评估非小细胞肺癌患者预后模型的构建及验证分析

Construction and Validation of Prognostic Models for Non-Small Cell Lung Cancer Patients

摘要: 目的：分析非小细胞肺癌(Non-Small Cell Lung Cancer, NSCLC)患者预后的影响因素，并建立评估NSCLC患者预后的列线图模型，为临床诊疗提供依据。方法：选择2017年1月至2020年12月初次就诊于青岛市市立医院经组织病理学确诊为NSCLC患者共429例为研究对象，分析患者的临床资料及初治前相关血液学指标，采用单因素及多因素COX回归分析影响NSCLC患者总生存期(Overall Survival, OS)的独立预后因素并建立列线图模型用于预测NSCLC患者的生存率，时间依赖性ROC曲线(Time-Dependent ROC, Time ROC)评价模型的判别区分能力，决策曲线(Decision Curve Analysis, DCA)分析评估预测模型的临床实用性，采用X-tile软件按照预后指数(Prognostic Index, PI)将患者分为高、低风险组。结果：多因素COX回归分析结果显示TNM分期、病理类型、神经元特异性烯醇化酶(Neuron-Specific Enolase, NSE)、前白蛋白(Prealbumin, PAB)及全身免疫炎症指数/血清白蛋白(Systemic Immune-Inflammation Index/Albumin, SII/ALB)、吸烟史是NSCLC患者OS的独立预后因素。基于以上独立预后因素建立评估NSCLC患者预后的列线图模型，建模组的1年、3年、5年生存率的ROC曲线下面积(Area Under Curve, AUC)分别为0.744、0.799、0.843。验证组分别为0.843、0.794、0.871，提示该预测模型具有较好的判别区分能力，决策曲线分析示该预测模型具有较好的临床实用性。PI的最佳截断值为0.48，Kaplan-Meier生存分析及Log-rank检验显示整体组、建模组及验证组中低风险组患者OS均显著优于高风险组患者OS (P < 0.001)。结论：TNM分期、病理类型、NSE、PAB、SII/ALB、吸烟史是影响NSCLC患者预后的因素，基于以上因素建立的列线图模型对NSCLC患者的生存率有较好的临床预测价值。

Abstract: Objective: To analyze the factors influencing the prognosis of NSCLC patients, and establish a nomogram model to evaluate the prognosis of NSCLC patients, so as to provide a basis for clinical diagnosis and treatment. Methods: A total of 429 patients with NSCLC diagnosed by histopathology in Qingdao Municipal Hospital from January 2017 to December 2020 were selected as the study objects. Clinical data and related hematological indicators before initial treatment were analyzed. Univariate and multivariate COX regression analysis was used to analyze the independent prognostic factors affecting the overall survival of NSCLC patients, and a nomogram model was established to predict the survival rate of NSCLC patients. The discriminant and distinguishing ability of the time-dependent ROC curve evaluation model was evaluated, and the clinical practicability of the prediction model was evaluated by decision curve analysis. X-tile software was used to divide patients into high and low risk groups according to prognostic index. Results: Multivariate COX regression analysis showed that TNM stage, pathological type, neuron-specific enolase, prealbumin and systemic immune-inflammation index/albumin, and smoking history were independent prognostic factors for OS in NSCLC patients. Based on the above independent prognostic factors, a nomogram model was established to evaluate the prognosis of NSCLC patients. The area under ROC curve of 1-year, 3-year and 5-year survival rates in the modeling group were 0.744, 0.799 and 0.843, respectively. The results of the verification group were 0.843, 0.794 and 0.871, respectively, suggesting that the prediction model had good discriminant and distinguishing ability, and the decision curve analysis showed that the prediction model had good clinical practicability. The optimal cut-off value of PI was 0.48. Kaplan-Meier survival analysis and Log-rank test showed that the OS of patients in the low-risk group was significantly better than that of patients in the high-risk group in the whole group, modeling group and verification group (P < 0.001). Conclusion: TNM stage, pathological type, NSE, PAB, SII/ALB and smoking history are the factors that affect the prognosis of patients with NSCLC. The nomogram model established based on the above factors has a good clinical prediction value for the survival rate of patients with NSCLC.