Mesh inlay, mesh kit or native tissue repair for women having repeat anterior or posterior prolapse surgery: randomised controlled trial (PROSPECT)

To compare standard (native tissue) repair with synthetic mesh inlays or mesh kits.


Introduction
Mesh use for prolapse surgery is controversial. Government policy is changing in the light of increased evidence of adverse effects. 1 In women having a first prolapse repair, our own multicentre randomised controlled trial (RCT) PROSPECT, set in the UK, 2 demonstrated in the short term, at 2 years, that more than 30% of women still reported either 'something coming down' or had anatomical prolapse extending beyond the hymen, irrespective of the use of mesh inlay or biological graft to reinforce the surgery. As a result of the findings from this trial, augmentation is no longer recommended for a first repair. 1 However, Olsen et al. 3 showed that 30% of women who have had one prolapse or incontinence operation required at least one more procedure, and the time intervals between repeat procedures decreased with each successive repair. This study failed to differentiate between repeat surgery for a recurrence in the same compartment and primary surgery for a de novo prolapse in another compartment, or new continence procedures. 4 Nevertheless, Olsen et al. 's study suggests that a third of women would eventually undergo at least one more procedure and some would require a third or fourth one. 3 Our study was preceded by priority assessment based on the relevant Cochrane review 5 and an Interventional Procedures review which investigated the use of mesh for women having anterior and/or posterior vaginal wall prolapse surgery. 6 These and other findings were presented to the Interventional Procedures Advisory Committee (IPAC) in January 2008 and their guidance published. 7 The committee recommended that mesh should be used only under special arrangements for clinical governance, consent and audit or research: hence, the PROSPECT Study was funded to fill the evidence gap.
We therefore compared, in an RCT, the effect of mesh inlay or mesh kit with native tissue repairs in women who had already experienced at least one failed previous prolapse repair in the same compartment.
We focused on those high-risk women whose specific compartment prolapse surgery had already failed, to try to reduce the chance that they would require further prolapse surgery. After consultation with gynaecologists and experts from specialist societies, we chose to compare mesh kits as well as mesh inlays with standard (native tissue) repairs, based on the scarcity of data about the safety and efficacy of mesh kits but their perceived potential to provide better support due to their method of insertion.

Participants
Women listed for transvaginal repair of an anterior and/or posterior prolapse were eligible if at least one of the compartments requiring surgery had been repaired previously. Women could have concomitant uterine, vault or continence surgery. Women under the care of 59 gynaecologists from 33 UK centres were enrolled into the trial between January 2010 and August 2013. All women provided written informed consent. The study was funded by the National Institute for Health Research Health Technology Assessment Programme (Project Number 07/60/18). The funder (through their peer review and funding board review process) approved the study proposal but had no role in the collection, analysis or interpretation of data, or writing of the report.

Randomisation
A remote web-based computer-generated randomisation system at the Centre for Healthcare Randomised Trials (CHaRT, University of Aberdeen, UK) was used for group allocation. We report two trials: the Mesh Inlay Trial compared standard (native tissue) repair with mesh inlay; and the Mesh Kit Trial compared standard repair with mesh kit. Not all gynaecologists offered all treatment options due to preferences or locally available resources. Therefore, women were randomised in three strata: Stratum A included women randomised to one of all three treatment options, standard repair, mesh inlay and mesh kit (in a 1:1:2 ratio); Stratum B compared standard repair with mesh inlay (in a 1:1 ratio); Stratum C compared standard repair with mesh kit (in a 1:2 ratio). Randomisation was unbalanced in the Mesh Kit Trial in favour of mesh kits to account for the number of surgeons who were trained in their use, in order to ensure adequate numbers in the groups. Because the analyses were carried out separately for each trial, data from some women in the standard repair group from Stratum A were included in both trials.
The minimisation algorithm included: age (<60 years or ≥60); planned prolapse repair (anterior, posterior or both); planned concomitant urinary continence procedure or not; planned concomitant upper vaginal prolapse procedure or not; and operating surgeon.
Further details of participants, masking and interventions are provided as online Supporting Information.

