Oral Phosphodiesterase-5 Inhibitors and Hormonal Treatments for Erectile Dysfunction: A Systematic Review and Meta-analysis

doi:10.1059/0003-4819-151-9-200911030-00150

Oral Phosphodiesterase-5 Inhibitors and Hormonal Treatments for Erectile Dysfunction: A Systematic Review and Meta-analysis

Alexander Tsertsvadze, MD, MSc;
Howard A. Fink, MD, MPH;
Fatemeh Yazdi, MSc;
Roderick MacDonald, MSc;
Anthony J. Bella, MD;
Mohammed T. Ansari, MBBS, MMedSc, MPhil;
Chantelle Garritty, MSc;
Karla Soares-Weiser, MD, PhD;
Raymond Daniel, BA;
Margaret Sampson, MLIS;
Steven Fox, MD, MPH;
David Moher, PhD; and
Timothy J. Wilt, MD, MPH

From Ottawa Health Research Institute, University of Ottawa, and Children's Hospital of Eastern Ontario Research Institute, Ottawa, Ontario, Canada; Geriatric Research Education and Clinical Center, Center for Chronic Disease Outcomes Research, Minneapolis Veterans Affairs Medical Center, and Minneapolis Evidence-based Practice Center, Minneapolis, Minnesota; Enhance Reviews, Kfar-Saba, Israel; and Agency for Healthcare Research and Quality, Rockville, Maryland.

Abstract

Background: Erectile dysfunction (ED) is a common male sexual disorder. The relative benefits and harms of pharmacologic therapies for ED, as well as the value of hormonal testing in men with ED, are uncertain.

Purpose: To evaluate the efficacy and harms of oral phosphodiesterase-5 (PDE-5) inhibitors and hormonal treatments for ED and assess the effect of measuring serum hormone levels on treatment outcomes for ED.

Data Sources: English-language studies from MEDLINE, EMBASE, Cochrane Central Register of Controlled Trials, PsycINFO, AMED, and SCOPUS through April 2009. Trial reference lists also were scanned.

Study Selection: Randomized, controlled trials (RCTs) of oral PDE-5 inhibitors and hormonal treatment for ED, and observational studies reporting measurement of serum hormone levels, prevalence of hormonal abnormalities, or both in men with ED.

Data Extraction: Two independent reviewers abstracted data on study, participant, and treatment characteristics; efficacy and harms outcomes; and prevalence of hormonal abnormalities.

Data Synthesis: Data, primarily from short-term trials (≤12 weeks), indicate that PDE-5 inhibitors were more effective than placebo in improving sexual intercourse success (69.0% vs. 35.0%). The proportion of men with improved erections was significantly greater among those treated with PDE-5 inhibitors (range, 67.0% to 89.0%) than with placebo (range, 27.0% to 35.0%). The PDE-5 inhibitors were associated with increased risk for any adverse events compared with placebo (for example, relative risk with sildenafil, 1.72 [95% CI, 1.53 to 1.93]). In 4 head-to-head RCTs comparing sildenafil, vardenafil, and tadalafil, improvement of ED and adverse events did not differ among treatments. Results from 15 RCTs evaluating hormonal treatment of ED were inconsistent on whether treatment improved outcomes. Evidence was insufficient regarding whether men with ED had a higher prevalence of hypogonadism than men without ED.

Limitations: Many RCTs were of low methodological and reporting quality, particularly those involving hormonal treatments or directly comparing different PDE-5 inhibitors. Most RCTs provided only short-term efficacy and harms data.

Conclusion: Oral PDE-5 inhibitors improved erectile functioning and had similar efficacy and safety profiles. Results on the efficacy of hormonal treatments and the value of hormone testing in men with ED were inconclusive.

Primary Funding Source: Agency for Healthcare Research and Quality.

Erectile dysfunction (ED) is a common male sexual disorder and is defined as the persistent inability to achieve or maintain penile erection sufficient for satisfactory sexual performance (1). Advanced age, diabetes, vascular diseases, psychiatric disorders, and possibly hypogonadism are associated with increased prevalence of ED (1–4). According to data from the Massachusetts Male Aging Study (2), the prevalence of ED in men aged 40 to 70 years is about 50%.

