A hypothesis-generating study of collateral reactivation: a reanalysis of Mycobacterium tuberculosis genomes from 119 recurrent tuberculosis cases.

Recurrent tuberculosis is traditionally classified as exogenous reinfection or endogenous reactivation. We previously proposed "collateral reactivation," in which distinct Mycobacterium tuberculosis subclones derived from a hypothetical original clone are involved in separate disease episodes, in contrast to traditional endogenous ("direct-descent") reactivation. To explore whether collateral reactivation occurs beyond the single case we previously reported, we reanalyzed publicly available whole-genome sequencing data from 123 recurrent tuberculosis cases. After curation, 119 cases (pairwise single-nucleotide variant [SNV] distance, 0-18) were classified as cases of collateral or direct-descent reactivation. In 21 cases (17.6%), a hypothetical original clone could be inferred, and these cases were classified as collateral reactivation. These cases showed a higher rate of SNVs/genome/year (median 3 vs. 0.4, p < 0.0001) and a higher proportion of drug resistance-associated SNVs (23.8% vs. 4.1%, p = 0.009) than direct-descent reactivation cases. These observations support the view that collateral reactivation is not restricted to the single case we previously reported, although mixed infection or exogenous reinfection caused by closely related strains cannot be excluded as alternative explanations. To determine whether collateral reactivation should be recognized as a concept distinct from direct-descent reactivation, future studies integrating whole-genome sequencing of single colonies, lesion-specific sampling, longitudinal sampling, deep sequencing, imaging, clinical follow-up, and epidemiological exposure information will be required.