Plasma host protein signatures correlating with Mycobacterium tuberculosis activity prior to and during antituberculosis treatment

Abstract

Abstract Background There is a need for rapid non-sputum-based tests to identify and treat patients infected with Mycobacterium tuberculosis (Mtb). In this study, performance of a human plasma protein signature for both TB triage and treatment monitoring has been evaluated. Methods A panel of seven host proteins CLEC3B, SELL, IGFBP3, IP10, CD14, ECM1 and C1Q were measured in the plasma of a prospective patient cohort undergoing anti-tubercular therapy to distinguish confirmed TB patients from control, to define culture slow and fast converters during an antituberculosis treatment and to monitor the treatment. To validate the protein signatures a Luminex xMAP® assay was used to quantify the proteins in unstimulated plasma isolated from the blood collected from HIV-negative pulmonary TB patients (ATB), at baseline and following 6-months of antituberculosis treatment, latently TB-infected individuals (LTBI) and healthy donors (HD). Protein signatures performances were evaluated using CombiROC algorithm and multivariate models. Findings: We measured the proteins in the plasma of 84 participants including 37 ATB, 24 LTBI and 23 HD. The seven plasma host proteins studied showed different levels between the TB clinical groups and when used separately or in combination, have variable performances for ATB triage and for treatment monitoring. Six of the plasma proteins (CLEC3B, SELL, IGFBP3, IP10, CD14 and C1Q) showed significant differences in normalised median fluorescence intensities when comparing ATB vs HD or LTBI while ECM1 revealed a significant association with early sputum culture conversion after 2 months following treatment (OR 0.004, p = 0.018). For both triage and treatment monitoring, a signature combining 4 host proteins markers (CLEC3B-ECM1-IP10-SELL) was identified, allowing to distinguish ATB from HD or LTBI (respectively, sensitivity = 94%, specificity = 92% and sensitivity = 89%, specificity = 91%) as well as the baseline distinction between slow vs fast sputum culture converters after two months of treatment (AUC = 0.87, sensitivity = 83%, specificity = 84%). Conclusion Our data demonstrate that using a combination of host plasma markers can generate a relevant biosignature for both TB triage and treatment monitoring meeting the WHO Target Product Profile for both purposes.