Detection of Pyrazinamide Heteroresistance in Mycobacterium tuberculosis

Abstract

Heteroresistance is defined as the coexistence of both susceptible and resistant bacteria in a bacterial population. Previously published data show that it may occur in 9 to 57% of Mycobacterium tuberculosis isolates for various drugs. Pyrazinamide (PZA) is an important first-line drug used for treatment of both drug-susceptible and PZA-susceptible multidrug-resistant TB. Clinical PZA resistance is defined as a proportion of resistant bacteria in the isolate exceeding 10%, when the drug is no longer considered clinically effective. The ability of traditional drug susceptibility testing techniques to detect PZA heteroresistance has not yet been evaluated. The aim of this study was to compare the capacity of Bactec MGIT 960, Wayne's test, and whole-genome sequencing (WGS) to detect PZA-resistant subpopulations in bacterial suspensions prepared with different proportions of mutant strains. Both Bactec MGIT 960 and WGS were able to detect the critical level of 10% PZA heteroresistance, whereas Wayne's test failed to do so, with the latter falsely reporting highly resistant samples as PZA susceptible. Failure to detect drug-resistant subpopulations may lead to inadvertently weak treatment regimens if ineffective drugs are included, with the risk of treatment failure with the selective growth of resistant subpopulations. We need clinical awareness of heteroresistance as well as evaluation of new diagnostic tools for their capacity to detect heteroresistance in TB.

Keywords: BACTEC MGIT 960; Mycobacterium tuberculosis; Wayne's test; drug susceptibility testing; heteroresistance; method evaluation; pyrazinamide; whole-genome sequencing.

FIG 1

Bactec MGIT 960 results. Histogram showing the mean growth units (GU) with standard deviations for the 10%, 5%, and 1% mixes of the in vitro-generated pncA Leu159Val and Leu85Pro mutants as well as the pncA Val21Gly mutant MDR isolate. A result of ≥100 GU (marked by the horizontal line) indicates pyrazinamide resistance. The tests were performed in replicates on one occasion and in duplicate on two separate occasions, i.e., 5 tests per mixed proportion (10%, 5%, and 1%) of each PZA-resistant strain (n = 45 tests in total).

FIG 2

Results of Wayne’s test with the mixes of H37Rv and the in vitro-generated isogenic H37Rv PZA-resistant mutant Leu159Val. From the left: 100% susceptible H37Rv and thereafter decreasing proportions of H37Rv (50%, 25%, 10%, 5%, and 1% H37Rv, mixed with the Leu159Val mutant.) A positive sample, i.e., a PZA-susceptible sample, forms a red band below the agar surface. The tubes containing 100% H37Rv and 50% H37Rv are clearly positive, while 25% H37Rv showed a weaker positivity. The remaining samples, i.e., those with 10%, 5%, and 1% H37Rv and with 100% Leu159Val mutant, are negative and regarded as PZA resistant.

FIG 3

Reproducibility of the whole-genome sequencing (WGS) assay to detect different levels of heteroresistance in cultures with 50, 25, 10, and 5% resistant bacteria (each proportion was tested for six different cultures). The horizontal lines in the boxes represent median values. Three of four mixtures with 5% resistant bacteria were detected only after a less strict variant filtering was applied. Data are based on sequencing of 5 cultures containing 50% resistant bacteria. One of the mixtures with 50% resistant bacteria did not yield enough DNA to be subjected to WGS.

FIG 4

(a and b) Bacterial suspensions prepared with mixed proportions of PZA-resistant M. tuberculosis for Bactec MGIT 960. Mixes for whole-genome sequencing were prepared similarly, with the addition of 25% and 50% PZA-resistant bacteria. (a) Isogenic H37Rv mutant and pansusceptible H37Rv. (b) A clinical MDR-TB isolate and a pansusceptible clinical isolate, both lineage 3.1.1. (c) Bacterial suspensions prepared with mixed proportions of PZA-resistant bacteria for Wayne’s test. Similar mixes were prepared for the Leu85Pro mutant.