Abstract
Isolated pancreatic and peripancreatic nodal tuberculosis may mimic pancreatic malignancy and may result in unnecessary surgery. Being a treatable disease, it is desirable to diagnose this condition without a laparotomy. Pancreatic tuberculosis should be considered in the differential diagnosis of a peripancreatic mass if the patient is young, residing in endemic tuberculosis regions, or has had a history of tuberculosis. We report our experience with pancreatic tuberculosis. Over a period of 12 years, we have managed 16 such patients, of whom 10 were diagnosed preoperatively. In six, the diagnosis was only made after extensive surgery. All 16 patients received antitubercular drugs and were well at a median follow-up of 52 months. Although pancreatic tuberculosis can be diagnosed by modern image-guided biopsy in a fair number of patients, if a high index of suspicion is maintained, a laparotomy may still be unavoidable in a subset of patients.
Introduction
Tuberculosis (TB) is still a major public health problem in low- and middle-income countries. Its incidence is increasing because of the emergence of human immunodeficiency virus (HIV) infection. Extrapulmonary TB accounts for 10–30% of all cases, 1 of whom >5% have abdominal involvement. 2 However, pancreatic or peri-pancreatic involvement is rare, usually occurring as a part of disseminated TB. 3 Isolated involvement of the pancreas is even rarer, 4 and because of this rarity, the natural course of the disease is currently unknown. Several case reports have included a detailed review of this subject, but a complete clinical picture is still lacking.
It is often very difficult to differentiate pancreatic TB from malignancy, both clinically and radiologically. As most cases of pancreatic and peripancreatic nodal TB respond well to antituberculous therapy, every effort should be made to diagnose this condition before subjecting a patient to an unnecessary laparotomy, or worse, a pancreatic resection. The aim of our study is to report our relative success in this endeavour.
Methods
Ours was an observational retrospective review study. All consecutive patients, who were diagnosed with isolated pancreatic or peripancreatic nodal tuberculosis between August 2007 and July 2019 in our Surgical Department, were included. Those with miliary TB were excluded. Approval was obtained from our local ethical committee. Informed patient consent was waived as data were anonymised.
Data pertaining to demographic features, signs and symptoms, duration of illness, imaging (chest radiography, transabdominal ultrasound, computerised tomography (CT) scan, magnetic resonance imaging, endoscopic ultrasound (EUS)), side-viewing endoscopy, cytology and/or histopathology, operative findings, therapeutic procedures and follow-up with anti-tubercular therapy (ATT) were obtained from our prospectively maintained data base.
Samples for cytology and/or histology were obtained by EUS or CT-guided aspiration and during surgery. The histological findings of granuloma with caseating necrosis (Figure 1) or identification of acid fast bacilli (Figure 2) were used as criteria for diagnosis. Our study was conducted in line with the PROCESS criteria.
5
May–Grunwald Giemsa stain showing epithelioid cell granuloma and caseation necrosis. Ziehl–Neelsen stain showing acid fast bacilli.
Results
Our 16 patients consisted of 12 men, and had a median age of 30 (range: 13–59) years. The median duration between the onset of symptoms and diagnosis was 4 (range: 1–12) months. Presentation is shown in Table 1. The most common symptom was epigastric pain (81%). One patient had a history of haemosuccuspancreaticus (bleeding from the ampulla of Vater). Two patients (12%) had a past history of pulmonary TB (10 and 30 years previously). Only one patient was HIV positive. Ten patients (62%) were anaemic, and the erythrocyte sedimentation rate was markedly raised in 11 patients (69%). Chest radiographs were performed in all patients but did not reveal any evidence of tuberculosis. Radiology and histopathological details of all patients are presented in Table 2. CT scan of the abdomen was likewise obtained in all 16 patients and EUS in 6. The bile duct was dilated in five and the pancreatic duct in three. Twelve (75%) showed enlarged peripancreatic lymph nodes with characteristic central necrosis and rim enhancement (Figure 3).
CECT abdomen showing peripancreatic lymph nodes with central necrosis and peripheral rim enhancement. Flow chart for selection of patients. Pan TB: pancreatic tuberculosis. Clinical presentation of Pancreatic TB patients (n = 16). Radiology and histopathological details of pancreatic TB patients (n = 16). HOP: head of the pancreas; CBD: common bile duct; MPD: main pancreatic duct; LN: lymph node; EUS: endoscopic ultrasound; TB: tuberculosis; FNA: fine needle aspiration; SMV: superior mesenteric vein; SMA: superior mesenteric artery; PV: portal vein; HP: histopathology; M: male; F: female.
As shown in Figure 4, 10 (62.5%) patients did not undergo surgery: in eight, a clinical suspicion of pancreatic or peripancreatic tuberculosis was made prior to tissue sampling and two were considered to be unresectable pancreatic head malignancy. Pancreatic TB was diagnosed in these after CT or EUS-guided fine needle aspiration (FNA). In six (37.5%) patients, diagnosis was revealed at laparotomy. Their preoperative diagnosis included pancreatic head malignancy in four and pseudocyst in two. Of the former, two had undergone palliative biliary-enteric by-pass and gastrojejunostomy on the presumption of malignancy. Two others underwent Whipple’s pancreaticoduodenectomy. The clinical parameters and imagery of these latter were in favour of a resectable pancreatic head malignancy. Subsequent histological examination of retrieved peripancreatic lymph nodes showed features of tuberculosis. In one of the two patients with a working diagnosis of pseudocyst, laparotomy revealed a pancreatic abscess, the pus from which showed many acid fast bacilli. In the other, a biopsy from the cyst wall showed features of tuberculosis. The first underwent external drainage of the abscess, and the second internal drainage.
