Ureteral endometriosis, the hidden enemy: multivariable fractional polynomial approach for evaluation of preoperative risk factors in the absence of ureteral dilation

Objective: To determine whether it is possible to predict the risk of ureteral endometriosis (UE) using a mathematical model based on preoperative findings. Design: Prospective observational study conducted between January 2017 and April 2020. Setting: Tertiary-level academic referral center. Patient(s): Three hundred consecutive women of reproductive age with a diagnosis of posterior deep infiltrating endometriosis (DIE) scheduled for laparoscopic surgery. Intervention(s): Before surgery, anamnestic data and the severity of endometriosis-related symptoms were evaluated, and all patients underwent a complete gynecological examination. Transvaginal and transabdominal ultrasound were performed to map the endometriotic lesion. Ureteral involvement was surgically and histologically confirmed. Main Outcome Measure(s): To select important risk factors for UE and determine a suitable functional form for continuous predictors, we used the multivariable fractional polynomial. Results: UE was surgically found in 145 women (48.3%). Based on our multivariable polynomial mathematical model, UE was significantly associated with adenomyosis, parametrial involvement, and previous surgery for endometriosis. A posterior DIE nodule with a transverse diameter >1.8 cm was associated with a higher probability of ureteral involvement. Conclusions: Posterior DIE nodule with a transverse diameter >1.8 cm, adenomyosis, parametrial involvement, and previous surgery for endometriosis appear to be good predictors of UE.