- Email: [email protected]
Meta-analysis of randomized trials of terazosin in the treatment of benign prostatic hyperplasia
ADULT UROLOGY
META-ANALYSIS OF RANDOMIZED TRIALS OF TERAZOSIN IN THE TREATMENT OF BENIGN PROSTATIC HYPERPLASIA PETER BOYLE, CHRIS ROBERTSON, RICH MANSKI, ROBERT J. PADLEY, CLAUS G. ROEHRBORN
AND
ABSTRACT Objectives. To determine the effectiveness of the long-acting alpha1-adrenergic receptor blocking agent terazosin compared with placebo on lower urinary tract symptoms and peak urinary flow rate in men with clinical benign prostatic hyperplasia. Methods. A formal meta-analysis of all nine randomized trials of terazosin using both an Empirical Bayes and a fully Bayesian approach was conducted. A pooled analysis was conducted on those studies in which patients had a baseline assessment of prostate volume by transrectal ultrasonography. Results. No evidence of heterogeneity was found in the estimated effects of terazosin on the change in peak flow rates in the studies. Terazosin treatment was associated with an increase in the peak flow rate of 1.4 mL/s (95% confidence interval [1.0, 1.7]) compared with placebo. Terazosin resulted in an average reduction of 2.2 points over placebo (95% confidence interval [1.6, 3.0]) regarding the common symptom score (range 0 to 36 points). A mild heterogeneity was found across the studies, with the decrease in symptom score slightly greater with longer treatment duration. No evidence was found that the baseline prostate volume influenced the effect of terazosin. Conclusions. Terazosin was effective and superior to placebo in reducing symptoms and increasing the peak urinary flow rate. The effect of terazosin on the peak urinary flow rate was apparent in studies as short as 8 weeks. Most importantly, the effect of terazosin on symptoms and peak urinary flow rate was independent of the baseline prostate size for the range of prostate volumes reported. UROLOGY 58: 717–722, 2001. © 2001, Elsevier Science Inc.
A
lpha1-adrenergic antagonists are a front line of therapy in the treatment of men with lower urinary tract symptoms (LUTS) suggestive of benign prostatic hyperplasia (BPH).1 A high density of alpha-receptors has been demonstrated in the areas of the bladder neck, prostatic capsule, and prostatic tissue itself. Furthermore, it has been The contribution of Peter Boyle and Chris Robertson to this research was within the framework of support from the Associazione Italiana per la Ricerca sul Cancro (Italian Association for Cancer Research). C. G. Roehrborn was the principal investigator for an unrestricted educational grant from Abbott Laboratories for one of the trials included in this meta-analysis. From the Division of Epidemiology and Biostatistics, European Institute of Oncology, Milan, Italy; Abbott Laboratories, Abbott Park, Waukegan, Illinois; and Department of Urology, University of Texas Southwestern Medical Center, Dallas, Texas Reprint requests: Peter Boyle, Ph.D., Division of Epidemiology and Biostatistics, European Institute of Oncology, via Ripamonti 435, Milan 20141, Italy Submitted: November 13, 2000, accepted (with revisions): June 13, 2001 © 2001, ELSEVIER SCIENCE INC. ALL RIGHTS RESERVED
demonstrated that although in young men the ratio between the stromal and glandular-epithelial tissue in the prostate is about 2:1, it changes with advancing age and the development of BPH and associated symptoms in favor of the stromal component to a ratio of approximately 5:1.2 The blockade of alpha1-receptors inhibits smooth muscle contraction and decreases the muscle tone in the prostate.3,4 Consequently, urethral pressure and resistance are decreased, thereby reducing bladder outlet obstruction.2,4 Thus, the use of alpha1-receptor antagonists such as terazosin would be expected to be effective in men with a high density of prostatic smooth muscle.5 Of the available alpha-blockers, terazosin has been most extensively studied in men with LUTS and clinical BPH.2,3,6 In large, double-blind, placebo-controlled trials and long-term open studies, it has been found to improve significantly the symptoms of BPH and to increase the peak urinary flow rate (Qmax) compared with the baseline values and placebo.2,3,7 Lepor,2 in a semiquantitative 0090-4295/01/$20.00 PII S0090-4295(01)01344-9 717
TABLE I. Summary details and intent-to-treat effect estimates (terazosin vs. placebo) for each study Intent-to-Treat Effect Estimates and Standard Errors Summary Details Study Reference
Study Length (wk)
M87-00510
12
M87-00911 M87-01212 M88-215IT13
24 24 8
M88-215UK14 M89-37715 C91-00516 M91-63917 P92-07618
8 26 24 52 52
Titration Fixed 2, 5, or 10 mg* Response Response Fixed 2, 5, or 10 mg* Fixed 2, 5, or 10 mg* Response Response Response Fixed 10 mg
Increase in Peak Flow
Subjects Country (n)
Drop Out (%)
Mean Mean Mean Prostate Peak Volume Flow Age (cm3) (mL/s) (yr)
USA
312
15.71
61.6
35.7
9.2
9.8
1.31
Europe USA Italy
102 199 136
19.61 23.12 7.35
65.4 62.0 63.3
NA 41.1 NA
8.5 8.7 8.5
9.3 12.3 10.2
86
6.90
63.5
NA
8.2
10.7
160 164 2084 615
16.25 18.29 18.43 22.44
64.1 64.5 65.7 65.1
33.8 NA NA 30.7
UK USA Canada USA USA
8.8 10.2 9.6 10.4
Mean Symptom Score
10.5 10.7/16.7† 20.1† 16.0†
Standard Effect Error
Decrease in Symptoms
Decrease in Rescaled Symptoms
Effect
Standard Error
Effect
Standard Error
0.48
1.40
0.43
1.58
0.56
1.87 1.26 1.42
0.88 0.51 0.54
0.71 1.24 0.67
0.61 0.51 0.55
0.16 1.11 0.88
0.86 0.69 0.71
⫺0.05
0.98
1.70
1.14
2.08
1.36
1.68 0.80 1.44 1.79
0.64 0.60 0.54 0.42
3.30 2.98‡ 3.49 3.07
0.73 1.11‡ 0.26 0.44
3.06 0.53 1.82 (3.33) 0.55 (1.24) 3.89 0.29 3.44 0.49
KEY: NA ⫽ not available; AUA ⫽ American Urological Association. Data in parentheses are AUA score in C91-005. Estimated effects in column 13 refer to the full symptom score used in the respective studies; estimated effects in column 15 refer to the common symptom score obtained by extracting the questions on the common domains and rescaling. * Patients were allocated randomly to placebo or one of the three fixed terazosin dose groups. Three quarters of patients received terazosin. † AUA Symptom Index, otherwise Boyarsky. ‡ Obtained by a linear transformation of the AUA score based on the relationship between the quasi-AUA and AUA Symptom scores in the two studies for which data on the individual AUA Symptom questions were available.
