Original Article

Role of Protective Defunctioning Stoma in Colorectal Resection forEndometriosis

J�er�emie Belghiti, MD, Marcos Ballester, MD, PhD, Sonia Zilberman, MD,Anne Thomin, MD, Chrysoula Zacharopoulou, MD, PhD, Marc Bazot, MD,Isabelle Thomassin-Naggara, MD, PhD, and Emile Dara€ı, MD, PhD*From the Department of Obstetrics and Gynecology, GRC 6-UPMC, Centre Expert En Endometriose (C3E), UMRS938 (Drs. Belghiti, Ballester, Zilberman,

Thomin, Zacharopoulou, and Dara€ı), and Radiology (Drs. Bazot and Thomassin-Naggara), Hopital Tenon, Assistance Publique des Hopitaux de Paris,

Universit�e-Pierre-et-Marie-Curie-Paris VI, Paris, France.

ABSTRACT Study Objective: To evaluate the role of protective defunctioning stoma (PDS) on the occurrence of digestive tract compli-

Drs. Belghiti and

The authors have

products or comp

Corresponding au

st�etrique, Hopital

E-mail: emile.dar

Submitted Octob

Available at www

1553-4650/$ - see

http://dx.doi.org/1

cations after colorectal resection to treat endometriosis.Design: Prospective cohort study (Canadian Task Force classification II-2).Setting: University hospital.Patients: All patients undergoing segmental colorectal resection to treat colorectal endometriosis with and without PDS be-tween 2003 and 2011 at Tenon University Hospital, Paris, France.Measurements and Main Results: Patients were assessed at 1, 6, and 12 months postoperatively and each year thereafter.Median follow-up was 60 months. Of 198 patients included for analysis, 53 (27%) had PDS. Overall, 15 (7.5%) digestive tractcomplications occurred: 9 (4.5%) rectovaginal fistulas and 6 (3%) anastomotic leakages. All rectovaginal fistulas occurred inpatients with a low colorectal anastomosis (p, .001) and 88% (8 of 9) in patients with a partial colpectomy (p, .001). PDSwas associated with a decrease in the number of rectovaginal fistulas in women undergoing partial colpectomy and low colo-rectal resection from 27% to 15%, without reaching significance (p5 .4). No anastomotic leakage occurred in patients withPDS.Conclusion: Our results support that PDS can be omitted in patients with mid-colorectal anastomosis without partial colpec-tomy. In patients requiring partial colpectomy or partial colpectomy plus low colorectal anastomosis, PDS remains question-able. Journal of Minimally Invasive Gynecology (2014) 21, 472–479 � 2014 AAGL. All rights reserved.

Keywords: Colorectal endometriosis; Digestive complication; Protective defunctioning stoma; Rectovaginal fistula

DISCUSS

You can discuss this article with its authors and with other AAGL members athttp://www.AAGL.org/jmig-21-4-JMIG-D-13-00568

Ballester contributed equally to this work.

no commercial, proprietary, or financial interest in the

anies described in this article.

thor: Emile Dara€ı, MD, PhD, Service de Gyn�ecologie-Ob-

Tenon, 4 rue de la Chine, 75020 Paris, France.

ai@tnn.aphp.fr

er 16, 2013. Accepted for publication December 4, 2013.

.sciencedirect.com and www.jmig.org

front matter � 2014 AAGL. All rights reserved.

0.1016/j.jmig.2013.12.094

Utoadth

se your Smartphonescan this QR code

nd connect to theiscussion forum foris article now*

* Download a free QR Code scanner by searching for ‘‘QRscanner’’ in your smartphone’s app store or app marketplace.

Endometriosis is defined by the presence of endometrialtissue outside the uterus, and affects 10% to 15% of womenof childbearing age [1]. Colorectal endometriosis is one ofthe most severe forms of the disorder and is a well-known

cause of altered quality of life and a source of infertility.Although the exact incidence of colorectal endometriosisis unknown, it is thought to affect 5% to 12% of patientswith endometriosis and up to 35% of patients with deep infil-trating endometriosis (DIE) [2].

