Novel type 2 diabetes prediction score based on traditional risk factors and circulating metabolites: Model derivation and validation in two large cohort studies

Importance The predictive value of metabolomics quantified by nuclear magnetic resonance (NMR) for type 2 diabetes risk has not been assessed before. In addition, previous studies with other metabolomics quantification methods did not have an external validation cohort, which leaves doubts about the robustness of the derived models in other settings. Objective This project aimed to evaluate the incremental predictive value of metabolomic biomarkers for assessing the 10-year risk of type 2 diabetes when added to the clinical Cambridge Diabetes Risk Score (CDRS), which includes HbA 1c . Design, Setting, and Participants We utilized 60,362 participants of the UK Biobank (UKB) for model derivation, 25,870 participants of the UKB for internal validation and 4,383 participants from the German ESTHER cohort for external validation. Exposures A total of 249 NMR-derived metabolites were quantified using nuclear magnetic resonance (NMR) spectroscopy. Main Outcomes and Measures The main outcome was 10-year type 2 diabetes incidence. Results Eleven metabolomic biomarkers, including glycolysis-related metabolites, ketone bodies, amino acids, and lipids, were selected with LASSO regression. In internal validation within the UKB, adding these metabolites Harrel’s C-index of the clinical CDRS from 0.815 to 0.834 ( P <0.001) and the continuous net reclassification index (NRI) was 39.8% ( P <0.001). External validation in the ESTHER cohort showed a comparable C-index increase from 0.770 to 0.798 ( P <0.001) and a continuous NRI of 33.8% ( P <0.001). Conclusions and Relevance Adding 11 biomarkers, mainly from glucose and lipid metabolism, to the clinical CDRS led to a novel type 2 diabetes prediction model, the “UK Biobank Diabetes Risk Score” (UKB-DRS), which substantially outperformed the clinical CDRS. As only very limited clinical information and a blood sample are needed for the UKB-DRS, and as high-throughput NMR metabolomics are becoming increasingly available at low costs, this model has considerable potential for routine clinical application in diabetes risk assessment. Key Points Question Can the inclusion of metabolites measured in blood samples with NMR spectroscopy improve the accuracy of the clinical Cambridge Diabetes Risk Score (CDRS), which is already a good prediction model including the main diabetes risk indicator HbA 1c ? Findings The novel UK Biobank Diabetes Risk Score (UKB-DRS), which includes traditional diabetes risk factors and 11 metabolites, demonstrated significantly enhanced predictive performance compared to the clinical CDRS, both in the UK Biobank and the German ESTHER cohort. Meaning The novel UKB-DRS could significantly improve the validity of early identification of individuals at risk for type 2 diabetes and guide clinicians and people at risk of diabetes in the choice of preventive measures.