A predictive model for identifying risk factors of early death in middle-aged and young adult hepatocellular carcinoma patients based on SEER database

The incidence of hepatocellular carcinoma (HCC) among young and middle-aged adults (18–50 years) has been increasing. This population often exhibits more aggressive tumor biology and atypical clinical presentations compared to older patients. This study aims to identify risk factors for early death in this population and develop a robust predictive model. We analyzed data from the Surveillance, Epidemiology, and End Results (SEER) database for 943 HCC patients aged 18–50 diagnosed between 2010 and 2022. Patients were randomly divided into a training cohort ( n = 661) and a validation cohort ( n = 282) at a ratio of 7:3. Univariate and multivariate logistic regression analyses were employed to identify independent risk factors for all-cause and cancer-specific early death. Nomograms were constructed based on the selected factors. Model performance was rigorously evaluated using the area under the receiver operating characteristic (ROC) curve (AUC), calibration curves, decision curve analysis (DCA), and clinical impact curves (CIC). Multivariate analysis revealed that married status, and receiving radiotherapy or surgery were independent protective factors against both all-cause and cancer-specific early death. Conversely, AFP positivity and older age were independent risk factors for both all-cause and cancer-specific early death. Advanced T stage was independently associated with cancer-specific early death and showed a borderline association with all-cause early death. Bone metastasis, although present in only 1.7% of patients, was associated with a markedly increased risk of all-cause early death. The nomograms demonstrated excellent discrimination, with AUC values of 0.893 and 0.891 in the training cohort, and 0.852 and 0.864 in the validation cohort, for all-cause and cancer-specific early death, respectively. The calibration curves demonstrated excellent agreement between predicted and observed outcomes. DCA and CIC confirmed the significant clinical utility and net benefit of the models across both cohorts. We developed and validated novel nomograms that accurately predict early death in middle-aged and young adult HCC patients. The model incorporated key sociodemographic, clinical, and treatment-related factors, demonstrating excellent performance and clinical applicability.