The study was approved by the University of Limpopo (Medunsa campus) Research and Ethics Committee (study approval number: MREC/P/177/2008:PG).

One hundred and five consecutive pleural fluid specimens submitted to the Dr George Mukhari Tertiary Chemical Pathology Laboratory in Pretoria, South Africa, between August 2008 and March 2009 were included in the study. The specimens were from 71 suspected tuberculosis cases (patients with suggestive symptoms of tuberculosis; symptoms included coughing for more than 2 weeks, night sweats, fever and loss of weight) and 34 non-tuberculosis cases/controls (patients with other clinical causes of pleural effusion). Pleural fluid of all suspected tuberculosis cases (n = 71) and 32 of the control patients were exudates and only 2 of the control specimens were transudates. Among the two patients with transudates, one was diagnosed with malignancy and the other one was diagnosed with diabetes. Patient demographic data, including age, sex and clinical history, were collected from the National Health Laboratory Services Data Intensive System and Applications database. Permission to use samples and access the database was granted by the Head of Chemical Pathology. Individual informed consent was not sought, because the study was conducted on routine samples only and it did not involve additional samples or change in the treatment of patients.

All pleural fluid specimens were stained using auramine O and cultured on Lowenstein-Jensen agar slants and MGIT BACTEC tubes containing Middlebrook medium (Becton Dickinson, Sparks, Maryland, United States). All the culture positive samples were confirmed by Ziehl-Neelsen staining. A diagnosis of pleural tuberculosis was made when the pleural fluid was culture positive and acid fast bacilli (AFB)-positive for Mycobacterium tuberculosis. The unconfirmed pleural tuberculosis cases were diagnosed by the clinical presentation, such as history of tuberculosis infection, chronic cough, night sweats and loss of weight. The non-tuberculosis pleural effusion cases were patients with pleural effusion caused by other medical conditions (Table 1).

Pleural fluid samples were concentrated by centrifugation at 3000 g for 15 min, the supernatant was discarded and the pellet resuspended in 2 mL of phosphate buffered saline buffer pH 6.8. The samples were mixed by vortexing for 5 seconds. The extraction of DNA was performed on concentrated pleural fluid specimens using the Amplicor respiratory sample preparation kit according to the manufacturer’s instructions (Roche Diagnostics, Mannheim, Germany). Briefly, 500 µL wash solution was added to 100 µL of concentrated pleural fluid sample in a 1.5 mL tube. The mixture was vortexed for 5 s, centrifuged at 12 500 g for 10 minutes, and the supernatant discarded. About 100 µL of lysis buffer was added to the pellet and the tubes were vortexed for 5 s to resuspend the pellet. The samples were then incubated at 60 °C for 45 min. Following incubation, the samples were spun down for 5 s, 100 µL of neutralising buffer was added and the mixture was vortexed for 5 s. The lysate was stored at 2 °C – 8 °C for later use.

Real-time PCR was performed using the Light Cycler Mycobacterium detection kit according to the manufacturer’s instructions (Roche Diagnostics, Mannheim, Germany). A total of 20 µL PCR mixture containing 0.25 µL of 8.5 uracil DNA glycosylase, 0.75 µL internal control, 4 µL master mix, 11 µL detection mix and 4 µL DNA sample was prepared in a capillary tube. An internal control was used to control for the presence of inhibitors. The master mix contained the enzyme Hot Start Taq polymerase, MgCl₂, deoxynucleoside triphosphate (dTTP, dCTP, dATP, dGTP) and PCR buffer. The detection mix contained the primers, hybridising probes, fluorescein and the acceptor probe LightCycler Red. The primers targeted the 16s rRNA sequence of the M. tuberculosis complex. Real-time PCR was performed on the Light Cycler 2.0 (Roche Diagnostics, Mannheim, Germany) real-time instrument according to the following protocol: incubation at 40 °C for 10 min, denaturation at 95 °C for 10 min; amplification consisted of 45 cycles of: 95 °C for 10 s, 50 °C for 10 s with a single acquisition mode and 72 °C for 20 s. This was followed by a melting curve analysis to determine melting temperatures for the sequences targeted by the hybridisation probes. This comprised three stages: 95 °C for 1 min, 40 °C for 2 min and 70 °C for 1 min, ramp rate 0.1 and a continuous acquisition mode, followed by cooling at 40 °C for 30 s.

