Lower gastrointestinal function after surgery for deep endometriosis: A prospective cohort study

UPD, MH, and CC contributed to the study concept and design; UPD, CC, MH, CI, and CR performed patient recruitment; and UPD and MH performed the statistical analysis. Data interpretation, drafting of the manuscript, and critical revision of the manuscript were performed by UPD, CR, CC, GL, CI, AM, and MH.

Figure  1 depicts the study participant flow and Table  1 shows their baseline characteristics. Mean age of study participants was 34.2 ± 6.1. Mean BMI was 27.4 ± 7.5; 66 patients (45.2%; missing data = 3) had a normal BMI, 42 (29.5%) were overweight (BMI 25–29.9) and 15 (10.3%), 10 (6.8%) and 12 (8.2%) had class 1 (BMI 30 to <35), 2 (BMI 35 to <40), and 3 (BMI ≥ 40) obesity, respectively. Of the 145 patients who reported their parity, 105 (72.4%) were nulliparous, 19 (13.1%) had one child, and 21 (14.5%) had 2 or more children. A previous diagnosis of irritable bowel syndrome was reported by 20 patients (13.4%). During the three follow‐up reviews in the 12 months post‐surgery, hormonal therapy was used by 34%–41% of participants. The study participant flow. Baseline characteristics of study participants ( n  = 149) a Data are presented as mean ± standard deviation or as number (percentage). Body mass index (BMI) is calculated as weight in kilograms divided by the square of height in meters. Missing data: BMI = 3, Ethnicity = 7, Gravidity = 4, Smoking = 4, Coffee/Tea = 10, Alcohol = 7, Past diagnosis of endometriosis = 6. As reported by patient. Tables  2 and 3 shows the adjusted modeling of the summary scores and three individual domains not included in the summary scores (with their bother scores), comparing before and after surgery. Note that a reduction in score means an improved function. This model has shown a significant improvement of: BPSS at all times after surgery, BCSS at 12 months and BIQOLSS at 6 weeks and 6 months after surgery. Of the three individual domains, need to strain improved at 6 months with bother scores improving at 6 weeks and 6 months; thinking of bowel accidents and its bother score improved at 6 months; and although sexual activity restriction due to bowel was not changed compared with before surgery, its bother score significantly improved at all time points (of note, we have excluded from this analysis patients who responded “not applicable” to the question on sexual activity). Table  S1 shows the frequencies of the various bowel symptoms before and after surgery. Almost half of patients reported pre‐operative bowel urgency and pain around the back passage. Embarrassment related to bowel causing change of plans according to their bowel was reported by 37 patients (25%), 21 (14.1%) reported nocturnal defecation, 19 (13.5%) bowel staining, and 22 (14.9%) reported a not full control on accidental loss of stool. Mixed effects linear regression modeling of lower gastrointestinal function summary scores a , b An increase or positive change in score reflects worse function after surgery, while a reduction or negative change in score reflects improved function after surgery. Values are presented as mean ± standard deviation unless stated otherwise. Adjusted for age, body mass index, ethnicity, smoking, parity, hormonal treatment, revised American Society for Reproductive Medicine score and residual disease. Mixed effects linear regression modeling of lower gastrointestinal individual domains a Included are individual domains that do not comprise any of the summary scores. An increase or positive change in score reflects worse function after surgery, while a reduction or negative change in score reflects improved function after surgery. Values are presented as mean ± standard deviation unless stated otherwise. Figure  2 shows the change in predicted summary scores after surgery according to the scores before surgery. In all three summary scores, the improvements in scores were more when the initial score was higher (worse). (a) Effect of pre‐operative bowel pattern summary score on predicted post‐operative levels. (b) Effect of pre‐operative bowel control summary score on predicted post‐operative levels. (c) Effect of pre‐operative bowel impact on quality of life summary score on predicted post‐operative levels. Predicted post‐operative change in bowel pattern summary score (BPSS, a), bowel control summary score (BCSS, b) and bowel impact on quality of life summary score (BIQOLSS, c) and 95% confidence intervals according to the summary scores before surgery. A negative score reflects improvement in function. QOL, Quality of Life. The pouch of Douglas was not obliterated, partially obliterated, and completely obliterated in 40 (27%), 30 (20%), and 79 (53%) cases, respectively. Pouch of Douglas obliteration (partial or complete) was associated with significant progressive improvement in BPSS at each time point and with BIQOLSS at 6 weeks and 6 months after surgery compared with before surgery (Figure  3a ). (a) Change in summary scores according to Pouch of Douglas involvement. (b) Change in summary scores according to bowel surgery. Change in bowel pattern summary score, bowel control summary score, and bowel impact on quality of life summary score at the three time points after surgery, compared with before surgery, for patients with and without an obliterated Pouch of Douglas (a) and for patients who have or have not had direct bowel surgery (b). BMI, body mass index (calculated as weight in kilograms divided by the square of height in meters); POD, pouch of Douglas; rASRM score, revised American Society of Reproductive Medicine Endometriosis Score. In 55 (36.9%) patients, bowel surgery was carried out. Of these cases, 46 (83.6%) underwent bowel shaving and 9 (16.4%) underwent bowel segmental or disc resection. The mean bowel lesion size was 2.2 ± 0.8 cm—recorded for 46 (83.6%) bowel surgery cases. Bowel lesion extent was recorded in 51 (92.3%) bowel surgery cases. The deepest bowel layer involved was the serose, muscularis, and mucosa in 20 (39.2%), 28 (54.9%), and 3 (5.9%) patients, respectively. Although there were no major complications, five patients had intraoperative complications: two bowel serosal tears and one bladder cystotomy that were diagnosed and repaired intraoperatively, one unplanned oophorectomy due to ongoing bleeding and one uterine perforation. Four of these complications occurred in patients who underwent bowel surgery. Figure  3b shows the changes in summary scores of patients with and without bowel surgery compared with before surgery. BPSS significantly improved at all time points after surgery for patients who underwent bowel surgery, but not for patients without bowel surgery. Pain around the back passage and its bother scores improved for patients who had not had bowel surgery (pre‐operative score: 1.3 ± 1.0; mean change of −0.3 ± 1.0 ( P  = 0.02), −0.3 ± 1.0 ( P  = 0.02) and − 0.4 ± 1.2 ( P  = 0.01) at 6 weeks, 6 months, and 12 months after surgery, respectively). For patients who had bowel surgery, a lower pain score was observed only 6 weeks after surgery (pre‐operative score: 1.3 ± 1.0; mean change −0.4 ± 0.9, P  < 0.01). BIQOLSS significantly improved at 6 weeks for both groups but only for the bowel surgery group at 6 months. Only patients who underwent bowel surgery reported significantly lower scores at 6 weeks and 6 months in response to the question “How much do your bowels interfere with your everyday life?” (scale of 1–10). Pre‐operative score: 2.5 ± 2.8; mean change at six‐weeks and six‐months = −0.7 ± 2.0 ( P  = 0.01) and − 1.0 ± 2.2 ( P  = 0.01), respectively. Of note, when assessing patients who underwent bowel shaving and not bowel resection (n = 46): BPSS showed a stable and significant improvement at all time points after surgery—pre‐operative score: 4.5 ± 2.0; mean change of −0.7 ± 1.4 ( P  = 0.002), −0.6 ± 1.5 ( P  = 0.04), and − 0.7 ± 1.7 ( P  = 0.04) at 6 weeks, 6 months, and 12 months, respectively; and BIQOLSS significantly improved at 6 weeks and 6 months after surgery—pre‐operative score: 4.3 ± 5.5; mean change of −1.2 ± 3.1 ( P  = 0.01), −1.9 ± 4.1 ( P  = 0.01) at 6 weeks and 6 months, respectively.

