Introduction

Urethrovaginal fistula (UVF) is an uncommon urological condition. Most of the causes are iatrogenic from vaginal surgeries in the developed world [1]. Currently, postoperative outcomes are largely confined to small studies with short-term follow-up, and functional outcomes are lacking. In particular, lower urinary tract (LUT) function and sexual function have not been much investigated in the postoperative setting. Much of this is impeded by (a) low case volumes and (b) high attrition rates on follow-up, making the long-term assessment of these patients difficult [2]. At the moment, there is no strong data comparing anatomic and functional outcomes from different etiologies in the treatment of non-irradiated UVF and the best method available for evaluation of these outcomes. We sought to review the composite outcomes (surgical and functional) in women having undergone UVF repair at our tertiary institution. In particular, we evaluated whether the etiology (mid-urethral sling vs. non-sling) impacted these outcomes.

Methods

Data collection and database use were approved by our institutional review board. We retrospectively reviewed records of consecutive adult women undergoing UVF repair at our institution from 1996 to 2013. Inclusion criteria included all UVF patients with a minimum 6-month follow-up and ≥18 years of age. Radiation-induced UVF were excluded. Medical record data were reviewed for (1) demographics, (2) clinical history, (3) etiology, (4) imaging [voiding cystourethrogram (VCUG), CT urogram, pelvic ultrasound], cystoscopy or retrograde pyelogram (RPG), (5) urodynamic study, (6) perioperative complications, (7) reoperation and (8) functional outcomes based on validated questionnaires.

Transvaginal UVF repair and any other associated indicated procedures such as tissue interposition were performed by a single fellowship-trained Urologist (PZ). A third-party investigator (DL), who did not participate in the surgeries, performed the data collection.

All patients had radiological assessment of fistula closure with lateral standing voiding cystourethrogram (VCUG) at 4- to 6-week follow-up. Patients were followed at 6 months and yearly thereafter. For functional outcomes, all patients were administered the following validated questionnaires at their latest clinical visit: Urogenital Distress Inventory 6 (UDI-6), Incontinence Impact Questionnaire (IIQ-7), Female Sexual Function Index (FSFI) and one global Quality of Life (QoL) question scored on a visual analogue scale (VAS) from 0 = excellent to 10 = terrible [3]. Patients lost to follow-up were reached by a structured phone interview incorporating the same questionnaires. Good quality of life (QoL) outcome on VAS was defined as ≤3. A score of ≤26.5 on the FSFI was classified as female sexual dysfunction (FSD). Two surgical groups were compared: (1) synthetic sling-related versus (2) non-sling-related UVF.

Descriptive statistics were used to characterize demographic data. Student’s t test was used for continuous variables between the surgical groups with criterion set at p < 0.05 for statistical significance. SPSS version 16 (Chicago, IL, USA) was used for all statistical analysis.

Surgical technique

The surgical technique has been previously published [4]. In summary, the procedure is performed in high lithotomy position. A urethrocystoscopy is done first to identify the fistula tract and cannulate it with an open-ended ureteral catheter over a guide wire. Then, a suprapubic tube catheter is inserted (punch technique or with a Lowsley retractor) (Fig. 1). Depending on the decision on using fascial patch or not for urethral closure and/or concurrent anti-incontinence procedure either a fascial patch (2 cm in width and 4 cm in length) or a fascial strip 2 cm × 6 cm for a pubovaginal sling (PVS) is harvested via a small longitudinal suprapubic incision. The fascial defect is closed with absorbable sutures. Additional tissue interposition with the Martius fat pad graft (MFPG) is infrequently utilized as it provides for tissue interposition, but does not aid in restoring continence. The technique for MFPG harvest has been previously described [5].

Fig. 1
figure 1

At cystourethroscopy: large urethral defect (red arrow, a) with glove suburethral (yellow arrow, b). Transvaginal view of fistula cannulated with 14-Fr Foley catheter (blue arrow, c, d)

Full size image

An 18-French urethral Foley catheter is inserted, and the patient is tilted in maximum Trendelenburg position. Following hydro-dissection, a broad-base inverted U-shaped anterior vaginal wall incision is raised to accommodate tissue interposition later on. The periurethral tissue is dissected to free the edges of the fistula enough for non-tensioned closure. Any synthetic sling material around the fistula tract is excised with special attention given to remove any single mesh strands. These strands can be easily found when the synthetic material is blue colored, but very challenging even with magnifying loupes when clear or transparent and covered by scar tissue.

The urethral Foley catheter is removed and a 25-French female sound is placed. The sound allows closure without incorporating the back wall of the urethra and avoids subsequent urethral narrowing. The closure of the urethral defect is performed in a longitudinal fashion using two sets of running fine 4/0 or 5/0 PDS sutures starting at each extremity of the urethral defect with a final knot around the midpoint of the fistula defect. A watertight test is performed with an 8-French feeding tube placed alongside the wall of the urethra or with a urethrocystoscope. When available, periurethral tissues can be closed over that first line fistula layer closure to reinforce the repair using interrupted fine absorbable sutures. A better reinforcement of this first suture line comes from an autologous fascial patch, which can be secured over the fistula repair site with a few absorbable sutures. Finally, the originally raised anterior vaginal flap is advanced to cover the repair and closed with absorbable sutures. The urethral Foley catheter is replaced and a vaginal pack soaked with antibiotics is inserted. The suprapubic incision is closed and the suprapubic tube is secured to the skin.

