A 57-year-old man was referred to the Emergency Room of our medical center for an episode of melena and multiple episodes of rectorrhagia during the last 3 days associated with intermittent abdominal pain. The patient had a history of chronic lithiasic pancreatitis, arterial hypertension, psoriasis, and a localized renal cell carcinoma treated with nephrectomy in 2017. Due to his oncologic surveillance program, he had recently performed an abdominal RM which had shown a fluid collection upon the body of pancreas compatible with a pancreatic pseudocyst. Actually, this patient was already known to our hospital, as 1 year before he had performed repeated accesses to our emergency department for repeated episodes of rectorrhagia, melena, and anemization and was hospitalized multiple times. First, in the april of 2019, he underwent an esophagogastroduodenoscopy (EGDS), a pancolonoscopy with retrograde ileoscopy, a videocapsule endoscopy, and a push-and-pull enteroscopy, which resulted negative for every gastrointestinal source of bleeding, and was therefore discharged; after few months, he was hospitalized again two more times for the same symptomatology, so he underwent again several angiography-CT scans, repeated endoscopic exams and even a scintigraphy with marked red blood cells, but no active bleeding sources or cause of gastrointestinal were found. Thus, he was discharged with a diagnosis of occult hemorrhage and prescribed long-acting octreotide. Then the patient was lost at follow-up due to the COVID-19 pandemic. He had self-suspended the administration of the drug after 6 months from last discharge.

At admission, he presented tachycardia (HR 113 bpm) but no hypotension or fever. At physical examination, the abdomen was not tender and there was no abdominal pain; digital rectal examination showed bright red blood. Laboratory tests showed hemoglobin (HB) 5.8 g/dL, white blood cell count (WBC) 8.19 × 10^9/L, platelets (PLT) count 401 × 10^9/L. Alanine transaminase was 45/45 UI while aspartate transaminase was 54/45 UI/L. Serum amylase were elevated (370 IU/L, range < 107 IU/L), blood urea nitrogen (BUN) was 23 mg/dl (range 10–23 mg/dl) while creatinine was 1.17 mg/dl (range 0.67–1.17 mg/dl) and C reactive protein (CRP) was 0.9 mg/L (range < 5 mg/L). An urgent EGDS was performed, which only reported an erosive duodenopathy without signs of recent bleeding; an abdominal angio-CT also did not show any active bleeding signs and confirmed the presence of the known pancreatic pseudocyst. However, after these exams and a hemotransfusion, an episode of hematemesis followed by abdominal pain occurred in the emergency room. The patient was immediately intubated and underwent another urgent EGDS, but even this time the exam did not show any bleeding cause, reporting a normal gastric mucosa with no signs of recent bleeding; duodenum was explored until the second portion and, again, no abnormalities were found except the known erosive duodenopathy. The patient was, therefore, hospitalized in our unit and performed a colonoscopy, which showed just some diverticula without bleeding or inflammatory signs. After this exam, the patient presented another self-limited episode of rectorrhagia with mild anemization and abdominal pain. Laboratory exams showed persistence of mild increase of amylase serum levels. In the suspicion of a pancreatic origin of the bleeding, even if a revision of the previous abdominal CT examination had confirmed the absence of any active bleeding sign, we planned to perform another urgent angiography-CT scan as soon as another episode of rectorrhagia would occur. When the following day another bleeding episode happened, the patient was immediately transferred while infusing supporting therapy with lactate ringer to perform the exam, which is presented in Fig. 1. The angio-CT showed a pseudoaneurysm of the left gastric artery (not documented at previous exams) with mild signs of contrast extravasation suggestive of a slight bleeding through the known pancreatic pseudocyst. A diagnosis of hemosuccus pancreaticus was, therefore, made (Fig. 1).

Fig. 1
figure 1

The abdominal angiography-CT scan (panels A, B) shows a pseudoaneurysm of the left gastric artery with blood extravasation into the pancreatic pseudocyst (yellow arrow). The diagnosis of hemosuccus pancreaticus was also confirmed byMR cholangio-pancreatography (MRCP, panels C and D), which confirmed the abnormal communication among the pseudocyst and the left gastric artery (see text below)

Full size image

After the diagnosis, two attempts of percutaneous sclerotization of the left gastric artery aneurysm were performed, but the procedures failed due to the impossibility of obtaining a stable catheterization of the artery. To plan the best therapeutic approach, an MR cholangio-pancreatography was performed which confirmed the persistence of the blood extravasation among the left gastric artery and the pseudocyst (Fig. 1).

Hemosuccus pancreaticus (HP) is considered the least frequent cause of upper gastrointestinal bleeding [1] and it is defined as a gastrointestinal hemorrhage via pancreatic duct from the papilla of Vater. HP is caused by an abnormal communication among pancreatic ducts and one or more branches of splanchnic arteries, usually provoked by chronic pancreatitis, pancreatic pseudocysts, vascular abnormalities or malignancies [1, 2]. HP can clinically present as a variety of conditions between chronic anemia and repeated acute hemorrhagic events, even with life-threatening hemorrhagic shock. The most frequent clinical presentation is composed of intermittent abdominal pain, upper gastrointestinal hemorrhage, and increase of amylase serum levels [1, 2]. HP can pass undetected through diagnostic endoscopic and radiologic exams usually performed for gastrointestinal bleedings, especially if not performed nearby the clinical manifestation; this aspect makes HP represent a very insidious diagnostic challenge. Most of the available data in literature derives from case reports or small retrospective studies [3]. The main therapeutic approaches are the endovascular embolization of the bleeding vessels and the surgery, even if an endoscopic-ultrasound-guided approach has also been proposed [1]. However, the management of these patients can be even more challenging than the diagnosis.

Indeed, after the failure of the endovascular approach, our patient was transferred to a surgical unit where he underwent an exploratory laparotomy with ligature of the gastroduodenal artery and of the left gastric artery, with a contemporary drainage and sclerosis of the pancreatic pseudocyst. The surgical intervention was complicated by an external pancreatic fistula and, furthermore, after some days, he presented again new episodes of rectorrhagia and hematemesis. Two new angiographies showed persistence of the bleeding through small collateral branches of the gastroduodenal artery stump, which this time were successfully embolized percutaneously. After these procedures, the hemorrhagic symptomatology finally stopped, and after a length of stay of 64 days, the patient was discharged home in stable conditions, with an abdominal percutaneous drainage catheter, and sent to surgical follow-up.

In conclusion, hemosuccus pancreaticus represents an extremely rare cause of gastrointestinal bleeding which should be suspected in patients with repeated occult gastrointestinal bleeding episodes and a known pancreatic disease or risk factors for pancreatic alterations. The management of these patients can be very challenging and often requires a multidisciplinary team with medical, radiological, and surgical expertise.