CDRH Director Dan Schultz's decision to override device and drug center review staff to approve Cyberonics' VNS Therapy neurostimulation device for treatment-resistant depression has scant precedent in recent history. All the same, he appears to have followed established FDA procedures.

A report released Feb. 16 by the Senate Finance Committee scrutinizes Schultz's decision to overrule more than 20 FDA staff members who concluded that data from Cyberonics did not meet standards for reasonable assurance of safety and effectiveness (¹ Also see "Senate Report Scrutinizes Schultz’ Role In Cyberonics’ Device Approval" - Medtech Insight, 20 February, 2006., p. 11).

In a June 12, 2005, memorandum released by the Finance Committee, Schultz acknowledges that Cyberonics' 12-week randomized controlled trial failed to reach statistical significance, but ultimately concludes that "the effectiveness of this therapy, perhaps unlike other treatments for depression, does improve over time."

Specifically, the director cites non-randomized trend data showing that the percentage of patients responding to the treatment after two years nearly doubled from the percentage that responded during the three-month randomized controlled trial.

The firm submitted the two-year follow-up data after initially receiving a "non-approvable" letter from FDA in August 2004. After appealing the decision to progressively higher levels of agency management, the firm bagged an "approvable" letter for treatment-resistant depression (TRD) in February 2005, based on Schultz's recommendations (² Also see "Cyberonics Depression Approval Contingent On Dosing Trial, Registry" - Medtech Insight, 7 February, 2005., p. 3).

The implantable neurostimulator, which was already on the market to treat epilepsy, was approved in a PMA supplement in July for use in patients with depression that have failed to respond to at least four other medical treatments.

While Schultz's conclusion differed from that of FDA staffers, the director's decision was consistent with FDA's Neurological Devices Panel, which recommended approval with conditions for the TRD indication following a June 2004 meeting (³ Also see "Cyberonics’ VNS Therapy Approvable As Last-Resort Depression Fix – Panel" - Medtech Insight, 21 June, 2004., p. 3).

However, the Senate report, based on an investigation that began in February, criticizes FDA for not representing the level of "scientific dissent" over the VNS Therapy in both the formal labeling for the product and the consumer information posted on FDA's website. According to the report, the product's labeling "does not make clear the director's own conclusions regarding the sponsor's short-term clinical study."

Director Override Is Rare, Despite Frequent Appeals

CDRH Office of Device Evaluation Director Donna-Bea Tillman observed that there is an established process by which the center director can be called on to make a final device approval decision.

"We have a whole appeals process where any decision made at any given level can be appealed by the company to the appropriate person at the next level," she explained. If the company is dissatisfied with Tillman's decision at the office level, then it goes to the center director. It is not atypical for an appeal to reach that level, she said in an interview.

Other sources suggest, however, that it is uncommon for a decision to be overturned after an application climbs so high in the review chain.

The Senate report noted that, in interviews with the Finance Committee, FDA management staff "could recall only one other instance where the center director made the final decision regarding a device's approvability in the past decade."

"Usually, when you have PMAs, they are either approved at the division level or, in the case of new, novel, first-of-a-kind products, they are approved at the office level," concurred Phil Phillips of Becker & Associates Consulting.

Phillips was the deputy director of science and regulatory policy at CDRH's Office of Device Evaluation up until February 2005. A final decision by the center director is "very, very unusual," he said.

According to the Finance Committee report, during the review of the VNS device, Tillman joined several Division of General, Restorative & Neurological Devices staffers - who had primary jurisdiction over the VNS Therapy review - in signing the review team's Jan. 6 memo recommending against approval for the device after it became clear that Schultz planned to override the decision.

The ODE director, who was not mentioned by name in the report, "wanted to make clear her concurrence with the rest of the review team that the VNS Therapy System for TRD should not be approved," the report said.

The staff was unconvinced that the data, including the two-year follow-up submission, showed that benefits outweighed risks.

Tensions Between Device, Drug Data Standards

At the center of the controversy is a debate over whether randomized control data is necessary to prove safety and effectiveness. In Cyberonic's TRD study, all participants were implanted with the neurostimulator device, but half were randomly assigned to have the device activated; the other half were not.

After 12 weeks, investigators could observe no statistically significant differences between the two groups. At that point, investigators turned on all the inactive devices, removing the study's control group. The sponsor then used a population of 124 patients from a different study of depression patients as a comparison group.

Among the most vocal opponents of relying on this longer-term "open-label" data were reviewers from the FDA's drug center, who were consulted during the review, the Senate report suggests.

A CDER medical reviewer assigned to examine data from Cyberonics' application wrote to his colleagues that he was "disturbed that VNS might actually get an approval for TRD... As an MD with an interest in science, it seems to me such an approval would be akin to approving an experimental product," the reviewer wrote in a Jan. 12 e-mail, reprinted in the report.

The report also cites an e-mail from a CDER "psychopharmacology expert" who reviewed some initial application materials. The CDER staffer said that if Cyberonics had submitted similar materials to the drug center, they would not have been filed as an application.

"The long-term claims [of efficacy by the sponsor] are based on open-label data," the expert writes. "We do not allow labeling claims based on open-label data. We do not allow labeling claims based [on] open-label studies that rely on historical controls in depression."

Conversely, in his override memo, Schultz said: "I believe that medical device statute, regulation and guidance all support the concept that data from a variety of sources and methods of analysis can and should be considered."

This difference of opinion can be explained in part by different data thresholds accepted at CDRH and CDER. Drug reviewers almost exclusively rely on large randomized, controlled data. But the short lifespan of many devices often necessitates more streamlined approaches, which are sanctioned by slightly more flexible regulations compared with those for drugs.

According to Tillman, the device center sometimes needs to rely on the expertise of other centers for assistance in product reviews. In this case, the drug center has experience with approving antidepressant medicines, while the device staffers have little expertise in clinical depression.

However, CDER reviewers often do not understand the data necessary to demonstrate a device's safety and effectiveness, she suggested. "The challenge...is making sure those people understand how devices and drugs are different," she said.

Committee Wants More Transparent Labeling

The Finance Committee does not call for VNS Therapy to be removed from the market, nor does it say outright that Schultz made the wrong decision. The report does, however, suggest that CDRH should take a lesson from the drug center's recently enunciated approach to labeling.

The report highlights language from a final drug-and-biologic physician labeling rule, issued in January, which stresses the fact that healthcare practitioners should be able to rely on labeling for "authoritative risk information and...health care practitioners should not be required to convey risk information to patients that is not included in the labeling."

The committee asserts that FDA did not fulfill its obligation to healthcare providers and patients in the Cyberonics case because it failed to disclose all the "relevant findings and conclusions" about the device by not making prominent enough the fact that the randomized controlled data was inconclusive.

In response to the committee's investigation, CDRH apparently included more language on its consumer information website, noting that the 12-week trial data were inconclusive. But the committee suggests the information should be expanded to include the full conclusions of the review team, office and division management staff who favored not approving the device.

Schultz’ Controversial Approval Of Cyberonics Device Has Little Precedent

More from Archive

More from Medtech Insight