Diagnostic value of pleural fluid adenosine deaminase in tuberculous pleuritis with reference to HIV coinfection and a Bayesian analysis.

OBJECTIVES

To evaluate the diagnostic use of pleural fluid adenosine deaminase (ADAPF) levels in tuberculous pleuritis (TBpl), with a special reference to HIV coinfection and a Bayesian analysis.

METHODS

We investigated a total of 216 patients with pleural effusion, including 100 with TBpl, 68 with malignant effusion, 6 with transudates, 19 with empyema, 15 with miscellaneous diseases, and 8 with diseases of unknown etiology.

RESULTS

The mean values (SE) of ADAPF were 110 (4.5) U/L in patients with TBpl vs 28 (5.3) U/L in patients with a malignancy, 18 (5.7) U/L in patients with transudates, 13 (2.1) U/L in patients with diseases of unknown etiology, 22 (5.1) U/L in patients with miscellaneous diseases, and 191 (26.3) U/L in patients with empyema (Kruskal-Wallis test, p < 0.001). The ADAPF level was 110 (4.5) U/L in 37 HIV-positive patients with TBpl vs 114 (4.1) U/L in 52 HIV-negative patients with TBpl (Mann-Whitney U test, p > 0.05). A receiver operating characteristic curve identified the best cutoff at 60 U/L, yielding measures for sensitivity (0.95), specificity (0.96), positive predictive values (PPVs; 0.96), and negative predictive values (0.95). A Bayesian analysis showed a posttest probability of PPV ranging from 0.5 to 0.99, resulting from a pretest probability of 0.05 to 0.9.

CONCLUSIONS

ADAPF is diagnostically useful across the various prevalences of TBpl, and its best diagnostic utility is in areas of intermediate prevalence of the disease. Moreover, the diagnostic value of ADAPF is independent of HIV serologic status.