Diagnostic performance of circulating microRNA signatures for differentiating tuberculosis disease from tuberculosis infection.

As regulators of innate and adaptive immunity, microRNAs (miRNAs) could aid in the discrimination between tuberculosis disease (TB) and (latent) TB infection (TBI). We analysed 754 circulating miRNAs in participants diagnosed with TB and TBI using TaqMan™ Advanced miRNA Human A and B cards. MiRNAs were normalized exogenously and endogenously via geometric means of selected reference miRNAs. Expression analysis was used to identify miRNAs that were significantly differentially expressed between individuals with TB and those with TBI. We utilised recursive feature elimination with a Random Forest model to identify the miRNAs most effective at discriminating TB from TBI and subsequently validated the miRNA in another group. 95 persons diagnosed with TB or TBI was divided into a discovery group (n = 36) and a validation group (n = 59). In the discovery group, we identified 495 distinct miRNAs in 36 persons with TB or TBI and by recursive feature elimination identified hsa-miR-148a-3p, hsa-miR-204-5p and hsa-miR-584-5p and created a three-miRNA-diagnostic model. In the validation group, the three-miRNA-diagnostic model had poorer performance. Expression analysis revealed 13 significantly differentially expressed miRNAs, including hsa-miR-148a-3p and hsa-miR-204-5p. Subsequent analysis in a validation group consisting of 59 persons revealed that six of the 14 miRNAs, including hsa-miR-148a-3p, exhibited the same pattern, albeit without statistical significance. Three circulating miRNAs showed potential for differentiating TB from TBI in the discovery cohort, but these differences were less pronounced in the validation cohort.