Medical Device Daily Washington Editor

GAITHERSBURG, Maryland — There were a number of factors that played into the negative outcome in Tuesday's hearing of the orthopedic and rehabilitative devices panel, but questions about the efficacy of the Oxiplex anti-adhesive gel tanked the application in the eyes of the panelists, who voted 5-2 against approvability.

The application may have suffered somewhat from having been in the FDA pipeline for several years, with the sponsor, FzioMed (San Luis Obispo, California), winning the investigational device exemption in 2001. Another factor that played into the failure of the application was a change of the statistical analysis from a univariate to a multivariate analysis in 2006, but FDA and Fzio also disagreed on whether data from patients with extended follow-ups should have been permitted in the final number-crunching.

The unmentioned 800-lb. gorilla in the room may have been the Adcon-L, which the panel recommended for approval in December 1997. Due to regulatory problems and FDA tests that indicated that the Adcon-L did not reduce post-surgical spinal scarring as claimed, Gliatech (Cleveland) pulled the device from the market and subsequently filed for bankruptcy.

Other problems with the firm included the fact that Derrick McKinley, Gliatech's medical director, pleaded guilty in 2005 to insider trading charges.

In the first open public hearing, Reginald Davis, MD, a neurosurgeon in private practice in the Baltimore area, told the panel the Oxiplex covers an unmet need. "A significant number of these patients, even those with successful outcomes, report residual pain," he said. "Any reduction in this pain would be significant improvement," Davis said, adding, "Oxiplex is an intuitive and logical solution to an unmet need."

In the afternoon public hearing, Patrick Fransen, MD, a neurosurgeon at Brussels Neurosurgery Center (Brussels), gave the same message to the panel, but added that the device has been available in Europe since 2001 and has been used more than 100,000 times since then, with no reports of adverse events that were attributed to the device.

John Krelle, president/CEO of FzioMed, said the two polymers from which the Oxiplex gel is made, carboxymethylcellulose and polyethylene oxide, are well characterized and that multivariate analyses of the clinical trial data were "unconditionally approved by FDA." These analyses "allowed identification of an important subgroup," patients with baseline leg and back pain.

The flashpoint between FDA and the sponsor over the number of patients included in the final analysis hinged on the fact that only 286 of the 335 original enrollees finished their follow-up within the specified six months. FDA's analysis was of all who completed the follow-up, including those whose follow-ups took a year.

In rebuttal, Ron Ehmsen, the firm's VP for clinical and regulatory affairs, said "attributing values collected beyond 28 weeks to a six-month value is prone to error." Ehmsen backed up the assertion with a citation from the 2006 edition of the European Spine Journal which, according to Ehmsen's presentation, concluded: "longer follow-ups may introduce error that influence patients' ratings of outcome, especially if based on self-ratings of current pain."

Kevin Lee, MD, the lead reviewer for the Oxiplex, observed that the Oxiplex pilot study employed a sample size that "may be too small to ... detect statistically significant differences," but nonetheless noted a higher incidence of leg and back pain in the pilot Oxiplex group. Lee also noted that FzioMed proposed the change to the statistical plan for the pivotal study at the date of a December 2006 interim analysis which took place after one-third of the subjects in each arm had completed their follow-ups to include lower back pain to the primary endpoint of reduced leg pain.

Jack Zhou, FDA's statistical reviewer for the application, said FzioMed's selection of a covariate analysis model was "unusual" and "a complex model that is difficult to interpret." He remarked that a univariate analysis of the pivotal study data indicated a treatment effect that did not reach statistical significance, and characterized Fzio's multivariate analysis as "post-hoc" and as "exploratory" in nature despite the agency's agreement to the change of analytical plan.

Earlier in the day, Richard Chiacchierini, PhD, the principal of R.P. Chiacchierini & Associates (Rockville, Maryland), who reviewed the company's analysis, said a multivariate analysis is "the approved and appropriate method to interpret this clinically complex condition." Chiacchierini acknowledged outcome differences between study sites, but said "there was no site-by-treatment interaction" and that any "differences between Oxiplex and control subjects were independent of sites."

The panel's discussions indicated no consensus on the use of both leg and back pain as endpoints, but after a discussion on variation of outcomes by study site, the panel chairman, Jay Mabrey, MD, of Baylor University Medical Center (Dallas), said "the panel generally believes that site variability ... is generally due to small sample size" per site.

Most of the panelists were not in favor of the use of multivariate analyses, although panelist Paul McCormick, MD, a neurosurgeon at Columbia University Hospital (New York) said "it was clearly appropriate that the analysis was done," in part because of the heterogeneity of the study population. He also said that this heterogeneity is not uncommon in clinical trials. However, McCormick said he was concerned "because it was a post-hoc multivariate analysis."

As for FDA's question of whether the device is safe, several said they could not answer, but the industry representative, Elizabeth George of Philips Medical Systems (Andover, Massachusetts), argued that this was not a tenable response given that the data indicated that no adverse events were attributed to the device.

Several panelists said they saw the device as safe, but that they could not be confident that the benefits outweighed the risks, and Mabrey concluded that "the panel seems evenly divided" between those who look at safety in isolation and those who see safety as tied to efficacy. Also, none of the voting members indicated a belief that the Oxiplex was effective other than in small subgroups of the study population.