After failure of medical treatment, colorectal resectionin patients with symptomatic endometriosis is a good optionto relieve symptoms [3,4] and enhance fertility [5–11].Moreover, since the first cases of laparoscopic colorectalresection to treat endometriosis [3,12,13], its use has beenwidely reported, which suggests that this major operationcan be considered a standard option. However, some

Belghiti et al. Defunctioning Stoma and Colorectal Endometriosis 473

authors have discussed the relevance of performing colorectalresection because of the risk of severe morbidity includingrectovaginal fistula, raising the issue of less extensivesurgery such as rectal shaving or discoid resection [14–17].Indeed, these less extensive surgical procedures, whenfeasible, are a good alternative to segmental resection[14–17]. Moreover, a recent meta-analysis has highlighteda lack of data concerning long-term results and fertilityoutcomes after a procedure that exposes patients to severecomplications including risk of neurogenic bladder anddigestive tract complications. Digestive tract complicationshave been observed in 2% to 10% of series, with a mean inci-dence of 2.7% [11]. In a meta-analysis of low anterior rectalresection to treat cancer, Tan et al [18] showed that a protec-tive defunctioning stoma (PDS) decreased clinical anasto-motic leakage and the rate of repeat operation to treatanastomotic leakage. In contrast to colorectal resection totreat cancer, few data are available on the relevance of PDSin preventing digestive tract complications in patients withendometriosis. Indeed, some authors recommend that aPDS be constructed systematically, imposing a second sur-gery [19], whereas other authors do not [11,20–23].

The objective of the present study was to evaluate the roleof PDS on the occurrence of postoperative digestive tractcomplications in patients requiring colorectal resection totreat DIE and thereby potentially clarifying its indications.

Materials and Methods

From March 2003 to December 2011, we performed aprospective cohort study that included all patients withsymptomatic DIE with colorectal involvement who under-went colorectal resection at Tenon University Hospital,Paris, France. All patients gave informed consent beforeinclusion, and the study was approved by the Ethics Com-mittee of the College National des Gyn�ecologues et Obst�etri-ciens Francais.

Preoperative Assessment of DIE

DIE was diagnosed clinically by 2 experienced surgeons(E.D. and M.B.) on the basis of the following criteria: visibledark blue nodules on the posterior vaginal fornix at speculumexamination or infiltration associated with palpable indura-tion at vaginal and rectal digit examination. Patients werethen referred to theDepartment ofRadiology for confirmationof the diagnosis. All patients underwent transvaginalultrasonography followed by magnetic resonance imagingto assess the presence of colorectal lesions, unifocality ormul-tifocality of bowel endometriosis, and location of associatedDIE lesions [2,24]. Anatomical locations of endometriosisand extent of colorectal endometriosis were also recorded.

Surgical Procedure

All laparoscopically assisted and open colorectal re-sections were performed with the objective of complete

resection, as previously described [25]. Procedures in-cluded adnexal surgery (ovarian cystectomy or salpingo-oophorectomy); uterosacral ligament, torus uterinum,parametrium, or vaginal resection; ureterolysis; and ure-teral reimplantation when required. The preferred surgicalroute was laparoscopy except from January 2006 toDecember 2008, when 52 patients were included in a ran-domized controlled trial comparing laparoscopy with opensurgery [4]. Laparoscopy was converted to open surgeryfor reasons of patient safety or because of technical diffi-culties. The creation of a PDS was recommended in all pa-tients requiring partial colpectomy or multiple bowelresections after our preliminary study of laparoscopic colo-rectal resection, which showed that partial colpectomy wasassociated with risk of rectovaginal fistula [25]. Moreover,for patients requiring en bloc resection including colorectalresection and hysterectomy, omentoplasty was performedwhen feasible and consisted of interposing omentum be-tween the vaginal suture and rectal staple line. Omento-plasty was not performed in patients wishing to preservetheir childbearing potential, to avert adhesions. Colorectalendometriosis was histologically confirmed in all patients.Patients were assessed at 1-, 6-, and 12-month postopera-tive follow-up and then each year thereafter.

Study Variables

Data including epidemiologic characteristics and radio-logic and surgical findings were prospectively recordedfrom patient medical records. The American Society forReproductive Medicine score was systematically calculated.A low colorectal anastomosis was defined as an anastomosis,7 cm from the anal margin. Digestive tract complicationsincluding anastomotic leakage and rectovaginal fistula wererecorded.

Statistical Analysis

Univariate analysis was performed using the Wilcoxontest for continuous variables and the Fisher exact test for cat-egorical variables. All reported p values were 2-sided. Sig-nificant difference was accepted as p , .05. All statisticalanalysis was performed using commercially available soft-ware (STATA version 11; StataCorp LP, College Station,TX).

