Endometriosis is a frequent gynecological condition affecting up to 10% of women of reproductive age, although its true prevalence remains uncertain [1]. It is characterized by ectopic growth of tissue similar to the endometrium beyond the uterine cavity [2]. More recently, broader definitions have emerged to capture the systemic and multifactorial nature of endometriosis [3]. Endometriosis is a well-known etiology of chronic pelvic pain, dysuria, dyspareunia, digestive symptoms, and infertility impacting on quality of life [4]. DPE is thought to affect about 20% of women with endometriosis [5, 6].

First-line management typically involves medical therapies such as analgesics, and hormonal suppression [7]. Surgical intervention is considered when medical treatment fails or when infertility or urinary or intestinal obstruction are present [8–10]. Surgical procedures should consists in complete removal of endometriotic tissue in order to maximize benefits and reduce the risk of recurrence [11–13]. Because these procedures are technically demanding and carry a risk of major complications such as urinary or rectovaginal fistulae [14–16], they should be restricted to experienced reference centers [17].

Several classification systems have been developed to aid preoperative evaluation and shared decision-making [18]. Although widely adopted in clinical and research settings, the revised ASRM classification remains limited as it mainly reflects the extent of superficial and ovarian disease, without adequately assessing deep endometriosis, thereby restricting its ability to guide surgical decision-making or predict functional outcomes (20–23). In 2021, the Enzian classification was proposed combining a complete staging of DPE with evaluation of peritoneal, ovarian, tubal locations, deep pelvic endometriosis lesions, and the presence of adenomyosis [19, 20]. When evaluating the reproducibility of the Enzian classification applied to MRI, Manganaro and al (2021) reported an excellent inter-reader agreement for peritoneal, ovarian endometriosis and adenomyosis but a moderate concordance for DPE [21]. The dPEI was designed as an MRI-based framework for standardized evaluation of the extent of deep pelvic endometriosis. It divides the pelvis into central, lateral, and extra-pelvic regions, allowing a structured description of lesion location and severity. Based on the number of compartments involved, disease is categorized as mild (≤ 2), moderate (3–4), or severe (≥ 5) [22]. This tool was externally validated in a multicenter cohort, showing its value for predicting surgical complexity and postoperative outcomes, and thus for supporting patient counseling and surgical planning [23]. However, dPEI was evaluated in patients undergoing conventional laparoscopy rending the extrapolation to robotic assisted laparoscopy (RAL) questionable. Indeed, previous studies have underlined the contribution of RAL to improve the vision, the precision and optimized gestures with a potential decrease in intra- and postoperative complications including the decrease in the incidence of high stage complications according to the Clavien-Dindo classification, decrease of voiding dysfunction while increasing the rate of complete resection [24, 25].

Therefore, the aims of the present study were to evaluate the relevance of the dPEI to predict operative time, postoperative complications, hospital stay, and voiding dysfunction after RAL for DPE.

All pelvic MRIs were acquired at 1.5 T or 3 T with a phased-array coil, following a high-resolution protocol (T2- and T1-weighted sequences with and without fat suppression). Gadolinium was administered only when clinically indicated, in accordance with French HAS–CNGOF recommendations [23, 26]. An experienced radiologist (SM), unaware of the patients’ clinical and surgical information, independently reviewed all MRI images.

Endometriosis lesions were classified according to the dPEI framework, which divides the pelvis into nine anatomical compartments—anterior, middle, and posterior regions, each with central and lateral subdivisions—plus an optional category for extra-pelvic disease [23]. For each affected compartment, one point was assigned, with an additional point given if the disease extended laterally beyond the parietal fascia. The DPE lesions were reported as belonging to 1 of the 10 compartments. In the presence of substantial tethering and anatomic distortion due to severe DPE causing projection of a structure into a different compartment, the compartment of the initial location was considered the involved structure [22]. Finally, a dPEI score was determined in accordance with the previous study [23]. The cumulative score defines the severity group : mild (≤ 2), moderate (3–4), or severe (≥ 5).

Intraoperative staging of endometriosis was performed using the revised American Society for Reproductive Medicine (rASRM) classification system [27]. Robotic-assisted procedures were carried out with the Da Vinci Xi following previously published protocols [25]. Depending on disease extent, surgery could include adnexal operations, uterosacral or parametrial resection, partial colpectomy, ureterolysis, cystectomy, hysterectomy, or bowel surgery such as shaving, discoid, or segmental resection.

All operations were conducted by senior surgeons (CT, YD, CF, ED), each with more than five years’ experience in endometriosis surgery. These surgeons had equivalent access to conventional laparoscopy and RAL. The choice between robotic and conventional laparoscopy depended on both RAL availability and surgeon preference. For RAL, an 8 mm trocar was placed trans-umbilically for the optics. On the same horizontal line, a 12- or 8-mm trocar was placed on the right for an ultrasonic energy scalpel, monopolar scissors or automatic stapling forceps and an additional 8 mm trocar on the right flank for a grasping forceps. On the left, an 8 mm trocar was placed for the bipolar forceps and a 5 mm conventional trocar for the operating aid [28]. A temporary protective stoma was created only when colorectal resection with associated colpectomy precluded safe interposition of peritoneal or mesorectal tissue between the vaginal and bowel anastomosis.

