Following the overview of the National Health and Nutrition Examination Survey (NHANES), the workshop considered different perspectives on returning genomic results to study participants, including a consideration of the philosophical foundations for returning results. Presenters reflected a vigorous recent debate about return of results, exemplified by a 2013 point-counterpoint exchange in Science with regard to incidental findings from clinical sequencing (see McGuire et al., 2013; Wolf, Annas, and Elias, 2013). Presenters included Henry S. Richardson, Georgetown University; Steven Joffe, University of Pennsylvania; and Susan M. Wolf, University of Minnesota, with the session chaired by Jeffrey Botkin, University of Utah.
The overriding question that underlies much of today’s debate centers on whether there is an obligation on the part of researchers and/or organizations to report or follow up on unexpected or incidental results and, further, whether there is an obligation to search for incidental or secondary findings of significance. This question may arise in various health-related research contexts, such as, for example, the discovery of HIV infection while doing a study of malaria treatment protocol. Henry Richardson began by discussing a “general duty of rescue,” a basic obligation that is incumbent on all persons at all times if one can save someone else from dire peril easily without expending enormous amounts of energy or put-
ting oneself at risk (Merritt, Taylor, and Mullany, 2010). This obligation covers simple warnings, referrals, and simple therapeutic efforts, but does not cover warnings that are logistically difficult, does not cover payments for care, and does not cover therapeutic efforts that are complex and difficult, which is often the case in the context of genetic findings.
Given the difficulties, why would it be obligatory to return results and what is the basis of the obligation? Richardson presented a partial entrustment model, developed initially to address ancillary care responsibility; that is, care that subjects need but which is not required for sound science or study safety. It is a partial model in that it does not cover everything that subjects need, but its scope does include information that comes to light via study procedures, and thus includes incidental findings. He suggested the model can be broadened to include care and help that subjects may need, including return of information (Belsky and Richardson, 2004; Richardson and Cho, 2012).
“Entrustment” comes about as a by-product of the informed consent process, Richardson said, noting that informed consent has a number of functions: a watchdog function, to prevent abuses; a more high-minded function of putting potential subjects in a position to decide autonomously whether to participate; and a basic permission function to proceed with research (e.g., special permissions to touch and sample bodies).1 Such permissions are rights waivers. People have privacy rights that they are waiving in the informed consent procedure. When researchers accept these waivers, they take on special responsibilities vis-à-vis research participants, Richardson asserted.
In Richardson’s view, the partial entrustment model explains why secondary researchers who use banked specimens, and hence have no direct contact with subjects, still may have obligations to report results (via some indirect chain) because the special permissions are inherited or passed along. Researchers have obtained permissions indirectly and therefore have accrued the special responsibilities. These obligations go beyond easy rescue, yet remain limited in strength and will not cover all cases of return of genetic findings (e.g., a case with some cost but little difference to a person’s well-being). The model does not support a duty to hunt for findings, he added.
Steven Joffe discussed some of the reasons that he said might suggest a cautious approach to returning results to study participants. He framed
1See National Research Council (2010) for additional aspects of informed consent.
his discussion in terms of two important questions: (1) ought researchers to offer to return results (or some subset) to participants and, if so, (2) can they actually carry out such a return policy?
Regarding the “ought” issue, he said common guidelines for both nongenetic and genetic results suggest that individual genetic results should be offered to study participants in a timely manner if the results meet all of the following criteria:
a. The finding(s) has important health implications for the participant, and the associated risks are established and substantial.
b. The finding(s) is actionable.
c. The test is analytically valid, and the disclosure plan complies with all applicable laws.
d. The participant has opted to receive his or her individual genetic results.
