New diabetes drugs should be subjected to studies to rule out
cardiovascular risk, the FDA's Endocrinologic and Metabolic Drugs
Advisory Committee decided at a meeting this week. But the specifics
of what exactly those studies should involve were left open. Such
safety information should also be sought for drugs already on the
market, if possible, it was suggested.

The committee met on July 1 and 2 to discuss whether long-term
cardiovascular safety data should be required for new and existing
diabetes drugs. This discussion was deemed necessary because
concerns about cardiovascular safety have been raised about several
drugs in this field (rosiglitazone [Avandia, GlaxoSmithKline] and
muraglitazar being recent examples), compounded by the recent
results of the ACCORD trial, showing an increased mortality in
patients undergoing intensive glucose lowering with multiple
diabetes drugs.

FDA official Dr Mary Parks explained to the press after the
meeting: "There have been calls for cardiovascular-outcome trials on
diabetes drugs. But it was unclear to the FDA what exactly this
meant." For example, she said the agency wanted to clarify whether
these trials would need to rule out harm or to show cardiovascular
benefit and whether the data would need to be in before or after
approval. These questions and other specifics of how such trials
should be conducted were therefore discussed by the advisory
committee over the past two days.

Vote of 14-2 in Favor of Long-Term Trial

After a lengthy discussion on how the question to be voted on should
be worded, the committee eventually settled on: "We are assuming
that a drug/biologic with a safety signal in phase 2/3 trials will
have to have an additional safety trial. But for those drugs without
such a signal, should there be a requirement to conduct a long-term
trial or provide equivalent evidence to rule out an unacceptable
cardiovascular risk?" And they voted 14-2 in favor of this concept,
with two abstentions.

The committee grappled with many other questions regarding specific
issues such as whether these cardiovascular assessments should occur
pre- or postapproval and design considerations such as what
magnitude of increased cardiovascular risk should be ruled out, what
the primary end point should be, what type of patient population
should be enrolled, which treatment comparator(s) should be used,
how deteriorating glycemic control should be handled, and how other
cardiovascular risk factors should be managed.

While no firm consensus was reached on many of these issues, there
was agreement on some points. Acting chair of the committee, Dr
Kenneth Burman (Washington Hospital Center, Washington, DC),
said: "I think we have consensus that we need either one large trial
or a series of smaller trials to give information on cardiovascular
risk. We don't need to show cardiovascular benefit given that these
drugs already have a benefit in lowering blood glucose levels, which
is known to be associated with a reduction in microvascular
complications. But we do need to show lack of hazard." He added: "In
my view, it is not possible to specify one size or duration of trial
for all drugs. This would depend on what adverse events had already
shown up in the development program and what other benefits the new
drug was offering."

Virtually everyone on the panel agreed that the primary end point
should be a composite of hard end points (preferably CV
death/MI/stroke). Many members suggested a duration of at least
three to five years would be required and that the population
involved should be mainly diabetics at high risk of heart disease to
ensure enough events.

There was also consensus that such a trial should evaluate the new
drug vs an active comparator, but this was problematic in that no
diabetes drug currently on the market was known definitely to be
free of cardiovascular risk. Most committee members suggested
metformin as the best comparator to start with, as this drug had the
most evidence of cardiovascular safety/benefit.

It was agreed that deteriorating glycemic control and other
cardiovascular risk factors (such as cholesterol and blood pressure)
had to be managed in both arms of the trial, although they were
confounders. Some panel members suggested that the goals for these
risk factors should be stated but it should not be insisted upon
that they be met, as there should be some room to show effects of
the new drug on these risk factors. On this issue, panel member Dr
Marvin Konstam (Tufts-New England Medical Center, Boston, MA)
said: "We need to try to ensure that if a drug has a favorable
effect on LDL we want that to show up, and that won't happen if all
patients have their LDL treated to target."

There was also extensive discussion on whether the cardiovascular-
outcome data needed to be available pre- or postapproval and how
this could be achieved in the most timely and least burdensome
manner, so not to delay too much the time taken for a drug to reach
the market.

Preapproval Screening Trial Followed by Larger Trial Postapproval?

