewhere, or that they are destroyed? It's up in the air.
GA: As a general rule, scientists don't like to throw out anything. And it's hard to blame them; you really don't know what the next research project will require.
PR: Once I was using a consent form for a study as an example in my class. I asked the class what their understanding was after reading that form, and their understanding was that if they gave a sample for this study, after a period of time the sample would be destroyed. And I asked, "What do you think 'destroy' means?" There happened to be someone in the class who was actually involved in the study, and she said, "Oh, I suppose you'd think that would mean that we would flush it down the toilet or something, but we don't do that. We just put it on the shelf." I asked why they wouldn't destroy it, and she said, "Because we think maybe the person would change their mind, and we would have lost it. If we already have it, we don't have to get another one from them. ... But now that I'm thinking about it, that's not really what we tell people, is it?" It's that mindset-"We don't want to destroy anything because it might be useful, we might need it, something else might happen."
GW: Are there any laws governing ownership of samples in biobanks or whether ownership can be transferred?
GA: Not really. There's not much doubt that you "own" your own DNA sample in the sense you can give it as a gift to a biobank; but just what other things you can (or can't) do with it are not as certain. There also aren't any gene bank transfer laws. You have to argue the legal status of DNA by analogy, and the analogies aren't very good here. Genetic information is unique, although most closely analogous to medical information; and the DNA sample itself could also be viewed as a medical information repository. And, of course, we will be able to derive more information from it in the future.
PR: As we go forward and as we learn more and more about the human genome, that sample is going to get more valuable. If I gave you a sample for a particular purpose, do I really have any idea what I gave you?
GW: Is there liability for whoever collects or stores the samples to protect participants' privacy?
GA: DNA bankers and researchers need to take reasonable steps to protect privacy. To protect, not to ensure; nobody's guaranteeing that.
PR: George Church's idea is that we need to get over treating DNA samples and genetic information like this is such sacred or harmful stuff. He theorizes that if volunteers in genomic studies provided unrestricted access to identifiable data from their DNA and medical records, nothing bad or harmful would happen to them. Consequently, the public would be less hesitant to take part in genetic research and researchers would have no need to make unrealistic guarantees that privacy will be maintained.
GA: It would be unreasonable to ask for a total guarantee of privacy. If you need a guarantee, don't give your sample.
GW: What do we mean when we refer to "de-identified" samples? Is there really such a thing?
RG: "De-identified" means the sample is given a code, and the link between the code and the actual names and identifying features of the subject are kept in a locked drawer or a computer file somewhere that's considered confident.
GA: "De-identify" is an inherently ambiguous term that could permit, for example, a trusted intermediary to keep sole possession of a code used to identify the samples. A much better term is "unlinkable"-even though you could argue that nothing's unlinkable, because someday we're going to be able to link everything.
SW: De-identification has for a long time been of concern to researchers; but the technology for re-identifying people has advanced. That has really, I think, led to the recognition that we probably shouldn't use words like "anonymous" or "anonymize" anymore, because with a reference sample you really can reidentify it. People more commonly talk about degrees of de-identification. I think that has made us cautious about the potential for re-identification and modest about what we accomplish when we de-identify.
PR: If I have this little bit of someone's sequence but I don't know the person's identity, the theory is that I could go into some public database, and when it hits on a match, I can figure out who it is. But I think all that would tell you is that whoever's sample it is that you started out with, they are also in this database. Depending upon what else I know about how folks were selected for inclusion in the database (e.g., only individuals with a history of heart disease) or how large the pool is, I might draw some conclusions about the source of my sample. But whether I can also readily figure out who they are is less clear.
GA: So, can you identify someone from an "unlinkable" or "anonymized" sample? Theoretically, but the odds are slim.
GW: If researchers need to communicate with participants later, who does the re-identification?
SW: The most common answer to that, in the past, has been that it should be the collecting site and the original researchers; but there is the question of whether the biobank or perhaps a trusted intermediary should hold the codes, so that the full burden of dealing with potential incidental findings from research results doesn't fall on the collecting research site-where the original researcher may be dead or retired or have run out of funding.
Incidental findings and returning results
RG: Say you've given me your sample for some kind of genetic testing, and I find- either by accident, or because I happen to be studying breast cancer-that you have a very prominent risk of breast cancer. Do I have a moral responsibility to tell you? Do you even want to know? Could I even contact you if you did want to know? And is it my responsibility as a researcher, on a research budget, to somehow hire the staff to call you up and contact you? At the present time, this is a huge ethical mess.
