Linking habitual vegetable intake and type 2 diabetes risk through metabolomic signatures

• Applicant: Dr. Catarina Schiborn
• Subject Area: Epidemiology and Medical Biometry/Statistics ; Nutritional Sciences
• Term: since 2025
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 572831361

Major guidelines recommend diets emphasizing the intake of vegetables for the prevention of type 2 diabetes (T2D). However, despite the overall beneficial nutritional profile of vegetables, epidemiological evidence suggests heterogenous associations with T2D risk across different vegetable groups. While leafy green vegetables have mainly been reported to be inversely associated, cruciferous vegetables appear to show a weak positive association with T2D risk. The mechanisms of these differential associations are largely unclear. We aim to gain mechanistic insights into the heterogenous associations of total vegetable and vegetable subgroup intake, namely leafy green vegetables, dark yellow vegetables, cruciferous vegetables, tomatoes, potatoes, other vegetables, by linking their habitual intake to metabolic signatures that may explain differences in the observed epidemiological associations between vegetable subgroups and T2D risk. We hypothesize that differential associations can be partially explained by differences in metabolomic signatures and that dietary modifications of habitual vegetable intake and its subgroups are associated with temporal changes in metabolic signatures. Specific Aim 1. Derive metabolic signatures that are associated with habitual vegetable intake and respective subgroups, compare them, and quantify their association with conventional blood biomarkers. Using subsamples from the NHS, NHSII and HPFS with available plasma metabolomics and dietary information, we will perform machine learning procedures to generate metabolic signatures that are associated with habitual vegetable intake and respective subgroups and compare them descriptively. We will additionally quantify their cross-sectional association with plasma concentrations of established conventional blood biomarkers in a subset of 37,544 individuals using multivariate linear regression. Specific Aim 2. Investigate if temporal changes in habitual vegetable intake and respective subgroups redict changes in derived metabolic signatures. Using data from 1850 female nurses from established nested case-control studies within the NHS with repeated metabolomics measurements, we want to determine with multivariable linear regression models if a change of habitual dietary intake of vegetables and subgroups is associated with a change in the respective metabolic signatures. Specific Aim 3. Investigate if derived metabolic signatures mediate the association between habitual vegetable intake and respective subgroups with T2D risk. In the data set previously used to derive metabolomic signatures, we will first quantify the prospective association of habitual vegetable intake and respective subgroups and their derived metabolic signatures with incident T2D over a follow-up time of more than 25 years using multivariable Cox proportional hazards regression and, second, assess if the derived metabolic signatures mediate the association of the respective vegetable group with T2D.

• DFG Programme: WBP Fellowship
• International Connection: USA
• Host: Professor Frank B. Hu, Ph.D.