Finasteride in the treatment of clinical benign prostatic hyperplasia: A systematic review of randomised trials

PubMed (to April 2001) and the Cochrane Library (Issue 2, 2001) were searched to identify full journal publications of randomised, double blind, placebo and active controlled trials of finasteride in the treatment of benign prostatic hyperplasia. Free text search terms used included 'finasteride', 'proscar', 'clinical trial', and 'benign prostatic hyperplasia'. Systematic reviews of finasteride [5, 11] were examined, as was a list of systematic reviews in benign prostatic hyperplasia (additional file 1) for possible references and reference lists of all obtained articles were checked to identify additional trials. Abstracts were not sought. Merck, Sharp and Dohme Ltd, UK, were asked for references of any published randomised trials for finasteride in the context of benign prostatic hyperplasia. Unpublished studies were not sought.

It was anticipated that patient information from major trials may have been published more than once, in part or in full, as information became available from longer use of finasteride. For each trial, the study that provided the fullest amount of information was included in the systematic review and any duplicated information was excluded. Duplicate studies were checked to ensure that relevant information for a particular outcome described in an excluded study was not missing from the included trial. A number of trials had an open-label extension in which men received finasteride only; no open-label information was analysed because the pre-hoc decision was to analyse only information from trials that were both randomised and double blind.

Each report which could possibly be described as a randomised controlled trial was read independently by both authors and scored using a commonly-used, three item, 1–5 score, quality scale [13]. Disagreements were discussed and consensus achieved. The maximum score of an included study was 5 and the minimum score was 2.

Outcomes were abstracted after discussion with a panel of two urologists and three general practitioners with an interest in urology to discuss outcomes likely to be of professional or patient interest. Information extracted from the double blind trials included: (i) number of men on finasteride and placebo, (ii) symptom score (total, obstructive, bother), (iii) prostate volume, (iv) urinary flow rate (maximum, mean), (v) discontinuation (total discontinuations, discontinuations because of lack of efficacy, discontinuations because of adverse effects), (vi) information on adverse effects including episodes of acute urinary retention and prostate surgery, (vii) prostate specific antigen (PSA), (viii) residual volume and (ix) total voided volume. Absolute values and/or mean or median values (with dispersion) were extracted. When provided, this information was extracted for the following time-points regardless of whether it was an intermediate or final assessment: baseline and 3, 6, 12, 18, 24, 36 and 48 months of treatment. The type of symptom score used was noted.

Not all men in a trial may have had each outcome assessed (e.g. prostate volume), or men may have discontinued. For efficacy analyses, therefore, the number of men for which information was available at a particular point in time was used if this differed from the number randomised; because results were rarely dichotomous, an intention-to-treat analysis was impossible for any beneficial outcome. For analysis of adverse effects and discontinuations, the number of men randomised to treatment was used to provide an intention-to-treat analysis.

When possible, patient information from different studies was pooled. The objective was to enter any continuous data in Review Manager (RevMan version 4.01; Update Software, Oxford), to calculate weighted mean difference from baseline for finasteride and for placebo, to generate standard deviations or 95% confidence intervals, and to determine statistical significance of differences between treatments at various time points. Weighted mean values (by group size) for continuous outcomes were calculated using Excel:mac 2001 on a Macintosh G4.

When dichotomous information was available, relative risk estimates, with 95% confidence intervals, were calculated using a fixed effects model [14]. A statistically significant difference between active treatment and placebo was assumed when the 95% confidence intervals (CI) of the relative risk did not include unity. Numbers-needed-to-treat, with 95% confidence intervals, were calculated [15]. The confidence interval of the NNT includes no benefit of one treatment over the other when the upper limit includes infinity. NNT is the reciprocal of the absolute risk reduction or increase; for instance, if 75 out of 100 men benefit with treatment and only 25 out of 100 benefit with placebo, the absolute risk increase is 0.75–0.25 = 0.5, and the NNT is 1/0.5 = 2.

