BOX WO-2FDA Trials for Antimicrobial Drugs: Plugging the Pipeline?

The FDA currently uses a model known as non-inferiority to evaluate virtually all experimental drugs for indications for which treatments already exist.a In a non-inferiority trial, patients are randomly assigned to one of two groups: one group receives a standard comparator drug already on the market; the other group receives the experimental drug. There is no direct comparison to placebo. If the two drugs prove similarly effective, there are two possible interpretations: either both drugs are better than placebo or neither drug is better than placebo. However, if the comparator drug has been demonstrated to be superior to placebo in previously conducted randomized, controlled studies, and the experimental drug is then shown to be noninferior in efficacy to the comparator drug, then the experimental drug can be inferred to be superior to placebo as well. For this reason, the FDA has come to insist that comparator drugs used in noninferiority trials be previously shown to be superior in efficacy to placebo.

This policy has far-reaching implications for the approval of novel antimicrobial drugs, according to Spellberg. Since the first antibiotics predate the advent of randomized, placebo-controlled studies by two decades (unlike most other drug classes, which were subject to placebo-controlled studies from the outset), and these antibiotics are unquestionably effective, it has never been ethical to test antibiotics against a placebo. Similar arguments complicate so-called superiority trials for antibiotics, which determine whether an experimental drug performs better than an approved comparator drug. A new antibiotic would most likely prove superior in patients who are infected with bacteria resistant to the comparator drug, but it would be unethical to enroll such patients in a trial in which some would receive a useless treatment for a resistant infection. “That’s like taking a patient with methicillin-resistant Staphylococcus aureus and giving them a 50 percent chance of being treated with methicillin,” Spellberg said. “You can’t [ethically] do that study.”

These dilemmas could be overcome, Spellberg said, by using analyses performed on early antibiotics in lieu of placebo-controlled randomized studies, so that these drugs could serve as comparators in noninferiority trials of novel antibiotics. He stated that between 1936 and 1950 at least 15 studies of antibacterial agents (sulfonamides or penicillin) were conducted on patients with community-acquired pneumonia (primarily but not exclusively pneumococcal in etiology), then a leading cause of mortality in the United States. Although neither randomized in the modern sense nor placebo controlled, these studies were sufficiently controlled to permit valid comparisons between patients who received antibiotics and those who did not, Spellberg asserted. Every study showed a significant decline in pneumonia mortality among patients given antibiotics (Spellberg et al., 2008b).

Another point of contention regarding the FDA approval process for antibiotics involves the choice of trial endpoints. Spellberg observed that some consider mortality to be the only acceptable endpoint for trials of potentially lifesaving drugs for syndromes such as pneumonia, although such clinical trial endpoints may be problematic. In theory, the clinical effects of antibiotics must be more significant than their effects on mortality, because “dead people don’t clinically respond,” he said. Practically, antibiotic trials using mortality as an endpoint for pneumonia would require huge enrollments, because mortality rates for that disease are less than 5 percent. “That means you’re going to need 5,000 patients in a study to adequately power the study,” he continued. “You have to do 2 of those studies to get an indication, so you need to enroll 10,000 patients into a Phase III program. That will cost $500 million and will take 5 to 10 years to enroll.”

“The critics believe that mortality is the most sensitive endpoint to detect a relatively ineffective drug and that if you don’t use mortality, you increase the risk of approving a relatively ineffective drug,” Spellberg observed. He contended, on the contrary, that relatively ineffective drugs (such as sulfonamides) can have huge mortality benefits, and that noninferiority studies using such clinical endpoints as symptom resolution have demonstrated when drugs are ineffective, as is the case with daptomycin, when partially inactivated by surfactant in the lung, or with tigecycline, when hypermetabolized in patients with ventilator-associated pneumonia. “The FDA simply has to move past radical skeptics and use available data to enable antibiotic noninferiority studies with clinical endpoints,” Spellberg asserted, even if it requires a statutory change recognizing the uniqueness of antibiotics. “Antibiotics are the only class of drugs that loses efficacy over time,” he concluded. “If you do not continually replace them, you will end up not having effective drugs.”

SOURCE: Spellberg et al. (2008b).

a

There is another mechanism. For diseases that are not typically fatal or are slowly progressive, one can do placebo trials with early escape leading to receipt of active therapy at the first sign of disease progression. This is rarely used and never for bacterial infections, because they are not slowly progressive and often fatal if untreated.

There is another mechanism. For diseases that are not typically fatal or are slowly progressive, one can do placebo trials with early escape leading to receipt of active therapy at the first sign of disease progression. This is rarely used and never for bacterial infections, because they are not slowly progressive and often fatal if untreated.

From: Workshop Overview

Cover of Antibiotic Resistance
Antibiotic Resistance: Implications for Global Health and Novel Intervention Strategies: Workshop Summary.
Institute of Medicine (US) Forum on Microbial Threats.
Washington (DC): National Academies Press (US); 2010.
Copyright © 2010, National Academy of Sciences.

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.