Outcomes
Women were followed up at 6 months, and 1 and 2 years after surgery by postal questionnaire and were clinically examined at 1 year. 8 We used a wide and comprehensive panel of validated core outcomes relevant to women, focusing primarily on women's symptoms. These were based on internationally agreed terminology and recommended core outcomes. 4 The primary outcome was women's report of prolapse symptoms at 1 year after surgery using the Pelvic Organ Prolapse Symptom Score (POP-SS), a validated measure which has been shown to be sensitive to change after treatment. 9 The POP-SS contains items relating to frequency of seven prolapse symptoms in the previous 4 weeks: each item is scored from 0 (never) to 4 (all of the time): the total score thus ranges from 0 to 28. Secondary outcomes included prolapse-specific qualityof-life measured using a visual analogue scale (VAS) and generic quality of life based on the EQ-5D-3L, 10 and an assessment of overall global improvement in symptoms (PGI-I). 11 Bladder, bowel and sexual function were measured using validated or adapted International Consultation on Incontinence Questionnaires (ICIQ). 12 Objective measurement of prolapse stage utilised the POP-Q system. 8 Adverse events, need for readmission/further treatment for adverse effects or prolapse recurrence were reported by surgeons or women and verified by Study Office staff from a second source when possible. Adverse events and complications of surgery were recorded using the IUGA/ICS complications classification which includes type, severity, time of occurrence and site. 13,14 Serious adverse events were defined using standard classification. 15 All definitions are in keeping with the recommendations of IUGA, ICS and ICI. 4,[12][13][14]16 The full Protocol is available on the funder's website. 17

Statistical analysis
The main analysis was conducted on an intention-to-treat basis (whereby women with observed outcome data remained in their allocated group). We did not follow up randomised women who did not receive any surgery. We made two comparisons: standard repair versus mesh inlay (Mesh Inlay Trial, data from women in Strata A and B) and standard repair versus mesh kit (Mesh Kit Trial, from Strata A and C) ( Figure S1). Study analyses were conducted according to a prespecified statistical analysis plan, using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA).
All outcome measures were presented as summaries using descriptive statistics (mean and standard deviation for continuous measures, and proportion for ordinal and dichotomous measures) and comparisons between randomised groups were analysed separately at 6 months and 1 and 2 years using generalised linear models. Models were adjusted for minimisation covariates, baseline measures where appropriate, and randomisation stratum. Continuous outcomes were analysed using linear mixed models, with surgeon fitted as a random effect. POP-Q stage and PGI-I were analysed using ordinal logistic regression (proportional odds models with cumulative logits). Dichotomous outcomes were analysed using log-binomial regression. Estimates of treatment effect size were mean differences in the linear mixed models (including the analysis of the POP-SS), odds ratios in the ordinal models and risk ratios in the binary models. For all estimates, 95% confidence intervals were calculated.

Sample size
Women were recruited opportunistically alongside those having a primary repair. 2 Based on the assumption that 30% of women requiring an anterior and/or posterior repair would receive a secondary or subsequent operation, 3 we expected that approximately 1240 women having secondary surgery would be available during the recruitment period for the primary trial. 2 Of those available, it was estimated that 50% (620 women) would agree to be randomised.
Pilot data indicated that women having secondary repairs had worse symptoms at baseline than women having their first repair (A. Elders & C. Glazener, unpubl data). 18 We considered it biologically plausible that these women might show a larger benefit from surgical treatment than would women having their first repair. We therefore calculated that, with an expected sample size of 620, it would be possible to detect, with 90% power and alpha equal to 0.025, a standardised effect size of 0.38, which equates to three points on the POP-SS scale (assuming a standard deviation of 8 units).

Results
Between January 2009 and August 2013, 396 women were found to have recurrent prolapse of the same compartment and were therefore potentially eligible for this trial. However, only 155 (39%) agreed to be randomised, of whom 154 were included in the analyses. 19

