Pelvic Floor Reconstruction After Radical Prostatectomy: A Systematic Review and Meta-analysis of Different Surgical Techniques

Radical prostatectomy (RP) is the gold standard for the treatment of localized PCa. A meta-analysis was conducted to evaluate the effect of different techniques of pelvic floor reconstruction on urinary continence. A comprehensive search was made for trials that evaluated the efficacy of pelvic floor reconstruction. Relevant databases included PubMed, Embase, Cochrane, Ovid, Web of Science databases and relevant trials from the references. Random-effects model was used to estimate risk ratios (RRs) statistics. Pooled results of patients treated with posterior reconstruction (PR) demonstrated complete urinary continence improved at 1–4, 28–42, 90, 180 and 360 days following catheter removal. Anterior suspension (AS) was associated with improvement only at 28–42 days. The anterior reconstruction (AR) + PR was associated with urinary continence at 1–4, 90 and 180 days. AS + PR was not associated with any benefit. And PR improved social urinary continence at 7–14 and 28–42 days. No benefit was associated with AS. AR + PR had better outcomes at 90 and 180 days. AS + PR was significant improved at 28–42 and 90 days. Patients who underwent RP and PR had the least urinary incontinence. No significant benefit was observed after AS. AR + PR and AS + PR had little benefit in the post-operative period.

et al. 8 and later combined with PR to yield an incremental benefit (AR + PR) [9][10][11] . A simple anterior suspension (AS) technique using sutures anchored to the pubic bone was first described by Sugimura et al. to improve early urinary continence 12 . The effect of anterior suspension combined with posterior reconstruction (AS + PR) has also been examined. Now the effect of different surgical techniques for improving urinary continence is not clear yet. Rocco et al. 13 reported a meta-analysis of posterior reconstruction technique and several trials have been conducted to evaluate the time to urinary continence after LRP and RARP. However, the previous study didn't evaluate other surgical techniques. The publication of new studies evaluating PR, AS, AR + PR, and AS + PR add to the power of a meta-analysis. We conducted a meta-analysis evaluating the continence rate at different time intervals after different surgical techniques.

Results
354 trials were identified by reviewing abstracts and articles. 159 duplicates were removed. Nine additional trials were excluded because there was no comparison group, outcome data was incomplete, it was a review article, or the article was not in English. The final set of trials eligible for analysis included 32 studies for the qualitative analysis 7,[9][10][11][12] . The selection strategy is shown in Fig. 1. The characteristics of the included trials are outlined in Table 1. A total of 4697 patients were included in this meta-analysis. 19 trials 7, 15-32 evaluated the efficacy of PR, 7 trials 12, 33-38 evaluated the efficacy of AS, 4 trials 9-11, 14 evaluated the efficacy of PR + AR, and 2 trials 39, 40 evaluated the efficacy of PR + AS. Seven of these trials were RCTs 9,15,31,32,37,38,40 . Six trials 11,18,25,29,32,33 evaluated IPSS and EPIC urinary domain scores.

Effect of surgical treatment on IPSS and EPIC urinary domain scores. IPSS and EPIC urinary
domain scores were reported in six studies 11,18,25,29,32,33 . Kojima et al. 33 reported a median IPSS score before surgery of 12.5 in the AS group and 7.0 in the control group. These values were 11.0 and 16.0, respectively, 4 weeks after surgery (P < 0.05). No benefit was also seen at week 12 or week 24. Sutherland et al. 32 reported that both the PR and control groups had a significantly improved IPSS score from postoperative week 6 to month 3 (P < 0.01). Krane et al. 29 found no difference in the IPSS score of the AS and control groups (8.2 vs 8.1, P = 0.97).
"Urinary function" and "urinary bother" subscale score from the EPIC urinary domain were also reviewed. Hoshi et al. 11 found that the proportion of recovery to baseline score was significantly improved in the "urinary function" subscale score at 12 months after surgery (P < 0.01) No significant improvement was found at other time points for the "urinary function" or at any time point for the "urinary bother" subscale score. Different outcomes were reported by Ito et al. 18 and Brien et al. 25 . Both found "urinary function" and "urinary bother" subscale scores to be significantly higher in the PR treated group, compared to a control group, at 3 months after surgery. Ito et al. 18 found a significant improvement in "urinary function" and "urinary bother" subscale scores at 6 months after surgery when PR was performed. In contrast, Brien et al. reported no benefit in these scores 6 months after catheter removal 25 .
Quality assessment of RCTs and historical cohort studies. The Jadad quality scores and methodological Newcastle-Ottawa scales are listed in Table 2. The quality of cohort studies was mostly high, but the level of evidence was low because of the nature of the study designs. Because of the lack of double blind for a surgery, the score for double blind in mostly studies was 0, expect one 9 . The quality of most RCTs was still high, and the level of evidence was stable expect one study 38 . Publication bias. Funnel plots of urinary continence at six time intervals showed only one publication with bias, in the AS treated group at 28-42 days (Begger test P = 0.089, Egger test P = 0.002). This bias could be due to the small number of patients with follow-up. No evidence of publication bias was found at any time interval with the other surgical treatments used (Figs S2-S8) ( Table 3).