overview, noted that symptom improvement with terazosin appeared to be dose dependent and the (higher) dose of 10 mg ameliorated obstructive, irritative, and total symptom scores to a significantly greater extent than did placebo. The Qmax increased by 1.5 to 4 mL/s in men treated with terazosin, representing a 20% to 40% increase that was generally twofold higher than the change seen with placebo.2,4,6 – 8 Terazosin appears to be effective for LUTS within a few weeks, and it has been speculated that the factors of a low Qmax, moderate to severe symptoms, a predominance of obstructive symptoms, and a high density of prostatic smooth muscle may predict a positive response to terazosin.2 A metaanalysis9 was undertaken to provide an overall estimate of the efficacy of terazosin and to investigate the potential predictive value of the baseline prostatic volume. MATERIAL AND METHODS Nine trials have been conducted in North America and Europe since 1988 (Table I), representing all randomized, double-blind, placebo-controlled clinical trials of terazosin.10 –18 The data and reports for all of these trials were supplied by Abbott Laboratories. A Medline search established that no other trials were eligible. The principal endpoints were the changes from baseline in Qmax and urinary symptom score. The Boyarsky Symptom Index19 was used in seven studies, and the most recent three studies used the American Urological Association (AUA) Symptom Index.20 One study used both. The study duration ranged from 8 to 52 weeks. Four studies used a fixed dose of terazosin, and the other five used a titration to response design. A total of 3948 men were studied (mean age ranging from 61.6 to 65.7 years). Prostate volume was measured by transrectal ultrasonography (TRUS) and estimated by digital rectal examination. The mean volume recorded varied between 30.7 and 41.1 g. The most widely accepted volume formula, namely for a prolate ellipsoid, was used to determine the volume by TRUS. The different symptom scores were taken into account through an appropriate statistical model (see below) and through calculation of a common symptom score, based on questions common to both instruments.9 Consequently, no systematic difference was present among the studies as a result of which symptom score was used. The common symptom score uses all questions except No. 2 from the AUA Symptom Index20 (“urinating less than 2 hours after you had finished urinating”), because this domain is not covered in the Boyarsky Index.19 Obstructive question 3 (“terminal dribbling”) and irritative questions 2 (“daytime frequency”) and 4 (“dysuria”) are contained in the Boyarsky but not the AUA Symptom Index. Questions from the Boyarsky Index are scored from 0 to 3 and those on the AUA Symptom Index from 0 to 5. A nonlinear rescaling of the AUA question responses was performed. This varied slightly from question to question; for example, the responses to question 1 (sensation of not emptying bladder) were adjusted to 0 (0), 1 (1, 2), 2 (3, 4), and 3 (5). The common symptom score, therefore, ranged from 0 to 18 points, which was then doubled to cover a range of 0 to 36 points, similar to the range of the AUA Symptom Index. UROLOGY 58 (5), 2001
STATISTICAL ANALYSIS We performed a meta-analysis21–24 of the standardized changes in symptom score and Qmax using an Empirical Bayes random effects model.25 It was necessary to use standardized scores (the effect divided by its standard error) because the variance in the scores throughout the studies was not constant. Parameter estimates were obtained using MLn.26 The parameter estimates for the full Bayesian model were estimated using BUGS27; these were not reported because they were similar to those obtained using the Empirical Bayes approach. The effect scores and standard errors from each study were derived from the least-squares means in a generalized linear model analysis of the original intent-to-treat data, with treatment group, center, and group by center interaction as the explanatory variables. This used the last available observation as the endpoint for subjects who did not complete the study. In one study,16 the complete data were not available for Qmax, and the effect was calculated from summaries in the technical report. Similar results were obtained in the analysis of the subjects available for evaluation. The pooled analysis model used in the investigation of the effects of baseline prostate volume is a three-level random effects model with subjects nested within the random investigator effect, which, in turn, was nested within the study. This model was fitted using Proc MIXED in Statistical Analysis System, version 6.12, using a components of variance model. The terms included in the fixed part of the model were baseline prostate volume, treatment group, and the interaction between them. The test of the interaction was used to determine whether prostate volume influenced the effect of terazosin. Complete details of the statistical analysis are available in the technical report.28
RESULTS PEAK URINARY FLOW The effects and their standard errors are presented in Table I. The effects of terazosin compared with placebo were similar in all the studies (heterogeneity statistic Q ⫽ 4.1 on 8 degrees of freedom). The overall estimated mean increase in Qmax with terazosin was 1.39 (0.19) mL/s greater than the effect seen with placebo. No evidence was found of any systematic trend in the study effects associated with any of the fixed study characteristics in Table I. SYMPTOM SCORES The Empirical Bayes estimate of the overall mean effect of terazosin compared with placebo corresponded to a 1.9-point reduction on the Boyarsky score (95% confidence interval [CI] [1.6, 2.3]) and a 3.5-point reduction in the AUA Symptom Index (95% CI [2.8, 4.1]). Some heterogeneity was evident in the individual study standardized effects (Q ⫽ 14.6 on 8 degrees of freedom, P ⫽ 0.07). Some evidence was present that the increase in symptom score was greater among studies of longer duration (P ⫽ 0.04). This conclusion was confounded because the studies that were not carried out in North America tended to be shorter. In addition, the two studies lasting the longest used the AUA Symptom Index. 719
FIGURE 1. Effect of terazosin compared with placebo on the common symptom score in the four studies with baseline prostate volume. The size of the plotted point is related to the sample size.
