Extrapulmonary tuberculosis presenting with proximal esophageal fistula and abdominal lymphadenopathy in a young man: A case report

Abstract

BACKGROUND

Abdominal and esophageal tuberculosis (TB) are uncommon manifestations of Mycobacterium infection and often mimic malignant or inflammatory conditions, leading to delayed diagnosis. Esophageal TB complicated by fistula formation is particularly rare, and pancreatic involvement can closely resemble pancreatic carcinoma or autoimmune pancreatitis.

CASE SUMMARY

A 32-year-old man from a TB-endemic region presented with persistent epigastric pain, progressive dysphagia and unintentional weight loss despite successful Helicobacter pylori eradication. Cross-sectional imaging revealed circumferential thickening of the proximal esophagus, diffuse pancreatic enlargement and pathological subdiaphragmatic lymphadenopathy. Barium swallow suggested a short, blind-ended proximal esophageal fistulous tract. Repeat upper gastrointestinal endoscopy confirmed a very short proximal esophageal fistula at 18 cm from the incisors, exposing the esophageal adventitia without a demonstrable communication to adjacent structures. Endoscopic ultrasound (EUS) demonstrated multiple enlarged subdiaphragmatic lymph nodes, which were sampled by fine-needle aspiration. Cytology showed florid granulomatous inflammation with multinucleated giant cells, highly suggestive of TB, although Ziehl–Neelsen staining for acid-fast bacilli was negative. Based on the clinical, epidemiological and imaging context, a diagnosis of extrapulmonary TB with intra-abdominal lymph node involvement, associated proximal esophageal fistula and probable pancreatic TB was made in a multidisciplinary team setting. The patient received standard four-drug anti-TB therapy (isoniazid, rifampicin, pyrazinamide and ethambutol). At 2 months, he developed worsening abdominal pain and radiological progression of lymphadenopathy, consistent with a paradoxical TB reaction; adherence was confirmed and there was no evidence of drug resistance. Anti-TB therapy was continued and extended to 12 months because of fistulating esophageal disease and bulky pancreatic involvement. He achieved complete clinical recovery with radiological resolution of the esophageal fistula, near-complete regression of lymphadenopathy and marked reduction in pancreatic size.

CONCLUSION

This unusual triad of proximal esophageal fistula, intra-abdominal tuberculous lymphadenitis and imaging-suggested pancreatic TB in an immunocompetent young man without pulmonary involvement underlines the need to consider TB in the differential diagnosis of unexplained esophageal fistula, abdominal lymphadenopathy and pancreatic enlargement, particularly in patients from endemic regions. EUS-guided tissue acquisition is a minimally invasive and highly accurate modality for establishing the diagnosis of intra-abdominal tuberculous lymphadenitis and guiding timely therapy.

Key Words: Esophageal tuberculosis; Esophageal fistula; Pancreatic tuberculosis; Abdominal lymphadenopathy; Endoscopic ultrasound; Fine-needle aspiration; Paradoxical tuberculosis reaction; Case report

Core Tip: Extrapulmonary tuberculosis (TB) may present with atypical and misleading features that mimic malignancy. We report a young man from a TB-endemic region with persistent abdominal pain, dysphagia and weight loss after successful Helicobacter pylori eradication who was found to have a very short proximal esophageal fistula, diffuse pancreatic enlargement and subdiaphragmatic lymphadenopathy. Endoscopic ultrasound (EUS)-guided fine-needle aspiration of abdominal lymph nodes revealed granulomatous inflammation that was highly suggestive of TB. Standard anti-TB therapy was extended to 12 months because of fistulating esophageal disease and probable pancreatic involvement. The patient reached complete clinical and radiological resolution, despite an interim paradoxical reaction. This case underscores the need to consider TB in patients presenting with unexplained esophageal fistula and abdominal lymphadenopathy, and illustrates the pivotal diagnostic role of EUS-guided tissue sampling.

INTRODUCTION

Patients presenting with vague abdominal pain and weight loss pose a considerable diagnostic challenge because these non-specific symptoms are attributable to a wide spectrum of gastrointestinal, neoplastic and infectious conditions. Abdominal tuberculosis (TB) is an uncommon but important manifestation of Mycobacterium infection, accounting for approximately 3%-5% of extrapulmonary TB cases worldwide and up to 11% of extrapulmonary TB when the entire gastrointestinal tract, peritoneum, lymph nodes and solid organs are included[1,2]. Abdominal TB frequently resembles neoplastic or inflammatory disorders, often resulting in delayed or missed diagnosis[3].

Helicobacter pylori (H. pylori) infection is a common cause of chronic gastritis and dyspeptic symptoms worldwide, and eradication therapy improves symptoms in a proportion of patients with functional dyspepsia[4,5]. However, many patients continue to have persistent or recurrent abdominal complaints following successful eradication, especially when alarm features such as weight loss, dysphagia or anemia are present; in such scenarios, alternative etiologies must be carefully considered[4,6].

Esophageal TB is one of the rarest forms of gastrointestinal TB, accounting for around 0.2% of gastrointestinal cases[7,8]. Most reported cases represent secondary involvement from adjacent mediastinal lymph nodes, pulmonary parenchyma or direct extension from contiguous structures[7-10]. Esophageal TB may manifest as ulcers, strictures, mass-like lesions or, rarely, fistulae to neighboring structures such as the trachea, bronchus or mediastinum[7-11]. Fistula formation involving the proximal esophagus is particularly uncommon and can mimic malignancy or other inflammatory disorders.

