A cross-sectional study on discordances between Anyplex™ II MTB/MDR and BACTEC MGIT 960 system for assessing susceptibility for Mycobacterium tuberculosis: analysis of clinical specimens from patients with isoniazid-monoresistant, rifampicin-monoresi

BACKGROUND: Rapid molecular tests have facilitated early diagnosis and guided treatment of tuberculosis (TB) by providing faster results than the gold-standard phenotypic drug susceptibility testing (DST). Anyplex™ II MTB/MDR is an RDT available in Thailand that simultaneously detects Mycobacterium tuberculosis (MTB) complex and mutations associated with isoniazid (INH) and rifampicin (RIF) resistance, enabling genotypic DST in various clinical specimens. However, discordance between the Anyplex™ II MTB/MDR and phenotypic DST creates a decision-making dilemma. METHODS: A retrospective review was conducted on clinical and microbiological data from patients with different types of tuberculosis (TB) at King Chulalongkorn Memorial Hospital (KCMH), a referral hospital for TB in Thailand. The study focused on patients with drug-susceptible TB (DS-TB), isoniazid mono-resistant TB (Hr-TB), rifampicin mono-resistant TB (RR-TB), and multidrug-resistant TB (MDR-TB) from January 2018 to June 2023. The correlation between genotypic and phenotypic DST using the Anyplex™ II MTB/MDR and BACTEC MGIT 960 assays, respectively, was investigated using clinical specimens and MTB clinical isolates recovered from specimens collected during the same period. RESULTS: 836 MTB clinical isolates from 796 patients had genotypic and phenotypic DST results. 44 MTB (5.2%) clinical isolates had discordant results. There was a significant discordance between the assays in assessing INH (p-value < 0.001) and RIF susceptibility (p-value < 0.022). The Cohen's kappa coefficient for assessing INH and RIF susceptibility was 0.837 and 0.772, respectively. The majority of INH susceptibility discordances (29/32, 90.6%) displayed genotypic susceptibility but phenotypic resistance. RIF susceptibility discordances (11/13, 84.6%) displayed genotypic resistance but phenotypic susceptibility. 52.2% were undertreated, 2.3% were overtreated, and 40.9% experienced no change in treatment in patients with genotypic-phenotypic discordant DST results. Undertreatment was due to false-negative genotypic DST. 6/11 RIF-susceptibility discordant isolates were susceptible by phenotypic DST; mutations in the rpoB gene were identified: Leu430Pro, Asp435Tyr, or Leu452Pro. CONCLUSIONS: The Anyplex™ II MTB/MDR assay demonstrated high agreement with the BACTEC MGIT 960 assay; however, both assays yielded substantial discordance in resistance results. Most of these discordances were false-negative genotypic results for INH resistance, which could lead to undertreatment decisions. Furthermore, some instances of genotypic resistance but phenotypic susceptibility to RIF were associated with borderline rpoB mutations, highlighting the need for careful interpretation of discordant results.