Busca e validação de assinatura gênica em amostras clínicas de pacientes com derrame pleural por tuberculose a partir da análise in silico de dados públicos transcricionais

Abstract

Pleural tuberculosis (PlTB), the main cause of extrapulmonary TB among immunocompromised individuals, presents paucibacillary characteristics and compartmentalized inflammatory response in the pleura. Previous studies show a transcriptional signature for pulmonary and extrapulmonary TB from peripheral blood samples. However, this approach has not been performed on clinical specimens originating from the infectious site of TBPl, in particular. In the present study, we evaluated a set of genes previously described in pulmonary TB in order to identify and validate a transcriptional signature for PlTB in peripheral blood (SP) and pleural fluid (PF) samples among patients with exudative pleural effusion due to TB and non-TB causes. Initially, we sought a set of target genes, differentially expressed between pulmonary TB and non-TB samples, selected by the “Random Forest” algorithm, using recursive feature elimination (RFE) applied to public data from DNA microarrays by Bloom et al., 2013. Then, clinical samples from patients with exudative pleural effusion, prospectively recruited from the Pneumology and Phthisiology Service of the Pedro Ernesto University Hospital-HUPE/UERJ (RJ, Brazil), were collected in PAXgene tubes during the thoracentesis procedure. Transcriptional analysis of previously selected genes was performed by quantitative RT-PCR (RT-qPCR). From public data, a ranking of 10 candidate genes (“Top-10”: CARD17, BHLHE40, FCGR1A, BATF2, STAT1, BTN3A1, ANKRD22, C1QB, GBP2 and SEPTIN4) was identified, which had an accuracy greater than 89, 5% to distinguish cases of pulmonary TB from other respiratory diseases. Our cohort of patients consisted of 41 men and 28 women, aged between 18 and 92 years, grouped into pl-TB (n=35) and non-TB (n=34). Our data showed that CARD17, GBP2 and C1QB showed a significant difference in gene expression between PlTB and non-TB groups in LP samples (p<0.001). Furthermore, the accuracy of these three highlighted genes, also in LP, was AUC=0.91; 0.90; 0.85, respectively. We observed that CARD17 and GBP2 gene expressions were higher in PlTB LP than in non-TB patients. GBP2 showed sensitivity and specificity above 80% (sens=0.89; spec=0.81), and CARD17 showed significant specificity (sens=0.70; spec=0.95) in group discrimination. C1QB expression was inverse, being higher in non-TB LP. After anti-TB therapy, GBP2 showed a significant reduction in gene expression (p<0.001). In conclusion, we identified that CARD17, GBP2 and C1QB are promising to discriminate PlTB and other causes of exudative pleural effusion, being possible to contribute to accurate diagnosis and rapid initiation of anti-TB therapy. In addition, our data provide a better understanding of the pathophysiological mechanisms of the disease in the search for new therapeutic targets.