In-depth profiling of proteins in plasma can provide valuable insights on the complex processes related to metabolic diseases such as type 2 diabetes (T2D). Within the framework of the EU IMI project DIRECT (www.direct-diabetes.org) a multi-center cohort was built on 3100 subjects of which 2300 were at risk of developing of developing T2D (HbA1c~ 5.6 - 6.5%) as well as 800 with early T2D (HbA1c> 6.5%) .
We used several multiplexed affinity proteomic assays to profile ~600 unique proteins in EDTA plasma collected from 3100 study participants at baseline and 2500 at the 3 years (early T2D) or 4 years (risk for T2D) follow-up. With access to extensive metadata, our initial analysis focused on possible sample-related confounders. This identified several pre-analytical variables and consequently, we applied a linear mixed model that included age, sex, study center and sample collection date for defining proteins associated with any of the >50 quantitative clinical traits.
At baseline, we found > 300 proteins in plasma that were associated with diabetes related traits (adjusted p-value < 0.0001), many of which were prominently associated with BMI, such as leptin. Further, IGFBP1 and IGFBP2 associated to Matsuda; adiponectin to basal insulin secretion rate and fasting HDL; LDL receptor proteins to fasting triglycerides; APOM to fasting cholesterol; or IL8 and MCP-1 to fasting liver AST. Making use of other omics data, we performed pQTL analysis to assess any connection between the protein values in plasma and genetic variants. We observed ~400 cis-pQTLs (q-value < 0.05), such as for APOM, which illustrated that many of the studied protein profiles are affected by a genetic component.
Our integrative, large-scale multi-omics analysis revealed insights about known and novel plasma proteins associated to pre- and early T2D, as well as indicators of progression and treatment response.