Erythrocyte membrane docosapentaenoic acid levels are associated with islet autoimmunity: the Diabetes Autoimmunity Study in the Young

Abstract

Aims/hypotheses: We previously reported that lower n-3 fatty acid intake and levels in erythrocyte membranes were associated with increased risk of islet autoimmunity (IA) but not progression to type 1 diabetes in children at increased risk for diabetes. We hypothesise that specific n-3 fatty acids and genetic markers contribute synergistically to this increased risk of IA in the Diabetes Autoimmunity Study in the Young (DAISY).

Methods: DAISY is following 2,547 children at increased risk for type 1 diabetes for the development of IA, defined as being positive for glutamic acid decarboxylase (GAD)65, IA-2 or insulin autoantibodies on two consecutive visits. Using a case-cohort design, erythrocyte membrane fatty acids and dietary intake were measured prospectively in 58 IA-positive children and 299 IA-negative children.

Results: Lower membrane levels of the n-3 fatty acid, docosapentaenoic acid (DPA), were predictive of IA (HR 0.23; 95% CI 0.09, 0.55), while α-linolenic acid (ALA), eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) were not, adjusting for HLA and diabetes family history. We examined whether the effect of dietary intake of the n-3 fatty acid ALA on IA risk was modified by fatty acid elongation and desaturation genes. Adjusting for HLA, diabetes family history, ethnicity, energy intake and questionnaire type, ALA intake was significantly more protective for IA in the presence of an increasing number of minor alleles at FADS1 rs174556 (pinteraction = 0.017), at FADS2 rs174570 (pinteraction = 0.016) and at FADS2 rs174583 (pinteraction = 0.045).

Conclusions/interpretation: The putative protective effect of n-3 fatty acids on IA may result from a complex interaction between intake and genetically controlled fatty acid desaturation.

Fig. 1

Scheme of the metabolic pathway of n-6 and n-3 long-chain polyunsaturated fatty acids

Fig. 2

Description of analysis populations used in Study 1 and Study 2 from the DAISY

Fig. 3

The decreased risk of IA with increased ALA intake differs by number of FADS1 and FADS2 minor alleles in the case-cohort of DAISY. The analysis cohort consisted of 87 children who developed IA and 297 children who did not. HRs and 95% CIs represent the relative hazard for a one SD difference in dietary intake. The SD for ALA intake was 0.46. The intake variable was analysed as a time-varying covariate, and the model adjusted for HLA-DRB1*03/DRB1*04,DQB1*0302, family history of type 1 diabetes, ethnicity, total energy intake and FFQ type. We detected significant interactions between dietary intake and FADS1 (rs174556) (p=0.017), and FADS2 (rs174570 and rs174583) (p=0.016 and p=0.045, respectively). The HRs and CIs shown in the figure are calculated from the estimates produced by the adjusted models containing the interaction terms