Delayed functional outcomes associated with surgical management of deep rectovaginal endometriosis with rectal involvement : giving patients an informed choice

Horace Roman1,5, Cecile Loisel1, Benoit Resch1, Jean Jacques Tuech2, Patrick Hochain3, Anne Marie Leroi4, and Loic Marpeau1 Department of Gynecology and Obstetrics, Rouen University Hospital, Rouen, France Department of Digestive Surgery, Rouen University Hospital, Rouen, France Gastroenterology, Clinique du Cèdre, Bois-Guillaume, France Department of Physiology, Rouen University Hospital, Rouen, France


Introduction
Rectal endometriosis is a chronic disease with a high risk of pain recurrence which can evolve over a 20 year period before diagnosis and management of rectal localizations (Roman et al., 2008).Surgical treatment of rectal endometriosis may be a difficult choice as patients are young, professionally active and plan to conceive.The removal of rectal endometriosis localizations is mainly conducted using either of two procedures: colorectal segmental resection and rectal nodule excision with or without opening the rectum.Both procedures are routinely performed by experienced teams worldwide and follow-up studies of 3-5 years indicate encouraging reductions in pelvic pain.
The choice of operative procedure depends on different parameters: size of the nodule, rectal circumference involved in the disease (Abra ˜o et al., 2008), frequency of multifocal intestinal nodules and of other associated deep lesions (Chapron et al., 2006) and the surgeon's experience and school of thought.The majority of surgeons on a daily basis choose to carry out colorectal resection in more than 90% of women presenting with symptomatic rectal endometriosis, strongly believing that the radical removal of digestive endometriotic foci is the most effective way of avoiding the risk of recurrences.This choice is supported by studies which show that microscopic endometriotic lesions usually exist around the main rectal nodule (Kavallaris et al., 2003;Remorgida et al., 2005).We recently showed that active glandular epithelium and stroma are responsible for deeper infiltration of rectal layers than that of fibrosis and smooth fibres related to rectal endometriotic nodules, and thus they are likely to be left out after fibrous nodule excision (Roman et al., 2009).The decision to perform colorectal resection is also supported by evidence of significant improvement in pain symptoms and in patient quality of life, shown by the results of numerous retrospective studies on women exclusively managed by this procedure (Chopin et al., 2005;Keckstein and Wiesinger, 2005;Daraı ¨et al., 2007;Minelli et al., 2009).
On the other hand, there are also several experienced teams primarily choosing to perform nodule excision rather than colorectal resection.In addition, surgeons such as Philippe Koninckx and Jacques Donnez have only on an exceptional basis, been performing rectal segmental resection in women presenting with rectal endometriosis (Donnez and Squifflet, 2004;Slack et al., 2007).This choice is based on strong arguments: surgical morbidity appears to be higher in women managed by colorectal resection (Daraı ¨et al., 2007;Mereu et al., 2007;Slack et al., 2007), post-operative functional digestive symptoms are expected to be less satisfactory after rectal removal (Ret Davalos et al., 2007), microscopic endometriotic foci may still be found on the limits of segmental resection suggesting that microscopically complete resection whereas always an aim, is rarely a reality (Roman et al., 2007;Anaf et al., 2009), and rectal resection does not prevent post-operative recurrences of pain (Vercellini et al., 2009).Furthermore, the clinical implications of leaving microscopic foci of endometriosis on the digestive tract remain unknown (Fanfani et al., 2009), with recent data suggesting that post-operative continuous medical treatment might be able to halt the progression of endometriotic implants left out and decrease the risk of recurrences (Seracchioli et al., 2009a,b).As far back as 1995, Donnez et al. wrote that it was time to curtail rather than encourage aggressive rectal surgery in rectovaginal endometriosis with rectal involvement (Donnez et al., 1995(Donnez et al., , 1997)).
