脑转移瘤放射治疗的预后因素分析及预后模型的建立

背景与目的：放射治疗是大多数脑转移瘤最主要的治疗手段，而预后却受多种因素的影响。本研究探讨影响肿瘤脑转移患者放射治疗生存期的预后因素并建立预后指数模型。方法：选择2008年1月—2011年7月符合入组条件接受放射治疗的脑转移患者140例，对各影响因素行单因素分析，差异有统计学意义的因素再行多因素分析，筛选出与预后最为相关的因素，并计算预后指数(prognostic index，PI)。同时，评估递归分区分析(recursive partitioning analysis classes，RPA)、脑转移基本评分(basic score for brain metastases，BS-BM)及等级预后评估标准(the graded prognostic assessment index，GPA)是否与预后相关。结果：全组中位生存时间为222 d。单因素检验提示卡氏评分(karnofsky performance score，KPS)、脑转移数目、中枢外转移、原发灶控制、放疗剂量、血红蛋白与预后相关；多因素分析筛选出KPS(P=0.002、Wald=9.700)、中枢外转移(P=0.018，Wald=5.604)、原发灶控制(P=0.001、Wald=10.212)3个因素影响总生存期。Log-rank检验提示，3种评分模式均与预后相关，而对于本组患者的3、6个月生存概率的预测，PI优于其他评分模式。结论：PI预后指数模型和3种预后指数均能反映预后，但PI更佳。

Analysis of prognostic factors and index model in patients with brain metastases after wholebrain radiotherapy

Background and purpose: Radiation therapy is still the most primary treatment of brain metastases, and prognosis is affected by many factors. The aim of this study was to identify the prognostic factors and to establish a prognostic index model in patients with brain metastases after whole-brain radiotherapy (WBRT). Methods: We reviewed the clinical date of 140 patients with brain metastases radiotherapy in our hospital from Jan. 2008 to Jul. 2011. The significance of prognostic variables in the survival was resulted from both univariate analysis and multivariate analysis. The prognostic index (PI) was established based on Cox regression analysis and subgrouping values. It was assessed whether recursive partitioning analysis classes (RPA), basic score for brain metastases (BSBM) and the graded prognostic assessment index (GPA) were related to prognosis. Results: The median survival time was 222 days. The univariate analysis showed that the independent prognostic factors were KPS performance status, number of brain metastases, presence of extracranial metastases, primary tumor status, radiation dose, hemoglobin. The multivariate analysis showed that KPS performance status (P=0.002, Wald=9.700), presence of extracranial metastases (P=0.018, Wald=5.604) and primary tumor status (P=0.001, Wald=10.212) were significantly correlated with overall survival. RPA, BS-BM and GPA were closely related to their prognosis by Log-rank test. In predicting 3 months and 6 months of survival for patients, PI was better than other modes. Conclusion: Our data suggest that the 3 indexes RPA, BS-BM and GPA are valid prognostic index models, but PI model is better.