Anatomical Compartment Involvement Predicts the Need for Rectal Resection in Deep Infiltrating Endometriosis: A Compartment-Based Analysis from Standardised Intraoperative Mapping at a High-Volume Tertiary Centre

Purpose To quantify how anatomical compartment involvement—derived from standardised intraoperative mapping at 29 pelvic sites—predicts the need for segmental rectal resection in deep infiltrating endometriosis (DIE). We also characterise the causal pathway from compartment pattern through rectal resection to surgical outcomes. Patients and Methods: We conducted a retrospective cohort study of 398 women who underwent surgery for DIE with ovarian endometrioma at a high-volume tertiary centre (June 2008–June 2016) where all lesions were recorded on a fixed 29-site intraoperative form. Twenty-nine sites were mapped to five anatomical compartments: posterior, central/uterine, adnexal, lateral/parametrial, and anterior. The primary outcome was rectal resection (yes/no). Multivariable logistic regression with bootstrap internal validation assessed the ability of compartment pattern (with age and rASRM stage) to predict rectal resection; model discrimination was compared by area under the receiver-operating characteristic curve (AUC). Secondary outcomes (blood loss, operating time, length of stay [LOS]) were analysed with generalised linear models. Causal mediation analysis (bootstrap, 1 000 replicates) quantified the indirect effect of compartment involvement on LOS transmitted through rectal resection. Results Of 398 patients, 93 (23.4%) underwent segmental rectal resection. Posterior compartment involvement was present in 212 patients (53.3%) and was strongly associated with rectal resection (adjusted OR 4.28, 95% CI 1.87–10.8, p  < 0.001). The primary logistic model (five compartments + age + rASRM stage) achieved an AUC of 0.819 (95% CI 0.765–0.873), with a bootstrap-corrected AUC of 0.804; Hosmer–Lemeshow goodness-of-fit p  = 0.576. In the surgical-outcomes analysis, rectal resection independently increased blood loss by 42% (rate ratio 1.42, 95% CI 1.14–1.78, p  = 0.003), prolonged operating time (exponentiated coefficient 1.36, 95% CI 1.24–1.50, p  < 0.001), and lengthened LOS by 46% (rate ratio 1.46, 95% CI 1.27–1.69, p  < 0.001). Causal mediation analysis showed that posterior compartment involvement increased LOS by a total of 5.3 days (95% CI 3.6–7.2), of which 2.1 days (95% CI 1.3–3.3; 40% of total effect) were mediated through rectal resection. Conclusion Compartment-based analysis of standardised intraoperative mapping identifies posterior compartment involvement as the dominant driver of rectal resection in DIE, with an AUC near 0.82. Rectal resection in turn mediates approximately 40% of the total effect of posterior involvement on hospital stay. These anatomically derived predictors translate directly to preoperative imaging targets, supporting compartment-based risk stratification and multidisciplinary operative planning.