Work

Validation of blood transcriptional signatures for tubeculosis infection in Thai population

Year: 2012

Type: article

Abstract: Background: Tuberculosis is one of the major global health problems, an estimate of 4,700 deaths per day is related to Tuberculosis. Improvement in diagnostic tools, treatment and prevention methods are keys to the success in combating with tuberculosis. Recently, transcriptome analysis of blood samples identified hundreds of gene transcripts associated with active tuberculosis. We used published transcriptomic data set available in public data base and re-analyze to narrow down to a small number of independent transcript set and identify top gene expression signatures that together can differentiate active tuberculosis from healthy individuals. Methods: The minimum list of gene expression were selected from 393 transcripts based on three criteria 1) based on the feature selection methods called “Minimum redundancy feature selection from microarray gene expression data (mRMR)”, 2) Transcipt with highest differentiated power for identifying active tuberculosis from three datasets (training, test and validate data sets from Berry et al.) 3) Genes with evidence of association from genome-wide association analysis of tuberculosis in Thais. Based on these 3 criteria, 12 TB-exclusive transcripts were selected to be tested for their association with active tuberculosis status in Thais. We employed a multi-target TaqMan gene expression assay to validate transcriptional level of the 12 target genes and 2 internal control genes HPRT1 and GAPDH. Blood samples were taken from 38 active TB patients during their first 2 month of anti-TB treatment and 36 healthy individuals were recruited. Quantiferon assays (QFT) were done in both TB patients and healthy controls in order to confirm TB infection status. Healthy controls with QFT positive were excluded from the association analysis. Results: From the 12 TB-blood signature transcripts, genes that are most differentially expressed between active TB as compared to normal individual are FCGR1B variant 1 (p = 1.49 × 10-8), STAT1(p = 5.66 × 10-7), FCGR1A (p = 1.82 × 10-5) and GBP5 (2.91 × 10-5). Conclusion: These gene expression patterns have potential for diagnosis of active TB especially when microbiological confirmation is not available.

Source: International Journal of Infectious Diseases

DOI: 10.1016/j.ijid.2012.05.969