卵巢上皮性肿瘤超声鉴别特征及预测模型

Ultrasonographic diagnostic characteristics and prediction model of benign, borderline and malignant ovarian epithelial tumors

目的探讨卵巢上皮性肿瘤的超声鉴别特征,并建立预测模型。方法按照国际卵巢协作组的超声术语规范,回顾性分析2015年1月至2019年7月427例卵巢上皮性肿瘤声像图表现,通过单因素分析得出有鉴别意义的超声征象,纳入多元Logistic回归分析,得出交界性、良性、恶性超声鉴别重要指标,并建立预测模型。结果单因素及多元Logistic回归分析结果显示微囊征是交界性肿瘤与良性、恶性肿瘤鉴别的特征性指标(OR值分别为10.97、19.22)。同时恶性肿瘤与良性肿瘤鉴别的重要指标以不规则、厚分隔、实性为主、富血供、腹水为主,其中不规则意义最大。恶性肿瘤与交界性肿瘤鉴别的重要指标还包括不规则、大乳头、厚分隔、实性为主、富血供。不规则在鉴别交界性肿瘤与良性肿瘤也具有相当的作用。本研究建立模型总的符合率72.4%,其中预测交界性模型的符合率为57.2%,良性为78.6%,恶性为80.7%。结论微囊征是交界性肿瘤的特异性超声表现。卵巢肿瘤是否规则、是否以实性为主、血供是否丰富及有无腹水在鉴别卵巢上皮性肿瘤中意义较大。本研究建立的卵巢肿瘤良性、恶性和交界性预测模型有较高诊断效能。

Objective To explore ultrasonographic diagnostic characteristics of ovarian epithelial tumors and establish prediction models.Methods The ultrasonographic images of 427 cases from multicenter with ovarian epithelial tumors confirmed by pathology from January 2015 to July 2019 were retrospectively analyzed according to the International Ovarian Tumor Analysis(IOTA).Ultrasonographic signs with distinguishing significance were obtained through univariate analysis and included into multivariate Logistic regression analysis to obtain important ultrasonagraphic indicators for distinguishing borderline,benign and malignant ovarian tumors,and to establish prediction models.Results The microcystic pattern of papillary projections and solid components was the diagnostic characteristic between borderline and benign,malignant ovarian epithelial tumors(OR value 10.97 and 19.22,respectively).Irregular morphology,septa thickness,solid lesions,rich blood supply and ascites were diagnostic characteristics between benign and malignant tumors,with the irregular morphology having the highest value.Irregular morphology,large papillary,septa thickness and rich blood supply could be used to identify borderline and malignant tumors.At the same time,irregular morphology was the valuable sign to distinguish borderline and benign tumors.In this study,the total coincidence rate of the proposed model was 72.4%,among which the predicted coincidence rate of the borderline model was 57.2%,78.6%for benign,and 80.7%for malignant.Conclusions The microcystic pattern of papillary projections and solid components are the specific sonographic characteristics of borderline ovarian tumors.Irregularity,solid lesions,rich blood supply and ascites have important value in differentiating ovarian epithelial tumors.The prediction models of benign,malignant and borderline ovarian tumors in this study have higher diagnostic efficacy.