Gastrointestinal function outcomes following radical and conservative colorectal surgery for deep endometriosis: A systematic review and meta-analysis

Ezgi Darici : conceived the study, designed the protocol, performed the literature search, selected the studies and extracted the relevant information, wrote the first draft of the paper. Attila Bokor : conceived the study, designed the protocol, critically revised successive drafts of the paper. Daria Pashkunova : designed the protocol, performed the literature search, selected the studies and extracted the relevant information, wrote the first draft of the paper. Birgit Senft : synthesized the data. Nilüfer Cimşit : designed the protocol, wrote the first draft of the paper. Gernot Hudelist: conceived the study, selected the studies and extracted the relevant information, critically revised successive drafts of the paper.

This work was supported by Austrian Society of Endocrinological Oncology (OEGEO).

A total of 128 articles were obtained following the database search. After removing duplicates, 79 studies were screened for title and abstract. The full text of 55 articles was assessed for eligibility. Thirty‐nine studies were excluded and the reason for the exclusion is depicted in Figure  1 . Eventually, 16 studies were included in the qualitative synthesis but only 13 of them reported eligible data to be analyzed in the meta‐analysis. Figure  1 shows a PRISMA flow diagram of the selection process and the exclusion criteria. PRISMA flow diagram of the selection process and the exclusion criteria. Nine retrospective cohort studies, 11 , 19 , 20 , 24 , 25 , 26 , 27 , 28 , 29 three prospective studies, 5 , 30 , 31 one cross‐sectional study 32 and three randomized controlled trials (RCTs) 33 , 34 , 35 were included in the systematic review enrolling 1846 patients; 799 (43.3%) underwent SR and 1047 (56.7%) underwent conservative surgery (FTDR and/or shaving). The mean age was 33.5 ± 4.1 vs. 32.9 ± 3.2 years ( p  = 0.61) and the mean BMI was 23.7 ± 2.1 vs. 24 ± 2.9 ( p  = 0.87) in SR group and conservative group, respectively. Median follow‐up interval was 53.4 ± 24.2 months in the SR group and 49.2 ± 18.8 months in the conservative group ( p  = 0.61). Thirty‐seven percent had previous surgery for endometriosis in the SR group and 30.3% in the conservative group ( p  = 0.13). Only six studies reported on the anatomical height of the lowest rectal DE lesion with a mean height of the lowest nodule of 92.5 ± 5 mm from the anal verge (SR group) vs. 90 ± 20 mm (conservative group), respectively ( p  = 0.82). 5 , 27 , 29 , 33 , 34 , 35 Major complication rates according to Clavien‐Dindo were comparable among the two groups with 15.9% in SR patients vs. 10.3% in the conservative cohort ( p  = 0.15). Characteristics of studies included according to PICOS (Population, Intervention, Comparison, Outcomes, and Study design) criteria are depicted in Table  2 . Characteristics of studies included according to PICOS (Patients, Intervention, Comparator, Outcomes, Study Design) criteria. Abbreviations: CPP, chronic pelvic pain; CRADI‐8, colorectal‐anal distress inventory‐8; FTDR, full thickness discoid resection; GIQLI, gastrointestinal quality of life; LARS, Low anterior resection syndrome; SR, segmental resection; VAS, visual analog scale. Fourteen studies were found to report eligible data to be included in the meta‐analysis. Three studies were the follow‐up data of an RCT (ENDORE trial). 33 , 34 , 35 The remaining 11 studies were non‐randomized trials including questionnaires to assess the bowel function before and after the surgical intervention. 5 , 11 , 20 , 24 , 25 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 Regarding the quality assessment of the included studies, one study was ranked as “excellent”, 31 7 studies were ranked as “good” 5 , 11 , 20 , 27 , 28 , 29 , 30 and three studies 24 , 25 , 32 were ranked as “fair” according to the Downs and Black Score. Three RCTs were assessed according to RoB2 were found to be low risk. 33 , 34 , 35 Six studies reported on prevalence of postoperative constipation. 24 , 25 , 28 , 32 , 34 , 35 Three studies included in the final analysis reflected the follow‐up of the only RCT on this subject (ENDORE‐Trial). 33 , 34 , 35 Conservative surgery was less associated with constipation than SR (OR = 2.02; 95% CI [1.12–3.66], p  = 0.02) I 2  = 0%; Figure  2 ). Forest plot depicting postsurgical constipation comparing segmental resection (SR) vs conservative surgery. No difference was found in the occurrence of gas and stool incontinence between SR and conservative surgery in the included 5 studies (OR = 1.13; 95% CI [0.59, 2.15], p  = 0.72, I 2  = 27%; Figure  S1 ). Defecation pain was reported as the number of affected patients in five studies 24 , 30 , 33 , 34 , 35 and reported as mean values according to the numeric analog scale (NAS) in four studies. 