Members of the Neurological Devices Panel of the Medical Devices Advisory Committee voted 11-0, with two members abstaining, to recommend approving the NeuroPace RNS System.
The RNS System lowered the rate of seizures by 37.9% during a 3-month blinded evaluation period when turned on, compared with 17.3% when not turned on (P=0.012), results of a randomized, double-blinded, sham-controlled trial of 191 people showed.
The device is surgically implanted under the skin on the skull and records electrocorticographic (ECoG) patterns via electrodes. The device delivers short electrical pulses intended to interrupt the triggers in the brain that cause epileptic seizures.
Physicians can review ECoG recordings and assess the relationship between the device’s detections and reported clinical seizures, and then adjust the RNS System’s electrical pulses accordingly.
While most advisory committee members agreed the efficacy data was robust enough to support approval, Michael Privitera, MD, a neurologist from the University of Cincinnati, noted the large benefit in the sham patients — whose RNS System was never activated — skewed the appearance of a benefit in the treatment arm.
Panelists asked many questions about how the data was interpreted. The manufacturer and FDA agreed that the prespecified form of analysis was flawed, given the wide range of seizure frequency of patients. The results weren’t statistically significant.
However, the two sides presented different ways to read the trial data.
For example, the manufacturer, Mountain View, Calif.-based NeuroPace, conducted a month-by-month analysis and found that by the end of the third month, patients with the device turned on experienced a 41.5% decrease in seizure frequency, compared with 9.4% of those who had the device off.
But those data included two outlying patients from the sham arm. When FDA excluded those two in a post-hoc analysis, the efficacy difference was 40.1% versus 22.9% for the treatment and sham groups, respectively, the agency found.
Some panelists noted the data showed a vast majority of the patients in the treatment arm were unchanged, negating any benefit for the device.
Michael Rogawski, MD, PhD, of the University of California Davis, said little is known about how to optimize the device’s effectiveness and in which populations it works well.
“If the device isn’t usable for a large portion of the people in which it might be implanted, that is a problem,” Rogawski said.
Others noted there is a subpopulation of patients in whom the device works well, but they just don’t know who that is.
Martha Morrell, MD, chief medical officer at NeuroPace and Stanford University neurology professor, equated the RNS System to cardiac devices from years ago, suggesting that exact use and specific treatment methods will be refined over time.
“Are we there right now?” she said. “Of course not.”
The panel also voted 12-0 that the device was safe. While there was much talk about the efficacy data, committee members gave little debate to the RNS System’s safety.
The FDA found the number of adverse events — mild and serious — didn’t differ significantly between the two trial arms. Serious adverse events (SAEs), including brain hemorrhaging and implantation-site infections, occurred in 4.2% of treatment patients and 5.4% of sham patients during the blinded evaluation period.
The FDA is proposing two post-approval studies should the RNS System gain approval, in order to follow up on the device’s safety and efficacy. A post-approval study could also help identify in which subgroups the device proves more effective.
NeuroPace is already offering to extend its ongoing, long-term study — which is at year 7 currently — another 2 years.
The FDA is not required to follow the opinion of its advisory committees, but usually does.
This article was changed to correctly state that NeuroPace included outlying patients in its analysis.
Source: By David Pittman, Washington Correspondent, MedPage Today