Direct-to-consumer genetic testing -- again

An opinion piece published in Science on Oct 8, entitled "Regulating Direct-to-Consumer Personal Genome Testing" calls for the Food and Drug Administration to enact a multi-tiered policy for regulating DTC genetic tests.  The strength of regulation would be determined by the level of risk for a given test, determined by the kinds of decisions test results might lead the consumer to make.  The determination of what kind of earwax you have based on your DNA, for example, would be considered essentially risk free while breast cancer testing might be high risk, since you might decide to have a double mastectomy if the results suggest that your chances are great.  High risk tests would require pre- and post-testing counseling by qualified professionals.

In addition,

All tests should be analytically valid (able to accurately and reliably measure what they say they are measuring), and any clinical claims made about the test must be accurate and substantiated. Safety and effectiveness data should be developed, but for many tests, this should be done through enhanced postmarket surveillance and clinical studies, rather than a more stringent premarket approval process. Premarket assessment would focus on identifying tests with potential for egregious harms (e.g., tests with uncertain validity or utility that could profoundly alter the course of medical treatment) and keeping those tests off the market until further studies show an acceptable risk-benefit ratio.

The idea of regulating these tests for the benefit of the consumer is all well and good, but as we've said before, the idea that regulation is based on ensuring the accuracy of risk prediction is predicated on the assumption that such predication can even be accurate.

Some genetic testing is highly accurate.  That would be prenatal tests for single gene diseases of childhood.  And the field of genetic counseling is successfully based on these kinds of risk assessment.  But that's not what the DTC genetic testing companies are selling.  Earwax genes, may well be harmless recreation but DTC's are primarily selling predictions of the chances you'll get things like type 2 diabetes, or heart disease, or age-related macular degeneration or Parkinson's disease.  These are conditions that have late ages of onset, and are most likely due to a combination of genetic and environmental factors.

And, it turns out that risk estimates for particular genes routinely vary wildly depending on the study, the study population, environmental exposures and other factors.  In addition, risk estimates for specific genes are generally quite low.  That is, they don't raise the risk of disease by very much.

There's another very important point here, too, that nobody wants to hear.  We know from extremely well documented data that risks of most of the relevant diseases (diabetes, cancer, heart disease, etc.) have increased greatly in living memory.  That's environment, not genes!  Even breast cancer risk fluctuates, even for people with the high risk alleles, depending on when a carrier was born.  Genetic effects depend on context.  Yet knowing as we do that the effect of context can change we know that we don't know the future environmental exposures -- and hence the future risks associated with the same genotypes.  Many or most DTC risk estimates are, quite literally, passé.

So, again as we've said before, regulation of what these companies is selling is a legal issue that must be thrashed out, but probably even more important is the epistemological issue of whether what they're selling means anything at all.  It's in nobody's interest -- except the consumers -- to acknowledge, much less deal with this serious issue.