Labeling of paclitaxel-coated balloons and stents should be changed to reflect current evidence linking the devices to a late mortality signal in patients with peripheral arterial disease (PAD), the US Food and Drug Administration's (FDA's) Circulatory System Devices Panel advisory committee agrees.
Members disagreed, however, on exactly what those changes should look like, with some favoring brevity and others calling for the totality of evidence to be included, not just the Katsanos et al meta-analysis that first sparked the late mortality concerns. At no time did the panel suggest the devices should be pulled from the market.
"At a very high level, there is a potential mortality excess that's been identified late, but I think what should also be included in there are additional datasets that suggest that isn't present," panelist Joaquin Cigarroa, MD, Knight Cardiovascular Institute, Oregon Health & Science University, Portland, said. "What I don't want is for it to be unidirectional — that there is only a singular dataset that raises a signal."
Panelist John Somberg, MD, countered, "I agree with some of what Dr Cigarroa said, but I don't think the label can be a recapitulation of the panel discussion — physicians aren't going to read it, patients aren't going to understand it."
Instead, he suggested the label say there "may be a late mortality signal, which must be balanced against an early and sustained benefit it terms of pain on walking and potential circulation to the extremities."
Panelist Kevin Kip, PhD, an epidemiologist and statistician at the University of South Florida, Tampa, responded, "When you use the terminology 'may have an increase in late mortality,' to the extent that the public can digest this, you're going to have to really wordsmith. You need to have relative risk and absolute risk because if you put relative risk of 40% to 50%, people are going to run for the hills."
On the opening day of the 2-day panel meeting, an updated FDA meta-analysis showed a 57% increased 5-year mortality risk with paclitaxel-coated devices vs uncoated devices in as-treated patients in three pivotal trials (risk ratio, 1.57; 95% confidence interval, 1.16 - 2.13). New data presented for Medicare beneficiaries and from individual device makers, however, failed to replicate the signal.
During today's session, the FDA provided new data showing that at 5 years, the number needed to treat (NNT) with a paclitaxel-coated device to avoid clinically driven target lesion revascularization was 13 patients, while the number needed to harm (NNH) was 14 to avoid one death.
At the request of the panel, industry representatives quickly calculated their own estimates over the lunch break. For example, Cook Medical reported a NNT of 6.5 patients for the Zilver PTX stent through 5 years and NNH of about 16 among patients who did not crossover and 62 patients among all Zilver-treated patients. For the Stellarex paclitaxel-coated balloon, Philips put the NNT in the mid- to high-teens and the NNH at 71 at 5 years.
Over the course of the 2 days, the panel debated 12 questions, but perhaps the most thorny was a potential dose-response relationship between paclitaxel and mortality. Based on literature reports, paclitaxel is cytotoxic at high concentrations and cytostatic at low concentrations but also may have proinflammatory and proangiogenic effects. In the FDA's analysis of as-treated patients from the three pivotal trials with 5-year mortality follow-up, however, only the LEVANT 2 trial showed a possible trend of higher mortality with increased dose.
The panel concluded the evidence was inconclusive, commenting frequently on the fact that 3% to 26% of data were missing from the trials at 5 years.
The panel spent much of the afternoon debating how to improve the quality of studies going forward to allow a better understanding of the data. Some suggested animal models might be useful if they more closely resemble PAD patients in terms of comorbidities and age than those used in previous preclinical work, but others saw this as a step backward. There were calls for new randomized trials, but others argued that valuable information can be gleaned from ongoing studies — including SWEDEPAD and BASIL-3 — or observational registries.
Panelist Karla Ballman, PhD, chief of biostatistics and epidemiology at Weill Cornell Medical College, New York City, said, "I do agree that registries are really good at picking up adverse events that are quite rare" and "acute events," but "not long-term adverse events that are influenced by prognostic variables."
Panelist Mitchell W. Krucoff, MD, Duke VA Health Care System, Durham, North Carolina, said the "biggest bang for the buck" would be to push trials inside and outside the US to close the gap in terms of missing data, as was done by Medtronic, which in recent months shrunk its percentages from 17% to 3%.
"The biggest bell to ring is that we still discuss things with registries and randomized trials as if they are separate entities," he added, noting that existing registries such as the National Cardiovascular Data Registry, the Society for Vascular Surgery's Vascular Quality Initiative, or the public-private RAPID partnership are already collecting data for 85% to 100% of sites.
"We can put a prospective randomized design into workflow that already exists and then we just have to attach how to do the long-term follow-up and how much granular information vs just mortality numbers really matter," Krucoff said. "And then the incremental burden of what we have to learn across manufacturers becomes a much more manageable conversation. To me that's the 21st century … this is not using lousy registry data retrospectively analyzed."
US FDA Circulatory System Devices Panel advisory committee meeting, Gaithersburg, Maryland. Presented June 19-20, 2019. Agenda
Medscape Medical News © 2019
Cite this: Patrice Wendling. FDA Panel Backs Labeling Changes for Paclitaxel Devices in PAD - Medscape - Jun 20, 2019.