Successful detection of an unrecognized outbreak of Mycobacterium tuberculosis in the modern Beijing subfamily through combined molecular epidemiological and population genetic analyses

Abstract

Introduction

While multi-locus variable number tandem repeat analysis (MLVA) as molecular epidemiological analysis has been used to detect Mycobacterium tuberculosis outbreaks, its discriminatory power for identifying strains is limited. Whole-genome sequencing (WGS) offers high discriminatory power but is expensive. This study was established to develop a strategy to overcome these limitations of molecular epidemiological analysis by combining it with population genetic analysis.

Methods

MLVA data from 2732 M. tuberculosis strains isolated in Chiba Prefecture, Japan, in 2008–2016 were subjected to Bayesian population genetic analysis to subdivide the strains into subfamilies and estimate subpopulations within each subfamily. Annual changes in the number of strains within subpopulations exhibiting linkage disequilibrium (LD) in MLVA data were examined. Only strains from subpopulations displaying significant increases were analyzed by WGS.

Results

Significant LD was observed in one subpopulation using Bayesian analysis (designated P3) within the modern Beijing subfamily, which exhibited a significant increase in strain number in 2016. WGS analysis of strains belonging to P3 from 2016 revealed that 17 out of 21 of them differed by three or fewer single-nucleotide polymorphisms from their most similar strain, indicating that they had a common origin (i.e., an outbreak). Among these common-origin strains, one exhibited a four-locus variant in the MLVA, which would not be suspected of being an outbreak-related strain based on MLVA alone without Bayesian analysis.

Conclusion

The combination of Bayesian population genetic analysis with MLVA successfully detected M. tuberculosis strains from an unrecognized outbreak by performing WGS on only a subset of the strains.