Establishment and evaluation of a malignant transformation risk prediction model for perimenopausal ovarian endometriosis
Objective To construct a predictive model for the malignant risk of perimenopausal ovarian endometriosis(OE).Methods Clinical data of 236 patients with OE and 41 patients with endometriosis-associated ovarian cancer(EAOC)who were in perimenopause and diagnosed pathologically from January 2013 to December 2021 attending the Third Affiliated Hospital of Zhengzhou University,Henan Provincial People's Hospital and Henan Provincial Cancer Hospital were retrospectively analysed.Multivariate logistic regression was used to determine the risk factors for malignant transformation in perimenopausal OE patients.R software was used to draw and evaluate nomogram models.Results The modeling OE set and modeling EAOC set exhibited statistically significant differences in menopausal status,previous cyst removal history,self-merging cancer,direct family history of cancer,dysmenorrhea duration of 5 years or more,progressive dysmenorrhea,menstrual disorders,preoperative disease management of 3 years or more,infertility,mass tenderness,OE type,severe internal endometriosis,lesion side under ultrasound,blood flow signals in ultrasound-detected masses,nipple-like structures under magnetic resonance imaging,and distribution of combined bands under magnetic resonance imaging,as well as age at first sex,parity,age at first pregnancy and age at first delivery(X2=17.527,16.255,1.634,7.609,20.822,15.313,5.140,4.751,13.303,5.264,8.448,8.920,5.248,8.385,7.029 and 4.223;t=2.228,2.052,3.863 and 3.046,P＜0.1).The validation OE set and validation EAOC set also exhibited statistically significant differences in menopausal status,previous cyst removal history,increased menstrual volume,concomitant dysmenorrhea,dysmenorrhea duration of 5 years or more,progressive dysmenorrhea,preoperative disease management of 3 years or more,infertility,immobile masses,OE typy,severe internal endometriosis,blood flow signals in ultrasound-detected masses,thickening of cyst walls under magnetic resonance imaging,and distribution of nipple-like structures under magnetic resonance imaging(X2=3.558,20.707,5.004,5.920,10.616,2.325,6.727,4.964,4.234,5.384,1.251,2.325,7.892 and 6.950,P＜0.1).Multivariate logistic regression analysis showed that menopausal status(OR=5.007,95％CI:1.363-18.394),progressive dysmenorrhea(OR=3.583,95％CI:0.909-14.122),infertility(OR=22.489,95％CI:3.552-42.382),severe endometriosis(OR=3.800,95％CI:0.894-16.147),and blood flow signal in ultrasound-detected masses(OR=3.431,95％CI:1.044-11.280)were independent risk factors for malignant transformation of ovarian endometriosis during perimenopause(P＜0.1).The area under the curve(AUC)of the modeling set was 0.848(95％CI:0.769-0.928),and the AUC of the validation set was 0.880(95％CI:0.725-0.941).The calibration curves of the modeling set and the validation set fitted well with the ideal curve.The Hosmer-Lemeshow goodness-of-fit test results were X2=3.691,P=0.884 and X2=4.166,P=0.842,respectively.Conclusions Menopausal status,progressive dysmenorrhea,infertility,severe endometriosis and blood flow signal in ultrasound-detected masses are independent risk factors for malignant transformation in perimenopausal patients with ovarian endometriosis.The nomogram model established based on these indicators has a good discriminatory degree,with high accuracy and consistency between predicted and actual values.It has good screening ability for high-risk groups of malignant transformation.
万方数据
维普
