Scientists Identify a Single Nucleotide Polymorphism (SNP) Responsible for Increased Risk of Diabetes

CAMBRIDGE, Mass. — Researchers at the Whitehead Institute have shown that a common genetic variant increases the risk of contracting type 2 diabetes. The variant, a single nucleotide polymorphism (SNP) in a gene called PPAR gamma, is carried by billions of people and helps to explain why some people are more likely than others to contract diabetes. The study, published in the September issue of Nature Genetics, has several implications: it offers new insights into the underlying causes of diabetes and more generally provides a blueprint for analyzing the role of SNPs in disease.

“By uncovering the genetic causes of diabetes, we can help understand the biologic basis of this common illness,” says researcher and endocrinologist Joel Hirschhorn, who co-authored the study at the Whitehead Institute Center for Genome Research. “Understanding in detail why diabetes occurs will hopefully point the way toward better and more specific treatments.”

The genetic variation—a very common, single-letter substitution in DNA—predisposes individuals to a modest but significant increased risk for type 2 diabetes. Because the risk allele is so common, the modest effects of this variant on an individual's risk translate into a dramatic effect on the population as a whole—influencing as much as 25 percent of type 2 diabetes in the general population.

Eric Lander, Director of the Whitehead Genome Center and senior author of the study, says that the study is relevant to a general understanding of the role of common genetic variation in disease. “In addition to the implications for diabetes, this study serves to illustrate the importance of common genetic variations in human disease. These variations may have a modest effect on individual risk but still affect a major portion of the human population,” says Lander. “The study also highlights the importance of identifying common DNA variations called single nucleotide polymorphisms, or SNPs, in human disease.”

The study also surmounted some of the difficulties inherent in studying SNPs and common disease. The PPAR gamma SNP had been previously analyzed in relation to diabetes, but because its effect on diabetes risk was modest, the earlier studies usually lacked adequate numbers of patients and inappropriately rejected a role for the PPAR gamma variant in diabetes. However, the Whitehead group analyzed DNA from multiple large groups of patients and was thus able to reliably detect the effect of this SNP on diabetes risk. Importantly, a metaanalysis revealed that the previous studies were all consistent with the Whitehead results, adding additional weight to the findings.

“Our results provide compelling evidence that this SNP plays a role in increasing risk for diabetes,” says David Altshuler, who along with Hirschhorn is first author on this study. Altshuler, like Hirschhorn, is a physician and endocrinologist working as a research scientist at the Whitehead Institute Center for Genome Research. “The association of a SNP in PPAR gamma with diabetes makes good biologic sense, since PPAR gamma is a target for thiazolidinediones, a class of medications currently used to treat diabetes.”

“Our results also have implications for the design and interpretation of genetic studies,” says Altshuler. “With the sequencing of the human genome and the discovery of large numbers of SNPs as part of the SNP consortium and other efforts, having proper guidelines for doing genetic studies with SNPs is critical.”

Most common diseases, like type 2 diabetes, are partly due to genetic variations in our DNA, to single letter substitutions in DNA called single nucleotide polymorphisms, or SNPs. Because any two humans are 99.9 percent similar genetically, the 0.1 percent difference in our DNA partly determines the traits that make us unique and also helps to explain why some of us are more susceptible to certain diseases than others. As a result, scientists have used comparisons of DNA in affected and non-affected individuals—patients with diabetes and healthy controls, for example—to determine the DNA variation that underlies risk for a disease. But such “association” studies—linking a gene to a disease—have been plagued by irreproducibility, and many turn out to be false positives or difficult to confirm.

In this study, the Whitehead scientists designed a multi-layered approach to control the confounding factors that can muddy genetic association studies. To eliminate a source of false-positive results called population stratification, they used a family-based approach, comparing affected and non-affected individuals within the same family (instead of comparing patients with and without diabetes from the random population). They tested 16 genetic variations that had previously been reported to be associated with type 2 diabetes or related conditions. They first analyzed these variants in parent-offspring trios with type 2 diabetes to eliminate false positives and identify the genetic variation variants that seemed to play a role in diabetes risk. To make sure that any positive results seen in the parent-offspring trios were real, they then double-checked these variations against a population of siblings where one had diabetes and the other didn't and against case-control pairs from two additional populations.

The Whitehead team found that of the 16 SNPs tested, three were rare or non-existent in their study populations. Of the remaining 13, 12 showed no discernible effect on the risk of diabetes. One variant turned out to have a reproducible effect—the SNP in PPAR gamma. Interestingly, this SNP, which encodes a substitution of alanine for proline, had previously been shown to affect the function of the PPAR gamma protein.

“Our results suggest a methodology for teasing out the genetic underpinnings of common disease while avoiding some of the problems that can confound genetic association studies,” says Hirschhorn. “First, they suggest that family-based controls and large, tiered samples need to be used to decrease false positive reports. Second, the previous difficulty demonstrating —that studies with modest sample sizes can fail to detect true associations. Third, although much can be learned from rare genetic variants, this example suggests that common variants with weak effects can have a large impact on the health of a population.”

Finally, say the researchers, despite the large population impact of common variations, their role in disease will be impossible to discover by linkage analysis. “In this case, for example, the genetic variation will typically be transmitted from both parents, requiring a genome scan of roughly 3 million sibling pairs to obtain a statistically significant result,” says Altshuler. As a result, genetic dissection of common diseases will surely involve association studies performed on large population samples.