Abstract
BACKGROUND: Pancreatic signet ring cell carcinoma (PSRCC) is a rare and aggressive subtype of pancreatic cancer, with a poor prognosis and limited evidence on the survival benefit of chemotherapy. From the perspective of conditional survival (CS) prognosis, this study sought to assess the effect of chemotherapy on PSRCC survival and to construct a predictive model integrating CS analysis. METHODS: Using the SEER database, 708 PSRCC patients diagnosed between 2000 and 2019 were analyzed. Propensity score matching (PSM) and Kaplan-Meier curves were employed to assess chemotherapy's impact on survival. The CS analysis was performed to evaluate dynamic survival probabilities. A nomogram was developed based on key prognostic factors identified through random survival forests (RSF), least absolute shrinkage and selection operator (LASSO) regression, and multivariate Cox analysis with a stepwise backward elimination procedure. And multiple evaluation methods were employed to assess the performance of the nomogram. RESULTS: The CS analysis for all cohort showed a rapid decline in survival probability within the first few years, dropping to 18% by year 1, 5% by year 3 and 3% by year 5. Chemotherapy improved short-term survival, with a 30% one-year survival rate compared to 8% in the non-chemotherapy group. However, long-term survival probabilities converged after the first year. Key prognostic factors included age, tumor size, stage, site, surgery, and chemotherapy were identified to develop a CS-integrated nomogram. And the nomogram was found to have strong predictive accuracy and clinical utility, validated by calibration, ROC, and decision curve analyses. CONCLUSION: Chemotherapy offered significant early survival benefits in PSRCC, although its long-term impact is limited. The developed nomogram provided a reliable tool for personalized survival prediction, with further validation needed in prospective studies.