摘要
目的:分析成人原发扁桃体弥漫大B细胞淋巴瘤(PTDLBCL)的临床特征,构建预后预测模型。方法:从美国SEER数据库获取2004—2015年诊断为PTDLBCL患者的临床资料、治疗记录及生存信息进行分析,共纳入符合条件的PTDLBCL患者711例。对这些患者的数据进行COX模型逐步回归分析筛选预后因素,采用rms R软件包拟合COX回归方程并建立列线图模型,利用Bootstrap法进行模型的内部验证,通过计算Harrell’s C指数验证模型的区分度,通过绘制校准曲线和计算Brier评分评价模型的准确度。结果:逐步回归分析结果显示,高龄(≥66岁)、Ann Arbor分期达Ⅲ或Ⅳ期、有B症状及未接受放化疗为PTDLBCL患者预后的不良影响因素。将5个预后因素纳入cph R软件函数构建的回归方程,据此绘制列线图,计算Harrell’s C指数为0.759,Brier分数为0.121。利用Bootstrap法重复抽样1000次进行内部验证得到Harrell’s C指数为0.753,Brier分数为0.124,3年、5年生存率校准曲线均接近理想曲线,表明该模型的精准度、区分度良好。结论:构建的PTDLBCL预后预测模型具有良好的预测价值,可协助临床对PTDLBCL患者的生存预后进行个体化分析。
Objective: To analyze the clinical characteristics of primary tonsillar diffuse large B-cell lymphoma(PTDLBCL), and built a prediction model to predict individual′s prognosis. Methods: The characteristic information, treatment record and survival information of 711 patients with PTDLBCL diagnosed from 2004 to 2015 were obtained from the Surveillance, Epidemiology, and End Results(SEER) Database of American National Cancer Institute. Prognostic factors were screened by COX model stepwise regression analysis in 711 cases. COX regression equation was fitted by rms R software package and a nomogram model was established. Bootstrap method was used to verify the model internally. The discriminant degree was verified by calculating Harrell’s C index, and its accuracy was evaluated by drawing calibration curve and calculating Brier score. Results: The stepwise regression analysis results showed that the elderly(≥66 years), Ann Arbor stage Ⅲ or Ⅳ, with B symptoms and did not receive radiation and chemotherapy for PTDLBCL patients were adverse factors. Five prognostic factors were incorporated into the regression equation constructed by cph R software function. Based on this, the nomograph was drawn, and the Harrell’s C index and Brier score were calculated as 0.759 and 0.121 respectively. Bootstrap method was used for internal verification, and the Harrell’s C index and Brier score were 0.753 and 0.124 respectively. The calibration curves of 3-year and 5-year overall survival were all close to the ideal curve, indicating that the accuracy and differentiation of this model were good. Conclusion: The prognostic prediction model of PTDLBCL we built has good predictive value, which can assist clinical individualized analysis for survival and prognosis of PTDLBCL patients.
作者
刘睿
黄方
厉嘉琪
万江波
郝思国
LIU Rui;HUANG Fang;LI Jiaqi;WAN Jiangbo;HAO Siguo(Department of Hematology,Xinhua Hospital Affiliated to Shanghai Jiaotong University School of Medicine,Shanghai,200092,China)
出处
《临床血液学杂志》
CAS
2021年第3期197-201,共5页
Journal of Clinical Hematology