Results

Epidemiologic and Surgical Characteristics

During the study, 212 patients underwent colorectalresection to treat symptomatic endometriosis (Table 1). Ofthese, 14 were excluded from the analysis for the followingreasons: rectal shaving (n 5 2), discoid rectal resection(n 5 6), lost to follow-up (n 5 6) (Fig. 1). The study popu-lation thus consisted of 198 patients with a median age of34 years (range, 23–53 years) and a body mass index of 22

Table 1

Epidemiologic and surgical characteristics of study population

Variablea Patients (N 5 198)

Age, yr 34 (23–53)

Body mass index 22 (16–38)

Previous surgery to treat deep

infiltrating endometriosis

116 (59)

Parity 0 (0–4)

Previous infertility 86 (44)

Surgical route

Laparoscopy 136 (69)

Conversion to open surgery 11 of 136 (8)

Open surgery 62 (31)

ASRM total score 85 (10–158)

Characteristics of colorectal lesions

Size, mm 30 (5–110)

Multifocal 33 (17)

Low rectal 84 (43)

Colorectal junction 111 (56)

Anatomic locations associated with colorectal

endometriosisb

Uterosacral ligament 181 (91)

Torus uterinum 185 (93)

Parametrium 105 (53)

Vagina 63 (32)

Ovaries 101 (51)

Bladder 11 (6)

Surgical procedures associated

with colorectal resectionc

Hysterectomy 70 (35)

Partial colpectomy 63 (32)

Ureteral reimplantation 17 (9)

Multiple bowel resections 11 (6)

Low colorectal anastomosis 84 (43)

Total resections (range) 6 (2–10)

Defunctioning protective stoma 53 (27)

Colostomy 32 (16)

Ileostomy 21 (11)

ASRM 5 American Society of Reproductive Medicine.a Values are given as median (range) or No. (%).b Some patients had multiple deep infiltrating endometriosis locations associ-

ated with colorectal endometriosis.c Some patients required multiple surgical procedures.

474 Journal of Minimally Invasive Gynecology, Vol 21, No 3, May/June 2014

(range, 16–38). Of the 136 patients (69%) who underwentlaparoscopic colorectal resection, conversion to open sur-gery was required in 11 (8%). Most patients had endometri-osis affecting the uterosacral ligament (n 5 181 [91%]) andtorus uterinum (n5 185 [93%]). Sixty-three patients (32%)had vaginal endometriosis requiring partial colpectomy. Ofthe 17 patients requiring a ureteral reimplantation, 12(71%) had severe ureteral stenosis alone and 5 (29%) hadboth bladder and ureteral endometriosis. Fifty-three patients(27%) had a PDS. Thirty-two patients (16%) had a colos-tomy, and 21 (11%) an ileostomy. Eighty-four patients(43%) had a low colorectal anastomosis, and 111 (56%) amid-colorectal anastomosis.

Incidence of Rectovaginal Fistula

During the study, 9 rectovaginal fistulas occurred (4.5%)(Table 2). The diagnosis was made by emission of gas orstool from the vagina. Two factors were associated withthe occurrence of rectovaginal fistula: partial colpectomy(p, .001) and low colorectal anastomosis (p, .001). Eightof 9 (89%) rectovaginal fistulas occurred in patients withcolpectomy, and all occurred in patients with a low colo-rectal anastomosis. The remaining rectovaginal fistulaoccurred in a patient without partial colpectomy (1 of 135patients, 0.7%) experiencing partial vaginal necrosis linkedto extensive coagulation, with spontaneous healing afterileostomy. A defunctioning stoma (3 colostomy and 1 ileos-tomy) was created in 4 patients with a rectovaginal fistulawithout a PDS and resulted in spontaneous healing withoutrecurrence after stoma reversal. Of the remaining 5 patientswith rectovaginal fistula in patients with PDS as the firstintent, a Hartmann procedure was necessary in 1 patient, fol-lowed by a second operation with coloanal anastomosis, andin the other 4 patients a second colorectal resection wasrequired. All 5 patients underwent omentoplasty.

Incidence of Anastomotic Leakage

Overall, 6 patients (3%) exhibited anastomotic leakageand had fever and infection. None of these patients had aPDS. No factors were associated with the occurrence ofanastomotic leakage (Table 3). Two patients required a Hart-mann procedure followed by a second colorectal resection.In the remaining 4 patients, a defunctioning stoma wascreated (3 ileostomy and 1 colostomy). The treatment wasinsufficient to obtain closure of the anastomotic failure in1 patient and required a second colorectal resection.

