The claims of DNA testing companies such as those I describe here are about to get some much-needed scrutiny. The House Committee on Energy and Commerce, chaired by Rep. Henry A. Waxman, has just sent letters to the big three (23andMe, Navigenics, and Pathway Genomics) asking for more information than the companies have in the past been willing to divulge. It will be interesting to see what kind of response the congressmen get, but one committee source is already expecting pushback from the companies' lawyers.
Waxman's interest was piqued by the move—quickly rescinded last week after the FDA objected—by Pathway to sell its DNA-collection kits in Walgreen's drugstores. The letters ask the companies for information on, among other things, how they analyze test results to determine someone's risk for any disease or drug response, and how accurately the DNA tests identify genetic risks. The two are obviously related, but there is a key difference. A DNA test can be perfectly accurate in identifying that someone has a particular genetic variant, and correctly say that this variant has been linked (in published studies) to a certain disease. But the next step is what matters to consumers: extrapolating from the presence of those genetic variants to assess the risk of disease.
Each company does that somewhat differently; the details of the algorithms they use are closely guarded. There are some red flags that customers may not be getting much for their hundreds of dollars (the DNA analyses typically cost $499), however: 23andme, for instance, offers a test for a gene variant that makes people susceptible to serious side effects from warfarin (a.k.a. Coumadin), a blood thinner. Yet published studies have thrown so much doubt on this that Medicare last year ruled that it will not cover the genetic test, since the test does not improve patient outcomes. As I described in my column, research has also cast doubt on whether the presence of other "disease genes" produces a more accurate assessment than traditional factors such as family medical history. Navigenics has a prominent page on its Web site arguing specifically that family history isn't enough. In fact, for some diseases—Alzheimer's being the most recent—the use of disease genes (with the exception of one rare one, called apoE4, in Alzheimer's) does not make predictions of whether someone will develop a disease any more accurate.
The committee's letters to the three companies (available here) ask for "internal and external communications regarding the accuracy of your testing," as well as "third-party communications validating the association between the scientific data your company uses for analyzing test results and the consumer's risk for each condition, disease, drug response, or adverse reaction." The companies have until June 4 to respond. A committee source says hearings are possible. If they happen, and if the companies are compelled to reveal and justify how they assess disease risk, consumers may finally learn whether DNA tests live up to all the hype.