Baseline characteristics and intervention received
The flow of women through the study is shown in the CONSORT diagram (Figure 1), in line with recommendations of the Consolidated Standards of Reporting Trials. 20 Women in the randomised groups were comparable at baseline and all were symptomatic based on at least one symptom on the POP-SS (Table S1). Two women did not receive surgery ( Figure 1). Most women received their planned surgery ( Figure 1).
In the standard repair group, more women (27% in the Mesh Inlay Trial and 20% in the Mesh Kit Trial) had a combined anterior/posterior repair than in the other two groups (synthetic mesh inlay 14% and mesh kit 13%; Table S2). Concomitant surgery included vaginal hysterectomy, which occurred in 7%/12% (respectively) of the standard repair groups, 8% with synthetic mesh inlay and 9% with mesh kit. A concomitant vault repair was more common in the native tissue group (25%/28%) compared with 10% in mesh inlay group and 13% in the mesh kit group. Finally, vaginal concomitant continence procedures were performed in 7%/8% of the native tissue repairs and in 4% of the mesh inlays; no continence procedures were performed with mesh kits.

Clinical symptoms and quality of life at follow up
Women's reports of prolapse symptoms (POP-SS) were less than half of the preoperative level (mean score before surgery 14.4 [SD 5.4], at 6 months 5.9 [5.7] Post randomisation exclusion: 1 woman had secondary prolapse surgery after consenting but prior to randomisation. She was followed up the cohort study (CC2). Other surgery includes women who did not have either an anterior or posterior repair, but did receive one or more of: tape for urinary incontinence, vaginal hysterectomy or suspension, cervical amputation, vault repair  at 1 year was maintained at 2 years, with respect to all the prolapse and quality of life outcomes measured.
In the Mesh Inlay Trial, the mean difference (MD) in the POP-SS score at 1 year for standard repair (mean 6.6, SD 6.0) versus synthetic mesh inlay (mean 6.1, SD 6.4), based on combined data from women in Stratum A (threeway randomisation) and Stratum B (two-way randomisation), was À0.41 (95% CI À2.92 to 2.11).
In the Mesh Kit Trial, the MD in the POP-SS score at 1 year for standard repair (mean 6.6, SD 5.5) versus mesh kit (mean 5.9, SD 5.3), based on combined data from women in Stratum A (three-way randomisation) and Stratum C (twoway randomisation), was À1.21 (95% CI À4.13 to 1.72).
At 2 years, the study found that women having a mesh kit had a better generic QoL score, measured with EQ-5D-3L, than did those who had a standard repair (MD 0.13 [95% CI 0.02-0.25]: P = 0.025; Table 1).
The other key symptoms of pelvic floor dysfunctionurinary, faecal, vaginal and sexual symptomsare presented in Tables 1 and S4. Although there was a decrease in the proportion of women with severe urinary incontinence after surgery, there was no difference between the randomised groups in either the Mesh Inlay Trial or the Mesh Kit Trial in respect of any of the urinary outcomes measured. Frequency of bowel movement and constipation were largely unchanged after prolapse surgery. There were no differences between the randomised groups in respect of any of the bowel outcomes measured.
Many women reported improvements in their vaginal and sexual function outcomes after surgery: this was evident from a reduction of the ICI-Q Vaginal Symptoms score (Table 1). After surgery, fewer women cited prolapse symptoms as a reason for not having a sex life (around 35% before surgery, versus around 10% after). Four women had severe dyspareunia at 1 year (Mesh Inlay group = 3, Mesh Kit group = 1), but only one at 2 years (Mesh Inlay group). However, there were no statistically significant differences between the randomised groups in respect of any of the vaginal or sexual symptom outcomes measured. 19

Satisfaction with treatment
Most women reported that their prolapse symptoms were very much or much better than before surgery, with no statistically significant differences between the groups in either trial (Table S4).

Objective outcomes
At 1 year, 83% of women attended for clinical review. Objective measurement showed improvement in each of the three prolapse compartments. The proportion of women with the leading prolapse edge beyond the hymen (POPQ >0 cm) reduced substantially. In the Mesh Inlay Trial, the difference between groups based on clinician's estimates of stage was RR 0.75, 95% CI 0.33-1.68, P = 0.479, and the proportion with more severe objective prolapse defined as 'leading edge of the prolapse at >0 cm beyond the hymen on POP-Q' was 14% in each group (RR 0.59, 95% CI 0.18-1.92, P = 0.380 (Table 2).
In the Mesh Kit Trial, women who had a standard repair were more likely to have prolapse compared with those who were randomised to mesh kit based on clinician's estimates of all stages (RR 0.24, 95% CI 0.07-0.83, P = 0.024;   Table 2). However, for more severe objective prolapse (defined as above), 3/18 (17%) of standard repair women had residual prolapse compared with none of 35 women after a mesh kit procedure.