Effects of PDS

Of the 63 patients with partial colpectomy, 17 (27%) hadno PDS. The rate of rectovaginal fistula in patients with par-tial colpectomy with and without PDS was 11% and 18%,respectively (p 5 .7).

Of 53 patients undergoing creation of a PDS, 37 had a lowcolorectal anastomosis. The rate of rectovaginal fistula in pa-tients with a PDSwith or without low colorectal anastomosiswas respectively 14% (5 of 37 patients) and 0% (none of 15patients) (p 5 .3).

Of 44 patients with partial colpectomy and low colorectalanastomosis, 33 (75%) had a PDS. Among these, 5 patients(15%) had a rectovaginal fistula. Three of 11 patients (27%)with partial colpectomy and low colorectal anastomosiswithout PDS experienced a rectovaginal fistula. No differ-ence in the rate of rectovaginal fistula was observed betweenpatients with partial colpectomy and low colorectal anasto-mosis with and without PDS (p 5 .39). Although no differ-ence was observed in the rate of rectovaginal fistula, adecrease in rectovaginal fistulas was observed in patients

Fig. 1

Description of the study population. DIE 5 deep infiltrating endometriosis; PDS 5 protective defunctioning stoma; *p 5 .4.

Belghiti et al. Defunctioning Stoma and Colorectal Endometriosis 475

undergoing partial colpectomy and low colorectal resection,from 27% to 15% (p 5 .4).

Effects of Omentoplasty

Of 70 patients undergoing en bloc hysterectomy andcolorectal resection, 26 (37%) had a low colorectal anasto-mosis. Thirty-one of the 70 patients (44%) underwent omen-toplasty. Nine of the 31 patients (29%) had a low colorectalanastomosis. The rate of rectovaginal fistula in patientsundergoing en bloc hysterectomy and colorectal resectionwith and without omentoplasty was 6% (2 of 31 patients)and 5% (2 of 38 patients), respectively (p 5 .99). Of the26 patients with low colorectal anastomosis, the rate of rec-tovaginal fistula in patients with and without omentoplastywas 22% (2 of 9 patients) and 12% (2 of 17 patients), respec-tively (p 5 .06).

Discussion

This study demonstrates that partial colpectomy and lowcolorectal anastomosis are risk factors for rectovaginal fis-tula in patients requiring colorectal resection to treat DIE.PDS contributes to decreasing, but not eliminating, the riskof rectovaginal fistula in patients requiring both partialcolpectomy and low colorectal anastomosis. In contrast, itwould seem that PDS is not necessary in patients withmid-colorectal anastomosis or low colorectal anastomosiswithout colpectomy.

The present study confirms that partial colpectomy is amajor risk factor for rectovaginal fistula in patients requiringcolorectal resection to treat endometriosis, with an overallrate of 13%. Our results are in agreement with those of pre-vious studies of colorectal surgery to treat cancer thatshowed that concomitant vaginal suture is a risk factor of

Table 2

Comparison between patients with or without a rectovaginal fistula (N 5 198)a

Variable

Patients with rectovaginal

fistula (n 5 9)

Patients without

rectovaginal fistula (n 5 189) p valueb

Age, yr 36 (29–40) 34 (23–53) .9

Body mass index 23 (20–29) 22 (12–38) .5

Parity 0 (0–1) 0 (0–4) .9

Previous surgery to treat deep infiltrating endometriosis 4 (44) 112 (60) 0.5

Surgical route .1

Laparoscopy 3 (33) 122 (65)

Conversion to open surgery 1 (11) 10 (5)

Open surgery 5 (56) 57 (30)

ASRM total score 82 (30–158) 86 (10–154) .9

Total deep infiltrating endometriosis lesions resected 6 (2–10) 6 (2–10) .1

Anatomic locations associated with colorectal endometriosisc

Uterosacral ligament 8 (89) 173 (92) .6

Torus uterinum 8 (89) 177 (94) .5

Parametrium 4 (44) 101 (53) .7

Vagina 8 (89) 55 (29) ,.001

Ovaries 4 (44) 97 (51) .7

Surgical procedures associated with colorectal endometriosisd

Hysterectomy 4 (44) 66 (35) .7

Ureteral reimplantation 1 (11) 16 (8) .6

Multiple digestive tract resections 2 (22) 9 (5) .08

Colpectomy 8 (89) 55 (29) ,.001

Low colorectal anastomosis 9 (100) 75 (40) ,.001

Ileostomy 4 (44) 17 (9) .07

Colostomy 1 (11) 31 (17) .1

ASRM 5 American Society of Reproductive Medicine.a Values are given as median (range) or as No. (%).b Wilcoxon test for continuous variables and Fisher exact test for categorical variables.c Some patients had multiple deep infiltrating endometriosis locations associated with colorectal endometriosis.d Some patients required multiple surgical procedures.