Pancreatic TB is likewise rare and often radiologically indistinguishable from pancreatic carcinoma or autoimmune pancreatitis[12-15]. It can present with a focal mass, diffuse enlargement or cystic lesions, typically accompanied by weight loss, abdominal pain and sometimes obstructive jaundice[12-14]. Distinguishing pancreatic TB from malignancy or autoimmune pancreatitis on imaging alone is difficult, and tissue diagnosis is crucial.

Intra-abdominal tuberculous lymphadenitis commonly accompanies abdominal TB. Endoscopic ultrasound (EUS)-guided fine-needle aspiration or core needle biopsy (EUS-FNA/B) has emerged as a minimally invasive, safe and highly accurate modality to evaluate intra-abdominal lymphadenopathy of unknown etiology, with reported sensitivity exceeding 85%-90% and specificity above 90%-95% for TB in experienced centers[16-18].

We report a diagnostically challenging case of extrapulmonary TB in an immunocompetent young man presenting with a rare triad of a very short proximal esophageal fistula, subdiaphragmatic tuberculous lymphadenitis and imaging-suggested pancreatic TB in the absence of pulmonary involvement. This case highlights the importance of reconsidering the differential diagnosis in patients with persistent symptoms after H. pylori eradication; illustrates how TB can mimic malignancy involving the esophagus, abdominal lymph nodes and pancreas; and demonstrates the central diagnostic role of EUS-guided tissue sampling. We also describe an interim paradoxical reaction (PR) during treatment and discuss therapeutic duration in the context of current guidelines.

CASE PRESENTATION

Chief complaints

A 32-year-old man presented with a 6-month history of persistent epigastric pain, progressive dysphagia to solids and liquids, and unintentional weight loss.

History of present illness

The patient first presented in October 2024 with refractory epigastric pain and dyspeptic symptoms. Upper gastrointestinal endoscopy at that time revealed H. pylori-associated chronic active gastritis. He was treated with standard triple therapy (clarithromycin-based regimen) for 14 days, with confirmed H. pylori eradication by urea breath test 8 weeks later. Despite successful eradication, he developed progressive dysphagia, initially to solids then to some liquids, recurrent epigastric pain and further weight loss over the subsequent 4 months. He reported nocturnal diaphoresis, anorexia and early satiety but denied odynophagia, hematemesis, melaena or change in bowel habit. There was no history of cough, hemoptysis, pleuritic chest pain or constitutional respiratory symptoms.

History of past illness

The patient had no previous history of pulmonary or extrapulmonary TB, no known contact with TB patients and no prior TB treatment. He had no history of chronic liver disease, diabetes mellitus, human immunodeficiency virus (HIV) infection, inflammatory bowel disease, pancreatitis, autoimmune disease or malignancy. He was not taking immunosuppressive medications, corticosteroids or biologic agents. There was no history of previous abdominal surgery.

Personal and family history

The patient was born and raised in a North-West African country with a high TB burden and had migrated 5 years prior to presentation. He had lived in an urban area but had regular visits to rural regions; there was no history of residence in pastoral nomadic communities. He denied known household or occupational contact with individuals treated for active TB. He was a non-smoker and consumed alcohol only occasionally. There was no history of intravenous drug use. He worked in hospitality and lived in shared accommodation, suggesting potential unrecognized community exposure. No evidence of a Bacillus Calmette-Guérin vaccination scar was documented; however, he reported routine childhood vaccinations in his country of origin. There was no family history of TB, chronic liver disease, or malignancy. There was no family history of TB, autoimmune disease, or gastrointestinal malignancy.

Physical examination

On examination, he appeared thin but was hemodynamically stable and afebrile. His body mass index was 19.2 kg/m², having decreased from an estimated 22.0 kg/m², equaling a weight loss of approximately 10 kg from a baseline of approximately 74 kg to 64 kg. The earlier self-reported 1.6 stone weight loss was converted for clarity. There was no cervical lymphadenopathy. Chest examination was unremarkable, with clear lung fields and normal breath sounds. Abdominal examination revealed mild epigastric tenderness without peritonism, organomegaly or palpable masses. There was no peripheral edema, clubbing, skin rash or stigmata of chronic liver disease. Neurological examination was normal.

Laboratory examinations

Baseline laboratory findings at the time of further evaluation (prior to initiating anti-TB therapy) were as follows: Hemoglobin: 12.4 g/dL (mild normocytic anemia), white blood cell count: 6.8 × 10⁹/L. Platelet count: 325 × 10⁹/L. Erythrocyte sedimentation rate: 48 mm/hour (elevated). C-reactive protein: 24 mg/L (elevated). Serum creatinine and urea: Within normal limits. Liver function tests: Alanine aminotransferase 28 U/L. Aspartate aminotransferase 25 U/L. Alkaline phosphatase: 92 U/L. Total bilirubin: 11 μmol/L (all within reference range). Serum albumin: 38 g/L. Serum amylase and lipase: Within normal limits. Serum calcium, fasting glucose: Normal. Viral serology: HIV-1/2 antigen-antibody combination assay negative; hepatitis B surface antigen and hepatitis C antibody negative.

Tuberculin skin testing was not performed at the time of acute workup. An interferon-γ release assay (IGRA; QuantiFERON-TB Gold) performed after EUS-guided sampling returned positive, supporting TB infection in the clinical context. Tumor markers, including carbohydrate antigen 19-9 and carcinoembryonic antigen, were not significantly elevated.

