On the first day of a two-day workshop on direct-to-consumer genetic testing co-sponsored by the Institute of Medicine and the National Academy of Science, Muin Khoury of the CDC raised the following question: if we speak out against the current crop of micro-array based genome-wide screens. are we allowing the perfect to be the enemy of the good? An answer, suggested the generally skeptical panel, might be based on how strictly you defined the word “good.” Several speakers, including Dr. Khoury, were unimpressed by the predictive capacity of today’s SNP-based disease-gene associations. Despite some notable successes where specific alleles have been demonstrated to increase relative risk in a meaningful way (age-related macular degeneration was usually the example), the massive number of genes identified through GWAS since 2006 have been, as Alan Guttmacher of NHGRI stated in his talk, “great for understanding the biology of the disease, but weak predictors accounting for only a small fraction of heritability.”
Complaints fell largely into three categories: validity, utility and the public health risks of giving out this information, right or wrong. As regards validity, a number of speakers pointed out that these early data were largely unreplicated and that standards did not exist to define what constituted a meaningful level of “association.” Utility was usually invoked to question whether or not the test would add information that was clinically significant – would it affect medical decision-making, or provide a better indicator than simpler tests or the underutilized gold standard of family history (oh that again!). These are excellent questions, since most GWAS-based associations don’t move the dial much on risk – increases tend to be in the range of relative risks under two, which, in technical terms, means you are a smidge more – or less – likely to get the disease or condition. But cost is also a question of utility, since changes that are not affordable cannot be incorporated into medical practice. The cost of scanning continues to fall and a dizzying pace – and multiple panelists insisted that we will see complete genome sequencing for under $1000 by the end of 2010 – but what about the costs of follow-up? What pot of money will the insolvent U.S. medical system find to pay for added MRI’s, or blood tests, or doctor’s visits for patients whose SNP profiles suggest an increased risk or this or that? Reading the websites for, Navigenics, 23andme, and deCODE, you might come to believe that your doctor will thank you for bringing this useful information to his or her attention, but Dr. Patricia Ganz of UCLA Medical School, brandishing the twenty-five page report from 23andme sent to her by a patient, wondered aloud if the average physician would in fact appreciate that lovely extra time together, reading pie charts.
And what does all of this mean for genetics counselors? Joe McInerney, executive director of NCHPEG, spelled out the good and the bad of it in his talk on understanding among health professionals. Genetics, he predicted, will be the first field in medicine to be de-centralized: moved from the realm of specialists back into the world of primary care. Will we move from being educators of patients to being educators of health professionals? McInerney suggested that pressure from patients interested in understanding what to do with their $1000 genomes will push physicians to seek expertise in genetics. Other participants pointed out that the generalized field of genetic counseling may be asked to provide more in the way of genetic counselor specialists to meet demand as the walls between Mendelian disease and common complex disease come crumbling down in an avalanche of new information. Forget the term “genetic disease,” McInerney declared! They are all genetic diseases now – and in a world that he described as “woefully unprepared” for the era of genomic medicine now approaching with all the subtlety and control of a locomotive off the tracks, the question lingers: where will we find ourselves in this new landscape? In the absence of a perfect answer, are we prepared to define when “good” is “good enough”?