Glucose metabolism-related genes shared by osteoporosis and frozen shoulder: exploratory diagnostic models and immune profiles

Osteoporosis (OP) and frozen shoulder (FS) are common musculoskeletal disorders with overlapping metabolic and inflammatory features. We examined glucose metabolism-related genes shared by the two conditions and developed four-gene models. GSE56815 and GSE140731 served as the OP and FS training datasets, respectively; GSE13850 and GSE238053 were used for secondary evaluation. Feature selection and model fitting were restricted to the training datasets, and secondary-evaluation scores used fixed model coefficients. MARCKS, MAP3K1, LTF, and ZMIZ1 were identified as shared candidate genes. The OP model achieved an AUC of 0.900 in training and 0.715 in the reported GSE13850 secondary comparison (95% CI 0.526–0.904); the sample labels require verification. The corresponding FS AUCs were 0.991 and 0.875 (95% CI 0.592–1.000); the validation estimate was based on four cases and four controls. The models showed preliminary discriminatory ability in the reported comparisons; the GSE13850 sample mapping requires verification, and both models require evaluation in larger independent cohorts. Exploratory docking prioritized three computational candidates for NFKBIA, but their binding and biological effects remain untested.