Intraductal papillary-mucinous neoplasms (IPMNs) of the pancreas have unique clinico-pathological characteristics and show a wide spectrum of histological types, ranging from adenomatous hyperplasia to invasive cancer. While non-invasive IPMNs show slow growth and good prognosis after resection, the outcome can become poor if they transform into invasive ductal cancers. There are three types of IPMN: main duct,, branch, and combined. The main duct and combined types are associated with higher rates of invasive cancer than the branch type 1.

A feature of IPMNs is occasional fistula formation with surrounding organs such as the duodenum, bile ducts, stomach and even the peritoneal or pleural cavity 2345678910. While such fistula formation is generally thought to be associated with invasive IPMNs, some non-invasive IPMNs also form fistulas. Thus, IPMN fistula formation raises intriguing issues concerning pathogenesis, diagnosis and patient prognosis.

We report here a rare non-invasive "borderline-malignant" IPMN that formed a pancreatogastric fistula. Fistula formation was triggered by acute pancreatitis following endoscopic retrograde cholangiopancreatography (ERCP).

A 72 year-old Japanese woman underwent abdominal ultrasonography (US) as part of a routine physical checkup in April 2001. This procedure incidentally detected marked dilatation of the main pancreatic duct (MPD) in the distal region of the pancreas. Magnetic resonance cholangiopancreatography (MRCP) and abdominal computed tomography (CT) showed MPD dilatation and a multi-lobular cystic lesion. ERCP confirmed these findings, and no extraluminal leakage of the contrast medium was observed (Figure 1). Cytology on pancreatic fluid indicated non-neoplastic epithelia. Endoscopic ultrasonography (EUS) identified 8 mm papillary projections in the cystic lesion, which was a key indication for surgery. A routine gastroscopy performed simultaneously with EUS showed no significant findings.

Following the ERCP, the patient complained of abdominal pain and showed hyperamylasemia and leukocytosis. Distinct swelling of the pancreas and thickening of the gastric wall were subsequently observed by abdominal CT (Figure 2), which indicated a mild post-ERCP pancreatitis. Gastroscopy performed for the assessment of gastric wall thickness revealed a small ulceration in the body of the stomach (Figure 3), which was accompanied by a mucinous discharge suggesting the presence of a fistula opening into the MPD. These findings, along with abdominal angiography data indicating encasement of the splenic artery suggested a malignant invasion into the stomach by the primary lesion. All other laboratory data, including tumor marker expression (e.g., CEA and CA19-9), were within normal limits.

A laparotomy was performed 1 month after the ERCP. Firm fibrous tissue between the pancreas and the mesenterium of the transverse colon suggested the pancreatic neoplasm was of an invasive nature. An en bloc resection of the distal pancreas, spleen, stomach and part of the transverse colon was performed. Postoperative histopathology showed the distal pancreas had an IPMN whose lining epithelia showed only "borderline-malignancy" and no apparent invasion of adjacent tissues or organs. Light microscopy examination revealed inflammatory cellular infiltration consisting mainly of neutrophils around a pancreatogastric fistula which had no neoplastic epithelia (Figure 4). The patient was discharged after an uneventful postoperative recovery and showed no signs of recurrence 3 years after surgery.

Fistula formation associated with an IPMN was first reported by Ohhashi et al in 1980 as a pancreatobiliary fistula 2. Since then there have been 41 cases reported between 1980 and 2004 (Table 1) 3456789. According to these reports, the organs most frequently affected by fistula formation were the duodenum (24 cases, 59%), common bile duct (21 cases, 51%) and stomach (7 cases, 17%). Apparent cancer invasion was the main cause of the fistulae in 16 cases (37%), while similar to the present case, 17 cases (41%) showed fistulae into neighboring organs in the absence of tumor invasion (Table 2). Five of the six previous IPMN cases associated with pancreatogastric fistulae were linked to apparent cancer invasion.

To our knowledge, the present report is only the second describing a case of IPMN associated with pancreatogastric fistulae without invasive cancer. Yago et al reported a rare case of non-invasive intraductal papillary adenocarcinoma of the pancreas that was associated with development of a pancreatogastric fistula in the absence of cancer invasion 5. In that case, neoplastic epithelia were observed only on the surface of the fistula and gastric mucosa, while the structure of the gastric wall beside the fistula was maintained without invasive cancer. Those authors speculated that the pancreatogastric fistula developed due to high pressure in the MPD. Baek et al and Watanabe et al reported the same phenomenon in mucinous cystic tumors (MCTs) 1112. The mechanism of such fistula formation without cancer spread could be explained by a combination of high pressure in the MPD and inflammatory stimulation 3413.

In the present case, ERCP played a key role in fistula formation. To our knowledge, this is the first report to show an absence of a pancreatogastric fistula immediately prior to ERCP, and then the presence of a fistula following post-ERCP pancreatitis. Post-ERCP pancreatitis, which is reported to occur in approximately 1.8–10% of patients, should be carefully investigated in IPMN cases 1415.

We emphasize here that development of an IPMN-associated fistula does not necessarily represent underlying invasive cancer. Fistula formation can result in an incorrect preoperative diagnosis of invasive cancer and lead to the undertaking of unnecessary surgical procedures. Although the lesion in the present case was a non-invasive tumor, surgery was appropriate given the indications of invasive cancer. While non-invasive IPMNs have a good prognosis, there remains the possibility of dissemination by tumor penetration. When treating IPMNs associated with fistula, the extent of resection should depend on the extent of cancer invasion. To avoid dissemination, the fistula should be removed independent to cancer invasion. The extent of resection of affected organs should be individually determined based on the results of preoperative images and precise intraoperative histological examinations.

The present case highlights that a pancreatogastric fistula can develop in the absence of invasive cancer. Further data regarding IPMN-associated fistulae are necessary in order to clarify the pathogenesis, diagnosis, appropriate surgical intervention and prognosis for this disorder.

The author(s) declare that they have no competing interests.

MK, NS, KY, MT, KK, YY and HN were involved in performing surgery, undertook the literature search and drafted the manuscript for submission.

HN supervised preparation of the manuscript and edited the final version for publication. All authors read and approved the manuscript.