476 Journal of Minimally Invasive Gynecology, Vol 21, No 3, May/June 2014

rectovaginal fistula [26]. Moreover, this high incidence ofrectovaginal fistula is in accordance with that in a reviewby Vercellini et al [27], who reported rectovaginal fistulain 2% to 10% of patients after rectovaginal septum surgeryperformed to treat endometriosis. Recently, Meulemanet al [11], in a meta-analysis of a series of 49 colorectal re-sections to treat endometriosis, noted that most authors didnot report the rate of rectovaginal fistulas, making it impos-sible to determine the risk factors for this severe complica-tion. However, the rate of rectovaginal fistula seemed tovary according to the use of PDS, reaching 18% in patientsundergoing colorectal resection without PDS (withoutreaching significance). This high incidence of rectovaginalfistula is in agreement with that of our preliminary studythat showed that all rectovaginal fistulas occurred after par-tial colpectomy, with an incidence of 25% in the absence ofPDS use [26]. However, despite the use of PDS, the rate ofrectovaginal fistula was 11%, which suggests that otherrisk factors such as partial colpectomymust be taken into ac-count. From a pathophysiologic point of view, the high inci-dence of rectovaginal fistula can be explained in that partialcolpectomy concerns the posterior vaginal fornix, resulting

in a vaginal suture facing the colorectal anastomosis. More-over, the relatively high incidence of digestive tract compli-cations associated with segmental colorectal resectionraises the issue of less extensive surgery such as rectalshaving or discoid resection, imposing evaluation in a ran-domized trial.

To date, there is no consensus about the indications forPDS in patients with endometriosis requiring colorectalresection [19–23]. Indeed, in a study of 100 patientsundergoing open surgery to treat bowel endometriosis andrequiring a low colorectal anastomosis, despite systematicuse of protective ileostomy, Dousset et al [19] reported a6% rate of digestive tract complications, primarily repre-sented by anastomotic leakage and rectovaginal fistula.Similarly, in a study of 750 patients undergoing laparoscopicresection to treat mid- and low rectal endometriosis, Ruffoet al [20] performed a systematic ileostomy when the anas-tomosis was ,5 cm or when hydropneumatic testing waspositive for gas leakage. In that study, the authors found3% of anastomotic leakage with intra-abdominal abscess,all in patients without PDS. Although these results suggestthat PDS might be recommended in the case of low

Table 3

Comparison between patients with or without anastomotic failure (N 5 198)a

Variable

Patients with anastomotic

leakage (n 5 6)

Patients without anastomotic

leakage (n 5 192) p valueb

Age, yr 34 (29–46) 34 (23–53) .6

Body mass index 23 (17–25) 22 (12–18) .6

Parity 1 (0–2) 0 (0–4) .2

Previous surgery to treat deep infiltrating endometriosis 4 (67) 112 (59) .99

Surgical route .09

Laparoscopy 5 (83) 120 (63)

Conversion to open surgery 1 (17) 10 (5)

Open surgery 0 62 (32)

ASRM total score 110 (46–154) 84 (10–158) .2

No. of deep infiltrating endometriosis lesions resected 6 (3–7) 6 (2–10) .7

Anatomic locations associated with colorectal endometriosisc

Uterosacral ligament 5 (83) 176 (92) .4

Torus uterinum 5 (83) 180 (94) .4

Parametrium 3 (50) 102 (53) .99

Vagina 0 63 (33) .2

Ovaries 4 (67) 97 (51) .7

Surgical procedures associated with colorectal resectiond

Hysterectomy 2 (33) 68 (75) .99

Uteral reimplantation 1 (17) 16 (8) .4

Multiple digestive tract resections 0 11 (6) .99

Colpectomy 0 63 (33) .2

Low colorectal anastomosis 2 (33) 82 (43) .6

ASRM 5 American Society of Reproductive Medicine.a Values are given as median (range) or as No. (%).b Wilcoxon test for continuous variables and Fisher exact test for categorical variables.c Some patients had multiple deep infiltrating endometriosis locations associated with colorectal endometriosis.d Some patients required multiple surgical procedures.

