Transvaginal Sonography for the Preoperative Assessment of Parametrial Deep Infiltrating Endometriosis: A Diagnostic Accuracy Study

Objective: To evaluate the diagnostic accuracy for parametria endometriosis (PE) of transvaginal sonography (TVS) performed following a systematic approach for the assessment of the lateral parametria. Design: Diagnostic accuracy study based on a prospective observational design Participants: All consecutive patients who underwent laparoscopic surgery for endometriosis between January 2016 and December 2020. Setting: Endometriosis referral hospital Methods: We prospectively collected clinical, imaging, and surgical data of all consecutive patients who underwent laparoscopic surgery for endometriosis between January 2016 and December 2020. A standardized technique with a systematic approach for the assessment of the lateral parametria following specific anatomic landmarks was used for the TVS. The diagnostic accuracy for PE of TVS was assessed using the intraoperative and pathologic diagnosis of PE as the gold standard. Results: In 476 patients who underwent surgery, PE was identified in 114 out of 476 patients (23.95%): 91 left and 54 right PE out of 476 surgical procedures were identified (19.12% vs. 11.34%; p=0.001); bilateral involvement in 27.19% (31/114 patients) cases. The sensitivity of TVS for PE was 90.74% (79.70%-96.92%, 95% CI) for the right side and 87.91% (79.40%-93.81%, 95% CI) for the left side. The specificity was almost identical for both sides (98.58% vs. 98.18%). For the right parametrium, the positive likelihood ratio (PLR) and negative likelihood ratio (NLR) were 63.82 (28.70-141.90, 95% CI) and 0.09 (0.04-0.22, 95% CI), respectively. On the left parametrium, the PLR and NLR were 48.35 (23.12-101.4, 95% CI) and 0.12 (0.07-0.21, 95% CI), respectively. The diagnostic accuracy for right and left PE was 97.69% (95.90%-98.84%, 95%CI) and 96.22% (94.04%-97.74%, 95% CI), respectively. Limitations: The principal limit is the high dependence of TVS on the operator experience. Therefore, although a standardized approach following precise definitions of anatomical landmarks was used, we cannot conclude that the observed accuracy of TVS for PE is the same for all sonographers. In this regard, the learning curve was not assessed. In the case of negative TVS for parametrial involvement with an absent intraoperative suspect, a complete dissection of the parametrium was not performed to avoid surgical complications; therefore, cases of minor PE may be missed, underestimating false negatives. Conclusions: TVS performed following a systematic approach for assessing the lateral parametria seems to have good diagnostic accuracy for PE with large changes in posttest probability of parametrial involvement based on the TVS evaluation. Considering the clinical and surgical implications of PE, further studies implementing a standardized approach for assessing the parametrium by TVS are recommended to confirm our observations and implement a standardized protocol in clinical practice.