Application of Venn's diagram in the diagnosis of pleural tuberculosis using IFN-γ, IP-10 and adenosine deaminase

Abstract

Background: Pleural tuberculosis (PlTB) is the most common extrapulmonary manifestation of this infectious disease which still presents high mortality rates worldwide. Conventional diagnostic tests for PlTB register multiple limitations, including the lack of sensitivity of microbiological methods on pleural specimens and the need of invasive procedures such as pleural biopsy performance. In this scenario, the search for biological markers on pleural fluid (PF) has been the target of several studies as a strategy to overcome the limitations of PlTB diagnosis. This study aims to evaluate the use either isolated or in combination with adenosine deaminase (ADA), interferon-gamma (IFN-γ), interferon-gamma inducible protein of 10-kD (IP-10) levels on PF in order to guide an accurate anti-TB treatment in microbiologically non-confirmed cases.

Methods and findings: Eighty patients presenting pleural effusion under investigation were enrolled in a cross-sectional study conducted at Pedro Ernesto University Hospital, Rio de Janeiro, RJ, Brazil. Peripheral blood (PB) and PF samples collected from all patients were applied to the commercial IFN-γ release assay, QuantiFERON-TB Gold In-Tube, and samples were analyzed for IFN-γ and IP-10 by immunoassays. ADA activity was determined on PF by the colorimetric method. Based on microbiological and histological criteria, patients were categorized as follow: confirmed PlTB (n = 16), non-confirmed PlTB (n = 17) and non-PlTB (n = 47). The Mycobacterium tuberculosis antigen-specific production of IFN-γ and IP-10 on PB or PF did not show significant differences. However, the basal levels of these biomarkers, as well as the ADA activity on PF, were significantly increased in confirmed PlTB in comparison to non-PlTB group. Receiver operating characteristics curves were performed and the best cut-off points of these three biomarkers were estimated. Their either isolated or combined performances (sensitivity [Se], specificity [Sp], positive predictive value [PPV], negative predictive value [NPV] and accuracy [Acc]) were determined and applied to Venn's diagrams among the groups. Based on the confirmed PlTB cases, IFN-γ showed the best performance of them at a cut-off point of 2.33 IU/mL (Se = 93.8% and Sp = 97.9%) followed by ADA at a cut-off of 25.80 IU/L (Se = 100% and Sp = 84.8%) and IP-10 (Cut-point = 4,361.90 pg/mL, Se = 75% and Sp = 82.6%). IFN-γ plus ADA (cut-point: 25.80 IU/L) represent the most accurate biomarker combination (98.4%), showing Se = 93.7%, Sp = 100%, PPV = 100% and NPV = 97.9%. When this analysis was applied in non-confirmed PlTB, 15/17 (88.2%) presented at least two positive biomarkers in combination.

Conclusion: IFN-γ, IP-10, and ADA in PlTB effusions are significantly higher than in non-PlTB cases. IFN-γ is an excellent rule-in and rule-out test compared to IP-10 and ADA. The combination of IFN-γ and ADA, in a reviewed cut-off point, showed to be particularly useful to clinicians as their positive results combined prompts immediate treatment for TB while both negative results suggest further investigation.

Fig 1. Flow chart of the study design and diagnostic testing performed.

QFT-GIT: QuantiFERON-TB Gold in-Tube; AFB: acid-fast bacilli; ADA: adenosine deaminase; TB: Tuberculosis; UND: undefined diagnosis. Numbers in parenthesis refer to the patients submitted to the correspondent diagnostic test. Gray boxes show the final diagnosis and the study groups.

Fig 2. Concentration of IFN-γ, IP-10, and ADA in pleural fluid.

(A) IFN-γ and (B) IP-10 were measured in the supernatants of the QFT-GIT system from PF and (C) ADA activity was measured directly in PF. Obtained levels from each biomarker were analyzed in a logarithmic scale and illustrated using boxplots to compare the groups: Non-PlTB (N = 46) and confirmed PlTB (n = 16). Small black dots represent individual cases and box plots represent the interquartile range and sample median (central solid gray line). Bigger black dots and vertical bars represent linear model estimated adjusted means and 95% confidence intervals (CI 95%). Comparisons of means among groups were performed by contrasts/differences obtained after both bi- and multivariate linear models fitted by ordinary least square regressions. The confounders for the parameters illustrated above were: (A) Nil/control (unstimulated) IFN-γ: gender, previous TB, cancer, chest X-ray, protein level, PMN percentage and MN percentage; Ag-Nil IFN-γ: gender, renal failure, and albumin; Mit-stimulated IFN-γ: smoke, chest X-ray, DHL, and albumin; (B) Nil/control (unstimulated) IP-10: previous TB, protein level, PMN percentage, and MN percentage; Ag-Nil IP-10: age, renal failure, and albumin; Mit-stimulated IP-10: previous TB, cancer, smoke, DHL, and albumin; and, (C) for ADA: protein level, PMN percentage, and MN percentage. Nil: Negative/control tube; Ag-Nil: Mtb-specific antigens minus Nil; Mit: Mitogen. * p = 0.001; ** p < 0.0001.

Fig 3. Receiver operating curves of ADA, IFN-γ, and IP-10 in pleural fluid.

Receiver operating curves were plotted and each biomarker had its area under the curve (AUC) calculated. The best cut-off value for PlTB diagnosis was established according to Youden’s Index. The statistical analysis was performed using confirmed PlTB (N = 16) and non-PlTB (N = 46).

Fig 4. Venn’s diagrams application on PlTB diagnosis using ADA_{(25.8 IU/L)}, IFN-γ and IP-10 positivity.

A, B and C Venn’s diagrams show the performance of the biomarkers positivity in three different groups of the study: confirmed PlTB (N = 16), non-PlTB (N = 46) and non-confirmed PlTB (N = 17), respectively. IFN-γ: interferon-gamma; IP-10: interferon-gamma inducible protein of 10-kD; ADA: adenosine deaminase; PlTB: Pleural tuberculosis. Numbers indicate the intersection positivity of the biomarkers.