5 , 11 , 28 , 31 In the studies reporting the number of affected patients, no significant difference was found between SR and conservative surgery (OR = 1.26; 95% CI [0.70, 2.25], p  = 0.44, I 2  = 4%; Figure  S2 ). Three studies reflected the follow‐up of the same RCT (ENDORE‐Trial). 33 , 34 , 35 In the studies where defecation pain was reported as mean values according to NAS, no difference was found between SR and conservative surgery (OR = 0.42; 95% CI [−0.35, 1.19], p  = 0,35, I 2  = 9%, Figure  S2 ). Conservative surgery was less associated with increased number of daily stool (>3/day) when compared to SR in the eligible six studies (OR = 2.67; 95% CI [1.55, 4.58], p = 0,0004, I 2  = 0%; Figure  3 ). Forest plot depicting postsurgical increased number of daily stool (>3/day) comparing segmental resection (SR) vs conservative surgery. No difference was found between two surgical intervention groups regarding time to defer defecation (≤ 15 min), (OR = 1.16; 95% CI [0.63, 2.12], p  = 0.64, I 2  = 0%) (Figure  S3 ). Few studies reported data using PROMs including GIQLI, KESS, Wexner and LARS to evaluate postsurgical GI function. Six studies reported on postoperative GIQLI 5 , 27 , 29 , 33 , 34 , 35 and found significantly increased GIQLI scores of patients treated with conservative surgery when compared to SR (OR = −5.76; 95% CI [−10.33, −1.18], p  = 0.01, I 2  = 66%). Four studies, three of which reflected the follow‐up of the ENDORE‐trial at 24 months, 60 months and 7 years reported data on KESS and Wexner scores without significant differences between SR and conservative surgery (OR = 0.74; 95% CI [−1.11, 2.60], p  = 0.07, I 2  = 58% and OR = 0.20; 95% CI [−0.08, 0.49], p  = 0.5, I 2  = 0%, respectively). 27 , 33 , 34 , 35 Five studies reported on postoperative LARS and found no statistically significant difference among the two groups (OR = 1.18; 95% CI [0.79, 1.76], p  = 0.41, I 2  = 25%). A detailed description of the studies is depicted in Table  3 and Forest plots of PROMs are listed in Figure  4 . Results of the patient‐reported outcome measures (PROMS) used to assess bowel function (GIQLI, KESS and Wexner and LARS) pre‐ and postsurgically. 10 (6.5–17.5) Roman et al. ENDORE Abbreviations: FTDR, full thickness discoid resection; GIQLI, gastrointestinal quality of life; LARS, low anterior resection syndrome; N/A, non‐applicable; SR, segmental resection. Scores were assessed by numeric analog scale. Forest plot depicting postsurgical PROMs used to assess GI function (GIQLI, KESS, Wexner and LARS score) comparing segmental resection (SR) vs conservative surgery. Out of 13 studies, three studies reported data on pre‐ and postoperative KESS and Wexner scores. All three studies were the follow‐up of the same RCT. 33 , 34 , 35 Four studies reported pre‐ and postoperative data on GIQLI 5 , 33 , 34 , 35 and one study on pre‐ and postoperative LARS (or LARS‐like symptoms). 5 Overall changes of pre‐ compared to postsurgical values of KESS, Wexner and GIQLI did not reveal significant differences between SR vs conservative surgery intervention groups (OR = −0.11; 95% CI [−2.77, 2.56], p  = 0.94, I 2  = 0%, OR = 0.10; 95% CI [−0.62, 0.83], p  = 0.78, I 2  = 0% and OR = 3.18; 95% CI [−2.57, 8.93], p  = 0.28, I 2  = 0%, respectively). It was not possible to conduct an analysis of pre‐ and post‐surgical LARS scores as there was only one study reporting pre‐surgical LARS Scores 5 (Figure  5 ). Forest plot of difference in pre and postsurgical PROMs used to assess GI function (GIQLI, KESS and Wexner score) comparing segmental resection (SR) vs conservative surgery. Regarding pre‐ and postoperative pain symptoms, defecation pain was the most frequently reported symptom and analyzed in 8 studies, 20 , 24 , 25 , 29 , 31 , 33 , 34 followed by diarrhea, 28 , 29 , 30 , 33 , 34 constipation, 28 , 29 , 30 , 33 , 34 increased number of daily stool, 28 , 33 , 34 time taken to evacuate 28 , 33 , 34 and gas and stool incontinence. 28 , 33 , 34 Preoperative rectal bleeding and cyclic rectal bleeding were reported in one study 29 (Table  3 ). In order to exclude studies that might entail a possible multiplication effect on the meta‐analysis, the follow‐up studies of the ENDORE Trial 34 , 35 published in 2019 35 and 2022 34 were excluded in a subgroup analysis. Following exclusion, a significant increase in postoperative GIQLI was observed in conservative surgery group when compared to the SR group (OR = −6.14; CI [−11.30, −0.99], p  = 0.02). However; when comparing changes of pre‐ and postoperative variables of GIQLI, no influential effect was observed when excluding these studies (OR = 0.46; CI [−6.76,7.67], p  = 0.9). 34 , 35