Unless contraindicated, oral phosphodiesterase-5 (PDE-5) inhibitors, such as sildenafil, tadalafil, or vardenafil, are currently first-line treatments of ED (5). Alternative treatments include hormones, vacuum constriction devices, intraurethral suppositories, intracavernosal injections, or surgery (for example, penile prosthesis implants) (5–7). Direct comparisons between treatments have been limited, and the relative long-term efficacy and safety profiles of different therapies have not been adequately explored.

Estimates from the National Health and Nutrition Examination Survey suggested that annual U.S. treatment costs of ED could reach $15 billion if all affected men sought care (8). In the past decade, the use of diagnostic tests for underlying causes of ED markedly decreased and use of pharmacologic therapy, especially with oral PDE-5 inhibitors, increased (8). In 2005, sales of sildenafil, tadalafil, and vardenafil were $1.6 billion, $747 million, and $327 million, respectively (9–11).

The value of hormonal blood tests (such as testosterone) in routine evaluation of men with ED is uncertain (12–15). The European Association of Urology and the British Society for Sexual Medicine guidelines recommend endocrinologic screening in the initial evaluation of all men with ED (15–20). In contrast, the American Urological Association recommends that hormone testing in men with ED be based on initial clinical assessment (for example, decreased libido, small testes, and reduced body hair) or failure of initial PDE-5 therapy management (5). Whether this targeted approach for identifying and treating hormonal disorders as underlying causes of ED is appropriate has not been rigorously evaluated (13, 15, 20, 21).

The aims of our review were to systematically identify and synthesize the published evidence to determine 1) the relative benefits and harms of oral PDE-5 inhibitors and hormonal treatments of ED, including the effect of patient characteristics and comorbid conditions on the likelihood of treatment success and 2) the clinical value of hormonal blood testing for identifying treatable causes of ED (for example, hypogonadism) and improving its outcomes.

We summarized and updated evidence from a technical review prepared for the Agency for Healthcare Research and Quality (22) and adapted the article in collaboration with the American College of Physicians' Clinical Efficacy Assessment Subcommittee to inform the development of its clinical practice guideline on this topic. The article focuses more specifically than the Agency for Healthcare Research and Quality technical review on treatments likely to be prescribed by primary care physicians. The topic of diagnosis and treatment of ED, as a subject for systematic review, was originally nominated by the American College of Physicians.

Previous Section Next Section

Methods

Data Sources and Searches

We searched for English-language articles in MEDLINE (1966 to May 2007), EMBASE (1980 to week 22 of 2007), Cochrane Central Register of Controlled Trials (second quarter of 2007), PsycINFO (1985 to June 2007), AMED (1985 to June 2007), and SCOPUS (2006). Search terms were impotence; erectile dysfunction; randomized, controlled trial; and controlled clinical trial. We also scanned reference lists of retrieved publications. We updated the review by searching MEDLINE and EMBASE (May 2007 to April 2009).

Study Selection

To assess the relative benefits and harms of pharmacologic treatments for ED, we selected randomized, controlled trials (RCTs) of pharmacologic treatments in men aged 18 years or older with ED. Treatments not generally prescribed by primary care physicians, such as vacuum constriction devices, intraurethral suppositories, intracavernosal injections, or psychotherapy, were considered beyond the scope of this review and are addressed in the technical report.

To assess the risks for nonarteritic anterior ischemic optic neuropathy (NAION) in PDE-5 inhibitor users, we selected RCTs; nonrandomized, controlled trials; and observational studies. To assess the clinical value of routine hormonal blood tests in men with ED, we selected studies that reported prevalence of hypogonadism, hyperprolactinemia, or both in men with ED and all RCTs comparing hormone treatment alone or in combination versus control in men with ED.

We excluded reviews, pooled analyses, editorials, commentaries, and letters. Two independent reviewers screened all identified titles and abstracts and, for articles considered potentially eligible, abstracted their full-text reports. Discrepancies were discussed and resolved by consensus. Appendix Figure 1 shows the literature selection process.

View larger version:

Appendix Figure 1. Literature search and selection.

NAION = nonarteritic anterior ischemic optic neuropathy.