Imaging examinations

Initial evaluation following persistent symptoms included a barium swallow and contrast-enhanced computed tomography (CT).

Barium swallow: This showed leftward deviation of the proximal esophagus at the upper thoracic inlet with a small, blind-ended outpouching at the level of the aortic knuckle, suspicious for either a proximal diverticulum or a short fistulous tract. There was no contrast leakage into the airway or mediastinum.

Contrast-enhanced CT of chest, abdomen and pelvis (baseline): CT demonstrated diffuse circumferential wall thickening of the upper esophagus over a segment approximately 3-4 cm in length, starting at around 16-17 cm from the incisors and extending distally. Multiple pathological subdiaphragmatic lymph nodes were identified along the left gastric artery, celiac axis and aortocaval region. The largest lymph node measured approximately 22 mm × 19 mm in short-axis diameter (left gastric station), with several additional nodes between 10 mm and 18 mm (Figure 1A). The pancreas appeared diffusely enlarged, with loss of normal lobulation but preserved enhancement and no discrete mass or ductal obstruction. The main pancreatic duct was not significantly dilated. There was no intrahepatic or extrahepatic biliary dilatation, free intra-abdominal fluid or focal liver lesions. No pulmonary parenchymal lesions, cavitation, tree-in-bud nodularity or pleural effusion were observed.

Figure 1 Contrast-enhanced computed tomography. A: The largest lymph node measured approximately 22 mm × 19 mm in short-axis diameter; B: Near-complete regression of subdiaphragmatic lymphadenopathy.

Endoscopic and EUS findings: A repeat upper gastrointestinal endoscopy was performed in November 2024 using an Olympus GIF-XP130N adult video gastroscope under conscious sedation. (1) Esophagogastroduodenoscopy: The upper esophagus appeared narrowed and rigid from approximately 16 cm to 20 cm from the incisors, with mucosal edema and subtle erythema. At 18 cm from the incisors, a very short blind-ended fistulous tract was identified on the left posterior wall. It appeared as a small, slit-like opening measuring approximately 4-5 mm in maximal diameter and extending 6-8 mm in depth, with exposure of the underlying adventitial layer. The tract terminated blindly without visible communication to adjacent structures. There was no ulceration, active bleeding, food impaction or obvious communication with the trachea, bronchus or mediastinum. No purulent discharge was observed. The remainder of the esophagus, stomach and duodenum appeared normal apart from mild antral gastritis. Because the fistulous tract exposed the adventitia and the surrounding esophageal wall appeared structurally compromised, mucosal biopsies were deferred to minimize the risk of full-thickness perforation. An EUS examination was performed in the same session using a linear echoendoscope (Olympus UCT-180) with an EU-ME2 ultrasound processor; and (2) EUS findings: From the distal esophagus and gastric cardia, multiple enlarged subdiaphragmatic lymph nodes were visualized along the left gastric artery and around the celiac axis and aortocaval region. The largest node at the left gastric territory measured approximately 21 mm × 15 mm, was oval, hypoechoic and well circumscribed with some central heterogeneity, suggestive of necrosis. Several additional nodes in the celiac and aortocaval regions measured between 10 and 18 mm along the short axis with similar echotexture. The pancreas was globally enlarged with relatively homogeneous hypoechoic parenchyma and preserved lobularity; there were no discrete focal masses, cystic lesions or obvious main duct irregularities. The common bile duct was normal in caliber. No peripancreatic fluid collections or vascular invasion were seen. The proximal esophageal wall at approximately 18 cm from the incisors appeared thickened with disruption of the normal layered architecture and focal thinning of the muscularis propria corresponding to the fistulous site. There was no clear fistulous communication with mediastinal lymph nodes or the tracheobronchial tree on EUS. No mediastinal abscess was identified (Video 1 and Video 2).

EUS-FNA procedure and cytopathology: EUS-guided FNA was performed targeting the largest left gastric lymph node, which was adjacent to the gastric wall and safely accessible. (1) Needle type and gauge: 22-gauge EUS-FNA needle (Cook EchoTip 22G); (2) Number of passes: Three passes using a standard suction technique (10 mL syringe), with fanning to sample different parts of the node; (3) Rapid on-site evaluation was not available at our center; therefore, sample adequacy was based on endosonographer experience and subsequent cytopathology; (4) Choice of FNA vs core biopsy: FNA was selected because the target node was relatively soft and hypoechoic with suspected necrosis, FNA has high diagnostic accuracy for tuberculous lymphadenitis, and because FNA was readily available at our institution, whereas dedicated core needles were not immediately accessible at the time of the procedure. In addition, FNA was deemed sufficient to differentiate TB from malignancy or lymphoma in this clinical context[16-18]. Specimens were processed as both direct smears and cell block preparations. Portions of the aspirated material were sent for Ziehl-Neelsen (ZN) staining, mycobacterial culture and nucleic acid amplification testing (NAAT) when feasible; and (5) Cytopathological findings: Cytology from the lymph node aspirate showed multiple well-formed, florid epithelioid granulomas with central caseous necrosis and numerous Langhans-type multinucleated giant cells at the periphery (Figure 2). Background lymphocytes and occasional neutrophils were present. There were no malignant cells, Reed–Sternberg cells or features suggestive of lymphoma. ZN stain: Negative for acid-fast bacilli (AFB). Periodic acid–Schiff and Grocott methenamine silver stains: Negative for fungal organisms.

