It is the question that almost never gets raised in polite company, although almost everyone has an opinion about it. Often, in fact, two opinions, one public and one private. There are many ways of asking it, but perhaps the most neutral way is: what, if any, are the significant genetic differences between racial groups, besides the obvious superficial ones, such as skin color, that we use to define "race" in the first place?

That sentence took me a half hour to compose, and I hope I got it right, because this is as touchy a subject as there is in America, or most other countries with populations of mixed ancestry. For decades, it was virtually a closed question in academia, where the reigning scientific paradigm, expressed by Harvard geneticist Richard Lewontin in 1972, held that almost all the variation in the human genome represented differences between individuals. This notion fit well with the social consensus that the question was better not asked, because it could lead toward racial comparisons that almost invariably end badly for one group or another. And there the matter might have rested, but for the march of medical progress, which has been accumulating data on the differential health risks and drug responses of population groups.

The question arose most prominently in the case of a drug called BiDil, which the FDA approved in 2005 to treat heart failure in blacks after a study found that it relieved hypertension in self-identified African-Americans, but not other ethnic groups. If the study findings were correct, then BiDil, made by NitroMed, can prolong the life of many of the 750,000 black Americans with heart failure, but the very idea of marketing a drug that way provoked outrage among people who saw it as a validation of the concept of "race" as a genetic, rather than a social, category. Perhaps because of that controversy, as well as other reasons, including the drug's high cost, it was recently estimated that only 3 percent of the patients who might benefit from BiDil were actually getting it.

But now it turns out that BiDil may only be the first such drug. A recent study in The Pharmacogenomics Journal counted up nine clinical trials around the world studying diseases or treatments in groups defined by race or gender or both, including chronic hepatitis B in blacks and Hispanics and respiratory syncytial virus in Native American infants. A conference recently convened by Timothy Caulfield, a professor of health law at the University of Alberta, drew together medical ethicists, geneticists and legal scholars to debate these issues. The conference report, soon to be published in Genome Medicine, urges researchers to tread carefully in designing, carrying out and reporting studies that involve racial categories. The broad social construct of "race" is, Caulfield points out, biologically not very useful. "Someone whose ancestors came from Nigeria is very different from a descendant of Kenyans, but if the two of them are walking down the sidewalk in New York, they're both 'black'," he says. "You can try to make those distinctions in your research, but once it gets in the hands of drug manufacturers, there's going to be slippage … marketers want to sell to the broadest possible categories."

Fortunately, that very tendency runs counter to another significant trend in medicine. The research that has allowed us to parcel out racial differences by ancestry will eventually outstrip even those categories, and identify specific vulnerabilities and drug reactions in the genomes of individuals. We will no longer be white or Asian or African, or even Northern European, Ashkenazi, Japanese or East African; we will be who we are, each one of us. And the sooner we reach that point, the better.

© 2009