Abstract
The challenges associated with diagnosing tuberculous pleural effusion have led to the use of pleural fluid adenosine deaminase (pfADA) as a biomarker for Mycobacterium tuberculosis infection. This study retrospectively reviewed the diagnostic performance of pfADA, the pleural fluid lactate dehydrogenase (LD)/ADA ratio, and combinations of these two parameters in 1,637 episodes of pleural effusion in the low-tuberculosis (TB)-incidence setting of Auckland, Aotearoa New Zealand, from between March 2008 and November 2014. The median pfADA in 57 TB pleural effusion episodes (58.1 U/liter) was significantly higher (P < 0.001) than in 1,580 non-TB pleural effusions (11.4 U/liter). The median LD/ADA ratio in TB pleural effusion (8.2) was significantly lower (P < 0.001) than in non-TB pleural effusions (30.5). The pfADA and pleural fluid LD/ADA ratio AUC(ROC) values (that is, receiver operating characteristic [ROC] curve analysis with determination of the ROC area under the curve) were 0.93 and 0.94, respectively. The pfADA thresholds of ≥15 and ≥30 U/liter demonstrated sensitivities of 100% (95% confidence internal = 93.7 to 100) and 93.0% (83.3 to 97.2), specificities of 62.7% (60.3 to 65.0) and 87.3% (85.6 to 88.9), positive predictive values (PPVs) of 8.8% (6.9 to 11.2) and 20.9% (16.4 to 26.4), and negative predictive values (NPVs) of 100% (99.6 to 100) and 99.7% (99.3 to 99.9), respectively. LD/ADA ratio thresholds of <25 and <15 demonstrated sensitivities of 100% (93.5 to 100) and 89.1% (78.2 to 94.9), specificities of 61.6% (59.1 to 64.0) and 84.8% (82.9 to 86.5), PPVs of 8.5% (6.6 to 10.9) and 17.3% (13.3 to 22.0), and NPVs of 100% (99.6 to 100) and 99.5% (99.0 to 99.8), respectively. A combination of pfADA ≥ 30 U/liter and an LD/ADA ratio < 15 increased the specificity and PPV to 97.8% (96.9 to 98.4) and 57.3% (46.5 to 67.5) but decreased the sensitivity to 85.5% (73.8 to 92.4). The primary value of pfADA in a low-TB-incidence setting, such as Auckland, is in utilization of its high NPV.