ABSTRACT
To compare diagnostic tests for tuberculous lymphadenitis by polymerase chain reaction (PCR) with histopathology and clinical diagnosis in sensitivity, specificity and predictive value. This retrospective analytic, single blind study was done at King Chulalongkorn Memorial Hospital. Paraffin-embedded specimens were classified into 2 groups. The study group contained 30 proved AFB positive paraffin-embedded specimens from patients who also had clinical diagnosis of tuberculosis and improved by antituberculous treatment. The control group contained 30 formalin-fixed, paraffin-embedded specimens of lymph node hyperplasia proved by histopathological and clinical review. All 60 specimens were slided, and systematically labeled and sent to PCR lab. Polymerase Chain Reaction method had sensitivity = 43.33 per cent, specificity = 100 per cent, positive predictive value = 100 per cent and negative predictive value = 63.83 per cent. The present findings revealed that the PCR results were related to the age of the paraffin-embedded tissues. No positive results were obtained from tissues kept since 1996. Positive results were obtained in 3/7 cases (42.86%), 2/3 (66.67%) and 8/10 cases (80%) from tissue of 1997, 1998 and 1999 respectively. Conclusion: Polymerase chain reaction has sufficient reliability best as a confirmatory diagnostic test for tuberculous lymphadenitis; however, it is not appropriate as a screening test.