基于中医证型的高级别浆液性卵巢癌铂耐药复发预测模型的构建与验证

Construction and Verification of Platinum-Resistant Recurrence Prediction Model for High-Grade Serous Ovarian Cancer Based on Traditional Chinese Medicine Syndrome

目的:构建并外部验证基于中医证型的高级别浆液性卵巢癌铂耐药复发患者的临床预测模型。方法:回顾性地收集高级别浆液性卵巢癌含铂化疗复发的患者273例,分为铂耐药组和铂敏感组,以调查表形式收集患者的相关病历资料。经单因素及多因素Logistic回归分析筛选出高级别浆液性卵巢癌铂耐药复发的独立影响因素作为模型预测因子,构建基于中医证型的铂耐药复发的临床预测模型,采用RStudio制作列线图进行模型呈现。前瞻性收集50例高级别浆液性卵巢癌患者的病历数据作为验证集对预测模型的应用效能进行外部验证。最终通过绘制受试者工作特征(ROC)曲线及校准曲线评价该模型效能。结果:高级别浆液性卵巢癌患者的单因素分析显示病程、是否满意减瘤手术、FIGO分期、末次化疗后CA125、脂代谢、是否淋巴结转移和中医证型等指标的P<0.1;多因素二元Logistic回归结果显示脂代谢、FIGO分期、中医证型、病程、末次化疗CA125水平是高级别浆液性卵巢癌患者发生铂耐药复发的独立影响因素。高级别浆液性卵巢癌的铂耐药复发诊断界值为0.465,此时对应的灵敏度(即敏感度)为84.6%,特异度为65.7%。HL检验的χ^(2)值为12.479,P>0.05;建模集与验证集校准曲线表明预测模型精确度较好;建模集AUC为0.777,验证集AUC为0.758,表示模型有较好的预测效能。结论:预测模型预测高级别浆液性卵巢癌的铂耐药复发效能良好,有助于临床工作中预测初诊高级别浆液性卵巢癌患者发生铂耐药的风险,对重点人群及早关注和干预,以延长无进展生存时间,但后续仍需更多有效的支持性数据,以增加预测模型性能的优良性。

Objective:To construct and externally verify a clinical prediction model for patients with platinum-resistant recurrence of high-grade serous ovarian cancer based on traditional Chinese medicine(TCM)syndrome.Methods:Totally 273 patients with high grade serous ovarian cancer who relapsed with platinum-containing chemotherapy were retrospectively collected and divided into platinum-resistant group and platinum-sensitive group.Medical records of patients were collected in the form of questionnaire.Independent factors influencing platinum-resistant recurrence of high-grade serous ovarian cancer were selected as model predictors by univariate and multivariate logistic regression analysis.Clinical prediction model of platinum-resistant recurrence based on TCM syndrome was constructed,and the model was presented by using RStudio to make a line graph.The clinical data of 50 patients with high-grade serous ovarian cancer were prospectively collected as a validation set for external validation of the predictive model.Finally,the receiver operating characteristic curve(ROC)and calibration curve were drawn to evaluate the efficiency of the model.Results:Univariate analysis of patients with high-grade serous carcinoma showed P<0.1 for the disease course,satisfaction with tumor reduction surgery,FIGO stage,CA125 after the last chemotherapy,lipid metabolism,lymph node metastasis and TCM syndrome.Multivariate binary Logistic regression equation analysis showed that lipid metabolism,FIGO stage,TCM syndrome,course of disease,and CA125 level in the last chemotherapy were independent factors influencing the occurrence of platinum resistance recurrence in ovarian cancer patients.The diagnostic threshold for platinum-resistant recurrence of high-grade serous ovarian cancer was 0.465,and the corresponding sensitivity(sensitivity)was 84.6%.And specificity was 65.7%.Theχ^(2) value of HL test was 12.479,with P>0.05.The calibration curves of modeling set and verification set showed that the prediction model was accurate.The modeling set AUC was 0.777,and the validation set AUC was 0.758,indicating that the model had good prediction efficiency.Conclusions:The prediction of platinum resistance recurrence of high-grade serous ovarian cancer has good efficacy,which is helpful to predict the risk of platinum resistance in newly diagnosed high-grade serous ovarian cancer in clinical work.Early attention and intervention should be paid to key groups to prolong the progression-free survival time,but more effective supporting data are still needed to increase the benign performance of the prediction model.