SILVER SPRING, Md. -- A bare majority of FDA advisory committee members agreed that cardiovascular outcomes trials (CVOTs) to exclude harm by new type 2 diabetes drugs should continue to be required, but many on the panel called for the process to be streamlined.
The key question was, "Should an unacceptable increase in cardiovascular risk be excluded" -- which so far has always been done through a specially designed trial -- "for all new drugs to improve glycemic control in patients with type 2 diabetes, regardless of the presence or absence of a signal for cardiovascular risk in the development program?" When it came to a vote, 10 of the 19 panel members said yes, with the remaining nine voting no.
Despite the close vote, there was nearly unanimous opinion on room for improvement to reduce the burden on drug developers and, in turn, patients. Many of those voting yes said the existing requirement may be overkill, and most of the "no" voters advocated replacing the current system with more aggressive efforts to uncover cardiovascular safety signals at the premarket stage.
Why CVOTs in the First Place?
The panel discussion hinged on the FDA's 2008 guidance for industry that require randomized, adjudicated data to establish hazard ratios for major adverse cardiovascular events (MACE). In particular, premarket data must be sufficient to exclude a 95% confidence interval upper bound of 1.8 for the MACE hazard ratio. Moreover, if those premarket data show an upper bound higher than 1.3, then postmarketing data must be obtained to exclude the possibility of a 30% excess risk, as indicated by a confidence interval upper bound under 1.3.
The rationale was that while improving glucose control would be expected to improve cardiovascular outcomes, "there have been instances where a signal of cardiovascular risk has been reported," according to FDA briefing documents, particularly pointing to:
- Rosiglitazone (Avandia), which a large meta-analysis suggested increased MI risk by 43% and cardiovascular mortality by 64% compared to placebo and other anti-diabetic agents (albeit since reevaluated)
- Intensive glucose lowering targeting 6% or less hemoglobin A1c, which in the ACCORD trial raised mortality 22% compared with a 7% to 7.9% target
In the decade since the guidance, eight CVOTs showing no excess risk and some finding cardiovascular prevention have been published. However, those trials also turned up some unexpected risks, such as excess amputations.
Industry representatives emphasized the cost of these trials -- $200-$400 million each -- that roughly doubles the cost of a diabetes drug development program and may stifle innovation.
Registries Not Adequate
Two of the three academic representatives from endocrinology and cardiology that spoke at the session emphasized that registry or other observational data simply wouldn't suffice.
Cardiologist Marc Sabatine, MD, MPH, of Harvard and chair of the TIMI Study Group, brought up the examples of antiplatelet therapy that decreased major adverse cardiovascular events and increased bleeding in the randomized TRITON trial while parallel data from the TRANSLATE-ACS registry suggested decreased risk of both endpoints but varied by adjustment technique. Even for the DPP-4 diabetes drugs, the point estimates in observational data for cardiovascular safety were "very different" from that of the randomized data. For other diabetes drug examples he showed, some elements were concordant between types of data but others weren't. Confounding ultimately undermines all this observational data, he argued.
"Adequately powered and randomized CVOTs of individual agents should continue. There is no substitute," agreed endocrinologist Jennifer Green, MD, of the Duke Clinical Research Institute in Durham, North Carolina.
On the opposite side, endocrinologist Robert Ratner, MD, of Georgetown University in Washington, D.C., argued that there is a preponderance of data showing cardiovascular safety to now move into a risk-stratified approach.
"If all we're doing is ensuring safety, then what is the minimal amount necessary to feel safe? Clearly FDA does that with every other safety indication within the diabetes field, they do it for all the other fields as well.... Diabetes is the only one where there's a requisite study to show cardiovascular safety," he noted.
He and the industry representatives argued for scrapping mandatory CVOTs and instead only demanding outcomes trials whenever a safety signal is noted in phase II or III or if a company wants to pursue a CV label indication.
However, panelist Thomas Wang, MD, of Vanderbilt Heart & Vascular Institute in Nashville, Tennessee, pointed to the low number of cardiovascular events accrued in phase II and III diabetes trials before 2008.
"We did not get useful information from those studies. If we were to go to the pre-2008 situation of relying on those studies to guide us as to whether to pursue follow-up studies, they would not have given adequate guidance because they were hopelessly underpowered," he said.
Room for Improvement
In the end, the votes split nearly down the middle but with broad consensus that adjustments are needed. Those fell into roughly two camps:
In explaining his vote for keeping mandatory CV safety trial requirements for all diabetes drugs, Kenneth Burman, MD, an endocrinologist at MedStar Washington Hospital Center in Washington, D.C., expressed an oft-repeated view:
"A standard phase II/III trial for efficacy and safety is not generally adequate to detect sensitivity and specificity and to detect the relevant cardiovascular signal. I think the next issue is whether a CVOT can be appropriately modified to improve efficiency, a so-called streamlined CVOT.... There are multiple suggestions that make sense, including increase of high volume centers, use of mobile devices, adoption of specific endpoints and a risk-adjusted monitoring schedule to name a few. I agree with these appropriate modifications which will hopefully increase efficiency and decrease cost without detracting from the high standards expected from such a trial."
Brendan Everett, MD, MPH, a cardiologist at Brigham and Women's Hospital in Boston, also voted yes but articulated a "compromise" position cited by many others.
He suggested a one-step process with more robust phase II and III data with more patients and longer follow-up to collect more endpoints that could exclude a stricter boundary for exclusion of cardiovascular harm to a "reasonable degree of satisfaction." A CVOT would be triggered if there were concerns raised by that preapproval data. It would take more resources than current preapproval approach but less than requiring all to have a CVOT, Everett noted.
One supporter of the "Everett approach" was Michael Blaha, MD, MPH, of the Johns Hopkins Ciccarone Center for the Prevention of Heart Disease in Baltimore, who said in explaining his no vote:
"Although we've learned a ton from the cardiovascular CVOTs that fundamentally changed my cardiology practice, for sure I use these drugs for cardiology patients in my practice, my vote was informed by the notion that no study to date has shown increased MACE, has shown an increased signal for cardiovascular safety, making it hard to mandate a CVOT for all new drugs in my view."
The landscape has changed -- guidelines from the American Diabetes Association now call for prioritizing drugs that have shown a cardiovascular benefit and companies will feel more pressure from that, he said. "I favor a simple but flexible one-step approach to premarket approval without a mandatory CVOT requirement. ... and, of course, a CVOT if there is a safety signal all of us agree on."
Another no-voter, Cecilia Low Wang, MD, an endocrinologist at the University of Colorado in Aurora, argued for not returning to pre-2008 regulation but aiming for a middle ground. She called for premarket data looking at signals of CV risk beyond just MACE, tightening the confidence interval boundary to 1.5 and possibly a point estimate of 1.2, with a minimum number of events in randomized trials preapproval but then perhaps long-term CV safety studies using registry and observational data.
"We would all like to see it simpler if possible," summarized panel chairperson Peter Wilson, MD, an epidemiologist at Emory University in Atlanta.