Nicely done Sharon Begley piece in today’s WSJ on the false scientism inherent in many claims of a genetic basis for various diseases. As she points out, there have been analyses of purported links between genetics and diseases and the number of such associations that are later found to be wrong is disheartening. One British study from 2003 estimated that as many as 95% of such associations were, in a word, wrong:
The list of genes whose link to a disease has not held up is a veritable alphabet soup. ADD1 and hypertension; APOE and schizophrenia; BLMH and Alzheimer’s; COMT and bipolar disorder; GSTM1 and breast cancer; PON1 and coronary-artery disease . . . I could go on. For hundreds of other genes, there seems to be a much weaker link to the disease than first report suggested, with the genetic variation accounting for a tiny bit of the disease risk.
“On a good day, it might explain only 1% or 2% of the variance,” says Irving Gottesman of the University of Minnesota, an expert on the genetics of psychiatric illnesses who has long been skeptical of simplistic gene-disease links.
One reason for what Dr. Wacholder calls “this unfortunate situation” is how easy it has become to analyze genes. Time was, that was so difficult that geneticists would first identify, based on knowledge of a disease, which genes might play a role. When they found that a suspect gene indeed raised the disease risk, it meant something.
Now, though, reading genes is so quick, taking a day rather than months, that scientists can and do scan thousands. By chance alone, some of those will seem to raise the risk of disease in a statistically significant way, says Joel Hirschhorn of the Broad Institute at Harvard University and the Massachusetts Institute of Technology in Cambridge, Mass. “Statistically significant” means less than a one-in-20 chance that the link was coincidental and not real.
“If you test enough different genes for an association with enough different diseases, you’ll automatically get a ‘significant’ result,” he says. But that might be only a coincidence: Flipping 10 heads in a row doesn’t mean you have a two-headed coin.
Related posts: