|Introduction and rationale: Tuberculous pleural effusion is a common cause of patient admission in Mulago hospital especially among patients infected with HIV.The routinely used diagnostic methods of pleural fluid analysis and ZN staining have a very low accuracy for the diagnosis of pleural TB. Histology and pleural tissue culture for MTB, the gold standard involves an invasive procedure which requires specific skills and is time consuming. We tested the diagnostic accuracy of Cepheid’s MTB-specific Gene Xpert, a PCR based real time test using pleural fluid with the aim of trying to get a quicker, more user friendly, accurate and less invasive method in the diagnosis of pleural TB in our setting.This test has been proven to improve diagnose of pulmonary TB with accuracy and in less than 2 hours. Varying levels of accuracy of this test in the diagnosis of pleural TB diagnosis have been reported from small pilot studies. A South African study by Sven et al 2011 however demonstrated a sensitivity of 25% and specificity of 100% on 20 samples of confirmed pleural TB patients. An additional study with a bigger sample size was therefore needed.
Objectives: To determine the accuracy of MTB-specific Gene Xpert test in the diagnosis of TB pleura among adult patients with suspected tuberculous pleural effusions attending Mulago Hospital Pulmonology Service over the study period.
Methods: We used a cross-sectional study and enrolled 116 patients with exudative pleural effusions in Mulago National Referral Hospital, pulmonology ward (4CP) and outpatient services from April 2012 to December 2012.
Clinical and radiological evidence of pleural effusion were used to select patients with suspected tuberculous pleural effusion, greater than 25% of hemithorax, and provided written informed consent.These patients were consecutively screened, their demographics and clinical data were recorded using a pre-tested questionaire. Thoracocentesis was done and pleural fluid examined usingxviii Light’s criteria for exudative effusions and we enrolled 116 patients for pleural biopsy. The biopsy specimens were sent for MGIT BACTEC 960 cultures and histology; at the same biopsy sitting, 10 mL of pleural fluid was collected for the Gene Xpert test. Patient characteristics were compared using Chi-Square test for dichotomous variables, while Students T-test and Wilcoxon rank-sum test were used for continuous normally distributed and non-normally distributed variables respectively. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), likelihood ratios and area under the curve for Gene Xpert were calculated using pleural tissue culture and histology as the reference gold standard.
Results: Out of 116 patients enrolled 87 (75%) were confirmed with pleural TB. 25(28.7%) patients tested positive by GeneXpert.The sensitivity, specificity, PPV, NPV, positive and negative likelihood raios were 28.7%, 96.6%, 96.1%, 31.1%, 8.44 and 0.738 respectively. Area under the ROC curve was 0.626 (p=0.0299). Pleural TB patients were significantly younger mean age 32 years (p<0.001), had higher temperatures (p=0.034) and more patients with CD4 counts <350cells/uL (p=0.015). GeneXpert positive peural TB had more patients with higher pleural fluid LDH, and this was more evident with pleural fluid LDH >1000IU/L (p=0.015), more patients with weight loss (p=0.085) and febrile patients (p=0.092).
Conclusion: MTB-Specific GeneXpert has low sensitivity and may not be used as an initial diagnostic test for pleural TB; however the high specificity and positive predictive values could make it an important confirmatory test where more evidence for pleural fluid TB is required.
Recommendations: GeneXpert may have no diagnostic utility as an initial test in pleural TB, but it could have a place in diagnosis of pleural TB as a confirmatory test in our setting. Further tests to determine the additional burden of TB in patients with very high LDH >1000IU/L, and neutrophil predorminant pleural effusions should be conducted.