Composite Metabolic Index Evaluated for Predicting Diabetes and Arthritis Comorbidity

Researchers have developed and evaluated DAP5, a routine blood-based composite metabolic marker designed to predict the co-occurrence of type 2 diabetes and arthritis, reporting that the tool effectively stratifies risk across two distinct ethnic cohorts. Type 2 diabetes and arthritis frequently coexist and are thought to interact bidirectionally through metabolic inflammation, yet simple indices to predict their co-occurrence have remained scarce. In an investigation published in BMC Endocrine Disorders, the five-factor index demonstrated consistent discriminative capacity in Chinese and English cohorts of middle-aged and older adults, while highlighting crucial limitations regarding external threshold calibration and arthritis-specific prediction.

The model development cohort drew from the China Health and Retirement Longitudinal Study (CHARLS) spanning 2011 to 2018, analyzing 5,733 disease-free middle-aged and older participants. To construct the Diabetes-Arthritis Prediction Factor 5 (DAP5), investigators pre-specified five routine biological markers selected on pathophysiological grounds: glycated hemoglobin (HbA1c), fasting blood glucose (FBG), triglycerides (TG), body mass index (BMI), and high-density lipoprotein cholesterol (HDL-C). The independent associations of these candidate indicators were verified through univariate regression and receiver operating characteristic (ROC) analysis prior to incorporating them into the composite index.

To evaluate the predictive performance and mathematical robustness of DAP5, the researchers utilized multivariate logistic regression, restricted cubic splines, and bootstrap resampling. Within the full CHARLS development cohort of 5,733 participants, the unadjusted area under the receiver operating characteristic curve (AUROC) for DAP5 was 0.723, with a 95% confidence interval of 0.676 to 0.770. This discrimination was comparable to the strongest single constituent marker, glycated hemoglobin, which recorded an AUROC of 0.725. The score was numerically higher than fasting blood glucose, body mass index, triglycerides, and high-density lipoprotein cholesterol, while demonstrating comparability to the established Metabolic Score for Insulin Resistance (METS-IR) with a p-value of 0.219.

The study authors clarified that DAP5 was not superior to glycated hemoglobin alone regarding predictive discrimination, noting that its practical value lies in integrating five routinely available clinical indicators into a single risk-ranking measure rather than achieving enhanced discrimination. Following multivariable adjustment, the index achieved an AUROC of 0.744 in the development cohort. After controlling for confounding variables including age, sex, and lifestyle, each one-unit increment in DAP5 was associated with significantly higher odds of diabetes–arthritis comorbidity, yielding an odds ratio of 5.933 (95% CI: 3.978–8.831, p < 0.001).

Cross-ethnic external validation was carried out in the English Longitudinal Study of Ageing (ELSA), which included 4,489 independent participants. For head-to-head comparison, researchers also developed an Extreme Gradient Boosting (XGBoost) machine learning model. In the ELSA validation cohort, the DAP5 index achieved an AUROC of 0.755, outperforming the XGBoost model, which achieved an AUROC of 0.670. While this provided preliminary evidence for the generalizability of the core metabolic indicators across ethnic groups, the pre-specified diagnostic cut-off was not supported in the external ELSA cohort, producing a sensitivity of 0% and demonstrating that cut-off values require further evaluation before broader application.

The findings also indicated that DAP5 does not function as an independent predictor of arthritis in the absence of metabolic disturbance. Among participants who were free of arthritis at baseline, DAP5 was not associated with incident arthritis, demonstrating an odds ratio of 1.04 (95% CI: 0.87–1.24, p = 0.643) and an AUROC of 0.582. An exploratory mediation analysis—noted by authors as unstable due to a low event rate—was consistent with the association operating mainly through incident diabetes, though the indirect effect itself was not statistically significant, against a total effect odds ratio of 5.86 (95% CI: 3.89–8.82). The investigators concluded that the composite index primarily captures the shared dysglycaemic substrate underlying both diseases rather than acting as an arthritis-specific predictor.

Subject of Research: Composite metabolic index for predicting diabetes-arthritis comorbidity

Article Title: DAP5: a composite metabolic index for predicting diabetes–arthritis comorbidity—development, external validation, and comparison with machine learning

Article References: Xu, Z., Zhang, Y., Zhu, K., Liu, X., Hu, C., Dong, X., Yang, F., Tang, L., Qin, K., Shen, J., & Zhou, Y. (2026). DAP5: a composite metabolic index for predicting diabetes–arthritis comorbidity—development, external validation, and comparison with machine learning. BMC Endocrine Disorders. https://doi.org/10.1186/s12902-026-02720-1

Image Credits: AI Generated

DOI: 10.1186/s12902-026-02720-1

Keywords: DAP5, type 2 diabetes, arthritis, comorbidity, CHARLS, ELSA, machine learning, metabolic index, composite, metabolic, index, predicting