Medical Device Daily Washington Editor
GAITHERSBURG, Maryland — Advisory committees often show an aversion to surgical outcomes as surrogates for clinical endpoints, but the outcome of Thursday's meeting of the neurological devices advisory committee swam in the other direction, resulting in a vote of approvability for a new indication for the DuraSeal Xact sealing system, made by Covidien (Mansfield, Massachusetts).
Covidien won the initial PMA for the bioresorbable hydrogel device in April 2005 as an adjunct to suture for sealing the dura mater in cranial applications, but the bid to expand the indication to use in spinal surgery required an advisory committee hearing because the DuraSeal would be the first such approved device.
The pivotal trial for the device rendered a safety profile that was not statistically worse than the current standard of care, but the numbers trended somewhat worse on several measures by the last follow-up visit at 90 days. On the other hand, the DuraSeal provided full sealing against CSF leaks in all patients after two attempts, whereas the current standard of care – which apparently is a poorly-defined set of other off-label uses of devices – results in full closure of cerebrospinal (CSF) leaks in all patients only after six attempts, a fact which helped win the day for the device.
Ryan Kretzer, MD, a medical officer at FDA, said that follow-up in the pivotal study took place at 30 and 90 days and noted that the study enrolled 102 in the study arm and 56 in the control arm. "The results in intraoperative outcomes," he said, "showed that DuraSeal was superior to the standard of care." However, he noted that at 90 days, "DuraSeal had a 7.8% CSF leak rate compared to 5.6% for controls." However, the trial did not enroll large enough numbers for this to constitute statistical significance.
"FDA's main issue is that the primary endpoint did not correlate" with 90-day data, Kretzer said, adding that he has "a problem with the surgical endpoint when I see the 90-day data as much more relevant."
The pivotal trial randomized patients to either the standard of care or the device only if the first suture failed to prevent leaks of CSF, but one of the unique elements of the device is that is is only needed in spinal applications when a surgery on the vertebrae accidentally damages the dura of the spinal cord.
Neal Wright, MD, of the Washington University School of Medicine (St. Louis), gave the sponsor's view of the trial, stating that "all untoward events were captured" including events not related to the device. Nervous system disorder events numbered nine in the study arm, but the independent adjudicating authority "determined that none of these were device-related."
"It's clear to me that these were related to the surgical procedure," not the device, Wright said.
Of the three patients who got the DuraSeal and nonetheless developed a fistula, one should have been disqualified, Wright said, because he went mountain biking within weeks of the procedure. "Stressing that with routine activities, let alone mountain biking," is extraordinary and should not be blamed on the device, Wright said.
Wright also cited extenuating circumstances for the other two patients, adding, "I'm not concerned about the 90-day leak because I think these three patients explain it." He also noted that surgeons will "use any and all methods to leave the OR with a sealed dura," remarking that the DuraSeal fixed 92% of leaks on the first application and the remaining 8% on the second pass. He observed that surgeons needed as many as six attempts to seal up the dura in patients in the control arm. Wright also said that the standard of care, which includes the use of fibrin glue, can add as much as 45 minutes to the procedure.
During the open public hearing, Peter Lurie, deputy director of the health research group at Public Citizen (Washington), attempted to dissuade the panel from recommending approvability. He said that the standard for a medical device requires that the device "provides clinically significant result," arguing that this is not what is seen in the pivotal trial. He said he was "concerned that the adequacy of the surrogate endpoint is taken" on faith. "I don't hear any conversation about the statistical validity" of the surrogate, he said.
As for the statistical analysis of clinical endpoints, Lurie made the case that "you would have to have had a tripling of the leakage rate" of the device "before you reach statistical significance."
As to claims by the sponsor that the device trims surgical times, Lurie argued, "they're unable to marshal any data" supporting the claim. Regarding concerns that current methods of enclosure – which often rely on tissue derived from human and bovine sources – may introduce deadly pathogens into the patient's body, he said that the incidence of any of infection by variants of spongiform encephalopathies are "vanishingly rare. They are so small that they pale in comparison to the safety and efficacy issue," he said.
The panel voted 4-0 in favor of approvability, but made no recommendation as to whether the trial should include a control group. The panel also opted not to suggest that the sponsor be required to provide training to use the gel, and that the PAS track patients to 90 days.
Danica Marinac-Dabic,PhD, chief of the Office of Statistics and Biometrics at FDA's Center for Devices and Radiological Health, told Medical Device Daily after the meeting that when a device has no clearly defined comparator and its use cannot be reliably predicted, a registry can be used to conduct post-approval studies, even though she declined to refer specifically to the DuraSeal.
"It's not micromanagement to give us some direction in regard to this," she said. Dabic declined to get into specifics as to how far back in time a sponsor could go to draw historical data for a Bayesian trial for this purpose.