One day in the near future, during a routine checkup, your doctor recommends that you have a simple genetic test to determine how you are likely to respond to a certain pharmaceutical treatment. Because the cost of sequencing has fallen to the point where collecting a full exome sequence is competitive with the one-off test for your condition, your doctor orders the more general test instead.
In addition to answering the clinical question at hand, the exome sequence reveals much more. For example, let us suppose that your susceptibility to Alzheimer’s disease is four times higher than expected in people of your ethnic background. Do you want to know? Should your doctor know? If he does know, is he ethically obligated to reveal this information to you? Should your insurance company know? Who should control the disclosure of this information? How can compliance with disclosure policies be audited and enforced?
The problem of incidental findings in genetic testing is ethically problematic, and there are no clear-cut, universally accepted guidelines. The Presidential Commission for the Study of Bioethical Issues recently issued a report that begins to lay out a framework for handling incidental findings, but the report’s recommendations might be summarized as:
- Anticipate incidental findings
- Have a plan for dealing with them
- Communicate clearly with the stakeholders involved
These recommendations are reasonable but toothless; the Commission does not make concrete recommendations as to what the proper policies should be. Other organizations, such as the American College of Medical Genetics, go further recommending automatic disclosure of specific, actionable markers to doctors and patients.
In a thoughtful post based on an article in the British Medical Journal, Ewan Birney (EMBL-EBI) and Caroline Wright (Wellcome Trust) propose that disclosure of incidental findings should be based on the evidence for and against the clinical significance of specific variants as well as their relevance to the specific disease being investigated. They argue that incidental findings should not necessarily be disclosed automatically:
Experience (and human nature) tells us that it is easy to overstate the benefits of screening while underestimating the harms from inappropriate treatment and worry.
Given the diversity of considered opinions regarding incidental findings, how should computer systems that process genomic information behave? We believe that systems should be flexible enough to encode and enforce a variety of policies, ranging from automatic full disclosure to total non-disclosure and encompassing all of the pragmatic middle ground. We believe that ultimately, the decision to learn or reveal particular information rests with the patient or subject.
Genecloud helps to address the issue of incidental findings in several ways:
- By making it possible for programs that interact with sensitive genomic information to ask for very specific information via a well-defined set of APIs, minimizing the chance that incidental findings are revealed;
- By applying digitally-encoded policies at the point of access to sensitive information, preventing programs from accessing prohibited or non-relevant information (and thus potentially disclosing it to inappropriate parties);
- By rigorously and securely auditing all data accesses so that patients are aware of who accessed their information, when, via what program, and for what purpose.
We do not believe that there is a single one-size-fits-all policy that is appropriate for every circumstance; rather, we believe that systems must be flexible enough to accommodate a variety of different policy choices as appropriate to the circumstances and the wishes of the stakeholders in the data. The architecture of Genecloud is an expression these beliefs, and in our opinion, a model for managing incidental findings without requiring the a priori resolution of all possible ethical issues.