Novel protein targets for type 2 diabetes prevention and their mediating role in the diet–disease relationship: evidence from 2 population-based cohorts and a Mendelian randomization analysis

Background
Circulating proteins are associated with type 2 diabetes (T2D) but causality has often not been investigated. Whether plasma proteins could be suitable targets for T2D prevention remains uncertain.
Objectives
We aimed to identify causal associations between plasma proteins and incident T2D and discover whether proteins may mediate associations between diet and T2D risk.
Methods
We measured 276 plasma proteins in 2 Swedish prospective cohorts. The discovery cohort consisted of 4103 females, and the replication cohort consisted of 1017 females and 4124 males. During follow-up, 189 females in the discovery cohort and 297 participants in the replication cohort developed T2D. Associations between proteins with incident T2D were estimated using Cox proportional hazards models. Mendelian randomization analyses using cis-protein quantitative trait loci were conducted to assess causal relevance, along with colocalization analyses to examine shared causal variants between proteins and T2D. We further assessed whether proteins could mediate associations between dietary patterns (DPs) derived from principle component analysis and incident T2D.
Results
Overall, 112 proteins were associated with incident T2D independent of age, sex and lifestyle factors in 2 cohorts. Genetically raised levels of tyrosine-protein kinase receptor TYRO3, intercellular adhesion molecule 2, sortilin (SORT1), and sialomucin core protein 24 were associated with increased T2D risk, whereas catechol O-methyltransferase, CD2-associated protein, matrix metalloproteinase-12, nectin-2 (NECTIN2), reticulon-4 receptor (RTN4R), and fibroblast growth factor 21 were inversely associated (false discovery rate-adjusted P < 0.05). Strong evidence (posterior hypothesis 4 of colocalization > 0.8) of shared genetic variants of NECTIN2, RTN4R, and SORT1 with T2D was found. Phenome-wide association investigation at the gene level confirmed associations of proteins with T2D traits. We identified a mediating role of proteins with genetic evidence in the association between a healthy DP (i.e., high consumption of vegetables, fruits, nuts, and whole grains) and T2D.
Conclusions
We provide novel and robust findings of proteins causally linked to T2D and their associations with habitual diet. These findings may guide nutritional strategies for T2D prevention.