The overall mean effect of terazosin compared with placebo on the common symptom score index was a 2.16 (0.40)-point reduction. There was evidence of heterogeneity (Q ⫽ 22.7, 8 degrees of freedom, P ⫽ 0.004) and strong evidence (P ⫽ 0.0003) that the decrease in the common symptom score with terazosin was greater in the studies that lasted longer [a decrease of 1.33 (0.26) points per 6 months]. This is one possible source of the heterogeneity among the effects over the studies. This effect may have been confounded by the location of the trial. BASELINE PROSTATE VOLUME AS A PREDICTOR OF RESPONSE A significant difference was found in the prostate volume in four studies (P ⫽ 0.001), which was partially confounded by the two different methods of measurement. The overall estimate of the effect of terazosin on the change in the common symptom score was 2.70 (0.41) points, comparable to the results from all nine studies. On the basis of the results of the meta-analysis only, large prostate volumes appear to be associated with a slight reduction in the effect of terazosin (Fig. 1). This had a borderline significant decrease in the magnitude of the effect (P ⫽ 0.05), and the study with the greatest effect and lowest average prostate volume had a large sample size. No effect of the mean prostate volume on the effect of terazosin in the change in the Qmax was found. In view of the importance of the trend in the effect of prostate volume on the change in symptom scores, it was prudent to carry out a pooled analysis of the raw data. The mean and standard deviations of the changes in Qmax and common symptom score by prostate volume are presented 720
in Table II. The differences between the two treatment groups—terazosin versus placebo—were relatively constant over the whole range of prostate volumes. Terazosin was associated with a greater increase in Qmax of 1.55 (0.26) mL/s (95% CI 1.03 to 2.07), and no evidence was found that the prostate volume mediated this effect (P ⫽ 0.82). Only 2% of the variability in the changes in the Qmax was associated with the investigators, and the betweensubject standard deviation of the changes in Qmax was estimated at 4.37 mL/s. For the common symptom score, 1% of the total variation in the changes in the index from baseline to the end of study was associated with the between-study variability and 6% with the betweeninvestigator variability. The between-subject standard deviation was 5.08 points. A greater decrease in the symptom score occurred with terazosin, 2.36 (0.30) points (95% CI [1.76, 2.96]). Again, no evidence was found that this was mediated by the prostate volume (P ⫽ 0.60). The indication from the meta-analysis, namely that terazosin might be more effective in smaller prostates, did not bear out when the more informative individual level analysis was carried out. COMMENT Terazosin is associated with an increase in the Qmax by 1.4 mL/s (95% CI 1.0 to 1.7) more than placebo. This effect did not depend on the length of the trial, and the full effect was attained in as short as 8 weeks. On the 36-point common symptom score, terazosin was associated with a greater reduction in the score by 2.2 points (95% CI 1.6 to 3.0) compared with placebo. Studies of longer duration had a greater response to terazosin, with an estimated reduction of 3.4 points after 1 year, according to the data from the two largest trials.17,18 Although the prostate volume was only available in four trials, and the range of the observed prostate volumes was not as great as in the finasteride meta-analysis,9 the pooled analysis demonstrated that the effect of terazosin is not dependent on the prostate volume. It had been speculated that the TRUS-measured prostate size might be related to the overall tone of the adrenergic nervous system, with men with a higher tone having smaller prostate dimensions (ie, the tone of the capsule confines the tissue in a smaller space) compared with men with a lower adrenergic nervous system tone. This particular theory was postulated to explain the noted increased cardiovascular mortality after transurethral resection of the prostate compared with open prostatectomy; that is, men with smaller prostates (ie, higher adrenergic tone) are usually UROLOGY 58 (5), 2001
TABLE II. Change in peak flow and common symptom score (terazosin vs. placebo) by prostate volume Prostate Volume (g) Peak flow (mL/s) Placebo Mean SD Number Terazosin Mean SD Number Common symptom score (points) Placebo Mean SD Number Terazosin Mean SD Number
6–<20
20–<30
30–<40
40–<50
50–<60
60–<130
1.50 4.76 52
1.50 4.15 176
1.24 3.56 144
0.81 4.08 65
0.78 3.44 42
1.15 3.63 48
3.10 5.03 78
2.59 4.29 230
2.88 4.90 189
2.33 5.65 91
2.59 4.14 44
3.66 5.08 46
⫺2.46 4.72 52
⫺2.74 5.20 177
⫺2.72 5.56 144
⫺2.46 5.36 65
⫺1.38 5.20 42
⫺2.80 5.82 48
⫺5.00 5.12 78
⫺5.20 5.16 230
⫺5.00 5.52 189
⫺3.94 5.40 89
⫺5.22 4.92 44
⫺4.56 4.06 46
Forty-two subjects had missing information on prostate volume and were not included.