Over the last 5 years, both surgical procedures have been performed by skilled surgeons in our Department of Gynecology and Obstetrics at Rouen University Hospital, France.Prior to November 2007, we systematically removed rectal endometriosis nodules using colorectal segmental resection and removal of involved posterior vaginal fornix.But November 2007 signalled a change in our choice of surgical procedure (Roman and Bourdel, 2009), and the decision was taken to primarily perform rectal nodule excision associated with systematic post-operative treatment by GnRH analogs followed by long-term continuous pill intake.Colorectal resection was reserved for any cases where nodules were of large volume rendering the excision and the suturing of the outer wall technically impossible or for women for whom post-operative amenorrhea was not desirable (desire to conceive immediately, or refusal of, or incompatibility with, prolonged hormone treatment).
These factors combine to form the basis of this retrospective study, which aims to evaluate delayed post-operative digestive and urinary functional outcomes associated with the surgical management of rectal endometriosis by colorectal resection or nodule excision, carried out by skilled surgeons in our department.The distinguishing feature of our study is that the choice of surgical procedure was not determined by the characteristics of the nodule.

Materials and Methods
Included in our series were women who benefited from surgical management of symptomatic rectal endometriosis in our department between January 2005 and June 2008, thus allowing for a follow-up period superior to 12 months.'Rectal endometriosis' was defined as being deep posterior endometriosis involving muscular, submucosal or mucosal layers of the rectum, up to 13 cm from the anus, indicated by MRI and endorectal ultrasound examination, and later intraoperatively confirmed.Deep endometriosis could also infiltrate the posterior vaginal fornix, torus uterinus and uterosacral ligaments and/or ureter, bladder and other abdominal organs.
All women were preoperatively examined by an experienced gynecologist.A detailed preoperative questionnaire was used to complete patient symptom history (Roman et al., 2008).Preoperative MRI and transrectal ultrasound examinations were performed for all women by physicians with considerable experience with deep endometriosis.When rectal involvement was suspected, the operative strategy was first discussed with both the patient and digestive surgeon before deciding on the surgical procedure to be used.
Prior to November 2007, we mainly performed colorectal segmental resection with colorectal anastomosis and omentoplasty, systematic resection of posterior vaginal fornix, and only in certain cases, transitory ileostomy for 10 -54 days.This choice was justified by a desire to provide microscopically complete removal of digestive nodules, expected to reduce the risk of rectal recurrences.During this period, only 6 patients out of 24 (25%) did not benefit from this radical procedure, as they specifically refused colorectal resection and requested nodule excision.The colorectal segmental resection procedure used was similar to that described in the literature by other teams, while colorectal anastomosis was performed laparoscopically using a single-use circular transanal stapler PCEA 28 device (Darai et al., 2005;Dubernard et al., 2006).Postoperative treatment by GnRH analogs and add back therapy was systematically prescribed for 3 -6 months, followed by continuous contraceptive pill intake in women not intending to conceive.
November 2007 marked a change in our surgical preference and general convictions about the disease.From this date onward, we considered that although with nodule excision the removal of microscopical rectal implants might be microscopically incomplete (Roman et al., 2009), a thorough relief of symptoms could be obtained by the surgical procedure associated with prolonged post-operative amenorrhea (Roman et al., 2007).Moreover, we felt that colorectal segmental resection was an overly complex procedure followed in some cases by pelvic nerve damage and subsequent unpleasant functional urinary and digestive symptoms in young patients (Roman and Bourdel, 2009;Vercellini et al., 2009).As a consequence, we recommended to our patients rectal nodule excision, instead of colorectal resection, associated with systematic post-operative treatment by GnRH analogs followed by long-term continuous pill intake.We reserved colorectal resection firstly, for any cases where excision was impossible, such as if there were large nodules requiring extensive opening of the rectum and where suturing would have obstructed the rectum and secondly, for women for whom post-operative amenorrhea was not possible.During the period from November 2007 to May 2008, 10 patients out of only carried out in seven nulliparas who refused prolonged amenorrhea by continuous pill intake.Colorectal resection was no longer performed in women who did not intend to conceive in the future and who accepted long-term post-operative medical treatment.In the case of women over 45 years old, post-operative medical treatment could be replaced by bilateral adnexectomy.The surgical procedure that we use in rectal nodule excision is similar to that performed by other surgical teams and as first described by Donnez et al. (1995).Both pararectal spaces are opened below the lateral limits of the rectal nodule and the rectovaginal space is reached under the inferior limit of the nodule.The nodule is then dissected away from the rectal wall, using the Ultracision Harmonic Scalpel (Ethicon Endosurgery, Cincinnati, USA).The dissection is made into the thickness of the rectal wall, in order to remove all abnormal fibrous lesions involving the rectal layers, using a high magnification endoscopic view.Partial or full thickness rectal wall defects are closed laparoscopically in one or two layers using resorbable sutures.At the end of the procedure, the site of rectal dissection is covered by an omentum flap, which is fixed by non-resorbable sutures (Roman et al., 2009).