This prospective study has shown that surgery for DE is associated with a significant improvement in LGF, specifically in bowel pattern and bowel impact on QOL. Higher initial scores were associated with larger improvements in scores post‐surgery, indicating a larger benefit from surgery when the initial function is worse. Even if not directly invading into the bowel, invasion of DE lesions into pelvic structures as well as extensive surgical pelvic dissection, may have an impact on pelvic organ function. 15 The ICIQ‐B is a validated questionnaire assessing colorectal function. Its BPSS improved significantly at all three time points after surgery with a slight trend of more improvement as time passed. A previous relatively small study using anorectal manometry and a defecatory function questionnaire in patients with colorectal DE found raised tone of the internal sphincter, an increase in the threshold to desire to defecate and incomplete evacuation, suggesting altered intestinal function in patients with colorectal DE. 16 They speculated that this may be related to chronic inflammation and pelvic pain. Our results demonstrating improved bowel pattern after surgery for DE may be related to the improvement in pain after surgery, less use of narcotics reducing constipation, restoration of normal anatomy, and to reduced inflammation. It is well established that general QOL can be severely impaired in patients with DE. 17 , 18 , 19 Recent cross‐sectional and retrospective cohort studies of 77 and 65 patients each, have reported an improved QOL after surgery for DE. 20 , 21 The ICIQ‐B questionnaire allowed us to specifically address the impact LGF has on QOL. Although the non‐significant improvement at 12 months could be attributed to the lower response rate at that final time point, it is also possible that the beneficial effect of surgery weakens as time passes. Colorectal resection for bowel DE is shown to have a positive long‐term influence on bowel symptoms and general QOL, 22 , 23 but it is associated with significant complication rates with possible long‐term functional implications, 24 driving many surgeons to explore other surgical methods to treat bowel DE. 25 Our cohort reflects the move toward less aggressive bowel surgery with shaving as the main surgical method of choice. Indeed, bowel pattern and impact on QOL significantly improved for patients who underwent bowel shaving. This is supported by a small study of 19 patients undergoing bowel shaving showing preservation of intestinal neurologic activity after surgery. 26 We are the first to use the ICIQ‐B questionnaire that specifically assesses bowel pattern, bowel control, and bowel impact on QOL. Although the small number of bowel resection procedures performed in our study has not allowed us to assess our secondary outcome and to compare bowel resection to bowel shaving, our findings of improvement in LGF after surgery are reassuring and support both the important role of surgery in the treatment of DE and the move from bowel resection to shaving. Our finding that patients with worse pre‐operative scores had the most benefit from surgery, supports a practice of limiting surgery to cases with more severely impaired LGF pre‐operatively. It follows that gynecologic surgeons should question patients considering surgical treatment of DE about the various aspects of bowel functional domains and counsel them on the possible effect of surgery on those functional domains. BCSS did not show post‐surgical improvement. This may be explained by the lower initial BCSS scores, reflecting less pre‐operative impaired function. This is supported by our finding that patients with an initial high BCSS, reflecting more severely impaired control before surgery, demonstrated a significant improvement after surgery. The main strength of this study is its prospective design. The ICIQ‐B questionnaire meticulously addresses a very wide range of LGF domains; however, there are no established ICIQ‐B normal values for LGF. The pre‐operative scores of the patients in our study were better than those reported in a study of 261 females and males with known bowel symptoms, 11 but the change in function following surgery is of clinical importance. Our relatively long follow up is another strength. However, we acknowledge the limited generalizability of a single‐centre study and the possible variation between surgeons from different institutions. Due to our clinic's move from segmental resection to more bowel shaving we were not able to compare functional outcomes between patients undergoing bowel shaving and patients undergoing segmental resection. We have not collected data on all treatments undertaken before surgery, nor on imaging studies before surgery. We also have not prospectively collected data on certain operative outcomes such as blood loss, operation time, and hospital stay and recognize these as limitations of the study. This was a paper‐based study with patients needing to post us their questionnaires. This may explain the lower response rate at 12 months after surgery, which is a limitation. The lower response rate may introduce a selection bias in which patients with worse function might be less likely to cooperate with completion of a study questionnaire. Although we have intentionally included the entire range of DE and not focused only on bowel DE, this resulted in a wider than usual range of disease severity making our findings able to be generalized beyond patients with bowel DE only. In conclusion, we found a positive influence of surgery for DE on LGF. Specifically, patients with initial worse function have shown greater improvement in LGF, lending further support to surgical intervention when needed. More and longer‐term prospective studies are needed to confirm our findings. Further research to characterize the groups of patients for which surgery may have a positive, equivocal, or negative effect on their bowel function, may aid clinicians to better counsel patients with DE who are contemplating surgery.

