子宫肌瘤经子宫动脉栓塞术治疗后再次复发的预测模型构建

Construction of the predictive model of recurrence of hysteromyoma after uterine artery embolization

目的:分析影响子宫肌瘤经子宫动脉栓塞术治疗后再次复发的危险因素,建立预测再次复发的列线图模型。方法:回顾性分析2014年3月-2017年7月本院和杭州师范大学附属医院行子宫动脉栓塞术治疗的子宫肌瘤患者314例,术后随访1~3年,统计复发率,按照是否复发分为复发组与未复发组,采用logistic回归模型筛选影响再次复发的危险因素,基于各危险因素使用R软件建立复发的列线图预测模型并评估模型的准确度与一致性。结果:314例中有67例复发,复发率21.3%;二元logistic回归分析表明,年龄、肿瘤直径≥6cm、肌瘤多发是影响子宫肌瘤术后复发的独立危险因素,OR=0.923(95%CI:0.868~0.980)、4.817(95%CI:2.482~9.347)、2.866(95%CI:1.544~5.320);将独立危险因素引入R软件(R3.6.3)构建列线图模型,ROC曲线评估列线图模型区分度ROC曲线下面积为0.860(95%CI:0.803~0.918),校准曲线为斜率接近1的直线,Hosmer-Lemeshow拟合优度检验χ^(2)=8.528,P=0.384。结论:本研究基于影响子宫肌瘤术后复发的危险因素建立的列线图预测模型具有良好的区分度与准确度,可为子宫肌瘤术后复发个体化预防措施的制定提供一定临床指导价值。

Objective: To analyze the risk factors of recurrence of hysteromyoma after uterine artery embolization, and to establish a nomogram model for predicting recurrence. Methods: 314 patients with hysteromyoma treated by uterine artery embolization from March 2014 to July 2017 were analyzed retrospectively. The patients were followed-up for 1-3 years after operation, and the recurrence rate was statistically analyzed. According to whether the recurrence occurred, these women were divided into recurrence group and non recurrence group. Logistic regression model was used to screen the risk factors of recurrence. Based on the risk factors, the nomogram prediction model of recurrence was established by R software, and the accuracy and consistency of the model were evaluated. Results: A total of 67 of 314 patients recurred with 21.3% recurrence rate. Binary Logistic regression analysis showed that age(OR 0.923, 95% CI 0.868-0.980), the tumor diameter≥6 cm(OR 4.817, 95% CI 2.482-9.347), and the multiple hysteromyoma(OR 2.866, 95% CI 1.544-5.320) were the independent risk factors of hysteromyoma recurrence. The independent risk factors were introduced into R software(R 3.6.3) to construct nomogram model, the area under ROC curve was 0.860(95%CI 0.803-0.918), and the calibration curve was a straight line with slope close to 1, Hosmer-Lemeshow goodness-of-fit test was 8.528(P=0.384). Conclusion: The nomogram prediction model based on the risk factors of postoperative recurrence of hysteromyoma has good discrimination and accuracy, which can provide certain clinical guidance value for the formulation of individualized prevention measures of postoperative recurrence of hysteromyoma.