Pathogen resistance can track with drug penetration, not just with drug number or variety

Here's an interesting paper from a multi-institution group looking at a perpetual problem in antibacterial and antiviral drug therapy - the emergence of mutations and resistance. As one of the authors describes it on her blog, the project started when the team realized that while the conventional wisdom about multiple drugs working better because of increased barriers to multiple mutations in the pathogen is correct, the mechanism for this seemed to be rather fuzzy, especially because the data seemed to show that multiple mutations arise not in tuples but as singles, one at a time.

The authors essentially divided the body into different compartments with viruses and bacteria and then assumed differential penetration of drugs in multidrug therapy into these compartments. They built a mathematical model to track the emergence of mutations as a function of exposure to various drugs. What they found is that imperfect drug penetration into a compartment can allow mutations against that drug to arise one at a time, and quite speedily at that. Thus, even if multiple drugs are being administered, the pathogen may not be exposed to these drugs simultaneously at the relevant concentrations in every compartment: as indicated in the title of the paper, even if the treatment may therefore consist of multitherapy in principle, it leads to "spatial monotherapy" in practice leading to the focused exposure of certain compartments only to single drugs at a time.

The study is interesting because it shows that mutations and resistance may not only be a function of drug number and variety but could also result from the drugs not making it to their site of action effectively - it's something that's not surprising in retrospect but it's nice to firm up this belief with a model. It also means that simply administering multiple drugs may not curb resistance if those drugs have vastly different penetration profiles.

If this hypothesis is true however, then you would see some correlation between drug penetration profiles and resistance mutations and not just with the diversity and number of drugs - in part it could explain why certain combinations of drugs even from similar classes work better than others (for instance, it could potentially explain why protease inhibitor X + RT inhibitor Y work better than protease inhibitor A + RT inhibitor B). I am assuming that this kind of cell and compartment penetration data does exist for certain drugs so it would be interesting to model it and track such potential correlations.