Effect of Surgical Treatment for Deep Infiltrating Endometriosis on Pelvic Floor Disorders: A Systematic Review with Meta-analysis

Objectives  To evaluate the impact of surgical treatment of deep infiltrative endometriosis (DIE) on pelvic floor dysfunction (urinary incontinence [UI], pelvic organ prolapse [POP], fecal incontinence [FI)] or constipation, and sexual function [dyspareunia]). Data Source  The present systematic review was performed in the PubMed database. For the selection of studies, articles should be published by January 5, 2021, without language restriction. Study Selection  Six randomized controlled studies that evaluated surgical treatment for DIE and the comparison of different surgical techniques were included. Data Collection  The studies were selected independently by title and abstract by two authors. Disagreements were resolved by a third author. All included studies were also evaluated according to the Cochrane risk of bias tool and the quality of the evidence was analyzed using the GRADE criteria. Subgroup analysis by different treatments and follow-up periods was also performed. Results  Six studies were included in the quantitative analysis. The risk of bias between studies showed an uncertain risk of bias for most studies, with concealment of allocation being the least reported category. The quality of the evidence was considered low. High heterogeneity was found between the studies. No study has evaluated UI or POP comparatively before and after surgery. Conclusion  Dyspareunia and FI have improved after the surgical procedure, but it was not possible to demonstrate which surgical technique was related to these outcomes as there was surgical heterogeneity. This diversity was found across data, with the recommendation of future prospective studies addressing pelvic floor disorders with DIE.


Introduction
Endometriosis affects 10% of the female population. Its main symptoms are pelvic pain and infertility. For pain, women may refer dysmenorrhea, chronic pelvic pain, dyschezia, dysuria, and dyspareunia. 1 For deep infiltrative endometriosis (DIE), gastrointestinal manifestations (between 3.8 and 37%) 2 can be more intense and with major repercussions. The urinary tract can be involved (bladder endometriosis) in between 0.3 and 12% of the cases and may also compromise the quality of life of women. 3 The literature has several systematic reviews on the impact of conservative and/or surgical treatment of DIE. However, pelvic floor dysfunctions before and after treatment of endometriosis are not so deeply explored. When surgery is performed without focusing on nerve-sparing techniques or without carefully revising the anatomy, the risk for urinary incontinence (UI), fecal incontinence (FI), and other dysfunctions are possibly increased. An observational study assessing 138 women with DIE has shown that the presence of endometriosis in the bladder was an independent predictor of low bladder compliance, whereas the presence of endometriosis in the parametrium was predictor of voiding dysfunction. 4-6 A recent systematic review has found that colorectal surgery for endometriosis has a significant impact on urinary function regardless of the technique. 6 We can even find in the literature an association between bladder endometriosis and UI. 5 However, we do not have data pooled and analyzed into a systematic fashion, with analysis of the quality of evidence about pelvic floor dysfunctions and DIE or bladder endometriosis. Given that, we sought to systematically review the literature for studies that addressed pelvic floor dysfunctions with DIE before and/or after treatment.

Search Strategy
The present systematic review was performed according to the preferred reporting items for systematic review and meta-analysis (PRISMA) guidelines 7 (►Fig. 1) and was registered in the PROSPERO database (CRD42020197049).
We have included randomized controlled studies that assessed surgical treatment for DIE and have compared the utilized techniques. We have excluded all studies that did not analyze pelvic floor dysfunctions, case reports, animal and/or experimental studies. The following outcomes were included: UI or FI, defined by self-report or any measurable, validated, or nonvalidated scale or questionnaire, following the IUGA/ICS recommendations for pelvic floor dysfunction terminology; pelvic organ prolapse (POP), whether symptomatic or by physical examination, constipation, and dyspareunia. Quality of life questionnaires related to pelvic floor dysfunctions were also included.

Resumo
We have consulted the PubMed database on 5 January, 2021; no studies were excluded due to language restrictions. The following strategy was utilized: (((((((urinary incontinence) OR (incontinence)) OR (fecal incontinence)) OR (constipation)) OR (pelvic organ prolapse)) OR (prolapse)) OR (urodynamics)) OR (pelvic floor muscle)) OR (dyspareunia) AND (endometriosis). We intended to produce a broad search strategy because we hypothesized that we would have difficulties to retrieve data.

