Editors: Norio Fujiki, M.D. & Darryl R.J. Macer, Ph.D.
Robert Mullan Cook-Deegan
Vice-Director, Institute of Medicine, National Academy of Sciences, Washington D.C., USA
Heightened attention to ethical issues in pedigree research can be traced directly to the Ethical Legal and Social Issues (ELSI) program of the National Institutes of Health. A grant to the American Association for the Advancement of Science (AAAS) was among the first funded in the ELSI program. This grant supported a series of three conferences. The second such conference was held in Charleston, South Carolina in March 1992, and will soon result in a report which raises the welter of complex ethical issues confronted when studying pedigrees (1). The National Center for Human Genome Research convened a follow-up workshop in October 1992 that built on the Charleston conference. Joan Porter from the NIH Office for Protection from Research Risks (OPRR) attended both meetings, and OPRR prepared a section of the Institutional Review Board Guidebook (2). These two publications, and another article on ethical issues in publishing pedigree data by Madison Powers of the Kennedy Institute of Ethics commissioned for the NIH workshop (3), are perhaps the best places to start for a review of the issues, beginning the slow of changing research practice. The OPRR guidebook, for example, has been distributed to Institutional Review Board directors at all major US research institutions, and is already in heavy use [personal communication, Gary B. Ellis, OPRR, 29 October 1993].
Pedigree research has been taking place for many decades, but the ethical issues in conducting such studies have been relatively neglected until quite recently. The rising importance of pedigree research had not, moreover, led to much public discussion of its legal and ethical issues. Different research groups were pursuing different policies and local Institutional Review Boards were, for the most part, simply not considering the potential for psychosocial harms inherent in such work. These initial publications are sure to be followed by others, greatly strengthened by other ELSI grants to study empirically the families involved in pedigree studies and genetic testing programs.
The issues distinctive to pedigree research arise from the fact that investigators must study a family rather than individuals, and the concomitant fact that information about one person yields information relevant to others in the family. This adds complexity to the ethical issues in research, because our conceptions of informed consent and clinical relationships are generally framed as dyadic doctor-patient or patient-professional terms, and cannot be directly applied. It is not clear what informed consent for a family means, or who can give it. Moreover, the disclosure of information is more complex in families, because information about an individual can be relevant to a child, or a sibling, a cousin, or even a parent. At some stages of genetic linkage analysis, for example, interpreting information about one person in the pedigree may require analysis of DNA from other family members. One person's desire for knowledge may be held hostage to another's desire for privacy, and vice versa.
Every family has stories that unite it. In families through which some genetic condition has cut a swath, the disease is often a prominent theme in the family lore. The disease is part of the family. Much of this paper is drawn on experience early in my research career. I worked in a laboratory that studied metachromatic leukodystrophy, retinitis pigmentosa and other neurological disorders with a genetic variant. Most of my background came from studying familial Alzheimer's disease, including two large families that are part of the so-called Volga German group whose genetic linkage has not yet been established, and is apparently either multi-allelic or arises from genes other than the Alzheimer's linkages established for regions on chromosomes 21, 19, and 14 in other families. I was far in the background on the other disorders, but for the Alzheimer's pedigrees, as a medical student I was the person most directly responsible for assembling the pedigree information, gathering blood and other samples, and educating the families about our research. This entailed very close contact over a period of three or four years, continuing into internship and residency, with repeated trips through the American West. The observations in the second part of this paper will draw heavily on experiences from this phase of my career.
The complexities of pedigree research start even as a study begins, and often before individuals agree to take part. The nature of pedigree studies is that information about one family member leads to questions about relatives who may also have the health problem in question. The initial proband is used to find others, in contact-tracing analogous to public health work for infectious diseases. Secondary contacts are linked to tertiary ones, and so on. Either the family member or investigators may then contact those others for further information. This can be a delicate matter. In one recent study, for example, women whose relatives were part of a study of breast cancer in the 1940s and 1950s are being contacted. The original study was a landmark in cancer epidemiology that called attention to the familial transmission of some cases of breast cancer. In going back to these same families decades later, investigators found that about forty percent of the people contacted did not know about family history of breast cancer. This raises several issues. First, just contacting the families can be a major intervention, introducing potent information with possibly strong psychological impact. Second, without due care, informing family members of possible genetic risk, even if they do not choose to participate in the subsequent study, can put them in jeopardy. The next time they fill out an insurance form that asks if there is a family history of cancer, they may have to either lie or disclose that they now know of a such a possible risk, inviting all sorts of further inquiries about the nature of that history. This can be avoided by including control groups NOT at genetic risk in the initial recruitment (so they are not lying if they say they do not know of family history of cancer), but investigators have to identify this issue before the study starts. Third, we know almost nothing about what families do with this information, or its impact on different groups. This is an important area for empirical study that is just beginning, borrowing methods from anthropology and sociology as well as clinical research. These issues of recruitment will be just as important for neurological studies as in cancer, because many devastating late-onset disorders have not been traditionally perceived as genetic, and so investigators pursuing a genetic line of inquiry can indeed be introducing new knowledge about genetic risk to family members, with all its potential for psychosocial harm.
