The recent advent of genome-wide association studies (GWAS) has led to major advances in the identification of common genetic variants contributing to diabetes susceptibility (40
). To date, at least 40 genetic loci have been convincingly associated with type 2 diabetes, but these loci confer only a modest effect size and do not add to the clinical prediction of diabetes beyond traditional risk factors, such as obesity, physical inactivity, unhealthy diet, and family history of diabetes. Many diabetes genes recently discovered through GWAS in Caucasian populations have been replicated in Asians; however, there were significant interethnic differences in the location and frequency of these risk alleles. For example, common variants of the TCF7L2
gene that are significantly associated with diabetes risk are present in 20–30% of Caucasian populations but only 3–5% of Asians (41
). Conversely, a variant in the KCNQ1
gene associated with a 20–30% increased risk of diabetes in several Asian populations (43
) is common in East Asians, but rare in Caucasians. It is intriguing that most diabetes susceptibility loci that have been identified are related to impaired β-cell function, whereas only a few (e.g., peroxisome proliferator–activated receptor-γ, insulin receptor substrate 1, IGF-1, and GCKR) are associated with insulin resistance or fasting insulin, which points toward β-cell dysfunction as a primary defect for diabetes pathogenesis. It should be noted that most of the single nucleotide polymorphisms uncovered may not be the actual causal variants, which need to be pinpointed through fine-mapping, sequencing, and functional studies.
Despite heterogeneity across populations in risk allele frequency or effect size in type 2 diabetes genes, the combined effects of multiple genetic variants using genetic scores based on the number of risk alleles appear to be similar across different ethnic groups. Typically, each risk allele increment is associated with a 10–20% increased risk of type 2 diabetes (41
). These data suggest that the overall contribution of the identified genetic loci to type 2 diabetes is similar between Caucasians and other ethnic groups, and that these loci do not appear to explain ethnic differences in diabetes risk. In predicting future risk of diabetes, the clinical utility of these cumulative genetic risk scores appears to be limited in either high- or low-risk populations.
Like other multifactorial diseases, type 2 diabetes is a product of the interplay between genetic and environmental factors. It is likely that the genetic factors that underlie individual susceptibility are amplified in the presence of certain environmental triggers. On the other hand, given the same dietary and lifestyle factors, some individuals may be more prone to type 2 diabetes than others because of different genetic backgrounds. In the Health Professionals’ Follow-up Study, we found a significant interaction between a Western dietary pattern derived from principle component analysis of 40 food groups and a genetic risk score (GRS) of type 2 diabetes susceptibility based on 10 established single nucleotide polymorphisms (P
= 0.02) () (45
). The multivariable ORs of diabetes across increasing quartiles of the Western dietary pattern were 1.00, 1.23 (95% CI 0.88–1.73), 1.49 (1.06, 2.09), and 2.06 (1.48, 2.88) among men with a higher GRS (≥12 risk alleles, P
for trend = 0.01). Among those with a lower GRS, the Western dietary pattern was not associated with diabetes risk. In addition, intake of processed meat, red meat, and heme iron, which characterize the Western dietary pattern, showed significant interactions with GRS in relation to diabetes risk (P
for interaction = 0.029, 0.02, and 0.0004, respectively). These results suggest that the detrimental effects of a Westernized diet may be enhanced by greater genetic susceptibility.
Figure 4 ORs of diabetes risk according to joint classification of Western dietary pattern scores (in quartiles, Q) and genetic risk scores (<10, 10–11, and ≥12). ORs and 95% CIs were calculated by using an unconditional logistic regression (more ...)