It’s been a busy week for genetic oversell.
Benjamin et al. examined the genetic architecture of economic and political preferences in PNAS. They conclude that “molecular-genetic-based heritability estimates…partially corroborate evidence of significant heritability from behavior genetic studies,” i.e., there is some evidence from molecules that there is some evidence from behavior genetics that political preference is genetic. And also, their analyses “suggest that these traits have a polygenic architecture, with the heritable variation explained by many genes with small effects,” i.e. that although behavior is mediated by proteins and therefore somehow involves genes, there is no single gene for being a bleeding liberal or an evangelical Republican.
Chris Mooney discussed the paper, recognizing that “finding a single ‘liberal gene’ or ‘conservative gene’ just ain’t gonna happen. The studies that we’ve heard about so far focusing on individual “political” genes may be suspect, and much larger samples will be needed. In all likelihood, there isn’t going to be any one gene that does a lot of work to explain ideology–the phenomenon is going to be polygenic…and very tough to study.”
All true, but we need to go deeper than this. Since the middle of last century it has been clear that there is no single gene for any complex behavioral trait. Granted, the headlines continue to scream this stuff (see below), but we need to go beyond refuting simple genetic determinism. The larger question is why the genetics of politics gets so much play, when all it concludes ultimately is that lots of genes plus lots of other stuff go into complex qualities of temperament.
A minor dust-up over personal genomics raised similar issues. Over at The Awl, Russell Brandom wrote a trenchant critique of personal genome tests. He points out that items on a personal genome screen such as, “Your odds of heroin addiction are 2.9X higher than average” are hard to interpret. Does he get a bigger than average reward from opiates and so is more likely to keep doing them if he tries them? Or does he get less reward, and so would have to do more and more to keep the sensation?
He points out that a large percentage of a low risk is still a low risk. Going from a 2.4% risk to a 3.2% risk doesn’t mean much for the average person. The public health geneticist Tony Holtzman has been collecting data and making this argument since the early days of PKU, sickle cell, and Tay Sachs screening in the 1970s and 1980s, and it’s still true.
Complex traits have many contributing factors, each of which contributes a tiny fraction of the total variation. No one argues this, not even 23AndMe. However, the fashionable emphasis on genes—the dizzying fact that we can now identify a gene that contributes, say, to IQ, or to left-handedness, or for crying out loud to a preference for pork, enhances the genetic component, making it seem more important.
“They don’t have any straightforward truths to offer users,” Brandom observes, “just a flood of vague and often frightening ambiguities. But wherever there’s a gap between what they promise and what they can deliver, 23andme fills it with social media.” All the Facebook- and Foursquare-style features are designed, of course, to keep users on the 23AndMe site, to create a sense of identity and community.
There’s nothing illegitimate about that, but there is something insidious. To the extent that one buys in to this scene, one enters Gene World, where your genes are You. Your identity is determined by your DNA. The complexity of gene-environment interactions that every sane person acknowledges is stripped away, and we slide back to the genetic equivalent of the Ptolemaic earth-centered cosmos: a simplistic, gene-centered universe that seems to account for more than it does.
Misha Angrist responded to Brandom with an impassioned and well-informed defense of genetic testing, in which he pointed out ways in which he has indeed gotten useful information from his genome. He brings up one of the best examples: BRCA1. No argument that knowing your BRCA1 status is useful; however, for cost and other reasons it is not recommended to have yourself tested for BRCA1 unless you have a family history of breast cancer. However, Angrist’s argument that some genetic testing is medically useful does not really refute Brandom’s central claims, that DTC testing risks over-simplifying and overselling the genetic basis of disease and personality.
Razib Khan then totted up the score and decided that personal genomics won this round. It’s not clear why. He emphasizes that a “non-trivial minority of people receive actionable information” from personal genomics—although by his own estimate, 90–99% of those who undergo thise sort of testing will not receive useful medical information—and that the recreational side of personal genomics deserves to be explored. Again, his points are valid but not salient as criticisms of Brandom’s piece.
The comments under Khan’s brief post are more interesting. Several readers seemed to put their arms round each other’s shoulders and say, “Why is everyone always hating on personal genomics?” They saw no reason to do so unless you were an anti-determinist, which they used the way an evangelical uses “pro-choice.” They suggest that skepticism of personal genomics reflects a lack of curiosity and even, bizarrely, a lack of interest in participating in scientific research. One reader suggested that authors such as Brandom who are skeptical of DTC testing are “wedded to a Blank Slate view” and “ideologically oppose any form of genetic determinism.”
