GSK - Imitrex: oh no, here comes trouble

Roy Poses writes:

Correspondence to the Lancet two weeks ago revealed another instance in which clinical research studies that failed to provide results favorable to a sponsor's product were suppressed. [Tfeft-Hansen PC. Unpublished clinical trials of sumatriptan. Lancet 2009; 374: 1501-2. Link here.]

Sumatriptan, sold as Imitrex by GlaxoSmithKline, and now available generically, is a commonly used treatment for acute migraine headaches. A Cochrane review from 2003 concluded that multiple clinical trials versus placebo showed that the drug is a safe and effective treatment of acute migraine [link here.] A comparison of multiple guidelines for headache treatment noted considerable variability in how guidelines were developed, but that a number recommended sumatriptan as a first-line agent based apparently mainly on trials against placebo, while noting a lack of head-to-head comparisons among sumatriptan (and other triptans) and older, simpler treatments, like non-steroidal anti-inflammatory drugs (NSAIDs). [Schuurmans A, van Weel C. Pharmacologic treatment of migraine: comparison of guidelines. Can Fam Physician 2005; 51: 838-843. Link here.]

Tfelt-Hansen had written a review of treatment of migraine with a combination of ergotamine and caffeine, which included a single trial comparing that combination to sumatriptan. He noted that the results of this trial appeared on "the homepage of the Swedish Medical Agency," but were not published in a peer-reviewed journal. The trial suggested that sumatriptan was inferior to the combination.

Tfelt-Hansen looked up this trial in the GlaxoSmithKline trial register. Recall that this database was the source of information on unpublished trials of Avandia (rosiglitazone, GSK) that Dr Steven Nissen and colleagues meta-analyzed to suggest that Avandia may produce adverse cardiovascular effects (see post here). Recall also that the registry was created as part of a settlement of a lawsuit by then New York Attorney General Elliot Spitzer that accused GSK of concealing clinical research unfavorable to its drug paroxetine (Paxil, GSK). [Steinbrook R. Registration of clinical trials - voluntary or mandatory? N Engl J Med 2004; 351:1820-1822. Link here .]

Tfelt-Hansen's main finding was that there were six trials comparing sumatriptan to other treatments, including paracetamol (acetaminophen) plus metaclopamide (2 trials), buclizine chloride, paracetamol and codeine (2 trials), ergotamine tartrate, cyclizine HCl, caffeine (1 trial), and ergotamine tartrate plus caffeine (1 trial). In 3 trials, sumatriptan treated patients were not significantly more likely to have relief of their headaches within 2 hours. In 1 trial, there was a non-significant trend favoring sumatriptan. In two trials, sumatriptan was superior to the comparison. In 4 trials, the rate of relief after sumatriptan treatment was 50% or less.

Tfelt-Hansen concluded:

It is easy to understand why these RCTs were never published when sumatriptan was introduced: in only one of the oral trials did more than 50% of patients have headache relief (the primary efficacy measure) after sumatriptan 100 mg for the first attack treated (table), and in the RCT with rectal sumatriptan, the drug was found inferior to ergotamine. These findings would at the time have spoiled the very positive picture of sumatriptan as a new wonder drug for migraine.

Of course, this points out that health care corporations may regard clinical trials more as marketing tools than as science. We have discussed numerous instances in which trials that did not show commercial sponsors' products in a favorable light were suppressed by these sponsors.

Individual trials only at best produce approximations of the truth about the drugs or devices they compare. Trials may be positive due to chance alone when the test or treatment under study actually has no good effects. Post-hoc suppression of "negative" trials therefore may exaggerate the benefits (and safety) of tests or treatments. Physicians and patients who try to uphold the ideals of evidence-based medicine, and base decisions on the best possible evidence can be misled when evidence unfavorable to vested interests is systematically suppressed. Suppression of evidence unfavorable to vested interests may lead to excess use of tests and treatments that really are less beneficial or more risky than the published evidence suggested, and to willingness to pay exaggerated prices for such tests and treatments. Thus suppression of evidence can lead to excess costs and bad outcomes.

Also, as we have said before, suppression of results of clinical research that are unfavorable to the vested interests of research sponsors violates the trust of research subjects. Research subjects are often assured that their participation is for the benefit of science and health care. Suppressing results unfavorable to vested interests distorts science and makes health care more dysfunctional.

Evidence that is purposefully suppressed is by definition hard to find. Nonetheless, we have seen several recent examples in which suppressed evidence was later revealed, and when combined with existing evidence, showed that previously hyped treatments were really not as safe and effective as was thought. In particular, the suppression by various drug companies of evidence unfavorable to new anti-depressants they were marketing has generated some discussion (see post here).

This parade of examples suggests that stronger measures are needed to assure that clinical research is not suppressed due to the vested interests of research sponsors. One seemingly radical, but increasingly plausible approach would ban corporations that sell health care products or services from influencing clinical research done to evaluate those products. Would be health care reformers who really want to improve outcomes, improve access and decrease costs might want to think about how to make the evidence available about the outcomes of health care interventions more honest.