In a presentation on the first day of the meeting, Dr Steven Nissen
(Cleveland Clinic, OH) suggested that two separate trials be
conducted--a preapproval screening trial to rule out a large
cardiovascular risk, followed by a large randomized trial to
estimate cardiovascular risk more accurately, which must be under
way as a condition of approval but could be completed after
approval. "This would provide more reliable preapproval data but
only delay approval by a few months and would also ensure that solid
data would be generated for all new drugs," Nissen said.

He added that it was essential that the large trial was under way
before approval was granted, as once a drug was approved, promises
by sponsors to do further trials are often not kept. "Only 14% of
phase 4 commitments actually get done. That is one of the things
that went wrong with rosiglitazone--they never conducted the trial
that they had committed to at the time of approval. I think we have
to learn from those mistakes."

Dr Robert Califf (Duke Clinical Research Institute, Durham, NC) then
gave a talk on the best way to perform large randomized trials to
get answers on cardiovascular safety as quickly and inexpensively as
possible, basically emphasizing the need to focus on major adverse
events and not to get bogged down on minor adverse events and
bureaucracy.

The second day of the meeting was taken up discussing Nissen's idea
or variations of it. The suggestion was raised that one large trial
could be conducted, which would be started preapproval, and that an
interim analysis after a specific number of events had been reported
could be used to gain an idea of safety for approval, with the trial
continuing after approval to give a better safety estimate. Many
panel members seemed to like this idea. Some voiced the suggestion
that the trial may be able to be stopped if there were no obvious
risk at further interim analyses.

But Nissen told heartwire he did not like the idea of doing one
study and taking a look at the interim results for approval. "Once
you take a look you have affected the behavior in the trial. And my
own view is that is not desirable," he said.

Some panel members also suggested that if the preapproval screening
trial showed a benefit, a larger trial may not be necessary, but
others disagreed, pointing out that a large trial was the only way
to know for sure what the cardiovascular safety profile was. Nissen
was also adamant that a long-term trial had to be conducted, even if
the preapproval screening trial was suggesting a cardiovascular
benefit. "If the preapproval trial suggested benefit, that trial
will still have very wide confidence intervals because it will not
be big enough to show a definitive result. It will simply reduce the
uncertainty. And if you really want to do this correctly, you have
to nail down the full answer in a large outcomes trial," he
commented to heartwire after the meeting.

At What Level Will the Bar Be Set?

A long time at the meeting was spent discussing statistical issues
of how many events would be needed in a preapproval trial to rule
out a large hazard and what the level of hazard ruled out at this
stage needs to be set at. No definite consensus was reached on these
issues, although one suggestion was for a screening trial to have
125 events, which could rule out an upper confidence interval of
harm of 1.8 (corresponding to a point estimate of 1.26).

Konstam commented: "I don't feel we as a panel should specify any
specific upper boundary. I would rather see a clinical assessment of
risk. Whatever the statistical bound of a trial is, my
interpretation of that trial will be determined by other things--the
point estimate, the number of events, whether there are other
signals of concern, the incremental value of the drug. If the drug
can achieve better glycemic control with reduced risk of
hypoglycemia, I may be more accepting of a higher upper boundary on
cardiovascular risk. And this may be different for a completely new
class of drug rather than another drug in an established class."

What Now?

FDA official Dr John Jenkins told the postmeeting press conference
that the agency now needs to digest the committee's discussions. "We
need to go back internally and think about the advice in more
detail. We have clearly heard that we need better [information on
the] cardiovascular safety of diabetes drugs before approval. The
FDA already has to decide whether the safety database of a drug in
the approval application is sufficient to rule out harm, but the
committee is now recommending that we should have more data to do
this and that an unacceptably high risk needs to be ruled out before
approval. This approach could also well be applied to drugs in other
classes such as the [nonsteroidal anti-inflammatory drugs] NSAIDs."

Old Drugs Also to Be Scrutinized

Parks noted that the FDA also intends to reexamine data on diabetes
drugs already on the market. "We will have to go back and look at
each individual drug and the amount of evidence available to see if
it is adequate or not to ensure cardiovascular safety. If not, we
may have to ask for more data," she told the press conference.