Comparative Performance of Genomic Methods for the Detection of Pyrazinamide Resistance and Heteroresistance in Mycobacterium tuberculosis.

Background: Pyrazinamide is an important component of both drug-susceptible and drug-resistant tuberculosis treatment regimens. Although approximately 50% of rifampicin resistant isolates are also resistant to pyrazinamide, pyrazinamide susceptibility testing is not routinely performed due to the challenging nature of the assay. We investigated the diagnostic accuracy of genotypic and phenotypic methods, and explored the occurrence of pyrazinamide heteroresistance. Methods: We assessed pyrazinamide susceptibility among 358 individuals enrolled in the South African EXIT-RIF cohort using Sanger and targeted deep sequencing (TDS) of the pncA gene, whole genome sequencing (WGS), and phenotypic drug-susceptibility testing. We calculated the diagnostic accuracy of the different methods, and investigated the prevalence and clinical impact of pncA heteroresistance. True pyrazinamide susceptibility status was assigned to each isolate using the Koser classification and expert rules. Results: We observed 100% agreement across genotypic methods for detection of pncA fixed mutations, only TDS confidently identified three isolates (0.8%) with minor variants. For the 355 (99.2%) isolates that could be assigned true pyrazinamide status with confidence, phenotypic DST had a sensitivity of 96.5% (95% CI: 93.8-99.3%) and specificity of 100% (95% CI: 100-100%); both Sanger sequencing and WGS had a sensitivity of 97.1% (95% CI: 94.6-99.6%) and specificity of 97.8% (95% CI: 95.7-99.9%); and TDS, sensitivity of 98.8% (95% CI: 97.2-100%) and specificity of 97.8% (95% CI: 95.7-99.9%). Conclusions: We demonstrate high sensitivity and specificity for pyrazinamide susceptibility testing among all assessed genotypic methods. The prevalence of pyrazinamide heteroresistance in Mtb isolates was lower than that identified for other first-line drugs.