Four patients required endoscopic biliary stenting (Table 2). Two stented patients underwent laparotomy for suspicion of malignancy with intent to primary resection (serial nos.: 12 and 13). The other two stented patients (serial nos.: 1 and 2) were initially managed conservatively after endoscopic retrograde stenting to relieve cholestasis prior to ATT. One of these developed a retropancreatic biliary stricture where stenting failed to resolve over a period of one year; thus, the obstruction was finally treated with a Roux-en-Y choledochojejunostomy.
Tissue for pathological confirmation was obtained from CT-guided FNA in five, EUS in five more and surgery in six. Caseating necrosis was found in 15 patients, but acid fast bacilli were only found in 6.
Conventional ATT for six months was used in all patients. One patient who developed drug-induced hepatitis was treated with a modified three-drug regimen for nine months. Ten patients showed symptomatic improvement within two weeks of starting ATT, whilst in the remainder, it took up to two months. All patients had resolution of their symptoms at a median follow-up of 52 (range: 6–138) months.
Discussion
Isolated pancreatic TB is extremely rare even in endemic areas, 4 the explanation for which may be due to the anti-mycobacterial effect of pancreatic lipase and deoxyribonucleases. 6 The pancreas may be involved in TB by direct extension, lympho-haematogenous dissemination or following reactivation of previous abdominal tuberculosis.
Its diagnosis is a real challenge partly because of its rarity but mainly because of its non-specific presentation, which mimics malignancy. Classic features of pancreatic TB have been described as mostly occurring in young people, especially female, having a history of TB or coming from an endemic zone, presenting with epigastric pain, fever and weight loss, with imagery demonstrating a pancreatic mass or peripancreatic nodules, some with focal calcification. 4 However, many other presentations such as cholestatic jaundice, acute or chronic pancreatitis, pancreatic abscess and portal vein thrombosis are reported.
The differentiation from malignancy is difficult. In a significant percentage, diagnosis is only made after extensive surgery.7–10 There are, however, some important diagnostic clues. In pancreatic TB, normal appearances of the common bile and pancreatic ducts are often seen, even if the mass is positioned centrally in the head of the pancreas (Figure 5).
11
This is in sharp contrast to adenocarcinoma of the pancreas where the pancreatic duct is dilated in centrally located tumours of the pancreatic head.
12
However, vascular involvement cannot be used as a criterion to discriminate pancreatic TB from malignancy as there are multiple reports of vascular invasion in pancreatic TB.13,14 The characteristic CT finding of extensive necrosis of peripancreatic lymph nodes with rim enhancement was a frequent feature in our series.
CECT abdomen showing a mass in the head of the pancreas with necrosis and normal appearing pancreatic duct.
However, an image-guided biopsy is virtually essential for a definitive diagnosis of pancreatic TB. EUS-guided FNA is our preferred method because of a lower chance of needle tract dissemination. 15 CT- or US-guided FNA can however safely be performed in experienced hands when EUS-guided FNA is not helpful or unavailable. Routine pre-operative tissue sampling, though, may not be in protocols for resectable periampullary or pancreatic head lesions. This will lead to failure to diagnose pancreatic TB preoperatively as occurred in two of our patients. Pre-surgical tissue sampling is normally done if CT reveals a locally advanced or metastatic disease or when clinical and imaging findings are suggestive of a lesion other than pancreatic adenocarcinoma (viz. chronic pancreatitis, TB or lymphoma).
Routine use of frozen biopsy in all cases of pancreatectomy is debatable. It is performed commonly to assess the biliary and pancreatic margins of excision. Differentiating between benign and malignant is, however, the least common reason. 16 Surgeons normally plan their surgery approach depending on accurate preoperative imaging and so depend much less on intraoperative frozen section for the diagnosis. 17 Furthermore, differentiating between chronic pancreatitis and adenocarcinoma can be very hazardous, where a false positive result is not unknown. 18 A TB diagnosis on frozen section is even more difficult, as necrotic tissue may only be obtained. Frozen section is, nonetheless, useful for evaluating lymph node status and suspected extra-pancreatic spread.
Polymerase chain reaction (PCR) is a promising new approach for the diagnosis of gastrointestinal TB. However, its use for the diagnosis of pancreatic TB has not been well studied. A recent systematic review 9 documented that TB was only identified by PCR in 9.6%. Indeed, positive results were obtained very variably (43–80%).19–21 The presence of PCR positivity however confirms the diagnosis, whereas the presence of granulomas is non-specific, occurring in other diseases such as sarcoidosis, auto-immune pancreatitis and Crohn’s disease. 22 Another advantage of TB–PCR is its rapidity giving a result in <24 h.
Some patients with cholestasis need either endoscopic or surgical intervention to relieve biliary obstruction as ductal narrowing may persist despite treatment with ATT.7,8,23,24 Furthermore, drainage of a tubercular pancreatic abscess or pseudocyst mandates some kind of surgical intervention.
Thus, surgery may not always be avoided in pancreatic TB, and these patients need to be included with the subset where the diagnosis is still not reached pre-operatively. However, ‘forewarned is forearmed!’ A determined search for TB even where the features of malignancy seem clearly present may well be worthwhile and prevent morbidity or even fatality following surgical intervention subsequently found to have been avoidable.
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
