Circulating 25-hydroxyvitamin D, IRS1 variant rs2943641, and insulin resistance: replication of a gene-nutrient interaction in 4 populations of different ancestries

Abstract

Background: Associations of either insulin receptor substrate 1 (IRS1) variants or circulating 25-hydroxyvitamin D [25(OH)D] with type 2 diabetes (T2D) and insulin resistance (IR) are inconsistent. This study sought to determine whether circulating 25(OH)D modulates the association of a potentially functional variant at IRS1 (rs2943641) with insulin resistance.

Method: Interaction between IRS1 rs2943641 and circulating 25(OH)D on homeostasis model assessment for IR (HOMA-IR) was examined in the Boston Puerto Rican Health Study (BPRHS) (n = 1144). Replication was performed in the African-American (n = 1126), non-Hispanic white (n = 1967), and Hispanic (n = 1241) populations of the Multi-Ethnic Study of Atherosclerosis (MESA) with genotypes of 3 IRS1 variants, rs2972144, rs1515104, and rs2673142, which are tag single nucleotide polymorphisms (SNPs) and in strong linkage disequilibrium with rs2943641.

Results: Higher circulating 25(OH)D was associated with lower risk of T2D and IR in BPRHS women homozygous for minor allele rs2943641T. Consistently, in each of 3 MESA populations, HOMA-IR and insulin decreased more evidently with higher circulating 25(OH)D in women of the rs2943641TT genotype than in carriers of the major allele (rs2943641C). Metaanalysis indicated significant and consistent interactions between circulating 25(OH)D and IRS1 variants on HOMA-IR (log transformed) [pooled β = -0.008, 95% CI: -0.016 to -0.001, P interaction = 0.004] and insulin (log transformed) (pooled β = -0.006, 95% CI: -0.011 to -0.002, P interaction = 0.023) in 3065 women of the 4 populations.

Conclusions: Participants with different genotypes of IRS1 rs2943641 exhibit differential benefit from high circulating 25(OH)D for the reduction of insulin resistance and T2D risk. This gene-nutrient interaction, which appears to be limited to women, warrants further examination in randomized controlled trials of vitamin D supplementation.

Fig. 1. Circulating 25(OH)D interacts with IRS1 variant on HOMA-IR in women in 4 populations of different ancestries. Predicted values for HOMA-IR (both natural log-transformed) were calculated with regression models, adjusting for potential confounders in the female participants of the BPRHS and MESA Hispanic, African-American, and non-Hispanic white populations

With higher circulating 25(OH)D, predicted HOMA-IR decreased more evidently in the IRS1 variant minor allele homozygotes than the major allele carriers in each population. The P values of the interaction for each population were calculated in a general linear model including an interaction term [genotype by 25(OH)D] in the model, and the P values are shown in the respective panel. Number of participants in each population: BPRHS, n = 79 for rs2943641 TT carriers and n = 729 for C allele carriers; MESA Hispanic, n = 62 for rs2673142 CC carriers and n = 573 for G allele carriers; MESA African-American, n = 44 for rs2972144 TT carriers and n = 557 for C allele carriers; and MESA non-Hispanic white, n = 118 for rs1515104 AA carriers and n = 903 for T allele carriers.

Fig. 2. Circulating 25(OH)D interacts with IRS1 variant on fasting insulin in women in 4 populations of different ancestries. Predicted values for HOMA-IR (both natural log-transformed) were calculated with regression models, adjusting for potential confounders in the female participants of the BPRHS and MESA Hispanic, African-American, and non-Hispanic white populations

With higher circulating 25(OH)D, predicted insulin decreased more clearly in the IRS1 variant minor allele homozygotes than the major allele carriers in each population. For each population, the P values of the interaction were calculated in a general linear model by including an interaction term [genotype by 25(OH)D] in the model, and the P values are presented in the respective panel. Number of participants in each population: BPRHS, n = 79 for rs2943641 TT carriers and n = 729 for C allele carriers; MESA Hispanic, n = 62 for rs2673142 CC carriers and n = 573 for G allele carriers; MESA African-American, n = 44 for rs2972144 TT carriers and n = 557 for C allele carriers; and MESA non-Hispanic white, n = 118 for rs1515104 AA carriers and n = 903 for T allele carriers.

Fig. 3. Interaction of circulating 25(OH)D, as a categorical variable, with IRS1 variant on HOMA-IR in BPRHS and MESA Hispanic women. In BPRHS, minor T homozygotes of rs2943641 had higher HOMA-IR than C carriers only when circulating 25(OH)D was higher than the population median (P = 0.036)

In the MESA Hispanic, minor C homozygotes of rs2673142 had higher HOMA-IR than G carriers only when circulating 25(OH)D was higher than the population median (P = 0.039). The P values of the interaction were calculated in a general linear model by including an interaction term [genotype by 25(OH)D] in the model. Number inside the bar indicates the number of subjects in that group. Values are means ± SE.

Fig. 4. Metaanalysis of the interaction between IRS1 variant and circulating 25(OH)D on HOMA-IR in 4 populations of different ancestries. Random-effects model was used to obtain the overall effect estimate

Gray square stands for study-specific β, with the square size reflecting the corresponding weight and horizontal bars reflecting 95% CIs.

Fig. 5. Metaanalysis of the interaction between IRS1 variant and circulating 25(OH)D on fasting insulin in 4 populations of different ancestries. Random-effects model was used to get the overall effect estimate

Gray square stands for study-specific β, with the square size reflecting the corresponding weight and horizontal bars reflecting 95% CIs.