ANTIPSYCHOTIC DRUGS FOR DEPRESSION?

Readers concerned about commercial influence know that the Devil is in the details when it comes to recognizing and exposing distortions like spin, bias and shills in journal publications. One guiding observation is that “For private companies, publication and a positive press have cash value…” (E.W. Campion, New England Journal of Medicine, 2004).

With an eye to the sly details, this posting will describe a current campaign to promote the atypical antipsychotic (AAP) drug risperidone as an augmenting agent in nonresponding major depression. The campaign aims to shape a favorable climate of opinion for the drug through experimercials (commercially strategic clinical trials) and journal publications that are really infomercials. The stakeholders are some major corporations, “key opinion leaders” (KOLs), leading medical journals, and several million patients who suffer from nonresponsive depression in the US. The winners are the KOLs and the corporations, while the big losers are the patients.

We begin with the November 2006 report of an experimercial testing Janssen’s AAP drug risperidone for nonresponding depression, published in the journal Neuropsychopharmacology (NPP). This is the house journal of the American College of Neuropsychopharmacology (ACNP), which aspires to be the premier organization in its field. Bottom line: short term, open label, uncontrolled use of risperidone appeared to be helpful, but placebo-controlled continuation beyond 6 weeks was not effective in preventing relapse. The lead author was Mark H. Rapaport, MD from Cedars-Sinai Medical Center in Los Angeles. A co-author was the controversial Martin Keller, MD from Brown University, and the senior (last) author was Charles B. Nemeroff, MD, PhD from Emory University, who was also editor in chief of the journal.

Dr. Nemeroff’s is a long-time KOL for Janssen’s efforts to create a favorable climate of opinion for risperidone in depression. He fronted for the company in presentations of the risperidone infomercial at the ACNP meeting in 2004, in a Janssen-financed journal supplement in 2005, and in co-authoring the 2006 infomercial in his own journal, NPP. Readers will recall that Dr. Nemeroff ran into problems in 2003 for failing to disclose his financial conflicts of interest in a Nature publication, and again in 2006 for failing to disclose his commercial ties in a review article he co-authored favorable to the Cyberonics vagus nerve stimulation (VNS) device for depression (also published in the journal he edited). As a result of that episode, Dr. Nemeroff resigned as editor of NPP, but not before publishing the infomercial mentioned above on risperidone in depression. One wants to ask whether these publications that he co-authored went through a credible peer review in Dr. Nemeroff’s journal. One also wants to ask whether Dr. Nemeroff abused his position as editor to give product placement to his corporate clients Cyberonics and Janssen. And one wants to ask why the risperidone infomercial was published in NPP, given that the data had already been published by Dr. Nemeroff himself in another journal, in a special supplement financed by Janssen. This prior publication of the risperidone trial, that by journal policy should have precluded publication in NPP, was not revealed (cited) by Dr. Nemeroff and his co-authors. Did they think nobody would notice?

The main outcome of the risperidone trial reported in NPP was negative: long term use of risperidone did not prevent relapse. By assiduous secondary analyses and data mining, the authors found something positive to salvage their effort. They claimed by two analyses that risperidone was effective in preventing relapse in a subgroup of patients. In short order, however, they qualified the original report. First they belatedly disclosed the authors’ commercial ties to Janssen. A second Corrigendum retracted one of the claims of efficacy: by some process of Immaculate Conception, an “error” had occurred whereby a p value of 0.4 was transformed into a p value of 0.05, first in the text and again in the Abstract. This retraction occurred only because alert readers asked probing questions of the lead author, Dr. Rapaport. No author took responsibility for this compound misrepresentation in NPP. The “error” was so obvious that all authors and any conscientious reviewer should have seen it. You begin to see the picture: we have the appearance of editorial self-dealing, including product placement for a corporate client of the editor; an incompetent or possibly dishonest journal review process; and the appearance that somebody went out of his way in two places to insert a false claim of efficacy into the report of a negative clinical trial. Finally, we have reasonable cause to suspect that the KOLs did not actually read the article they were credited with authoring. For busy KOLs, tradecraft is a thing of the past.

Soon, more questions arose. In a letter to NPP published in 2007, I challenged the second claim of efficacy for risperidone in preventing relapse in a subgroup: the claim made in the journal and in two of Dr. Nemeroff’s prior publications did not match the corporation’s analysis reported on ClinicalTrials.gov. Rapaport, Keller and Nemeroff had “nudged” the statistical result of Janssen’s analysis from borderline to unqualified statistical significance. My letter also detailed how these KOLs downplayed the disturbing data on metabolic toxicity (weight gain): the raw data were stated but the unfavorable statistical analysis of those data that appeared on ClinicalTrials.gov was suppressed. The KOLs chose not to reveal that patients who received risperidone gained significantly more weight than those who received placebo. The appearance of bias and spin was unmistakable. In response to my letter, the lead author, Mark H. Rapaport, MD, who seemed to have been deserted by his KOL colleagues, abjectly disowned any claim of efficacy whatsoever for risperidone, while being nonresponsive to the charge that they had manipulated the risk-benefit profile of the drug. So now we have high profile academic KOL authors putting lipstick on the pig with both false and exaggerated claims of efficacy in a negative study, misstating the results of statistical analyses, and “burying” inconvenient toxicity analyses. Were the reviewers for NPP so incompetent that they could not see what any reader could see? One wants to ask whether this report, with the editor in chief as senior author, received any review at all.