Colorectal surgery for DE is a radical treatment option that includes SR techniques such as nerve‐ and vessel‐sparing SR as well as conservative procedures such as FTDR and shaving. 5 There is an ongoing debate regarding the benefits of one procedure over the other in terms of severe complications and GI function outcomes. 36 , 37 PROMs used in high‐quality studies to evaluate GI function in patients with endometriosis predominantly include GIQLI, LARS, KESS and Cleveland Clinic Fecal Incontinence Severity Scoring System/Wexner scores. 38 However, a recent consenus on the use and validity of PROMs to be used in studies evaluating outcomes of interventions in endometriosis therapy suggested GIQLI as the first choice instrument for evaluation of GI function. 39 The present work analyzed the available evidence on GI function outcomes following SR and conservative techniques by analyzing previously mentioned PROMs and pain symptoms. The results of the present work demonstrate lower postsurgical constipation rates and frequency of daily stools in women following conservative surgeries when compared to SR. Differences in constipation rates in favor of conservative surgery have also been reported in a meta‐analysis published by Quintairos et al. 37 This observation is based on the theory that shortening of the rectum by SR may influence colonic water absorption and cause greater damage to the intestinal autonomous innervation leading to increased constipation after SR. 40 The present work included three additional studies on post‐surgical constipation rates with the largest caseload analyzed by Mabrouk et al. 28 , 32 , 34 Interestingly, a sensitivity analysis excluding the work by Mabrouk et al. resulted in similar constipation rates following any type of intervention demonstrating that the statistical significance regarding lower rates of constipation following conservative surgeries was shifted primarily by this study. Mabrouk et al. reported on 297 patients who underwent rectal shaving, 33 patients who underwent FTDR and only 62 patients following SR. A possible explanation may be that patients with shaving surgeries will undergo more “superficial operations”, resulting in a lesser degree of constipation afterwards. However, surgical preference and experience towards one selected surgical approach may also influence postsurgical complication rates and sequelae due to the factor of surgical experience suggesting that these results need to be interpreted with caution. In addition, a recent follow‐up study of the RCT (ENDORE‐Trial) by Roman et al. reported insignificant postoperative results regarding constipation in the conservatively managed group when compared to SR. 33 , 34 , 35 The most common PROM used in included studies was GIQLI, 5 , 29 , 33 , 34 , 35 followed by the KESS and Wexner questionnaires which were used in four studies and three studies included follow‐up data from the same RCT, published by Roman et al. 33 , 34 , 35 Hudelist et al. provided data on pre‐ and postoperative LARS scores (LARS‐like symptoms). 5 Four papers reported exclusively on post‐surgical LARS scores. 11 , 20 , 29 , 31 Studies exclusively reporting postsurgical PROMs show a significant increase in postoperative GIQLI scores in favor of conservative surgery when compared to SR. No relevant difference could be observed between SR and conservative surgery for other PROMs. However, when baseline, that is, presurgical GIQLI scores were considered and the changes between the pre‐vs postsurgical scores were analyzed, no difference could be observed between the radical surgery, that is, SR vs conservative approaches. The lack of relevant differences in changes of GI function following SR and conservative surgery as demonstrated by this analysis may be explained by several factors. First and foremost, postsurgical GI function may not solely depend on the surgical technique but on the degree of overall radicality of removal of disease including adenomyosis and DE affecting other sites. Residual adenomyosis and peripheral and central sensitization has been shown to influence autonomous nerve function and will therefore also have implications on GI function. 41 , 42 Secondly, factors such as a general inflammatory state caused by DE and fixation of the rectum by secondary adhesions have been suggested to contribute to GI function impairment. 43 As a consequence, the effect of surgical adhesiolysis as part of colorectal surgery will also contribute to changes in GI function independent of the approach used. Thirdly, the notion of a high presurgical prevalence of GI function impairment reflected by LARS‐like symptoms present in nearly 30% of colorectal DE patients before surgery and the postsurgical worsening of LARS sores in only 4.1% of women following colorectal surgery by either SR or FTDR suggest that GI function needs to be interpreted in the light of preexisting presurgical impairment already present and caused by DE. 5 Reporting preexisting PROMs on GI function is relevant as a variety of GI symptoms are common in patients with colorectal endometriosis. 44 , 45 The present work may have some limitations. Firstly the studies included do have different follow‐up intervals. Since bowel function (or dysfunction) may change over time we can not exclude a possible influence of heterogenous follow‐up intervals on the results of this meta‐analysis. Another possible limitation of the present meta‐analysis is the influence of studies including multiple follow‐up data from the same study cohort. 34 , 35 In order to exclude a possible multiplication effect, we performed a subgroup analysis and observed a benefical effect of conservative surgery in favor of GIQLI. However, we want to underline that this effect still was not present when comparing pre‐ and post operative means which may be more important in terms of clinical relevance. Finally, surgical experience is a factor that will always influence surgical outcome data. This factor was not evaluated in the present anaylsis and could not be evaluated for given reasons. To best to our knowledge this study is the first systematic review and meta‐analysis analyzing changes in pre‐vs post‐surgical GI function outcomes. Nevertheless, the data presented in this systematic review and meta‐analysis have to be interpreted in the light of a relevant heterogeneity that exists throughout the included studies as different PROMs were used by different groups (KESS, Wexner 33 , 34 , 35 and LARS 5 , 11 , 20 ) at different follow‐up intervals.

