Blackwell Science, LtdOxford, UKBJUBJU International1464-410XBJU InternationalApril 2005 956 Review Article PHARMACOLOGICAL CLASS EFFECT IN MANAGING PROSTATIC DISEASES EVANS et al. An evidence-based approach to understanding the pharmacological class effect in the management of prostatic diseases CHRISTOPHER P. EVANS, NEIL FLESHNER*, JOHN M. FITZPATRICK† and ALEXANDER R. ZLOTTA‡ University of California, Sacramento, CA, USA, *Princess Margaret Hospital, Toronto, Canada; †Mater Hospital and Conway Institute, University College, Dublin, Ireland and ‡University Clinics of Brussels, Erasme Hospital, Brussels, Belgium Accepted for publication 5 December 2004 KEYWORDS reviews or meta-analyses of well-designed Questions have been raised as to whether randomized controlled trials (RCTs), followed study sponsorship may also introduce bias evidence-based medicine, class effect, benign by individual RCTs and well-designed non- and there are concerns that the process of prostatic hyperplasia, prostate cancer randomized studies are considered the most publication itself can lead to bias in favour of robust, reliable and therefore valuable [3]. Less positive results. robust evidence from case reports, clinical INTRODUCTION examples, or consensus meetings may also be Although many clinical trials are conducted considered. ‘blind’ to minimize observer bias, this is not Ever more alternative products are available always a realistic option. For example, in for each drug type commonly used to treat Systematic reviews integrate otherwise prostate cancer trials, it would not be prostatic diseases, i.e. a-blockers, 5a- unmanageable amounts of information from reasonable to carry out sham orchidectomy or reductase inhibitors, antiandrogens and LHRH primary investigations in a way that limits radiotherapy, and characteristic treatment agonists. Once a urologist has decided which bias and random error; meta-analyses allow effects, such as hot flushes with LHRH type of drug to use, their decision about an evaluation of consistency of findings and, agonists, can effectively ‘un-blind’ a study. which specific agent to prescribe will depend if quantitative, may increase the accuracy The more patients in a trial, the more on several factors, including dosing regimen, of estimates of treatment effects [4]. As accurately the size of a clinical effect can be delivery mechanism, speed of onset, such, both should be useful tools in clinical assessed and the less likely the result is to be treatment costs, local prescribing habits, decision-making. However, systematic a result of random chance. The amount of marketing, patient choice, personal reviews and meta-analyses are not always information needed to avoid an incorrect experience and published reports. Ideally, possible. For trials to be combined in a conclusion depends on the size of the effect evidence-based medicine (EBM) should be the systematic review or meta-analysis, there being studied and the level of certainty main factor in treatment choice. This review needs to be a sufficient number of similar required. To be valid, a trial must be of an sets out the principles of EBM and examines studies, i.e. with analogous study designs appropriate design to answer the question the best available evidence for drugs that are in patients with similar disease states and addressed; validity is based on criteria such as commonly prescribed to treat BPH or prostate assessing comparable clinically relevant use of clinically important outcomes and cancer. We consider whether a class effect can outcomes. duration of intervention and observation. be shown for any of these groups of drugs and if class effects should be accepted in Whether trials are combined in a systematic clinical practice. We focus on efficacy, but review or meta-analysis, or examined for their IS THE CLASS EFFECT EVIDENCE-BASED? tolerability can also be an important factor individual merit, their design, endpoints and when choosing between drugs of the same reporting quality need careful examination Drugs are generally considered to be in the class, and will be discussed where relevant. to determine the most appropriate and same class if they have a similar chemical robust data. Trials must be of a high quality structure and mechanism of action, and if to avoid bias, sufficiently large enough to give they confer similar pharmacological effects. PRINCIPLES OF EBM a reliable answer, of good validity, and the However, compounds with very similar population studied should allow the results to structures can have different properties. EBM has been described as the ‘conscientious, be clinically applicable. For example, dihydrotestosterone differs from explicit and judicious use of current best testosterone by only one hydrogen atom, evidence in making