Analysis of the samples was done in two steps: PCR amplification, where the target amplicon for each sample was detected between the annealing and elongation steps as sigmoid curves at 640 nm. The melting temperature calling, where the melting temperature specific for each subtype in the sample, was determined using the Light Cycler 2.0 software (Roche Diagnostics, Mannheim, Germany). Samples were regarded positive for mycobacteria if they had exponential amplification with a crossing point value of less than 35 with a signal intensity of more than 0.02 and negative with no amplification on the 640/back 530 nm channel and exponential amplification for the internal control in the 705/back 530 nm channel having a signal intensity of more than 0.02. A melting temperature of 53.5 °C – 56.5 °C indicated M. tuberculosis, 48 °C – 51 °C M. avium and 57 °C – 60 °C M. kansasii.

The ADA was analysed using a commercial colorimetric assay kit according to manufacturer’s instructions (Diazyme General Atomics, Poway, California, United States). The activity in the patient’s specimen was calculated with the formula: activity in specimen = (optical density specimen – optical density specimen blank)/(optical density standard – optical density reagent blank) × 50 with the result expressed in units/L. An ADA value of 30 units/L or higher was considered to be positive. The assay was done on a DxC 600 Synchron analyser (Beckman Coulter, Brea, California, United States) using user-defined methods.

The results of the ADA assay and qPCR against the culture, the gold standard for diagnosis of tuberculosis, were entered into 2 × 2 tables, wherein the sensitivity was calculated by using the formula TP/(TP+FN), specificity by using the formula TN/(FP+TN), positive predictive value by using the formula TP/(TP+FP) and negative predictive value by using the formula TN/(TN+FN), where TP = true positives, FN = false negatives, FP = false positives and TN = true negatives. The mean and standard deviation of ADA and the mean age were calculated using Epi Info version 3.3 (Centers for Disease Control and Prevention, Atlanta, Georgia, United States).

Of the 21 confirmed cases, 14 (67%) were qPCR and culture positive, while 7 (33%) were qPCR negative. None of the specimens from the control group and the unconfirmed tuberculosis cases were positive by qPCR. The sensitivity of the qPCR was 67% and specificity was 100% (Table 2). The positive predictive value was 78% and negative predictive value was 86%.

Of the 21 confirmed cases, 17 (81%) were ADA and culture positive, and 4 (19%) were ADA negative. Of the 50 unconfirmed cases, 39 (78%) were ADA positive and 11 (22%) were ADA negative. The ADA levels among the confirmed pleural tuberculosis cases ranged from 8 units/L to 134.3 units/L with a mean ADA value of 52.2 ± 21.66 units/L. The ADA levels among the unconfirmed pleural tuberculosis cases ranged from 5.68 units/L to 200 units/L with a mean ADA value of 50.12.2 ± 21.77 units/L. The ADA levels among the control group ranged between 2.4 units/L and 98 units/L, with a mean ADA value of 12.7 ± 8.64 units/L. Only two (6%) of the 34 non-tuberculosis pleural effusion cases were ADA positive with values of 39 units/L and 98 units/L. Both patients with elevated levels of ADA had pneumonia. The mean ADA level among the pleural tuberculosis cases was significantly higher than that of the control group, p < 0.0001. The sensitivity of ADA was 80% and specificity was 94% (Table 2). The positive predictive value was 89% and negative predictive value was 88%.

The ADA assay has a high sensitivity (80%) and specificity (94%) and hence it is still a useful tool for the diagnosis of pleural tuberculosis. The sensitivity of qPCR was 67% and specificity was 100%. The low sensitivity of qPCR suggests that this test should not be used for excluding a diagnosis of pleural tuberculosis. However, the high specificity of qPCR suggests that a combination of the two methods may be useful for the diagnosis of pleural tuberculosis. Adenosine deaminase assay (ADA) activity in pleural fluid can differentiate between pleural disease due to tuberculosis and effusion due to non-tuberculous lymphocytic effusion. In addition, the ADA result is available on the same day compared to culture, which takes about two weeks. More studies on the diagnostic value of qPCR are needed.