Direct Detection of Rifampin and Isoniazid Resistance in Sputum Samples from Tuberculosis Patients by High-Resolution Melt Curve Analysis

Abstract

Drug-resistant tuberculosis (TB) is a major threat to TB control worldwide. Globally, only 40% of the 340,000 notified TB patients estimated to have multidrug-resistant-TB (MDR-TB) were detected in 2015. This study was carried out to evaluate the utility of high-resolution melt curve analysis (HRM) for the rapid and direct detection of MDR-TB in Mycobacterium tuberculosis in sputum samples. A reference plasmid library was first generated of the most frequently observed mutations in the resistance-determining regions of rpoB, katG, and an inhA promoter and used as positive controls in HRM. The assay was first validated in 25 MDR M. tuberculosis clinical isolates. The assay was evaluated on DNA isolated from 99 M. tuberculosis culture-positive sputum samples that included 84 smear-negative sputum samples, using DNA sequencing as gold standard. Mutants were discriminated from the wild type by comparing melting-curve patterns with those of control plasmids using HRM software. Rifampin (RIF) and isoniazid (INH) monoresistance were detected in 11 and 21 specimens, respectively, by HRM. Six samples were classified as MDR-TB by sequencing, one of which was missed by HRM. The HRM-RIF, INH-katG, and INH-inhA assays had 89% (95% confidence interval [CI], 52, 100%), 85% (95% CI, 62, 97%), and 100% (95% CI, 74, 100%) sensitivity, respectively, in smear-negative samples, while all assays had 100% sensitivity in smear-positive samples. All assays had 100% specificity. Concordance of 97% to 100% (κ value, 0.9 to 1) was noted between sequencing and HRM. Heteroresistance was observed in 5 of 99 samples by sequencing. In conclusion, the HRM assay was a cost-effective (Indian rupee [INR]400/US$6), rapid, and closed-tube method for the direct detection of MDR-TB in sputum, especially for direct smear-negative cases.

Keywords: DNA sequencing; high-resolution melt curve analysis; molecular diagnostics; multidrug resistance; sputum; tuberculosis.

FIG 1

HRM curves for the detection of M. tuberculosis drug resistance. (A, C, and E) Normalized melting curves of rpoB, katG, and inhA, respectively. (B, D, and F) Temperature-shifted difference plots of rpoB, katG, and inhA generated by normalizing the melting profiles of all samples (mutant plasmid controls and sample DNA in color) against the WT profile (in black). RFU, relative fluorescence units.

FIG 2

Categorization of samples in mutant/WT category. (A) M. tuberculosis isolates (n = 25); 2 katG 315 mutants determined by sequencing (S315N and S315T) were incorrectly denoted as WT by HRM. Three samples were categorized as an S315N mutant by sequencing but were incorrectly denoted as an S315T mutant by HRM. (B) Smear-positive sputum samples (n = 15). (C) Smear-negative sputum samples (n = 84); out of 9 rpoB H526N mutant samples (determined by sequencing), three samples showed discrepant results. One sample was incorrectly categorized as an S531L mutant, and 2 samples were categorized as WT by HRM. Out of 4 katG S315N mutant samples (determined by sequencing), 3 samples were incorrectly denoted as S315T mutants, 1 was categorized as WT by HRM, and 2 S315T mutants (determined by sequencing) were wrongly denoted as WT by HRM.

FIG 3

Representative electropherograms for heteroresistant samples (at codon 526 of rpoB). (A) Wild-type (CAC) and mutant (AAC) sequence (sample 42). (B) Wild-type (CAC) and mutant (CAG) sequence (sample 89).

FIG 4

Flow chart of the study. M.tb, M. tuberculosis.