Belghiti et al. Defunctioning Stoma and Colorectal Endometriosis 477

colorectal anastomosis, no consensus exists on its indication,because the risk of leakage is low and imposes a second sur-gery to close the PDS [14–17]. In contrast, in 2 series,respectively including 92 and 115 patients undergoingsegmental colorectal resection to treat endometriosiswithout PDS, the authors reported a low rate of digestivetract complications, which raises the question of theinterest of systematic PDS imposing a second operation[20,22]. The risk-benefit ratio of a systematic PDS requiringa second operation compared with a second operation in asmall number of patients with a rectovaginal fistula requiresfurther investigation. In contrast to colorectal endometriosis,the role of PDS has been evaluated in rectal cancer surgery.Indeed, a meta-analysis including 4 randomized control tri-als of low rectal cancer surgery showed that creating a PDSresulted in significantly lower anastomotic leakage andrepeat operation rates [18]. The authors consequentlyrecommend that a PDS be performed systematically in lowrectal cancer surgery. However, the analogy between rectalendometriosis and rectal cancer surgery is imperfect andshould be interpreted with caution because of important dif-ferences between populations. Patients requiring colorectalresection to treat cancer have more often associated risk fac-tors for anastomotic complications such as radiotherapy,

malnutrition, preoperative weight loss, advancing age, orobesity [28]. In contrast, most patients with colorectal endo-metriosis are young and in good health without malnutrition;however, colorectal resection is more often associated withopening of the vagina. Although a decrease in the numberof rectovaginal fistulas was observed in patients with PDSin cases of partial colpectomy and low colorectal resection,our results did not support the use of a systematic PDS in thisspecific situation. Moreover, most patients with colorectalendometriosis wish to preserve their childbearing potential.This point is crucial in the management of patients withcolorectal endometriosis because digestive tract complica-tions might affect fertility. Whereas Kondo et al [29] showedthat fertility remains relatively high in these patients, Dara€ıet al [30] reported a reduced probability of conception aftera major complication of colorectal endometriosis, and noneof these patients have become pregnant spontaneously.

Omentoplasty is thought to be an alternative to PDS. Inthe present study, patients requiring concomitant hysterec-tomy and colorectal resection underwent omentoplastywhen possible. The rate of rectovaginal fistula in patients un-dergoing en block hysterectomy and colorectal resectionwith and without omentoplasty did not differ. These resultsare in contrast with those of previous prospective studies

478 Journal of Minimally Invasive Gynecology, Vol 21, No 3, May/June 2014

of the role of omentoplasty in protecting colorectal anasto-mosis, resulting in less anastomotic leakage and lower repeatoperation rates [31,32], However, these studies includedprimarily patients with colorectal cancer, rending itdifficult to extrapolate these results to our population. In ameta-analysis of the use of omentoplasty, Hao et al [33]showed no difference in radiologic anastomotic leakage,repeat operation, or death between the groups with andwithout omentoplasty, although a difference in clinical anas-tomotic leakage was observed (relative risk, 0.36; 95% con-fidence interval, 0.16–0.78). Although in the literatureomentoplasty seems to reduce the risk of digestive tract com-plications associated with colorectal resection, its use inyoung patients with colorectal endometriosis who wish topreserve their childbearing potential remains questionable.Indeed, omentoplasty might be associated with risk of adhe-sions, potentially affecting tubo-ovarian function. However,no published data support this assertion. Therefore, in pa-tients undergoing colorectal resection to treat endometriosis,the role of omentoplasty on the risk of digestive tract compli-cations and its effect on fertility remain to be explored.

Some limitations of the present study must be underlined.First, we cannot exclude biases related to changes in surgicaltechniques such as the advent of the harmonic scalpel andthe increasing experience of surgeons during the studyperiod. At the beginning of the study, a rectovaginal fistulaoccurred after vaginal necrosis subsequent to extensive bipo-lar coagulation. Second, despite a decrease in the number ofrectovaginal fistulas in patients undergoing partial colpec-tomy with PDS compared with patients without PDS, thedifference was not significant. This is probably related tosample size; however, confounding factors cannot beexcluded including the degree of bowel endometriosis, thelength of bowel resection, and the extent of colpectomy.Moreover, removal of pararectal endometriotic lesionsinvolving parametrium and extraserous pelvic fascia mightaffect rectal vascularization, contributing potentially to ahigh incidence of rectovaginal fistula. However, it is clearthat patients with no colpectomy and mid-colorectal resec-tion received no advantage from PDS. In the present study,these patients corresponded to two-thirds of the population.Third, because of the sample size and the relatively lownumber of rectovaginal fistulas, multivariate analysis wasnot feasible, and it was therefore not possible to identify con-founding factors contributing to this severe complication.