scheduled for transurethral resection of the prostate and men with larger prostates (ie, relaxed tone) are usually scheduled for open enucleation of the prostate. After treatment for BPH, men with a higher tone (ie, smaller glands) have a higher propensity toward cardiovascular events. Given this theory, one might have expected that terazosin would have higher efficacy in men with smaller glands, but this was not the case. It is a widely accepted belief that alpha-blocker therapy does not affect the size of the prostate. In fact, in the Veterans Affairs Cooperative Study, the overall change in the TRUS-measured prostate volume during 1 year of treatment with terazosin was a 1.3% increase in size, identical to the change in the placebo arm of this trial.18 Recent reports indicate, however, that alpha-blockers do elicit an apoptotic effect.29 The clinical consequences of this effect on the prostate need to be assessed. Given that TRUS to measure the prostate volume is subject to intra-examiner and inter-examiner variability, more precise modalities of measurement might indicate subtle changes in prostate size and structure, particularly as it may pertain to the differential effects on the transition zone versus the total gland volume. Treatment with terazosin also does not affect serum prostate-specific antigen levels,30 again supporting a lack of effect of alpha-blocker therapy on volume by way of prostate-specific antigen as an indirect measure. It remains unclear whether prostate size might influence long-term compliance with therapy. When choosing alpha-blocker medUROLOGY 58 (5), 2001
ical therapy for men with LUTS and clinical BPH, physicians do not have to be concerned about efficacy variability as a result of prostate volume, and a prostate volume estimate or measurement is not necessary before initiation of therapy. REFERENCES 1. Clifford GM, and Farmer RD: Medical therapy for benign prostatic hyperplasia: a review of the literature. Eur Urol 38: 2–19, 2000. 2. Lepor H: Alpha blockade for the treatment of benign prostatic hyperplasia. Urol Clin North Am 22: 375–386, 1995. 3. Jonler M, Riehmann M, and Bruskewitz RC: Benign prostatic hyperplasia: current pharmacological treatment. Drugs 47: 66 – 81, 1994. 4. Wilde MI, Fitton A, and Sorkin EM: Terazosin: a review of its pharmacodynamic and pharmacokinetic properties, and therapeutic potential in benign prostatic hyperplasia. Drugs Aging 3: 258 –277, 1993. 5. Shapiro E, and Lepor H: The pathophysiology of clinical BPH. Prog Clin Biol Res 386: 253–261, 1994. 6. Eri LM, and Tveter KJ: Alpha-blockade in the treatment of symptomatic benign prostatic hyperplasia. J Urol 154: 923– 934, 1995. 7. Roehrborn CG, Oesterling JE, Auerbach S, et al: The Hytrin Community Assessment Trial: a one-year study of terazosin versus placebo in the treatment of men with symptomatic benign prostatic hyperplasia. Urology 47: 159 –168, 1996. 8. Lepor H: Long-term efficacy and safety of terazosin in patients with benign prostatic hyperplasia. Urology 45: 406 – 413, 1995. 9. Boyle P, Gould AL, and Roehrborn CG: Prostate volume predicts outcome of treatment of benign prostatic hyperplasia with finasteride: meta-analysis of randomized clinical trials. Urology 48: 398 – 405, 1996. 10. Lepor H, Auerbach S, Puras-Baez A, et al: A randomized, placebo-controlled multicenter study of the efficacy and 721
safety of terazosin in the treatment of benign prostatic hyperplasia. J Urol 148: 1467–1474, 1992. 11. Anonymous: Technical Report of Abbott Study M87– 009. Waukegan, Illinois, Abbott Laboratories, 1991. 12. Anonymous: Technical Report of Abbott Study M87– 012. Waukegan, Illinois, Abbott Laboratories, 1991. 13. Di Silverio F: Use of terazosin in the medical treatment of benign prostatic hyperplasia: experience in Italy. Br J Urol 70(suppl 1): 22–26, 1992. 14. Lloyd SN, Buckley JF, Chilton CP, et al: Terazosin in the treatment of benign prostatic hyperplasia: a multicentre, placebo-controlled trial. Br J Urol 70(suppl 1): 17–21, 1992. 15. Brawer MK, Adams G, and Epstein H, for the Terazosin Benign Prostatic Hyperplasia Study Group: Terazosin in the treatment of benign prostatic hyperplasia. Arch Fam Med 2: 929 –935, 1993. 16. Elhilali MM, Ramsey EW, Barkin J, et al: A multicenter, randomized, double-blind, placebo-controlled study to evaluate the safety and efficacy of terazosin in the treatment of benign prostatic hyperplasia. Urology 47: 335–342, 1996. 17. Roehrborn CG, Oesterling JE, Auerbach S, et al, for the HYCAT Investigator Group: The Hytrin Community Assessment Trial Study: a one-year study of terazosin versus placebo in the treatment of men with symptomatic benign prostatic hyperplasia. Urology 47: 159 –168, 1996. 18. Lepor H, Williford WO, Barry MJ, et al, for the Veterans Affairs Cooperative Studies Benign Prostatic Hyperplasia Study Group: The efficacy of terazosin, finasteride, or both in benign prostatic hyperplasia. N Engl J Med 335: 533–539, 1996. 19. Bolognese JA, Kozloff RC, Kunik SC, et al: Validation of a symptoms questionnaire for benign prostatic hyperplasia. Prostate 21: 247–254, 1992. 20. Barry MJ, Fowler FJ, O’Leary MP, et al, and the Mea-
722
surement Committee of the American Urological Association: The American symptom index for benign prostatic hyperplasia. J Urol 148: 1549 –1557, 1992. 21. Whitehead A, and Whitehead J: A general parametric approach to the meta analysis of randomized clinical trials. Stat Med 10: 1665–1677, 1991. 22. Smith TC, Speigelhalter D, and Thomas A: Bayesian approaches to random effects meta analysis: a comparative study. Stat Med 14: 2685–2699, 1996. 23. Boyle P, Robertson C, Lowe F, et al: Meta analysis of clinical trials of Permixon威 in the treatment of symptomatic benign prostatic hyperplasia. Urology 55: 533–539, 2000. 24. Boyle P, Bellanti JA, and Robertson C: Meta-analysis of published clinical trials of ribomunyl in prevention of respiratory infections. Biodrugs 14: 389 – 408, 2000. 25. Bryk AS, and Raudenbush SW: Hierarchical Linear Models. London, Sage Publications, 1992. 26. Woodhouse G, Rasbash J, Goldstein H, et al: MLn Command Reference. London, Multilevel Models Project, Institute of Education, 1995. 27. Speigelhalter D, Thomas A, Best N, et al: BUGS 0.5. Bayesian Inference Using Gibbs Sampling. Cambridge, MRC Biostatistics Unit, 1996. 28. Robertson C, and Manski R: Meta Analysis and Pooled Analysis of Terazosin Trials. Technical Report. Milan, European Institute of Oncology, 2001. 29. Kyprianou N, Litvak JP, Borkowski A, et al: Induction of prostate apoptosis by doxazosin in benign prostatic hyperplasia. J Urol 159: 1810 –1815, 1998. 30. Roehrborn CG, Oesterling JE, Olson PJ, et al: Serial prostate specific antigen measurements in men with clinically benign prostatic hyperplasia during a 12 month placebo controlled study with terazosin. Urology 50: 556 –561, 1997.
