The diagnostic utility of pleural fluid adenosine deaminase for tuberculosis in a low prevalence area.

Abstract

Pleural fluid adenosine deaminase (pfADA) is not routinely measured in patients with undiagnosed pleural effusion due to limited evidence of its diagnostic utility in areas of low tuberculosis (TB) prevalence. We conducted a retrospective consecutive case series analysis of all patients who underwent pfADA testing from 2009 to 2015 at a tertiary service pleural centre in south London. Using receiver operating characteristic (ROC) curve analysis, we identified the optimal threshold at which maximal sensitivity and specificity were achieved. Of the 132 patients tested for pfADA, 27 had confirmed pleural TB and 105 did not, with median pfADA levels of respectively 63 IU/l (interquartile range [IQR] 47-88) and 12 IU/l (IQR 7.5-22.5). ROC curve analysis determined the optimal pfADA cut-off to be 30 IU/l, which had positive and negative predictive values of respectively 60.5% and 98.9%, 96.3% sensitivity (95%CI 0.892-1.000) and 83.8% specificity (95%CI 0.768-0.909). The calculated area under the ROC curve was 0.934 (95%CI 0.893-0.975). A pfADA level <30 IU/l makes a diagnosis of TB highly unlikely in the South London population. Its high sensitivity and negative predictive values make pfADA a valuable screening test for excluding suspected pleural TB.