If industry is to learn anything from the three-day Anti-Infective Drugs Advisory Committee meeting on how to correctly use non-inferiority margins in trials, it is to adhere to the panel's recommendation of 10 percent (at least for now).

On the first day of the Nov. 18-20 meeting, the 20-member panel voted unanimously that non-inferiority trials are acceptable for the indication of complicated skin and skin structure infections, effectively ending years of FDA ambiguity on antimicrobial development. Members generally agreed that a 10 percent non-inferiority margin would be suitable in most cases, although it would sometimes depend on the trial setting and patient population.

In the following days, however, the committee followed its own advice quite strictly, voting in favor of companies that used a 10 percent margin and against those that did not.

On the second day the committee voted in favor of approving Theravance's telavancin, which used a 10 percent non-inferiority margin, but voted against Targanta's oritavancin, which used a 10 percent margin in one study and a 15 percent margin in another ¹ Also see "FDA Panel Votes For Theravance’s Telavancin, Against Targanta’s Oritavancin" - Pink Sheet, 19 November, 2008.). When the committee was asked to look at Targanta's trials separately, they voted that the 10 percent non-inferiority margin trial was efficacious and safe, but that the 15 percent non-inferiority margin trial was not.

As if to drill it home, the committee voted on the last day that Arpida's iclaprim, which used a 12.5 percent non-inferiority margin, was not safe or efficacious (² Also see "FDA Panel Says Arpida’s Iclaprim Inferior To Comparator" - Pink Sheet, 20 November, 2008.).

Too Historical To Be Helpful

A major reason for FDA's indecisiveness on how best to evaluate the use of non-inferiority margins in anti-infective trials stems from the fact that no placebo-controlled trials in cSSSI exist from which to compare (³ Also see "Guidance on Anti-Infective Non-Inferiority Margins May Not Be Clear-Cut" - Pink Sheet, 17 November, 2008., p. 27). On top of that, the available historical trials that can be used date back to the early 1900s, a "pre-antibiotic era" that bears little resemblance to today.

FDA, for example, presented findings from three studies conducted in the 1930s and one from 1924.

Not only are patients today often more elderly, diabetic, and immunocompromised, but the bacteria have changed as well. S. pyogenes is a new toxin but has a stable resistance profile and S. aureus is more virulent and resistant.

"Antimicrobials continually lose effectiveness over time," pointed out temporary voting member Thomas Fleming, University of Washington. "We are using comparator trials from the 1900s and comparing them to today's trials. How does that not undermine the interpretability of this?" he asked, stressing the importance of what he termed "biocreeping," or the loss of efficacy for antibiotics over time.

Arpida took a stab at getting around the limitations of historical data by using Pfizer's Zyvox (linezolid), which was approved in 2000 and thus has more recent trial data available. Plus, the firm argued, while vancomycin (the comparator of choice, and that used by Theravance and Targanta) is superior against methicillin-resistant S. aureus, linezolid was a better choice for methicillin-sensitive S. aureus. The choice of comparator, however, contributed to the committee's rejection.

All Roads Lead To 10-15 Percent NI Margin

During the meeting, each sponsor was allotted 20 minutes to justify the non-inferiority margins used in their trials. While each cited ICH E10 guidelines on choice of control group in efficacy trials as factoring into their margin development, they all employed different strategies - though they arrived at strikingly similar margin choices.

Theravance used a meta-analysis of vancomycin (its comparator in pivotal trials) to reach a 10 percent non-inferiority margin. Targanta modeled exposure-response evidence of drug effects from levofloxacin and tigecycline, and vancomycin trial data, to calculate a 10 percent margin for their ARRI trial and a 15 percent margin for their ARRD trial. Arpida chose its non-inferiority margin of 12.5 percent based on other non-inferiority trials whose margins fluctuated between 10 and 15 percent; it also performed a meta-analysis of four linezolid (its comparator drug) trials.

The Infectious Disease Society of America, which was a strong endorser of FDA allowing non-inferiority designs, did a more detailed analysis based on infection types and concluded that a non-inferiority margin of 14 percent could be justified.

"The bottom line is that it seems that you've heard four or five different strategies and they're all imperfect," IDSA's Brad Spellburg said. "We've all tried to do our best in an imperfect world. Now you have to decide," he charged FDA.

It's A Complicated Matter

While members backed non-inferiority trials for cSSSI, they voted against their use in trials for uncomplicated SSSI. The panel voted 19-1 that the trials were not acceptable for uSSSI given that the data evaluated for determining treatment effect in skin infections included data from various types of skin infections, including ulcers, cellulitis and abscesses.

For uSSSI, members said superiority trials would be much more effective. "If there's any chance of doing a superiority trial, you should, because the non-inferiority procedure is so flawed as we've come to understand," said voting member W. Kemper Alston, University of Vermont.

But committee members grappled with defining the differences between cSSSI and uSSSI (a distinction rooted more in regulatory rules than in clinical practice), only coming to the conclusion that when an infection is accompanied by bacteremia, then it is complicated.

"I think part of what we're struggling with here is the definition of complicated and uncomplicated," said temporary voting member Melvin Weinstein. "Complicated I think is more than just deeper infection or something that requires surgery. That's a complicated infection and it's going to require a more aggressive approach, but we need a better definition."

Getting Down To Specifics

When asked whether it is acceptable to justify a non-inferiority margin for cSSSI as a group or whether it should be justified by specific infection types, such as cellulitis, wound infections or abscesses, members agreed that it depends on the trial setting and patient population, but most infections could be grouped together with a 10 percent margin (⁴ The RPM Report, June 2007).

"The margin has to be specific to the trial in which it's being conducted," said voting member Dean Follman, National Institute of Allergy and Infectious Diseases. "It depends on what kind of infections we're looking at."

"The number 10 gets used a lot because we have ten fingers and ten toes," said industry representative and non-voting member John Rex, AstraZeneca. "But let's think about what it means to do a trial and stay at a 10 percent margin."

Even though more than half the members said they were comfortable grouping together cSSSI infections, they stressed that abscesses and diabetic foot patients should be left out of non-inferiority studies, as they would dilute the rigor of the trials.

As with cSSSI and uSSSI, the committee wrestled with how to identify the severity of an abscess. Voting member Matthew Goetz, University of California, Los Angeles suggested defining an advanced abscess as one having the presence of fever greater than 101 degrees, a white blood cell count greater than 15,000, or being in a critical anatomical location, such as the head or neck. Generally, committee members agreed.

Patients with diabetic foot infections should also be studied in separate clinical trials, panelists said. "They have a completely different epidemiology; the microbiology is different; there are circulation issues that have to be addressed," explained temporary voting member Peter Katona, University of California, Los Angeles.

The Future Of Anti-Infectives

Ultimately industry will benefit from the firmness of the advisory committee's decision to allow non-inferiority trials for cSSSI indications, and FDA's long-awaited guidance will be able to take a more precise stance.

There are, however, still issues to consider in going forward, Arpida representative L.J. Wei, Harvard University, said. Industry's future efforts should focus on the retention of efficacy in different scenarios, accounting for the mix of pathogens, proper comparator choices and better surveillance data.

- Lauren Smith (lauren.smith@elsevier.com)

Industry’s Lesson On Non-Inferiority Margins In Anti-Infective Trials: Use 10%

More from Archive

More from Pink Sheet