Surgical outcomes included the evaluation of operative time, hospital stay, urinary, digestive, vascular complications, and reoperations. The complications were stratified according to the Clavien-Dindo classification as low grade (I-II) or high grade (III-IV) during the post operative hospitalization and late up to the first postoperative month.

A total of 170 patients were included on between February 2019 and December 2024.

The epidemiologic characteristics of the study population are summarized in the Table 1. The median patient age was 36 years old ,with most being nulliparous. Among the 157 patients with prior surgery, 70 patients (44.5%) had a history of gynecological surgery. The main symptoms reported by the patients were dysmenorrhea (87%), dyspareunia (70%), and chronic pelvic pain (48%). All the patients had multiple symptoms suggestive of endometriosis.

Surgical procedures and complications according to compartment involved by endometriosis are summarized in the Table 2. The median operative time and hospital stay were 181 min (± 95.3) and 4.5 days (± 4.1), respectively. Colorectal surgery was the most frequent intervention, performed in 141 patients (82.9%), encompassing 54 discoid excisions, 44 rectal shavings, and 43 segmental resections. Conversion to open surgery occurred in 2.9% of cases. The rate of digestive stoma was 5.9%. Postoperative complications were absent in 128 patients (75.3%). Among the remaining 42 patients (24.7%), the distribution of the postoperative complications according to the Clavien-Dindo classification was 69% with grade II (29 patients), 23.8% with grade III (10 patients), 1.8% with grade IV (3 patients), and no patient underwent a grade V complication.

The overall rate of de novo voiding dysfunction was 10.6% (18 patients) and the rate of voiding dysfunction beyond one month was 4.1% (7 patients).

The distribution of operative time, complication rate according to the Clavien-Dindo classification, and hospital stay according to the pelvic compartment involved by endometriosis is summarized in the Table 2.

Mediocentral compartment was associated with a higher operative time (p = 0.007), a longer hospital stays (p = 0.04), but without increased incidence of grade III complication neither in the rate of voiding dysfunction.

Posterocentral compartment was associated with a higher operative time (p < 0.001), a longer hospital stays (p < 0.001), but without increased incidence of grade III complication neither in the rate of voiding dysfunction.

Anterocentral compartment was not associated with an increase in operative time, hospital stay, grade III complication, nor in voiding dysfunction.

Mediolateral compartment was associated with a higher operative time (p < 0.001), a longer hospital stays (p < 0.001), a trend for a higher incidence of voiding dysfunction (p = 0.05), but without increased incidence of grade III complication.

Posterolateral compartment was associated with a higher operative time (p = 0.03), a longer hospital stays (p < 0.001), a higher incidence of voiding dysfunction (p < 0.001), and a trend for in increased incidence of grade III complication (p = 0.1).

The number of patients with involvement of the anterolateral compartment was too low to evaluate its impact on operative time, hospital stay, incidence of grade III complication nor on voiding dysfunction.

Extrapelvic endometriosis was associated with a higher operative time (p = 0.02), a longer hospital stays (p = 0.04), but without increased incidence of grade III complication neither in the rate of voiding dysfunction.

Operative time significantly increased with higher dPEI categories: 140 min (mild), 202 min (moderate), and 240 min (severe) (p < 0.001). This trend showed a positive correlation (Spearman ρ = 0.40, p < 0.001) (Fig. 1). Similarly, length hospital stay was associated with the dPEI category score, with an increase in median length of hospital stays: 3 days ,5 days and 6 days respectively for mild, moderate and severe endometriosis, respectively (p < 0.01). The difference was significant regardless of the groups of dPEI compared, with a positive correlation (Spearman’s ρ = 0.43, p < 0.001) (Fig. 2). The frequency of severe complications rose with the severity of the dPEI: 1.6% for mild, 7.8% for moderate, and 16.7% for severe cases (p = 0.02). Severe complications occurred more frequently in patients with severe endometriosis compared with patients with mild endometriosis (OR = 13.1; 95% CI [1.54–111.3]). De novo voiding dysfunction occurred in 4.7%, 7.8%, and 23.8% of mild, moderate, and severe dPEI categories, respectively (p = 0.01). Voiding dysfunction occurred more frequently in patients with severe endometriosis compared with patients with mild endometriosis (OR = 5.9; 95% CI [1.48–23.5]).

This retrospective study confirms that the dPEI, initially validated for laparoscopic surgery, can also predict postoperative risk when applied to RAL for DPE.

