Bayesian reconstruction of Mycobacterium tuberculosis transmission networks in a high incidence area over two decades in Malawi reveals associated risk factors and genomic variants.

Benjamin Sobkowiak,Amelia C Crampin,Taane G Clark,Judith R Glynn,Themba Mzembe,Louis Banda

doi:10.1099/mgen.0.000361

Abstract

Understanding host and pathogen factors that influence tuberculosis (TB) transmission can inform strategies to eliminate the spread of Mycobacterium tuberculosis (Mtb). Determining transmission links between cases of TB is complicated by a long and variable latency period and undiagnosed cases, although methods are improving through the application of probabilistic modelling and whole-genome sequence analysis. Using a large dataset of 1857 whole-genome sequences and comprehensive metadata from Karonga District, Malawi, over 19 years, we reconstructed Mtb transmission networks using a two-step Bayesian approach that identified likely infector and recipient cases, whilst robustly allowing for incomplete case sampling. We investigated demographic and pathogen genomic variation associated with transmission and clustering in our networks. We found that whilst there was a significant decrease in the proportion of infectors over time, we found higher transmissibility and large transmission clusters for lineage 2 (Beijing) strains. By performing evolutionary convergence testing (phyC) and genome-wide association analysis (GWAS) on transmitting versus non-transmitting cases, we identified six loci, PPE54, accD2, PE_PGRS62, rplI, Rv3751 and Rv2077c, that were associated with transmission. This study provides a framework for reconstructing large-scale Mtb transmission networks. We have highlighted potential host and pathogen characteristics that were linked to increased transmission in a high-burden setting and identified genomic variants that, with validation, could inform further studies into transmissibility and TB eradication.

Highlights

Establishing patterns of transmission – who infected whom – for infectious pathogens is critical for controlling outbreaks and informing health management strategies to prevent the spread of infection
High-quality whole-g enome sequencing (WGS) data were available for 2129 Mycobacterium tuberculosis (Mtb) genomes collected between 1995 and 2014
The majority of isolates belonged to major lineage 4 (1279 isolates; 68.7 %), followed by lineages 1 (n=290; 15.6 %), 3 (n=217; 11.6 %) and 2 (n=71; 3.8 %) (Fig. 1a), with their proportions changing through the study period, but lineage 4 strains always being predominant (Fig. 1b)

Summary

Introduction

Establishing patterns of transmission – who infected whom – for infectious pathogens is critical for controlling outbreaks and informing health management strategies to prevent the spread of infection. Tuberculosis (TB), caused by Mycobacterium tuberculosis (Mtb), remains a major global health concern, responsible for 1.6 million deaths in 2017 alone, including nearly 400 000 deaths attributed to HIV-associated infection [1]. Despite initiatives aimed at reducing global incidence rates, such as the World Health Organization’s ‘End TB Strategy’ [1], many regions are falling behind set targets for case reduction. Understanding and preventing transmission is fundamental to disease control, but the accurate characterization of networks is difficult, especially as the onset of active disease follows a long and highly variable latency. There is a paucity of long-t erm studies in high-incidence areas [4,5,6,7] to assist with providing much needed biological and epidemiological insights into transmission

Objectives

Methods

Results

Discussion

Conclusion