Genexpert Mtb/Rif Assay Compared with Non-Molecular Methods on Fine Needle Aspiration Samples for Diagnosis of Tuberculous Lymphadenitis at University Teaching Hospital of Kigali (CHUK)

Abstract

Background Tuberculosis remains one of the major public health problems with a considerable number of new cases of rifampicin resistance. Extra pulmonary tuberculosis (EPTB) accounts for a significant proportion of tuberculosis cases, and tuberculous lymphadenitis (TBL) is one of the most common form of EPTB whose diagnosis still faces many challenges. Conventional methods like auramine stain which is mostly used in our settings, generally has a low sensitivity and specificity. The gold standard culture takes long time to grow, delaying the treatment. Recently, WHO recommended GeneXpert to be used as the initial diagnostic test of Extrapulmonary tuberculosis and there are limited data about GeneXpert test on fine needle aspiration samples. GeneXpert is a rapid molecular test which gives results within 2 hours as well as the status of rifampicin resistance. The aim of this study was to determine the diagnostic performance of GeneXpert and compare GeneXpert with auramine and combined auramine/GeneXpert tests on fine needle aspiration samples. Method The study was prospective cross-sectional of 10 months period from June 2020 to March 2021. We have enrolled patients with palpable lymph node suspected to be TBL at the University Teaching Hospital of Kigali (CHUK). The total of 98 patients were received and lymph node specimen were collected for FNAC (Fine Needle Aspiration Cytology), auramine-O, GeneXpert MTB/RIF assay and culture on Lowenstein-Jensen medium for each patient. Analysis was done by Statistical Product and Service Solutions (SPSS) version 27.0 (IBM Corporation, Armonk, NY), GraphPad Prism (GraphPad Software, Inc., CA 92037 USA) version 9 and MedCalc statistical software (MedCalc Software Ltd, Ostend, Belgium) version 20.010. Descriptive statistics, chi-square, Fisher’s exact, odd ratio, Cohen’s kappa coefficient, positive/negative likelihood ratio and receiver operator characteristic (ROC) with area under the curve (AUC) were used accordingly. Results In our study the sensitivity, specificity, PPV and NPV of GeneXpert taking culture as gold standard were 100%, 88.4%, 54.5% and 100%, respectively. Considering composite reference standard (positive auramine, positive culture or both) specificity and PPV were improved significantly and the values of sensitivity, specificity, PPV as well as NPV were as follow: 94.7%, 94.9%, 81.8% and 98.7%, respectively. A moderate level of agreement between GeneXpert and auramine was observed with k=0.74 and the almost perfect level of agreement between GeneXpert and combined auramine/GeneXpert was noted with k=0.97. Taking culture as gold standard, GeneXpert was compared with auramine and combined auramine/GeneXpert. The GeneXpert showed the excellent accuracy with AUC of 0.942, followed by combined auramine/GeneXpert (AUC: 0.936, excellent accuracy) and auramine with AUC of 0.787 (fair accuracy), p <0.001. This has been supported by the value of likelihood ratios where GeneXpert showed a higher LR+ of 8.620 than combined auramine/GeneXpert (8.196) and auramine (7.172). The values of LR- are 0, 0.367 and 0 for GeneXpert, auramine and combined auramine/GeneXpert respectively. During this study there was no rifampicin resistance detected by the GeneXpert MTB/RIF assay. Conclusion The study highlighted that GeneXpert MTB/RIF assay is more accurate than both combined auramine/GeneXpert and auramine in diagnosis of TBL on fine needle aspiration samples. The implementation of GeneXpert may improve early and accurate diagnosis of TBL and guide for proper treatment.