Discussion
This meta-analysis included 7 randomized studies and 25 historical cohort studies of different urethral reconstruction methods after radical prostatectomy, including PR, AS, PR + AS and PR + AR. A quantitative synthesis of the evidence can be really helpful for urologist because urinary incontinence is the major problem after radical prostatectomy.
Urinary incontinence could be improved by many techniques, such as pelvic floor reconstruction, bladder neck preservation 41 or intussusceptions 42 , preserving the fascia covering the levator ani muscle 43 and preserving neurovascular bundles 44 . Among these techniques, pelvic floor reconstruction was reported most. The reconstruction prolonged a little surgery time and gained benefit in improving urinary continence. And the hot point for reconstruction is which layers to be sutured and how to suture. So many studies used different methods to improve the urinary continence compared to the common technique in this meta-analysis.
Patients were evaluated at a large number of time points for both complete and social continence, and a large number of surgical techniques were evaluated. Evaluation of pooled results demonstrated an improvement in urinary continence using these techniques. PR group outcomes in this meta-analysis were similar to the results in Rocco et al. 13 , but two different points should be noticed. First, we analyzed complete continence and social continence, respectively. Second, we used 1-4, 7-14, 28-42, 90, 180 and 360 day after catheter removal as cut-off point. Meanwhile, no differences in PSM and cystogram leakage were identified.
Treatment of patients with PR improved the complete urinary continence rate at 0-4, 28-42, 90, 180 and 360 days after catheter removal, but not at 7-14 days. These findings are similar to those reported by Grasso et al. 8 and Rocco et al. 13 . Rocco et al. 13 found no improvement in the urinary continence rate at 3 and 6 months after catheter removal. This finding was similar to the improvement in social urinary continence rate seen with the pooled data. The different inclusion criteria used and different number of trials evaluating different outcomes could have contributed to some of the different findings. AS provided no benefit of complete or social urinary continence, There are some kinds of potential heterogeneity in this meta-analysis. First, surgical technical differences were reported in each of the surgical reconstructions, although these were felt to be minor. For example, Patel et al. 34 anchored the anastomosis to the pubic bone, while Noguchi et al. 36 anchored to the dorsal venous complex (DVC) and puboprostatic ligaments. Second, different methods were used to evaluate continence including a self-administrated questionnaire, EPIC questionnaire, valsalva leak-point pressure, and pad weighing. Third, different study designs including the variable use of a nerve-sparing technique, variations in reporting times, and differences in the historical cohorts used as control groups could have influenced the outcomes. We did not distinguish randomized studies from historical cohort studies because of the small number of reported trials. Finally, the difference in the number of patients treated in each study could introduce bias into our analysis. These potential effects make high heterogeneity of results. It's impossible to control these differences in each trial.
Bias due to different study designs may be greater in subgroup analyses. Both complete and social urinary continence was present only at 1-4 days in RCTs, where heterogeneity was generally low. Complete urinary incontinence was observed at 7-14 days and social urinary incontinence at 180 days in historical studies. These differences could occur because RCTs better control patient related bias and also because there may be small differences in the surgical technique used in the two groups. The IPSS and EPIC urinary domain score was analyzed in this meta-analysis. Because the scale scores were not well described using RR, and so were individually described by report. This is another method to assess the postoperative urinary continence.
There were several limitations to this study. First, only publications reported in English were included because of the lack of a translator. Second, the individual patient data was not available for each study which is the gold standard for meta-analysis. Third, conference abstracts were also not included because of lack of available data. These factors could have reduced the number of trials evaluated in this meta-analysis. Fourth, heterogeneity and variation in study quality, as described above, could also have affected results. Lastly, different time intervals among the included studies also influenced the outcomes despite of grouping sections. These limitations may make the results unstable, so further studies are still needed to explore the effect of these surgical techniques in RP.

