Multimodality treatment and survival in primary cardiac malignancies: evidence from competing-risk and machine learning analyses

Background Primary cardiac malignancies are rare and highly aggressive, with limited clinical evidence to guide optimal treatment. This study evaluated the association between multimodal treatments and survival outcomes under competing-risk conditions. Methods A retrospective cohort analysis was conducted using SEER-18 data (2000–2018) to identify patients with cardiac malignancies. Overall survival and cause-specific mortality were assessed using Kaplan–Meier and Fine–Gray competing-risk models. Multimodal treatment effects were examined using adjusted regression analyses. Prognostic performance was further evaluated using machine learning methods, including random survival forests and survival trees, with discrimination, calibration, and Brier scores assessed in a validation cohort. Results Among 416 patients, angiosarcoma predominated, most had regional/distant disease, surgery and chemotherapy were common, and over 80% experienced either primary cardiac malignancy-specific mortality or other-cause mortality. Chemotherapy plus surgery was associated with lower primary cardiac malignancy-specific mortality (AHR: 0.72; 95% CI 0.54–0.95) and other-cause mortality (AHR: 0.58; 95% CI 0.36–0.93). Surgery with radiotherapy was not significantly associated with primary cardiac malignancy-specific survival but with lower other-cause mortality (AHR: 0.44; 95% CI 0.22–0.87). Compared with surgery alone, multimodal combinations were not significantly associated with primary cardiac malignancy-specific mortality; however, surgery with chemotherapy (AHR: 0.54), surgery with radiation (AHR: 0.29), and triple therapy (AHR: 0.34) were linked to lower other-cause mortality. Regression tree revealed age- and stage-specific heterogeneity, with greater survival likelihood among patients ≤ 65 years with regional/distant disease who underwent surgery. Conclusion Surgery, particularly when combined with chemotherapy, was associated with improved outcomes. Benefits varied by age and stage, underscoring the need for individualized multimodal treatment strategies.