Genetic variation sways risk of diabetes.

Scientists have found that a gene carried by up to 85 percent of the people in the world increases the risk of developing diabetes by about 25 percent.

The increase, compared with that of people having a subtly different form of the gene, is relatively modest but significant, say investigators who report their study in the September NATURE GENETICS. "Although [the gene] has a very

weak effect on the individual, because it is so common, it has a pretty big effect on the general population," notes coauthor David Altshuler of the Whitehead Institute for Biological Studies in Cambridge, Mass.

The gene, which encodes a protein called peroxisome proliferator-activated receptor-gamma (PPAR-gamma), seems to influence a person's chance of developing the most common form of diabetes. Known as noninsulin-dependent, or type II, diabetes, this disorder generally occurs when the body has trouble responding to the hormone insulin, which controls blood sugar.

PPAR-gamma, which is the target of a new class of diabetes drugs, governs the maturation and function of fat cells (SN: 12/9/95, p. 390). In 1997, Alan R. Shuldiner of the University of Maryland in Baltimore and his colleagues reported that the PPAR-gamma gene comes in two forms that spawn slightly different proteins. The more common version has the amino acid proline at a spot where the rarer version has the amino acid alanine.

Because it involves the alteration of just one DNA subunit, scientists call the variation in the PPAR-gamma gene a single-nucleotide polymorphism, or SNP. Investigators analyzing the human genome are compiling huge databases of such variations. One of these, at the National Institutes of Health, already contains more than 800,000 examples.

Scientists plan to use SNPs (pronounced "snips") to identify gene variants that influence disease susceptibility and to tailor drugs to a person's unique genetic heritage. There's a vigorous debate, however, about whether this strategy will work, notes Altshuler.

His group's work focused on 16 SNPs already tentatively associated with diabetes. In 1998, for example, Samir S. Deeb of the University of Washington in Seattle and his colleagues reported that people's diabetes risk is increased if they have the proline-encoding PPAR-gamma gene.

Such associations have proven to be notoriously difficult to reproduce, often because follow-up studies have focused on ethnic groups different from the ones in the original study. Moreover, if a gene's effect on disease susceptibility is modest, initial studies may have too few people to reveal the SNP's influence. A handful of studies of the polymorphism in the PPAR-gamma gene had in fact failed to confirm a diabetes connection.

Altshuler and his colleagues first analyzed their 16 SNPs in 333 parent-offspring trios in which the offspring has type II diabetes--or impending signs of the disease--but the parents don't. The results suggested that just two of the SNPs influence diabetes risk. When the researchers studied three other groups, encompassing more than 2,000 people, only the SNP within the PPAR-gamma gene consistently revealed an effect.

Altshuler's team also reanalyzed the earlier negative studies and found that they had hinted at the same effect. "When we looked at all of our data, and all of the previously published data, it actually told a very consistent story, which was that there was a weak effect of this variant," says Altshuler. Deeb calls this confirmation "gratifying."

The new analysis makes a compelling case for the polymorphism's effect, agrees Richard S. Spielman of the University of Pennsylvania Medical School in Philadelphia. While the 25 percent increase in diabetes risk may seem modest, there's no guarantee that researchers will find genes that have any bigger impact, he adds.