Summary
A statistical model estimated that testing for recent tuberculosis infection could reduce the median number of people needed for a vaccine trial by 4.2- to 6-fold compared with long-lasting IGRA positivity under favourable conditions. The estimates are preprint model results, and the authors say candidate tests still need to be characterised.
A statistical model suggests that testing for recent Mycobacterium tuberculosis infection could reduce the number of people needed to recruit for some tuberculosis vaccine trials. In a medRxiv preprint posted October 21, 2026, researchers estimated a 4.2- to 6-fold reduction in median enrolment compared with a strategy based on long-lasting positive interferon-gamma release assays (IGRAs), under favourable conditions. The model aimed to recruit enough participants to observe 50 tuberculosis disease endpoints.
Why the timing of infection matters
Many people infected with M. tuberculosis do not progress to tuberculosis disease. Trials that rely on disease cases as endpoints can therefore need large participant groups and long follow-up to accumulate enough cases.
IGRAs detect an immune response to tuberculosis bacteria, but a positive result can persist for years and does not show whether infection is recent or remote. The researchers propose that a test identifying recent infection could help trial organisers select participants more likely to contribute disease endpoints during the trial period.
Tests for recent infection are in development. The model treated the proposed test as remaining positive for two years, and compared it with IGRA-based enrolment, modelled as having lifelong positivity. These are assumptions for the recruitment analysis, not results from testing a diagnostic in trial participants.
What the recruitment model found
The researchers modelled how many people would need to be contacted, tested and enrolled in South Africa, Kenya and India. They compared five approaches: recruiting people aged 15–49, recruiting those aged 12–34, recruiting household contacts of recent tuberculosis cases, enrolling people with positive IGRAs, and using a test for recent infection.
For the same target of 50 disease endpoints, the recent-infection approach reduced the median number enrolled by 4.2- to 6-fold compared with the long-duration-positive test strategy under favourable conditions. It also performed better in the model than the non-test-based strategies, including recruiting household contacts. Achieving that smaller enrolment would require substantially more testing.
A crude cost estimate in the preprint suggests the approach could still cost less overall if the new test costs no more than roughly five to ten times as much as an IGRA. That estimate depends on the modelled balance between testing more people and enrolling fewer participants.
The findings are model-based estimates in a preprint, rather than results from a vaccine trial using a recent-infection test. The authors say further work is needed to establish candidate tests’ time to positivity, duration of positivity, sensitivity and specificity, and to evaluate their use in this setting.