Junkfood Science: Can we trust the published results of clinical trials?

November 14, 2008

Can we trust the published results of clinical trials?

News coverage of the long-awaited session of this week’s American Heart Association’s scientific meetings that had promised to constructively address uncomfortable issues surrounding recent statin clinical trials is virtually nowhere to be found. There was no AHA press release and only one reporter even covered it at all.

The session — “What Is the Emperor Wearing? Unbiased Evidence from Clinical Trials” — was going to try and get to the bottom of the increased cancer deaths seen in the SEAS trial and learn the positions of both academic investigators and the industry sponsors of statin trials. More importantly, for researchers, it was going to explore constructive suggestions for maintaining the integrity of clinical trials when conducted in partnership with pharmaceutical companies; and for healthcare professionals and the public, it was going to examine hidden agendas and biases of published research and what we can trust.

This session was in response to issues reviewed in depth here. Although the SEAS trial had shown the statin Vytorin to be of no additional benefit in its primary clinical outcomes, reducing cardiovascular events or all-cause mortality, it had raised concerns about heightened cancer risks. It also brought to a head growing trepidation over biases in industry-sponsored trials, the use of surrogate endpoints versus clinical outcomes, trials unblinded or stopped prematurely, and the frustrations of healthcare professionals unable to make sound clinical decisions with problematic and incomplete data.

Staff writer Charles Bankhead with MedPage Today appears to be the only reporter who wrote on this session from New Orleans. He said that in the wake of the SEAS incident, Dr. Robert Califf, M.D., of Duke University in Durham, NC, had proposed the establishment of universally-recognized standards of conduct to ensure trial outcomes reflect the most objective, unbiased evidence. Mr. Bankhead reported:

…The first step in developing those standards should be to bolster the independence of data monitoring committees, he said. "Members [of the committees] should have no financial interests in sponsors or competing sponsors in which personal wealth is tied to the fate of the company," said Dr. Califf. "They must have access to the raw database and the ability to analyze it independently of sponsor employees… NIH and academia need to step up," he said.

The recommendations proposed by Dr. Califf included:

● A balanced executive committee that includes representatives of the sponsor but whose majority consists of investigators. Only the sponsor's representatives should have any ties to the sponsor.

● Transparent reporting of financial relationships.

● Identification of conflicts of interest that are not directly financial in nature (deeply held professional beliefs about a scientific issue, working for competitors, previous request for funding turned down by sponsor).

● Commitment to publish.

● Appropriate airing of differences of opinion (not in the media).

● Prohibition of direct-to-consumer advertising until definitive data are available.

● Increased public funding of clinical trials to preempt claims of bias.

Mr. Bankhead went on to review the incidences surrounding the SEAS controversy and the discussions that followed at the meeting. An Oxford reviewer said that the 50% increase in overall cancer incidences in SEAS was ruled out by the interim data from SHARP and IMPROVE-IT and that the incidences failed to satisfy the principle criteria for a causal relationship. Dr. Allen Taylor, M.D., of the Uniformed Services University of the Health Sciences in Rockville, Maryland, was reported to have pointed out that the upcoming SHARP and IMPROVE-IT trails won’t conclusively answer questions about cancer risks, because they weren’t designed to address them. All three trials though, he said, have the effect of creating a negative bias towards cancer. “Despite the negative bias, a cancer signal has emerged from SHARP and IMPROVE-IT during relatively brief follow-up, one year in the case of IMPROVE-IT,” Dr. Taylor observed. “The confidence intervals suggest as much as an 84% increased risk of cancer death,” he said. In contrast, he noted that compared with other recent trials, the benefits appear modest. As Bankhead reported, the FDA is still reviewing the SEAS trial.

The bottom line, doctors and consumers are left in a conundrum for now. “The net benefit remains unmeasured, and the absence of harm is uncertain.”

Bookmark and Share