5 Environmental Factors Linked to Type 2 Diabetes Using Novel Approach

Jacquelyn K. Beals, PhD

May 21, 2010

May 21, 2010 — A new study is borrowing the methodology of genome-wide association studies (GWAS) to investigate associations between environmental substances and type 2 diabetes (T2D). Published online May 20 in PLoS ONE, the investigation found and validated 5 environmental substances whose odds ratios (ORs) regarding T2D are comparable to those previously reported for genetic loci.

"We really initiated this kind of effort to put the environment on an equal footing with the genetics world," said senior author Atul J. Butte, MD, PhD, assistant professor, Department of Pediatrics and Medicine and Stanford Center for Biomedical Informatics Research, Stanford University School of Medicine, California, in a telephone interview with Medscape Diabetes & Endocrinology.

"With all the press and everything with the Human Genome Project over the past decade, a lot of investigators certainly use genetic tools to study the role of genes and variation in DNA with human disease. But the largest component is always going to be the environment. We have an increasing level of, for example, T2D and obesity. Our genes aren't changing that fast — it's the environment," Dr. Butte explained.

The investigation, termed an environment-wide association study (EWAS), used data sets from 4 cohorts from the Centers for Disease Control and Prevention's National Health and Nutrition Examination Survey (NHANES): 1999 to 2000, 2001 to 2002, 2003 to 2004, and 2005 to 2006. Because the 4 cohorts were analyzed individually, the number of environmental factors varied among cohorts (111, 146, 211, and 75, respectively).

Overall, 266 specific environmental factors were present, with 157 of them present in 2 or more cohorts. The study looked for significant association between factors such as levels of heavy metals or pesticides in blood or urine and cases of T2D, defined as fasting blood glucose of 126 mg/dL or greater. The study's criterion for significant association was P ≤ .02.

Characteristic GWAS significance values are P < 5 × 10−8. Medscape Diabetes & Endocrinology asked the authors whether the P values in the EWAS study reflected the relatively small number of environmental variables analyzed compared with the numerous single nucleotide polymorphisms in GWAS.

"That is correct," said lead author Chirag J. Patel, in an email. "Our higher P values are also related to smaller cohorts of individuals used in comparison to a typical GWAS," he explained. Mr. Patel, a graduate student from the Department of Pediatrics and Medicine and Stanford Center for Biomedical Informatics Research, Stanford University School of Medicine, is credited in a Stanford press release with conceiving, designing, and executing the analytic software for EWAS.

Stringent Validation of Environmental Factors

GWAS include a replication phase for validation. In the present study, environmental factors significantly associated with T2D in the initial analysis were validated if significant T2D association was found in 2 or more cohorts. "In some ways, it's a little bit more stringent than what they do for GWAS," said Dr. Butte. "They'll do a chip kind of approach for 1 cohort, and then they'll just take a couple of those hits and validate in a larger second cohort. Here, we're basically doing screening across independent cohorts, and really only talking about those that are positive in 2 or more."

The present study found 5 environmental factors significantly associated with T2D in 2 or more NHANES cohorts. The T2D-associated factors were cis-β-carotene, trans-β-carotene, γ-tocopherol, heptachlor epoxide (an oxidation product of the pesticide, heptachlor), and polychlorinated biphenyl PCB170.

"γ-tocopherol [a form of vitamin E] has a lot of different functions," said Dr. Butte. "Think about these positive hits the same way [as] positive hits from GWAS.... When you get a positive hit, you're not 100% sure it's exactly that spot. For GWAS, it's a cue to look around there to do deep sequencing to find the causal factor. Even for γ-tocopherol, there are so many kinds of forms of it: it's activated, it's oxidized, methylated.

"We see this association [of T2D] with γ-tocopherol. It could be the raw γ-tocopherol, it could be even the stuff that's actually ingested, or it might be some particular form of it, or breakdown product.... We should study this to see exactly what the factor is and, of course, whether it's causal or not, or is it merely associated," Dr. Butte observed. Analysis of combined 1999 - 2006 NHANES cohorts showed a significant detrimental association between γ-tocopherol and T2D (P < .001; OR, 1.5; 95% confidence interval [CI], 1.3 - 1.7).

A preventive association was found between the β-carotenes and T2D, confirming results of earlier studies. The combined NHANES 2001 - 2006 cohorts demonstrated significant associations of both cis- and trans-β-carotenes with T2D (for both forms, P < .001; OR, 0.6; 95% CI, 0.5 - 0.7). Nevertheless, past attempts to use these vitamins in T2D therapy have been largely ineffective.

Linking Toxins With Disease and Phenotype

"We've known about the toxins and the level of the NHANES studies for all these years...but it's a continual frustration as to what this data means," commented David Ewing Duncan, director of the Center for Life Science Policy, and visiting researcher at the Graduate School of Journalism, University of California–Berkeley, in a telephone conversation with Medscape Diabetes & Endocrinology. "What Atul and his team have done is really rather brilliantly come up with a method for at least taking the known information and beginning to link these toxins with disease and phenotype.

"The other interesting thing is that [Dr. Butte] is speaking to geneticists and other molecular biologists in their own language," said Dr. Duncan. "It's basically an environmental story, but it's spoken in a language that these other scientists will understand and hopefully will begin to link together the molecular biologists and people that deal with environmental toxicity."

Dr. Butte is already looking ahead: "We have tons of biotech [companies] in this area that focus on endogenous targets like proteins and receptors. Why can't we think about pharmaceutical companies someday targeting these environmental factors?" he asked. "I can totally imagine a small molecule someday that might block the effect of this pesticide. Pure speculation, but we should be thinking along those lines."

Dr. Butte has disclosed no relevant financial conflicts of interest with the present work but has been scientific advisor and/or consultant to NuMedii, Genstruct, Prevendia, Tercica, Eli Lilly and Company, and Johnson and Johnson. Mr. Patel and Mr. Duncan have disclosed no relevant financial relationships.

PLoS ONE. Published online May 20, 2010.