S83 Genetic diagnosis of bedaquiline resistance – an individual isolate meta-analysis

Abstract

<h3>Introduction</h3> Bedaquiline is an important treatment for drug-resistant tuberculosis. The main bedaquiline resistance mechanism is Rv0678 mutations. Many individually rare mutations are reported, with many having a variable relationship to phenotypic resistance - causing small minimum inhibitory concentration (MIC) increases to at/near the critical concentration (CC). Sequencing approaches could diagnose resistance if genotypic-phenotypic correlations can be established. However, the WHO mutation catalogue does not contain any bedaquiline resistance-associated variants (RAVs). Here, we collate all genotypic and phenotypic resistance reported from clinical isolates to assess effectiveness of sequencing resistance-associated genes. <h3>Methods</h3> We screened public databases for articles published up until June 2022. Studies were included if they performed whole genome sequencing of clinically-sourced isolates with bedaquiline resistance and measured MIC by MGIT, 7H11, broth microdilution. Candidate resistance genes were Rv0678, atpE, pepQ, and atpB. We deemed isolates with MICs&gt;CC to be resistant, those with MICs=CC to be intermediate, and those with MICs&lt;CC to be susceptible. To calculate the association of individual mutations with resistance, we used previously established methods on a combination of all extracted isolates from included studies and the ‘reuse’ Cryptic study dataset. <h3>Results</h3> Twelve studies were included yielding 896 isolates containing ≥1 potential RAV, with 277 (30.9%) phenotypically intermediate/resistant. 261 intermediate/resistant isolates were identified with no candidate gene mutations. 943/1175 (80.3%) of all isolates were derived from Cryptic. Sensitivities and positive predictive value of taking an ‘any mutation’ approach in candidate genes (assuming all mutations may cause resistance) are shown (table 1). Twenty mutations had a significant association with an intermediate/resistant MIC (p&lt;0.05) (Table). All were strong associations (OR &gt;10), except 192_indel (OR=8.8). When considering only resistant MICs, 11 mutations had a significant association, all a strong except 192_indel (OR=4.6). <h3>Conclusions</h3> The any candidate gene mutation approach has limited sensitivity for identifying bedaquiline resistance (45–63%) but high specificity and negative predictive value (&gt;97%) in this cohort with a high prevalence of drug resistance (61% rifampicin-resistant). We identify 20 mutations that had a strong association with intermediate/resistant MICs (13 resistant), not included in the current mutation catalogue. Genetic methods to identify bedaquiline resistance are unlikely to be sufficient in isolation.