Diagnosing tubercular pleural effusions.

We read with interest the article by Hikari et al (March 2004)1 comparing the markers of tuberculosis in pleural effusions. We wish to express our disagreement with their statement that interferon (IFN)should be measured routinely in all suspected cases of pleural tuberculosis. They have based their conclusion on the basis of a perfect area under the curve of 1.000 on receiver operator characteristic analysis for IFNas compared to 0.958 for adenosine deaminase (ADA). The authors have failed to adequately review the fairly large body of literature on biological markers of tubercular pleural effusion. ADA has been reported with perfect values in the literature (100% sensitivity,2–4 and also 100% specificity, positive predictive value, and negative predictive value5) in studies with larger sample sizes (n 221, 48 tuberculous2; n 405, 91 tuberculous3; n 350, 76 tuberculous4; and n 138, and 61 tuberculous5) than the present study (n 55, 20 tuberculous).1 Valdes et al,3 using simultaneous measurement in the same set of patients (n 405), reported a higher sensitivity for ADA (100%) than IFN(94.2%) and a higher specificity (95% for ADA and 91.8% for IFN). Villegas et al6 compared ADA and IFN(along with polymerase chain reaction [PCR]) simultaneously in 140 patients with 42 confirmed TB cases and reported a higher sensitivity (88.1% for ADA vs 85.7% for IFN) and better negative predictive value than IFN in the whole prevalence range. Valdes et al7 reported that 253 of a total of 254 tuberculous pleuritis patients had ADA levels 40 IU/mL, and in the 82 patients in whom both ADA and IFNwere done, the sensitivity of IFN was 89% (73 of 82 patients) against at least 98.78% (81 of 82 patients) for ADA. Studies comparing ADA and IFNsimultaneously in the same set of patients have reported both ADA better than IFN2–7 and IFNbetter than ADA1,8–9 as diagnostic markers. In fact, a meta-analysis by Greco et al10 regarding the diagnostic accuracy of ADA vs IFNincluded 31 studies in favor of ADA (total, n 4,738) and 13 studies in favor of IFN(total, n 1,189). Using summary receiver operating characteristic curve, they found only a marginal difference in overall sensitivity and specificity: 93% for ADA, and 96% for IFN. Using Bayes theorem, the posttest probability of a negative test result was calculated. The minute difference in posttest probabilities (ADA vs IFN, 0.4% vs 0.22%, 2.4% vs 1.2%, and 24% vs 17%) was maintained over a wide prevalence range of 5 to 85%. The authors concluded that “ADA and IFNappear to be reasonably accurate at detecting TB pleurisy.” Virtually similar sensitivity and specificity coupled with lower cost should favor the use of ADA as a diagnostic tool compared to IFN. Lastly, the authors suggest that PCR should be compared with IFN, etc. Such a study comparing PCR, IFN, and ADA simultaneously in pleural effusion patients has already been published in CHEST.6