UROLOGY 58 (5), 2001
META-ANALYSIS OF RANDOMIZED TRIALS OF TERAZOSIN IN THE TREATMENT OF BENIGN PROSTATIC HYPERPLASIA PETER BOYLE, CHRIS ROBERTSON, RICH MANSKI, ROBERT J. PADLEY, CLAUS G. ROEHRBORN
AND
ABSTRACT Objectives. To determine the effectiveness of the long-acting alpha1-adrenergic receptor blocking agent terazosin compared with placebo on lower urinary tract symptoms and peak urinary flow rate in men with clinical benign prostatic hyperplasia. Methods. A formal meta-analysis of all nine randomized trials of terazosin using both an Empirical Bayes and a fully Bayesian approach was conducted. A pooled analysis was conducted on those studies in which patients had a baseline assessment of prostate volume by transrectal ultrasonography. Results. No evidence of heterogeneity was found in the estimated effects of terazosin on the change in peak flow rates in the studies. Terazosin treatment was associated with an increase in the peak flow rate of 1.4 mL/s (95% confidence interval [1.0, 1.7]) compared with placebo. Terazosin resulted in an average reduction of 2.2 points over placebo (95% confidence interval [1.6, 3.0]) regarding the common symptom score (range 0 to 36 points). A mild heterogeneity was found across the studies, with the decrease in symptom score slightly greater with longer treatment duration. No evidence was found that the baseline prostate volume influenced the effect of terazosin. Conclusions. Terazosin was effective and superior to placebo in reducing symptoms and increasing the peak urinary flow rate. The effect of terazosin on the peak urinary flow rate was apparent in studies as short as 8 weeks. Most importantly, the effect of terazosin on symptoms and peak urinary flow rate was independent of the baseline prostate size for the range of prostate volumes reported. UROLOGY 58: 717–722, 2001. © 2001, Elsevier Science Inc.
A
lpha1-adrenergic antagonists are a front line of therapy in the treatment of men with lower urinary tract symptoms (LUTS) suggestive of benign prostatic hyperplasia (BPH).1 A high density of alpha-receptors has been demonstrated in the areas of the bladder neck, prostatic capsule, and prostatic tissue itself. Furthermore, it has been The contribution of Peter Boyle and Chris Robertson to this research was within the framework of support from the Associazione Italiana per la Ricerca sul Cancro (Italian Association for Cancer Research). C. G. Roehrborn was the principal investigator for an unrestricted educational grant from Abbott Laboratories for one of the trials included in this meta-analysis. From the Division of Epidemiology and Biostatistics, European Institute of Oncology, Milan, Italy; Abbott Laboratories, Abbott Park, Waukegan, Illinois; and Department of Urology, University of Texas Southwestern Medical Center, Dallas, Texas Reprint requests: Peter Boyle, Ph.D., Division of Epidemiology and Biostatistics, European Institute of Oncology, via Ripamonti 435, Milan 20141, Italy Submitted: November 13, 2000, accepted (with revisions): June 13, 2001 © 2001, ELSEVIER SCIENCE INC. ALL RIGHTS RESERVED
demonstrated that although in young men the ratio between the stromal and glandular-epithelial tissue in the prostate is about 2:1, it changes with advancing age and the development of BPH and associated symptoms in favor of the stromal component to a ratio of approximately 5:1.2 The blockade of alpha1-receptors inhibits smooth muscle contraction and decreases the muscle tone in the prostate.3,4 Consequently, urethral pressure and resistance are decreased, thereby reducing bladder outlet obstruction.2,4 Thus, the use of alpha1-receptor antagonists such as terazosin would be expected to be effective in men with a high density of prostatic smooth muscle.5 Of the available alpha-blockers, terazosin has been most extensively studied in men with LUTS and clinical BPH.2,3,6 In large, double-blind, placebo-controlled trials and long-term open studies, it has been found to improve significantly the symptoms of BPH and to increase the peak urinary flow rate (Qmax) compared with the baseline values and placebo.2,3,7 Lepor,2 in a semiquantitative 0090-4295/01/$20.00 PII S0090-4295(01)01344-9 717
TABLE I. Summary details and intent-to-treat effect estimates (terazosin vs. placebo) for each study Intent-to-Treat Effect Estimates and Standard Errors Summary Details Study Reference
Study Length (wk)
M87-00510
12
M87-00911 M87-01212 M88-215IT13
24 24 8
M88-215UK14 M89-37715 C91-00516 M91-63917 P92-07618
8 26 24 52 52
Titration Fixed 2, 5, or 10 mg* Response Response Fixed 2, 5, or 10 mg* Fixed 2, 5, or 10 mg* Response Response Response Fixed 10 mg
Increase in Peak Flow
Subjects Country (n)
Drop Out (%)
Mean Mean Mean Prostate Peak Volume Flow Age (cm3) (mL/s) (yr)
USA
312
15.71
61.6
35.7
9.2
9.8
1.31
Europe USA Italy
102 199 136
19.61 23.12 7.35
65.4 62.0 63.3
NA 41.1 NA
8.5 8.7 8.5
9.3 12.3 10.2
86
6.90
63.5
NA
8.2
10.7
160 164 2084 615
16.25 18.29 18.43 22.44
64.1 64.5 65.7 65.1
33.8 NA NA 30.7
UK USA Canada USA USA
8.8 10.2 9.6 10.4
Mean Symptom Score
10.5 10.7/16.7† 20.1† 16.0†
Standard Effect Error
Decrease in Symptoms
Decrease in Rescaled Symptoms
Effect
Standard Error
Effect
Standard Error
0.48
1.40
0.43
1.58
0.56
1.87 1.26 1.42
0.88 0.51 0.54
0.71 1.24 0.67
0.61 0.51 0.55
0.16 1.11 0.88
0.86 0.69 0.71
⫺0.05
0.98
1.70
1.14
2.08
1.36
1.68 0.80 1.44 1.79
0.64 0.60 0.54 0.42
3.30 2.98‡ 3.49 3.07
0.73 1.11‡ 0.26 0.44
3.06 0.53 1.82 (3.33) 0.55 (1.24) 3.89 0.29 3.44 0.49
KEY: NA ⫽ not available; AUA ⫽ American Urological Association. Data in parentheses are AUA score in C91-005. Estimated effects in column 13 refer to the full symptom score used in the respective studies; estimated effects in column 15 refer to the common symptom score obtained by extracting the questions on the common domains and rescaling. * Patients were allocated randomly to placebo or one of the three fixed terazosin dose groups. Three quarters of patients received terazosin. † AUA Symptom Index, otherwise Boyarsky. ‡ Obtained by a linear transformation of the AUA score based on the relationship between the quasi-AUA and AUA Symptom scores in the two studies for which data on the individual AUA Symptom questions were available.