Partial or complete resection of uterosacral ligaments was conducted whenever infiltrated, however, care was taken to spare the parasympathetic bladder nerves particularly on the side corresponding to the ligament least affected by the disease.
Although our stance concerning the surgical management of rectal endometriosis may be challenged by other authors, this present series allows a comparison of delayed functional outcomes related to both procedures, providing a long enough follow-up to evaluate outcomes which are most likely definitive.In June 2009, all patients were asked to complete a detailed questionnaire consisting of 25 items, including multiple choice questions concerning pelvic pains, digestive symptoms, additional procedures required after the primary surgery, fertility and global patient satisfaction.In accordance with French regulations, this retrospective study was exempted from IRB approval.
Statistical analysis was performed using Stata 9.0 Software (Stata Corporation, 4905 Lakeway Drive, TX, USA).Median values, percentiles, range, mean values and SD were calculated for continuous variables, and percentages were calculated for the qualitative variables.Parameter distributions, stratified on surgical procedure were compared by univariate analysis (Fischer's exact test in qualitative parameters and Mann-Whitney U-test in continuous variables).Pain evaluation using a 10-points analog rating scale before and after surgery was compared using a paired t-test.The actuarial life-table analysis and Kaplan-Meier method were used to estimate the probability of being free-of-recurrence depending on length of follow-up, and their 95% CI.A comparison of Kaplan-Meier curves was carried out using the LogRank test when appropriate.Box plots were used to represent scores of pre-and post-operative pain.P , 0.05 was considered statistically significant.

Results
Over 40 consecutive months, 41 women underwent surgical treatment of rectal endometriosis.They answered a post-operative questionnaire at the end of the follow-up (from April to May 2009).Patient characteristics are presented in Table I and are stratified by surgical procedure.Mean age was 33.6 + 6.7 years (range 23 -47).All patients presented preoperative dysmenorrhea for which the score using a 10-points analog rating scale was 7.7 + 1.7 (median 8, range 4-10).Of the 36 women (88%) who reported having had sexual intercourse during the previous 5 years, 33 presented deep dyspareunia, with a score of 5.8 + 2.9 (7, 0-10).There were 34 patients (83%) who presented non-cyclic pain, for which the score was 5.6 + 3.1 (7, 0 -10).Defecation pain during menses was a systematic complaint (100%), associated with either cyclic constipation or diarrhoea.There were 24 women who were nulliparas (59%), and primary infertility was previously recorded in nine cases (22%).
