The brouhaha over Steve Nissen’s meta-analysis and near-fatal attack on GlaxoSmithKline’s Avandia resurfaced this week at the American Diabetes Association meeting in Chicago where an extra-ordinary special two-hour symposium was slotted into the programme.

Dr Nissen and Philip Home, one of the UK’s most respected diabetologists and chairman of the RECORD trial steering committee, were pitched head to head to examine the question of whether or not Dr Nissen was right to generate the panic that has run through the diabetes community, the US Congress, GSK and its shareholders. RECORD is the only long-term prospective cardiovascular outcome trial examining the safety of Avandia (rosiglitazone) and was not due to report until 2009 but Prof Home has been forced into publishing an interim analysis to instil calm.

Attendees flocked to the ADA forum filling a massive auditorium and three overflow rooms while presumably leaving the unfortunate presenters in parallel symposia presenting their science to thin air. The normally dry and dull scientific setting where the highlight is usually a Phase III trial or a set of new guidelines for once gave delegates all the excitement of a courtroom drama.

The event was certainly great theatre with Dr Nissen arguing his case largely from a moral and emotional standpoint while Prof Home used scientific logic and professional conduct issues to trounce the meta-analysis, accusing Dr Nissen of “data snooping on a grand scale” and essentially denouncing the use of the New England Journal of Medicine as a vehicle to spread undue alarmism. The meta-analysis was a highly irregular communication, not mentioning meta-analysis in the title or testing a hypothesis, but lent credibility by a couple of “half-baked editorials” he said. Studies such as Dr Nissen’s are important for raising issues but are a poor basis for decision-making, he added. “Drug regulatory bodies have a difficult task in assessing safety recommendations but they rather than individual physicians are the people who should do so.”

The affair has left the scientific community reeling over just how a safety concern should be brought to the awareness of clinicians, pending a rational evaluation by regulators, without launching a witch hunt for big pharma in the media. Calls for all clinical trial data to be made available have gained new momentum. Prof Home concedes that it is right to generate hypotheses when concerns are felt but this has to be done in a way where regulators can coolly evaluate the risks and benefits of medicines on the basis of science without pressure from the media or Congress. The FDA is reviewing the safety of the drug class on July 30.

Dr Nissen defended his science, saying he included only hard endpoints of heart attacks and cardiovascular deaths in the study. Sir Richard Peto of Oxford University, the pioneer of use of meta-analysis to give power to collective data from small trials, had re-analysed the data and endorsed the conclusions, Dr Nissen stressed. RECORD on the other hand has serious design flaws largely because it is open-label and unblinded and includes softer endpoints, he suggested.

Performing a public service?

Dr Nissen believes he has performed a public service saying the signal of cardiovascular harm he uncovered might never have surfaced if GSK had not been forced by a court case to disclose its full database including numerous small unpublished trials. “The alternative was to leave the diabetic community in the dark” he argued and his conscience could not permit that. He did not call for the withdrawal of Avandia, he emphasises, merely for patients and physicians to consider the collective evidence.

If Dr Nissen blames anyone, it is the FDA for rushing through an approval for Avandia, after the earlier thiazolidinedione troglitazone was rejected, despite “a signal from the very beginning that there was a problem”. This was a mistake, he insists. Drugs can only be approved on the basis of surrogate endpoints (HbA1c levels in diabetes) in the absence of adverse events, he argued. Dr Nissen says he first heard alarmbells on seeing that rosiglitazone raised LDL by almost 20% but says numerical excesses of cardiovascular events in DREAM and ADOPT received little attention. The fact that thiazolidinediones involve gene transcriptions that are not fully understood also generates unease.

Prof Home argues that Dr Nissen used data from trials not designed to examine cardiovascular outcomes to draw conclusions rather than generate hypotheses. The so-called hard endpoints included unadjudicated reports of chest pain, he pointed out. Judgments have been made on the basis of small observational trials and insurance research databases rather than randomised trials where cardiovascular outcomes are the primary endpoint. RECORD includes only adjudicated events but even these can be open to question, he added and on the evidence base, Avandia should continue to have a role in treating diabetes.

Dr Nissen drew applause but also strong criticism from the audience over his stance but the feeling is that the whole debacle has irrevocably damaged the thiazolidinedione class. Data presented on Tuesday highlighted the risks of fragility fractures in women seen with Avandia in the ADOPT trial and these have also been found to occur with Takeda’s Actos (pioglitazone). Takeda’s share price has risen on the back of Actos having an apparent cardiovascular benefit in patients with chronic kidney disease reducing cardiovascular risk by 44% and Actos sales are projected to grow by 20%. However with new drug classes like the dipeptidyl peptidase-4 inhibitors in the wings, and the uncertainty surrounding TZDs and heart failure, physicians say they are likely to become a much less preferred option in future. By Olwen Glynn Owen