Readmission, adverse effects and further treatment
Five women were readmitted in the first 6 months after surgery (Standard group = 2, Mesh Inlay group = 3; Table 3). Two subsequent readmissions were for revision of prolapse surgery (Fenton's operation)-one in the Mesh Kit group and one in the Mesh Inlay group.
In the first 2 years, 7/52 women had vaginal mesh exposure in the mesh inlay arm (13%), of whom four required surgical revision; and 4/46 women had vaginal mesh exposure in the mesh kit arm (8%), of whom two needed surgical revision. Hence in total, six women needed further surgery to address an area of mesh exposure (all but one <1 cm 2 ) 2 years after surgery. There were no reports of mesh perforation of the bladder or bowel at insertion but one woman in the Mesh Inlay Trial experienced a bowel perforation during mesh removal. The other mesh exposures were managed conservatively by observation, topical estrogens or cautery.
At 2 years after surgery, around 20% of women who had a standard repair required further treatment for prolapse compared with 11% who had a mesh inlay and 5% who had a mesh kit (Table 3): however, there was no statistical difference between the randomised groups.

Main findings
There were no statistically significant differences at 1 year in the primary clinical outcomes after prolapse surgery using native tissue, polypropylene non-absorbable mesh or a mesh kit to reinforce the repair in either trial. The uncertainty around this finding is reflected in the wide confidence intervals around the primary outcome (POP-SS) at 1 year (RR À0.41, 95% CI À2.92 to 2.11 in the Mesh Inlay Trial and RR À1.21, 95% CI À4.13 to 1.72 in the Mesh Kit Trial). Women in the Mesh Kit Trial were less likely to have prolapse beyond Stage 2 at 1 year and had a higher (better) EQ-5D-3L score at 2 years. However, these may have been chance findings and their clinical significance is uncertain as there were no differences in any other subjective outcomes between the randomised groups at any time point.
The overall incidence of non-mesh-related serious adverse events was around 10% and comprised primarily pain, infection and urinary retention. As women could only have a mesh-related complication if they received mesh,

Strengths and limitations
PROSPECT is rare in being one of the few RCTs in the field to distinguish rigorously between primary and secondary surgery. Unfortunately, our secondary trials on their own did not attain sufficient power to detect a difference. In future, studies should report prolapse surgery trials using the subgroups of Primary and Secondary (the latter defined as 'repeat surgery in the same compartment').
Another strength was the pragmatic reflection of actual practice in the UK. We included surgeons from a large number of hospital settings. It was not possible for all surgeons to randomise women between all three options, but the analysis by strata accommodated this.
Our secure and effective randomisation programme ensured that women were comparable at baseline and that concomitant surgery and other confounding variables were accounted for. We used validated outcome measures to measure women's symptoms of pelvic floor dysfunction. We captured a wide range of adverse effects and made efforts to verify these from alternative sources when possible. Essential missing data were actively sought from the women. Participants, outcome assessors and data entry clerks were blinded to randomisation as far as possible. Readmissions 0-6 months 4.0% 2/50* 6.4% 3/47** 1.76 0.30-10.37 0.532 0% 0/22 0% 0/43 n/a n/a n/a Readmissions 6-12 months 0% 0/49 0% 0/44 n/a n/a n/a 0% 0/21 2.3% 1/44*** n/a n/a n/a
Furthermore, we identified fewer women than expected (396 rather than 1240) because of our more rigorous definition of repeat surgery (same compartment rather than any compartment, 30%). 1 In addition, fewer women than expected were randomised (39% rather than 50%): this was more often due to a clinical decision rather than the women's choice (54% versus 39%). This resulted in fewer women than estimated (155 rather than 620) being randomised. Thus, we were not able to recruit to the sample size that would have given us enough power to identify a difference of three points on the POP-SS.
We and other researchers have suggested that prolapse beyond the hymen (>0 cm on POP-Q) is a sign of severe objective failure. 21 However, we recognise that women with worse anatomical findings may not have symptoms and, vice versa, women with an objective 'cure' may still have prolapse symptoms.
Longer follow up is required: the average time to a repeat operation (in any compartment) is around 12 years. 1 While we did not identify differences in the repeat surgery rate between the groups, it is likely that 2 years is too short a time scale to provide a definitive answer. Both the natural history of prolapse and the longterm ongoing tissue interactions with polypropylene indicate that it is important that trials pursue longer term follow up of outcomes and complications, ideally over 12 years. 3 We have commenced follow up of the PRO-SPECT women for at least 6 years after surgery and also plan electronic data linkage to capture outcomes from non-responders about further admission for prolapse surgery.