The type of surgical procedure was decided on the basis of clinical and radiological features prior to surgery, and this decision remained unchanged intraoperatively.Surgical procedures were performed  entirely by laparoscopy in 29 cases (71%), and by laparotomy in five cases (12%).In seven cases (17%), the gynecological stage of the procedure was carried out laparoscopically, followed by laparoconversion conducted by the surgeon performing digestive suture.Rectal endometriosis was removed by colorectal segmental resection in 25 women (61%) and by nodule excision in 16 (39%).Temporary ileostomia was performed in 16 women managed by colorectal resection and was maintained for 9-54 days.The posterior vaginal fornix was consistently excised, as its involvement in deep posterior endometriosis appears systematic (Donnez et al., 1997;Matsuzaki et al., 2009).To prevent rectovaginal fistulae, the rectum and vaginal suture were separated by interposition of fat tissue from omentum, mesorectum or pararectal space.Ureteral segmental resection followed by anastomosis and partial cystectomy were complementary procedures carried out, respectively, in two and one patients, as a result of associated ureteral and bladder infiltrations by the disease.Total hysterectomy was performed in eight cases due to a large infiltration of the posterior wall of the uterus or cervix in women who no longer intended to conceive and whose MRI results indicated suspected adenomyosis.The ages (mean, SD and range) of women who did and did not benefit from hysterectomy were, respectively, 38.1, 6.0, 27 -47 and 32.4,6.4, 23-43 years (P ¼ 0.03, Wilcoxon test).No post-operative rectovaginal or ureteral fistulae were recorded.Characteristics of the surgical procedures are presented in Table II.Follow-up was completed over 26 + 13 months (range 12 -53).Post-operative evolution of pelvic pain and functional digestive symptoms are presented in Table III.Regarding functional digestive symptoms, severe constipation (,1 stool/5 days) was recorded in three women having undergone segmental resection.Despite several endoscopic dilatations of colorectal anastomosis, secondary incomplete relief was observed in only one case.The number of patients presenting 3 stools/day was significantly increased in the group of women managed by colorectal resection 13 (52%) versus 3 (19%), P ¼ 0.02.The comparison of the daily number of stools was significantly superior in the colorectal resection group (Wilcoxon test, P ¼ 0.024).Two women managed by segmental resection presented bladder dysfunction requiring systematic or occasional bladder catheterization, respectively, at 40 and 12 months after surgery.
An additional 'intention to treat' analysis was carried out comparing the 24 women operated prior to November 2007 to the 17 women operated afterwards.The percentages of colorectal resections   12) 2 ( 12) 3 ( 12) 1 Ureter resection þ anastomosis 2 ( 5) 1 ( 6) 1 ( 4) 1 Temporary ileostomy 16 (39) 0 16 (64) ,0.001 performed before and after that date were, respectively, 75% (18 women) and 41% (7 women).Due to the small sample size and the carrying out of excisions before November 2007 and resections after, the comparison between the two groups with regard to the variables presented in Table III revealed no statistically significant differences.

Discussion
Through this retrospective study, bearing in mind the possible methodological weakness inherent to it, we observed that women presenting with rectal endometriosis and undergoing colorectal resection, when compared with women managed by nodule excision, were more likely to present several unpleasant functional digestive outcomes, such as a significant increase in the daily number of stools or rare but severe post-operative constipation.Urinary dysfunction might also be more frequent following colorectal resection, due to the enlarged dissection of pararectal spaces and vegetative nerve injuries.Conversely, we did not observe any significant difference in the improvement of pelvic pain between the two types of rectal surgery.This study was not randomized and therefore we were unable to control all confounding factors that might influence the rate of either constipation or frequent stools in each group.However, it is most likely that these outcomes are closely related to the surgical procedure itself (colorectal resection, excision, inadvertent section of pelvic vegetative nerves), and not influenced by other factors such as hysterectomy or surgical approach (laparoscopy or laparotomy).As with other reported series, our sample is too small to allow multivariate analysis and therefore randomization remains the best way to control for a majority of confounding factors.However, randomized    trials require preliminary data from retrospective studies such as this one, in order to accurately estimate sample sizes and expected outcomes.The results of this study therefore represent valuable data for both surgeons involved in the management of rectal endometriosis and researchers designing a randomized trial.