Conclusion
Patients with PCa who underwent RP with PR had the least urinary incontinence. PR is currently one of the most widely used surgical reconstructive techniques to improve the adverse effect of RP. No benefit was observed after AS. AR + PR, while AS + PR, might have little influence at early time points, but had the best outcomes at 180 or more days. More RCTs are needed to better assess the efficacy of different surgical reconstructions after RP.

Methods
Selection Criteria. Studies that were published in English were selected if they met the following criteria: (1) all patients were diagnosed with PCa by clinical examinations and prostate biopsy; (2) all patients underwent radical prostatectomy; and (3) the surgical modification was AS, AR, PR, AS + PR or AR + PR. Studies of patients who received neoadjuvant treatment were excluded.
Search Strategy. This meta-analysis was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement 15 . To identify studies that met the above selection criteria, we searched the PubMed, Embase, Cochrane Central Register of Controlled trials, Ovid and Web of Science databases for trials published before June 6, 2016. The search strategy was followed using all possible combinations of the medical subject headings (MeSH) or non-MeSH terms including prostate neoplasm, prostatic neoplasm, and prostatic cancer; posterior reconstruction, anterior reconstruction, anterior suspension, pelvic floor reconstruction and total reconstruction; urinary incontinence and incontinence or urinary continence and continence. Each    Outcome Measures. The primary outcome measure in this meta-analysis was complete urinary continence rate. Complete urinary continence was defined as using 0 pad per day. The secondary outcome measure was social urinary continence. Social urinary continence was defined as using 0-1 pads per day. The study group Randomized controlled trial (Jadad score)  Statistical Analysis. RRs with 95% CIs were used to evaluate the primary outcome and secondary outcome. A RR > 1 indicated an advantage of reconstruction over non-reconstruction (NR). Heterogeneity across studies was quantified using the I 2 statistic and the Chi-square (Cochrane Q statistic) test. Studies with an I 2 statistic greater than 40% and a P value less than 0.1 for the Chi-square test had a high level of heterogeneity. A random-effects model was used to pool estimates regardless of high or low levels of heterogeneity in order to better deal with the heterogeneous nature of the different surgical modifications. Study designs, surgical modifications and other confounding factors were not consistent between studies. Therefore, there was a significant advantage of a random-effects model compared with a fixed-effects model in accounting for heterogeneity between studies 16 . A p value less than 0.05 was considered statistically significant. All statistical analyses were performed using STATA version 13.0 (College Station, Texas, USA).
Quality Assessment. The methodological quality of each randomized controlled trial (RCT) was evaluated using the Jadad scale 17 . Quality was assessed using presence of randomization (0-2 points), used of double blind (0-2 points) and presence of patient withdrawals and dropouts (0-1 point). The 2 reviewers classified studies into two quality grades: low (0-2 points) and high (3-5 points). The methodological quality of each cohort study was evaluated according to the Newcastle-Ottawa Scale (NOS) 18 . Method of selection of the study groups (0-4 points), comparability of cohorts (0-2 points) and ascertainment of the outcome (0-3 points) were the three major aspects used for calculating the quality score of included reports. The studies were classified into three quality grades: low (0-3 points), moderate (4-6 points) or high (7-9 points). All studies were evaluated using the level of evidence (LOE) defined by Phillips et al. 19,[45][46][47][48][49] . Two independent reviewers evaluated each study. Disagreements were resolved through discussion.  Table 3. Pooled results of complete urinary continence, social urinary continence, PSM rates and publication bias of comparing different surgical techniques and time points.