Endometriosis, estimated to affect 10% of reproductive aged women, 1 is an under‐recognized cause of gastrointestinal symptoms. 2 Endometriosis‐associated gastrointestinal symptoms include abdominal pain, cyclical rectal bleeding, cyclical dyschezia, pain with a full bowel, abdominal bloating, constipation, and diarrhea. 3 Deep endometriosis (DE), considered to be the most severe form of endometriosis, describes infiltrative forms of the disease and may involve any pelvic or abdominal structure. DE may infiltrate the colon directly or effect it indirectly, such as when the pouch of Douglas is obliterated. As such, DE is very commonly staged as stage 4 disease, 4 as per the revised American Society for Reproductive Medicine (rASRM) endometriosis scoring system. 5 Even without direct bowel involvement, DE is strongly associated with gastrointestinal symptoms. 6 Although pharmaceutical treatments are a viable option, when medical treatment fails and deep lesions do not improve, laparoscopic surgery is the treatment of choice. 7 Surgery for DE is associated with relatively high incidence of surgical morbidity 8 and therefore poses a therapeutic dilemma for the treating gynecologist. Although surgery to treat DE has been shown to improve pain symptoms, there are limited data on lower gastrointestinal functional outcomes before and after surgery for DE. Hence, the aim of this study was to prospectively document and compare long‐term lower gastrointestinal function (LGF) before and after surgery for DE.

The authors have none to declare.

