Transmission dynamics study of tuberculosis isolates with whole genome sequencing in southern Sweden

Abstract

Epidemiological contact tracing complemented with genotyping of clinical Mycobacterium tuberculosis isolates is important for understanding disease transmission. In Sweden, tuberculosis (TB) is mostly reported in migrant and homeless where epidemiologic contact tracing could pose a problem. This study compared epidemiologic linking with genotyping in a low burden country. Mycobacterium tuberculosis isolates (n = 93) collected at Scania University Hospital in Southern Sweden were analysed with the standard genotyping method mycobacterial interspersed repetitive units-variable number tandem repeats (MIRU-VNTR) and the results were compared with whole genome sequencing (WGS). Using a maximum of twelve single nucleotide polymorphisms (SNPs) as the upper threshold of genomic relatedness noted among hosts, we identified 18 clusters with WGS comprising 52 patients with overall pairwise genetic maximum distances ranging from zero to nine SNPs. MIRU-VNTR and WGS clustered the same isolates, although the distribution differed depending on MIRU-VNTR limitations. Both genotyping techniques identified clusters where epidemiologic linking was insufficient, although WGS had higher correlation with epidemiologic data. To summarize, WGS provided better resolution of transmission than MIRU-VNTR in a setting with low TB incidence. WGS predicted epidemiologic links better which could consolidate and correct the epidemiologically linked cases, avoiding thus false clustering.

Figure 1

Flow chart of sample collection. Graph show number of isolates/sequences included in the study.

Figure 2

Phylogenetic tree constructed by pairwise distance (maximum likelihood) of 93 Mycobacterium tuberculosis isolates including CDC1551 as reference genome.

Figure 3

ML SNP tree with CLC. A ML SNP-tree of the five largest clusters in the study, comprising of 26 patient isolates was created using the reference genome CDC1551. Within the four smaller clusters the strains differ by 0–2 SNP in a genetic pairwise comparison. In the largest cluster the patient isolates are separated by 0–7 SNP.