Results

We identified a total of 18 women from our prospective database that had UVF repair during the study period from 1996 to 2013. Baseline demographic and preoperative data are presented in Table 1. Not surprisingly, Group 1 numbers increased over the past 10 years. Prior failed UVF repair was recorded in 11 women (62 %). Surgical characteristics and relevant surgical reconstruction surgeries are summarized in Table 2. Patients presented with variable symptoms including urinary incontinence 17 (94 %), dyspareunia/pelvic pain 4 (22 %), voiding dysfunction 4 (22 %) and recurrent UTI 3 (17 %). Size of UVF defect ranged from 1 mm to larger defects up to 3 cm. Of the 18 women, 17 had tissue interposition. Overall repair success rate was 95 %. One case of recurrence was in a renal transplant woman on immunosuppressant who eventually required a cystectomy and ileal conduit.

Table 1 Urethrovaginal fistula patient characteristics (n = 18)
Full size table
Table 2 UVF surgery characteristics
Full size table

Of the 18 women, we obtained 12 (67 %) responses to our questionnaires of which 7 were from clinic visits and 5 from structured telephone interviews. The results of the postoperative UDI-6, IIQ-7 and FSFI are presented in Table 3. There was no statistical difference on UDI-6 questionnaire outcomes between the two groups among responders for Q2 (UUI) and Q3 (SUI); but there was a statistical difference for Q4 small leak: 1.9 versus 0.8 (p = 0.03) and Q5 difficulty emptying: 1.3 versus 0 (p = 0.02). Mean score (SD) for the VAS QoL was 3.8 (3.6), and statistical difference was noted between the two groups, favoring the non-sling group: 1.5 (0.6) versus 5 (4) (p = 0.05). No differences in IIQ-7 were noted between the two groups (p = 0.09) with a mean score (SD) of 6.5 (9.5). Of the 18 patients, 5 (28 %) remained sexually active and of those, two responded to FSFI (40 %) with low scores.

Table 3 Outcomes following urethrovaginal fistula (UVF) repair
Full size table

No major perioperative complications were noted. Six of the 18 women (33 %) required a secondary procedure: three required injectable agents for recurrent SUI, two had transurethral holmium laser to remove residual urethral sling mesh strands and one had urethral dilation.

Discussion

Urethrovaginal fistula (UVF) is fortunately a rare event in developed countries, but it still poses a significant challenge for pelvic reconstructive surgeons. The paucity of reported outcomes from surgical management presents a distinct disadvantage as there is no consensus with regards to management to guide both treating physicians and patients. Much of the current treatment and outcome is based on anecdotal small case series. In our series, we have attempted to overcome these deficiencies by combining both objective and subjective outcome measures using validated tools in a reasonable-sized cohort of patients with a standardized surgical approach.

Our patient population is heterogenous owing to differing etiologies. Symptoms were predominantly associated with incontinence with a quarter having pain or dysfunctional voiding. Over two-thirds of the patients had had previous attempts at fistula repair from our referral source before being referred to our institution. Majority of the patients were postmenopausal except in cases of urethral diverticulae and trauma patients, where this represented a younger cohort of women. In contrast to developing countries, we documented no UVF cases from obstetric complications, but for the last decade, there has been an increase in cases related to use of synthetic mid-urethral sling (MUS) [6–12]. The estimated risk for sling erosion from MUS in the general literature varies from 0.07 to 1.5 %. Various attributable factors have been proposed including tissue factors with vaginal atrophy and estrogen deficiency, and technical factors may also be involved such as submucosal placement and excessive sling tension. Some authors suggest that fibrosis resulting from the rejection process around the prosthetic material is a factor in the occurrence of urethral erosion [13]. During MUS removal, a urethral injury can occur and may not be recognized, leading to secondary incontinence from a UVF. This diagnosis is not always easily recognized unless the clinician has a high index of suspicion. Lateral view voiding urethrocystogram and urethroscopy is very useful to confirm the UVF diagnosis.

