Phylogeography and transmission of Mycobacterium tuberculosis spanning prisons and surrounding communities in Paraguay

Abstract

Recent rises in incident tuberculosis (TB) cases in Paraguay and the increasing concentration of TB within prisons highlight the urgency of targeting strategies to interrupt transmission and prevent new infections. However, whether specific cities or carceral institutions play a disproportionate role in transmission remains unknown. We conducted prospective genomic surveillance, sequencing 471 Mycobacterium tuberculosis complex genomes, from inside and outside prisons in Paraguay's two largest urban areas, Asunción and Ciudad del Este, from 2016 to 2021. We found genomic evidence of frequent recent transmission within prisons and transmission linkages spanning prisons and surrounding populations. We identified a signal of frequent M. tuberculosis spread between urban areas and marked recent population size expansion of the three largest genomic transmission clusters. Together, our findings highlight the urgency of strengthening TB control programs to reduce transmission risk within prisons in Paraguay, where incidence was 70 times that outside prisons in 2021.

Fig. 1. Genomic surveillance within and outside prisons in Paraguay’s urban centers.

a Map of South America, with Paraguay highlighted, and as an inset. Paraguay’s departments are colored by population size density and points indicate the two largest urban centers in Paraguay, where we conducted focused genomic surveillance. b Notification rate of TB per 100,000 people in prisons (red) and in the general population (blue) from 2009 to 2020. Source data are provided as a Source Data file.

Fig. 2. M. tuberculosis isolates from incarcerated and non-incarcerated people are closely related across Paraguay.

A maximum likelihood phylogeny of 471 tuberculosis isolates from Lineage 4 inferred from a multiple sequence alignment of 9966 SNPs and rooted on three sub-lineage 4.8 samples from this study. Branch lengths are in units of substitutions per site. Branches are colored by sub-lineage. From the inside, rings are colored by antimicrobial resistance; city of sampling; and incarceration status at the time of TB notification. Other isoniazid includes isoniazid mono-resistant isolates without an ahpC promoter mutation. Rifampicin indicates rifampicin mono-resistant isolates. Other resistance includes isolates with mutations associated with resistance to pyrazinamide, streptomycin, or fluoroquinolones and not isoniazid or rifampicin. Source data are provided as a Source Data file.

Fig. 3. Genomic transmission clusters spanning prisons and neighboring communities have recently expanded.

a Effective population size (Ne) estimates for the three largest genomic clusters in our sample over time. Black lines indicate Ne inferred in a Bayesian Skyline Coalescent model and grey shading indicates 95% high posterior density estimates. b Median clade credibility trees inferred from the Bayesian Skyline Coalescent model. Branch lengths are in years and grey bars indicate 95% high posterior density estimates of node date. The heatmap to the right of the phylogeny indicates patient incarceration status at the time of TB notification. Source data are provided as a Source Data file.

Fig. 4. Frequent gene flow of M. tuberculosis connects Paraguay’s major urban centers.

a–c We used discrete ancestral state reconstruction to reconstruct migration between the two cities for the three dominant sub-lineages in our sample; lineages 4.1.2.1, 4.3.3, and 4.4.1.1. Bayesian maximum clade credibility trees of samples in the three dominant sub-lineages with tip points colored by city of sampling and pie charts at nodes indicating the inferred ancestral location. Branch lengths are in years and grey bars indicate 95% high posterior density estimates of node date. A model of asymmetric rates of movement between the two cities was supported for sub-lineages 4.1.2.1 a model of symmetric rates of movement was supported for 4.3.3 and 4.4.1.1. d Map of Paraguay with pie charts indicating the genomic diversity sampled in Asunción (at Paraguay’s western border) and Ciudad del Este (eastern border). Arrows are colored by sub-lineage and are weighted by the relative rate of migration between cities. Bi-directional arrows indicate equal rates of migration in each direction. Source data are provided as a Source Data file.

Fig. 5. Emergence of a putative resistance-associated ahpC promoter mutation.

Time-scaled Bayesian maximum clade credibility tree for 58 samples in sublineage 4.4.1.1. Branch lengths are in years and grey bars indicate 95% high posterior density estimates of node date. The heatmap to the right of the phylogeny indicates patient incarceration status at the time of TB notification, ahpC mutation frequency within an individual’s infection, and the occurrence of a rifampicin resistance-conferring mutation (rpoB His445Leu). Source data are provided as a Source Data file.