Data Extraction and Quality Assessment

Two reviewers independently abstracted data on study, population, and treatment characteristics. Treatment efficacy outcomes were the proportion of successful sexual intercourse attempts based on either participants' diaries or event logs (erection sufficiently hard and long-lasting for satisfactory intercourse) or participants' responses to question 3 of the Sexual Encounter Profile (SEP) (erection lasted long enough for successful intercourse) (23); the improvement in erectile function based on either participants' self-reports of improved erections (global assessment or efficacy question 1) or the mean Erectile Function domain score of the International Index of Erectile Function (IIEF) (24); and participants' responses to IIEF questions 3 (successful penile penetration) and 4 (maintenance of erection after penetration) (24).

Abstracted adverse events data were the number of patients with any adverse event, specific adverse events, withdrawals due to adverse events, serious adverse events, and serious cardiovascular adverse events. We assessed the prevalence of hypogonadism or hyperprolactinemia by using the definitions provided by study authors, even though these may have differed between studies.

We evaluated the overall strength of evidence by using a method developed by the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) group (25). We used the Jadad scale to assess the methodological and reporting quality of RCTs (score range, 0 to 5; higher score indicates better quality) (26). We judged the adequacy of allocation concealment to be adequate, inadequate, or unclear by using the approach proposed by Schulz and Grimes (27). We assessed the quality of studies reporting serum hormonal levels by using a subset of the Quality Assessment Tool of Diagnostic Accuracy Studies (questions 1, 2, 8, 12, and 14) (28). To explore the overall risk for bias, we generated risk-for-bias graphs (29). Appendix A includes all evidence and quality assessment tables and risk-for-bias graphs for the included studies.

Data Synthesis and Analysis

We qualitatively summarized data on study (design, reporting quality, and sample size), population (age, severity of ED, and comorbid conditions), and treatment characteristics (dose, frequency, and duration). We considered studies suitable for pooling if they used the same design (RCT), enrolled similar populations (trials restricted to participants with a specific comorbid condition vs. those enrolling participants with a heterogeneous profile of comorbid conditions), evaluated the same type of treatment (for example, sildenafil), and reported the same efficacy or safety outcomes. We used DerSimonian and Laird random-effects models to generate pooled estimates of relative risks (RRs) and weighted mean differences (WMDs) with 95% CIs (30). To avoid double-counting during pooling of a trial with several PDE-5 dose groups versus placebo, we used a generic inverse variance method to combine data from these groups for a single estimate of mean response rate versus placebo. Statistical heterogeneity was evaluated using a chi-square test and the I ² statistic (low = 25.0%; moderate = 50.0%; high = 75.0%) and was explored through subgroup and sensitivity analyses (for example, study quality and random- or fixed-effects model) (31). We defined the subgroups a priori with respect to severity (mild, severe, or moderate) and cause (psychogenic, mixed, or organic) of ED, treatment (for example, dose [50 mg or 100 mg], dosing [fixed or flexible], type [sildenafil, tadalafil, or vardenafil], and duration [≤12 weeks vs. >12 weeks]), and underlying or concurrent condition (for example, diabetes, cardiovascular disease, depression, or prostatectomy).

When studies did not adequately report summary statistics (such as treatment group mean score and SD), we calculated the needed variables if data for individual patients were reported. If a study reported only an SE of the mean response, we converted it to an SD. Trials were not incorporated into meta-analyses if the needed data (means [SDs]) could not be derived, and crossover trials did not report precrossover data. We included trials with no events for harms in the meta-analyses.

We examined the extent of publication bias through visual inspection of funnel plot asymmetry (32) and linear regression–based tests proposed by Egger and colleagues (33). We performed analyses with R software, version 2.4.0 (www.r-project.org), and STATA software, version 11 (StataCorp, College Station, Texas).

Role of the Funding Source

The Agency for Healthcare Research and Quality provided funding. The funding source suggested the initial research questions and provided copyright release for this manuscript but had no role in the design, conduct, analysis, or reporting of the data or in the decision to submit the manuscript for publication.

