Phosphodiesterase type-5 inhibitors for erectile dysfunction following nerve-sparing radical prostatectomy

Supplemental Digital Content is available in the text


Introduction
Prostate cancer ranks first among the most commonly diagnosed cancers in the elderly population and may also be the most commonly non-cutaneous cancer in men. In 2014, in the USA it was estimated that 29,480 men died of prostate cancer. [1] Radical prostatectomy is still widely accepted for treating early localized prostate cancer. However, erectile dysfunction (ED) is common after nerve-sparing radical prostatectomy (NSRP) notwithstanding the increasing experience in surgical techniques. [2] Although several methods are used in penile rehabilitation after NSRP, the most common treatment is oral phosphodiesterase type 5 inhibitors (PDE5is) daily or on demand despite that the fact PDE5is is less effective in populations who suffer ED after NSRP compared with the general population. [3] Tadalafil, Vardenafil, Sildenafil, and Avanafil are all used for penile rehabilitation following NSRP, and the efficacy of these PDE5is has been demonstrated by more than one pair-wise metaanalysis when compared with placebo. [4][5][6] However, which strategy is better in clinical practice is still lacking evidence, thus making it difficult to provide a recommendation.
Due to the rarely head-to-head trials among different PDE5is treatments, it is difficult to answer such a question comprehensively using only the pair-wise meta-analysis method. Surprisingly, the network meta-analysis (NMA) method [7] can indirectly compare these treatments with the same comparator, and can also compare different treatments by combining direct and indirect comparisons simultaneously without losing randomization in individual trials.
Therefore, the aim of this study was to explore which PDE5is strategy is better in the treatment of ED following NSRP. A systematic review will be conducted to screen randomized controlled trials (RCTs) for further analysis. Then, a two-stage meta-analysis, namely traditional pair-wise meta-analysis and random-effects NMA, will be performed to answer this question considering all available data from included RCTs.

Methods
This study was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement. [8]

Search strategy
According to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, we conducted a systematic literature search in MEDLINE, Web of Science and Cochrane Central Register of Controlled Trials database to identify eligible studies from the startup of these databases to 1 November, 2019. Restricting to the human participants and English language study was the limitation of the search strategy. Population, Intervention, Comparison, Outcome and Study design (PICOS) principle was used to search. The MeSH terms and text words used in Medline were ((((("phosphodiesterase 5 inhibitors"[Mesh]) OR phosphodiesterase 5 inhibitors))) AND ("Erectile Dysfunction"[Mesh] OR Erectile Dysfunction))) AND ((((randomized OR random * ) OR ((Randomized Controlled Trial [Publication Type]) OR "Randomized Controlled Trials as Topic"[Mesh])))) AND (Prostatectomy [Mesh] OR Prostatectomy). Two independent reviewers performed the literature search. If disagreements appeared, two authors discussed the first. If they could not reach consensus, it was resolved by consulting with a senior author.

Eligible criteria
Studies included in systematic review that matched the eligible criteria were screened based on PICOS evidence listed below: (1) Participants: patients after NSRP, (2) Intervention: PDE5is treatment; (3) Comparison: compared with the placebo or PDE5is; (4) Outcomes: the efficiency evaluations; (5) Study design: only RCTs were included in meta-analysis.
The following types of literature were excluded from our study: meta-analysis, review, editorials, letters, comments, case reports, congress reports, and meeting abstracts.

Data extraction and Quality Assessment
Two authors independently extracted the information from each eligible study. Disagreements were resolved by discussion first. The third senior author will arbitrate if the disagreements still exist after discussion. The detailed data extraction information was shown in Table 1. The primary outcome was ED recovery rate according to individual results based on the International Index of Erectile Function (IIEF) questionnaires including IIEF, IIEF-EF and IIEF-5. The quality of included trials was evaluated according to the previously reported guidelines and the judgments for each quality item were classified as three grades: "high," "unclear," and "low." [8,9] Two independent authors assessed the quality of study and disagreements were resolved as the same as describe above.