Figure 2 Histological slide showing granulomatous changes (arrow).

Mycobacterial culture of the EUS-FNA specimen did not yield growth, likely due to the paucibacillary nature of the lymph node sample. TB polymerase chain reaction (PCR) (NAAT) was not locally available on cytology specimens at the time of this case. Blood cultures were sterile. Overall, the cytomorphological appearance was highly suggestive of tuberculous lymphadenitis in the appropriate clinical context.

Further investigations and diagnostic reasoning: In view of the imaging findings (circumferential proximal esophageal thickening, subdiaphragmatic lymphadenopathy, bulky pancreas), the initial differential diagnosis included the following: Lymphoma (particularly non-Hodgkin lymphoma), metastatic malignancy involving lymph nodes and esophagus, primary esophageal carcinoma with regional lymphadenopathy, autoimmune pancreatitis with associated IgG4-related sclerosing disease, sarcoidosis, fungal infection (e.g., histoplasmosis), Crohn’s disease with upper gastrointestinal involvement, tuberculous lymphadenitis with esophageal and pancreatic involvement.

Features supporting TB included: (1) Epidemiological background: Origin from a TB-endemic region with likely latent TB exposure; (2) Constitutional symptoms: Weight loss, anorexia and night sweats; (3) Imaging: Multiple discrete enlarged abdominal lymph nodes with central hypoattenuation/necrosis rather than confluent bulky masses, which is more typical for lymphoma or metastatic disease; diffuse pancreatic enlargement without a discrete mass or duct cut-off, which may be seen in pancreatic TB or autoimmune pancreatitis[12-15]; and (4) EUS-FNA cytology: Florid necrotizing granulomatous inflammation with Langhans-type giant cells, in the absence of malignant cells, highly suggestive of TB.

Positive IGRA supporting Mycobacterium infection: Lymphoma and metastatic malignancy were considered less likely because of the absence of atypical cells on cytology and the distinct granulomatous pattern. Autoimmune pancreatitis was considered but was less favored because of the absence of characteristic imaging features, such as “sausage-shaped” pancreas with capsule-like rim and irregular ductal narrowing, lack of extra-pancreatic IgG4-related disease, and because pancreatic histology was not obtainable without undue risk[14]. Serum IgG4 levels were not elevated (measured subsequently; 0.9 g/L, within normal range). Sarcoidosis was less likely because of the presence of caseating rather than non-caseating granulomas and the absence of pulmonary or other organ involvement. Fungal infections were not supported by the clinical context, negative special stains and lack of epidemiologic exposure.

Consequently, after multidisciplinary discussion involving gastroenterology, radiology, infectious diseases, thoracic surgery and microbiology, extrapulmonary TB involving intra-abdominal lymph nodes with associated proximal esophageal fistula and probable pancreatic involvement was considered the most likely unifying diagnosis. The absence of pulmonary parenchymal disease on CT increased the diagnostic challenge but did not exclude TB.

FINAL DIAGNOSIS

Extrapulmonary TB presenting as intra-abdominal tuberculous lymphadenitis (subdiaphragmatic, left gastric and aortocaval lymph nodes) confirmed by EUS-FNA cytology; proximal esophageal involvement with a very short, blind-ended esophageal fistula at 18 cm from the incisors, highly suggestive of esophageal TB in this context; and probable pancreatic TB based on imaging (diffuse enlargement without focal mass) and its regression with anti-TB therapy.

TREATMENT

The patient was started on standard first-line anti-TB therapy following national guidance and in alignment with World Health Organization (WHO) and national (e.g., Index-TB) recommendations for extrapulmonary TB[19-21]. Dosing was weight-based according to his baseline weight of approximately 64 kg, as follows: Isoniazid: 300 mg once daily. Rifampicin: 600 mg once daily. Pyrazinamide: 1500 mg once daily. Ethambutol: 1200 mg once daily. Pyridoxine (vitamin B6) 10-25 mg daily was co-prescribed to prevent isoniazid-related neuropathy. The initial intensive phase consisted of all four drugs for 2 months, followed by a planned continuation phase with isoniazid and rifampicin.

Given the extent of disease (intra-abdominal lymphadenitis, esophageal fistula and probable pancreatic involvement) and after multidisciplinary discussion, a 9-month total treatment duration (2-month intensive phase + 7-month continuation phase) was initially planned. This was later extended to 12 months, as detailed below.

The patient received adherence counseling and was followed closely in the TB clinic. Baseline liver function tests and renal function were normal, and monthly monitoring was instituted. He did not develop clinically significant hepatotoxicity, optic neuritis or other serious adverse drug reactions. Transient mild elevation of alanine aminotransferase (peak 48 U/L at 4 weeks) resolved spontaneously without treatment interruption.

OUTCOME AND FOLLOW-UP

PR during therapy

At approximately 8 weeks after initiation of anti-TB therapy, the patient was re-admitted with worsening epigastric pain and abdominal discomfort, without fever or new systemic symptoms. He reported good adherence to therapy, supported by pill counts and pharmacy refill records. There were no new respiratory symptoms. Repeat laboratory tests showed stable hemoglobin (12.6 g/dL), normal white cell count, slightly reduced C-reactive protein (18 mg/L) and normal liver and renal function. HIV status remained negative. An interval contrast-enhanced CT of the abdomen and pelvis demonstrated a slight increase in the size and number of subdiaphragmatic lymph nodes (largest left gastric node 22 mm × 19 mm vs 22 mm × 16 mm initially), with more central low attenuation suggestive of necrosis (Figure 1B). No new lymph node stations were involved. Pancreatic size remained similar, with no new focal lesion or duct obstruction. There were no new pulmonary lesions or mediastinal lymphadenopathy. The clinical and radiological picture raised concern regarding potential treatment failure or drug resistance vs a PR (immune reconstitution inflammatory syndrome-like phenomenon) commonly described in TB, particularly in lymph node disease. A comprehensive evaluation was therefore undertaken.

