Background:
Early diagnosis of tuberculous pleural effusion (TPE) remains difficult, especially in hospitalization. While some inflammatory markers in pleural effusion (PE) are helpful in diagnosis, the implication of anti‐inflammatory cytokines and effector molecules of cytotoxic T lymphocytes have never been investigated.
Methods:
Lymphocyte‐predominant exudative PE samples were assayed for inflammatory and anti‐inflammatory cytokines, and effector molecules of cytotoxic T lymphocyte. The results were transformed into bi‐level categorical variables based on the determined cutoff values. Logistic regression analysis was applied to identify factors independently associated with TPE and to predict probability of TPE. Receiver operation characteristic (ROC) curve analysis was then applied to determine the optimal cutoff value for the predicted probability.
Results:
Of 95 patients enrolled, 35 had TPE, 46 malignant PE, and the remaining 14 had PE due to other etiologies. Logistic regression analysis revealed that interferon‐g ≥75 pg/ml, adenosine deaminase (ADA) ≥40 IU/ml, decoy receptor (DcR) 3 ≥9.3 ng/ml and tumor necrosis factor‐soluble receptor 1 (TNF‐sR1) ≥3.2 ng/ml were independent factors associated with TPE. The predicted probability based on the four predictors has an area under ROC curve of 0.920 (sensitivity: 82.9%; specificity: 86.7%). Among TPE group, patients with positive PE culture for Mycobacterium tuberculosis had higher pleural intereron‐g, monocyte chemoattractant protein‐1 and perforin than those with positive sputum but negative PE culture.
Conclusions:
While pleural IFN‐g and ADA (inflammatory biomarkers) are conventional markers for diagnosing TPE, simultaneously measuring DcR3 and TNF‐sR1, the two anti‐inflammatory biomarkers, can improve diagnostic efficacy.