免疫检查点抑制剂治疗非小细胞肺癌患者疗效影响因素分析及预测模型的构建

Prognostic factors after non-small cell lung cancer patients treated with immune checkpoint inhibitors and establishment of a prediction model

目的探讨影响免疫检查点抑制剂(ICI)治疗非小细胞肺癌(NSCLC)患者疗效的因素,并建立预测模型及验证。方法回顾性分析某三甲医院2017年1月—2022年1月656例使用过ICI治疗NSCLC患者的一般资料和临床数据。Log-Rank单因素分析与Cox多因素回归分析影响无进展生存期(PFS)的因素。使用R stutio 4.2.0绘制软件构建列线图模型,通过一致性指数(C-index)、受试者工作曲线(ROC)、校正曲线和决策曲线分析(DCA)对模型进行内部与外部验证。结果Cox多因素结果显示体质量指数(BMI)、治疗方案、免疫表达、脑转移、肾上腺转移、吸烟指数、KRAS突变、合并抗菌药物、合并PPI、合并降压药、合并糖皮质激素、合并镇痛药、PLR和CRP为PFS的独立影响因素。PFS预测模型的训练集和验证集的C-index分别为0.849 vs 0.884。3、6、12和18个月的曲线下面积分别为0.900 vs 0.964,0.885 vs 0.891,0.835 vs 0.860和0.804 vs 0.846。3、6、12和18个月的Brier指数分别为0.093 vs 0.039,0.110 vs 0.050,0.116 vs 0.083和0.118 vs 0.085。ROC曲线表明模型具有较好的临床鉴别能力。校准曲线的实际曲线接近理想曲线,表明该模型具有较好的临床区分度和校准度。决策曲线表明,该预测模型有较好的临床净获益。结论①高BMI基线、吸烟指数>400、选择ICI合并化疗治疗方案、免疫表达、KRAS突变是免疫治疗预后的独立保护因素。脑转移、肾上腺转移、合并抗菌药物、合并PPI、合并降压药、合并糖皮质激素、合并镇痛药、高PLR基线和CRP水平异常是免疫治疗疗效的独立风险因素;②构建的预测模型可以为临床上使用ICI治疗NSCLC患者疗效预测提供一定的参考。

Objective To determine the factors affecting the efficacy of immune checkpoint inhibitors(ICI)for non-small cell lung cancer(NSCLC),establish a prediction model and verify it.Methods The general and clinical data of 656 patients with NSCLC treated with ICI from January 2017 to January 2022 in a top comprehensive hospital were retrospectively analyzed.Log-Rank univariate analysis and Cox multivariate regression analysis were used to identify the factors that affected progress free survival(PFS).A Nomogram model was established with the R stutio 4.2.0 rendering software,and the model was validated internally and externally by consistency index(C-index),receiver operating curve(ROC),correction curve,and decision curve analysis(DCA)curve.Results Cox multivariate results showed that body mass index,treatment regimen,immune expression,brain metastases,adrenal metastases,smoking index,KRAS mutation,antibiotic combination,proton pump inhibitor combination,antihypertensive combination,glucocorticoids combination,analgesic combination,platelet-to-lymphocyte ratio(PLR)and C-reactive protein(CRP)were all independent influencing factors for PFS.The C-index of the training set and the validation set of PFS prognostic model was 0.849 vs 0.884,respectively.The area under curve at 3,6,12 and 18 months was 0.900 vs 0.964,0.885 vs 0.891,0.835 vs 0.860 and 0.804 vs 0.846,respectively.The Brier index at 3,6,12 and 18 months was 0.093 vs 0.039,0.110 vs 0.050,0.116 vs 0.083 and 0.118 vs 0.085,respectively.ROC curve showed that the model had good clinical differential ability.The actual curve of calibration curve was close to the ideal curve,indicating good clinical differentiation and calibration degree of the model.The decision curve showed good clinical net benefit of the prediction model.Conclusion①High body mass index baseline,smoking index>400,choice of ICI combined with chemotherapy,immune expression,and KRAS mutation are independent protective factors for the prognosis of immunotherapy.Brain metastasis,adrenal metastasis,concomitant use of antibiotics,proton pump inhibitor,antihypertensive drugs,glucocorticoids,and analgesics,high PLR baseline,and abnormal CRP levels are independent risk factors for the efficacy of immunotherapy;②The established prediction model can provide a certain reference for the efficacy of ICI treatment for NSCLC patients in clinical practice.