overview, noted that symptom improvement with terazosin appeared to be dose dependent and the (higher) dose of 10 mg ameliorated obstructive, irritative, and total symptom scores to a significantly greater extent than did placebo. The Qmax increased by 1.5 to 4 mL/s in men treated with terazosin, representing a 20% to 40% increase that was generally twofold higher than the change seen with placebo.2,4,6 – 8 Terazosin appears to be effective for LUTS within a few weeks, and it has been speculated that the factors of a low Qmax, moderate to severe symptoms, a predominance of obstructive symptoms, and a high density of prostatic smooth muscle may predict a positive response to terazosin.2 A metaanalysis9 was undertaken to provide an overall estimate of the efficacy of terazosin and to investigate the potential predictive value of the baseline prostatic volume. MATERIAL AND METHODS Nine trials have been conducted in North America and Europe since 1988 (Table I), representing all randomized, double-blind, placebo-controlled clinical trials of terazosin.10 –18 The data and reports for all of these trials were supplied by Abbott Laboratories. A Medline search established that no other trials were eligible. The principal endpoints were the changes from baseline in Qmax and urinary symptom score. The Boyarsky Symptom Index19 was used in seven studies, and the most recent three studies used the American Urological Association (AUA) Symptom Index.20 One study used both. The study duration ranged from 8 to 52 weeks. Four studies used a fixed dose of terazosin, and the other five used a titration to response design. A total of 3948 men were studied (mean age ranging from 61.6 to 65.7 years). Prostate volume was measured by transrectal ultrasonography (TRUS) and estimated by digital rectal examination. The mean volume recorded varied between 30.7 and 41.1 g. The most widely accepted volume formula, namely for a prolate ellipsoid, was used to determine the volume by TRUS. The different symptom scores were taken into account through an appropriate statistical model (see below) and through calculation of a common symptom score, based on questions common to both instruments.9 Consequently, no systematic difference was present among the studies as a result of which symptom score was used. The common symptom score uses all questions except No. 2 from the AUA Symptom Index20 (“urinating less than 2 hours after you had finished urinating”), because this domain is not covered in the Boyarsky Index.19 Obstructive question 3 (“terminal dribbling”) and irritative questions 2 (“daytime frequency”) and 4 (“dysuria”) are contained in the Boyarsky but not the AUA Symptom Index. Questions from the Boyarsky Index are scored from 0 to 3 and those on the AUA Symptom Index from 0 to 5. A nonlinear rescaling of the AUA question responses was performed. This varied slightly from question to question; for example, the responses to question 1 (sensation of not emptying bladder) were adjusted to 0 (0), 1 (1, 2), 2 (3, 4), and 3 (5). The common symptom score, therefore, ranged from 0 to 18 points, which was then doubled to cover a range of 0 to 36 points, similar to the range of the AUA Symptom Index. UROLOGY 58 (5), 2001
STATISTICAL ANALYSIS We performed a meta-analysis21–24 of the standardized changes in symptom score and Qmax using an Empirical Bayes random effects model.25 It was necessary to use standardized scores (the effect divided by its standard error) because the variance in the scores throughout the studies was not constant. Parameter estimates were obtained using MLn.26 The parameter estimates for the full Bayesian model were estimated using BUGS27; these were not reported because they were similar to those obtained using the Empirical Bayes approach. The effect scores and standard errors from each study were derived from the least-squares means in a generalized linear model analysis of the original intent-to-treat data, with treatment group, center, and group by center interaction as the explanatory variables. This used the last available observation as the endpoint for subjects who did not complete the study. In one study,16 the complete data were not available for Qmax, and the effect was calculated from summaries in the technical report. Similar results were obtained in the analysis of the subjects available for evaluation. The pooled analysis model used in the investigation of the effects of baseline prostate volume is a three-level random effects model with subjects nested within the random investigator effect, which, in turn, was nested within the study. This model was fitted using Proc MIXED in Statistical Analysis System, version 6.12, using a components of variance model. The terms included in the fixed part of the model were baseline prostate volume, treatment group, and the interaction between them. The test of the interaction was used to determine whether prostate volume influenced the effect of terazosin. Complete details of the statistical analysis are available in the technical report.28
RESULTS PEAK URINARY FLOW The effects and their standard errors are presented in Table I. The effects of terazosin compared with placebo were similar in all the studies (heterogeneity statistic Q ⫽ 4.1 on 8 degrees of freedom). The overall estimated mean increase in Qmax with terazosin was 1.39 (0.19) mL/s greater than the effect seen with placebo. No evidence was found of any systematic trend in the study effects associated with any of the fixed study characteristics in Table I. SYMPTOM SCORES The Empirical Bayes estimate of the overall mean effect of terazosin compared with placebo corresponded to a 1.9-point reduction on the Boyarsky score (95% confidence interval [CI] [1.6, 2.3]) and a 3.5-point reduction in the AUA Symptom Index (95% CI [2.8, 4.1]). Some heterogeneity was evident in the individual study standardized effects (Q ⫽ 14.6 on 8 degrees of freedom, P ⫽ 0.07). Some evidence was present that the increase in symptom score was greater among studies of longer duration (P ⫽ 0.04). This conclusion was confounded because the studies that were not carried out in North America tended to be shorter. In addition, the two studies lasting the longest used the AUA Symptom Index. 719
FIGURE 1. Effect of terazosin compared with placebo on the common symptom score in the four studies with baseline prostate volume. The size of the plotted point is related to the sample size.
