Pfizer’s selective norepinephrine re-uptake inhibitor reboxetine (Edronax) is “overall, an ineffective and potentially harmful antidepressant”, an analysis of published and unpublished clinical trials by German researchers has concluded.

The systematic review and meta-analysis featured in the BMJ also found that the available evidence on reboxetine had been “substantially affected by publication bias”. This prompted further calls for mandatory disclosure of all clinical trial data and some soul-searching about the effectiveness of meta-analyses when large chunks of relevant information may be missing – in the case of reboxetine, data on 74% of patients were previously unpublished.

The reboxetine analysis is also the latest twist in a battle between Pfizer and Germany’s health technology assessment body, the Institute for Quality and Efficiency in Health Care (IQWiG), to secure access to undisclosed study data on the antidepressant.

The conclusions published in the BMJ will be seen as a vindication of IQWiG’s efforts and another blow to industry’s declared commitment to transparency in clinical research. Pfizer, though, insists it discloses the results of its clinical trials to regulatory authorities worldwide and that reboxetine remains an effective acute treatment option for patients with depressive illness or major depression.

IQWiG-Pfizer battle

In September 2009, IQWiG said the German government should push for an EU regulation making it compulsory to register and publish within tight deadlines the results of all clinical trials. The Institute cited its attempts to secure a full dataset on reboxetine for a health technology assessment alongside two other antidepressants, mirtazapine (MirtaLich, Winthrop Arzneimittel) and bupropion XL (Wellbutrin XL, GlaxoSmithKline).

According to IQWiG, this process was “greatly hindered” by Pfizer, which refused “for a long time” to make trial data on reboxetine available to the Institute and only caved in “when put under public pressure”. The full data showed that some 5,100 patients in total were treated with reboxetine in the 17 trials included in the benefit assessment, IQWiG noted at the time. Yet “sufficiently transparent” published data were available for only 1,600 of those patients.

The BMJ paper explores those findings in more detail as well as giving IQWiG’s final verdict on reboxetine’s safety and efficacy. As BMJ editor Fiona Godlee and associate editor Elizabeth Loder point out in an accompanying editorial, that verdict “starkly contradicts the findings of other recent systematic reviews and meta-analyses published by reputable journals” – not so surprising, Godlee and Loder add, as these studies “did not combine all of the existing evidence from clinical trials”.

The German team led by Beate Wieseler, deputy head of the department of drug assessment at IQWiG, analysed 13 acute treatment trials with reboxetine that were placebo-controlled, SSRI antidepressant-controlled or both, including 10 provided by Pfizer. Of these 10 trials, seven were unpublished and three were incompletely published, Wieseler et al reported. In all, a substantial proportion – 74% - of the patient data on reboxetine had not previously been published.

No significant benefits

The researchers found that reboxetine was ineffective as an antidepressant as it showed no significant benefits over placebo in terms of remission rates. The Pfizer antidepressant lagged behind selective serotonin re-uptake inhibitors (SSRIs: fluoxetine, paroxetine and citalopram) on both remission and response rates. It was also inferior to placebo on harm outcomes and to fluoxetine on withdrawals due to adverse events.

Addressing publication bias, the research team noted that, for both of the benefit outcomes assessed, the addition of unpublished data “changed the superiority of reboxetine versus placebo shown in published data to a non-significant difference and also changed the non-significant difference between reboxetine and SSRIs to an inferiority of reboxetine”.

A comparison of the published data on reboxetine with the full dataset showed that the published data overestimated the full beneficial effect of reboxetine versus placebo by 99-115% and of reboxetine versus the SSRIs by 19-23%, Wieseler et al added.

The addition of unpublished data also changed the relationship between reboxetine and placebo in terms of harm outcomes from a non-significant difference to inferiority of reboxetine, with the same effect on the relationship between reboxetine and fluoxetine in terms of withdrawals due to adverse events.

As Wieseler and colleagues observe in an analysis of their difficulties in obtaining data for IQWiG’s health technology assessment of reboxetine, the drug was approved in a number of European countries in 1997 and was granted a preliminary licence by the US Food and Drug Administration in 1999. Final FDA approval was subsequently declined in 2001, though, because of “a lack of compelling evidence of efficacy”.

In the UK, the National Institute for Health and Clinical Excellence concluded, on the basis of published studies, that reboxetine was “superior to placbo and as effective as other antidepressants in the treatment of depression”, Wieseler et al note, adding: “In our opinion, this conclusion can no longer be upheld”.

Effective treatment

In the UK market, reboxetine is licensed for the acute treatment of depressive illness/major depression and for maintaining clinical improvement in patients initially responding to treatment. “This medicine presents an effective treatment option to clinicians for use in patients suffering from these conditions,” a Pfizer spokesperson stated.

The company “discloses the results of its clinical trials to regulatory authorities all around the world”, it added. “These regulatory authorities carefully balance the risks and benefits of each medication, and reflect all important safety and efficacy information in the approved product labelling.”
 Pfizer says it will review in detail the reboxetine meta-analysis published in the BMJ and “will provide further comment after completing the review”. 

In their editorial, Godlee and Loder note that the Food and Drug Administration Amendments Act of 2007 and parallel European efforts will increase the accessibility of clinical trial results and make it more difficult to conceal information. However, these efforts “do not solve the problem of our current evidence base, which contains incomplete and questionable evidence”.

Indeed, the authors suggest, the “reboxetine story and similar episodes must call into question the entire evidence synthesis enterprise. Meta-analyses are generally considered the best form of evidence, but is that a plausible world view any longer when so many of them are likely to be missing relevant information?”
Godlee and Loder also accept there has to be “a shared commitment to set the record straight”.

The “sometimes rancorous debate over research transparency, and the reasons for publication and non-publication” can overshadow the crucial need for efforts to restore trust in existing evidence, they comment. “To that end, the BMJ is more interested in constructive use of data than finger pointing or blame.”