Prediction of overall survival in patients with hepatocellular carcinoma and second primary malignancies: a prognostic model based on the SEER database.

Background: Survival in hepatocellular carcinoma (HCC) has improved with advances in diagnosis and treatment, resulting in a growing population of long-term survivors. As survivorship increases, second primary malignancies (SPMs) have become an important clinical issue and may substantially affect long-term outcomes. However, prognostic determinants and individualized overall survival (OS) prediction tools for HCC patients who subsequently develop SPMs remain limited. This study aimed to identify factors associated with OS and to develop and validate a multivariable prognostic model for OS prediction in this population. Methods: Patients diagnosed with primary HCC between 2004 and 2015 who later developed a single SPM, were ≥20 years old, and had an interval of ≥6 months between the two primary cancers were identified from the Surveillance, Epidemiology, and End Results (SEER) database. The SPM cohort was randomly divided into a training set and a validation set at a 7:3 ratio. Independent prognostic factors were identified using a Cox proportional hazards model in the training set. A nomogram was then constructed based on these factors. Model performance was evaluated using the concordance index (C-index), area under the receiver operating characteristic (ROC) curve (AUC), calibration curves, and decision curve analysis (DCA). To further strengthen internal validation, 1,000 bootstrap resamples of the training set were performed. This approach was used to assess potential model overfitting and to obtain more reliable calibration results. Results: A total of 1,608 eligible patients with SPM were included, comprising 1,126 in the training set and 482 in the validation set. Multivariable analysis identified age, marital status, radiotherapy (RT), surgery, and latency period as independent prognostic factors for OS. The nomogram achieved C-indices of 0.731 in the training set and 0.717 in the validation set. In the training set, the AUCs for predicting 8-, 9-, and 10-year OS were 0.81, 0.79, and 0.77. The model demonstrated good calibration and potential clinical utility, as indicated by DCA. Conclusions: A nomogram predicting OS in HCC patients with subsequent SPM was developed and preliminarily validated using SEER data. The model demonstrated moderate discrimination and calibration during internal validation and may serve as a tool to support individualized follow-up and treatment decisions in this patient population.