Reply by Drs. Rosenbaum and Fava

1. The topic of withdrawal or discontinuation-emergent symptoms with selective serotonin reuptake inhibitors (SSRIs) has become a focus of controversy, in part because of disagreement among experts as to its frequency and clinical significance and partly because of the marketing competition between manufacturers of SSRIs with long or short half-lives. It would level the playing field to assert this is a problem of equal significance for all. This appears to be Mr. Medawar’s position.

2. Clinical observations in our own patients made as early as six years ago led us to design and develop studies to address what appeared to us to be an important clinical issue that was under-recognized by psychiatrists and not appreciated by pharmaceutical companies. The issue was the apparent association of significant symptoms with the abrupt treatment interruption or discontinuation of shorter half-life agents -- symptoms such as nausea, dizziness, fatigue, myalgia which would typically be attributed to the "flu" or to some other medical illness.

3. Further, as co-investigators in a multi-center, placebo-controlled study of a short-acting antidepressant (not an SSRI), we assessed the emergence of spontaneously-reported withdrawal symptoms among patients completing the study at our site. In this blind study, we found that these symptoms were significantly more frequent and more severe upon discontinuation of drug than placebo. These results were published in the American Journal of Psychiatry.

4. Whenever investigators identify meaningful differences in safety profile across drugs of the same class (for example, a relatively higher rate of sedation/drowsiness or dry mouth/constipation with tricyclic antidepressants compared to SSRIs), it should not be surprising that the pharmaceutical companies advantaged by the findings are likely to support and publicize the investigations, while others may ignore or refute them.

5. Mr. Medawar mistakenly asserts that the SSRI withdrawal studies "originated" with Eli Lilly; in fact, the idea, design, methods, and key instrument for these studies came from us. In the case of the treatment interruption studies, we actually designed the studies many years ago and then sought a sponsor. Eli Lilly initially considered the possibility of sponsoring these studies, but then declined to pursue the investigations, in part because of uncertainty that the SSRIs would differentiate. (So with hindsight, Mr. Medawar may believe the design was a sure win, but until a controlled study is accomplished, that is only opinion.) It is standard procedure to design studies with a hypothesized outcome in mind. For example, placebo is typically compared with active drugs in many efficacy and safety studies in medicine. The assumption is that placebo will do less well…this doesn’t make the comparison a rigged or unfair one, but creates a model for comparison of outcomes. We believed, based on our clinical experience, and on a number of case reports in the literature, that withdrawal symptoms with treatment interruption would be more common and severe with the short-acting SSRIs. After many anecdotes were published in the world literature and reports by the UK’s Committee on Safety of Medicine about discontinuation-emergent adverse events of shorter acting SSRIs, Eli Lilly finally reached the point of having an interest (?) in sponsoring the study we designed. We were notified that they had reconsidered our project, but now wanted to go forward with multi-center studies rather than an MGH project. After waiting so long, they were wanting a faster completion of the study than we could accomplish at our site. Thus the MGH-based project we hoped to accomplish here or with two selected academic partners was then replaced by two studies that could enroll subjects more quickly. One study was carried out across several academic sites (including ours) over the course of two years and another study, with a simplified design, was carried out by a CRO organization. Again, it was our idea, our design, our authorship of the protocol, our choice of the psychometric instruments used in studies, and our development of a new instrument used to record symptoms of discontinuation from SSRIs (DESS). The IRB’s of the several academic sites involved reviewed and approved the protocol. We worked with the statistician to analyze data for the published manuscript, revised multiple iterations of the manuscript and responded to the peer reviewers’ critiques and suggestions.

6. That a research psychiatrist collaborator from Lilly was given first authorship on the interim-data abstract, reflected the fact that she presented the data at the meeting session cited (we were not there), and bears little on the appropriateness of order of authorship for the final study manuscript. It is customary for the presenting author to be the first author of the particular presentation. Since the creative process of methods, design, and instruments was ours, the authorship of the full, referee-reviewed, published paper reflected that intellectual contribution, in accordance with the guidelines on authorship proposed by the New England Journal of Medicine.

7. Mr. Medawar states that the study published in Biological Psychiatry did not report spontaneously reported adverse events. This statement is incorrect; the study reports spontaneously reported adverse events (as did the Zajecka paper) and not just solicited ones.

8. Mr. Medawar speculates that the JCP "Highlights" were written by Eli Lilly. That belief is also incorrect, as the JCP "Highlights" were meeting proceedings written by JCP (Journal of Clinical Psychiatry) writers ("Physicians Postgraduate Press") for their journal, not written by Lilly, although Lilly did give an unrestricted educational grant to the journal to run the meeting, chaired by the Chairman of Psychiatry at Stanford University.