In the current study, RAL was primarily selected for patients presenting with severe disease according to the dPEI classification. In contrast to the multicenter external validation, where only 13.1% of patients had a dPEI ≥ 5, severe cases represented nearly one quarter of our series representing a potential bias to adequately evaluate the complication risk for this specific subpopulation. The prior validation study did not stratify outcomes by surgical approach, likely due to the predominant use of conventional laparoscopy. This limitation prevented extrapolation to RAL, although previous studies and meta-analyses have shown reduced hospital stay and postoperative complications, likely owing to better visualization and nerve identification with the robotic system, albeit at the cost of increased operative time [14, 24, 28–30]. In our study, when considering the pelvic compartments, the mediocentral and the postrocentral compartments were the most frequently involved in 91.8% and 59.4%, respectively. Interestingly, the mediolateral compartment, defined by the involvement of parametrium, ureteral dilatation, and pelvic wall, and the posterolateral compartment, defined by the involvement of distal portion of uterosacral ligaments, sacrorectal septum, and pelvic wall were involved in 47.1% and 30% respectively. This also contrast with the multicenter external validation of the dPEI where the mediolateral compartment involvement was noted in only 21.6%. This difference is crucial as lateral compartments are well recognized to be associated with the risks of ureteral, vascular and nervous injuries with a higher risk of voiding dysfunction [31–34].

In the present study, when comparing patients according to dPEI score categories, an increase in operative time, hospital stay, incidence of grade III complications according to the Clavien-Dindo classification was observed from mild to severe endometriosis (p < 0.001). A positive correlation was found with longer operative time (Spearman’s ρ = 0.40) and hospital stay (Spearman’s ρ = 0.43). The increment in operative time, hospital stay, and complication rate observed in our series aligns with findings from the multicenter external validation of the dPEI [23]. These data support the relevance of the dPEI scoring system as a predictive tool of surgical outcomes after RAL for DPE thus contributing to improve decision making. To date, only a few tools can be used to tailor preoperative counseling for women with DPE. Poupon and al. proposed a nomogram for predicting complication risk in deep endometriosis without bowel involvement but its discriminative performance remained modest (AUC = 0.72) [35]. Bazot et al. introduced the Endostage MRI classification, which aims to correlate the anatomical spread of lesions with surgical difficulty, although it still awaits multicenter validation [36].

Overall, 10.6% of patients developed de novo voiding dysfunction, rising to 23.8% among those with severe endometriosis—mostly associated with medolateral and posterolateral compartments involvement. Moreover, when considering the duration of voiding dysfunction inferior or superior to one postoperative month for the whole population, the rates were 10.6% and 4.1%, respectively supporting that about two-thirds of the patients recovered normal micturition in the first postoperative month. In contrast, in the dPEI multicenter external validation, the occurrence of a voiding dysfunction was observed in only 7.4% of the whole population and was considered a grade II complication while this complication is not included in the Clavien-Dindo classification. The higher rate of urinary dysfunction observed here may relate to the substantial proportion of colorectal resections, a known risk factor for postoperative voiding disorders [15, 16, 33, 37, 38]. Transient (< 1 month) voiding dysfunction may be considered minor, whereas persistence beyond one month reflects a severe complication that substantially affects quality of life and may require neuromodulation therapy [15, 32, 39–44]. In addition, a previous meta-analysis on colorectal resection for endometriosis has emphasized that the rate of voiding dysfunction depended on the type of colorectal surgery. Indeed, rectal shaving was less associated with voiding dysfunction than segmental colorectal resection (OR : 0.34; CI : 0.18–0.63, p < 0.001) or discoid excision (OR : 0.22; CI : 0.09–0.51, p < 0.001) while no difference was found between discoid and segmental surgery [15]. In our study, among the 150 patients with colorectal surgery (88% of the whole population), only 17.3% underwent a rectal shaving while the remaining patients underwent discoid excision or segmental resection. Moreover, previous studies have underlined the high rate of voiding dysfunction in case of vaginal and/or parametrial endometriosis that were required respectively in 22% and 41% of our patients [32, 33, 37]. Finally, in our study, when considering complications rates according to compartments involved, no differences between patients with a mediolateral and posterolateral compartment endometriosis were observed.

This study has several limitations that warrant consideration. First, its retrospective design limits the ability to exclude residual biases. Second, no attempt was made to evaluate among patients with a severe dPEI whether the risk of complication varied according to the type of colorectal surgery (rectal shaving, discoid excision and segmental resection). Third, no attempt was done to evaluate the predictive value of 4th postoperative day CRP in the detection of complication after colorectal surgery for endometriosis [45]. However, our rates of pelvic abscess and of rectovaginal fistulae (0.04% and 0.006%, respectively) were too low to allow a statistical evaluation. Lastly, the study did not include a cost-effectiveness analysis to determine whether the dPEI could serve as a practical selection tool for robotic surgery.

The dPEI proved to be a reliable MRI-based index for predicting postoperative outcomes in patients undergoing RAL for DPE. Applying this index preoperatively may enhance patient counseling and multidisciplinary decision-making, particularly for individuals with extensive disease (dPEI ≥ 5). Future prospective studies are warranted to determine whether the robotic approach can effectively reduce the incidence of postoperative voiding dysfunction. Finally, in the absence of solid evidence demonstrating the cost-effectiveness of robotic surgery, the dPEI could help identify patients with severe forms of endometriosis who might derive the greatest benefit from a robotic approach.