Background Type 2 diabetes (T2D) is a major risk factor associated with cardiometabolic diseases, and a major contributor towards mortality and morbidity, given its rapidly rising prevalence worldwide. In experimental studies, trans-fatty acids (TFAs) exert harmful biologic effects that may affect T2D risk, but findings from observational studies remain inconclusive, especially for biomarkers which provide an objective advantage with less recall bias and estimation errors. By pooling multiple studies, we may also increase generalizability, statistical power, and address potential interactions by subgroups. Therefore, we assessed prospective associations between circulating biomarkers of individual TFAs and incident T2D in a large, diverse sample.

Methods We pooled ten prospective cohort or nested-case-control studies from Australia, Germany, Iceland, UK, and the USA to perform an analysis using harmonized individual level data for TFA biomarkers and incident T2D. Fatty acids (FAs) were measured in plasma phospholipid, red blood cell membrane phospholipid, or total plasma collected between 1990–2008 from 22,711 participants aged ≥18 years without prevalent diabetes. Evaluated TFAs included trans-16:1n-9, sum of trans-18:1 isomers (trans-18:1n6 to trans-18:1n12), sum of trans-18:2 isomers (cis/trans-18:2, trans/cis-18:2, trans/trans-18:2), and individual trans-18:2 isomers. The multivariable-adjusted relative risk or odds ratio was estimated for each cohort by lipid compartments using a pre-specified protocol for definitions of exposures, covariates, and outcomes for statistical analysis. Association estimates were pooled using fixed-effects inverse-variance weighted meta-analysis.

Results During an average maximum of 14 years of follow-up, 2,244 cases of incident T2D were identified. Median levels of TFAs across cohorts were 0.05–0.18% total FAs for trans-16:1n-9, 0.09–2.05% for total trans-18:1, 0.10–0.73% for total trans-18:2, and 0.01–0.36% for individual trans-18:2 isomers. In overall pooled analysis, TFAs evaluated per inter-quintile range were not significantly associated with risk of T2D. Relative risks for individual TFAs were 1.02 (0.78–1.32) for trans-16:1n-9, 0.92 (0.79–1.08) for total trans-18:1, 1.16 (0.98–1.37) for trans/trans-18:2, 0.98 (0.79–1.21) for cis/trans-18:2, 0.93 (0.76–1.14) for trans/cis-18:2, and 0.90 (0.78–1.04) for total trans-18:2. Findings were consistent when TFAs were assessed categorically by study-specific quintiles, and when associations were pooled within lipid compartment (phospholipids or total plasma).

Conclusion We found that biomarker levels of TFAs were not significantly associated with risk of incident T2D in this international pooling project. Findings may reflect no effect of circulating TFA on T2D or be influenced by mixed TFA sources (industrial or ruminant), or to a general decline in TFA exposure during this period. Associations with T2D for higher levels of TFA biomarkers should be investigated.