At first sight, radical surgery, i.e., SR appears to be associated with increased postsurgical constipation rates and increased number of daily stool (>3/day) and lower GIQLI scores when compared to conservative surgery. However, when presurgical GI function parameters and their changes following surgery are taken into account, the results of this analysis do not suggest a benefit of conservative surgery over SR.

Bowel endometriosis can be observed in 3% to 37% of the women presenting with endometriosis. 1 Women suffering from colorectal deep endometriosis (DE) frequently report gastrointestinal (GI) function impairment. 2 Within this, up to 70% of patients with rectal endometriosis report defecation pain and over 50% of the patients experience cyclic constipation. 3 Furthermore, the presence of colorectal DE does cause symptoms similar to lower anterior resection syndrome (LARS) per se. 4 , 5 Patients who undergo surgery for symptomatic colorectal endometriosis may continue to experience digestive complaints. 6 Dubernard et al. evaluated the quality of life following colorectal surgery for DE and observed significant improvements in gynecological and some digestive symptoms. However, constipation and tenesmus increased in up to 27% of these women 7 which is supported by Benbara et al. regarding an increased prevalence of tenesmus postoperatively. 8 As a consequence, neither SR (SR) nor conservative colorectal surgery like full thickness discoid resection (FTDR) or rectal shaving may provide amelioration of digestive complaints. 9 Although the exact pathomechanisms causing digestive dysfunction following colorectal surgery are still a matter of controversy, impaired function of the autonomous nerve supply of the rectum, moderate stenosis of the colorectal anastomosis and colorectal intussusception are suggested etiologies that may contribute to impairment of GI function. 10 Another constraining component is the absence of preoperative normative data on GI function of patients who undergo colorectal surgery for DE, since LARS‐like symptoms are observed in 18.5% (major LARS) and 27.8% (minor LARS) of patients with untreated colorectal DE. 11 In line with this, Roman et al. reported that baseline rates of severe constipation are relevant factors exerting influence on postsurgical functional outcomes. 12 The true impact of different surgical approaches on the digestive complaints of patients with colorectal DE appears increasingly complex. We aimed to evaluate the current evidence comparing GI function outcomes following radical, that is, SR vs conservative surgery such as FTDR and shaving for colorectal DE.

The authors have no conflicts of interest to declare.

