We analysed liver disease patients admitted to our tertiary care centre from January 2012 to December 2016 in this retrospective study. Patients were screened for TB only if patients presented with: fever, cough for >2 weeks, weight loss, haemoptysis, and unresponsive ascites, arcane bowel symptoms (constipation, diarrhoea, or subacute intestinal obstruction), and radiological features reminiscent of TB and past or family history of TB. Samples received were alienated as pulmonary (sputum, bronchoalveolar lavage) and extrapulmonary (lymphnode samples, tissue biopsy, urine, and body fluids as ascitic fluid, pleural fluid etc.,). The samples were subjected to microscopy with Zeihl Nelsen staining, TB was diagnosed based on one of the following: (i) histological evidence of caseating granulomas; (ii) smear positivity for AFB; (iii) growth of Mycobacteria on MGIT culture (BD Microbiology Systems, Cockeysville, MD); or (iv) positive PCR for Mycobacterium tuberculosis in tissues (MTB qPCR) (Cobas TaqMan MTB assay). Mycobacteria other than tuberculosis (MOTT) were identified by negative MPT64 immunochromatography assay (SD Bioline) or MGIT culture positive and MTB PCR negative samples. Statistical data analyses were performed using SPSS Statistics 19 (SPSS Inc., Chicago, IL, USA). Sensitivity and specificity of the tests were calculated using culture results as gold standard. The study was approved by Institute ethical committee.

A total of 816 samples were processed for mycobacterial identification. These included 260 pulmonary and 556 extra-pulmonary samples. A total of 118 were positive by MGIT culture or PCR inclusive of pulmonary 31/260 (11.92%) and extra pulmonary 87/556 (15.65%) samples. The underlying liver disease was predominantly cirrhosis 72/118 (61.01%), alcoholic liver disease 25/118 (21.1%), post liver transplant 6/118 (5.08%), non-alcoholic steatohepatitis 3/118 (2.54%), liver malignancy or secondaries 2/118 (1.69%), hydatid disease of liver 1/118 (0.84%), viral hepatitis 2/118 (1.69%) and pancreatic malignancy 2/118 (1.69%), pancreatitis 2/118 (1.69%), an ulcerative colitis 3/118 (2.54%) amongst other causes. The system wise distribution of positive samples as respiratory (11.92%), lymphnodes (32.67%), tissue biopsy (10.41%), urine (7.4%) and body fluids (4.44%) are given in [Table/Fig-1]. The demographic profile revealed 78/118 (66.1%) male and 40/118 (33.8%) female patients with positive results. The distribution of age and sex in pulmonary and extrapulmonary diseases is depicted in [Table/Fig-2]. Smear microscopy was positive in 32/118 (27.11%) which was higher in pulmonary 19/31 (61.2%) than extrapulmonary samples 13/87 (14.9%). The samples exclusively positive on MGIT culture system were 108/118 (91.52%) and those positive on MTB qPCR were 97/118 (82.2%). Mycobacterium tuberculosis was identified in 103/118 (87.28%) and MOTT was detected in 15/118 (12.71%) samples. The pulmonary and extrapulmonary distribution of these species are given in [Table/Fig-2]. The sensitivity of MTB qPCR when compared to MGIT culture as gold standard was 90.3% (95% CI, 73-97%) for pulmonary disease and 76.6% (95% CI,65.3-85.2) for extrapulmonary disease [Table/Fig-3]. The specificity of MTB qPCR when compared to MGIT culture was 100% (95% CI, 97.9-100%) in pulmonary samples and 97.9% (95% CI, 96-98.9) in extra-pulmonary samples [Table/Fig-3]. In our study the positive predictive value of MTB qPCR in pulmonary disease was 100% while in non-pulmonary disease was 85.5% (95% CI, 74.4- 92.4%). The negative predictive value was 98.7% (95% CI, 95.9-99.6%) and 96.3% (95% CI, 94.1- 97.7%) in pulmonary and extrapulmonary samples respectively [Table/Fig-3].

[Table/Fig-1]:

Systemwise distribution of positive samples.

TB is still epidemic in India carrying more than 20% of the global TB burden, despite of efforts taken to minimise it [1]. The stigma linked with TB in society has harsh consequences on infected individuals. Although, the treatment approach using drugs in combination is successful in treating TB, it may lead to increased side effects with toxicity.

Hepatotoxicity is common with the use of anti-tubercular drugs and often exasperated by the cumulative toxicity of such drug combinations. TB and the liver have complex relationship. The direct involvement of liver by the disease itself is common but marked impairment of liver functions is seldom seen [2]. Chronic liver disease patients are prone to develop TB posing management problems. The underlying disease process may be aggravated as potent anti-tubercular drugs are hepatotoxic. Thus, we need to modify the drug regimens to avert further hepatic insult. Moreover, chronic liver disease is one of the most important risk factor for the development of ATT induced hepatitis also [3]. The absorption and distribution of drugs metabolised in the liver such as rifampicin and isoniazid can be altered due to hepatic dysfunction [4]. Chronic liver disease, being an immunocompromised state augments the risk of activation of latent TB [3,4]. Conversely, hepatic involvement by TB can lead to deterioration of underlying liver injury in liver disease patients [4]. EPTB is difficult to diagnose with concomitant liver disease. The characteristic features of ascitic fluid in peritoneal TB is raised LDH (>90 U/L), increased protein (>2.5 g/L) and low SAAG (<1.1 g/L) levels. However, the diagnosis of tubercular ascites may be confounded by a lower protein and higher SAAG levels in the ascitic fluid with coexistent chronic liver disease [5]. The fluctuations in the liver function profile owing to a pre-existing liver disease causes difficulty in biochemical monitoring the degree of drug induced hepatic injury [6,7].

