Abstract
BACKGROUND: Frailty and comorbidity substantially influence clinical outcomes in patients with multiple myeloma (MM), yet existing tools such as the IMWG frailty score and Charlson Comorbidity Index (CCI) have limitations in real-world applicability and disease specificity. OBJECTIVE: To develop and externally validate a multiple myeloma-specific comorbidity index (MM-CI) using real-world data from Korean and Japanese cohorts. MATERIALS AND METHODS: This retrospective study was conducted in two parts: (I) development of MM-CI using a nationwide Korean claims cohort of 17,273 MM patients diagnosed between 2007 and 2022; and (II) external validation using two independent multicenter registry cohorts: 1473 Korean patients (2010-2021) and 314 Japanese patients (2008-2023). Multivariable Cox regression was performed incorporating key clinical factors, including ECOG performance status, R2-ISS stage, and frontline treatment intensity. RESULTS: The MM-CI was derived from eight variables: male sex; age 60-69, 70-79, and ≥80 years; congestive heart failure, cerebrovascular disease, hepatic disease, and cancer: 1 × (Sex: Male) + 2 × (Age: 60-69 years) + 4 × (Age: 70-79 years) + 6 × (Age ≥80 years) + 2 × (Congestive heart failure: Yes) + 1 × (Cerebrovascular disease: Yes) + 1 × (Hepatic disease: Yes) + 1 × (Cancer: Yes). Risk scores stratified patients into four groups: low (0-2), intermediate-I (3-4), intermediate-II (5), and high (≥6), with corresponding median overall survival (OS) of 72.5, 43.8, 30.9, and 20.3 months, respectively. Validation analyses demonstrated superior predictive performance of the MM-CI (AUC: 0.637) compared to the conventional age-adjusted CCI (AUC: 0.613) and the concise CCI used in the IMWG frailty score (AUC: 0.569). The MM-CI remained independently prognostic after adjustment for ECOG, R2-ISS, and treatment intensity. CONCLUSION: The MM-CI provides a simple, objective, and clinically applicable tool for comorbidity-based risk stratification in MM. It outperforms existing models and may support treatment decisions in real-world practice.