Study Selection, Data Extraction, and Risk of Bias
Studies were independently selected by title and abstract by two authors (Fraga M. V. and TAAM). Discordances were solved by a third author (Brito L. G. O.). Data extraction was performed in a previous spreadsheet pilot-tested and blinded for both authors. All included studies were also assessed according to the Cochrane risk of bias tool 8,9 and the quality of evidence was analyzed by the GRADE criteria. 9 Risk of bias analyzes five domains (selection, attrition, report, and other biases). The GRADE criteria consider the strength of other recommendations according to the presented variables.

Data Analysis
Meta-analysis was considered when at least two studies could be pooled. Heterogeneity was classified according to the i 2 test 10 and a random-effect model was applied to data when i 2 was > 50%. Continuous variables were described as mean difference plus standard deviation (SD). Some outcomes were described as median plus interquartile ranges (IQRs) and their data were transformed into mean plus SD according to the following formula (median ¼ mean, SD ¼ IQR/1.35). Dichotomous variables were transformed into odds ratio (OR) plus 95% confidence intervals (CIs) with lower and upper limits. A subgroup analysis before and after treatment was performed for each treatment and pooled into forest plots. No funnel plots were built to assess publication bias as we did not have enough studies to perform this analysis. Statistical analysis was revised by Review Manager version 5.4 (The Cochrane Collaboration, Copenhagen).
The surgical techniques compared were laparoscopically assisted or open colorectal resection surgery 11 ; conservative surgery (shaving or disc excision) or radical rectal surgery (segmental resection) [12][13][14] ; laparoscopy treatment using electroablation versus CO 2 laser ablation, 15 and conservative surgery alone or conservative surgery and presacral neurectomy. 16 The primary outcomes of the studies were characterized by the relief of dyspareunia and the evaluation of gastrointestinal symptoms (constipation and fecal loss). The other outcomes proposed by the review were not analyzed as they were not found during data selection/extraction.

Risk of Bias and GRADE Evaluation
Three studies reported having performed a sample size calculation. 12,14,16 Two studies 12,13 included the intention to treat analysis. Three studies [14][15][16] presented uncertain risk for randomization. For allocation bias, Daraï et al. 11 had an uncertain risk and Roman et al. 12 presented a low risk, while the others had a high risk. One study 15 reported high risk of bias and two were categorized as low-risk. 11,12 Regarding the   Roman et al. 12 evaluated the results after conservative surgery (shaving or disc excision) and radical rectal surgery (segmental resection), with no difference between groups after 24 months (median dyspareunia of 3 (2 to 3) and 4 (3 to 6), respectively; p ¼ 1.00). In another study by the same group, 13 comparing shaving, disc excision or segmental resection 5 years after the surgery, they demonstrated a reduction in dyspareunia, with no statistically significant difference between the surgical techniques.

Results from Individual Studies
Posadzka et al. 15 compared electroablation versus laparoscopy CO2 laser ablation and found an improvement in dyspareunia at 3 months after surgery; however, at 6 months, there was an increase in the symptom score within both groups.
Candiani et al. 16 compared the surgical techniques of conservative surgery alone or conservative surgery with presacral neurectomy. The authors have found a reduction in moderate and severe dyspareunia and an increase in the number of asymptomatic women in both groups. However, they have concluded that presacral neurectomy did not add significant improvement in the performance of conservative surgery alone.

Gastrointestinal Symptoms
Only three studies 12-14 evaluated the complaint of FI, classifying it as an involuntary loss of gas or feces, and they are from the same group. Roman et al. 12 compared shaving/disc excision versus segmental resection and, after 24 months, they found an improvement in FI symptoms within both groups, but with no difference between them (p ¼ 0.83). They also used assessment of gastrointestinal symptoms using the GIQLI score (low scores are related to a worse result) and, after treatment, the scores increased in both groups, but with no significant difference between them (p ¼ 0.64). Wexner scores before and after treatment behaved the same way (p ¼ 0.42).
The second study 13 presented a longer follow-up period (5 years) and the authors have also noticed symptom improvement, but with no difference between groups (p ¼ 0.42). The presurgical evaluation using the GIQLI, KESS, and Wexner score questionnaires showed improvement in the functional results, but with no difference between the groups. A third study 14 has found the same results.

