Epidemiological cut-offs for Sensititre susceptibility testing of Mycobacterium tuberculosis: interpretive criteria cross validated with whole genome sequencing.

Universal drug susceptibility testing (DST) is an important requirement of the End TB Strategy. The Sensititre broth micro-dilution assay (BMD) tests multiple drugs quantitatively. We defined interpretive criteria for this assay and analysed genotypic-phenotypic relationships. 385 Mycobacterium tuberculosis clinical isolates were processed for BMD and whole genome sequencing. The epidemiological cut-off value 99% (ECV) amongst genotypically wild type (gWT) strains defined susceptibility. Minimum inhibitory concentration distributions of the resistance-associated variants (RAVs) for each drug were analysed. Susceptibility (µg/mL) criteria were determined as follows: rifampicin (≤0.125), isoniazid (≤0.25), ethambutol (≤2.0), moxifloxacin (≤0.5), levofloxacin (≤1.0), amikacin (≤2.0), kanamycin (≤8.0), capreomycin (≤4.0), clofazimine (≤0.25) and linezolid (≤2.0). Most drugs showed clear separation between gWT and RAV. Isoniazid showed a tri-modal pattern with 14/17 strains at ECV harbouring a fabG1 c. -15C > T RAV. Ethambutol RAVs at embB codons 306, 405 and 497 were responsible for resistance and showed differential distributions. Moxifloxacin RAVs (gyrA codon 90) were a dilution or two higher than the ECV while gyrB RAVs were uncommon and showed drug specific resistance propensity. Interpretive criteria established were robust facilitating progress towards universal DST and individualised precision medicine. This study demonstrates the value of quantitative DST to accurately interpret mutation data.