We conducted a prospective cohort study between November 2016 and January 2020 at a tertiary university‐affiliated hospital in Victoria, Australia. The study is part of a larger project also exploring bladder function, sexual function, and quality of life (QOL) after DE surgery. The primary outcome was to assess change in LGF following surgery for DE. We also aimed to compare LGF among patients who underwent colorectal shaving and those who underwent resection. Patients who attended the Endometriosis clinic were offered the chance to participate in this trial. This unit consists of more than 10 gynecologists who specialize in pelvic pain, endometriosis, and gynecologic laparoscopy. All surgeons are highly experienced with endometriosis surgeries and use similar surgical methods. Recruitment included patients aged 18–50 years with pelvic pain or infertility, planned to undergo laparoscopy for suspected DE, based on imaging or clinical findings. Our clinic has a policy of combined decision making between doctor and patient after education and offering all reasonable treatment options. We usually offer our patients the options of no treatment, at least two types of medical treatment, and surgery. The decision to perform surgery is taken with the patient only after a thorough consultation in our endometriosis and pelvic pain clinic. Each patient undergoing surgery is then presented and discussed in our weekly meeting in which all unit consultants participate. Only after this meeting is surgery approved. All patients undergo imaging before surgery, which may include a normal ultrasound scan, an endometriosis dedicated ultrasound scan or magnetic resonance imaging. DE was defined as endometriosis situated more than 5 mm deep to the peritoneum. 9 If DE was not found at surgery, the patient was excluded from follow up. The study was approved by the Royal Women's Hospital Human Research and Ethics Committee and all participants provided informed consent. Before surgery, patients completed a questionnaire including demographic details and the International Consultation on Incontinence Questionnaire Anal Incontinence Symptoms and QOL Module (ICIQ‐B). Although originally designed to evaluate bowel incontinence symptoms, it provides a robust measure of LGF. 10 The questionnaire includes 21 items. Most items can be used to comprise summary scores as follows: (1) five items comprise the bowel pattern summary score (BPSS, scale 1–21); (2) seven items comprise the bowel control summary score (BCSS, scale 0–28); and (3) four items comprise the bowel impact on QOL summary score (BIQOLSS, scale 0–26). Four additional items do not form part of any summary score because they are considered essential extra items for assessment. 11 Each individual item is divided into two sections: presence of the specific symptom (scale of 0–4) and a bother scale (0–10). In all cases, higher score indicates worse function. At laparoscopy, the pelvis and abdomen were systematically examined, adhesions were divided, and endometriosis lesions were excised. Although we do not operate using the nerve‐sparing surgical technique, when pelvic nerves were identified, they were preserved. In cases of bowel involvement, bowel surgery was performed if this was the pre‐operative plan. Our practice is to recommend bowel surgery in patients with significant gastrointestinal symptoms thought to be attributed to endometriosis. If bowel surgery was performed, one of the following surgical methods was used: (1) segmental resection; (2) disc resection: full thickness removal of a bowel lesion smaller than 5 cm, with stapling of the anterior bowel wall 12 ; (3) bowel shaving: removal of disease from the bowel wall without breaching the lumen. 13 Bowel surgery—(1) and (2)—were performed by a team comprising a colorectal surgeon and a gynecologist. Immediately following surgery, a study operation report was filled, including a detailed description of surgical findings and the rASRM score 14 To assess for change in LGF after surgery, the ICIQ‐B questionnaire was sent to participants to complete at 6 weeks, 6 months, and 12 months post‐surgery. Continuous variables were summarized as mean ± standard deviation, and categorical variables were presented as absolute counts and percentages. Bowel scores reported at four time points over 12 months were correlated for each patient. Linear mixed effects modeling with subjects as a random effect and time point as a fixed effect were used throughout to adjust for this correlation. Each of the three summary scores was modeled initially with time (pre‐surgery, 6 weeks, 6 months, 12 months) as a fixed effect and patient identity as the random effect adjusting for the correlated nature of the measures. This model was adjusted with the potential confounders of age, body mass index (BMI; calculated as weight in kilograms divided by the square of height in meters), ethnicity, smoking, parity, hormonal treatment, rASRM score, and residual disease. Sub‐analyses of adjusted models were performed for: (1) patients with or without pouch of Douglas obliteration; (2) patients who did or did not have bowel surgery; and (3) assessing impact of pre‐surgery score on post‐surgery scores. For each time point, the coefficients with 95% confidence intervals and two‐sided P values were reported. When comparing the scores before and after surgery, a lower score after surgery indicates improvement in function. Sample size was calculated for the comparison of functional outcomes after bowel shaving and bowel resection. However, this outcome was abandoned due to slow recruitment and therefore this study became an exploratory study of the effect of surgery on LGF. The statistical software packages SPSS 25.0 (IBM) and STATA 16 (Stata Corporation) were used for analyses.

Table S1 Click here for additional data file.