Neither heterogeneity tests nor funnel plots were used since they lack the power to reliably detect statistical heterogeneity or publication bias [16‐18]. Instead, pre-planned sensitivity analyses were conducted to detect possible variations in effect of study treatments in men with differing aetiology or baseline severity of their condition.

The four sensitivity analyses were:

Analyses based on fewer than 500 men are not presented in this report, because the results would not be robust [20].

One hundred and one reports were identified as potential randomised trials of finasteride in the treatment of benign prostatic hyperplasia. Several of these were duplicate publications, with or without reference to previous publications. Eighty studies were excluded and reasons are given in additional file 2. Twenty-two double blind randomised trials from 21 reports were included in the systematic review [19, 21‐40]. Nineteen trials were placebo controlled, and three compared finasteride with another active treatment. Full details of the included studies, with description of men included, study duration, outcomes, and results are shown in additional file 3, and all the baseline characteristics of study groups in additional file 4.

Finasteride 5 mg daily was assessed in all included trials, though some also included finasteride 1 mg or 10 mg. One trial pooled information from 17 men with finasteride 1 mg and 15 men with finasteride 5 mg [37]; this information was included in the analyses described below.

In placebo controlled trials, 8820 men received finasteride 5 mg given once daily and 5909 men received placebo. Information was available for men followed up for three months in two trials (74 men), six months in nine (366), 12 months in six (6364), 18 months in one (55), 24 months in four (4286), and for 48 months in one (3082). Three trials used an active control; 903 men received finasteride 5 mg once daily, 553 permixon (serenoa repens), and 358 alfluzosin 10 mg daily. Men were followed up for six months in two studies (1870 men) and for 12 months in one trial (489 men). Trials of active comparisons were not followed up further because the comparators were all different and numbers of men were small.

Men included in the trials had a clinical diagnosis of benign prostatic hyperplasia, mostly based on symptoms and urine flow rates. For symptoms, for instance, the American Urological Association symptom scoring scale uses seven questions which can be scored from 0 (no problem) to 5 (severe problem); the scale can be from 0 to 35, split into mild (0–7 points), moderate (8–19 points) or severe (20–35 points) disease. At baseline, symptoms were moderate to severe in 17 trials, and in two studies men with mild symptoms were included. General exclusions in the trials were men with the suggestion of prostate cancer, urinary tract infection, previous prostate surgery, haematuria, or those who required catheterisation for acute urinary retention.

A summary of the included trials, their size at randomisation, quality score, duration, intention-to-treat (by number randomised) or per protocol analysis, and the statistical significance at longest double-blind estimate is given in Table 1. Size varied from 36 to 3040 randomised men; nine trials randomised at least 500 men and four at least 2000. The largest trials were of at least 12 months duration. Most of the larger trials demonstrated a significant superiority of finasteride 5 mg over placebo at the p < 0.01 level for symptom score, maximum urinary flow rate and prostate volume; the notable exception was the VA Cooperative study for symptom scores and maximum urinary flow rate because the inclusion criteria for this study included men with small prostates [31].

Trial reports usually provided dispersion measures (standard deviation, interquartile ranges) at baseline for measures like symptom score, urine flow rate and prostate volume. In almost no case did they provide standard deviations or 95% confidence intervals for intermediate and/or final assessment for these outcome data. Consequently statistical significance of differences between finasteride and placebo could not be assessed for continuous measures.

Three different scoring systems were used. In placebo-controlled trials, the Boyarsky scale (0–54) was used in two studies, modified Boyarsky scale (0–36) in nine, and the American Urological Association (AUA; 0–35, similar to the current International Prostate Symptom Score, IPSS) in eight; higher scores show worse symptoms in all three scales. Symptom severity was recorded at baseline and at various stages throughout the study, though not all studies reported this information for all time points, including baseline. Figure 1 shows the weighted mean total symptom scores at different time points with the American Urological Association scale which was used for most men, and the results for all scales are in additional file 5. Total symptom scores were similar at baseline with finasteride (17) and placebo (16). They then fell (improved), and by 12 months there was a greater reduction in symptom score with finasteride (by 3.7 points) than with placebo (by 2.3 points), and maintained for 24 to 48 months. Scores continued to decrease with finasteride up to 48 months, whereas they started to increase with placebo after 18 months.

