False hope: narrow-spectrum bacteria killers

Indeed, it appears that much of the (preclinical) field has moved in this direction.¹³⁸ These strategies focus on generating “narrow-spectrum bacteria killers” to solve AMR and deal with the problem of insufficiently selective broad-spectrum antibiotics. The technologies to do this are genuinely innovative and exciting. They range from targeted small molecules (often employing AI in the process), to antibodies, to engineered bacteriophages (bacteria-specific viruses), to lysins (engineered molecules sourced from bacteriophages which kill bacteria directly), and more. Despite the diverse technological approaches, these techniques share the objective of killing a specific pathogen while sparing the rest.

Narrow-spectrum bacteria killers have a certain appeal, because they are conceived of as “personalized medicine”. A patient could be tested, the specific pathogen identified, and the drug administered, exactly matched to the pathogen causing the infection. The idea has elegance.

Our view, though, is that as a strategy, this is a dead end, independent of the technology harnessed to kill the bacteria.

First, this strategy relies on a robust diagnostic infrastructure operating very quickly and at scale. We do not have that in any meaningful sense, and should not expect to in the foreseeable future.

Current diagnostics deployed in healthcare settings are slow and unreliable. Slow, meaning they usually take two days or more to deliver results.¹³⁹ Unreliable, meaning that they often do not return an actionable result (either a null, or a polymicrobial report).¹⁴⁰ Respiratory infections only identify a culprit pathogen about half the time. Hiding bacteria (biofilm, intracellular, persisters) may not be collected by the biopsy and/or identified in the diagnostic. Many new technologies in development could conceivably speed up results, fixing the “slow” part of the diagnostic problem if deployed at scale.¹³⁹ But these technologies, even if widely deployed, cannot address the unreliability of sample collection.

Partly as a result of the limitations in sampling and diagnostic technology, severe, resistant infection patient populations are small. As mentioned before, Carbapenem-resistant Acinetobacter infections cause 8,500 hospitalizations and 700 deaths annually.³⁴ Recruiting enough patients for a narrow-spectrum bacteria killer clinical trial presents a serious challenge: not only are the patient numbers small, but the nature of the acute infection means that they are hard to enroll in a trial in the small window of time when medication needs to start—often just a few hours.

Assuming a trial can be completed and drug approved, patient populations defined in this way result in extremely narrow commercial markets, which must still compete against generic, approved alternatives. Generic competition limits pricing power. As a drug product, narrow-spectrum bacteria killers suffer from low volume and low price. Most of these technologies have complex manufacturing processes and high cost of goods. As commercial products, most (or all) will struggle to break even, and are unlikely ever to recover development costs.

The problem gets worse. To cover all the resistant pathogens of concern, the world would require a portfolio of narrow-spectrum bacteria killers. We would probably need a minimum of five such drugs to cover the specific pathogens of highest clinical importance. More realistically, 10–20 would be required. Each with the same commercial dynamics.

Such a program seems difficult, at best. At worst, wasteful of scarce talent and capital.

At Unison, we seek to change the entire paradigm for how bacterial infections get treated, managed, and controlled. Narrow-spectrum bacteria killers are unlikely to be anything beyond that: narrowly applied. We focus our energy on bacterial defense disablers, which we believe can and will change the paradigm for infection control.