Background: In liver tuberculosis, demonstration of acid bacilli by conventional methods remains futile. Since the definitive diagnosis of liver tuberculosis is based on the histologic evidence of granulomatous process with caseation necrosis, seen in only a third of cases, the diagnosis is made retrospectively by response to empirical anti-tuberculous drug therapy.
Aims: Our objective is to establish a polymerase chain reaction assay for detection of Mycobacterium tuberculosis affecting the liver using the paraffin-embedded liver biopsy specimens obtained from patients with hepatic granulomas.
Methods: As positive control, patients having either "definitive" (n=8) or "presumptive" (n=9) tuberculosis on the basis of clinical, microbiological, histologic data and their positive response to empirical treatment of anti-tuberculous drugs were used. Patients with hepatic granulomas secondary to schistosomiasis (n=6), sarcoidosis (n=2) and other liver diseases (n=10) were used as negative control.
Results: Of those patients who were diagnosed as having "definitive" and "presumptive" liver tuberculosis, positivity by one-step polymerase chain reaction was 100% and 44%, respectively. Using the nested polymerase chain reaction, positivity increased to 78% with "presumptive" liver tuberculosis. In contrast, the polymerase chain reaction assays were negative among all patients with hepatic granuloma due to non-tuberculous-in-origin and other liver diseases.
Conclusions: The overall positivity of this polymerase chain reaction assay (88%) compares favorably with that of other conventional methods (12%). Thus, this polymerase chain reaction assay may be a reliable diagnostic tool for liver tuberculosis in a patient population in which the prevalence of diseases associated with hepatic granuloma is common.