A whole-genome sequencing approach to targeting Mycobacterium tuberculosis outbreak management

Abstract

BackgroundEpidemiological investigations into Mycobacterium tuberculosis outbreaks use 24-locus genotyping (MIRU-VNTR typing). Where no epidemiological link can be found between patients, the importance of shared genotypes remains unclear. This issue is especially problematic and time-consuming when tracing contacts within some social groups at high tuberculosis risk, in which unwillingness to volunteer information is common. We investigated whether whole-genome sequencing (WGS) could delineate outbreaks with greater resolution than MIRU-VNTR typing has done. MethodsWe sequenced 390 M tuberculosis isolates from 254 patients from the UK Midlands (1994–2011) using Illumina technology (appendix). We estimated the expected genomic diversity between isolates within a transmission chain by measuring pairwise nucleotide differences between genomes within hosts (79 individuals with pulmonary and extrapulmonary disease, or multiple pulmonary episodes) and between hosts within 25 household outbreaks (63 individuals). We then investigated 11 MIRU-VNTR-based community clusters (168 patients, 157 transmission events) to assess whether WGS could delineate outbreaks more effectively. For each cluster we reconstructed the most plausible transmission chain based on epidemiological data collected by tuberculosis nurses, pairwise nucleotide distances, and times of diagnosis, and compared the genomic diversity across these constructed links with that within individuals and within household outbreaks. Findings109 (96%) of 114 isolates were within five SNPs of another isolate taken from the same individual or from an individual in the same household outbreak. On the basis of longitudinal isolates from individuals or households, we estimated an evolutionary rate of 0·5 SNPs per genome per year, consistent with a maximum of five SNPs between related isolates 3 years or less apart. Using a greater than five SNP threshold to assess 11 MIRU-VNTR-based community clusters, we found that none of 69 epidemiologically related pairs of MIRU-VNTR-matched cases plausibly related by transmission, two of 13 possibly related pairs, and 13 of 75 pairs with no known epidemiological relation were separated by more than five SNPs (p<0·0001). Seven MIRU-VNTR-matched pairs with no epidemiological relation had more than 30 SNPs, five of seven belonging to the same immigrant community cluster. WGS also showed that 62 of 75 MIRU-VNTR-matched pairs for which no epidemiological relation had been identified from contact tracing were highly likely to indicate transmission: in one substance misuse cluster, 38 individuals were linked by five or fewer SNPs without a single epidemiological link having been established previously. Further analysis suggested that microevolutionary divergence of lineages within outbreaks could signal possible super-spreaders, corroborated by clinical and epidemiological data in two clusters. InterpretationWGS can delineate tuberculosis outbreaks with greater resolution than has previously been possible. These findings offer public health teams the potential to limit outbreak investigations to patients who are likely to be linked by recent transmission, irrespective of whether it has been possible to identify epidemiological links, and to save resources where they are not, even in the context of matched MIRU-VNTR genotypes. Uniquely, WGS also provides information about the genetic structure of outbreak clusters, thereby providing the potential to direct public health resources towards individuals most likely to have infected the largest number of secondary cases. As a consequence, the Health Protection Agency is considering introduction of WGS technology for routine tuberculosis public health practice in England. FundingNIHR Oxford Biomedical Research Centre and the UKCRC Modernising Medical Microbiology Consortium (UKCRC Translational Infection Research Initiative supported by MRC, Biotechnology and Biological Sciences Research Council, and NIHR on behalf of the Department of Health [grant G0800778] and the Wellcome Trust [087646/Z/08/Z]).