Differentiation between PR and treatment failure/drug resistance

The following considerations favored a paradoxical TB reaction: Good self-reported adherence corroborated by pharmacy records. Clinical stability without high-grade fever, new focal deficits or signs of systemic deterioration. Partial improvement in symptoms (reduced dysphagia, modest weight gain of approximately 2 kg) before the transient worsening of abdominal pain. Radiology revealed increased necrosis within previously involved lymph nodes rather than widespread new disease sites. There was no microbiological or radiological evidence of new active infection elsewhere.

Risk factors such as young age and lymph node TB are associated with PR, occurring in approximately 20%-30% of patients with tuberculous lymphadenitis[22,23]. No clear risk factors for drug-resistant TB were identified.

Repeat blood cultures were negative. Repeat sampling of lymph nodes for culture or GeneXpert MTB/RIF was considered ultimately deferred because of the patient’s improving clincal status, the procedural risks, and the high likelihood of paucibacillary disease. There was no evidence of rifampicin resistance in the limited microbiological tests that had been performed on sputum (induced sputum NAAT was negative; culture was negative). Although these investigations could not definitively exclude drug resistance, the overall clinical picture was most compatible with a PR.

Consequently, the anti-TB regimen was continued unchanged, with close clinical and radiological follow-up. Adjunctive corticosteroids were discussed in the multidisciplinary team meeting. Given the absence of life-threatening complications (e.g., airway compromise, central nervous system involvement or severe systemic deterioration) and the risk of steroid-related adverse events, corticosteroids were not initiated. Non-steroidal anti-inflammatory drugs (NSAIDs) were used briefly for analgesia.

Treatment duration and follow-up

The initial plan for 9 months of therapy was revised during follow-up MDT discussions considering several factors, including the complexity of disease with fistulating proximal esophageal involvement and probable pancreatic TB, the Interim PR at 2 months, and the need to ensure durable radiological and clinical resolution of the fistulae and intra-abdominal lymphadenopathy.

Current WHO guidelines and several national recommendations suggest that a 6-month regimen is adequate for most forms of drug-susceptible extrapulmonary TB, but longer courses (9-12 months) should be considered for complicated sites such as central nervous system, skeletal, disseminated disease, or when clinical response is suboptimal[19-21]. For gastrointestinal TB, including complex fistulating or obstructive disease, many experts suggest longer treatment durations (9-12 months), supported by observational data and expert consensus[1,2,19-21].

In this case, given the absence of pulmonary disease but the presence of esophageal fistula and probable pancreatic involvement, the multidisciplinary team elected to extend total therapy to 12 months (2 months of isoniazid, rifampicin, pyrazinamide and ethambutol, followed by 10 months of isoniazid and rifampicin). This decision was individualized and is not intended to suggest that all esophageal TB requires prolonged treatment.

The patient was reviewed at 1-2 monthly intervals in the TB clinic. Key clinical milestones are summarized in Table 1 and Figure 1B (timeline schematic). At 4 weeks, dysphagia improved from solids and liquids to occasional difficulty with large boluses only; abdominal pain intensity decreased by approximately 30%-40% on a visual analogue scale. Weight stabilized at 65 kg. At 8 weeks (PR), abdominal pain transiently worsened; dysphagia remained improved, weight was 66 kg. Between 3 and 6 months, the abdominal pain progressively improved with near-complete resolution by month 6; dysphagia completely resolved by approximately 3-4 months. Weight increased steadily to 70 kg by 6 months (net gain of 6 kg from baseline). At 9 months, the patient was asymptomatic and his weight was 72 kg. He did not experience any adverse drug reactions, and all liver function tests remained within the normal range. At 12 months (end of therapy), the patient’s weight was 74 kg, representing a full return to pre-morbid weight. He did not have any abdominal pain or dysphagia, and his appetite and energy levels were normal.

Table 1 Timeline of clinical course, imaging and treatment.

Time point	Key events and findings
Month-6 to 0	Onset and progression of epigastric pain, dyspepsia and weight loss; initial endoscopy showing Helicobacter pylori gastritis; successful eradication but persistent symptoms and new dysphagia
Month 0	Barium swallow: Proximal esophageal deviation and outpouching; baseline CT: Circumferential proximal esophageal thickening, subdiaphragmatic lymphadenopathy (largest node 22 mm × 16 mm), diffuse pancreatic enlargement (head AP diameter 32 mm); baseline weight about 64 kg
Month 1	Repeat endoscopy: Very short blind-ended proximal esophageal fistula at 18 cm from incisors; EUS: Multiple enlarged subdiaphragmatic nodes; EUS-FNA: Necrotizing granulomatous lymphadenitis consistent with TB; IGRA positive; initiation of 4-drug anti-TB therapy
Month 2	Transient paradoxical reaction: Worsening abdominal pain; CT: Increase in lymph node size (largest 26 mm × 18 mm) with more central necrosis; no new sites of disease; adherence confirmed; regimen continued unchanged; weight about 66 kg
Month 3–4	Improvement in dysphagia (complete resolution by 3-4 months); progressive reduction in abdominal pain; weight gain to 68-69 kg
Month 6	Near-complete resolution of abdominal pain; weight about 70 kg; interim imaging (not shown) indicating decreasing lymph node size and pancreatic dimensions
Month 9	Patient asymptomatic; weight about 72 kg; continued HR therapy; MDT decision to complete 12 months in total
Month 12	Completion of 12-month anti-TB course; barium swallow: No residual fistula; CT: Near-complete regression of lymphadenopathy (largest node about 7 mm), pancreatic head AP diameter about 24 mm with normal contour; weight about 74 kg
Month 18 (follow-up)	Continued clinical well-being; no recurrence of symptoms (telephone follow-up)

