Nomograms for predicting cancer-specific survival in patients with primary central nervous system lymphoma: a population-based analysis

BACKGROUND: This study identified the risk factors for survival in patients with primary central nervous system lymphoma (PCNSL). Nomograms were developed and validated to predict individualized overall survival (OS) and cancer-specific survival (CSS) in this particular cohort. METHODS: Patients diagnosed with PCNSL between 1975 and 2016 were selected from the Surveillance, Epidemiology, and End Results (SEER) database for this study. The Cox regression model, the Fine and Grey’s model, and the backward method were applied to determine the risk factors for OS and CSS. Nomograms were established accordingly. Internal and external validation was performed in an Asian population to examine the accuracy of the nomograms. RESULTS: A total of 5,900 patients with PCNSL were identified from the SEER database. A further 163 patients with PCNSL from the Beijing Tiantan Hospital between 2004 and 2018 were included. Age at diagnosis, tumor site, pathological subtype, surgery, chemotherapy, coexisting malignancies, and HIV infection were independent risk factors of CSS. In addition to the risk factors of CSS, gender, marital status, and radiation were also independent factors of OS. Nomograms were developed to estimate the 1-, 3-, and 5-year OS and CSS. The discrimination and calibration of the nomograms performed well. The C-indexes of the nomograms for OS and CSS prediction were 0.728 [95% confidence interval (CI): 0.703–0.753] and 0.726 (95% CI: 0.696–0.756), respectively. In addition, compared with previously published OS nomograms, the newly established nomograms displayed superior prediction for OS. CONCLUSIONS: Nomograms predicting the 1-, 3- and 5-year OS and CSS of patients with PCNSL were established in this study. The validated nomograms showed relatively good performance and may be used clinically to evaluate patients’ individualized risk and prognosis with PCNSL. Free software for individualized survival prediction is provided at http://www.pcnsl-survivalprediction.cn.