Tubercular gastric fistula: apropos of two cases

Introduction

The stomach is an uncommon site of involvement in tuberculosis (TB). Gastric involvement due to TB may occur in the form of ulcer, nodules, hypertrophied folds or submucosal lesions, usually manifesting as gastric outlet obstruction, haematemesis, and loss of weight and appetite. The diagnosis is difficult and usually based on demonstration of granulomatous inflammation (and rarely acid-fast bacilli [AFB]) from mucosal biopsies.^1–3 We report two cases of tubercular gastric fistula: one where a patient on anti-tubercular therapy (ATT) developed a fistula between stomach and perihepatic collection; and another where a tubercular left liver lobe abscess, initially thought to be amoebic, ruptured into the stomach resulting in a hepato-gastric fistula.

Case 1

A 26-year-old man had been on ATT for disseminated TB (pulmonary, peritoneal, and lymph node). He had been diagnosed with sputum-positive TB with evidence of a peri-hepatic collection along with ascites (Figure 1a). There was no family history of TB, he was a non-smoker and non-alcoholic. Tests for diabetes (HbA1c, blood sugar fasting) and human immunodeficiency virus (HIV) were negative. He had developed ATT-related hepatitis on a daily four-drug regime and was therefore commenced on modified ATT with levofloxacin, streptomycin and ethambutol, with a good compliance. After five months, he complained of renewed epigastric pain, which was moderate but continuous without relation with meals. OPEN IN VIEWER

Repeat computed tomography (CT) revealed reduction in size of the peri-hepatic collection, but the presence of air foci and a possible communication with the gastric lumen (Figure 1b). On endoscopy, the suspicion of fistulisation close to the pylorus was confirmed (Figure 1c and 1d). The patient improved with continuing modified ATT.

Case 2

A 45-year-old man was evaluated for abdominal pain and fever. On the basis of abdominal CT showing a liver abscess (Figure 2a) with caecal thickening, colonoscopy showing colonic ulcers (Figure 2a) and positive serology for amoebiasis, he received therapy with metronidazole. At re-evaluation four weeks later, he had continued to lose weight, now presenting with severe abdominal pain, despite an ultrasound scan showing reduction in abscess size. Gastroscopy revealed a fistula with whitish pus-like material (Figure 2c). Colonoscopy demonstrated that his colonic ulcers were still present, from which biopsy showed evidence of granuloma without caseation or AFB (Figure 2d). Ileoscopy was normal. Mantoux skin test was positive and the patient was started on four-drug ATT. There was no past history of TB but his brother had received compete treatment two years before for pulmonary TB. Screening for HIV and diabetes were negative; the patient was a non-smoker and non-alcoholic. After two months of ATT, he had gained 5 kg of weight and the colonic ulcers had healed. Control gastroscopy showed the fistula had closed and ultrasound scan demonstrated the hepatic collection had resolved. OPEN IN VIEWER

Discussion

In the first case, fistulisation was possibly related to the loculated collection which did not respond to ATT, rupturing into the stomach. In the second case, fistulisation was probably due to rupture into the stomach which had resulted in abdominal pain and reduction in size of liver abscess. Unfortunately, in both of our cases, we did not biopsy the mucosa around the fistulous tract, not expecting granulomas after five months of ATT in the first case. In the second case, the possibility of TB was entertained only later when colonic biopsies yielded granulomatous inflammation. Interferon gamma release assay testing was not done in either patient owing to its expense and that its additional value beyond Mantoux test is limited.

Other reports of gastric fistula reported include communications occurring with the spleen, the pancreas, the gall-bladder, the bile duct, the colon, the left lower lobe bronchus, pericardium and multiple organs including the skin.^4-13

Several of these cases were treated by surgical intervention, although in our cases surgery was not needed. Although one of our patients developed his gastric fistula before initiation of ATT while the other developed it during ATT, both improved with continued ATT.