Are placebo controls necessary to test new antidepressants and anxiolytics?

Transcription

1 International Journal of Neuropsychopharmacology (2002), 5, Copyright 2002 CINP DOI: S Are placebo controls necessary to test new antidepressants and anxiolytics? ARTICLE Arif Khan 1,2, Shirin Khan 1 and Walter A. Brown 3 Northwest Clinical Research Center, Bellevue, WA, USA Department of Psychiatry and Behavioral Sciences, Duke University Medical Center, Durham, NC, USA Department of Psychiatry, Brown University, Providence, RI and Tufts University, Boston, MA, USA Abstract One measure of a s effectiveness is the regularity with which it proves superior to placebo. That measure also tells us about the consequences of using a as a standard against which to test a new agent. To assess the frequency with which approved and presumably effective antidepressants and anxiolytics show superiority over placebo, we reviewed placebo-controlled clinical trials of antidepressants and anxiolytics in a singularly large database free of publication bias. We evaluated clinical-trial data from the nine antidepressants approved by the FDA between 1985 and These trials comprised depressed patients who participated in 52 antidepressant clinical trials evaluating 93 arms of a new or established antidepressant. Similarly, we examined clinical trials data from the 13 anxiolytics approved by the FDA between 1985 and These trials comprised 8340 anxious patients, 40 anxiolytic clinical trials and 75 arms of a new or established anxiolytic. Fewer than half (48%, 45 93) of the antidepressant arms showed superiority to placebo. Among anxiolytics, 48% (36 75) of anxiolytic arms showed superiority over placebo. These data suggest that conventional psychopharmacologic s for depression and anxiety are superior to placebo less than half the time and call into serious question the widely propagated notion that placebo controls can be dispensed with in clinical trials of these agents. Exclusion of placebo controls in favour of non-inferiority trials would result in a high likelihood that ineffective antidepressants and anxiolytics would be foisted on the public and, less dangerous but also problematic, that potentially effective agents would be missed. Received 3 March 2002; Reviewed 15 May 2002; Revised 20 May 2002; Accepted 22 May 2002 Key words: Antidepressants, anxiolytics, clinical trials, Food and Drug Administration, placebo. Introduction Restrictions on placebo use in medical research, including psychopharmacology research, are being considered in several quarters. For example, according to the revised version of the Declaration of Helsinki (2000): The benefits, risks, burdens and effectiveness of a new method should be tested against those of the best current prophylactic, diagnostic, and therapeutic methods. For now, the USA s Food and Drug Administration (FDA) continues to require two or more positive placebocontrolled trials to approve a new agent. But the European Agency for the Evaluation of Medicinal Products (EMEA, 2001) has incorporated and implemented a broad interpretation of the revised Declaration of Helsinki and many IRBs (or ethics committees) in the USA and Europe are reluctant to allow placebo controls in anxiolytic and Address for correspondence: Dr A. Khan, th Ave NE no. 112, Bellevue, WA 98004, USA. Tel.: (425) Fax: (425) akhan nwcrc.net antidepressant trials or simply disallow placebo in all such trials. Recent articles, editorials, and letters have articulated the ethical and scientific issues pertinent to the use of placebo controls (Emanuel and Miller, 2001; Lavori, 2000; Leber, 2000; Miller, 2000). Among the controversies of particular importance to the conduct of antidepressant and anxiolytic clinical trials is whether the new agent should be compared to placebo or standard (non-inferiority trial). As is often the case in such matters, the debate is fuelled less by information than by misconceptions, intuition, and the posturing of stakeholders. One piece of information central to this debate, and as yet missing, is the regularity with which the available standard antidepressants and anxiolytics prove superior to placebo. This paper offers such information. Methods We reviewed the Summary Basis of Approval (SBA) data from the FDA archives, under the Freedom of Information Act (US Congress, 1996), that describes clinical trial