As rectal endometriosis is a benign disease and patient choice should be an important criterion in the decision concerning surgical management, we believe that women must be aware of two important findings.First, the expected rate of post-operative delayed unpleasant functional outcomes seems to be lower where nodule excision is carried out than with colorectal resection.Second, the advantage of a lower morbidity might not be at the cost of a higher rate of pain recurrences.To date, on the basis of small amount of available data from patients followed-up over 3 -5 years, post-operative pain recurrence rates appear to be comparable between women managed by the two techniques (Fanfani et al., 2009).
It may be argued that in daily practice, women with advanced stage diseases might have worse functional outcomes and be more likely to benefit from colorectal resection.If true, this could generate an artificial relationship between the type of surgery and functional outcomes.This may have been the case with a recent study by Fanfani et al. (2009), who found that segmental resection was associated with a significant risk of bladder dysfunction, post-operative fever and a tendency towards constipation.The patients included in their two groups were not however totally comparable due to the systematic choice of discoid resection in women whose nodule size was 3 cm, although segmental resection was performed in women with larger nodules.Consequently, this study did not allow the authors to demonstrate whether or not performing disc resection in large nodules is associated with less post-operative functional morbidity than that observed in the group undergoing segmental resection.It should be emphasized that in our series the choice of surgical procedure was at no time related to the size of rectal nodule but was determined by our convictions at the time of surgery.This situation prevents the hidden confounding effect of stage of disease on functional outcomes.The authors of this study therefore consider the post-operative outcomes to be closely related to the type of surgery, independent of the stage of the disease.
The change in our surgical preference was based on reports from other experienced surgical teams, either published in the literature or presented in various international meetings (Donnez and Squifflet, 2004;Ret Davalos et al., 2007;Slack et al., 2007).However, in the literature, a majority of authors have been generally less interested in post-operative functional digestive outcomes than in the relief of pelvic pain and the improvement in quality of life.As regards the functional digestive outcomes of rectal surgery, available data is mainly provided by series of patients managed for colorectal cancer, whose mean age and health features are significantly different from those of young women presenting with endometriosis.For patients having undergone colorectal resection for cancer, the rates of post-operative diarrhoea, constipation and anal incontinence has been reported as, respectively, 30, 18 and 18% (Rauch et al., 2004).These symptoms may significantly impact a young woman's quality of life and are related to rectal anterior resection syndrome (Desnoo and Faithfull, 2006), to the surgical autonomic denervation of the colon (Lee et al., 2008) or to other unknown mechanisms.Indirectly, in a series of 52 women managed by laparoscopic colorectal resection for rectal endometriosis and followed-up for 2-55 months, Dubernard et al. (2006) reported a post-operative increase in constipation, tenesmus and diarrhoea in, respectively, 24, 27 and 7% of cases.
Severe constipation requiring additional endoscopic procedures occurred in three women.Endoscopic dilatation relieved constipation in one woman, who then presented an increase in frequency of stools (up to 6/day).Severe constipation persisted in two women, respectively, 34 and 12 months after surgery.It is important to note that severe constipation (,1 stool/5 days) was only observed in women managed by colorectal resection, which might be explained by inadvertent section of pelvic nerves due to extensive dissection around the rectum (Possover et al., 2000;Ferrero et al., 2006;Landi et al., 2006) or to surgical autonomic denervation resulting in altered colonic motility (Lee et al., 2008).
It has been suggested that complete recovery of gastrointestinal function should not be evaluated immediately, but rather several months after surgery (Ferrero et al., 2006).Thus choosing to include only women with a follow-up superior to 12 months has most likely increased the validity of our data, particularly in women managed by colorectal resection whose mean follow-up averages 28 months (Table III).We believe that functional digestive symptoms attributable to surgery and verified in our post-operative questionnaires are very unlikely to change.