Pulmonary TB accounts for approximately 80-85% of all cases in the general population [8]. However, in patients with cirrhosis, EPTB is more common. In a Korean study, 31% cirrhotics had EPTB as compared to 12% in the non-cirrhotic control group with predominance of peritoneal TB [9]. Our study reiterates this finding with 15.65% EPTB cases as compared to 11.92% pulmonary cases. Liver cirrhosis as a risk factor for EPTB has been suggested in a previous study [10-12]. Immunopathogenesis of TB in such clinical conditions is still ambiguous. It has been proposed that although most of the host defence mechanisms, particularly the reticuloendothelial system clearance capacities, are dampened in patients with cirrhosis, nonetheless it is yet not clear how this immune dysfunction results in patients being more prone to extrapulmonary TB than pulmonary TB [5,13].

A definitive diagnosis of TB by culturing Mycobacterium tuberculosis organisms from samples obtained from the patient. However, EPTB diagnosis remains challenging as clinical samples obtained from relatively inaccessible extrapulmonary sites may be paucibacillary, which decreases the sensitivity of diagnostic assays. Since the conventional smear microscopy has a low sensitivity with a range of 0%–40%, negative results cannot exclude the presence of TB [14]. Our study shows overall sensitivity of staining as 27.12%, thus, reliance only on microscopy could lead to wrong diagnosis in more than 2/3^rd of cases. Smear positivity was much lower in extrapulmonary disease (14.9%) than the pulmonary disease (61.2%). The reported yields of mycobacterial culture vary from 30% up to 80%, but it usually takes 2–6 weeks to receive the results, which is too slow to help treatment decisions [14]. In this study we had culture positivity in (89.83%) and majority of the cultures were positive by third week of incubation. Our culture results were promising as the processing involved stringent decontamination of samples.

Diagnosis of EPTB from tissue samples is usually made by histopathological examination that depends on the presence of granulomatous inflammation and caseous necrosis [15-17]. However; histopatholology does not distinguish between EPTB and infections from other granulomatous diseases such as nontubercular mycobacteria, leprosy, sarcoidosis, and systemic lupus erythematosus [16,17].

Molecular techniques yield rapid results with high sensitivity and shown new vigor in TB diagnosis [15]. Real-time PCR is a novel and robust assay primarily used to quantify the nucleic acid molecules in specimens [18,19]. This technique is advantageous owing to shortened turnaround time, aiding quantification of bacterial load and automation of the procedure reducing hands-on time and decreased risk of cross contamination [19,20]. Because of extremely high sensitivity of PCR, false positivity is also common due to the carry over contamination of amplicon, previous infection or asymptomatic EPTB infection at another site [21,22]. The false positivity of PCR in cases without clinical findings challenges the diagnosis in EPTB cases [23]. Several meta-analyses of commercial Nucleic Acid Amplification Tests (NAATs) have documented sensitivities of 90% to 100% and specificities of 71% to 96% in sputum smear positive samples, but significantly lower sensitivities of 22% to 89% and specificities of 97% to 99% in sputum smear-negative samples [24-26]. For diagnosis of EPTB, performance of NAATs generally is unsatisfactory, signifying pooled sensitivity and specificity of 53% and 95%, respectively for serum, and 30% to 76% and 73% to 99%, respectively for pleural TB [24]. In our study the qPCR was positive in 97/118 (82.2%) with higher sensitivity in pulmonary (90.3%, 95% CI 73-97) than EPTB (76.6%, 95%CI 65.3-85.2). Similarly the specificity for pulmonary disease was 100% compared to EPTB 97.9%. Our study revealed an acceptable positive and negative predictive value of the real time PCR assay [Table/Fig-3]. We adopted a policy to process only those samples with appropriate quality and quantity followed by adequate decontamination when required. The sample collection is most important part for diagnosis of EPTB and we have a system of repeated instructions and training for appropriate sample collection, which has helped increased the positivity of tests. It is noteworthy to mention that, PCR and culture, rather than either test alone could be a useful method to accurately identify TB cases.

Several studies have shown that period prevalence of pulmonary MOTT disease has been highest among Japanese, Chinese, and Vietnamese patients (>300/100,000 persons) and lowest among Native Hawaiians and Other Pacific Islanders (50/100,000) [27,28]. In previous reports, persons identified as Asian American/Pacific Islander were at increased risk for pulmonary MOTT disease, independent of geographic area of residence [27,28]. In an Indian study on MOTT in tertiary care centres in north India, the isolation rate was 62/227 (27.4%) in extra pulmonary samples [29]. Various studies from India showed MOTT prevalence of 17.4% from clinical specimens in patients with fibro cavitary disease and 7.4% from various clinical specimens [30]. In our study the incidence of MOTT disease was 12.71%, isolated from extrapulmonary samples only. Hence, it is necessary to suspect non tubercular mycobacteria in extrapulmonary disease especially with an underlying liver disease. This difference in occurrence of MOTT in extrapulmonary disease needs further evaluation. The studies on MOTT in liver diseases are meagre and more data generation in this subject is needed.

Thus, our study gives important insights into the characteristics of mycobacterial infections in liver diseases. Our study demonstrates the predominance of smear negative, EPTB and MOTT in our patients with underlying liver diseases. A combination of MGIT culture and TB PCR has additive advantage over either test alone. A high index of suspicion and good screening methods are needed to identify TB and MOTT in such patients, owing to similar presentation and anti-tubercular drug toxicity issues. Rapid speciation that distinguishes MTB from MOTT is an important prerequisite for the proper management of patients with mycobacterial infections in liver diseases and may be introduced as a required standard into routine laboratory diagnostics.