SW: This whole question of incidental results and return of findings has really exploded and become a very pressing question in the design of research biobanks-both prospectively and also for established biobanks, facing the question of what to do with the incidental findings and potentially actionable results they've already got in hand. There are people who argue that biobanks shouldn't be in the business of returning incidental findings and research results at all; that these are important, crucial resources for the conduct of genetic and genomic research, particularly large scale research; that they have limited resources, they are not clinical care entities, and their resources shouldn't be used for the purposes of clinical care. Furthermore, many biobanks are constructed based on the receipt of de-identified samples and data, so they really are not very well set up to return information to participants. So there are a lot of practical barriers, resource barriers, and, some people would argue, ethical barriers to returning incidental findings at the biobank level The pushback is that we have growing data to suggest that participants are interested and want clinically important and perhaps reproductively important information. In Europe, in particular, there has been a lot of movement toward suggesting that biobanks and researchers owe participants clinically important information as an expression of reciprocity, solidarity, and respect.
GW: Are there very many research biobanks currently that return findings to participants?
SW: We've discovered that there are studies and biobanks that do, but I do not think that is the dominant approach right now. I think there is a great deal of concern about the cost of doing this and whether biobanks are set up to do it. There is not widespread consensus yet on exactly what incidental results and research findings are appropriate for return. I think the question of how to handle incidental findings and return of results is still very much up for grabs.
GW: How might the determination be made whether or not a particular result or incidental finding should be reported to the individual who provided the sample?
SW: I think there is a trend in the ethics literature toward recognizing a duty to consider some results for return; for example, where there is high clinical importance and actionability. "Actionability" has become a term of art, and is defined somewhat variously; but the basic idea is that if you return these results, does it have the potential to change the clinical course for this participant, either by opening up treatment options or by increasing surveillance to look for disease, for example the development of cancer. I think that there is a developing consensus that high clinical importance plus actionability suggests considering return.
PR: When it comes to that issue of whether to give results back to participants, the first thing researchers have to ask themselves is: what's the reliability of the results to begin with? A lot of research is so preliminary that they really don't know whether they will find something that will have any utility. And if they were going to give the information back to people, they need to know how confident they are about the results-if not in terms of the clinical validity, at least in terms of analytical validity. I don't have a problem with somebody saying upfront, "We may find something, but until we do subsequent research and confirm either the test we're coming up with is a valid way of doing this or that the results have significance, we're not going to give you back information."
GW: Committing to return results seems to open up another ethical can of worms.
PR: To say to people, "We're not going to give you back any information, is that OK?" is one thing; it's another thing to know that the way you're designing it, you're likely to find something that might be significant to people, and not say to them upfront, "By the way, we're likely to find something that would be beneficial to you, but we're not going to tell you-is that OK?"
GA: This really goes to the question of what the person giving the DNA sample thinks the deal is. Is that person simply giving a "gift" to science with no expectation of any individual knowledge or information coming back to him or her, or is the person making a healthcare agreement with the DNA bank in the expectation that any information found about my particular genome will be shared with the donor? This makes all the difference, because if it is the latter expectation, and if you don't contact me, am I supposed to think I'm OK? Should I think, "They didn't contact me, my genome must be clean"?
SW: I should introduce the caveat that almost nobody talks about imposing this information on research participants. So "return of results" is really a little bit of a misnomer; it's really offering back results and incidental findings to participants who are interested in receiving them.
RG: Say I perform a research study and I say to you explicitly that I am going to take and store your biological sample, and I would like your permission to give it to any researcher I want. It will stay de-identified-meaning it's theoretically possible to hook up to your identity, but we'll make every effort to keep it confidential- and you will never learn anything back from me. Is that enough? The alternative is almost unthinkable to many people who actually run biobanks, because it starts to turn the researcher into a population screening machine and to conflate the role of researcher with the role of the clinician.
SW: Researchers conceive of themselves as researchers, not clinicians. So the idea that researchers owe some kind of duty to participants to find information of clinical importance, and then to offer that back to participants, kind of exists in tension with the research mission. It takes money, it takes resources, and it really is not what research is set up to do, at its core. It challenges the traditional line that we've drawn between research duties and clinical care duties. Clinical care duties are very robust and broad; research duties have really been conceived as primarily pursuing generalizable knowledge, rather than caring for individual research participants and returning information to them of their own particular clinical care or reproductive decisions.
GW: Research biobanks make a lot of very important research possible. How much is really at stake in how an individual's sample is handled?
PR: I think the worst that could happen is for people to give away samples without knowing what they are signing up for. Firstly, they may be exposed to a risk that they wouldn't accept; but I also think that it's insulting and disrespectful not to inform participants. If I'm going to ask you for something, and you're not going to get any benefit from it, and I need it to do something that I think is going to be important, then it seems to me the very least that I can do is be honest with you, to tell you what it is I want to do.
GA: There's no excuse for a researcher not to tell you what they're doing. And not just for the DNA donor's sake. Anthropologists used to warn their colleagues: "You have to be nice to the group you're studying, for the next anthropologist who comes along. Don't irritate (or exploit) them too badly or nobody else will be able to study them again." You don't want to get the public to view researchers as suspect creatures who believe that participating in their research is a matter of civic duty.