RESUMO
PURPOSE: Discrimination of uterine leiomyosarcoma (LMS) and leiomyoma (LM) prior to surgery by basic preoperative characteristics and development of a preoperative leiomyosarcoma score. METHODS: A predominantly prospective cohort of 826 patients with LM from a clinical institution and an outpatient center was included in the study. Further a predominantly retrospective cohort of 293 patients with LMS was included from the counseling database of the German Clinical Center of Excellence for Genital Sarcoma and Mixed Tumors (DKSM, University Medicine Greifswald, Germany). We analyzed and compared anamnestic, epidemiological and clinical findings between both cohorts. Tenfold cross-validated logistic regression and random forest was performed on the 80% training set. The preoperative LMS score (pLMS) was developed based on logistic regression and independently evaluated by analyzing the area under the receiver operating characteristic curve (AUC) with the 20% test set. RESULTS: In the LMS cohort, 63.1% had initially surgery for presumed LM and only 39.6% of endometrial biopsies revealed LMS. Key features for LMS discrimination were found to be bleeding symptoms: intermenstrual bleeding [RRc = 2.71, CI = (1.90-3.49), p < 0.001], hypermenorrhea [RRc = 0.28, CI = (0.15-0.50), p < 0.001], dysmenorrhea [RRc = 0.22, CI = (0.10-0.51), p < 0.001], postmenstrual bleeding [RRc = 2.08, CI = (1.30-2.75), p < 0.001], suspicious sonography [RRc = 1.21, CI = (1.19-1.22), p < 0.001] and the tumor diameter (each centimeter difference: ß = 0.24, SD = 0.04, p < 0.001). pLMS achieved a mean cross-validated AUC of 0.969 (SD = 0.019) in the training set and an AUC of 0.968 in the test set. CONCLUSIONS: The presented score is based on basic clinical characteristics and allows the prediction of LMS prior to a planned surgery of a uterine mass. In case pLMS is between - 3 and + 1, we suggest subsequent diagnostics, such as endometrial biopsy, color Doppler sonography, LDH measurement, MRI and transcervical biopsy.
