Diagnóstico da tuberculose pleural utilizando ensaios de produção de interferon gama, proteína indutora do interferon 10-kD, adenosina deaminase e modelagem de árvores de classificação

Abstract

The nosological entity called pleural effusion (PE) presents more than 50 etiologies, being heart failure, cancer, pneumonia and pleural tuberculosis (PlTB) the most frequent causes. Difficulties in determining the cause of PE are a reality whether in the primary health care setting or in tertiary centers. In spite of its frequency, PlTB has a great difficulty in its diagnosis because it is a paucibacillary presentation and because of the need for invasive procedures for its confirmation, which are sometimes contraindicated or not performed due to the lack of structure. This thesis aims to offer two diagnostic strategies for PlTB. The results were presented in the form of two articles, both submitted to periodicals with concepts B1 by the Sucupira Platform. The first strategy, presented in article 1, study three biomarkers quantified in the pleural fluid (PF) of patients with PE under investigation: interferon-gamma (IFN-γ), interferon-inducible protein 10-kD (IP-10) from supernatants of IFN-γ release assay QuantiFERON-TB Gold in Tube®, and adenosine deaminase (ADA), measured according to Giusti, 1974. All these markers showed higher values in patients with PlTB when compared with patients with diseases other than TB. Based on confirmed PlTB cases, IFN-γ showed the best performance at a cut-off point of 2.33 IU/mL [sensitivity (Se) = 93.8% and specificity (Sp) = 97.9%) followed by ADA with a cut-off point of 25.8 IU/L (Se = 100% and Sp = 84.8%) and IP-10 (cut-off = 4,361.9 pg/mL, Se = 75% and Sp = 82.6%). IFN-γ + ADA (cut-off: 25.8 UI/L) represented the most accurate association of biomarkers (98.4%) showing Se = 93.7%, Sp = 100%, positive predictive value = 100% and negative predictive value = 97.9%, and suggesting this combination as an excellent predictor for PlTB diagnosis. The second article presents as strategy a classification tree model using simple clinical and laboratory variables for the diagnosis of tuberculous PE. The proposed predictive model used three clinical and laboratorial variables that could be used even a basic health units (hyporexia, polymorphonuclear percentage and protein dosage in PF), identified by multivariate analysis [hyporexia: adjusted odds ratio (aOR) 27.39 (95% CI 6.26 - 119.89), polymorphonuclear in two categories: 3-14% aOR 26.22, 95% CI 7.11 - 96.68 and < 3% aOR 28.67, 95% CI 5.51 - 149.25 and proteins &#8805; 5g/dL: aOR 7.24, 95% CI 3.07-17.11]. The yield of this model in the cases of confirmed PlTB revealed values of Se, Sp and accuracy of 89.2%, 84.5% and 87.6%, respectively. Both strategies proved useful for the diagnosis of TB, respecting the more limited structures of the scenarios where they could be applied, suggesting that despite the difficulty of diagnosing PlTB, this challenge may be on the way to being overcome.