2 194 A. Khan et al. Table 1. Outcome for antidepressants (new and established) compared to placebo in 52 placebo-controlled pivotal trials New antidepressant vs. placebo Established antidepressant vs. placebo New antidepressant placebo arms* No. adequate dosing No. (%) of adequate arms inadequate dosingf arms No. (%) of arms 1. Fluoxetine (33) 0 1a 1a (100) 2. Sertraline (80) 0 2b 1b (100) 3. Paroxetine (55) 0 6a 3a (50) 4. Venlafaxine IR (75) 2 3a,c 1c (33) 5. Nefazadone (25) 2 5a 3a (60) 6. Mirtazapine (42) 1 5b,c 4b (80) 7. Bupropion SR (0) 4 0 na 8. Citalopram (43) 1 0 na 9. Venlafaxine ER (50) 0 2d,e 0 (0) Total (45) (58) patients a Imiprimine; b amitriptyline; c trazodone; d venlafaxine IR; e paroxetine; f none of the inadequate dosing met significance to placebo. * Indicates number of placebo-controlled trials or data for placebo-treated patients. Statistical significance of p 0 05 compared to placebo. na, not applicable. outcomes for recently approved antidepressants (n 9) and anxiolytics (n 13). These data were obtained by a specific request to the FDA (Freedom of Information Staff, Room 12A-16, 5600 Fishers Lane, Rockville, MD 20857). The data include pivotal clinical trial data that were used for assessing the effectiveness of new agents approved from 1985 to Under Title 21 of the Code of Federal Regulations (FDA, 1997), these pivotal studies by convention and practice are randomized, double-blind, and placebo controlled, with established criteria for the sample under study and defined criteria for response. Typically, these are phase 2 and 3 studies with several hundred patients. Studies are designated as pivotal without regard to outcome as many do not demonstrate superiority of the antidepressant or anxiolytic over placebo. We have previously described methods of data collection, details of trial designs as well as methods of analysis. Factors such as number of patients, duration of trial, specific diagnostic instruments used are described and available in FDA SBA reports and have been published earlier for the antidepressant trials (Khan et al., 2000, 2001). Among the 52 antidepressant clinical trials conducted during the development of the 9 antidepressants, there were 79 new (test) arms for the antidepressant under investigation as several trials evaluated multiple doses of the new antidepressant; of these, 10 arms evaluated doses of antidepressant that are not approved for marketing. An additional 24 arms evaluated an established antidepressant (active comparator) for a total of 93 antidepressant arms (Table 1). Among the 40 anxiolytic clinical trials, there were 62 new (test) arms for the anxiolytic under investigation; of these, 1 panic disorder arm evaluated a dose that is not approved for marketing. An additional 14 arms evaluated an established anxiolytic (active comparator) for a total of 75 anxiolytic arms (Table 2). We examined the results from each of the new and established arms for frequency of superiority over placebo (p 0 05) (Tables 1 and 2). Results Table 1 describes the 3462 depressed patients assigned to placebo in 52 clinical trials compared to the 4967 patients assigned to a new antidepressant evaluating a dose that is currently approved for market and the 1601 patients assigned to an established antidepressant. Forty-five of the 93 (48 4%) trials of new and established anti-

3 Are placebo controls necessary to test new drugs? 195 Table 2. Outcome for anxiolytics (new and established) compared to placebo in 40 pivotal placebo-controlled trials New anxiolytic vs. placebo Established anxiolytic vs. placebo New anxiolytic placebo arms* adequate dosing No. (%) of adequate arms arms No. (%) of arms (A) Panic disorder 1. Alprazolam (50) 2a,b 1a (50) 2. Paroxetine (17) 2a,b 0 (0) 3. Clonazepam (20) 0 na 4. Sertraline (40) 0 na Total (30) 4 1 (25) patients (B) Generalized anxiety disorder 5. Buspirone (50) 6e,f 3e,f (50) 6. Venlafaxine XR (40) 1g 0 (0) Total (45) 7 3 (43) patients (C) Social anxiety disorder 7. Paroxetine (60) 0 na Total (60) 0 na patients na (D) Obsessive compulsive disorder 8. Clomipramine (100) 0 na 9. Paroxetine (60) 2b 2b (100) 10. Fluvoxamine (100) 0 na 11. Fluoxetine (50) 0 na 12. Sertraline (50) 1b 1b (100) Total (62) 3 3 (100) patients (E) Post-traumatic stress disorder 13. Sertraline (50) 0 na Total (50) 0 na Patients na a Imipramine; b phenylzine; c clomipramine; d alprazolam; e diazepam; f clobazam; g buspirone. * Indicates number of placebo-controlled trials or data for placebo-treated patients. Statistical significance of p 0 05 compared to placebo. There was one additional arm evaluating an inadequate dose that was not ly significant to placebo. na, not applicable. depressants showed superiority over placebo. There were an additional 10 antidepressant arms that evaluated doses deemed inadequate or below therapeutic dose by the FDA. None of them showed significance over placebo. Table 2 describes the 2428 patients with five types of anxiety disorders assigned to placebo in 40 clinical trials, compared to the 5062 patients assigned to a new anxiolytic evaluating a dose that is currently approved for market and the 850 patients assigned to established anxiolytics. Thirty-six of the 75 (48 0%) trials of new and established anxiolytics showed superiority over placebo. Discussion Currently available antidepressants and anxiolytics do not regularly show superiority to placebo. In fact, in more than half of the trials (87 168, 51 8%), response to antidepressants and anxiolytics was indistinguishable

