获得性免疫缺陷综合征相关原发性中枢神经系统淋巴瘤的预后研究

Prognostic study of acquired immune deficiency syndrome-related primary central nervous system lymphoma

目的构建获得性免疫缺陷综合征(AIDS)相关原发性中枢神经系统淋巴瘤(AR-PCNSL)的短期预后生存分析模型并进行内部验证。方法由首都医科大学附属北京地坛医院电子病例系统中收集2013年1月至2022年3月经病理确诊的42例AR-PCNSL患者组成1个回顾性队列,死亡为终点事件,主要评价指标为总体生存率(OS),通过单因素Cox回归分析筛选与OS相关因素,构建多因素Cox生存分析模型并绘制列线图,模型区分度和准确度分别用C指数和校准曲线评价(采用bootstrap重抽样1000次的方法绘制)。生存资料采用Kaplan-Meier法进行生存分析和log_rank检验。结果42例患者年龄为[34(28,51)]岁,其中男性39例(93%)。17例(40.5%)患者出现死亡终点,中位生存时间未达到,中位随访时间为945(230,1518)d。Karnofsky功能状况评分(KPS)(HR=0.27、95%CI:0.073~1.0、P=0.012)和深部脑结构受累(HR=2.42、95%CI:1.747~7.8、P=0.040)两个变量纳入多因素Cox回归分析构建预后模型,C指数为0.72,预测校准曲线与理想曲线接近。根据列线图得分的三分位数进行危险分层,低、中风险组生存曲线交叉,与高风险组生存曲线无交叉(P=0.0023)。结论KPS低于70分、深部脑结构受累是影响AR-PCNSL预后的危险因素,基于二者Cox回归分析模型区分度一般但预测准确性高,仍需要进一步优化。

Objective To develop and validate a short-term prognostic model of acquired immune deficiency syndrome(AIDS)-related primary central nervous system lymphoma(AR-PCNSL).Methods 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 1000 bootstrapped resampling to qualify any overfitting.Survival data were analyzed by Kaplan-Meier survival analysis and log_rank test.Results 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).Conclusions 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.