Previous Section Next Section

Results

Literature Flow

Our literature search (Appendix Figure 1) identified 10 882 publications, of which 191 unique studies (222 publications) were included in the review. Of these included studies, 32 (23 PDE-5 inhibitor trials, 1 trial of hormonal treatment, 5 prevalence studies, and 3 NAION case reports) were identified through the updated EMBASE and MEDLINE search (May 2007 to April 2009).

Oral PDE-5 Inhibitors

Study and Population Characteristics

In total, 130 RCTs (160 main and secondary publications) evaluating oral PDE-5 inhibitors met eligibility criteria. These included 72 RCTs of sildenafil (34–102), 27 RCTs of vardenafil (23, 103–128), 28 RCTs of tadalafil (129–156), 2 RCTs of mirodenafil (157, 158), and 1 RCT of udenafil (159). In addition, 4 RCTs directly compared PDE-5 inhibitors (160–163).

Most trials (>70.0%) were parallel-group and placebo-controlled and had only short-term follow-up (≤12 weeks). Most were conducted in North America or Europe and were funded by the pharmaceutical industry. Trials enrolled heterosexual adult men with ED of various causes. Exclusion criteria common to most trials were penile or testicular deformity, cardiovascular conditions, use of nitrates, prostate cancer, HIV/AIDS, major hepatic or renal disease, spinal cord injury, and major psychiatric disorder. Nearly half of vardenafil trials and one third of tadalafil trials excluded men with ED that did not respond to previous PDE-5 inhibitor therapy or those who had discontinued PDE-5 inhibitor therapy because of adverse events. The efficacy and harms for 2 novel PDE-5 inhibitors, mirodenafil (157, 158) and udenafil (159), were evaluated only in Asian men. The PDE-5 inhibitors were administered at flexible doses (initial dose could be titrated up or down on the basis of individual participant response, with the following dose ranges: sildenafil, 25 to 100 mg/d; vardenafil, 5 to 20 mg/d; and tadalafil, 5 to 20 mg/d) or fixed (sildenafil, 25 to 100 mg/d; vardenafil, 5 to 40 mg/d; tadalafil, 5 to 20 mg/d; and mirodenafil, 50 to 150 mg/d) doses. About 80.0% of the trials were double-blind, and 30.0% used an appropriate randomization method. The adequacy of allocation concealment and blinding method was not clear for a high proportion of trials (85.0% and 60.0%, respectively) (Appendix A: Figure 1).

Among 4 RCTs that directly compared PDE-5 inhibitors, all had a crossover design and reported patient treatment preference as their primary outcome measure (160–163). It was not clear whether these trials used an appropriate randomization method, 2 trials were not described as double-blind (160, 163), and 2 restricted sildenafil to less than the standard maximum dosing (161, 162). All 4 trials reported reasons for and proportions of withdrawals or dropouts. Three trials were sponsored by the manufacturer of tadalafil (Appendix A: Figure 2) (160–162).

Efficacy

PDE-5 Inhibitor Versus Placebo.

In 116 RCTs, all PDE-5 inhibitors consistently improved erectile functioning more than placebo. The mean scores for the Erectile Function domain and questions 3 and 4 of the IIEF were statistically significantly greater in PDE-5 inhibitor–treated men than in placebo recipients (data not shown).

On the basis of participants' diaries or event logs in 16 trials that enrolled men with a wide spectrum of comorbid conditions, the mean per-patient percentage of successful sexual intercourse attempts for sildenafil-treated patients was 69.0% (range, 52.0% to 85.0%) versus 35.5% (range, 19.0% to 68.0%) for placebo recipients (Table 1). In 4 trials from which data pooling was possible (Appendix B: Figure 1), the WMD in improvement from baseline in the percentage of successful intercourse attempts was 34.3 (95% CI, 25.8 to 42.8) in favor of sildenafil (85, 88, 97, 98).

View this table:

Table 1. Summary of Findings of Effectiveness

In 13 trials of vardenafil that enrolled men with a wide spectrum of comorbid conditions, the percentages of successful sexual intercourse attempts based on SEP question 3 in vardenafil and placebo groups were 68.0% (range, 50.0% to 88.0%) and 35.0% (range, 20.0% to 49.0%), respectively (23, 104, 106, 109, 110, 115, 117–119, 121, 122, 125, 127). The diary or event log–based, weighted mean per-patient percentage of successful sexual intercourse attempts for vardenafil-treated patients was 73.0% versus 38.0% for placebo recipients (107, 112) (Table 1). In 2 trials from which data pooling was possible (Appendix B: Figure 2), the WMD in improvement from baseline in the percentage of successful intercourse attempts was 33.2 (CI, 26.0 to 40.3) in favor of vardenafil (106, 117).

