帕博利珠单抗治疗恶性黑色素瘤致免疫相关性甲状腺功能异常的影响因素及预测模型的建立

Factors affecting immune-related thyroid foundation abnormality and establishment of prediction model for the treatment of malignant melanoma with pembrolizumab

目的 探讨恶性黑色素瘤(MM)患者经帕博利珠单抗治疗后发生免疫相关性甲状腺功能异常(TFA-irAEs)的影响因素,并构建预测模型。方法 回顾性分析2018年1月1日至2020年1月31日期间在北京大学肿瘤医院接受帕博利珠单抗治疗的MM患者临床资料,包括性别、年龄、合并基础疾病、体表面积及分期等,采用单因素分析和多因素logistic回归分析帕博利珠单抗治疗发生TFA-irAEs的影响因素,并建立预测模型。采用受试者工作特征(ROC)曲线评价预测模型。结果 本研究共纳入126例病例,其中发生不良反应91例(72.2%),TFA-irAEs 36例(28.6%)。经logistic回归分析显示,TFA-irAEs影响因素包括性别、有无基础疾病、疗程、合并靶向药物治疗及合并化疗±靶向药物治疗,ROC曲线下面积为0.934±0.024(95%CI:0.886~0.979),预测特异度为77%,灵敏度为91.4%,阳性预测准确度为68.6%,阴性预测准确度为92.8%。当预测模型方程logit(P)≥0.2879152时,可作为判断使用帕博利珠单抗后是否发生TFA-irAEs的标准。结论 患者性别、有无基础疾病、疗程、合并靶向药物及合并化疗药物治疗是帕博利珠单抗发生TFA-irAEs的影响因素,拟合的预测模型检验结果可靠。

Objective To investigate the influencing factors of immune-related thyroid foundation abnormality adverse(TFA-irAES) in patients with malignant melanoma(MM) treated with pembrolizumab and construct a predictive model.Methods We conducted a retrospective analysis of MM patients treated with pembrolizumab in Peking University Cancer Hospital from January 1, 2020 to January 31, 2020. Clinical data including gender, age, underlying diseases, body surface area, and stage, etc. Univariate analysis and multivariate logistic regression analysis were used to determine the influencing factors of TFA-irAEs treated with pembrolizumab, and a predictive model was established. Receiver operating characteristic(ROC) curve was used to evaluate the prediction model. Results A total of 126 patients were included in this study. Among them, 91cases(72. 2%) had adverse reactions, including 36 cases(28. 6%) of TFA-irAEs. The logistic regression analysis showed, the influencing factors of TFA-irAEs included gender, underlying diseases, course of treatment, combined targeted drugs, combined chemotherapy or chemotherapy plus targeted drugs, the area under the ROC curve was 0. 934±0. 024(95%CI:0. 886~0. 979), with a specificity of 77%, sensitivity of 91. 4%, positive predictive value of 68. 6%, and negative predictive value of 92. 8%. When the predictive model equation logit(P) ≥0. 2879152, it could be used as a criterion to determine whether TFAirAEs occurred after the use of pembrolizumab. Conclusion Gender, underlying diseases, course of treatment, combined treatment with targeted drugs and chemotherapy drugs are the influencing factors of TFA-irAEs in patients with pembrolizumab. The results of the fitted prediction model are reliable.