Some questions that arise in pedigree research cannot be avoided by forethought. The issues surrounding insurance and genetics are often reduced to sharing test results with insurance companies and other third parties. They surely include this, but are much broader. Many research groups have included a warning that insurance may be at risk in the research as part of the informed consent process. This may be honest, but it may not be fair. It may merely shift the onus to the potential participant to make a tragic choice. Prospective participants must then choose between research that may help to understand the nature of their families' medical plight for future generations - a task that many find quite meaningful - or risking loss of insurance. This can happen even without genetic testing, as noted above, by mere participation and labelling as part of a genetic study. Some genetic investigators have set about to maintain laboratory and clinical records entirely separate from their associated hospital, or have come up with coding terminology that does not indicate participation in a genetic study. Many believe that the problem would go away if the United States shifted to a more fair health care system. The problem would diminish, but it would not go away. The problem arises not only in health insurance, but also life insurance, disability insurance, auto insurance, and other areas.
Just imagine the problems if it became possible to detect genetic predisposition to Alzheimer's disease in a substantial fraction of cases, for example. This alone would completely wipe out the possibility of a private market for long-term care insurance. More than half those likely to use benefits heavily will suffer from Alzheimer's. If individuals could get tested, they would surely get insurance if testing positive. Insurers would therefore also have to know in order to adjust premiums and keep from going bankrupt. This would escalate cost, making it more likely that only those at risk would buy coverage. In the end, this is not insurance, but prepayment.
Once contacted, family members can agree to participate or not. But the nature of participation is not binary "in or out," but admits of many intermediate points. The proband case is generally similar to other research, but secondary contacts are not necessarily so. Even the proband cases can be complex. In the largest Alzheimer family that we studied, for example, the index case was mute and completely demented at first contact. Informed consent was obtained from the spouse. Since the person in question was in the VA system (and thus not at risk of losing health benefits due to a "pre-existing condition") and the physical risk was minimal, this seemed a straight-forward case. Even here, however, there were tough calls. We decided to establish a cell line at the Camden repository, for example, which entailed obtaining a skin biopsy from the proband. The person was utterly unable to communicate or make decisions, but when I as a sophomore medical student did the biopsy, the patient clearly found it painful. I, belatedly, agonized over whether the marginal benefit to future science justified the immediate pain to this person under my care, and about whose life I can come to learn a great deal.
I would rather have been anywhere else than in that hospital room causing that pain without having thought in advance about whether it was justified. The procedure did indeed have very low medical risk, in the sense of health risks, but it clearly caused acute suffering that the patient could not understand, and to which he could not assent. His spouse had given proxy consent, but significantly (and understandably) declined to be there when I did the biopsy.
Similar questions of proxy consent will often be present in the study of neurological and psychiatric conditions, or diseases of childhood. This is especially important in conditions that can influence how families and schools handle children. Many neurological and psychiatric conditions are associated with learning disabilities or emotional problems that may be flagged by teachers or parents. Many studies document the potency of labelling, and the opportunities for self-fulfilling prophesies once children are identifying as having cognitive or emotional problems. A recent proposal to do screening for fragile X syndrome in the State of Colorado has raised this issue once again, and has not been fully thrashed out. It is telling that one of the main justifications for such screening is a cost-effectiveness analysis that tallies the State's cost of caring for retarded persons. This implies that the screening program, to be judged effective on grounds of cost-effectiveness, will have to prevent such children from being born or will have to otherwise remove care and special education costs from the public roles. Skepticism of such motives is understandable, as this is similar to the public health reasoning that lead to killing of the mentally infirm in Nazi Germany, the prelude to the Holocaust. The Colorado program does not propose killing those born with fragile X, but it may raise the question of prenatal diagnosis and abortion in a public health framework, and in some circles these arguments are morally similar to their German predecessors. It is not clear how early identification of fragile X children will benefit those children. While shrouded in the garb of economic argument, one can surmise that we have not heard the last word about the public policy content of the Colorado program.
Beyond the proband, informed consent becomes even more complex. Who should contact other relatives? Using a "family facilitator" seems ethically cleaner, in that the investigator cannot be blamed for intruding on privacy. But this approach also relinquishes control of the accuracy of information, and can unleash coercive forces within the family. Family members can be manipulated by other family members, and the family facilitators can easily use their information and status as a source of power. Even without such issues of power, there is ample opportunity for information warp. Unless the family facilitator fully understands the nature of the risk and the techniques being used - and this is rare - it is quite easy for the nature of the research project to be distorted subtlely, and sometimes grotesquely. Many groups now have the family facilitator make initial contact, and ask interested secondary contacts to reach the investigators (by phone, post card, or letter). Investigators then ensure that once a person expresses tentative interest, someone from the research team takes responsibility for explaining the nature of the risks, the purpose of the study, and the policies of the research institution. This is likely the optimum, but cannot be made a matter of fixed policy because of the need to make judgments on a case-by-case basis in particular families (for example, if there is an excessively enthusiastic family facilitator). The fact that neurological and psychiatric conditions are genetically complex and often involve diagnostic uncertainty makes such clarification all the more important.
