Drug tolerance predicts clinical outcomes in mycobacterial lung infections

Abstract

<b>Background:</b> Antimicrobial susceptibility testing (AST) is critical for guiding treatment in infectious diseases. However, in chronic pulmonary infections, AST often correlates poorly with treatment success. AST evaluates growth across different antibiotic concentrations, but not bacterial killing (drug tolerance), which may be required to efficiently clear long-lasting infections. <b>Aims:</b> By overcoming the poor scalability and reproducibility of killing assessments, we establish the clinical relevance of sterilising treatment in <i>Mycobacterium abscessus</i>, a particularly difficult-to-treat pathogen causing increasing rates of chronic pulmonary infections globally. <b>Methods:</b> We assessed drug susceptibility (MICs) and bacterial killing in regard to clinical outcomes in an international cohort of 142 cystic fibrosis patients with M. abscessus lung disease. To evaluate bacterial killing, we developed a live-cell imaging platform that allowed testing thousands of conditions. <b>Results:</b> Using our imaging approach, we analysed bacterial killing in 142 M. abscessus isolates using 11 drugs. We observed various time-kill profiles across and within drugs tested, indicating that bacterial killing is a fundamental bacterial phenotype. Cefoxitin and imipenem kill-kinetics were predictive for clearing infections independent of MICs, wheras&nbsp;cefoxitin killing&nbsp;(in contrast to MICs) was also associated with a faster lung function decline. <b>Conclusion:</b> We present a novel method for assessing bacterial killing, demonstrate highly diverse killing kinetics in M. abscessus and validate its importance via predicting clinical outcomes; thereby challenging current drug susceptibility measures.