decisions about the care One of the most important factors is but has a much greater binding affinity for of individual patients’ [1]. For the clinician, randomization to exclude selection bias; it has the androgen receptor, resulting in different this means identifying the best available been estimated that not randomizing can lead effects on gene expression. Consequently, the evidence from a vast number of published to an overestimation of treatment effect by notion of a pharmacological ‘class effect’ medical reports, assessing whether it is 40% [5]. Other factors (and their percentage should be considered with caution. Indeed, applicable to the individual patient and then overestimation) include small trials (30%), there is no universally accepted definition. using it in clinical practice [2]. Of the types of poor reporting quality (25%), duplicate A class effect is usually taken to mean that published evidence available, systematic reporting (20%), and lack of blinding (17%). drugs in a class have similar therapeutic © 2005 BJU INTERNATIONAL | 95, 743–749 | doi:10.1111/j.1464-410X.2005.05390.x 743 EVANS ET AL. effects and similar safety and tolerability, includes tamsulosin, terazosin, alfuzosin, compared and no definitive conclusions can both in nature and extent [3]. If such a class doxazosin and prazosin, which have different currently be drawn about any differences effect exists, it would be likely that the least selectivity for the a1-adrenoceptor subtypes. between them. While dutasteride is said to costly agent in each class would be the first Systematic reviews have been published for inhibit serum 5a-reductase to a greater choice. However, it is clear that no matter how tamsulosin and terazosin, and a pooled extent than finasteride, the significance of strong the pathophysiological rationale or analysis for alfuzosin (Table 1) [6–8]. These this may be limited, as it is prostatic stromal indirect evidence, the efficacy and safety of a reports, in agreement with an earlier meta- intracellular 5a-reductase that mediates gene new drug must be established in clinical analysis by Djavan and Marberger [9], transcription for growth factor genes. outcome studies, and the equivalency of conclude that the a-blockers are effective Therefore, although the data support the untested drugs even in a well-established and consistently improve LUTS and urinary efficacy of both finasteride and dutasteride in ‘class’ should be considered unconfirmed. flow compared with placebo. BPH, there is as yet insufficient evidence to address the question of whether a class effect For some products routinely used in prostate a-Blockers have been directly compared in a exists for the 5a-reductase inhibitors. medicine, comparative data are limited, but few small trials (involving 50–256 patients) prescription is still widespread based on the [9–12]. A review of trials directly comparing a-BLOCKERS COMBINED WITH assumption of a class effect. To take an tamsulosin with terazosin found that these 5a-REDUCTASE INHIBITORS evidence-based approach to establishing a agents are equally effective in improving class effect, RCTs of direct comparisons of symptoms [7,8]. No definitive conclusions The long-term efficacy of the a-blocker drugs within the class are needed [3]. about differences in efficacy can be made doxazosin and the 5a-reductase inhibitor However, this level of evidence is rarely from these studies; all a-blockers, whether finasteride, as monotherapy or combined, was available. selective or not, seem to have similar efficacy evaluated in a randomized, long-term, in short-term trials [9]. The data suggest that double-blind placebo-controlled trial [15]. The next best level of evidence includes a1-blockers, such as terazosin or doxazosin, While each agent reduced the risk of overall indirect comparisons across two or more give similar improvements as subtype- clinical progression, combined therapy was placebo-controlled trials. In this case, only selective a1a-blockers, like tamsulosin, in peak significantly more effective than either proportional effects such as the relative risk urinary flow rates and symptom scores after monotherapy. This trial was of excellent reduction can be compared. A class effect is 4 weeks of treatment. From these studies it design, yet raises several interesting issues considered to be present when drugs with might be concluded, on the basis of efficacy, about the class effect. For example, is it similar mechanisms of action generate that there possibly is a class effect. However, appropriate to extrapolate the role of relative risk reductions (or odds ratios) that there are differences in tolerability, with doxazosin to other a-blockers? The trial are similar in direction and magnitude [3]. tamsulosin better tolerated