Prevalence and characterization of heterogeneous variablenumber tandem-repeat clusters comprising drug-susceptible and/or variable resistant mycobacterium tuberculosis complex isolates in the netherlands from 2004 to 2016

Abstract

The variable-number tandem-repeat (VNTR) typing method is used tostudy tuberculosis (TB) transmission. Clustering of Mycobacterium tuberculosis isolateswith identical VNTR patterns is assumed to reflect recent transmission. Hence, clusters are thought to be homogeneous regarding antibiotic resistance. In practice,however, heterogeneous clusters are also identified. This study investigates the prevalence and characteristics of heterogeneous VNTR clusters and assesses whether isolates in these clusters remain clustered when subjected to whole-genome sequencing (WGS). In the period from 2004 to 2016, 9,072 isolates were included.Demographic and epidemiological linkage data were obtained from the NetherlandsTuberculosis Register. VNTR clusters were defined as homogeneous when isolatesshared identical resistance profiles or as heterogeneous if both susceptible and (variable) resistant isolates were found. Multivariate logistic regression analysis was performed to identify factors associated with heterogeneous clustering. Isolates from2016 were subjected to WGS, and a genetic distance of 12 single nucleotide polymorphisms (SNPs) was used as the cutoff for WGS clustering. In total, 4,661/9,072(51%) isolates were clustered into 985 different VNTR clusters, of which 217 (22%)were heterogeneous. Patient characteristics associated with heterogeneous clustering were non-Dutch ethnicity (odds ratio [OR], 1.46 [95% confidence interval {CI},1.22 to 1.75]), asylum seeker (OR, 1.51 [95% CI, 1.24 to 1.85]), extrapulmonary TB(OR, 1.26 [95% CI, 1.09 to 1.46]), previous TB diagnosis (OR, 1.38 [95% CI, 1.04 to1.82]), and not being a contact of a TB patient (OR, 1.35 [95% CI, 1.08 to 1.69]). WithWGS, 34% of heterogeneous and 78% of homogeneous isolates from 2016 remainedclustered. Heterogeneous VNTR clusters are common but seem to be explained by asubstantial degree of false clustering by VNTR typing compared to WGS.