In 15 trials of tadalafil that enrolled men with a wide spectrum of comorbid conditions, the percentages of successful sexual intercourse attempts based on SEP question 3 were 69% (range, 50.0% to 85.0%) for tadalafil versus 33% (range, 23.0% to 52.0%) for placebo (131, 132, 134–136, 138, 139, 142, 144, 148–152, 154). The mean per-patient percentage of successful intercourse attempts was 48.0% for tadalafil-treated patients and 9.0% for placebo recipients, based on diary or event logs (1 trial) (Table 1) (153). In 5 trials from which data pooling was possible (Appendix B: Figure 3), the WMD in improvement from baseline in the percentage of successful intercourse attempts was 35.1 (CI, 26.9 to 43.3) in favor of tadalafil (132, 135, 139, 142, 148).

All 5 agents (sildenafil, vardenafil, tadalafil, mirodenafil, and udenafil) consistently increased the proportion of men with improved erections (range, 73.0% to 88.0%) more than placebo (range, 26.0% to 32.0%) (see overall grade of evidence in Table 1). Appendix Figures 2, 3, and 4 provide the corresponding pooled RR estimates and studies.

View larger version:

Appendix Figure 2. Improvement in erections: sildenafil versus placebo.

CAD = coronary artery disease; CVD = cardiovascular disease; RR = relative risk.

View larger version:

Appendix Figure 3. Improvement in erections: vardenafil versus placebo.

RR = relative risk.

View larger version:

Appendix Figure 4. Improvement in erections: tadalafil versus placebo.

RR = relative risk.

Men with specific medical conditions were significantly more likely to experience improved erections with PDE-5 inhibitors relative to placebo, on the basis of trials limited to men with diabetes (37, 48, 49, 53, 55, 99, 102, 103, 121, 122, 141, 156), depression (35, 47, 61, 86, 116), cardiovascular disease (40, 56, 59, 69, 75, 77, 90, 114, 128), prostate cancer (68, 91, 96, 105, 126, 137, 153, 164), multiple sclerosis (57, 101), colorectal cancer (66), schizophrenia (87), liver failure (58), and renal failure (84, 123) (Appendix Figures 2, 3, and 4).

Improvement in sexual intercourse success and erectile function was greater with higher versus lower doses of sildenafil (50 mg vs. 25 mg, but not 100 mg vs. 50 mg) and vardenafil (20 mg vs. 10 mg vs. 5 mg) but not tadalafil (20 mg vs. 10 mg vs. 5 mg) or mirodenafil (50 mg vs. 100 mg vs. 150 mg) (42, 48, 51, 72, 103, 107, 108, 110, 112, 122, 156–159). In 1 trial, men preferred tadalafil on-demand therapy rather than scheduled dosing 3 times per week, although the mean per-patient proportion of successful intercourse attempts or mean IIEF Erectile Function domain score did not differ (129). In another trial, on-demand and once-daily vardenafil, 10 mg, produced similar treatment effects on the mean Erectile Function domain score in men with mild to moderate ED (124).

Among men assigned to PDE-5 inhibitors, those with severe baseline ED experienced significantly greater absolute improvements in the IIEF scores than did those with mild baseline ED. However, men with severe ED still had worse end-of-treatment Erectile Function domain scores than did men with mild baseline ED (37, 46, 66, 110, 112, 131, 149, 151, 152). There was no obvious treatment effect modification by the duration or cause of ED (42, 44, 51, 70, 74, 112).

PDE-5 Inhibitor Versus PDE-5 Inhibitor.