It is, rather, the ideology of genetic determinism that Brandom and I and others are concerned about: the troubling attitude that anyone who is wary of some of the implications of molecular soothsaying must be an extreme environmental determinist. Paired with a refusal to consider ways in which, unchecked, the profit motive may corrupt the dissemination of medical data, this is a dangerous view.
There are good reasons to be skeptical of DTC genome testing, and those reasons have nothing to do with one’s general support of science, one’s curiosity, or one’s interest in genetics. DTC genome testing is a large and potent social experiment. It has already brought some benefits, and we can expect it to bring more. However, it also has associated risks, which can and sometimes do mitigate those benefits.
Let me just take two examples, which in my mind encompass many of the smaller criticisms and risks. First, there is a creeping genetic determinism in much of the information on the relationship between genes and disease and behavior. This occurs at several levels. First, it’s worth contemplating why papers like the Benjamin et al. article even get published. There’s a man-bites-dog quality to them. Consider the converse, a paper that purported to show that political and economic preferences had a cultural basis. Ho-hum. But if you find a gene that correlates with such a complex “trait” you have a paper in PNAS. In other words, novelty creates a kind of ascertainment bias that makes it easy to think genes are more important than they are. A slight lapse in one’s critical thinking is all it takes to go from that paper to thinking, “Political and economic preferences just are genetic.” That’s a risk.
That risk is magnified by the publication process, from article authorship all the way to headline-writers. At each step, the contrast of the conceptual image is enhanced, in order to attract the notice of readers. Eventually, you end up with blog posts entitled, “Your IQ depends on a single gene,” and “Can a single pill save your marriage?” While everyone up to speed on genetics understands it doesn’t no, it can’t—that many genes and many environmental variables contribute to any complex biological trait—that complexity tends to get lost when the information is translated into accessible language.
Or consider 23AndMe’s page on “variation.” Using physical performance as an example, they begin modestly:
What does it take to become an Olympic champion? Hard work and perseverance are no doubt part of the equation, but DNA may play a role too.
Nothing to argue with there. The next paragraph introduces ACTN3, alpha-actinin skeletal muscle isoform 3, a protein associated with so-called “fast-twitch” muscle fibers. One form of ACTN3 has been correlated with stellar athletic performance. Its role in athletic performance, the 23AndMe page continues, “has led some to dub it the ‘gene for speed.’” Who is this “some”? Some scientists? Some journalists? Some nutjob bloggers? Some evidence.
At any rate, the page continues in a sweet vein, noting a long-jumper with two non-working copies of ACTN3 as a “testament to the fact that genes are not destiny.”
But the page concludes by asking, “Want to know if you have the ‘gene for speed’? 23andMe’s Health and Traits reports can tell you what versions of the ACTN3 gene you have in your DNA. You can also learn about how your genes may influence a variety of other traits, including your risk for certain diseases. Join 23andMe now!”
The potential customer is left with a clear if implicit message that there is a gene for speed and that for $299 you can find out if you have it. The discussion of variation and the importance of training and other factors is lost.
Where it gets lost, not surprisingly, is in the sales pitch, which brings up my second risk: commercialization. 23AndMe is not a charity. Although I have no reason to doubt the humanitarian impulses of anyone at the company, it’s no secret that their first allegiance is to their shareholders. Their primary task is making money. This of course does not immediately invalidate it as a socially beneficial enterprise, but it does and should make us alert to potential conflicts between profit and benevolence. As an advocate for patients, one feels responsible for raising questions and maintaining transparency so that the medical benefits Angrist rightly emphasizes don’t get lost in the commercialism Brandom exposes.
DTC testing is a fascinating, perhaps revolutionary development in medicine. It democratizes medical diagnostics, putting the first step in prevention and treatment into patients’ hands. It has the potential to empower patients, and thus to begin to counter the twentieth-century trend toward an ever more hierarchical healthcare establishment, in which objectification of the patient was justified on grounds of therapeutic potency. DTC testing may be one of those developments that medical historians a century hence will note as pivotal, the beginning of a new model of more participatory healthcare.
Advocates of DTC testing need to address concerns about hype, creeping genetic determinism, and commercialization in order for DTC testing to flourish. Waving away critics as anti-intellectual, anti-science, or blindly ideological merely deepens concerns about overselling this potentially transformative technology.