Date Published: September 23, 2008
Publisher: Public Library of Science
Author(s): David R Karp, Shelley Carlin, Robert Cook-Deegan, Daniel E Ford, Gail Geller, David N Glass, Hank Greely, Joel Guthridge, Jeffrey Kahn, Richard Kaslow, Cheryl Kraft, Kathleen MacQueen, Bradley Malin, Richard H Scheuerman, Jeremy Sugarman
Abstract: David Karp and colleagues discuss the ethical and practical concerns that arise when data are shared in aggregated databases.
Partial Text: The goal of “personalized medicine” relies upon defining the genetic variation responsible for disease susceptibility and response to therapy . For most common human diseases, the contribution of a single sequence variant to disease susceptibility is typically small, and can only be detected with data from large numbers of people . Practically, this necessitates collaboration among investigators who either have DNA and phenotypic information previously collected, or have access to populations from which to recruit participants. It also requires that data be shared among the collaborators. Modern bioinformatics platforms have the capacity to combine datasets and store them for re-analysis. This is scientifically advantageous since it makes possible studies with enhanced validity in a cost-effective fashion. However, this data storage can complicate the already vexing practical, scientific, and ethical issues associated with gene and tissue banks. Research participants’ data may have been collected without authorization that meets today’s standards for informed consent. Research participants may not have consented to participation in genetics research in general, to inclusion in genetics databases specifically, or to use of their samples in genetic analyses that were unanticipated, unknown, or nonexistent at the time samples were collected . Participants who consented to the collection of their data for use in a particular study, or inclusion in a particular database, may not have consented to “secondary uses” of those data for unrelated research, or use by other investigators or third parties . There is concern that institutional review boards (IRBs) or similar bodies will not approve of the formation of aggregated databases or will limit the types of studies that can be done with them, even if those studies are believed by others to be appropriate, since there is a lack of consensus about how to deal with re-use of data in this manner.
Two of the authors (DRK and JS) organized the two-day meeting. Potential participants who were known to the organizers either as stakeholders in ImmPort or for their work in the fields of research methodology and regulation were invited to participate. These included NIH program staff working with ImmPort, bioethicists, NIH-supported researchers who would be required to submit data to ImmPort, and content experts. We sought people with expertise in the areas of informed consent, protection of vulnerable populations, privacy protections, research regulation, and intellectual property. Lay members of the University of Texas Southwestern Medical Center IRB and research participants were also included. A total of 32 people were invited, and 20 participated. Six discussants were asked to present the issues and controversies relating to three general themes: Assuring Data Privacy and Security, Informed Consent, and Challenges of Collaboration. Audio recordings of the presentations and discussion, including controversial issues and dissenting opinions, were prepared for reference. After the presentations and discussion by all participants, the major concepts were summarized by the organizers along with a listing of problem areas and possible solutions, which were discussed in detail by the entire group. Consensus on the recommendations was achieved and refined through a collaborative writing process by all authors. The strengths of this meeting include the focus on a particular research topic, the ImmPort project, with potential wide applicability to similar initiatives, by a small group of persons with relevant expertise. The weakness of this method is also related to these aspects, as the selected participants may not have addressed all possible pitfalls in database aggregation and sharing.
The panel developed seven broad recommendations, which are summarized in Box 1.
While there is considerable activity focused on providing public access to clinical trial data and on merging multiple databases, a set of best practices for this type of research is clearly needed. It is hoped that the recommendations offered here will facilitate scientifically and ethically sound research. Since it is unlikely that the issues and proposed solutions described here will be static over time, they will need to be reviewed periodically. The consistency of data sharing and data protection must be evaluated and maintained. Data protection standards will evolve, and a methodology that was appropriate at one time may not be appropriate later. Equally likely to change are standards imposed by the user community and donor-participant community, and privacy protection methods will need to reflect those changes in order to retain the trust of all stakeholders. This is necessary to help realize the potential for personalized medicine.