The present study was conducted following the Preferred Reporting Items for Systematic Reviews and Meta‐ Analyses (PRISMA) guidelines. 13 A PRISMA checklist detailing the adherence to reporting standards is provided as supplementary information. The protocol of this review has been registered in the international prospective register of systematic reviews (PROSPERO, CRD42023476251). To identify eligible studies, a literature search was performed in the electronic databases United States National Library of Medicine and National Institutes of Health (PUBMED Medline), Excerpta Medica database (EMBASE), Web of Science, Clinical Trials.gov and the Cochrane Database, covering the period from January 1, 2010 until April 1, 2024. The search strategy used the following MeSH terms: (“bowel endometriosis” or “colorectal endometriosis”) AND (“surgery for endometriosis” or “conservative management” or “radical management” or “colorectal resection” or “shaving” or “full thickness resection” or “disc excision”) AND (“bowel function”). The inclusion criteria for this review were set using the PICOS (Patients, Intervention, Comparator, Outcomes, Study Design) formulation, and studies were considered if they met the following criteria: (1) population type: patients with colorectal endometriosis who underwent colorectal excision of the disease; (2) intervention type: segmental resection (SR); (3) control type: full thickness discoid resection (FTDR); (4) outcome type: bowel function reflected by either symptom reporting and reported by validated tools evaluating bowel function; (5) study design type: clinical trials and observational studies. Studies were excluded from this review if one or more of the following exclusion criteria were present: (1) studies reporting a review of literature or only reporting one arm of surgical intervention without the control group; (2) studies in another language than English; (3) studies describing a novel technique; (4) case‐series and case reports. Two independent reviewers (E.D. and D.P.) screened the titles and abstracts of all identified articles for relevance. Any discrepancies were resolved through discussion or consultation with a third reviewer (G.H.). The bibliographic records were downloaded and imported into Endnote software program (Ver X9.3.3) to save, remove duplicates, and analyze the search results. Data extraction was performed from included studies as follows: study characteristics (name of the authors, year of publication, design of the study, sample size), characteristics of study participants (mean age, body‐mass index, previous surgery for endometriosis, mean follow‐up time in months), data on surgical technique, that is, type of the anastomosis, height of the lowest nodule (mm from the anal verge), major complication rates according to Clavien‐Dindo classification, 14 abnormal bowel symptoms (constipation, gas/stool incontinence, defecation pain, increased number of daily stool >3/day, time to defer defecation ≤15 min) and the scores of the validated instruments used in the extracted studies ( Knowles Eccersley Scott Symptom Score [KESS]), 15 Wexner score, 16 Low Anterior Resection Syndrome (LARS) 17 and Gastrointestinal Quality of Life Index (GIQLI). 18 The Knowles‐Eccersley‐Scott Symptom questionnaire (KESS) was used to diagnose constipation, which is composed of 11 individual items with a maximum of 39 points. Higher scores indicate increased symptom severity. 15 The Wexner scale is composed of 5 items and scores from 0 to 4. It is used to measure the functional impact of fecal incontinence. The score 0 corresponds to continence and 20 represents anal incontinence. 16 LARS Score was originally designed to assess quality of life following low anterior resection for rectal cancer and includes 5 questions regarding bowel function, consequently dividing patients into three groups: no LARS with 0–20 points, minor LARS: 21–29 and major LARS: 30–42. 17 This tool was shown to be also useful in evaluating patients after radical colorectal surgery for DE. 11 , 19 , 20 The GIQLI was a self‐administered questionnaire comprising 36 questions that scored from 0 to 4, being 0 classified as worse and 4 as best quality of life. Two independent reviewers assessed the risk of bias in the included studies using validated instruments. The quality of included studies was assessed by Downs and Black Checklist. 21 This checklist includes 27 criteria, widely covering areas reporting quality, external and internal validity, and power. Downs and Black score ranges were given quality levels as: excellent (26–28); good (20–25); fair (15–19); and poor (<14) (Table  1 ). Conflicts regarding study quality were resolved by the authors. Risk of bias in the included RCT's was assessed by Cochrane risk of bias tool for randomized trials (RoB 2) (Table  S1 ). 22 Risk of bias regarding assessment of the included non‐randomized controlled studies according to the Downs and Black Score. Y Y Y Y Y Y N N Y Y Y Y N N Y Y Y Y N N N Abbreviations: N/A, non‐applicable; N, no; Y, yes. Analyses were performed using Cochrane Collaboration's Review Manger Version 7.5.0. Comparisons were performed when at least 2 studies could be pooled into a forest plot (random effects model). For all metric outcomes, mean values and standard deviations were extracted pre‐ and post‐surgery from the studies. If medians and quartiles or medians with min and max were available, mean values and standard deviations were calculated using the Excel file by Wan et al. 23 The standard deviation of the differences of pre‐ and post‐surgical values for the two intervention groups was calculated using correlation coefficients from the study by Hudelist et al. 5 The results are expressed as Odds Ratio (95% CI confidence interval) for dichotomous data and mean differences (95% CI) for metric data.

Table S1. Figure S1. Figure S2. Figure S3.