The question of what to do with those who do not participate is also important. If someone chooses not to give samples, that is a fairly simple matter, but it does not put the issue to rest. Nonparticipation clearly means that clinical records will not be sought and stored, and samples will not be analyzed. But analysis of participating family members will often reconstruct the genotype of family members who do not participate. It seems that those who choose not to participate should not in general be able to stop this from happening, but if there is risk of third party disclosure, or leakage of information into the family, then the situation can change if this information could harm the nonparticipating person. This is not necessarily far-fetched. In one Alzheimer's family, the children of a nonparticipating affected person were informed of their risk despite the explicit wishes of the affected person's spouse. (This was done by family members, not the investigators.) This caused a major rift in that family, with resentment of the nonparticipant spouse among some children, who argued for a "right to know," and resentment of the intrusion by the family facilitator by other children, who argued for a "right not to know." This is no simple matter.
Withdrawal from a study is also complex. While no one would dispute that one's clinical records and identifying information (name, address, etc.) should be expunged upon request, even if one has previously participated, it is less clear what do to with genotype information already derived, or (usually) nonidentifying information relating to affected/unaffected status or number of children. The only precedent in US law, to my knowledge, is the Moore case in California, which would suggest that such nonidentifying information could be legally retained by the investigators. But this is a legal interpretation, and not necessarily the most ethical action nor the same interpretation that might be reached in other courts (in other states or abroad). One suspects that a European court might put a higher premium on privacy interests.
Neurological and psychiatric pedigrees are extremely valuable, and often are used and reused for decades. The importance of diagnostic rigour is quite important for neurological and psychiatric studies, and this often entails quite extensive clinical corroboration that may include expensive imaging techniques, neuropsychological testing, and repeated examinations by physician specialists. The trend now is to organise these into stable repositories at least for the most prevalent conditions, the model for which is the Huntington's registry in Indiana. This raises questions about how to protect family members from unwarranted intrusions by investigators, and the repositories generally have a scientific advisory group that screens research proposals and makes contact with the family, retaining contact information only at the registry, and sending out materials (such as stored DNA, cell lines, and clinical information) only with family members' consent. The accuracy of clinical information can be a sticky point here, as corroboration of clinical information almost always requires direct family contact and may require face-to-face examination by the investigating team. Again, it is hard to envision a comprehensive and fixed policy without case-by-case review.
The re-use of data and materials can also be complex. If identifying information is not used, this is most often not controversial. If extensive clinical or identifying information is necessary to the new study, in general this is grounds for obtaining informed consent again before proceeding. This may once again raise the issues of recruitment noted above, but this risk seems on balance smaller than the risk of proceeding without such consent.
The situation with central repositories is in most ways simpler than among smaller research groups, which in aggregate do most pedigree research. What happens to the data generated at considerable expense when a grant ends? Many call for destroying such information to protect family privacy, but this can fly in the face of the purpose for which family members contributed data and samples - to find a disease-associated gene as quickly as possible. We faced this question ourselves when our Alzheimer's grant ended. We had thousands and thousands of pages of information. Much of this consisted of clinical records - diagnostic procedures, autopsy reports, and physician examination notes - that had since been destroyed by the clinics and hospitals that originally housed them. Destroying the information would have irreversibly lost valuable clinical data, most of which related to dead affected individuals in the pedigree. When our grant ended, I suggested that we ship the clinical records and our incomplete genetic data (which predated the RFLP technique and were mainly derived from protein markers and phenotypes such a tongue-rolling) to other investigators we knew were working on Alzheimer's disease. The principal investigator disagreed, arguing that our institution might someday obtain another grant to continue the work. After several years, and when this investigator's laboratory was being closed down due to retirement, I made a unilateral decision to ship the core clinical records to the familial Alzheimer's Research Foundation, to the Gajdusek group at NIH, and to another group already working with the largest German Volga family. I did not confer with others in advance, although I did inform the department chair about my intent to share the data. In retrospect, I believe I should have shared the data even more widely.
I shipped off the clinical records knowing that there was no way to get the consent of those who initially consented to my getting the data. Many had died or moved so that I could not contact them. By this time, eight years or so after our research commenced, I did not even know how to contact the proband's spouse. My reasoning was that this valuable information had been given with the intent of expediting a search for the gene, and my forwarding it to other clearly qualified investigators was more in harmony with that goal than destroying the information or storing it somewhere it might get lost. I believe I did the right thing, but I also know others have good reasons to disagree.
Another class of ethical dilemmas can arise when investigators are drawn across the line into clinical care or social gate-keeping because of their research. Many investigators stress the need to distinguish research from clinical duties, but this boundary can be impossible to guard. In our work on Alzheimer's disease, for example, we were pursuing genetic linkage as a research matter. This entailed examining family members, often in settings outside clinics and hospitals. I was a medical student, and on field trips I was not accompanied by a more experienced (or board-certified) neurologist. Yet in several cases, I was the person who was asked if a family member should be allowed to drive, should sign a will, or should confer legal power of attorney to another family member. While it has never to my knowledge been formally published, a wealth of clinical experience suggests that those with dementia do have a much higher than normal frequency of minor auto accidents, for example, and some not so minor ones. (My grandmother who had Alzheimer's disease, for example, was said to "park by feel," coming to rest in her space by careening between the cars in front of her behind her, relying on noise and tactile feedback.) I was also drawn into a case of disability certification in a Texas social security proceeding, in which the judge stupidly decided that dementia was not legally disabling. I was also called upon to mediate family squabbles where some question about who had Alzheimer's or what it portended was at stake. We were conservative in demanding formal clinical evaluation before establishing an "affected" case, and we were very careful not to speculate about or share provisional or incomplete clinical data, but this did not stop me from being called in to referee occasional family spats. Even public safety can become an issue. One group studying Huntington's disease has discussed a case in which an air traffic controller is at risk. The person appears to be responsible and the investigators have talked about when it might be necessary for the patient to notify the employer, but what should investigators do if the patient refuses despite symptoms of cognitive decline? These social and gatekeeping functions are more indicative of a clinical relationship than research. The long-term relationships and specialized knowledge of the condition under study make it impossible to draw bright lines between the domains of research, clinical care, and nonmedical social and gatekeeping functions.