The overall mean effect of terazosin compared with placebo on the common symptom score index was a 2.16 (0.40)-point reduction. There was evidence of heterogeneity (Q ⫽ 22.7, 8 degrees of freedom, P ⫽ 0.004) and strong evidence (P ⫽ 0.0003) that the decrease in the common symptom score with terazosin was greater in the studies that lasted longer [a decrease of 1.33 (0.26) points per 6 months]. This is one possible source of the heterogeneity among the effects over the studies. This effect may have been confounded by the location of the trial. BASELINE PROSTATE VOLUME AS A PREDICTOR OF RESPONSE A significant difference was found in the prostate volume in four studies (P ⫽ 0.001), which was partially confounded by the two different methods of measurement. The overall estimate of the effect of terazosin on the change in the common symptom score was 2.70 (0.41) points, comparable to the results from all nine studies. On the basis of the results of the meta-analysis only, large prostate volumes appear to be associated with a slight reduction in the effect of terazosin (Fig. 1). This had a borderline significant decrease in the magnitude of the effect (P ⫽ 0.05), and the study with the greatest effect and lowest average prostate volume had a large sample size. No effect of the mean prostate volume on the effect of terazosin in the change in the Qmax was found. In view of the importance of the trend in the effect of prostate volume on the change in symptom scores, it was prudent to carry out a pooled analysis of the raw data. The mean and standard deviations of the changes in Qmax and common symptom score by prostate volume are presented 720
in Table II. The differences between the two treatment groups—terazosin versus placebo—were relatively constant over the whole range of prostate volumes. Terazosin was associated with a greater increase in Qmax of 1.55 (0.26) mL/s (95% CI 1.03 to 2.07), and no evidence was found that the prostate volume mediated this effect (P ⫽ 0.82). Only 2% of the variability in the changes in the Qmax was associated with the investigators, and the betweensubject standard deviation of the changes in Qmax was estimated at 4.37 mL/s. For the common symptom score, 1% of the total variation in the changes in the index from baseline to the end of study was associated with the between-study variability and 6% with the betweeninvestigator variability. The between-subject standard deviation was 5.08 points. A greater decrease in the symptom score occurred with terazosin, 2.36 (0.30) points (95% CI [1.76, 2.96]). Again, no evidence was found that this was mediated by the prostate volume (P ⫽ 0.60). The indication from the meta-analysis, namely that terazosin might be more effective in smaller prostates, did not bear out when the more informative individual level analysis was carried out. COMMENT Terazosin is associated with an increase in the Qmax by 1.4 mL/s (95% CI 1.0 to 1.7) more than placebo. This effect did not depend on the length of the trial, and the full effect was attained in as short as 8 weeks. On the 36-point common symptom score, terazosin was associated with a greater reduction in the score by 2.2 points (95% CI 1.6 to 3.0) compared with placebo. Studies of longer duration had a greater response to terazosin, with an estimated reduction of 3.4 points after 1 year, according to the data from the two largest trials.17,18 Although the prostate volume was only available in four trials, and the range of the observed prostate volumes was not as great as in the finasteride meta-analysis,9 the pooled analysis demonstrated that the effect of terazosin is not dependent on the prostate volume. It had been speculated that the TRUS-measured prostate size might be related to the overall tone of the adrenergic nervous system, with men with a higher tone having smaller prostate dimensions (ie, the tone of the capsule confines the tissue in a smaller space) compared with men with a lower adrenergic nervous system tone. This particular theory was postulated to explain the noted increased cardiovascular mortality after transurethral resection of the prostate compared with open prostatectomy; that is, men with smaller prostates (ie, higher adrenergic tone) are usually UROLOGY 58 (5), 2001
TABLE II. Change in peak flow and common symptom score (terazosin vs. placebo) by prostate volume Prostate Volume (g) Peak flow (mL/s) Placebo Mean SD Number Terazosin Mean SD Number Common symptom score (points) Placebo Mean SD Number Terazosin Mean SD Number
6–<20
20–<30
30–<40
40–<50
50–<60
60–<130
1.50 4.76 52
1.50 4.15 176
1.24 3.56 144
0.81 4.08 65
0.78 3.44 42
1.15 3.63 48
3.10 5.03 78
2.59 4.29 230
2.88 4.90 189
2.33 5.65 91
2.59 4.14 44
3.66 5.08 46
⫺2.46 4.72 52
⫺2.74 5.20 177
⫺2.72 5.56 144
⫺2.46 5.36 65
⫺1.38 5.20 42
⫺2.80 5.82 48
⫺5.00 5.12 78
⫺5.20 5.16 230
⫺5.00 5.52 189
⫺3.94 5.40 89
⫺5.22 4.92 44
⫺4.56 4.06 46
Forty-two subjects had missing information on prostate volume and were not included.