CT: Computed tomography; TB: Tuberculosis; HR: Isoniazid (H) and rifampicin (R); MDT: Multidisciplinary team; AP: Anteroposterior; EUS-FNA: Endoscopic ultrasound-guided fine-needle aspiration; IGRA: Interferon-γ release assay.

Post-treatment imaging and outcome

Barium swallow (post-treatment, 12 months): Demonstrated normal caliber and contour of the esophagus without residual outpouching or contrast retention. There was no evidence of persistent fistula or diverticulum (Figure 1B).

Contrast-enhanced CT (post-treatment, 12 months): Resolution of circumferential thickening of the proximal esophagus.

Near-complete regression of subdiaphragmatic lymphadenopathy: The largest left gastric node decreased from 22 mm × 19 mm at baseline (peak 26 mm × 18 mm at PR) to 3 mm in short-axis diameter at 12 months (> 70% reduction) (Figure 1B).

Marked reduction in pancreatic size: The anteroposterior diameter of the pancreatic head decreased from 32 mm at baseline to 24 mm at 12 months, with restoration of normal contour and echotexture. There was no focal lesion, and the main pancreatic duct remained normal.

These findings confirmed effective response to therapy. The proximal esophageal fistula was no longer visible endoscopically or radiologically. At 18 months after therapy completion, telephone follow-up reported continued well-being without symptom recurrence.

DISCUSSION

Esophageal TB is extremely rare, accounting for only approximately 0.2% of gastrointestinal TB[7,8]. Most cases are secondary to contiguous spread from infected mediastinal lymph nodes or pulmonary disease and typically involve the mid- or distal esophagus. Reported manifestations include ulceration, strictures, pseudotumours and fistulas to the trachea, bronchi or mediastinum[7-11,24,25]. Pancreatic TB is also rare and usually presents as a mass-forming lesion mimicking pancreatic carcinoma[12-15].

This case is unique in several respects. First, the proximal esophageal fistula at 18 cm from the incisors was localized to the cervical/upper thoracic esophagus, a less common site for TB involvement. It presented as a very short blind-ended tract without demonstrable communication to the airway or mediastinum, distinct from the more commonly reported esophago-pulmonary or esophago-mediastinal fistula[10,11,24,25]. Second, the triad of proximal esophageal fistula, intra-abdominal tuberculous lymphadenitis and probable pancreatic TB is rare. To our knowledge, only a few reports have described concomitant esophageal TB and pancreatic TB, and even fewer involved a short, blind-ended proximal esophageal fistula in an immunocompetent, HIV-negative young man without pulmonary involvement[10-12,24,25]. Third, the absence of pulmonary TB involvement on CT increased diagnostic uncertainty, as clinicians often associate TB with primary pulmonary disease. Finally, the paradoxical worsening of lymphadenopathy at 2 months in an otherwise improving young man with lymph node TB illustrates a recognized but challenging phenomenon that can be mistaken for treatment failure or drug resistance[22,23].

By comparing this case with previously published reports of esophageal TB, including cases with esophago-pulmonary fistulas, mediastinal abscesses and esophageal masses mimicking malignancy[7-11,24,25], our report adds to the spectrum of presentations, underscores the possibility of proximal esophageal fistulation and highlights the utility of EUS-guided sampling in differentiating TB from malignancy.

The patient’s symptoms and CT findings lacked specificity and overlapped with several conditions; circumferential proximal esophageal thickening and dysphagia raised concern for malignancy. However, the patient’s young age, lack of risk factors (heavy smoking, alcohol misuse), absence of a discrete mass or ulcerating lesion on endoscopy, and granulomatous lymph node histology argued against carcinoma. Generalized lymphadenopathy and systemic symptoms could suggest lymphoma, particularly non-Hodgkin lymphoma. Nevertheless, cytology showed classic necrotizing granulomas without atypical lymphoid cells or Reed-Sternberg cells, and nodes were well circumscribed with central necrosis typical of TB rather than homogeneous large masses common in lymphoma[17,18]. No primary tumor was identified, and cytology remained negative for malignant cells.

Upper gastrointestinal Crohn’s involvement can cause strictures and fistulae. However, there was no history of diarrhea, perianal disease or other intestinal involvement; colonoscopy did not show inflammatory changes (performed as part of workup, normal), and granulomas were caseating rather than non-caseating. Diffuse pancreatic enlargement and abdominal pain raised suspicion for autoimmune pancreatitis (AIP). Yet, the absence of typical imaging features (capsule-like rim, “sausage-shaped” pancreas), normal serum IgG4 levels, lack of other IgG4-related organ involvement and the presence of necrotizing granulomatous lymphadenitis supported TB over AIP[14].

