Evaluation In The Preoperative Diagnosis Of Ovarian Tumors, The Malignant Risk Index

Objective:To assess the ability of a risk malignancy index(RMI) based on serum levels of CA125,ultrasound findings,and menopausal status to discriminate between benign and malignant pelvic masses, and propose the reasons and directions of improving RMI ' s diagnose function levels.Methods:This is a retrospective study of 214 women admitted at the First Affiliated Hospital of Guangxi Medical University between January 2003 and September 2006 for surgical exploration of ovarian tumor,among the total,52 Gases of malignant tumor,17 cases of borderline tumor,145 cases of benign tumor.The serum CA 125 levels,ultrasound findings,and menopausal status were combined into the RMI1,RMI2,and RMI3 to diagnose ovarian cancer.Drown the ROC curve of CA125,Doppler ultrasound,RMI1,RMI2 and RMI3;compared the areas of under ROC curve of CA125,RMI1,RMI2 and RMI3 by parametric method;compared the areas of under ROC curve of CA125,RMI1,RMI2,RMI3 and Doppler ultrasound by nonparametric method;contrasted the diagnostic capability of CA125,RMI1,RMI2 and RMI3 at different levels.Contrasted the diagnostic capability of RMI and the combination diagnose involving CA125 and Doppler ultrasound; compared their sensitivity by chi-square criterion.Find the influence factors of ovarian tumor diagnose by binary logistic regression,to explore the method to improve RMI model.Drown the ROC curve of improved RMI model;compared two areas of under ROC curve of RMI model before and after improved by parametric method;find the best suitable cut-level of improved RMI1,RMI2,RMI3.Compared the areas of under ROC curve of improved RMI1,RMI2 and RMI3 by parametric method.Results:The areas under ROC curve of CA125,Doppler ultrasound,RMI1,RMI2 and RMI3 are 0.767,0.707,0.835,0.808,0.805 separately. The results of the comparison of two areas of ROC curve by parametric methods are:The areas under ROC-curve of CA125,RMI1,RMI2 and RMI3 all have differences(P＜0.05),the diagnostic performances order from high to low is:RMI1,RMI2,RMI3,CA125.The results of the comparison of two areas of under ROC curve by nonparametric method are:The areas under ROC curve of RMI1,RMI2 and RMI3 are bigger than Doppler ultrasound with differences(P＜0.05);The areas under ROC curve of Doppler ultrasound and CA125 have no differences(P=0.200).The diagnostic performance of RMI is better than CA125 at any different cut-off level.The diagnostic capability of RMI is higher thanthe combination-diagnose involving CA125 and Doppler ultrasound(P＜0.001).The result of binary logistic regression is:Postmenopausal state,multicystic,materiality,anomal-TSGF are influence factors of ovarian-mass diagnose.We improved the RMI model by adding anomal-TSGF factor into calculating. The areas under ROC curve of improved RMI1,RMI2 and RMI3 model are: 0.848,0.826,0.824 separately,higher than before.By parametric method comparing two areas of under ROC curve of RMI model between before and after improved are:P=0.0054,0.0015,0.0003 separately. the best suitable cut-off level of improved RMI is 200.Improved RMI1 is better than RMI2 and RMI3,the two latter have no different. Conclusions:The risk of malignancy index is able to correctly discriminate between malignant and benign ovarian tumor.It is a scoring system which can be introduced easily into clinical practice to facilitate the selection of patients for primary surgery.The diagnose performance of RMI1,RMI2 and RMI3 is better than CA125, which is proved by the comparison between two areas under ROC curve. The diagnose performance of RMI is better than the combination-diagnose involving CA125 and Doppler ultrasound.RMI need to be improved and it has widely improved space。...