Joffe continued that the issue then arises as to whether it should be permissible for investigators and participants to agree that findings will not be returned, as has often been the case in past consent procedures (including NHANES). Informed consent, as Richardson noted previously, spells out terms of agreement between an investigator and participant, and is a quasi-contract. As a general matter, competent adults have wide latitude to set the terms of their agreements. Joffe asked if this should be true in the research arena as well. He noted that some terms are or could be considered unconscionable, while others are expressly prohibited by applicable regulations or laws. In the case of return of results, an agreement not to return could be unconscionable if subjects agree to accept a risk of serious harm, but it is more likely the case that the agreement to forego results relates to some potential side benefit. As to whether such an agreement is expressly prohibited, Joffe pointed out there does not appear to be anything in the Common Rule2 (which governs biomedical and behavioral research with human subjects conducted by federal agencies), or in other laws or regulations at the federal level, that would proscribe it.
This leads to the “can” question, he said: Can researchers carry out the duty (if they accept a duty) to return results given the state of current technologies, which includes not only genomic technologies such as sequencing, but also database technologies, informatics, and curation technologies? It seems extraordinarily difficult to ask a group of researchers to identify pathogenic variants in an unselected population, stated Joffe. To illustrate the difficulties, he used data from a study of participants in ClinSeq, a large-scale medical sequencing clinical research pilot study that
seeks to identify “likely pathogenic” and “definitely pathogenic” variants in 37 high-penetrance cancer-susceptibility genes (Johnston et al., 2012; National Human Genome Research Institute, 2012). There is a potentially large problem with false-positive results, which may stem from several sources: technical error in reading sequences; false-positive associations in the literature (e.g., due to multiple hypothesis testing); incorrect penetrance estimates, especially because most data are based on variants from individuals with known phenotypes or family histories; and inaccurate annotation in reference databases. In Joffe’s view, asking research teams to navigate these and related issues would be feasible only if a project were to prospectively specify the genes of interest and the particular variants within those genes, such that all of the curation work could be done prospectively and was not something that research teams had to confront as they dealt with participants and samples.
Joffe said he sees four choices regarding return of genomic results: (1) have no requirement for return as a general matter; (2) require return, but accept a lesser standard of interpretive validity than in a clinical genetics lab, particularly with regard to false positives; (3) expect research teams to achieve clinical-lab standards (which he said seems most unlikely); and (4) send every sample (or every screen-positive sample) to a clinical lab. He concluded that an informed agreement between investigator and prospective participant that no results will be returned is ethically and legally permissible. Further, he said, unless a list of returnable variants (not genes) is specified in advance, the requirement to return secondary findings would be immensely costly, do more harm than good, or both.
Susan Wolf added another dimension to the discussion by focusing her remarks around the notion of public accountability. NHANES is a publicly funded, population-based study that relies on public trust, she noted, and NHANES should approach the return-of-results question in a way that will fulfill its responsibilities to the public and maintain public trust. Involvement of the public in setting policy will be important, in part to understand what study participants regard as valuable information and, in their own lives, actionable information.
Her work on return of results and incidental findings in the context of a biobank or a biorepository suggests that the key concept is that of a biobank research system. One cannot adequately address the return-of-results question if biobank responsibilities are analyzed in isolation. Rather, she stated, biobanks are part of an information flow that often begins with primary research and collections sites (stage 1) feeding data and specimens into a central biobank or similar resource (stage 2). In
addition to storing data and specimens, biobanks can do their own data and sample collections, curation, annotation, analyses and reanalyses, perform quality control, and confirm pathology. And importantly, biobanks can supply secondary research sites (stage 3) with data and specimens. Incidental findings and individual research results that may be considered for return to participants can arise at all three stages (Wolf et al., 2012; Wolf, 2013). A decade or so ago, Wolf said the conventional view was that biobanks should not be concerned with return of results. If something were discovered in a biobank or secondary research site, it would, at most, be returned to the primary research site. More recently, there has been substantial pushback on this view, acknowledging a growing public (participant) interest in results. NHANES should address the role of the biobank research system, she suggested, which in this case is a biobank research system with the responsibilities of a public entity doing a population-based study. NHANES has been returning nongenetic findings for decades, and there is no sharp distinction between genetic and nongenetic findings for the purpose of potential return. Indeed, NHANES has the potential to pioneer publicly accountable return of results and incidental findings in the context of public health. Wolf said that NHANES should undertake a deep consultation with the public in terms of what results and findings they are interested in being offered. She noted that withholding data from subjects may make them “passive purveyors of biomaterials and data” rather than research partners (Kohane et al., 2007).