In conclusion, the location of the colorectal anastomosisand the association of partial colpectomy are risk factors forrectovaginal fistula in patients requiring colorectal resectionto treat DIE. Although our results support that PDS can beomitted in patients with mid-colorectal anastomosis withoutpartial colpectomy, indication for PDS in patients requiringpartial colpectomy or partial colpectomy plus low colorectalanastomosis remains questionable. Future studies shouldfocus on effective ways to separate the vaginal suture linefrom the rectal staple/suture line, which is the determinantfactor for rectovaginal fistula.

Acknowledgment

We thankRomanRouzier,MD,PhD, for statistical analysis.

References

1. Ballweg ML. Impact of endometriosis on women’s health: comparative

historical data show that the earlier the onset, the more severe the dis-

ease. Best Pract Res Clin Obstet Gynaecol. 2004;18:201–218.

2. Bazot M, Darai E, Hourani R, et al. Deep pelvic endometriosis: MR im-

aging for diagnosis and prediction of extension of disease. Radiology.

2004;232:379–389.

3. Redwine DB, Wright JT. Laparoscopic treatment of complete obliter-

ation of the cul-de-sac associated with endometriosis: long-term

follow-up of en bloc resection. Fertil Steril. 2001;76:358–365.

4. Dara€ı E, Dubernard G, Coutant C, Frey C, Rouzier R, Ballester M. Ran-

domized trial of laparoscopically assisted versus open colorectal resec-

tion for endometriosis: morbidity, symptoms, quality of life, and

fertility. Ann Surg. 2010;251:1018–1023.

5. Weed JC, Ray JE. Endometriosis of the bowel. Obstet Gynecol. 1987;

69:727–730.

6. Coronado C, Franklin RR, Lotze EC, Bailey HR, Vald�es CT. Surgicaltreatment of symptomatic colorectal endometriosis. Fertil Steril.

1990;53:411–416.

7. Dara€ı E, Carbonnel M, Dubernard G, et al. Determinant factors of

fertility outcomes after laparoscopic colorectal resection for endometri-

osis. Eur J Obstet Gynecol Reprod Biol. 2010;149:210–214.

8. Fedele L, Bianchi S, Zanconato G, Bettoni G, Gotsch F. Long-term

follow-up after conservative surgery for rectovaginal endometriosis.

Am J Obstet Gynecol. 2004;190:1020–1024.

9. Ferrero S, Anserini P, Abbamonte LH, Ragni N, Camerini G,

Remorgida V. Fertility after bowel resection for endometriosis. Fertil

Steril. 2009;92:41–46.

10. Vercellini P, Pietropaolo G, De Giorgi O, Daguati R, Pasin R,

Crosignani PG. Reproductive performance in infertile women with rec-

tovaginal endometriosis: is surgery worthwhile? Am J Obstet Gynecol.

2006;195:1303–1310.

11. Meuleman C, Tomassetti C, D’Hoore A, et al. Surgical treatment of

deeply infiltrating endometriosis with colorectal involvement.Hum Re-

prod Update. 2011;17:311–326.

12. Redwine DB, Sharpe DR. Laparoscopic segmental resection of the sig-

moid colon for endometriosis. J Laparoendosc Surg. 1991;1:217–220.

13. Nezhat C, Nezhat F, Pennington E. Laparoscopic treatment of infiltra-

tive rectosigmoid colon and rectovaginal septum endometriosis by the

technique of videolaparoscopy and the CO2 laser.Br J Obstet Gynaecol.

1992;99:664–667.

14. Nezhat C, Nezhat F, Pennington E, Nezhat CH, Ambroze W. Laparo-

scopic disk excision and primary repair of the anterior rectal wall for

the treatment of full-thickness bowel endometriosis. Surg Endosc.

1994;8:682–685.

15. Mohr C, Nezhat FR, Nezhat CH, Seidman DS, Nezhat CR. Fertility

considerations in laparoscopic treatment of infiltrative bowel endome-

triosis. JSLS. 2005;9:16–24.

