Clinical trials for infectious diseases are somewhat unique in the world of drug development because, in most cases, a drug’s activity cannot be established through comparison to placebo.
There are a few exceptions: conditions that are uncomfortable and unpleasant, but not deadly, such as traveler’s diarrhea, sinusitis, bronchitis or cystitis, can generally be studied in a placebo-controlled setting where the aim of the new drug is to resolve symptoms. The same is true for conditions in which there is no acceptable existing treatment, such as a novel viral disease.
But in most cases, placebo-controlled studies of anti-infectives are unethical because these diseases carry measurable mortality that can be lessened, if not averted, through the use of existing drugs. For example, in pneumonia, it is not acceptable to perform a placebo-controlled trial when a variety of antibiotics are available that could be used to reduce the risk of death.1
Yet both drug developers and regulators recognize that the continuous development of new anti-infectives is essential to combat the emergence of drug-resistant bacterial strains. The system that has evolved to ensure that new anti-infectives are more effective than placebo, without actually comparing them to a placebo, is the non-inferiority trial.
Non-inferiority trials use an approved drug as a comparator, with the goal of proving within a certain statistical probability, that the new drug is not significantly worse than the approved drug.
For example, in many cases, a new drug may be considered to be not inferior to a current drug if the new drug is no more than 10% “worse” than its comparator.2 Of course, the new drug likely brings other benefits to the table, such as a better safety profile or the ability to overcome bacterial strains that have become resistant to the older comparator drug.
A problem with non-inferiority trials, however, is that years and years of comparisons can lead to a phenomenon called “biocreep.”
Consider, for example, that Drug A is 80% effective against pneumonia. Drug A is FDA-approved and is thus known to be better than placebo. Using a 10% non-inferiority margin, Drug B could be 10 percent worse than Drug A—effective in only 72% of the population—but still be considered non-inferior to Drug A.
This situation is deemed appropriate because of Drug B’s other benefits over Drug A, such as fewer side effects or activity against bacterial strains resistant to Drug A. But suppose Drug C is in development, and its developer chooses to establish non-inferiority against Drug B. Drug C only needs to be effective in 65% of the population. Then the developer of new Drug D picks Drug C as its comparator, and thus only has to show effectiveness in 58% of the population. Yet by inferential deduction, Drug D with its 58% cure rate is considered non-inferior to Drug A with its 80% cure rate.
Continually comparing a new drug to the weakest comparator results in biocreep. The 14th drug in this series could be considered equal to the original gold standard drug, when in fact it probably has no benefit at all. Yet as previously established, it cannot be directly compared to placebo to ensure that it has at least some benefit.
The FDA has attempted to avoid biocreep by encouraging makers of new drugs to use a gold standard drug as their comparator, rather than another new drug. The problem with this approach is, because gold standard Drug A might have been developed 10 years ago, bacteria have likely developed resistance to it over that time. If a newer drug is then tested against Drug A, and 40% of patients in the trial have a strain of infection that is resistant to Drug A—but not to the newer drug—then the newer drug could show non-inferiority and without being as effective as it appears.
So what is the best way to avoid biocreep and prove to the FDA that your new drug is more effective than placebo? As illustrated through the examples above, one important factor is to choose the best drug to use for a comparator. This is not necessarily the gold standard from a decade ago and not necessarily the newest approved drug.
Picking the right endpoints is also critical. For example, if nafcillin and vancomycin were compared in treating methicillin-susceptible Staphylococcus aureus, using a primary endpoint of time to sterilization of the blood, the nafcillin group would achieve sterile blood in two days while the vancomycin group would require perhaps seven. Vancomycin would fail this trial. But while vancomycin is slower in achieving this endpoint than nafcillin, it is effective against methicillin-resistant S. aureus (MRSA) while nafcillin is not—making it a critically important part of the modern anti-infective drug arsenal.
In addition to selecting the right comparator and the right endpoint, it is important to establish the right non-inferiority margin by which to compare the new drug and the approved drug.
The difference between a new drug and a placebo is known as M1. But since anti-infectives are not usually compared to placebo, a historic value is often used. This can be data from a previous failed trial of a drug known to have very little efficacy—in essence, a surrogate placebo—or a good study from the pre-antibiotic era. In evaluating new anti-infectives for acute bacterial skin and skin structure infections, the FDA uses a 1937 study comparing sulfonamides to ultra-violet light in the treatment of erysipelas.3 The M1 is then further reduced to account for trial variability and other uncertainties inherent in using historic controls, resulting in M2, the non-inferiority margin.4
Recently, the agency has been trending toward the application of tighter non-inferiority margins—often 10%. In special cases, a drug developer may be able to move forward with a wider margin,5 while in other others, a company may be required to use an even tighter delta than 10%.
To select the best comparator, the most appropriate endpoints and the proper non-inferiority margin in a pivotal trial, the FDA is encouraging drug developers to conduct an extensive Phase 2 trial that looks at response rates for several variables. Once a company has decided on these, gaining the agency’s imprimatur on these elements is essential. Following this guidance provides the best chance to avoid biocreep and gain approval.
About the Author
Michael L. Corrado, MD, FIDSA is an established infectious disease expert through his involvement in the development of numerous infectious disease compounds, from preclinical work through marketing application submission. Prior to INC Research, he co-founded Advanced Biologics and previously worked at Merck and Johnson & Johnson, where he was the vice president of regulatory affairs.
References
1. Murphy TF. Placebo-controlled trials of treatments for community-acquired pneumonia: review of the literature and discussion of feasibility and potential value. Clin Infect Dis. 2008; 47(Suppl 3):S145.
2. https://www.fda.gov/OHRMS/DOCKETS/98fr/FDA-2009-D-0136-gdl.pdf
3. Snodgrass WR and Anderson T. Sulphanilamide in the Treatment of Erysipelas. Br Med J. 1937;2(4014):1156.
4. https://www.fda.gov/ohrms/dockets/ac/08/briefing/2008-4394b1-01-FDA.pdf
5. https://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM202140.pdf
Filed Under: Drug Discovery
