High-resolution mapping of tuberculosis transmission: Whole genome sequencing and phylogenetic modelling of a cohort from Valencia Region, Spain

Abstract

Background

Whole genome sequencing provides better delineation of transmission clusters in Mycobacterium tuberculosis than traditional methods. However, its ability to reveal individual transmission links within clusters is limited. Here, we used a 2-step approach based on Bayesian transmission reconstruction to (1) identify likely index and missing cases, (2) determine risk factors associated with transmitters, and (3) estimate when transmission happened.

Methods and findings

We developed our transmission reconstruction method using genomic and epidemiological data from a population-based study from Valencia Region, Spain. Tuberculosis (TB) incidence during the study period was 8.4 cases per 100,000 people. While the study is ongoing, the sampling frame for this work includes notified TB cases between 1 January 2014 and 31 December 2016. We identified a total of 21 transmission clusters that fulfilled the criteria for analysis. These contained a total of 117 individuals diagnosed with active TB (109 with epidemiological data). Demographic characteristics of the study population were as follows: 80/109 (73%) individuals were Spanish-born, 76/109 (70%) individuals were men, and the mean age was 42.51 years (SD 18.46). We found that 66/109 (61%) TB patients were sputum positive at diagnosis, and 10/109 (9%) were HIV positive. We used the data to reveal individual transmission links, and to identify index cases, missing cases, likely transmitters, and associated transmission risk factors. Our Bayesian inference approach suggests that at least 60% of index cases are likely misidentified by local public health. Our data also suggest that factors associated with likely transmitters are different to those of simply being in a transmission cluster, highlighting the importance of differentiating between these 2 phenomena. Our data suggest that type 2 diabetes mellitus is a risk factor associated with being a transmitter (odds ratio 0.19 [95% CI 0.02–1.10], p < 0.003). Finally, we used the most likely timing for transmission events to study when TB transmission occurred; we identified that 5/14 (35.7%) cases likely transmitted TB well before symptom onset, and these were largely sputum negative at diagnosis. Limited within-cluster diversity does not allow us to extrapolate our findings to the whole TB population in Valencia Region.

Conclusions

In this study, we found that index cases are often misidentified, with downstream consequences for epidemiological investigations because likely transmitters can be missed. Our findings regarding inferred transmission timing suggest that TB transmission can occur before patient symptom onset, suggesting also that TB transmits during sub-clinical disease. This result has direct implications for diagnosing TB and reducing transmission. Overall, we show that a transition to individual-based genomic epidemiology will likely close some of the knowledge gaps in TB transmission and may redirect efforts towards cost-effective contact investigations for improved TB control.

Author summary

Why was this study done?

• To facilitate public health intervention and to design new tuberculosis (TB) control strategies, there is a need to identify when TB is transmitted and by whom.

• Whole genome sequencing combined with phylogenetic modelling has the potential to fill knowledge gaps on TB epidemiology.

What did the researchers do and find?

• We analyzed a population-based cohort of TB patients in Valencia Region, Spain, between 2014 and 2016.

• We systematically sequenced the whole genomes of culture positive isolates and identified transmission clusters. We combined genomic and epidemiological data to understand how TB is transmitted.

• We showed that in many cases the index case is likely either not sampled or not the first diagnosed.

• For a fraction of TB individuals, we could accurately predict when transmission happened. For several transmitters we showed that transmission happened well before diagnosis and symptom onset.

What do these findings mean?

• Our findings provide novel insights into TB transmission going beyond cluster delineation.

• Our results highlight the limitation of contact tracing to identify index cases and can be used to design new TB control strategies.

• The finding that TB can be transmitted by individuals before they have symptoms, very likely during sub-clinical or incipient disease, has important implications for TB control strategies. More studies are needed to understand the dynamics of TB transmission in different clinical settings.

• This study highlights the importance of combining genomic data with epidemiological data, in order to gain new insight into how and when TB is transmitted.