Negative fungal stains and lack of epidemiologic exposure reduced the likelihood of histoplasmosis or other fungal infections; sarcoidosis is characterized by non-caseating granulomas and rarely causes esophageal fistulas or pancreatic enlargement. In this context, the combination of epidemiologic risk (endemic region), constitutional symptoms, imaging features of necrotic lymphadenopathy and bulky pancreas, IGRA positivity and EUS-FNA-confirmed necrotizing granulomatous lymphadenitis made TB the most plausible diagnosis.

The pancreatic TB diagnosis was based on imaging and treatment response rather than histopathology, and we therefore describe it as “probable” pancreatic TB or “imaging-suggested pancreatic involvement” to avoid overstating diagnostic certainty.

Pancreatic TB may present as: (1) A focal mass, often in the pancreatic head; (2) Diffuse enlargement mimicking AIP; or (3) Cystic/abscess-like lesions[12-15]. Radiologically, it can be virtually indistinguishable from pancreatic carcinoma or AIP, and tissue diagnosis via EUS-FNA/B or surgical biopsy is often required for confirmation[12,13,15]. However, obtaining pancreatic tissue may be challenging in patients with coexisting comorbidities or high procedural risk.

In our patient, the pancreas was diffusely enlarged without a discrete mass or duct obstruction, and EUS did not reveal a safe focal lesion amenable to targeted FNA without excessive risk. Because the patient already had conclusive evidence of TB from lymph node FNA and showed marked radiological regression of pancreatic enlargement with anti-TB therapy, obtaining pancreatic tissue was deemed not essential. This reflects common real-world practice, in which extrapulmonary TB at one site plus compatible imaging and response to therapy at another site can be sufficient to infer involvement, especially where risks of biopsy outweigh potential benefits.

This case highlights the diagnostic complexity of pancreatic TB and underscores the importance of integrating epidemiological data, imaging characteristics, EUS findings and therapeutic response when histological confirmation is not feasible.

Role and performance of EUS-guided tissue acquisition

EUS-FNA/B has become a first-line tool for evaluating intra-abdominal lymphadenopathy of unknown etiology because it allows high-yield sampling of nodes in anatomically challenging regions (e.g., peripancreatic, celiac, para-aortic) with low complication rates[16-18].

Several studies have demonstrated its value in diagnosing tuberculous lymphadenitis. For example, Rao et al[17] reported a sensitivity of 83.3% and specificity of 94.7% for EUS-FNA/B in diagnosing abdominal TB lymphadenitis. Pausawasdi et al[16] described sensitivity and specificity above 85%-90% in mixed cohorts with malignant and benign nodes. Lee et al[18] showed that EUS-guided core needle biopsy (ProCore) could provide ample tissue for histology and culture in intra-abdominal TB, with diagnostic yields exceeding 85%.

In our center, 22-gauge FNA needles are routinely used for suspected TB because they provide adequate cytological material with minimal trauma and low cost. Although core biopsy may offer higher yields for culture and NAAT, FNA remains a pragmatic and effective choice in many settings. In this case, FNA yielded characteristic necrotizing granulomas and giant cells sufficient to establish a working diagnosis and exclude malignancy. We also attempted mycobacterial culture, which was negative, a common limitation in paucibacillary nodal TB. Where available, TB PCR on FNA material can further increase sensitivity and provide rapid drug-resistance information; however, this was not accessible at our institution for cytology specimens at the time.

Microbiological testing and AFB-negative disease

AFB smear negativity on ZN staining is common in extrapulmonary and nodal TB because of low bacillary loads[1,2,19]. In our patient, ZN staining on lymph node aspirates was negative, and the culture did not grow Mycobacterium. Sputum cultures and NAAT were also negative. Despite this, the combination of IGRA positivity, necrotizing granulomatous inflammation, epidemiologic risk and characteristic imaging strongly favored TB over other granulomatous conditions.

International guidelines emphasize that extrapulmonary TB is often diagnosed on a composite of clinical, radiological, histopathological and immunological data, especially when microbiological confirmation is elusive[19-21]. Our case illustrates this pragmatic approach and the frequent need to initiate empirical therapy in AFB-negative but histology-positive disease.

PR: Frequency, timing and management

PRs are defined as clinical or radiological worsening of pre-existing TB lesions or the development of new lesions after initial improvement on effective anti-TB therapy, in the absence of alternative explanations such as non-adherence, drug resistance or other infections[22,23]. PRs are thought to reflect an exaggerated immune response to mycobacterial antigens after therapy initiation.

In lymph node TB, PRs are reported in approximately 20%-30% of patients, particularly in young adults, females and those with high baseline inflammatory burden[22,23]. PRs usually occur within 1-3 months of starting therapy, aligning with the 2-month onset in our patient. The typical manifestations include enlargement of existing lymph nodes, appearance of new nodes, suppuration, sinus formation or worsening systemic symptoms.

In our case, the PR manifested as transient worsening of abdominal pain and radiological enlargement of subdiaphragmatic lymph nodes with increased central necrosis, without systemic deterioration. We systematically assessed for alternative causes and confirmed the patient’s adherence and lack of drug interactions, did not identify any clinical or laboratory evidence of sepsis or other infections, new organ involvement on imaging, risk factors for multidrug-resistant TB, or evidence of rifampicin resistance on limited NAAT.

Because the patient remained clinically stable with evidence of prior partial response (improved dysphagia, weight gain), the multidisciplinary team concluded that the findings were most consistent with a PR rather than treatment failure. Anti-TB therapy was continued unchanged, and the patient improved without the need for regimen modification.

