Abstract
BACKGROUND: Recent studies have reported that overall survival of elderly patients with primary central nervous system lymphoma (PCNSL), who have the highest incidence of this disease, had failed to benefit from the advancements in treatment strategies over the past decades. This highlights the necessity for intensified research to guide treatment decisions for this specific patient population. METHODS: The Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute (NCI) was used to extract data of elderly PCNSL patients (age ≥ 60) who were divided into training and validation groups at the ratio of 7:3, for our analysis. Conditional survival [CS(y|x)] was defined as the probability at survival additional y years given that the patient had not died of PCNSL at a specified period of time (x years) after initial diagnosis. The CS pattern of elderly PCNSL patients was analyzed. The least absolute shrinkage and selection operator (LASSO) regression and multivariate Cox regression analysis were applied to develop a novel CS-based nomogram. RESULTS: A total of 3315 elderly patients diagnosed with CNS lymphoma between 2000 and 2019 were extracted from the SEER database, of whom 2320 patients were divided into the training group and 995 into the internal validation group. CS analysis revealed a noteworthy escalation in the 5-year survival rate among elderly PCNSL patients for every additional year of survival. The rates progressed from an initial 21-49%, 63%, and 75%, culminating in an impressive 88% and the survival improvement over time was nonlinear. The LASSO regression identified nine predictors and multivariate Cox regression was used to successfully construct the CS-based nomogram model with favorable prediction performance. CONCLUSION: CS of elderly PCNSL patients was dynamic and increased over time. Our newly-established CS-based nomogram can provide a real-time dynamic survival estimation, allowing clinicians to better guide treatment decision for these patients.