In 4 trials (160–163), improvements in outcomes between PDE-5 inhibitors were inconsistent. Between-group differences in the mean IIEF Erectile Function domain scores were either statistically nonsignificant or significant but of small magnitude. In 1 trial (160), the use of 10- to 20-mg tadalafil was associated with a small but statistically significantly greater improvement in the mean proportion of successful sexual intercourse attempts compared with 25- to 100-mg sildenafil (76.9% vs. 72.2%; P = 0.003). In the same trial, mean change in Erectile Function domain scores did not differ between groups. In a second trial, mean Erectile Function domain scores were similar between men receiving 100-mg sildenafil and men receiving 20-mg tadalafil (163) (Table 1). More men preferred 20-mg tadalafil (range, 52.2% to 73.0%) to 50-mg sildenafil (range, 27.0% to 33.7%) or 20-mg vardenafil (20.0%).

Extent of Publication Bias

Visual inspection suggested asymmetry in the sildenafil, vardenafil, and tadalafil funnel plots for the rates of improved erection (Appendix C). The linear regression test confirmed statistically significant asymmetry for all 3 funnel plots: sildenafil (P < 0.001), vardenafil (P = 0.003), and tadalafil (P < 0.001).

Harms

PDE-5 Inhibitor Versus Placebo.

A greater proportion of men treated with PDE-5 inhibitors than men who received placebo had at least 1 adverse event (Table 2). The most commonly reported adverse events were headache, flushing, rhinitis, and dyspepsia (Appendix D). Other reported events were visual disturbances, myalgia, nausea, diarrhea, vomiting, dizziness, and chest pain. In general, these events were mild to moderate and were transient. Serious adverse events were reported in fewer than 2.0% of participants, and incidence did not differ between PDE-5 inhibitor recipients and placebo recipients. In 58 placebo-controlled trials reporting data, risk for a serious cardiovascular event (such as fatal or nonfatal myocardial infarction, stroke, or severe angina) seemed higher in men treated with sildenafil (0.5%) than in men receiving placebo (0.1%). The risk for serious cardiovascular events was similar in men treated with vardenafil (0.2%) or tadalafil (0.3%) compared with placebo (range, 0.1% to 0.2%). Because many PDE-5 inhibitor trials were unbalanced with respect to the number of men randomly assigned in each treatment group, the estimated Peto odds ratios may have been prone to bias (29) and therefore are not presented here.

View this table:

Table 2. Summary of Findings on AEs

PDE-5 Inhibitor Versus PDE-5 Inhibitor.

Differences in the incidence of any adverse events among men treated with sildenafil (range, 24.0% to 34.0%), tadalafil (range, 28.0% to 35.0%), and vardenafil (27.0%) were not statistically significant (Table 2) (160–163). Discontinuation due to adverse effects ranged from 0.5% to 3.8% during tadalafil treatment, 0.5% to 3.8% during sildenafil treatment, and 1.0% during vardenafil treatment. The frequency of specific adverse events (headache, flushing, dyspepsia, and nasal congestion) seemed similar among treatments. Tadalafil may have been associated with more frequent myalgia (range, 2.3% to 4.4%) than sildenafil or vardenafil (range, 0% to 0.5%).

NAION or Priapism.

Evidence on the incidence of NAION associated with use of PDE-5 inhibitors was limited to 10 case reports (165–174), 2 case series (175, 176), and 1 retrospective cohort study (177). In the cohort study, NAION and “possible” optic neuropathy were identified by using medical diagnostic codes, with NAION defined as ischemic nonarteritic optic neuropathy in the absence of temporal arteritis and polymyalgia rheumatica and “possible” NAION defined as papillitis, optic neuritis, or both in the absence of temporal arteritis, polymyalgia rheumatica, and previous optic neuropathies. Among more than 4 million male veterans aged 50 years or older, those prescribed PDE-5 inhibitors (11.5% of the cohort) had a risk for NAION similar to that of those who were not prescribed PDE-5 inhibitors (absolute rates of 4.6 and 4.5 cases per 10 000 men per year, respectively; RR, 1.02 [95% CI, 0.92 to 1.12]), but they had an increased risk for “possible” NAION (RR, 1.34 [CI, 1.17 to 1.55]) (177). Trials did not report the incidence of priapism, although the incidence of prolonged erection and priapism has been reported infrequently in PDE-5 inhibitor users during postmarketing surveillance (178).