Once the data are in and some scientific finding is in hand, one faces the question of publication. This will often, although not always, entail the publication of pedigrees. This often causes a conflict between the privacy interests of family members, the scientific needs of the investigators and readers, and novel questions of informed consent. Some have taken to publishing incomplete or altered pedigrees. The critical issue here is the degree to which individuals or whole families can be identified by dint of merely making the pedigree public. The questions are far from clear, however, since information is neither fully private nor fully public, but often somewhere in between. If only those in the family can figure out their identity, where does this leave the investigators? Clearly, if some family members can find themselves in the pedigree, they may find surprises. In several of our families, for example, it became clear that information about who was affected and who was not was not spread uniformly throughout different branches of the family. Some were not surprised by black circles and squares others were. I made a mistake once by trying to elicit pedigree information from a family member. I was poring over the original pedigree (not the one published), which showed one case of adoption, several instances of artificial insemination by donor, and a case of nonpaternity. I had to feign that I could not recall precisely which parents were in question, and awkwardly move the pedigree off the table before confidences were violated. I learned to bring out only the published version of our pedigree after that, and I eliminated some information from that pedigree also. To my knowledge, we never had a case where the child of an affected person learned of their own risk by looking at our pedigree, but we could have, and those studying Huntington's disease and other disorders have reported this.
How public is public? I used to show a picture of 14 siblings, 11 of whom developed Alzheimer's disease. The photo was taken thirty years ago, but I stopped using it for fear that at some point, a family member or acquaintance would be in the audience and recognize someone, and would be appropriately offended. We always knew not to publish this photograph, but we did use the photo for lectures or scientific presentations, mainly because it made the family's plight more palpable to the audience. It is easier to imagine the impact of the disease if one can see a group of people rather than an abstract pedigree. It was good for a dramatic effect, and this was consonant with the family's interest, since they clearly wanted others to understand why they believed the search for the Alzheimer's gene in their family was a matter of some urgency, as a stepping-stone to understanding the mysterious but devastating disease. But over the years, I began to leave this slide out because I felt an inchoate sense of impropriety.
This matter of deciding what information to publish is again far from straight-forward. When we began our research on familial Alzheimer's disease in 1976, I culled through every article I could find in the international literature. At the time, the main British neurology textbook said flatly that Alzheimer's disease was not inherited, citing a report of twin discordance (4). (One of our early papers on familial Alzheimer's disease was about a case of such discordance that turned out to be concordance, but with over a decade difference in age of onset.) A very large Nordic study of families with Alzheimer's disease had concluded that the mode of inheritance was autosomal recessive, based on inspecting dozens of pedigrees. It was important to me to be able to go to the original paper and look at the ages of death and ages of onset. It turned out that early onset families that appeared to skip generations, in all pedigrees except one, could be explained if a parent died before living through the period of risk. Skipped generations might thus be due to a gene passed on by a parent who died before displaying symptoms. This made me more confident of our own interpretation that the mode of inheritance was autosomal dominant with high penetrance. Autosomal dominant inheritance now seems obvious, but it was far from obvious from the pedigrees published before 1976.
One way to handle the conflict between privacy and scientific accuracy is to eliminate "inessential" information, like the gender of family members, or age of onset or death of those affected. Yet the maternal inheritance of the genetic ophthalmoplegias (such as Kearn-Sayre syndrome), the paternal transmission of the Westphal variant of Huntington's disease, and the "anticipation" (earlier onset in later generations) of triplet repeat disorders like myotonic dystrophy and fragile X depended on noticing patterns not known to be significant at the time that pedigrees were first published. The birth order of siblings can even be important. Historical examples suggest caution in doctoring pedigrees, or even substantially limiting the amount of data displayed in them. Yet it is also clear that investigators can transgress privacy interests.
The problem of privacy may be even more important for neurological and psychiatric disorders than most other medical conditions. The risk of stigma is much higher for traits that affect cognition or emotion; epilepsy and movement disorders (Parkinson's, Huntington's, and others) have been stigmatised since there have been writers to record it. Many have speculated that some burned as witches actually had Huntington's disease or epilepsy, and the first victims of Nazi eugenics were not Jews but the mentally infirm. This raises the stakes for protecting privacy and ensuring confidentiality, but it paradoxically increases the importance attached to scientific understanding so necessary to eradicating stigma. For many mysterious disorders, the study of families can lead to genetic linkage and thence to identified genes, with the hope of finding a molecular handle to improve therapy and detection. Genetics may be the fast track to understanding, but may also place pioneering families at risk of stigma during the period of transition, before stigma abates (assuming it ever will).
