To develop and validate a short-term prognostic model of acquired immune deficiency syndrome (AIDS)-related primary central nervous system lymphoma (AR-PCNSL).

Total of 42 patients confirmed AR-PCNSL pathologically were enrolled retrospectively from January, 2013 to March, 2022 in Beijing Ditan Hospital, Capital Medical University. The primary endpoint was overall survival (OS). Clinical variables associated with survival were assessed based on clinical importance scientific knowledge, and predictors identified in previously published article. The prognostic factors included in multivariate Cox regression analysis were used to develop a predicted model and a nomogram was established accordingly. The discrimination of the model were evaluated by C index and 1 000 bootstrapped resampling to qualify any overfitting. Survival data were analyzed by Kaplan-Meier survival analysis and log_rank test.

The age of 42 patients was [34 (28, 51)] years old; 93% (39/42) patients were male; 40.5% (17/42) patients died due to AR-PCNSL terminal period, and the median survival time was not available. The median follow-up time were [945 (230, 1518)] days. Karnofsky Performance Status (KPS) score (HR = 0.27, 95%CI: 0.073-1.0, P = 0.012) and deep brain structure (HR = 2.42, 95%CI: 1.747-7.8, P = 0.040) were included in multivariate Cox regression analysis to fit a prognostic model with C index 0.72, and the predicted calibration curve was close to the ideal curve. Risk was stratified according to the tertiles of the nomogram scores. The Kaplan-Meier curves of the low and medium risk groups crossed partially, but paralleled with the high risk group (P = 0.0023).

KPS lower than 70, deep brain structure involvement are two risk factors for AR-PCNSL patients to effectively predict the individual short-term mortality risk. The Cox regression model with moderate discrimination and accurate prediction needs to be optimized further.