Interpretation (in light of other evidence)
The most recent Cochrane review 22 identified 37 trials of mesh or graft in women having anterior, posterior or apical prolapse surgery. Only two of those RCTs published separate data from women having repeat surgery for recurrent anterior or posterior prolapse, 23,24 although most trials included some such women. Both relevant RCTs compared native tissue repair with a mesh kit (Prolift â Gynaecare, inserted with trochars, which is no longer available). Altman reported only a composite failure rate at 1 year in 53 women (20/25 versus 14/28). 23 Withagen reported longterm prolapse symptoms at 7 years in up to 142 women (sensation of bulge at 7 years: 17/76 with native tissue versus 14/66 with mesh kit). 24 Although anatomical failure rates were less in the mesh group (47/67 versus 28/53 at 7 years), there was little difference in the further prolapse surgery rate (11/69 versus 14/56). The mesh exposure rate in the mesh kit group (42% at 7 years) was high and 54% were symptomatic, while a third of the women required surgical revision. Neither trial was conclusive.
Meta-analysis of the PROSPECT data with the two trials examining mesh kits showed that more women had prolapse symptoms with native tissue (68%) than with mesh kits (42%) in the first year (RR 1.56, 95% CI 1.11-2.18) but this difference did not persist in the longer term (23 versus 27%: RR 0.92, 95% CI 0.55-1.52). More women had persistent objective prolapse at 1 year (40 versus 7%, RR 4.97, 95% CI 2.52-9.81) but this was not reflected in repeat prolapse surgery rates (17 versus 18%, RR 0.85, 95% CI 0.46-1.54) at up to 7 years' follow up. Four of 52 women required surgery for mesh exposure after a mesh inlay, and 9/99 women after a mesh kit, although many mesh exposures were small and asymptomatic.
PROSPECT was the only trial to compare native tissue repairs with synthetic mesh inlays.
In any case, in the light of increased evidence of adverse effects, it is unlikely that further trials of mesh will be conducted. 1

Conclusions
Based on the evidence available in this trial, we are unable to say whether a mesh inlay or a mesh kit confers more benefit to women having a repeat prolapse repair than native tissue surgery in the first 2 years after surgery. Some women required an additional surgical procedure to remove exposed mesh, which may be considered to be an unnecessary risk. However, long-term follow up may reveal whether the excess risks are offset by a potential decrease in the need for repeat surgery, which in itself is associated with higher risks.
PROSPECT is rare in being one of the few RCTs to distinguish rigorously between primary and repeat surgery. We would strongly encourage future studies to use our approach. Although our trial did not have sufficient power to demonstrate a statistical difference, the information is available for meta-analysis with other trials. 19 Further longterm follow up will ultimately determine whether the use of synthetic mesh in vaginal prolapse repair confers any long-term benefits in women whose prolapse surgery has already failed at least once. Large international datasets will be required to make true progress in this field.

Disclosure of interests
Mr Elders reports that his institution (Glasgow Caledonian University) received payment from the University of Aberdeen for statistical analysis undertaken by him. Prof. Norrie reports grants from the University of Aberdeen during the

Supporting Information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Figure S1. Extended CONSORT diagram for secondary trial. Table S1. Baseline characteristics of the intention-totreat population. Table S2. Surgery actually carried out, including concomitant procedures. Table S3. Clinical symptoms and quality of life outcomes at 6 months. Table S4. Condition-specific quality of life measures and satisfaction at 1 and 2 years. Table S5. Serious and related adverse effects within first and second years. Table S6. Other (non-serious) related adverse effects within first and second years. &