Transmission analysis of a large tuberculosis outbreak in London: a mathematical modelling study using genomic data

Abstract

Outbreaks of tuberculosis (TB) – such as the large isoniazid-resistant outbreak centred on London, UK, which originated in 1995 – provide excellent opportunities to model transmission of this devastating disease. Transmission chains for TB are notoriously difficult to ascertain, but mathematical modelling approaches, combined with whole-genome sequencing data, have strong potential to contribute to transmission analyses. Using such data, we aimed to reconstruct transmission histories for the outbreak using a Bayesian approach, and to use machine-learning techniques with patient-level data to identify the key covariates associated with transmission. By using our transmission reconstruction method that accounts for phylogenetic uncertainty, we are able to identify 21 transmission events with reasonable confidence, 9 of which have zero SNP distance, and a maximum distance of 3. Patient age, alcohol abuse and history of homelessness were found to be the most important predictors of being credible TB transmitters.

Methods Combining Genomic and Epidemiological Data in the Reconstruction of Transmission Trees: A Systematic Review

“…As such, these methods ignored any uncertainty in the estimation of the phylogeny [18] and therefore did not take the full uncertainty in the evolutionary process into account [26]. Thus, the Transphylo package was extended to reconstruct transmission trees from multiple phylogenetic trees [55]. However, another strategy was to infer transmission trees and phylogenetic trees simultaneously; we grouped these methods in the simultaneous phylogenetic family.…”

“…The Transphylo package was chosen to study (Table S4) bacterial transmission (such as M. tuberculosis [48][49][50] and K. pneumoniae outbreaks [51,52]), as well as viral transmission (e.g., part of the recent SARS-CoV-2 pandemic [53] and a large mumps outbreak in Canada [54]). Recently, the Transphylo package [40] was extended to infer transmission trees from multiple phylogenetic trees [55].…”

“…None of these transmission tree reconstruction methods explicitly modeled sequence mutation since the method is applied to an already fully reconstructed phylogenetic tree (hence the "not applicable" in Table 3). However, some articles [39,40,48,51,[53][54][55] have used substitution models to reconstruct the phylogenetic tree prior to the implementation of their method.…”

“…Among these five methods, Transphylo, outbreaker, and outbreaker2 could allow for a long latency period by selecting an appropriate generation time distribution (Table S6), as has previously been demonstrated with the Gamma generation time density in Transphylo [40]). M. tuberculosis outbreaks have been reconstructed using five methods from phylogenetic and non-phylogenetic families: Seqtrack (NPF) [46], Didelot et al (2014) [17], Eldholm et al (2016) [39], and Transphylo [40,[48][49][50]55] from the SeqPF and phybreak (SimPF) [26] (Tables S3-S5).…”

Comparing transmission reconstruction models with Mycobacterium tuberculosis whole genome sequence data

“…To allow for the presence of non-sampled hosts within clusters, we randomly down-sampled the Epidemiological parameters for the simulated clusters were chosen to reflect clinical Mtb clusters, thus we chose the same gamma generation and sampling time distributions as described previously. The sequence length was set as length of the H37Rv Mtb reference stain at 4.4M base pairs and we allowed for within-host evolution at the rate of 1.48 as previously estimated 16,40,44 and a mutation rate of 0.5 SNPs per genome per year, taken from estimates in our clinical data using BEAST2 v2.6.3 35 , which is in line with previous estimates of the Mtb mutation rate 45 . The simulator takes these epidemiological estimates and produces simulated sequence data and transmission networks, including the time and direction of transmission.…”