Back grey_arrow_rt.gif
Diabetes Docs Skeptical of FDA Drug Safety Policy, FDA's Demand for Extensive CVD Safety Data on New Drugs
  MedPage Today
Published: May 14, 2010
PRAGUE -- At least some in the diabetes community remain unconvinced that the FDA's demand for extensive cardiovascular safety data on new drugs will ultimately help patients, researchers indicated here.
In a platform presentation at the World Congress on Controversies to Consensus in Diabetes, Obesity, and Hypertension, an AstraZeneca research executive provided an industry perspective on the policy, telling attendees that it would increase development costs and impose delays.
Later, a prominent academic researcher told MedPage Today that, on balance, the negatives of the FDA's stance appear to match the positives, and perhaps outweigh them.
"It won't help us," said Stefano Del Prato, MD, of the University of Pisa in Italy, co-chair of the meeting.
He had invited Boaz Hirshberg, MD, director of clinical research for AstraZeneca in Wilmington, Del., to describe for attendees the impact of the FDA policy.
The policy was published in 2008 in the form of a so-called guidance document, which "represents the agency's current thinking" on a topic. Although guidances are officially nonbinding, companies usually regard them as equivalent to regulation.
This one was issued in the wake of studies suggesting that rosiglitazone (Avandia), at that time the largest-selling branded drug for diabetes, increases the risk of cardiovascular mortality.
The guidance told the drug industry that new submissions for diabetes therapies needed to include far more cardiovascular safety data than had been the standard for these or, indeed, any other class of drug.
Hirshberg -- who emphasized that he was delivering his own views, not those of AstraZeneca -- said that diabetes drug trials had previously focused on glycemic efficacy endpoints, with most patients having relatively early disease, and were either treatment-naive or had failed one previous medication.
Patients at high risk for cardiovascular events were often excluded, he said. Lipids and blood pressure were tracked as safety indicators, but events such as myocardial infarctions or strokes were reported only as adverse events, not predefined objectives.
But the FDA guidance turned that almost completely around, he said.
It specified that companies should include data on patients with advanced diabetes, elderly patients, and those with some degree of renal impairment.
At least two years of cardiovascular safety data were also mandated, with independent adjudication of events. Generally speaking, the risk of "major adverse cardiovascular events" -- cardiovascular death and nonfatal MI and stroke -- should be reported.
The guidance also set forth the limits of what relative-risk estimates of cardiovascular safety would be allowed for an approved drug. If the upper bound of the 95% confidence interval was less than 1.3 (relative to placebo or an approved active comparator), that would be acceptable.
An estimate with an upper bound of less than 1.8 could also pass muster, but the manufacturer would have to sponsor a postmarketing study to determine the risk more precisely.
Any drug with a cardiovascular risk estimate where the upper limit of the 95% confidence interval was higher than 1.8 would be rejected, because it would mean either that the actual relative risk was well above 1.0, or the studies were underpowered to provide an accurate estimate.
In order to keep the upper bound of the interval below 1.8, Hirshberg said, trial data with 80% power to detect a difference in safety risk would need to have 91 events, if the actual relative risk was 1.0. If the actual risk was 1.1, then 129 events would be needed.
Hirshberg also explained that the expected rate of cardiovascular events in the target population is in the range of 0.5% to 1%. Trial data covering 5,000 patient-years -- a large development program under the traditional model -- would therefore produce around 25 to 50 events.
Thus, he said, "even large phase III programs have limited chances of meeting the number of events likely to be required" under the guidance.
Hirshberg counted six ongoing studies of six different drugs -- four for unapproved drugs and two postmarketing studies on approved products -- each seeking at least 18,000 patient-years of data.
Three of the studies aim to collect data on 60,000 patient-years or more.
He told MedPage Today in an interview that the requirement had delayed the introduction of some products "by a year or two."
However, he did not know of any examples of drugs that were killed because their sponsors decided the time and costs were too great. "It's a theoretical concern," he said.
Del Prato said he recognized there were arguments to be made both ways. "It's a difficult choice," he told MedPage Today.
But, he said, one consequence for clinicians was that the company-sponsored trials would be less likely to answer the questions of greatest urgency for them.
At another session here, several hundred attendees were asked to vote on their greatest concerns about diabetes medications. Cardiovascular safety came in last, after issues such as glycemic efficacy and weight gain.
Del Prato said that, in theory, later trials could address such clinical issues, but he doubted they will happen. "Who's going to pay for it?" he asked.
Another diabetologist was overheard to say that companies should ge
  icon paper stack View Older Articles   Back to Top