Bladder dysfunction was only observed in the colorectal resection group, however, the difference was not found to be statistically significant due to the small sample size.It has been reported that transitory urinary retention or dysuria, which is generally the result of bladder parasympathetic nerve damage following section of uterosacral ligaments (Slack et al., 2007;Dubernard et al., 2008), may concern up to 16% of patients managed by colorectal resection (Darai et al., 2005;Fanfani et al., 2009).Although we often section uterosacral ligament insertions on the torus due to frequent infiltration by the disease, care was taken to spare bladder nerves particularly on the side corresponding to the ligament least affected by the disease.In our opinion, the risk of bladder dysfunction is higher in women undergoing colorectal resection, as the dissection carried out around the rectum is larger and deeper in the lower area underlying the nodule and on all sides of the rectum.Consequently the risk of section and thermal nerve damage to the bladder is probably higher when surgeons perform colorectal resection than with rectal nodule excision.
There were no rectovaginal fistulae in our series.In our opinion, the absence of rectovaginal fistulae in our sample could be attributed to the particular care taken to thoroughly close the vaginal wound by laparoscopic suture using resorbable stitches, to the care taken to separate vaginal suture and the rectal wound by fat tissue taken from omentum, mesorectum or pararectal space, and to not hesitating to perform temporary ileostomia to prevent anastomotic leakage.
The principal argument held by surgeons who choose to carry out only colorectal resection in women suffering for endometriosis, is that this procedure ensures a more complete resection of the rectal disease.This hypothesis was confirmed by a study comparing colorectal resection to the removal of rectal nodules by full thickness resection of the rectal wall (Remorgida et al., 2005).We also recently showed that active glandular endometrial foci are likely to be left out after fibrous rectal nodule excision (Roman et al., 2009).However, microscopically complete resection of digestive implants might remain incomplete even if large segmental resection is conducted, as indicated by the presence of endometriosis foci found on the margins of colorectum specimen removed (Roman et al., 2007;Anaf et al., 2009).Moreover, the question remains as to whether complete long-term relief of endometriosis pain systematically requires complete resection of digestive foci, when taking into consideration the risk of post-operative complications and unpleasant functional symptoms.
Contrary to rectal cancer, rectal endometriosis does not threaten patients' lives; however, it is usually extremely damaging to a woman's health and well being.In this article, we chose to focus on delayed post-operative functional outcomes resulting from the surgical procedure employed in the management of rectal endometriosis, based on the belief that treating endometriosis should not mean replacing pain with other unpleasant post-operative symptoms.However, this paper does not seek to endorse rectal nodule excision rather than colorectal segmental resection, because a definitive recommendation must take into account the long-term risk of recurrence associated with each surgical procedure.As both expected recurrence rates appear to be close (Fanfani et al., 2009), a comparative study focusing on the risk of recurrences would require several hundred patients, with a follow-up of several years (Roman and Bourdel, 2009).To our knowledge, no such randomized or prospective comparative study will be available within the next few years.Consequently, both our study and that of Fanfani et al. provide useful information regarding the functional outcomes relating to each surgical procedure, which is indispensable when deciding on the most appropriate course of treatment in each individual case of rectal endometriosis.

Figure 1
Figure1Post-operative evolution of dysmenorrhoea.Kaplan -Meier curves describe post-operative risk of recurrence of dysmenorrhoea, stratified by surgical technique (red line-colorectal resection; blue line-nodule excision; censored cases are indicated by numbers shown above the curve).Box plots present preoperative (blue box) and post-operative (red box) scores of dysmenorrhoea, using the 10-points analog rating scale, in women managed by nodule excision (up) and colorectal resection (down).

Figure 2
Figure 2 Post-operative evolution of dyspareunia.Kaplan -Meier curves describe post-operative risk of recurrence of dyspareunia, stratified by surgical technique (red line-colorectal resection; blue line-nodule excision; censored cases are indicated by numbers shown above the curve).Box plots present preoperative (blue box) and post-operative (red box) scores of dyspareunia, using the 10-points analog rating scale, in women managed by nodule excision (up) and colorectal resection (down).

Table II
Characteristics of surgical procedures (N 5 41)