【摘要】 目的 分析老年弥漫大B细胞淋巴瘤(CGA)情况，并构建预后不良的预测模型。方法 选取2020年3月至2023年2月南京医科大学第一附属医院收治的初诊老年弥漫大B细胞淋巴瘤(DLBCL)患者183例行回顾性研究，以2∶1比例分为训练集(122例)及验证集(61例),以1年生存情况评估预后，生存为预后良好，死亡为预后不良，比较其简化CGA分层、工具性日常生活活动能力及年龄-Charlson合并症指数-白蛋白(IACA)指数等CGA情况及临床分期等基线资料。在训练集中，采用单因素及多因素logistic回归分析筛选老年DLBCL患者预后不良的影响因素，构建老年DLBCL患者预后不良的预测模型；在验证集中，采用Hosmer-Lemeshow检验评估预测模型的拟合度，采用受试者工作特征(ROC)曲线评估预测模型的预测效能。采用SPSS 25.0软件进行数据分析。根据数据类型，组间比较分别采用t检验、χ~2检验及Fisher精确概率检验。结果 训练集中有21例(17.21%)发生预后不良，验证集中有11例(18.03%),差异无统计学意义(P=0.891)。训练集单因素分析发现，简化CGA分层、IACA指数、临床分期、化疗疗效共4个变量在老年DLBCL患者发生预后不良上存在显著不同(P<0.05)。多因素logistic回归分析显示，IACA指数≥3分、Ann ArborⅢ～Ⅳ期均为老年DLBCL患者预后不良的独立危险因素(P<0.05),简化CGA分层为适合化疗、化疗疗效为完全缓解则为独立保护因素(P<0.05),构建预测模型为Y=-4.221+(-1.212)×简化CGA分层+1.451×IACA指数+1.269×临床分期+(-1.036)×化疗疗效。Hosmer-Lemeshow检验显示，预测模型的拟合度χ~2=12.580,P=0.125,校准度良好。在验证集中，ROC曲线分析显示，预测模型ROC曲线下面积为0.886,95%CI 0.832～0.940,灵敏度为72.73%,特异度为88.00%。结论 初诊老年DLBCL患者近期预后与简化CGA分层、IACA指数、临床分期、化疗疗效有关，以此建立的预测模型对筛选1年死亡高危患者有积极作用。更多还原

【Abstract】 Objective To analyze the comprehensive geriatric assessment(CGA) status of elderly patients with diffuse large B-cell lymphoma(DLBCL) and construct a prediction model of poor prognosis. Methods A retrospective study was conducted on 183 elderly patients with newly-diagnosed DLBCL in our hospital from March 2020 to February 2023. They were divided into a training set(122 cases) and a validation set(61 cases) at a ratio of 2∶1. The prognosis was evaluated by 1-year survival status, and survival was taken as good prognosis and death was considered as poor prognosis. Simplified CGA stratification, instrumental activities of daily living, and age, Charlson comorbidity index, albumin(IACA) index and baseline data such as clinical staging were compared. In the training set, univariate and multivariate logistic regression analyses were used to screen the influencing factors of poor prognosis in elderly DLBCL patients, and a prediction model for poor prognosis was then constructed. In the validation set, Hosmer-Lemeshow test was adopted to evaluate the fitting degree of the prediction model, and receiver operating characteristic(ROC) curve was adopted to evaluate the predictive efficiency of the model. SPSS statistics 25.0 was used for statistical analysis. Data comparison between two groups was performed using t test, Chi-square test or Fisher exact probability test depending on data type. Results There were 21 cases(17.21%) with poor prognosis in the training set and 11 cases(18.03%) with poor prognosis in the validation set, without statistical difference between two sets(P=0.891). Univariate analysis of the training set found that four variables, including simplified CGA stratification, IACA index, clinical staging, and chemotherapy efficacy, were significantly different in the poor prognosis of elderly patients with DLBCL(P<0.05). Multivariate logistic regression analysis showed that IACA index ≥3 and Ann Arbor stage Ⅲ-Ⅳ were independent risk factors for poor prognosis in elderly DLBCL patients(P<0.05), and simplified CGA stratified as suitable chemotherapy and chemotherapy efficacy obtaining complete remission were independent protective factors(P<0.05). The prediction model was constructed as Y=-4.221+(-1.212) × simplified CGA stratification+1.451 × IACA index + 1.269 × clinical staging +(-1.036) × chemotherapy efficacy. Hosmer-Lemeshow test revealed that the fitting degree of the prediction model was shown as Chi-square=12.580, P=0.125, with good calibration. In the validation set, ROC curve analysis showed that the area under curve(AUC), 95%CI, sensitivity and specificity of the prediction model were 0.886, 0.832-0.940, 72.73% and 88.00%, respectively. Conclusion In elderly patients with newly-diagnosed DLBCL, short-term prognosis is related to simplified CGA stratification, IACA index, clinical staging and chemotherapy efficacy. The prediction model established based on these variables has a positive effect on screening patients with high-risk of one-year death.更多还原