子宫内膜癌深肌层浸润相关因素分析及预测

目的:探讨临床病理特征与子宫内膜癌深肌层浸润的关系,同时建立Logistic回归预测模型以更好预测子宫内膜癌深肌层浸润情况。方法:回顾性收集274例经手术病理确诊的子宫内膜癌患者的术后临床病理资料,运用单因素和多因素Logistic回归分析确定子宫内膜癌深肌层浸润的高危因素并建立预测模型,用受试者工作曲线(ROC)确定截断值。结果:274例子宫内膜癌患者中,共有101例有深肌层浸润。多因素分析结果显示术前CA125血清水平>35 U/ml、肿瘤最大直径>3 cm、中/低分化是子宫内膜癌深肌层浸润的高危因素,其OR值分别为3.980(95%CI 1.998~7.926)、2.526(95%CI 1.470~4.339)、1.796(95%CI 1.035~3.116)。建立的Logistic回归预测模型为:Logit P=-2.521+1.381×术前血清CA125水平+0.927×肿瘤最大直径+0.585×组织分级。绘制回归模型预测概率的ROC曲线,其AUC为0.702(95%CI 0.636~0.768);取深肌层浸润概率P=0.32为截断值,其灵敏度和特异度分别为57.4%和79.2%。P>0.32时,深肌层浸润的可能性大。结论:联合术前血清CA125水平、肿瘤最大直径和组织分级这3个因素,利用Logistic回归预测模型有助于评估子宫内膜癌患者深肌层浸润的可能性。当深肌层浸润概率P>0.32时,深肌层浸润的可能性大。

Analysis of Related Factors and Prediction of Deep Myometrial Invasion in Endometrial Cancer

Objective:To explore the relationship between clinicopathological features and deep myometrial invasion of endometrial cancer,and to establish a Logistic regression prediction model.Methods:The postoperative clinicopathologic data of 274 patients with endometrial cancer diagnosed by surgical pathology were retrospectively collected.Univariate and multivariate Logistic regression analysis were used to determine the risk factors of deep myometrial invasion in endometrial cancer and to establish a predictive model.The cutoff point was determined using the receiver operating curve(ROC).Results:Among the 274 patients enrolled,101 were diagnosed with deep myometrial invasion.Multivariate analysis showed that preoperative CA125 serum levels>35 U/ml,tumor maximum diameter>3 cm,low/moderate differentiation were risk factors for deep myometrial invasion of endome-trial cancer.The OR values were 3.980(95%CI 1.998-7.926),2.526(95%CI 1.470-4.339),1.796(95%CI 1.035-3.116),respectively.The Logistic regression prediction model was established as Logit P=-2.521+1.381×preoperative serum CA125 level+0.927×maximum tumor diameter+0.585×tissue grade.The ROC curve of the regression model prediction probability was drawn with the AUC 0.702(95%CI 0.636-0.768.When P=0.32 was taken as the cut-off point,the sensitivity and specificity were 57.4%and 79.2%,respectively.When P>0.32,the possibility of deep muscle layer infiltration increased.Conclusions:Combined with preoperative serum CA125 level,tumor maximum diameter and tissue grade,the Logistic regression prediction model was feasible to assess the possibility of deep myometrial invasion in patients with endometrial cancer.When P>0.32,the possibility of deep muscle layer infiltration could increase.