So far so good: with Dr. Rapaport’s latest retraction, Janssen and its hired KOLs now disown any claim that risperidone is effective in long term prevention of relapse in nonresponding depression. What about short term treatment? The infomercial in NPP contained only uncontrolled, open label data on the short term efficacy of risperidone. That is the weakest level of evidence, even though it was touted by the KOLs with the message that many patients responded “rapidly and robustly.” Inconveniently, many of them relapsed even with continuation risperidone therapy. So much for robustness.

As luck would have it, Janssen reported a placebo-controlled short term study of risperidone in nonresponsive depression in the November 6, 2007 issue of Annals of Internal Medicine (AIM) (Mahmoud RA et al. Ann Intern Med 2007; 147: 593-602). The authors are Janssen employees who had been co-authors with Rapaport, Keller and Nemeroff on the now retracted NPP report. (Possibly in response to static over the earlier report, Nemeroff and Rapaport plus a few other KOLs were simply acknowledged as advisors instead of receiving co-author status on the AIM report).

The bottom line? Though there were some statistically significant differences favoring short term risperidone over inactive placebo by conventional rating scales, evidence of clinical utility was underwhelming. Once again, Janssen’s presentation of the results was biased: the authors used completer data for efficacy, which improved the appearance of benefit for risperidone, whereas they used Intent to Treat (ITT) data for toxicity, which minimized the apparent metabolic adverse effects of risperidone. Dr. Ralph Koek and I pointed out in an on-line response to the journal that when ITT data are used consistently the Number Needed to Treat (NNT) for remission was 15.6 at 4 weeks and 10.2 at 6 weeks. Those NNTs do not indicate compelling clinical benefit for risperidone: an NNT of 15 means one would need to treat 15 patients with risperidone to obtain 1 remission that would not have occurred anyway with placebo. Moreover, the metabolic toxicity of risperidone (weight gain) was significantly increased over placebo, but the authors downplayed that problem by keeping it out of the Abstract, and it did not appear in the “Summaries for Patients” posting by the journal. Once again, Janssen used questionable manipulations to maximize the appearance of efficacy while minimizing the apparent risk of risperidone in nonresponding depression. This is called “augmenting” the risk-benefit profile.

So now there is evidence from two controlled trials that risperidone confers little significant benefit and has a dubious risk-benefit profile either in short term use or in longer term use for nonresponsive depression. In light of this evidence it is difficult to justify the broad and early use of risperidone or, probably, any other atypical antipsychotic drug in such patients. Richard Shelton, MD from Vanderbilt and George Papakostas from MGH concur with this conclusion, in a thoughtful review article that has just appeared on-line in Acta Psychiatrica Scandinavica. These two investigators conducted several of the early trials of other AAP drugs in depression. They gave a timely warning about the late-appearing but often devastating side effect of tardive dyskinesia that can result from exposure of depressed patients to AAP drugs. Needless to say, that tissue was not addressed in either of the Janssen publications talking up risperidone for depression.

Studies of this issue routinely take the de minimis approach of testing AAP agents only against an inactive placebo, ignoring the problems of unblinding that occur when these sedating agents are used, and avoiding comparisons with other options. Quite possibly, any sedating or anxiolytic agent like off-patent trazodone would have effects indistinguishable from risperidone. The corporation’s goal is not to answer such important clinical questions, however, but only to clear the regulatory bars needed to expand the market for its drug. Patients lose out when an expensive atypical antipsychotic drug with an NNT of 10-15 is promoted while an inexpensive option like lithium augmentation, which has an NNT of 4, is ignored. But, as these experimercials are driven by marketing rather than by scientific or educational goals, the evidence that risperidone is inferior to lithium, the best established augmenting agent, is routinely ignored in the clinical trial reports.

So, here is the litany of sly tactics used by Janssen with the willing cooperation of KOLs like Nemeroff , Rapaport and Keller. Remember, these KOLs are paid directly or indirectly by Janssen for their association with the marketing campaign: design a clinical trial in which unblinding is guaranteed and ignore this confound which favors the drug; secure high profile product placement in a journal like NPP, with the cooperation of the editor; make false claims of efficacy in a negative trial report; when challenged, lamely try to pass off this compound misrepresentation as an “error” for which nobody accepts responsibility; take self-serving liberties with the results of statistical analyses so as to claim efficacy where none exists; suppress inconvenient statistical analyses of toxicity; manipulate the apparent risk-benefit profile of the drug; constructively plead nolo contendere to the charge of manipulating key data analyses; focus on the statistical significance of psychometric outcome measures while ignoring clinimetric measures of usefulness like NNT; bias the analysis of efficacy by using completer data; bias the analysis of toxicity by using ITT data; and ignore existing data that suggest the corporation’s drug is inferior to established alternatives.

Any one of these sly tactics might be viewed individually as an unfortunate slip, for which some will incline to give the authors the benefit of the doubt. The concatenation of deceits, however, speaks for itself. Readers ought never to underestimate the cunning and the venality of marketeers and their co-opted academic KOLs in corrupting the medical literature. In a future posting I will continue this sordid tale with illustrations from the equally compromised world of CME programs.