Figure 2 shows the results for maximum urinary flow rate. At baseline the weighted mean maximum urinary flow rate was 11.2 mL/s with finasteride (3960 men) and 10.5 mL/s with placebo (3893 men). This increased (improved) with time on treatment and by 24 months weighted mean urinary flow rates were 12.5 mL/s with finasteride (2592 men) and 11.3 mL/s with placebo (2523 men).

Figure 3 shows the results for prostate volume, which declined by 25% over 24 months with finasteride compared with a 4% decline with placebo. At baseline the weighted mean prostate volume was 43.7 cm³ with finasteride (2847 men) and 44.8 cm³ with placebo (2853 men). This decreased (improved) with duration of finasteride treatment, but generally increased (worsened) with placebo. By 24 months prostate volume was 32.7 cm³ with finasteride (1705 men) and 43.0 cm³ with placebo (1724 men). Prostate volumes of about 30 cm³ or less are generally assumed to be normal in older men and not associated with prostatic hyperplasia [41].

Discontinuation was well reported in the trials.

Significantly fewer men discontinued with finasteride than with placebo at 12 months or 48 months (Table 3A). After 12 months, all cause discontinuation rates were 13% (553/4098 men) with finasteride and 17% (299/1764) with placebo; number-needed-to-treat to prevent one discontinuation was 29 (18 to 71). After 48 months, discontinuation rates were 34% (524/1524 men) with finasteride and 42% (633/1516) with placebo; NNH to prevent one discontinuation was 13 (9.0 to 27).

There was no significant difference between groups at any time point for discontinuation because of lack of efficacy (Table 3B), with 24 month discontinuations because of lack of efficacy at 4% with finasteride and 5% with placebo. There was no significant difference between groups at any time point for discontinuation because of adverse effects (Table 3C), with 24 month discontinuations because of adverse effects at 8% with finasteride and 14% with placebo.

The most commonly reported adverse effects were impotence, decreased libido and ejaculation disorder. Definitions of these adverse effects were generally not provided in the trials. Since cumulative adverse effect information was not available after year one in some studies, analyses were conducted for different time points up to one year for most adverse effects. Numbers-needed-to-harm are shown in Table 4 for analyses for which there were adequate data. Serious adverse effects occurred at similar frequencies with finasteride (12%; 437/3557 men) as with placebo (13%; 150/1175 men).

Significantly more men reported any sexual dysfunction, decreased libido, impotence, or ejaculation disorder with finasteride than with placebo at 12 months of treatment. Incidence rates ranged between 2–14% with finasteride and 0.6–7% with placebo, and NNHs for particular adverse events ranged between 14 and 55 (Table 4). This means, for instance, that for every 24 men treated with 5 mg finasteride for 12 months impotence would occur in one in whom it would not have occurred with placebo.

Significantly fewer men suffered acute urinary retention or surgery related to benign prostatic hyperplasia after 24 months (Table 4), though much of this information was derived from the PLESS trial. Acute urinary retention and surgery related to benign prostatic hypertrophy occurred in less than 1% of men over the first 12 months of treatment. By 24 months their occurrence was significantly lower with finasteride than with placebo. The NNTs for avoiding acute urinary retention were 49 (31 to 112) over 24 months and 26 (19 to 44) over 48 months. The NNTs for avoiding prostate related surgery were 31 (21 to 61) over 24 months and 18 (14 to 27) over 48 months. This means, for instance, that for every 31 men treated with 5 mg finasteride for 24 months prostate surgery would be avoided in one in whom it would have occurred with placebo.

It was not possible to calculate results for a combined end point of avoiding acute urinary retention or surgery.

There was no statistically significant difference in the incidence of prostate cancer with finasteride compared with placebo. Rates were under 0.5% at 12 months, and in one four year study [19] rates were 1–2% at 24 months and 0.5% at 48 months.