Madison Powers of the Kennedy Institute of Ethics rightly argues that families should play a greater role in deciding about publication of pedigree data. This can be generalized to other potential intrusions on privacy or breaches of confidentiality. While the principle is laudable, its practical meaning is not yet clear. The basis for collective decision-making is a notorious gray zone in ethics. Do investigators have to get signed consent forms from literally hundreds of people listed on pedigrees, even when their names are not disclosed and identification may be possible only among the families themselves or among a select group of investigators? But is such publication warranted if there is ANY risk to someone who has not expressly given consent? Powers notes the adverse consequences of having the most reluctant family members have veto power over publication, for example, but sees no way around it.
Several issues are unclear here. One is who needs to give consent, and what it means to be a participant. Clearly those who contribute samples are participants, but does everyone need to be consulted anew each time a pedigree is published? How should those who refuse be handled? The default may be to use a gender-neutral place-holder, but this cannot always work, if surrounding relatives still make it possible to discern the identity of the nonparticipant. How identifiable is the information? In many cases, listing even birth order and number of siblings is sufficient to unequivocally identify main branches of a pedigree. Ages of onset, age at death, and affected status add considerably more information. Genotypes add even more, although this kind of information is generally accessible only to geneticists, not other family members or third parties. The standard that if individuals are identifiable, then informed consent is needed thus leaves open two unresolved issues: to what degree individuals are identifiable by whom, and what it means to have informed consent from a family, short of assent from every person in it.
The duty to disclose information of potential interest can also arise in pedigree research. Suppose that investigators succeed in establishing genetic linkage to a marker or find the actual gene. Clearly families that participated in research have to be informed that there may be information of interest to them that could be genetically tested. The simple solution is to corroborate the information by a separate clinical genetic test with separate informed consent. Many cannot be contacted in a typical pedigree, however, because data get old, leaving open the question of how hard scientists have to work to perform their ethical duty to educate family members. In one famous case, the separation between research and clinical testing was morally inappropriate. In a University of Michigan pedigree afflicted with familial breast cancer, two sisters and the mother had developed breast cancer, and a third sister had planned a prophylactic mastectomy. Yet the research genotype clearly indicated she had inherited the unaffected chromosome region from her mother - the wild type copy her two unfortunate sisters did not get. It would clearly have been unethical for the investigators to permit major unnecessary surgery, even though their initial intent was to use linkage to locate a gene, not to do clinical testing.
The clean way to handle this is to stipulate how clinically relevant information will be handled before starting the study, but many studies are started without thinking about what the investigators will do if they actually succeed. Clearly, the magnitude of harms that can be avoided and benefits that can be achieved are relevant, as well as the degree to which family members can act on the information. A treatable condition that is life-threatening implies a greater duty to track down and inform family members. This duty to disclose potential harm (or prospect of clinical benefit), clearly flies in the face of the "right not to know." Again, this is best handled prospectively, in the initial informed consent, but that requires that the investigators anticipate that information of interest to family members could become an issue and think through how they will handle it in advance. And some unanticipated cases, such as the sisters with breast cancer, will surely arise. Anticipatory agreement is much more the exception than the rule under current practice.
Thus far, we are treading on territory over which bioethics has tread, however lightly and recently. I will now proceed into even more treacherous terrain, flanked by land mines in the form of sanctimonious oversimplification and Polyanna-like naiveté on one side, and venal financial interests and careerism on the other. Daniel Pollen is one of the few scientists to honestly confront this issue, in a chapter of his marvellous book Hannah's Heirs: The Quest for the Genetic Origins of Alzheimer's Disease (New York and Oxford: Oxford University Press, 1993). Dr. Pollen was gentler towards some of his colleagues than I intend to be.
I want to talk about conflicts among and between the interests of those engaged in gene hunts - scientists and clinical investigators - and those whose genes are being hunted - the families. I will borrow some of the same general principles that Madison Powers and others have applied to pedigree research more generally, but with a new domain of application - the social contract between investigators and families affected with genetic disease.
One of the hallmarks of human genetics during the 1980s and into the 1990s has been its conspicuousness. The first Nobel Prize has not yet been awarded for work in this area, but surely one, or more likely several, will be. It beyond imagining that the Nobel committee remain forever oblivious to the shifting centre of gravity of biomedical research. Its track record suggests it will recognize at least one or a few among the relevant accomplishments: the Botstein and Davis insight of 1978 (and independently generated notion of Bodmer and Solomon) that led to RFLP mapping; the mapping and subsequent decade-long pursuit of the Huntington's gene; the trailblazing work on chronic granulomatous disease and Duchenne muscular dystrophy in Boston and Toronto (esp. Orkin, Kunkel, and Worton); the immensely competitive and sometimes nasty race for the cystic fibrosis gene, to map it in 1985 and to identify the gene itself in 1989; the dead-heat sprints for neurofibromatosis, colonic polyposis, and now early-onset breast cancer. Alzheimer's disease has been a succession of races for mapping and identification of one gene, and a prototype of the quest for multiple genes that share a clinical phenotype. It seems likely that the highest form of scientific recognition must come for some of this work.
