Genome-wide expression for diagnosis of pulmonary tuberculosis: a multicohort analysis

摘要: Summary Background Active pulmonary tuberculosis is difficult to diagnose and treatment response effectively monitor. A WHO consensus statement has called for new non-sputum diagnostics. The aim of this study was use an integrated multicohort analysis samples from publically available datasets derive a diagnostic gene set in the peripheral blood patients with active tuberculosis. Methods We searched two public expression microarray repositories retained that examined clinical cohorts infection whole blood. compared either latent or other diseases versus using our validated framework. Three were used as discovery meta-analytical methods assess effects these cohorts. then capacity three remaining 11 datasets. Findings total 14 containing 2572 10 countries both adult paediatric included analysis. Of these, (N=1023) discover genes ( GBP5, DUSP3 , KLF2 ) are highly power separate healthy controls (global area under ROC curve (AUC) 0·90 [95% CI 0·85–0·95]), (0·88 [0·84–0·92]), (0·84 [0·80–0·95]) eight independent composed children adults ten countries. Expression three-gene not confounded by HIV status, bacterial drug resistance, BCG vaccination. Furthermore, four additional cohorts, we showed score declined during Interpretation Overall, yielded robustly tuberculosis, multiple potential application diagnosis monitoring response. Prospective laboratory validation will be required before it can setting. Funding National Institute Allergy Infectious Diseases, Library Medicine, Stanford Child Health Research Institute, Society University Surgeons, Bill Melinda Gates Foundation.