Adjunctive corticosteroids are sometimes recommended for severe PRs, particularly when vital structures (e.g., central nervous system, airway) are threatened[22,23]. In our case, the PR was moderate and self-limiting; therefore, NSAIDs were used for symptom relief and corticosteroids were withheld to avoid immunosuppression and additional adverse effects. This case underscores the importance of recognizing PRs, differentiating them from treatment failure and drug resistance, and counseling patients about this potential complication to avoid premature discontinuation or inappropriate escalation of therapy.

Treatment duration, regimen and outcomes

Most guidelines, including WHO and national protocols, recommend a 6-month regimen (2HRZE/4HR) for drug-susceptible extrapulmonary TB, except for specific forms such as TB meningitis and bone/joint TB, where 9-12 months is advised[19-21]. For abdominal TB, several studies and expert opinions support at least 6 months of therapy, with extension to 9-12 months in complex or slow-responding cases, especially those with strictures, fistulas or mass-forming lesions[1,2,19-21].

Evidence specific to esophageal TB is limited to case series and case reports. Most esophageal TB cases managed with 6-9 months of standard therapy show good outcomes[7-11,24,25]. However, in instances of fistulizing disease, mediastinal abscess or delayed healing, prolonged therapy up to 12 months has been used with favorable results[10,11,24,25].

In our patient, the decision to extend therapy to 12 months was based on the following: Fistulating proximal esophageal disease with initial structural disruption of the esophageal wall, probable pancreatic involvement that can be slow to resolve, intercurrent PR at 2 months, and the desire to maximize the chance of permanent radiological and clinical resolution without recurrence.

We emphasize that this extended duration was an individualized decision made by a multidisciplinary team and does not imply that all esophageal TB requires 12 months of treatment. In uncomplicated esophageal TB (e.g., small ulcers without fistula or strictures), a standard 6-month regimen may be adequate, consistent with guideline recommendations[19-21].

Our patient tolerated therapy well, with only mild transient liver enzyme elevation that did not necessitate treatment interruption. No optic neuritis, significant gastrointestinal intolerance or rash occurred. Adherence remained high throughout.

Anatomical considerations of the esophageal lesion

The fistulous lesion at 18 cm from the incisors corresponds anatomically to the cervical/upper thoracic esophagus, just distal to the upper esophageal sphincter and above the aortic arch. Proximal esophageal TB is rare because most secondary spread originates from lower mediastinal or hilar lymph nodes; however, cervical or upper mediastinal nodes can occasionally be involved[7-11,24,25].

In this case, there was no clear EUS or CT evidence of a direct communication between the fistulous tract and mediastinal lymph nodes, trachea or bronchus, and no mediastinal abscess was identified. Therefore, the lesion was best described as a very short blind-ended fistulous tract or diverticulum-like outpouching, likely representing a sequela of transmural granulomatous inflammation and micro-perforation that subsequently sealed off during healing. Pre- and post-treatment endoscopic and radiologic images (Figures 1B and 2) illustrate the evolution from a structurally disrupted segment with a blind-ended tract to a normal-appearing esophageal lumen without residual deformity.

Esophageal TB typically presents with dysphagia, odynophagia, retrosternal pain or weight loss and may mimic malignancy on endoscopy and imaging[7-11,24-26]. Endoscopic findings include ulcerative lesions, nodularity, strictures or pseudotumor masses[7-11]. Biopsy is diagnostic in many cases, but may be limited by risk of perforation in deeply ulcerated or fistulating lesions. Adjacent mediastinal lymphadenopathy is frequent, and concomitant pulmonary TB is seen in many but not all cases[7-11,24-26]. Anti-TB therapy is usually effective, with most lesions healing without the need for surgery, although strictures may require dilatation or stenting.

Pancreatic TB, though rare, is an important differential for pancreatic masses, especially in young patients from endemic regions[12-15]. Clinical presentation often mimics pancreatic carcinoma (abdominal pain, jaundice, weight loss) or autoimmune pancreatitis. EUS-FNA/B plays a crucial role in obtaining tissue; histology typically shows granulomatous inflammation, and AFB may or may not be demonstrable[12-15,18]. Anti-TB therapy leads to substantial regression or complete resolution in most reported cases, obviating the need for major pancreatic surgery[12-15]. Our case reinforces these points and further demonstrates that concomitant esophageal and pancreatic manifestations may occur in the absence of pulmonary disease, making tissue diagnosis and high clinical suspicion essential.

Ethical considerations

Written informed consent for publication of clinical details and anonymized images was obtained from the patient. The case report was conducted in accordance with the ethical principles of the Declaration of Helsinki (2013 revision) and complied with local institutional policies for case report publication. Formal institutional review board approval was not required for a single anonymized case report according to our institutional guidelines.

CONCLUSION

This case illustrates an unusual presentation of extrapulmonary TB as a combination of a very short proximal esophageal fistula, intra-abdominal tuberculous lymphadenitis and probable pancreatic involvement in an immunocompetent young man from a TB-endemic region, without pulmonary disease. The non-specific nature of his symptoms and imaging findings initially suggested malignancy or other inflammatory conditions.

ACKNOWLEDGEMENTS

We thank the multidisciplinary team, including colleagues in radiology, infectious diseases, thoracic surgery and pathology, for their contributions to the diagnosis and management of this patient. We are grateful to the patient for consenting to share his case for educational purposes.