Statistical Analyses
At the first stage, a traditional pair-wise meta-analysis was performed. Dichotomous variables in our study were expressed as odds ratio (OR) with 95% confidence interval (CI). The Q test and I 2 statistics were used to evaluate heterogeneity. I 2 > 50% or P < .1 indicated significant heterogeneity. [10] Referring to the previous studies, [11] random-effect model via the DerSimonian and Laird method was applied regardless of whether heterogeneity was high or low.
In the second stage, an NMA with random-effects approach was applied as previously described. [12] A network graph was plotted to show the network of the included comparisons. Global inconsistency and local inconsistency were tested using the Higgins and Dias model respectively. If the p value was greater than or equal to 0.05, a consistency model was performed. Otherwise an inconsistency model would be applied [13,14] because of the high risk of inconsistency. The biggest contribution of this model is that it introduces inconsistent parameters into the model, thereby theoretically avoiding the impact of confounding factors on the results. Based on the results of NMA, surface under the cumulative ranking curves (SUCRA) probabilities which is a commonly reported method in NMA, was used to evaluate the efficacy of different treatments. [7] Sensitivity analysis was performed by omitting the study mainly contributed to the inconsistency in the NMA. Publication bias and small-study effects were demonstrated by comparisonadjusted funnel plots. All data from our meta-analysis were analyzed by Review Manager 5.3 and Stata 14.

Direct Comparision between PDE5is and Placebo
Pooled evidence suggested that PDE5is followed by NSRP had a benefit for penile rehabilitation compared to placebo (OR = 2.67, 95%CI: 1.98, 3.59; P < .001) (Fig.1). We then performed subgroup analyses based on the PDE5is used, namely on demand (OD) or daily (nightly), kinds of PDE5is (Tadalafil, Vardenafil, Sildenafil and Avanafil), questionnaire categories, and therapy duration period. A stronger benefit was detected in the OD subgroup (OR = 3.00, 95%CI: 1.83, 4.91; P < .001) (Fig. 1A). significantly higher compared to placebo (Fig. 1B). Similar results were obtained in both subgroup analyses according to the questionnaire categories (Fig. 1C) and the therapy duration time (Fig. 1D). Table 1 Summary of included studies using oral PDE5Is for penile rehabilitation after nerve-sparing radical prostatectomy.   Figure S4, http://links.lww.com/MD/F792). Therefore, the following NMA was performed by inconsistency model, [13,14] which introduced inconsistent parameters into the model, thereby theoretically avoiding the impact of confounding factors on the results. Figure 2B showed the results of SUCRA enrolling all included studies. Avanafil 200 mg OD ranked first (SUCRA 83.5), thus having the highest probability of being the best intervention for improving ED recovery. The supporting material of the SUCRA, called the graph of the area under the curve was shown in Supplementary Figure S5, http://links.lww.com/MD/F793.

Sensitivity analysis and publication bias
The inconsistency in our NMA maily came from one study. [17] After omitting this study in sensitivity analysis, both the global inconsistency (chi 2 = 1.24, Prob > chi 2 = 0.538) and local inconsistency (Supplementary Figure S6 Figure 2C and Figure 3C, most of the plots were symmetrically distributed inside the 95% CIs in comparisonadjusted funnel plots which indicated a low risk of publication bias.

