High frequency of bedaquiline resistance in programmatically treated drug-resistant TB patients with sustained culture-positivity in Cape Town, South Africa

Aims & objectives: Potentially transformative new tuberculosis (TB) drugs like bedaquiline are undergoing roll-out, however, this is largely in the absence of programmatic drug susceptibility testing (DST). We lack information on how susceptibility changes during treatment in patients on bedaquiline-containing regimens, especially in at risk patients who have complex treatment histories, are in programmatic rather than trial environments, and have a delayed treatment response (defined here as sustained culture-positivity). Methods: Serial isolates from 51 patients with drug resistant (DR-)TB who were culture-positive after ≥4 months of a programmatically administered bedaquiline-containing regimen, were collected. Bedaquiline phenotypic DST in MGIT 960 (1μg/ml), targeted deep sequencing (Rv0678, atpE, pepQ) and whole genome sequencing was done on paired isolates (pre-bedaquiline initiation, post-four-month). Results: 24/51 (47%) patients were phenotypically and genotypically resistant (39% acquired resistance). Excluding one patient with an unknown history, prior clofazimine exposure was associated with bedaquiline-resistance [21/24 (88%) bedaquiline resistant cases had prior clofazimine vs. 12/26 (46%) susceptibles; p=0.002]. Diverse combinations of single nucleotide polymorphisms (SNPs) and indels were in the Rv0678 promoter region and the Rv0678 and atpE genes. Examples of newly described resistance associated variants (RAVs) include Rv0678 -8 T/G and atpE 223 C/T. RAVs were not in defined hotspots and sometimes occurred concurrently with atpE RAVs. Discussion: The rate of bedaquiline resistance acquisition in this population is alarmingly high and associated with prior clofazimine exposure. The diverse RAVs pose challenges to molecular test development. This study highlights the existence of a potentially infectious pool of bedaquiline-resistant patients present under programmatic conditions in a resource-constrained setting and illustrates the danger of starting patients with complex histories on a novel drug without routinely available DST.