The stakes are quite high in medical terms also. A successful mapping project for dread diseases - those that are severe, highly prevalent, or publicly known for some other reason - is sure to land one at least a feature in the New York Times science section on some Tuesday, and often on the front page of multiple papers. Many of the principal investigators of large collaborations have been featured in television documentaries, TV news, and weekly news magazines such as Newsweek, Time, and The Economist. There has even been a special issue of Consumer Reports. Public esteem is matched by academic accolades. Gene jockeys who ride the winning horses (or, more precisely, drive teams of postdoctoral fellows, graduate students, and clinical colleagues) have become among the hottest properties in the academic Monopoly game. They become the big fish in their university ponds, commanding laboratory space, research resources, high salaries, and other concessions. They may even do well in study sections and tenure committees the next year, although this is less certain (the academic community's generosity does have its limits, and the further from public view the more obvious those limits are). One rule is stable, although not entirely invariant - the spoils go to the winner.
Competition is inherent to modern science, and perhaps no field is more competitive than human molecular genetics right now. In this respect, it follows retroviral oncogenes in the early 1980s, AIDS retrovirology soon thereafter, the gene regulation field in the mid-1980s, and a succession of other "hot" subfields within molecular genetics. Human gene therapy is now heating up to the same boiling point, and may become the next hypercompetitive subfield.
This intensity of competition severely stresses the ideals of science, which promotes the sharing of data and materials after publication so that science as a whole can progress faster. These norms are observed, for the most part, in the technical fields that gave rise to the technologies of the genome project. The phage group was famous for friendly competition and wide sharing of results. When Jim Watson's Double Helix (New York: Norton, 1980) profiled the possessive and competitive faces of science, Max Delbrück and Salvador Luria of the phage group made no secret of their displeasure. The other fields that share this intellectual heritage from the phage group, such as yeast and
The race to map the CF gene was a study in tensions between collaboration and conflict. A US company, notably named Collaborative Research, guided by the work of Helen Donis-Keller, contributed a series of probes to Lap-Chee Tsui of Toronto, who had the pedigree resources to do linkage analysis. The Collaborative Research probes, developed at the company's expense, included the one that first linked to the gene. The chromosomal origin of this probe was not clear, however, and delay in nailing it down led to several months of rumours. Collaborative Research and the Toronto group were extremely anxious, but the company was awaiting confirmation from a third research group that all had agreed would do the corroborative in situ hybridization. That group was preoccupied with another gene hunt. As I understand it, French geneticist Jean Frezal spilled the beans about preliminary chromosome 7 linkage at a Paris press conference, and this information made its way to Ray White at Utah and Bob Williamson in London. Each proceeded to check their chromosome 7 markers, and each found markers closer to the gene than the initial one used by Tsui. They quickly prepared papers, but the Toronto-Collaborative Research team got wind of it, and prepared its own paper. The risk of a very nasty priority squabble, and perhaps a corollary patent battle, was high, but the story came out - first in Nature, and then in two long Science features by Leslie Roberts, among the most gripping pieces of science journalism about human gene hunts (5, 6).
The tension did not end with successful mapping, but merely became rechanneled into the hunt for the CF gene itself. I remember sitting at a Gordon Conference in 1987, when Stuart Orkin talked about chronic granulomatous disease, Lou Kunkel about Duchenne muscular dystrophy, and Bob Williamson about cystic fibrosis. I was sitting with Ray White on one side and Helen Donis-Keller on the other, and the question and answer period after Williamson's paper was extremely tense, as Williamson had reported linkage disequilibrium, suggesting he was very close to the CF gene. Those assembled more or less assumed he had the gene, but he was waffling, and it turned out that he did not. White politely but pointedly asked about the inconsistent order of markers on the different genetic linkage maps he displayed, and Donis-Keller asked point blank why after the initial linkage papers had already been published, Williamson had not published the relevant sequence data.
When the Toronto and Michigan groups led by Tsui and Francis Collins, respectively, found the gene in 1989, some of the rancor had gone out of the race. The Collins-Tsui collaboration was notable for its smoothness most of the time. There is no way of knowing the role of Leslie Roberts's news features, but perhaps they made clear that practices even appearing to diverge from the scientific ideals might damage careers as much as coming in second.
The race for the neurofibromatosis was another cliffhanger, with Francis Collins's group and Ray White's group the first finishers, again chronicled by Leslie Roberts (7). A similar race led to a photo finish for the colonic polyposis gene. Clearly, this was becoming the norm, with a handful of extremely able teams throughout the world working feverishly in competition to map and identify the most salient human disease-associated genes. In the face of such intense competition, one can surely forgive a reflex to protect one's lead.
The race mentality has several features to recommend it. It keeps armies of postdocs and senior investigators working through the night. It puts a premium on not making a major mistake, although molecular biology is considerably sloppier than other fields such as mathematics. Many major laboratories have made embarrassing errors of commission (e.g., reporting a linkage not later replicated by others) or omission (e.g., both the Utah and Michigan groups initially missed a biological similarity between neurofibromatosis and a known gene because the data were new and because they were in a horrendous rush to publish). The competitive mold must allow for mistakes, leaving follow-up work by discoverers, along with the plethora of groups pursuing parallel research to clean up the messes. This competitive spirit enlivens meetings, gives great urgency to daily laboratory work, and makes for good stories. The high stakes of gene quests make even better news. A bit of controversy and rivalry always makes better copy.
