Introduction The assumption that infection with M tuberculosis is life-long in the absence of treatment underlies the “statistic” that one third of people worldwide are currently infected. Exploration of the validity of this statement and the contribution of this reservoir to later disease are critical, given the WHO target of elimination of TB by 2050. HIV-positive individuals are assumed to be at high risk of “remotely” acquired M tuberculosis infection when their immunity declines. Understanding individuals who clear or control infection is key to identifying correlates of protection and requires a well defined long term epidemiological cohort.

Methods A cohort of 2000 individuals with strong evidence of M tuberculosis infection (tuberculin skin test [TST]>20 mm without active disease at baseline) in Malawi were followed up after 25 years. Outcomes (active TB, death, departure) were recorded. A sample of survivors were investigated for HIV status and CD4 count, TST and commercial interferon gamma release assay (T-Spot.TB). A sub-sample were re-tested immunologically after isoniazid treatment.

Results Life-time risk of active TB was approximately 8%, with highest incidence in the first 2 years. We identified groups who appear to have resolved infection: HIV-positive immunocompromised survivors with no evidence of TB, and HIV-negative individuals with complete absence of characteristic immunological responses. Some individuals showed exaggerated immune response indicating ongoing immune stimulation.

Conclusions Latent M tuberculosis infection may resolve without treatment and natural, identifiable, “protected” phenotypes may exist. These findings have implications for both control of the reservoir of latent infection and the potential for evaluation of vaccines.