4 196 A. Khan et al. from response to placebo. Based on these data, if new antidepressants and anxiolytics were tested against standard rather than placebo, about half of the trials would yield invalid results: ineffective s mistakenly accepted as effective and effective s mistakenly designated inactive. To our knowledge, ours is the first report assessing how often antidepressants and anxiolytics are superior to placebo in clinical trials. However, our findings are similar to earlier publications assessing the magnitude of difference between drug and placebo in antidepressant trials (Khan et al., 2000, 2001; Kirsch and Sapirstein, 1998; Storosum et al., 2001). These data, in combination with other recent reports showing increasing magnitude of placebo response (Walsh et al., 2002) and two failed trials (Hypericum Depression Trial Study Group, 2002; Shelton et al., 2001) with St John s Wort, bolster the idea of using placebo. For example, exclusion of placebo by the Hypericum Depression Trial Study Group could easily have led to the incorrect conclusion that sertraline and St John s Wort are equivalent s for major depressive disorder. It is worth noting that the less than impressive showing of antidepressants and anxiolytics in these clinical trials data does not necessarily reflect the effectiveness of these agents in clinical practice. The depressed and anxious patients participating in clinical trials are a unique segment of the population with these conditions. They are mild to moderately ill, not suicidal, and without significant psychiatric or medical co-morbidity. The clinical features and the clinical trials process itself, which includes considerable evaluation and attention and creates high expectations for improvement, renders clinical trials participants particularly likely to improve with placebo. Data from several sources, for example, suggest that drug placebo differences are robust in the chronic and severely ill depressed patients who are not usually the subjects of clinical trials. Other disorders such as obsessive compulsive disorder and schizophrenia may have a lower placebo response than depression or anxiety disorders allowing for larger drug vs. placebo differences. The FDA SBA reports consist of a large database, include all data from all clinical trials and are free of publication bias. In this way they are ideally suited for answering questions, such as the one we posed, requiring a large number of trials. But these data have some limitations. For example, they do not allow us to calculate the proportion of patients who experienced a clinically meaningful response, compare results across trials or calculate effect sizes. In addition, we note that the data are a continued analysis of earlier published reports and are not from a new source (Khan et al., 2000). Also, the variability in individual study inclusion exclusion criteria, not discernible via SBA reports, may have contributed to these findings. In conclusion, the risks of placebo do warrant the serious consideration they have received. The elimination of placebo controls calls for serious consideration as well. It is in the public interest for both matters to be informed not only by broad ethical and scientific principles but also by the available data. References Emanuel EZ, Miller FG (2001). The ethics of placebocontrolled trials a middle ground. New England Journal of Medicine 435, European Agency for the Evaluation of Medicinal Products (2001). EMEA CPMP position statement on the use of placebo in clinical trials with regard to the revised Declaration of Helsinki. Available at: http: www. emea.eu.int pdfs human press pos en.pdf (accessed 21 January 2002). Food and Drug Administration (1997). Code of Federal Regulation, 21. Parts Revised as of 1 April 1997 (codified at 21 CRF ). Vol 5. Washington, DC: US Government Printing Office; 1997, Freedom of Information Act (1996). 5 US Congress. 552 (1994 & Supp. II 1996). Available at: http: 04foia (accessed 21 January 1999). Hypericum Depression Trial Study Group (2002). Effect of Hypericum perforatum (St John s Wort) in major depressive disorder. Journal of the American Medical Association 287, Khan A, Khan SR, Leventhal RM, Brown WA (2001). Symptom reduction and suicide risk in patients treated with placebo in antidepressant clinical trials: a replication analysis of the Food and Drug Administration database. International Journal of Neuropsychopharmacology 4, Khan A, Warner HA, Brown WA (2000). Symptom reduction and suicide risk in patients treated with placebo in antidepressant clinical trials: an analysis of the Food and Drug Administration database. Archives of General Psychiatry 57, Kirsch I, Sapirstein G (1998). Listening to prozac but hearing placebo: a meta-analysis of antidepressant medication. Prevention and Treatment, vol. 1, Article American Psychological Association. Available at: http: journals.apa.org prevention volume1 pre a.html (accessed 14 September 1998). Lavori PW (2000). Placebo control groups in randomized trials: a statistician s perspective. Biological Psychiatry 47, Leber P (2000). The use of placebo control groups in the assessment of psychiatric drugs: an historical context. Biological Psychiatry 47, Miller FG (2000). Placebo-controlled trials in psychiatric

5 Are placebo controls necessary to test new drugs? 197 research: an ethical perspective. Biological Psychiatry 47, Shelton RC, Keller MB, Gelengerg A, Dunner DL, Hirschfeld R, Thase ME, Russel J, Lydiard RB, Crits-Cristoph P, Gallop R, Todd L, Hellerstein D, Goodnick P, Keitner G, Stahl SM, Halbreich U (2001). Effectiveness of St John s wort in major depression: a randomized controlled trial. Journal of the American Medical Association 285, Storosum JG, Elferink AJA, van Zwieten BJ, van der Vrink W, Gersons BPR, van Strik R, Broekmans AW (2001). Shortterm efficacy of tricyclic antidepressants revisited: a metaanalytic study. European Neuropsychopharmacology 11, Walsh BT, Seidman SN, Sysko R, Gould M (2002). Placebo response in studies of major depression. Journal of the American Medical Association 287, World Medical Association Declaration of Helsinki (2000). Ethical Principles for Medical Research Involving Human Subjects. Available at: http: e policy 17-c e.html (accessed 27 November 2000).