One disadvantage of the race mentality, however, is the prisoner's dilemma. All groups would be better off, that is they would in aggregate have more information to narrow the search for genes more quickly, if all pooled their pedigrees, DNA samples, probes, and other resources. But each group has an incentive not to share at the margin if it believes it has an edge or potential edge. This is particularly acute when one group worries about competitors who have more postdocs, more automated equipment, better analytical software, or some other advantage. In the middle of battle, one focuses on the adversaries' advantages, minimizing ones own strengths because they are not the immediate threat. In this frame of mind, sharing information and materials may seem a suicidal act of senseless altruism.
The limits to the sharing ethic are also embedded in human genetics more than the fields that gave rise to genome research technologies. Clinical research has always been more territorial, perhaps because it is larger with more unknown competitors, or perhaps because doctors are just more worldly and territorial than folks who care about yeast or phage. The human genetics research that grew out of somatic cell hybrids was a notorious hotbed of contention, with some groups widely known to demand coauthorship and explicit collaboration in return for sharing cell lines. This practice, once pursued by one or a few major laboratories, can spread through a subfield. These practices, particularly when they do not involve materials garnered from individuals or patients whose intentions should be considered, may not be frankly unethical, but merely cause inefficiency and tarnish science. Indeed, human genetics has long been notable for testing the limits of propriety, and those who confront its nasty side when they are used to yeast or
Human pedigree research extending back through the decades, however, has missed many opportunities for progress through sharing. Perhaps I generalize too freely, but I have certainly been struck in perusing the more than 80 major Alzheimer's families reported since the 1930s how rare it is to find different groups studying the same pedigrees with new technologies, handing off their resources to groups who can more quickly do the analysis. With the advent of RFLP mapping in the 1980s, and particularly after the mapping of the Huntington's gene in 1983, the Alzheimer's field seems to have settled into contending camps. Family resources have, for the most part, been handled as the territory of specific research groups, who do not share the DNA or access to family members. Access to the DNA and clinical records is generally only bought by negotiation or explicit alliances. The justification for this practice can take the guise of protecting family privacy interests, but many opportunities to follow the Huntington's registry model, where this issue does not arise, have been allowed to lapse. Alzheimer's disease genetics, like many other areas of human genetics, has been Balkanized.
This came to a head in connection with linkage of many early-onset Alzheimer's families to the long arm of chromosome 14. In the fall of 1992, the Alzheimer's genetic community was rife with rumours of impropriety when a paper by Peter St. George-Hyslop's group in Toronto group was reviewed for Nature. One of the groups sent the paper for review recommended its rejection by Nature. The group then proceeded to submit a paper of its own with weaker linkage data for the same chromosome. This delayed publication of the initial paper, which was resubmitted to the new journal Nature Genetics, and priority initially appeared to be shared among four groups. The Toronto group was joined by a group from Seattle, another in Belgium, and a third at the University of South Florida. As it turns out, the discovery was apparently made independently by the Toronto group, the Seattle group, and the Belgian group (8). The first publication, in Science, came from the Seattle group. The other three groups submitted papers within days of one another to Nature Genetics. (For the Toronto group, the Nature Genetics paper was a resubmission, following the rejection by Nature.) The Belgian group, directed by Christine van Broeckhoven, had been compiling linkage data for some time, and pulled together the data for publication when she got wind of the other groups' linkages. John Maddox, the editor of Nature, saw fit to publish what could charitably called a weak and elliptical defense of the Florida team, which was suspected of the review transgression (9). The exact chronology of events is in dispute. John Hardy and Mike Mullan, the two involved in the Nature review process, were suspected by many in the Alzheimer's research community of having used their review copy of the Nature paper to redirect their research effort and to generate a seemingly independent publication while delaying publication by their competitor (10). Hardy and Mullan contend otherwise, saying they were alerted to a possible chromosome 14 linkage by a reporter. On another tangent, a Duke University group under Allen Roses was perturbed that Gerard Schellenberg's group in Seattle used some of their family materials without acknowledgment. Schellenberg says he did acknowledge it, and had been studying the families independently anyway (10).
Ethical canons of science exhort scientists to share data and promote cooperation, but details about how to do this are generally lacking, and in particular human geneticists have been conspicuously silent. Formal mechanisms for investigating alleged breaches of scientific ethics are weak, except in the area of human subjects protections (Nature's cursory examination is accorded little credibility by most in the field). Perhaps the mechanisms are not important as long as such matters become public, but often they do not. Even in this case, where part of the story was picked up by The Journal of NIH Research (10), the whole story will likely never come to light. The scientists involved must continue to meet at conferences and the field must progress, including those whose work's priority is being disputed, even after a transgression. Perhaps it is enough that a shadow hangs over the scientific reputations of John Hardy and Mike Mullan. And then again, perhaps it is not.
One of the puzzling aspects of the current system for according scientific prestige is that the lion's share of credit goes to the gene jockeys. Those who pursue the clinical information, pedigree background data, and assemble the family data often labour for years. The amount of effort to construct a pedigree with clinical data sufficiently accurate to use for linkage analysis is generally more intense than the amount of labour devoted to scanning for linkage analysis using modern techniques. The current regime is not entirely unfair, however, as it does place a premium on applying the newest technologies, and this encourages groups to work faster and to innovate. The pleasures of pedigree construction, moreover, go far beyond scientific credit.