16. Nezhat C, Hajhosseini B, King LP. Laparoscopic management of bowel

endometriosis: predictors of severe disease and recurrence. JSLS. 2011;

15:431–438.

17. Donnez J, Squifflet J. Complications, pregnancy and recurrence in a

prospective series of 500 patients operated on by the shaving technique

for deep rectovaginal endometriotic nodules. Hum Reprod. 2010;25:

1949–1958.

18. Tan WS, Tang CL, Shi L, Eu KW. Meta-analysis of defunctioning

stomas in low anterior resection for rectal cancer. Br J Surg. 2009;96:

462–472.

19. Dousset B, Leconte M, Borghese B, et al. Complete surgery for low

rectal endometriosis: long-term results of a 100-case prospective study.

Ann Surg. 2010;251:887–895.

Belghiti et al. Defunctioning Stoma and Colorectal Endometriosis 479

20. Ruffo G, Sartori A, Crippa S, et al. Laparoscopic rectal resection for se-

vere endometriosis of the mid and low rectum: technique and operative

results. Surg Endosc. 2012;26:1035–1040.

21. Meuleman C, Tomassetti C, D’Hooghe TM. Clinical outcome after lapa-

roscopic radical excision of endometriosis and laparoscopic segmental

bowel resection. Curr Opin Obstet Gynecol. 2012;24:245–252.

22. Pereira RMA, Zanatta A, Preti CDL, de Paula FJF, da Motta ELA,

Serafini PC. Should the gynecologist perform laparoscopic bowel resec-

tion to treat endometriosis? results over 7 years in 168 patients. J Minim

Invasive Gynecol. 2009;16:472–479.

23. Ribeiro PAA, Rodrigues FC, Kehdi IPA, et al. Laparoscopic resection of

intestinal endometriosis: a 5-year experience. J Minim Invasive Gyne-

col. 2006;13:442–446.

24. BazotM, Lafont C, Rouzier R, RoseauG, Thomassin-Naggara I, Dara€ı E.Diagnostic accuracy of physical examination, transvaginal sonography,

rectal endoscopic sonography, and magnetic resonance imaging to diag-

nose deep infiltrating endometriosis. Fertil Steril. 2009;92:1825–1833.

25. Darai E, Thomassin I, Barranger E, et al. Feasibility and clinical

outcome of laparoscopic colorectal resection for endometriosis. Am J

Obstet Gynecol. 2005;192:394–400.

26. Kosugi C, Saito N, Kimata Y, et al. Rectovaginal fistulas after rectal

cancer surgery: incidence and operative repair by gluteal-fold flap

repair. Surgery. 2005;137:329–336.

27. Vercellini P, Crosignani PG, Abbiati A, Somigliana E, Vigan�o P,

Fedele L. The effect of surgery for symptomatic endometriosis: the

other side of the story. Hum Reprod Update. 2009;15:177–188.

28. H€user N, Michalski CW, Erkan M, et al. Systematic review and meta-

analysis of the role of defunctioning stoma in low rectal cancer surgery.

Ann Surg. 2008;248:52–60.

29. Kondo W, Dara€ı E, Yazbeck C, et al. Do patients manage to achieve

pregnancy after a major complication of deeply infiltrating endometri-

osis resection? Eur J Obstet Gynecol Reprod Biol. 2011;154:196–199.

30. Dara€ı E, Lesieur B, Dubernard G, Rouzier R, Bazot M, Ballester M.

Fertility after colorectal resection for endometriosis: results of a pro-

spective study comparing laparoscopy with open surgery. Fertil Steril.

2011;95:1903–1908.

31. Tocchi A, Mazzoni G, Lepre L, et al. Prospective evaluation of omen-

toplasty in preventing leakage of colorectal anastomosis. Dis Colon

Rectum. 2000;43:951–955.

32. Agnifili A, Schietroma M, Carloni A, et al. The value of omentoplasty

in protecting colorectal anastomosis from leakage: a prospective ran-

domized study in 126 patients. Hepatogastroenterology. 2004;51:

1694–1697.

33. Hao X-Y, Yang K-H, Guo T-K, Ma B, Tian J-H, Li H-L. Omentoplasty

in the prevention of anastomotic leakage after colorectal resection: a

meta-analysis. Int J Colorectal Dis. 2008;23:1159–1165.