Prognostic utility of natriuretic peptides and incremental value of the HFA-ICOS risk score in patients with multiple myeloma

Abstract

Multiple myeloma (MM) is associated with an elevated cardiovascular risk, necessitating effective risk stratification prior to chemotherapy. The Heart Failure Association–International Cardio-Oncology Society (HFA-ICOS) risk score has been proposed as a comprehensive assessment tool; however, it requires multifactorial evaluation and has insufficient external validation. In contrast, natriuretic peptides have been reported as single biomarkers for the prognostic prediction of MM. In this study, we compared the predictive performance of baseline B-type natriuretic peptide (BNP) levels and the HFA-ICOS risk score for predicting adverse events after chemotherapy. This retrospective cohort study included consecutive patients who initiated first-line treatment for MM between March 2015 and May 2025. Patients were stratified into high and low BNP groups (≥100 pg/mL and <100 pg/mL, respectively) according to baseline BNP levels before treatment and followed up for up to 5 years. The primary outcome was CTCAE grade ≥2 heart failure, and the secondary outcome was all-cause mortality. For heart failure, cumulative incidence functions were estimated using Gray's test, and multivariable Fine–Gray regression analysis was performed, accounting for competing risks. Kaplan–Meier analyses and Cox proportional hazard models were used to assess all-cause mortality. Discriminative performance was assessed using the time-dependent AUC (C-statistic), comparing BNP levels and the HFA-ICOS risk score. A total of 66 patients were included (median age: 72.5 years; 45.5% men). During follow-up, 20 patients (30.3%) developed heart failure, and 19 (28.8%) died. Cumulative incidence analysis showed that the high-BNP group had a higher heart failure risk than the low-BNP group (Gray's test: p=0.002) (Figure 1A). BNP remained an independent predictor of heart failure in the multivariable Fine–Gray regression analysis (sHR 3.75 [95% CI 1.45–9.76]; p=0.007). For all-cause mortality, the high-BNP group had a higher risk (log-rank test: p=0.026) (Figure 1B), and BNP was an independent predictor in the multivariable Cox regression analysis (HR 2.71[95% CI 1.04–7.10]; p=0.042). The time-dependent C-statistics for heart failure and all-cause mortality were 0.718 and 0.728 for BNP and 0.708 and 0.730 for HFA-ICOS, respectively, with no significant differences between the models (heart failure: ΔAUC 0.010, p=0.883; all-cause mortality: ΔAUC 0.002, p=0.986) (Figure 2A, 2B). In patients with MM, baseline BNP levels demonstrated fair discrimination for predicting heart failure and all-cause mortality, and equivalent predictive performance compared with the HFA-ICOS risk score. These findings support the use of BNP levels as an efficient tool for risk stratification in patients with MM.