Abstract
BACKGROUND: Major Depressive Disorder (MDD) drives global disability burden with suicide as a critical concern, yet multi-system biomarkers integrating erythroid, inflammatory, and metabolic pathways for suicide risk stratification remain underexplored. METHODS: This cross-sectional study analyzed 357 DSM-Ⅴ diagnosed MDD patients, excluding acute infections, autoimmune disorders, immunomodulatory treatments, or malignancy. Fasting blood samples quantified erythroid parameters, inflammatory indices including pan-immune-inflammation value (PIV), the triglyceride glucose (TyG) index calculated as ln (TG×FBG/2), and thyroid hormones. Patients’ suicide risk was assessed via structured interviews, and were categorized into three groups: non-suicidal ideation (non-SI), suicidal ideation (SI) only, and suicide attempt (SA) history. RESULTS: MDD patients with suicidal tendencies showed significantly higher unmarried proportions, red blood cell counts, years of schooling, logPIV, and mood stabilizer use alongside lower age, triglyceride glucose index, fasting glucose, thyroxine, and hypertension prevalence versus patients without suicidal tendencies. Binary regression identified logPIV and more complex conditions (use of mood stabilizers) as risk factors with odds ratios (OR) of 2.38 and 3.86 respectively, while marriage reduced risk (OR = 0.29). Ordinal regression across suicide risk tiers confirmed illness duration, logPIV, and more complex conditions (use of mood stabilizers) as escalating risk predictors, whereas marriage remained protective. Combined discriminatory performance (AUC) of disease duration, years of schooling, logPIV, TyG index, T4 levels, marital status, hypertension, and mood stabilizer usage was 0.64 (non-SI vs. SI), 0.82 (SI vs. SA), and 0.85 (non-SI vs. SA). CONCLUSION: Dysregulated inflammation via logPIV, metabolic dysfunction, marital status, and mood stabilizer use constitute validated biomarkers for suicide risk stratification, supporting integrated multi-system models for early clinical intervention. CLINICAL TRIAL NUMBER: Not applicable. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12888-025-07616-3.