As a young medical student, I forged intimate ties with families in a half dozen states. I shared their homes, joined their family reunions, learned their stories, and experienced their pain at a personal level that has enriched my life. I intervened for family harmony, despite being a threat to some members; I arranged for free medical care; and for some in those families, I was their doctor. The person who first brought our largest family to light was a special case. I eventually lifted his brain out of his cranium and held it in my hands; I was the first person to physically see it. I knew his entire family on a first-name basis. I stayed in the farm house he built. His life was an important part of my life. It was a true honour to do his autopsy.
Afterward, I looked at the plaques and tangles in his brain under the microscope, and helped confirm the diagnosis. That experience gave Alzheimer's disease new meaning, and the quest for its genetic origins a new intensity. These are experiences that those who only run the gels, read the genotypes, or run the sequencing machines cannot share. It would have been nice if we found linkage in 1977 or 1978, by some miracle of luck, but it does not diminish the experience of pedigree research that we did not.
I was vicariously thrilled when St. George-Hyslop and his group found the first linkage to chromosome 21, and have found the identification of the APP gene on that chromosome, and subsequent linkages on chromosomes 14 and 19 extremely exciting to observe at a distance. I felt the families had in some way been vindicated. They believed in science, and it made the promised incremental progress. Science fed hope. Given the choice of roles, starring in the gene hunt or meeting the people whose genes were hunted, I would choose the role I played without a moment's hesitation. I got to know the people. Those who finally identify the genes for Alzheimer's disease will, however, have indeed made the more scientifically significant, if less sociologically enriching, work. The unfairness is the current way of allocating scientific credit is thus not that it unfairly recognizes true accomplishments, but that it fails to acknowledge the underpinnings necessary to genetic linkage and mutation analysis. This is thus a small inequity, acceptable for the
expeditious conduct of science.
Whatever the sociological and legal mechanisms for encouraging good behaviour in collective science, as almost all biomedical research has become, there is a larger issue at stake. The families who have contributed their blood and information - who have given of their time and given investigators extremely intimate, often intimidating personal data - deserve better. It is one thing to hoard crystallographic data, bacterial clones, or scientific data that are not attached to individuals. It is quite another when materials and information derive from individuals who suffer from a disease. The families that live with the ravages of Alzheimer's disease want to beat it. They give of themselves hoping future generations will not face what they did, and if the families I worked with are typical, they expect the investigators to have the families' best interests at heart. This is not a doctor-patient relationship, but its collective analog in the research arena, a tacit agreement between a group of family members and a group of investigators. Squabbles that are transparently related to credit-mongering can endanger this moral bond. The obvious first resort to resolve this situation is informed consent. If investigators intend to treat "their" families as chattel, as part of their scientific domain, then this should be agreed up front. Family members should know how their materials will be treated, if it is other than freely available to those with legitimate scientific interests, including competitors. Those who would share their resources can also make that plain, but this should be the default assumption. If families are offered a choice, they are likely to encourage sharing, as constrained by agreed-upon mechanisms, such as outside advisory committees to judge the scientific merit of requests for data and materials. Families may not know there are options at first, and many are exhilarated merely to find someone interested in their problem. But over time, large families that are valuable for research will soon learn, particularly if Institutional Review Boards ask for prospective agreements about the disposition of data and materials. The question of confidentiality need not be conflated with this issue of data sharing, as there will only be a half dozen groups involved in even the most populous races, and all will be bound by confidentiality constraints with respect to other family members and third parties. Over time, the norms may change, confining the incentive to win within moral boundaries agreed in advance. This would return some authority to those who really have the most to win or lose in the medical quest - the families.
1. Frankel, M. S. & Teich, A. H., Ethical and Legal Issues in Pedigree Research. (Washington, DC, American Association for the Advancement of Science, 1993).
2. Office for Protection from Research Risks, National Institutes of Health (1993). Chapter 5, Section H: Human Genetic Research. Protecting Human Subjects: Institutional Review Board Guidebook. Ed. Series Ed. Bethesda, MD 20892, National Institutes of Health. Second, ed., pp. 5-42 to 5-63.
3. Powers, M. (1993) Publication-Related Risks to Privacy: Ethical Implications of Pedigree Studies. IRB: A Review of Human Subjects Research 15 (July-August): 7-11.
4. Walton, J. Brain's Diseases of the Nervous System. (Oxford, Oxford University Press, 1977).
5. Roberts, L. (1988) Race for Cystic Fibrosis Gene Nears End. Science 240: 282-285.
6. Roberts, L. (1988) The Race for the Cystic Fibrosis Gene. Science 240: 141-144.
7. Roberts, L. (1991) The Rush to Publish. Science 251: 260-263.
8. Pollen, D. Hannah's Heirs: The Quest for the Genetic Origins of Alzheimer's Disease. (New York, Oxford University Press, 1993).
9. Maddox, J. (1992) Conflicts of Interest Declared. Nature 360: 205.
10. Hooper, C. (1992) Of Rumors, Reviews, and Rip-Offs in Alzheimer's Research. Journal of NIH Research 4 (December): 34.