9. Mr. Medawar insinuates that manufacturers of the comparator treatments were not informed of the protocol.Before Lilly tabled our study five years ago, we were informed it had been held up because of difficulty obtaining a study drug supply agreementfrom the manufacturer of one, but not from the manufacturer of the other comparator agent.. We were told the one company was reluctant to provide study drug when they were notified about the protocol. Thus, someone at each of the other companies was notified, despite Mr. Medawar’s assumption and implication otherwise.

10. There was not evidence of an early fluoxetine discontinuation syndrome in our published study as Mr. Medawar asserts. Patients on fluoxetine did not have an increase in discontinuation emergent signs and symptoms. Mr. Medawar assumes that an increase in depression rating at one point in time in a small number of patients in treatment for depression means discontinuation. As clinicians know, in a population of patients with depression, even those who have been in remission at first assessment, some will be more symptomatic on re-evaluation at a later time. Symptoms of the discontinuation syndrome were not present in the fluoxetine group during this interruption period.

11. The "second" trial (Zajecka et al), which Mr. Medawar implies was rigged to not disprove the hypothesis, was actually conducted several years earlier and before there ever was a hypothesis. The paper was post-hoc data mining to see if a dataset that had been obtained without consideration of our hypothesis would show evidence of a discontinuation syndrome 2, 4, and 6 weeks after stopping fluoxetine. (The authors of that paper included those who had conducted the original long-term clinical trial in 1991 at 5 academic sites.) There was no evidence of a discontinuation syndrome but some mild dizziness (4% and 6% of those discontinued to placebo) was noted at weeks 4 and 6. Mr. Medawar states that the study's methodology was sloppy and there was less scrutiny on withdrawal symptoms. Although it is true that the analysis could only examine spontaneously-reported adverse events, the study we recently published in the Am J Psychiatry had the same methodology as the Zajecka paper, had a smaller sample size, had no systematic assessment of withdrawal symptoms (the symptoms listed were only the ones spontaneously reported by patients), but still showed a marked difference (77% vs 22%) in the rate of withdrawal reactions between drug and placebo. Finally both the Biological Psychiatry paper by Rosenbaum et al. and the Zajecka et al., paper ultimately relied on spontaneous reports of adverse effects and not on "close and systematically" asked questions. The Zajecka paper is criticized for not counting pre-existing adverse effects in rating discontinuation symptoms. If patients have the same symptoms before discontinuation while on the drug as after, the latter should not be deemed caused by discontinuation

12. The Biological Psychiatry study and another were both submitted to the FDA which gave Lilly permission to detail (market) the findings. This has fueled a marketing war, and reprints of the study abound in several languages. There are certainly many examples in clinical research of scientific and objective findings being utilized in reprint form and by citation for marketing purposes.

13. Using similar methods to the study published in Biological Psychiatry, another study was accomplished, and the findings in this population, who were on the same SSRIs but for a much shorter period than in the first study, were much less robust...In fact for one time point, sertraline performed better than fluoxetine...all testimony to the objectivity of the study design and investigators, and presumably to the fact that risk for the syndrome goes up when interrupting treatment of longer duration.

14. That we have authored papers involving fluoxetine reflects, in large measure, the fact that we are mood disorder researchers, and that this agent was the first of a new class, has been around for over a decade, and has been the subject of over 6000 papers in refereed journals.It also reflects the fact that Lilly, like all pharmaceutical companies, has supported clinical research about the effects of its molecule. It is important to note that much of our research with this agent has been sponsored with peer-reviewed grants from the National Institute of Mental Health.

15. A review of our full portfolio of current and past research at MGH will reveal a very balanced and extensive list of funding sources, reflecting what we always assert: our conflicts of interest are in equipoise. So we call it as we see it, and sometimes observers and colleagues like the results of that process and sometimes they don't.

16. Mr. Medawar implies singularity of our relationship to Lilly and fails to represent our extensive and important relationships with essentially all of its competitors including the makers of the comparators in the study under question for whom we have served or are serving as consultants, investigators and advisory board members.

17. Finally Mr. Medawar seems particularly excited by the news about consequences of abrupt discontinuation of antidepressant medication despite the fact that there are potential adverse consequences of abrupt discontinuation for many classes of medicine to which the organism adapts with chronic use: e.g. beta-blockers and other anti-anginal and anti-hypertensive medications, anti-epilepsy drugs, steroids, pain treatments, oral decongestants, and others. Enhancing physician awareness of the need to taper when treatment is ending and patient education to minimize abrupt interruptions from treatment would be a more useful application of Mr. Medawar’s bandwidth.

CLICK HERE TO READ PRELIMINARY RESPONSE

CLICK HERE TO READ SUBSTANTIVE RESPONSE

	Contents page
	What's New?