scheduled for transurethral resection of the prostate and men with larger prostates (ie, relaxed tone) are usually scheduled for open enucleation of the prostate. After treatment for BPH, men with a higher tone (ie, smaller glands) have a higher propensity toward cardiovascular events. Given this theory, one might have expected that terazosin would have higher efficacy in men with smaller glands, but this was not the case. It is a widely accepted belief that alpha-blocker therapy does not affect the size of the prostate. In fact, in the Veterans Affairs Cooperative Study, the overall change in the TRUS-measured prostate volume during 1 year of treatment with terazosin was a 1.3% increase in size, identical to the change in the placebo arm of this trial.18 Recent reports indicate, however, that alpha-blockers do elicit an apoptotic effect.29 The clinical consequences of this effect on the prostate need to be assessed. Given that TRUS to measure the prostate volume is subject to intra-examiner and inter-examiner variability, more precise modalities of measurement might indicate subtle changes in prostate size and structure, particularly as it may pertain to the differential effects on the transition zone versus the total gland volume. Treatment with terazosin also does not affect serum prostate-specific antigen levels,30 again supporting a lack of effect of alpha-blocker therapy on volume by way of prostate-specific antigen as an indirect measure. It remains unclear whether prostate size might influence long-term compliance with therapy. When choosing alpha-blocker medUROLOGY 58 (5), 2001
ical therapy for men with LUTS and clinical BPH, physicians do not have to be concerned about efficacy variability as a result of prostate volume, and a prostate volume estimate or measurement is not necessary before initiation of therapy. REFERENCES 1. Clifford GM, and Farmer RD: Medical therapy for benign prostatic hyperplasia: a review of the literature. Eur Urol 38: 2–19, 2000. 2. Lepor H: Alpha blockade for the treatment of benign prostatic hyperplasia. Urol Clin North Am 22: 375–386, 1995. 3. Jonler M, Riehmann M, and Bruskewitz RC: Benign prostatic hyperplasia: current pharmacological treatment. Drugs 47: 66 – 81, 1994. 4. Wilde MI, Fitton A, and Sorkin EM: Terazosin: a review of its pharmacodynamic and pharmacokinetic properties, and therapeutic potential in benign prostatic hyperplasia. Drugs Aging 3: 258 –277, 1993. 5. Shapiro E, and Lepor H: The pathophysiology of clinical BPH. Prog Clin Biol Res 386: 253–261, 1994. 6. Eri LM, and Tveter KJ: Alpha-blockade in the treatment of symptomatic benign prostatic hyperplasia. J Urol 154: 923– 934, 1995. 7. Roehrborn CG, Oesterling JE, Auerbach S, et al: The Hytrin Community Assessment Trial: a one-year study of terazosin versus placebo in the treatment of men with symptomatic benign prostatic hyperplasia. Urology 47: 159 –168, 1996. 8. Lepor H: Long-term efficacy and safety of terazosin in patients with benign prostatic hyperplasia. Urology 45: 406 – 413, 1995. 9. Boyle P, Gould AL, and Roehrborn CG: Prostate volume predicts outcome of treatment of benign prostatic hyperplasia with finasteride: meta-analysis of randomized clinical trials. Urology 48: 398 – 405, 1996. 10. Lepor H, Auerbach S, Puras-Baez A, et al: A randomized, placebo-controlled multicenter study of the efficacy and 721
safety of terazosin in the treatment of benign prostatic hyperplasia. J Urol 148: 1467–1474, 1992. 11. Anonymous: Technical Report of Abbott Study M87– 009. Waukegan, Illinois, Abbott Laboratories, 1991. 12. Anonymous: Technical Report of Abbott Study M87– 012. Waukegan, Illinois, Abbott Laboratories, 1991. 13. Di Silverio F: Use of terazosin in the medical treatment of benign prostatic hyperplasia: experience in Italy. Br J Urol 70(suppl 1): 22–26, 1992. 14. Lloyd SN, Buckley JF, Chilton CP, et al: Terazosin in the treatment of benign prostatic hyperplasia: a multicentre, placebo-controlled trial. Br J Urol 70(suppl 1): 17–21, 1992. 15. Brawer MK, Adams G, and Epstein H, for the Terazosin Benign Prostatic Hyperplasia Study Group: Terazosin in the treatment of benign prostatic hyperplasia. Arch Fam Med 2: 929 –935, 1993. 16. Elhilali MM, Ramsey EW, Barkin J, et al: A multicenter, randomized, double-blind, placebo-controlled study to evaluate the safety and efficacy of terazosin in the treatment of benign prostatic hyperplasia. Urology 47: 335–342, 1996. 17. Roehrborn CG, Oesterling JE, Auerbach S, et al, for the HYCAT Investigator Group: The Hytrin Community Assessment Trial Study: a one-year study of terazosin versus placebo in the treatment of men with symptomatic benign prostatic hyperplasia. Urology 47: 159 –168, 1996. 18. Lepor H, Williford WO, Barry MJ, et al, for the Veterans Affairs Cooperative Studies Benign Prostatic Hyperplasia Study Group: The efficacy of terazosin, finasteride, or both in benign prostatic hyperplasia. N Engl J Med 335: 533–539, 1996. 19. Bolognese JA, Kozloff RC, Kunik SC, et al: Validation of a symptoms questionnaire for benign prostatic hyperplasia. Prostate 21: 247–254, 1992. 20. Barry MJ, Fowler FJ, O’Leary MP, et al, and the Mea-
722
surement Committee of the American Urological Association: The American symptom index for benign prostatic hyperplasia. J Urol 148: 1549 –1557, 1992. 21. Whitehead A, and Whitehead J: A general parametric approach to the meta analysis of randomized clinical trials. Stat Med 10: 1665–1677, 1991. 22. Smith TC, Speigelhalter D, and Thomas A: Bayesian approaches to random effects meta analysis: a comparative study. Stat Med 14: 2685–2699, 1996. 23. Boyle P, Robertson C, Lowe F, et al: Meta analysis of clinical trials of Permixon威 in the treatment of symptomatic benign prostatic hyperplasia. Urology 55: 533–539, 2000. 24. Boyle P, Bellanti JA, and Robertson C: Meta-analysis of published clinical trials of ribomunyl in prevention of respiratory infections. Biodrugs 14: 389 – 408, 2000. 25. Bryk AS, and Raudenbush SW: Hierarchical Linear Models. London, Sage Publications, 1992. 26. Woodhouse G, Rasbash J, Goldstein H, et al: MLn Command Reference. London, Multilevel Models Project, Institute of Education, 1995. 27. Speigelhalter D, Thomas A, Best N, et al: BUGS 0.5. Bayesian Inference Using Gibbs Sampling. Cambridge, MRC Biostatistics Unit, 1996. 28. Robertson C, and Manski R: Meta Analysis and Pooled Analysis of Terazosin Trials. Technical Report. Milan, European Institute of Oncology, 2001. 29. Kyprianou N, Litvak JP, Borkowski A, et al: Induction of prostate apoptosis by doxazosin in benign prostatic hyperplasia. J Urol 159: 1810 –1815, 1998. 30. Roehrborn CG, Oesterling JE, Olson PJ, et al: Serial prostate specific antigen measurements in men with clinically benign prostatic hyperplasia during a 12 month placebo controlled study with terazosin. Urology 50: 556 –561, 1997.
UROLOGY 58 (5), 2001