基于子宫全切术的宫颈癌竞争风险预后模型的构建及评价

Construction and evaluation of a competing risk prognostic model for cervical cancer patients with hysterectomy

目的构建及评价基于子宫全切术的宫颈癌竞争风险预后模型。方法基于监测、流行病学和最终结果(SEER)数据库,选取于2010年至2013年诊断为宫颈癌并选择基于子宫全切术治疗的3 246例患者为研究对象,收集人口学特征和肿瘤相关信息。运用竞争风险模型进行单因素和多因素分析,找出独立的预测因素构建列线图。采用C指数、校准度图和决策曲线评价模型的预测能力和临床收益。结果多因素分析结果显示,人种(黑人:HR=1.50,95%CI:1.11~2.04)、肿瘤分化程度(中度分化:HR=1.95,95%CI:1.20~3.16;低度分化:HR=3.67,95%CI:2.24~5.99;未分化:HR=4.65,95%CI:2.39~9.05)、肿瘤大小(HR=2.61,95%CI:2.08~3.27)、区域淋巴结受累(HR=1.69,95%CI:1.12~2.53)和国际妇产科联盟(FIGO)分期(II期:HR=1.95,95%CI:1.38~2.76;III期:HR=4.94,95%CI:3.18~7.67;IV期:HR=10.19,95%CI:3.68~28.21)是基于子宫全切术的宫颈癌患者预后的独立预测因素(P<0.05)。以其建立的列线图的C指数为0.83(95%CI:0.82~0.84),校准度图中宫颈癌特异性死亡的预测发生率与实际发生率具有较为良好的一致性,决策曲线显示列线图具有良好的净收益。结论基于子宫全切术的宫颈癌竞争风险预后模型具有良好的预测性能和临床收益,可用于宫颈癌的临床决策和预后判断。

Objective To construct and evaluate a competing risk prognostic model for cervical cancer patients with hysterectomy. Methods Based on the surveillance, epidemiology and end result(SEER) database, a total of 3246 patients diagnosed as cervical cancer between 2010 and 2013 and undergoing hysterectomy, were selected as research objects. The demographic characteristics and tumor-related information were collected. The competing risk model was applied to conduct univariable and multivariable analysis, to select out the independent prognostic variables to build the nomogram. C-index, calibration plots and decision curve analysis were used to evaluate the predictive ability and clinical utility of the nomogram. Results In the multivariable analysis, race(Black: HR=1.50,95%CI:1.11-2.04),grade(moderately differentiated: HR=1.95,95%CI:1.20-3.16;poorly differentiated: HR=3.67,95%CI:2.24-5.99;undifferentiated: HR=4.65,95%CI:2.39-9.05),tumor size(HR=2.61,95%CI:2.08~3.27),regional lymph node involvement(HR=1.69,95%CI:1.12-2.53),and international federation of gynecologists and obstetricians(FIGO) stage(II:HR=1.95,95%CI:1.38-2.76;III:HR=4.94,95%CI:3.18-7.67;IV:HR=10.19,95%CI:3.68-28.21) were independent predictors for the prognosis of cervical cancer patients with hysterectomy(P<0.05). The C-index of the nomogram was 0.83(95%CI:0.82-0.84),the predicted incidence of cervical cancer specific mortality was in good agreement with the actual incidence in the calibration plots, and the decision curve showed that the nomogram had good net benefit. Conclusion The competing risk nomogram for cervical cancer patients with hysterectomy was of good predictive ability and clinical utility, which could be used in the decision making and prognosis judgment of cervical cancer.