Measuring recent in-country TB transmission using a classification model with whole genome sequencing data

Molecular typing of Mycobacterium tuberculosis isolates provides insight into TB transmission by identifying clustering isolates. However, clustering alone does not necessarily indicate ongoing in-country transmission, as infections may have occurred abroad or long ago.

We developed a classification model categorising clustering TB patients into three groups: i) likely, ii) possibly, and iii) unlikely to have been recently infected in the Netherlands. The model consists of two stages: i) individual labelling, followed by ii) cluster labelling for the remaining cases, and includes epidemiological, whole genome sequencing (WGS) data, and confirmed epidemiological links based on cluster investigation.

We found that 28% of TB patients with WGS results in 2018-2023 had a clustered isolate. However, when we classified cases by individual labelling (48% of cases) and cluster labelling (52% of cases), only 11% to a maximum of 15% were likely to have been infected within the last 2 years in the Netherlands.

Our WGS-based classification model provides a valuable tool for monitoring progress towards TB pre-elimination by enabling estimation of recent in-country transmission. Importantly, our findings indicate that a substantial proportion of clustered cases were likely the result of infections acquired abroad or from non-recent transmission events.