构建可切除T_(2)~T_(4)期肺腺癌根治术后复发生存的列线图模型

A nomogram model of recurrence survival after radical resection of stage T_(2)-T_(4)lung adenocarcinoma

目的分析可切除T_(2)~T_(4)期肺腺癌患者根治术后复发生存的影响因素及构建列线图模型。方法收集2019年1月至2022年12月间在泉州市第一医院行肺癌根治术的患者进行回顾性分析。共纳入400例连续患者,根据Zstats统计学软件生成随机序列号,将患者按7∶3分为训练集(n=280)和验证集(n=120)。记录患者的临床、病理、手术和随访信息,随访截至2024年5月31日,记录患者的无复发生存期(RFS)。采用单因素及多因素Cox风险比例回归模型分析影响肺腺癌患者RFS的因素。结果训练集患者中位随访时间为14.27个月,复发率为40.36%,验证集中位随访时间为14.10个月,复发率为50.83%。训练集与验证集的人口统计学和临床特征均衡可比(P>0.05)。单因素Cox回归及LASSO Cox分析,C反应蛋白-白蛋白比(CAR)、体质量指数(BMI)、分化程度、肿瘤大小、淋巴血管侵犯(LVI)、血小板与淋巴细胞比值(PLR)、癌胚抗原(CEA)是影响肺腺癌患者术后RFS的因素。多因素Cox回归分析结果显示,低分化/中-低分化、LVI、CAR≥1.09×10^(-3)是影响肺腺癌患者术后RFS的独立危险因素。根据上述结果,构建了3年和5年的RFS列线图预测模型。训练集中,RFS预测模型的C-index为0.783(95%CI:0.744~0.822),验证集采用RFS列线图模型对队列中每例患者进行评分,C-index为0.717(95%CI:0.651~0.784)。训练集和验证集中,预测3年及5年内RFS的时间依赖性受试者工作特征曲线下面积均>0.700。在训练集和验证集中列线图的校准曲线显示出预测和观察到的生存率高度一致。Kaplan-Meier法结果显示,在训练集及验证集中,低风险肺腺癌患者的RFS均显著长于高风险患者。结论本研究构建包含CAR、分化程度及LVI在内的预测肺腺癌患者根治术后RFS的列线图模型,该列线图模型可准确预测患者复发生存,有助于对高危患者进行适当的术后管理。

Objective To analyze the feature recognition of recurrence survival in patients with resectable stage T_(2)-T_(4)lung adenocarcinoma after radical surgery and construct a nomogram model.Methods The patients who underwent radical resection of lung cancer in Quanzhou First Hospital from January 2019 to December 2022 were analyzed retrospectively.A total of 400 consecutive patients were included.Random serial numbers were generated by Zstats statistical software,and the patients were divided into a training set(n=280)and a validation set(n=120)according to 7∶3.Clinical,pathological,surgical,and follow-up information were recorded,and relapse-free survival(RFS)was recorded as of May 31,2024.Univariate and multivariate Cox proportional regression models were used to analyze the factors affecting RFS in patients with lung adenocarcinoma.Results The median follow-up time of the training set was 14.27 months,and the recurrence rate was 40.36%.The follow-up time of the verification center was 14.10 months,and the recurrence rate was 50.83%.The demographic and clinical characteristics of the training set and the validation set were basically balanced(P>0.05).Univariate Cox regression and LASSO Cox analysis showed that C-reactive protein-albumin ratio(CAR),body mass index(BMI),differentiation degree,tumor size,lymphovascular invasion(LVI),platelet-to-lymphocyte ratio(PLR),and carcinoembryonic antigen(CEA)were the factors influencing postoperative RFS in patients with lung adenocarcinoma.Multivariate Cox regression analysis showed that poorly differentiated/medium-poorly differentiated,LVI and CAR≥1.09×10^(-3)were independent risk factors for postoperative RFS in lung adenocarcinoma patients.Based on the above results,3-year and 5-year RFS nomogram prediction models were constructed.In the training set,the C-index of the RFS prediction model was 0.783(95%CI:0.744-0.822);in the validation set,each patient in the cohort was scored using the RFS nomogram model,and the C-index was 0.717(95%CI:0.651-0.784).In both the training set and the validation set,the area under the time-dependent subject operating characteristic curve of the predicted RFS within 3 and 5 years was>0.700.Calibration curves in the training set and validation set nomogram showed a high degree of agreement between predicted and observed survival rates.Kaplan-Meier method showed that the RFS of low-risk lung adenocarcinoma patients was significantly longer than that of high-risk patients in both the training set and the validation set.Conclusion In this study,a nomogram model including CAR,differentiation degree and LVI was constructed to predict RFS of lung adenocarcinoma patients after radical surgery.This nomogram model can accurately predict the recurrence survival of patients,and is helpful for appropriate postoperative management of high-risk patients.