Subgroup Meta-analysis
In the subgroup meta-analysis, it can observed that there was a decrease in dyspareunia after surgical intervention (MD: -0.82 [-1.05--0.59] (p < 0.00001) (►Fig. 3); however, an important heterogeneity is found as each study represents a different intervention (Chi 2 : 30.31; I 2 : 87%). The same can be observed for constipation (►Fig.

Discussion
Although the present review found studies that addressed the effect of the surgical treatment of DIE on pelvic floor dysfunctions, the heterogeneity of the studies did not make it possible to gather and analyze all the data. Within the subgroup analysis, it was possible to observe the benefits of surgical treatment for some pelvic floor disorders (dyspareunia and FI), but without superiority for a technique. According to the GRADE tool, the quality of the evidence was very low for both symptoms evaluated, that is, reduction of dyspareunia and improvement of gastrointestinal symptoms. None of the selected studies evaluated the presence and/or alteration of UI and POP. Among the studies that analyzed dyspareunia, although most of them suggested a reduction in this symptom after surgical treatment, one of them 15 revealed a resurgence of the symptom at the same level after 6 months, indicating the need for long-term evaluations. Through the meta-analysis, it was possible to confirm the results presented individually by the authors; however, the high heterogeneity among them is noteworthy. In the same direction, a recent systematic review that included only two surgical techniques (laparoscopic excision compared with laparoscopic ablation) for endometriosis and their effects on dyspareunia showed that both reduced the symptom, with no difference between the two techniques. 17 Likewise, we can point out that studies are scarce in the analysis of the dyspareunia response; they are even more restricted to gastrointestinal symptoms, such as FI. Although we have demonstrated, through meta-analysis, the improvement of symptoms of FI, the evidence is also not robust enough to indicate the superiority of one technique over another, with important heterogeneity between studies. Considering noncomparative studies, Erdem et al., 18 in a cohort study of 48 women with DIE, assessed long-term functional results (postoperative bowel movement and FI) after rectal resection, showing improvement in FI. A cohort study by Riiskjaer et al. 19 that evaluated 128 patients, before and after laparoscopic intestinal resection, also observed an improvement in the evacuation procedure 1 year after surgery.
Gastrointestinal symptoms usually present before surgical intervention, according to some authors, can predict postoperative results, which are worse the greater the severity of symptoms, indicating that surgical removal of the lesions may not completely reduce the symptoms. 20,21 Such data indicate that symptoms related to the pelvic floor should be evaluated before the surgical procedure. Their presence can directly interfere with functional results after surgery, requiring long-term follow-up.
We did not find data regarding dysfunctions related to UI and POP that could be included in a robust methodological analysis, although the literature draws attention to the risk of impaired urinary control when DIE is surgically treated. 22 Considering the extent of endometriotic lesions and the extent of surgical procedures performed, a potential effect on such pelvic floor dysfunctions may occur.
Considering that one of the most important indications for the surgical treatment of DIE is the control of pain symptoms, the present review has its main strength in demonstrating that surgeries, regardless of the technique used, can reduce dyspareunia and intestinal complaints, but also it has its greatest weaknesses when it demonstrates the great heterogeneity between the studies about the comparator group and the different instruments used to evaluate the results, as well as differences between the follow-up period across studies. Thus, groups of experts must meet and indicate methodologies that guide the authors when planning and executing prospective controlled studies to treat symptomatic women with DIE, evaluating the possible implications on pelvic floor dysfunctions.

Conclusion
Dyspareunia and FI improved after the surgical procedure, but it was not possible to demonstrate which surgical technique was related to these outcomes, as there was surgical heterogeneity. This diversity was found in the data, recommending future prospective studies addressing UI, POP and FI so that more robust evidence can be provided to health professionals about the association of DIE and pelvic floor disorders.