Patients
We retrospectively reviewed the databank at National Taiwan University Hospital for 2003–2018. This study was approved by the institutional review board of the hospital. We searched for electronic medical records of patients who received the Whipple procedure or PPPD. Patients who had delayed PPH and underwent EVT were included. Patients were excluded if (1) the culprit vessel was not branched from the common hepatic artery or (2) the clinical or image data were missing or were insufficient for analysis. In total, 19 patients had delayed PPH and underwent EVT. Because one patient with a splenic artery pseudoaneurysm after the Whipple procedure was excluded, 18 patients were included in the final analysis.
Clinical management and data assessment
Hemorrhage was detected by the presence of either sentinel bleeding, defined as blood in the abdominal drain, or hematemesis and melena. All patients presented with delayed PPH, defined as hemorrhage occurring more than 24 h postoperatively. Most patients underwent multidetector computed tomography (MDCT) angiography for culprit lesion detection.
Clinical data, including age, sex, pathologic diagnosis, coagulation profile, clinical presentation, and onset time of bleeding after surgery, were obtained from the available medical records. On the basis of the International Study Group for Pancreatic Fistula (Bassi et al. 2017), we defined pancreatic leakage as an amylase concentration greater than three times the upper limit of normal serum amylase concentration in the drain tube after postoperative day 3. Moreover, we defined coagulopathy as a serum platelet count of less than 50,000 × 106/L or an international normalized ratio of > 1.5 (Hasegawa et al. 2017).
Endovascular procedures
After a vascular sheath was introduced through either the right or left common femoral artery, a 4- or 5-Fr angiographic catheter was navigated at the celiac artery for angiography, and angiographic findings regarding the culprit lesion were identified as either contrast spillage or a pseudoaneurysm. Three EVT strategies, namely destructive, superselective, and constructive approaches, were applied on the basis of the MDCT findings, available embolization materials, and duty doctors’ experience.
Superselective approach
This approach, was frequently employed at the hospital between 2005 and 2008, involved the application of embolization to the culprit lesion (Fig. 1). Hepatic artery patency was preserved intentionally. A microcatheter was typically navigated into the lesion, which was either a pseudoaneurysm or active bleeding site, and pushable coils (Cook, Bloomington, IN, USA) or a 40%–50% N-butylcyanoacrylate (NBCA)–lipiodol mixture were used.
Destructive approach
This approach has been employed at the hospital since 2003. The destructive approach involved hepatic artery sacrifice (Fig. 2). The catheter was navigated to the distal part of the culprit lesion, and pushable coils were deployed from the distal to the proximal part of the lesion (sandwich technique). When navigation to the distal portion was difficult, a 40%–50% NBCA–lipiodol mixture was used to occlude the proper hepatic artery or common hepatic artery from the proximal portion.
Constructive approach
Covered stent placement has been performed at the hospital since 2016. A 45-cm, 6–8-Fr-long vascular sheath (Cook Medical, Bloomington, IN, USA) was placed at the celiac origin. After the distal intrahepatic artery was wired, a self-expandable polytetrafluoroethylene-covered stent (Viabahn; W. L. Gore and Associates, Flagstaff, AZ, USA) was navigated onto the lesion and deployed across the culprit lesion (Fig. 3). Poststenting angioplasty was not conducted routinely. Antiplatelet medications were not administered considering the current bleeding condition.
Outcome assessment
Technical success was confirmed after a review of angiography after EVT. We defined technical success as the cessation of active contrast extravasation or nonopacification of a pseudoaneurysm. Hemorrhage recurrence rates, 30-day and 1-year mortality rates, and major and minor hepatic complication rates were recorded. Recurrent hemorrhage was defined as any evidence of bleeding, such as increased bloody drain or hematemesis, unstable vital signs necessitating the consultation of interventional radiologists, and evidence of bleeding in follow-up MDCT or angiography.
We defined major and minor hepatic complications according to the standards of the Society of Interventional Radiology (Sacks et al. 2003) and Hasegawa et al.’s (2017) study. Major hepatic complications included hepatic failure and hepatic abscess. Hepatic failure was defined as an increase in aspartate and/or alanine aminotransferase levels by 1000 U/L or an increase in the total bilirubin level by 10 mg/dL within 7 days after EVT. Hepatic abscess was defined as the combination of liver abscess observed in contrast-enhanced computed tomography (CT) and clinical signs and symptoms of infection. Minor hepatic complications involved abnormal hepatic function, defined as a temporary increase in the level of aspartate and/or alanine aminotransferase three times above the normal upper limit of the serum level within 14 days after EVT.
Statistical analysis
Because of the small sample size, inferential statistical analysis could not be performed. Descriptive statistics related to the clinical data were obtained. Technical success, hemorrhage recurrence rates, 30-day and 1-year mortality rates, hepatic failure and abscess rates, and abnormal liver function rates of the three approaches are provided.