Discussion
Montorsi et al. first proposed penile rehabilitation in 1997. [31] A large number of studies including clinical trials and meta-analyses have been reported on this topic. However, it is difficult for pairwise meta-analysis to highlight this whole topic comprehensively because of the rarely head-to-head trials among different PDE5is treatments. To the best of our knowledge, this was the first NMA to verify which PDE5is strategy is better in the treatment of ED following NSRP. There were three core assumptions that should be considered in the meta-analysis, including homogeneity, consistency and similarity. In our meta-analysis, differences in ED recovery definition, kinds of PDE5is, dosage, usage, and treatment time might mainly contribute to the homogeneity. To reduce the homogeneity, we tried our best effort to remove the differences in the types of PDE5is, dosage, and usage. However, the residual homogeneity could not be avoided. Owing to the random effect model performed, our results became more conservative by considering homogeneity. Publication bias was not significant in our study. However, inconsistency existed in NMA. An inconsistency model was applied, [13,14] which introduced inconsistent parameters into the model, thereby theoretically avoiding the impact of confounding factors on the results. Surprisingly, there was no obvious inconsistency in the sensitivity analysis. In our study, similarity was also evaluated by detailed examination of each RCT. The quality of partly RCTs was relatively low under the guidance of the Cochrane library guideline. This led to the potential uncertainty risk of bias in our meta-analysis. In general, our results should be treated with caution.
Avanafil is a highly selective, quickly action and potent PDE5is that has been approved for the treatment for ED in general population. [30] In our study, pooled evidence suggested that Avanafil 200 mg OD might be most likely to be the best treatment option for ED recovery after NSRP according to the first rank of SUCRA both in NMA (83.5) and sensitivity analysis (90.2). This result is in accordance with a previous meta-analysis, which reported that Avanafil on demand was the most effective PDE5is in EF recovery. In addition, our study found that Avanafil 200 mg OD might be more effective than Avanafil 100 mg OD. Due to the rapid improvements after administration, [32] Avanafil OD seems to be more likely to be accepted by patients than planning to take it daily. However, it should be noted that the available evidence reporting Avanafil came from one trial. [30] Although the quality of this trial was relatively high, a greater proportion of patients discontinued the study (24%) in the placebo group than the Avanafil group (11%). In addition, this trial was sponsored by industry. Therefore, the efficacy should be interpreted with caution.
Sildenafil is also a short-term PDE5i like Avanafil, the SUCRA of Sildenafil 100 mg daily was second only to Avanafil 200 mg OD in our NMA. In sensitivity analysis, Sildenafil 50 mg OD was comparable with Sildenafil 100 mg daily. Therefore, we recommend that Sildenafil 100 mg daily and Sildenafil 50 mg OD might be both suitable regimens for ED recovery after NSRP. However, these studies related to Sildenafil were small sample size. In addition, the methodological quality of the included studies reporting Sildenafil was relatively low. Other limitations, such as differences in patient selection and baseline score, existed for recommending Sildenafil as a treatment option. Therefore, this result should be treated with caution.
Our NMA suggested that SUCRA of two regimens, namly Vardenafil 10 mg OD (48.6) and Vardenafil 5 mg daily (50.1) took the middle place. In the following sensitivity analyses, the removal of data contributing to inconsistency did not change significantly. A previous meta-analysis reported that Vardenafil significantly improved ED recovery only when used OD. [5] This bias was mainly caused by the different results of two included RCTs. [23,24] In the multicenter, large sample and double-blind head-to-head comparison, Vardenafil OD treatment resulted in significantly greater recovery than Vardenafil daily used. Both showed better ED recovery compared with placebo. [23] However, another single center and small sample RCT [24] reported that there was no significant difference between Vardenafil daily (10 mg) and placebo. In the NMA, the placebo acted as the same comparator, so the OD and daily regimen seemed to have a comparable effect.
In total, more well-designed RCTs need to be enrolled in our analysis to answer this question. The same bias existed in the indirect comparison between Tadalafil. After the study mainly contributing to inconsistency in NMA was excluded, [17] the sensitivity analysis in our study showed that Tadalafil 5 mg daily was more significant than Tadalafil 20 mg OD, which was consistent with the high quality RCT. [19] Some limitations need to be addressed when interpreting our results. First, the evaluation questionnaires were not the same in our study. Second, the different treatment periods among single studies could cause heterogeneity and influence therapeutic efficacy. Third, patient inclusion criteria after NSRP differed would limit the systematic interpretation of the efficacy in the treatment of ED recovery. Fourth, due to certain methodological deficiencies and low quality RCTs, bias existed in our study. Finally, although we removed the differences in PDE5is compound, dosage and usage, the inconsistency and homogeneity caused by the factors mentioned above, including patient selection in the included individual study, baseline erectile score, and relatively low quality of partly RCTs could not be reduced.

Conclusion
This is the first NMA to explore which PDE5is strategy is better for ED recovery following NSRP. We found that all PDE5is were effective compared to placebo. Avanafil 200 mg OD has the highest probability of being the best intervention according to this network meta-analysis based on the currently available evidence. Thus, it could be recommended for these patients in clinical practice. However, several limitations should not be ignored as described above. Besides, in consideration of the published data regarding Avanafil in the ED after radical prostatectomy is relatively small scale, larger sample and welldesigned head-to-head trials are needed to validate the role of Avanafil 200 mg OD for ED recovery after NSRP in the future.