there’s a risk in 23andMe calling genetic alleles (a type of variation) incorrectly. 23andMe’s fundamental difference when compared to the many hundreds of laboratories performing genetic tests is that they are going directly to consumers rather than through the health care establishment. Unlike clinical laboratories, however, they approached the FDA, had numerous meetings, and then stopped talking.
Did the FDA overreach? Or, is 23andMe playing out a brilliant cat and mouse strategy to increase public awareness and support of consumer genetic testing? Because genetic tests are broad in scope and complex in detail, a single test provides thousands of results. The real issue is that the vast majority of “diagnostic” alleles have not been validated for their utility as biomarkers for a disease or therapeutic protocol in a clinical trial. An example, cited in the FDA letter, was related to warfarin dosing based on “false genotyping results” that could have “significant unreasonable risk of illness, injury, or death to the patient due to thrombosis or bleeding events” because of improper dosing.
This was a particularly poor example because the benefit of genotype-guided warfarin dosing remains unclear, even after several trials. So, the question here is not whether an allele is reported correctly as much as does it have the effect we think it does. In short, genetic tests can be interpreted with variable results because our knowledge of the relationship between genetic differences and the mechanisms of disease are incomplete. Filling in this knowledge requires that millions of individuals have their DNA analyzed.
23andMe is bringing value in terms of increasing public awareness and increasing participation. The FDA has legitimate issues with the company’s tendency to overstate the diagnostic potential of the tests. The argument of incorrect results is flawed, but the underlying issue of misinterpreting results due to a lack of scientific rigor was missed or not discussed, at least partly because it affects a major portion of the clinical testing industry. This latter point could be an opportunity for 23andMe.
In the current flurry of headlines, the 23andMe business model has escaped notice. When you sign up for their service, you give 23andMe rights to use your data anyway they see fit. You can have your data, which satisfies the personal genomics folks, but not that of the others who have signed up, unless they specifically share with you. Many interpret this activity as data sharing. It is not. 23andMe has all the data, you get to see a little. If the company opened their data to the research community they could truly make real advances and possibly turn this setback into a new opportunity.