Wolf mentioned governance innovation already going on via a number of consortia, including the GENEVA Committee on Incidental Findings,3 the eMERGE Network Return of Results Oversight Committee,4 the UK Biobank Ethics and Governance Council,5 and the Coriell Personalized Medicine Collaborative’s Informed Cohort Oversight Board.6 Dynamic
3The Gene Environment Association Studies (GENEVA) consortium was initiated by the National Human Genome Research Institute (NHGRI) to identify genetic variants related to complex diseases; to identify variations in gene-trait associations related to environmental exposures; and to ensure rapid sharing of data through the Database of Genotypes and Phenotypes (dbGaP). See https://www.genome.gov/27541319 [June 2014].
4The eMERGE (electronic MEdical Records and GEnomics) Network is an NHGRI-supported consortium of five institutions exploring the utility of DNA repositories coupled to Electronic Medical Record (EMR) systems for advancing discovery in genome science. eMERGE also includes an emphasis on the ethical, legal, and social issues related to these endeavors. See http://www.biomedcentral.com/1755-8794/4/13 [June 2014].
5The UK Biobank is a national health resource established by the Wellcome Trust Medical Charity, the UK Medical Research Council, the UK Department of Health, the Scottish Government, and the Northwest Regional Development Agency. See https://www.ukbiobank.ac.uk/ [June 2014].
6See http://cpmc1.coriell.org/about-the-cpmc-study/advisory-boards [June 2014].
consent processes are being developed to use informatics to track preferences over time.
The cost of return is an important issue, though little research has been published to date rigorously specifying that cost. Another important issue is whether to offer results and incidental findings to a study participant’s family members, including after the death of the participant, as genetic and genomic findings may have health and reproductive implications for family members.
Wolf said her main message is that NHANES needs more participatory governance and a governance structure that brings the public into the discussion on return of results and incidental findings. There are many research opportunities in this rich study, she suggested. To simply say “no return” due to technical difficulties is no longer feasible because, as she characterized it, NHANES is a public study with major public responsibilities. The next steps, in her view, involve figuring out how to calibrate the return of results, obtaining accurate costs for different options, and beginning a serious process of public engagement. She suggested that telephonic genetic counseling may be an avenue to explore for containing costs.
Wolf went on to discuss a temporal aspect of DNA and genomic information, one that involves the notion of durability. Basic genomic information does not appreciably change over an individual’s lifespan. Wolf said one issue for NHANES would be identifying the types of analyses that can be done, and the resources needed, in order to produce participant information in the 12- to 16-week window after the survey, if NHANES continues to adhere to that timeframe. A second issue, Wolf said, is that science does not know nearly as much about the genome now as it will in two or five years in the future. Gene variants that do not appear actionable at this point in time will be interpreted as actionable going forward.
When thinking about ethical responsibilities in the NHANES study context, Wolf said it is important to recognize the roles of the different actors in the process: data collectors, data curators, and secondary researchers (the three stages she identified in a biobank research system). The National Center for Health Statistics (NCHS) is tasked and funded to collect data, return individual test results to the subjects, make summary statistics available to the public, and provide anonymized data to researchers. NCHS does very little research and is not funded to come up with findings or to determine whether something is actionable or not actionable. After the survey data collection, the genomic specimens go to the Centers for Disease Control and Prevention (CDC) laboratory. The laboratory’s duty is curation; its duty does not extend to analysis, evaluation, or a responsibility for reporting results to participants. The third set of actors is researchers, those who analyze the data in the CDC biobank.
Wolf urged careful consideration of the ethical responsibilities in the context of each of these three roles. Those occupying these roles have distinct capabilities and limitations when it comes to determining when there is a responsibility to get back to a participant, and in actually doing so. Although this is the reality of how the NHANES program has operated in the past, Wolf suggested considering how the program should operate going forward. There is widespread national and international consensus that biorepositories have duties of stewardship and ethical responsibility. Specifying NHANES responsibilities with regard to return of results and incidental findings is a crucial next step.