We have often mentioned the central importance of microbiological response as an endpoint in trials of antibacterials. This is contrary to other therapeutic areas, in which the etiology is less understood and where surrogate markers are needed.
In bacterial infectious diseases we have the luxury of knowing exactly what causes the disease, and an organism-response outcome can be established quite unambiguously. When EOT cultures cannot be obtained, the term ‘presumed eradication’ was used, if the clinical response was favorable. When there was a need to switch antibiotics because of non-response, and only then, did we call these cases clinical failures. The microbiological response was labeled ‘persistence’ or ‘presumed persistence’. These terms reflected clinical practice and made intuitive sense.
With antibiotics eradication of the causative organism is really all we care to achieve in order to regain health, and antibacterials do just that. First and foremost, antibiotics are not symptomatic treatments but are curative by their mode of action, namely attacking the bacterial causative pathogens. We are not using them for any anti-inflammatory activity they may possess to stop the cytokine response that comes with infections. We are using them to eradicate the organism. More than 120 years after Pasteur and Koch’s work and thousands of clinical trials, we don’t have to prove any more that eradication of a bacterial pathogen is central to controlling the disease process: clinical improvement invariably follows bacterial eradication. Of course, one can track outcome using various surrogate markers, but we already know that leukocytosis, inflammatory markers, LFTs will improve on their own, unless there is an MDR pathogen not covered by the antibiotic selected.
That said, why are we looking at other markers of improvement in registration trials, knowing that none of them carries much weight and little is added in value? How many times will patients with pneumococcal pneumonia improve clinically if the organism persists? This is a rhetorical question, of course, as it hardly ever happens. How many times will pneumococcal pneumonia worsen despite proper antibiotic choice in the case of a fully susceptible pathogen? Also an unlikely scenario. How many times is there a discordant result between bacteriologic response and clinical outcome? Not often at all.
Exceptions exist, of course, when dealing with pus collections that need surgical drainage or in immunocompromised hosts, but the primary driver of the outcome rests with organism eradication. Is that so hard to comprehend?
Do we believe that it is the bacteria that are causing the disease, or do we need to invoke some kind of miasma that needs attention?
Call our emphasis on microbiology biased, call it simplistic, but it is our firm belief that a dead bug is a good bug and the goal of antibiotic therapy is eradication of the causative pathogen. Everything else is epiphenomena, nice to collect in clinical trials but adding little of substance.
The disease has not changed, the science has not changed, and the organisms have not changed. Therefore why have Guidances changed every 10 years? Does anyone ever go back to see if anything has been gained by the effort? Has the latest iteration of antibacterial guidelines improved the safety of drugs in development, eliminated ineffective drugs, or improved patient well-being in any way, shape or form?
None of what we discuss here is new, of course. Most clinical investigators resigned themselves to the FDA mantra of ‘feel-function-survive’ and gave up arguing with this logic, although begrudgingly. Of course, regulated industry had to fall in line with the new rules. It was not that everyone followed ‘Big Brother’ like lemmings, just the opposite: industry chose to walk away from the field of anti-infective drug development. There was a tacit understanding to just stay on the sidelines and wait until the pendulum swings back.
Given this historical debate and our current regulatory status, it was refreshing to read a publication by Yates et al. from earlier this year that I would like to bring to everyone’s attention (including Regulators working on the next set of Guidances).
Yates, studying tuberculosis trials [1], recommended that “primary efficacy endpoints in phase 3 non-inferiority trials …should only include microbiological outcomes”. This sounds unorthodox, almost heretical. Of note, the authors, supported by the Gates Foundation, are from London’s School of Hygiene & Tropical Medicine and the London University in the UK, respected centers not known for old-world eccentricity.
Here are some quotes and our comments:
| Yates et al. | Allphase Pharma Consulting |
| Composite primary efficacy endpoints have typically included treatment failure, relapse, death, early discontinuation of treatment, loss to follow-up, and treatment switches, with each of these components carrying equal weight. …giving equal weight in a composite primary outcome to microbiological endpoints and soft outcomes, such as stopping treatment a few weeks early or having to change components of a tuberculosis treatment regimen, makes little sense. | We agree: It really makes little sense to combine criteria of uneven weight. |
| Primary efficacy endpoints in phase 3 non-inferiority trials to establish new tuberculosis treatment regimens should only include microbiological outcomes | We agree: Feel-function-survive should take a back seat to microbiologic outcomes. Why only in TB treatment regimens? We can’t think of any reason why the same would not apply to other bacterial infections as well |
| relative efficacy and safety of tuberculosis treatment regimens can only be reliably ascertained through randomised controlled trials | Fully agree – and so would FDA and most regulators |
| In most trials, non-microbiological endpoints were much more frequent than microbiological endpoints. Precision was markedly improved using only microbiological endpoints | Of course, with less discriminatory power, these ancillary response criteria just dilute the treatment effect captured in the micro endpoint |
| Many people who stop treatment early experience no ill consequences. If stopping treatment does lead to treatment failure or relapse, these outcomes will anyway be captured as microbiological endpoints. | This does not apply only to TB but to all mild/moderate infections.
|
| Re: deaths: By including events in a composite primary outcome that cannot be impacted by the intervention, there is a risk of falsely concluding noninferiority. …our preference would be to include death as a safety outcome | Well stated – we agree. No need for double-counting deaths as ‘lack of efficacy’ AND as safety events |
The FDA felt that clinical judgement is biased and needs to be replaced programmatically by algorithms, like the ECR (early clinical response) assessment for CABP. The ECR criteria are listed below [2]; it is a composite omitting bacteriology; it is an efficacy read-out when the patient is just a few days into treatment. Thankfully, EMA did not follow the FDA logic and stuck by the older TOC endpoint.
Here is the rub: even if we can acknowledge that clinical judgement is subjective to a degree, microbiological response is not subjective. When dealing with bacterial infections, we therefore agree with Yates that microbiology provides more solid information than the subjective criteria of chest pain, sputum production, cough or dyspnea taken individually or together. The FDA changes did not help patients or investigators; instead they made the CROs rich. CROs benefit the most from complex guidelines requiring the collection of myriads of data points (which for the most part are not contributory!) to evaluate a new antibiotic’s safety and efficacy in an indication.
Yates and colleagues have reopened a discussion, which those of us in the trenches have long given up on. We are happy to see our ‘biases’ confirmed by researchers that have not resigned themselves to the ‘Brave New World’ of FDA mandates that relegate microbiology to second tier information in antibiotic efficacy evaluations.
In antibacterial trials, bacterial response as the primary efficacy parameter has served us very well in the past. Yates and colleagues point out the drawbacks of the current regulatory framework. Let’s hope the pendulum swings back quickly making antibiotic development feasible and reasonable again.
ABBREVIATIONS
CABP community-acquired pneumonia
ECR early clinical response
EOT end-of-treatment
MDR multidrug-resistant
TOC test of cure
REFERENCES
[1] Yates T. Primary efficacy endpoints in phase 3 non-inferiority trials to establish new tuberculosis treatment regimens should only include microbiological outcomes. Lancet Microbe 2025; 6: 101117
[2] “Response at this early time point requires improvement in at least 2 of 4 specific symptoms (chest pain, cough, sputum production, and dyspnea) without any symptom worsening”. From: US Food and Drug Administration. Guidance for Industry; Community-Acquired Bacterial Pneumonia: Developing Drugs for Treatment Guidance for Industry, 2020
