Measurements of the in vitro anti-mycobacterial activity of ivermectin are method-dependent

Sir, We recently discovered surprising in vitro activities of avermectins against Mycobacterium tuberculosis. The conclusions of our report were based on conventional antimicrobial assays against representative M. tuberculosis strains, including drug-resistant clinical isolates. The antimycobacterial activities of avermectins, commonly used to treat helminthic (nematode) infections, may have important clinical implications for tuberculosis therapy. Muhammed Ameen and Drancourt subsequently questioned our work based on a small set of data that detected low activity for ivermectin. While they interpreted their MIC results as being largely contradictory to our data and conclusions, they did not replicate our experiments in broth but instead adopted a different methodology based on bacterial growth on a solid surface (agar reference proportion method). This method is standardized for drug susceptibility comparisons of clinical isolates but not for drug screening. In fact, it was originally optimized to recover maximal numbers of bacteria from sputum samples. It is clear that neither broth nor agar surface-based assays accurately reflect the complex environment of M. tuberculosis residing in human lungs; as a result, the MICs of drugs measured in vitro often do not correlate well with in vivo protective activity. While there is an urgent need for new drugs to treat tuberculosis, it is still unclear what media or culture conditions are most predictive of in vivo activity when screening compound libraries. Therefore, we believe that it is premature to rule out a new potential antituberculosis drug candidate based on a single method using 11 M. tuberculosis isolates from French clinics. Our studies were carried out in two different laboratories using 36 mycobacterial strains (laboratory and clinical isolates) from at least five different geographical locations. We used microdilution broth methods coupled to the MTT assay and two independent kinetic kill curve experiments (dose–response and time dependence) to evaluate the activities of avermectins. The MTTassay directly measures metabolic activity. It correlates well with the reference proportion and other well-established methods for determining the MICs of drugs. – 7 An additional misunderstanding relates to the use of the term MIC90 to describe dose– response experiments. The term MIC90, widely used in both drug development (including tuberculosis drug development) and clinical studies, has different meanings in these fields and therefore must be interpreted in context. In our drug discovery studies, we used MIC90 in a conventional manner to describe minimal inhibitory drug concentrations for single strains/isolates (MIC90 indicates 90% growth inhibition). In addition, contrary to the allegations of Muhammed Ameen and Drancourt, we did not interpret these MIC90 values as proof that these strains were clinically ‘susceptible’ to avermectins. Instead, we used this standardized analysis to better understand the in vitro antimycobacterial activities of avermectins, which are bactericidal and exposure dependent, a characteristic of the most effective antibiotics. We would like to avoid such misunderstandings between colleagues addressing clinical or drug development challenges. To resolve this issue, we rigorously assayed the activity of ivermectin using the experimental conditions described by Muhammed Ameen and Drancourt in parallel with our methodology. We reconfirmed our MTT results and were also able to reproduce their data showing that the MIC of ivermectin was much higher in solid media compared with liquid media (Table 1). Similar results were obtained by Dr Norio Doi using agar dilution and 7H9 microdilution methods on 30 clinical M. tuberculosis isolates, including 10 that were drug resistant and the reference H37Rv strain (Dr Norio Doi, Research Institute of Tuberculosis, Japan, personal communication). The unambiguous conclusion of Muhammed Ameen and Drancourt, that ‘ivermectin lacks antituberculous activity’, was also based on an oversimplified interpretation of ivermectin’s pharmacokinetic properties. We agree that the low peak plasma concentrations of ivermectin after single oral dose administration (in the ng/mL range) to treat nematode infections and potential toxicity at higher dosages could jeopardize the development of ivermectin as an antituberculosis drug. However, since antibacterial activity is correlated with drug concentrations at the site of infection, the localization of tuberculosis infections in pulmonary tissue could minimize the relevance of ivermectin plasma concentrations. In addition, avermectins might even be more active against intracellular M. tuberculosis; ivermectin was recently reported to inhibit the obligate intracellular bacteria Chlamydia