The repair of urethrovaginal fistula (UVF) poses a significant challenge for pelvic reconstructive surgeons. Successful anatomical closure depends on good tissue quality and tissue interposition given the restricted space for reconstruction. Often there is considerable scarring within the vagina, and the lack of pliant tissue makes for difficult non-tension closure. This situation is often compounded by involvement of the urinary sphincter complex that may potentially compromise outcomes with recurrent stress urinary incontinence. For these reasons, the closure and outcomes of UVF need to be anatomical and functional and cannot be extrapolated from repairs for vesico-vaginal fistula (VVF). Even though the mechanism of injury may be the same, the outcomes can be vastly different. From a technical standpoint, we approached the repair vaginally as this is the most minimal invasive irrespective of the size and location of the fistula as this varied considerably in our patients. It is important that a wide-based U-shaped vaginal flap be raised for two reasons: (1) it allows for maximal sling excision in the case of MUS and (2) it allows for sufficient room to accommodate any tissue interposition that may be required in the reconstruction. Our series documents a high fistula closure rate that is comparable to most contemporary series [14, 15]. We were able to achieve a 95 % anatomical success due to near exclusive use of tissue interposition for our repairs. We used mainly autologous rectus fascia as it allows not only to cover over the urethrotomy closure defect but also to prevent secondary SUI associated with intrinsic sphincter defect induced by the UVF. In our series, three patients had both a rectus fascia and MFPG interposition. This combination is seldom needed unless there is a large urethral defect or poor vascularity compromising healing. One consideration is the MFPG should be less bulky than would be for a vesico-vaginal fistula repair given the restricted space around the urethra, and possibly a skin island can be accompanied to breach any vaginal wall defect if present. In addition, for recurrent cases, tissue interposition is imperative to reduce probability of operative failure. Pushkar et al. published a heterogeneous UVF series on 71 women including obstetric cases. Twenty-one women had mean follow-up of 99.6 months and only 9 had MFPG as part of their reconstruction. Primary success was 90 % with a postoperative SUI rate of 52 % (37 patients), all of which proceeded to anti-incontinence surgery [15]. Their experience certainly outlines the need to consider tissue interposition at the time of UVF repair to reduce the rate of secondary SUI. Interestingly, recurrent cases did no worse in outcomes than primary repairs, suggesting surgical technique and high volume referrals centers may contribute to the overall outcome.

The functional outcome is problematic for the management of UVF. Although most patients are amenable to anatomical closure, the same cannot be said of their functional outcomes. Our cohort, however, is novel as we compared outcomes from different etiologies (sling vs. non-sling) utilizing validated questionnaires, which is lacking in most surgical series. Our results from the UDI-6 demonstrated a higher rate of voiding dysfunction in the sling group. This is somewhat alarming that many of these women continued to have symptoms even at a mean of 4 years following their repair. We postulated that these persistent complaints resulted from either undiagnosed bladder outlet obstruction following initial sling placement and/or a direct consequence of the surgery itself, predisposing women to persistent storage voiding symptoms. Our posit may be supported by published data to suggest persistent bladder dysfunction despite adequate urethrolysis in patients with bladder outlet obstruction (BOO) following sling surgery. Leng et al. [16] have reported their findings of worsened functional outcomes with persistent voiding dysfunction in those with delayed versus early urethrolysis, suggesting that this may be associated with irreversible bladder dysfunction. Similarly Starkman et al. [17] retrospectively evaluated 40 patients undergoing urethrolysis for iatrogenic bladder outlet obstruction following anti-incontinence surgery and reported that overactive bladder symptoms may remain refractory in >50 % of patients despite an effective urethrolysis procedure, which impacted negatively on quality of life and the impression of improvement after surgery.

With respect to sexual function, we were unable to deduce much from our data due to small number of sexually active women in the postoperative setting, but we are concerned that this may be compromised and further research is necessary to address this issue.

In our study, 1 in 3 women required repeat procedures. Most interestingly, they were confined almost exclusively to the sling-related group with either recurrent SUI (3) or voiding dysfunction (2). From an anatomical perspective, this seems plausible, as the reconstruction often require extensive urethral dissection to remove the synthetic material before closing the fistula, and this dissection places both the intrinsic and extrinsic striated sphincters at risk of damage. More recently, there is data suggesting women who underwent repeat MUS for recurrent SUI have worsened urethral function alluding to the same underlying pathophysiological mechanism of sphincter compromise in these women. Stav et al. [18] reported a higher incidence of urodynamically confirmed intrinsic sphincter deficiency (ISD) in the repeat synthetic sling patient cohort (77 patients) compared to those who had primary slings (1035 patients) (31 vs. 13 %, p < 0.001) with mean follow-up duration of 50 months (range 12–114).

As with any retrospective data set, there are inherent biases including the lack of preoperative symptom questionnaire or knowledge on baseline voiding dysfunction in these women referred to our center from outside institutions. Furthermore, there was no comparable published study using validated tools to study functional outcomes after UVF repair. Finally, while we were able to document anatomical closure in all patients with postoperative VCUG, we were not able to report on all functional outcomes. Lost to follow-up from long-distance referral, lack of insurance coverage or relocation meant that we had a questionnaire responder rate of >60 %, which was quite acceptable but not complete. Despite these few limitations, our long-term results in this unique study argue for durable results and acceptable functional outcomes after a successful UVF repair, with some concerns nonetheless for sling-related UVF. Given the lack of outcome measures in general for UVF patients in current literature, we believe our case series do offer reconstructive surgeons the surgical approach coupled with validated outcomes that can be utilized as reference for counseling of patients with regard to surgical success.

Conclusion

This large series of non-radiated UVF indicates a satisfactory outcome in UVF closure repair. At long-term 4- to 5-year follow-up, differences in lower urinary tract outcomes for continence and voiding domains were noted more in the synthetic sling-related UVF group. Sexual function could not be fully explored as sexual activity among responders was low.