- (2008) Volume 9, Issue 2
Shirin E Khorsandi1, Paolo Limongelli1, James E Jackson2, Paul Tait2, Robin C Williamson1, Nagy A Habib1, Long R Jiao1
1HPB Unit, Department of BioSurgery and Surgical Technology, and 2Department of Radiology, Imperial College London, Hammersmith Hospital. London, England
Received December 12th, 2007 - Accepted January 29th, 2008
Context Delayed arterial hemorrhage is a rare complication after pancreaticoduodenectomy (frequency 2-4%) but carries a high mortality, with up to a third of patients dying as a consequence. Its ideal management remains unclear. Case series Between 1993 and 2007, 317 head of pancreas resections were performed at our institution; there were 5 cases of delayed arterial hemorrhage (frequency 1.6%). Conclusion This paper presents our experience in the management of delayed arterial haemorrhage post pancreaticoduodenecomy and discusses the role of angiograms, endoscopy and laparotomy in its management.
Pancreaticoduodenectomy; Radiology, Interventional
PD: pancreaticoduodenectomy; PPPD: pylorus preserving proximal pancreaticoduodenectomy
Over the past two decades, mortality and morbidity after pancreaticoduodenectomy (PD) have improved significantly from a shift of care to higher volume centers which has allowed for improved surgical experience and technique, better surgical intensive care and critical care, better antibiotic and infectious disease services, better and faster crosssectional imaging with interventional radiology establishing a crucial role in the management of the complications following PD [1, 2, 3, 4].
Delayed arterial hemorrhage is an uncommon complication (incidence 2-4%) but carries a high mortality, with up to a third of patients dying as a consequence [5]. Because delayed arterial hemorrhage post PD is rare, its management remains unclear. Interventional radiology has become more sophisticated over the past two decades, consequently delayed arterial hemorrhage is increasingly managed by this approach, rather than surgery. The present paper discusses our experience in the management of delayed arterial hemorrhage post PD and the role of angiograms, endoscopy and laparotomy in its management.
Between 1993 and 2007, 317 head of pancreas resections were performed at our institution. For this analysis, delayed arterial hemorrhage was defined as bleeding greater than 48 hours after PD, presenting either as revealed blood loss: via the abdominal drain or the gastrointestinal tract as hematemesis or melena, or as concealed blood loss: in the form of cardiovascular collapse or a post operative drop in hemoglobin necessitating blood transfusion. On retrospective analysis, five cases of delayed arterial hemorrhage post PD (frequency 1.6%) were identified.
The principles of the Declaration of Helsinki were adhered to throughout this review.
Case 1
A 70-year-old male underwent a pylorus preserving proximal pancreaticoduodenectomy (PPPD) for a distal common bile duct cholangiocarcinoma. On the 14th postoperative day, the patient collapsed after an episode of melena, followed by two bouts of hematemesis. In view of the patient being hemodynamically unstable, the decision was made to proceed straight to laparotomy. Operative findings were disruption of the pancreaticojejunal anastomosis, with fresh blood in the proximal jejunum. No active bleeding point was able to be identified. Technically, it was not possible to reconstruct a new pancreaticojejunostomy because the pancreatic stump was grossly oedematous and pancreatic duct non-dilated. Completion pancreatectomy, with spleen preservation and over sewing of the jejunum was performed. The patient continued to bleed on the intensive care unit and 24 hours later, the patient returned to the operating room, where a distal gastrectomy, proximal jejunal resection and refashioning of bilioenteric and gastrojejunal anastomoses was done. The patient became septic and then died of multiorgan failure twenty days post-operatively.
Case 2
A 56-year-old man underwent a PPPD and right nephrectomy for biliary papillomatosis and renal cell carcinoma. On post-operative day 21, he had an episode of hematemesis but remained hemodynamically stable. Upper endoscopy demonstrated an ulcer at the duodenojejunal anastomosis which was not actively bleeding, so no therapeutic procedure was performed. However, a subsequent visceral angiogram did demonstrate active bleeding from the distal branches of the third jejunal artery, at the site of the duodenojejunal anastomosis. The vessel was selectively catheterized with a coaxial 3 French catheter (FasTracker-18MX, Boston Scientific, Natick, MA, USA) and embolization performed with a small amount of polyvinyl alcohol (Contour, Boston Scientific, Natick, MA, USA), initially with a particle size of 150-250 μm and then 355-500 μm (Figure 1ab). Complete hemostasis was achieved and the patient was discharged one week later.
Case 3
A 57-year-old male underwent a PPPD for pancreatic adenocarcinoma, arising from an intraductal papillary mucinous tumour. He was discharged on post-operative day 19. Nineteen days after discharge the patient collapsed with hematemesis and was admitted to his local hospital. Upper endoscopy and CT were both unremarkable. He received a blood transfusion and was about to be discharged after an inpatient stay of seven days, when he had a further episode of hematemesis. Repeat upper endoscopy demonstrated a large clot at the duodenojejunal anastomosis, no therapeutic procedure was undertaken and the patient was transferred back to our institution. On arrival a visceral angiogram was performed which demonstrated a pseudoaneurysm of the splenic artery. The splenic artery was catheterized, with a 7 French sidewinder catheter (Cordis Corporation, Miami, FL, USA), to a point beyond the pseudoaneurysm and MR compatible platinum coils (William Cook Europe, Bjaeverskov, Denmark) were placed on either side of the pseudoaneurysm neck (Figure 2ab). The patient remained well and was discharged seven days later.
A 63-year-old female had a PPPD for pancreatic adenocarcinoma. On the 12th day after surgery she collapsed with a massive hematemesis. Upper endoscopy in the operating room found a stomach full of blood but the bleeding point was unable to be identified, a subsequent laparotomy also failed to localise a bleeding point. A day later on the intensive care unit, the patient had further bleeding, manifested by fresh blood in the nasogastric tube and per rectum. Visceral angiogram was performed but again no active bleeding point could be localised. The patient had no further episodes of bleeding but became progressively septic and died of multi-organ failure 25 days after PPPD.
Case 5
A 55-year-old male had a Whipple procedure for pancreatic adenocarcinoma. On the 4th post-operative day, the abdominal drain was found to contain 400 mL of blood. He was hemodynamically stable but a drop in hemoglobin necessitated a four unit blood transfusion. Visceral angiogram was performed and a 2.5 cm diameter pseudoaneurysm arising from the transverse branch of the dorsal pancreatic artery was found. The transverse pancreatic artery and an epiploic branch vessel, beyond the pseudoaneurysm were selectively catheterized with a coaxial 3 French (FasTracker-18MX, Boston Scientific, Natick, MA, USA) and embolized with platinum microcoils (Boston Scientific, Natick, MA, USA). Further platinum microcoils were then deployed proximal to the arterial defect (Figure 3ab). After embolization, the patient had a swinging temperature with raised inflammatory markers. CT was performed which demonstrated a small collection in the gastrosplenic ligament which was managed with antibiotics. The patient otherwise had an uneventful course and was discharged on the 18th post-operative day.
Figure 3. a. Angiogram performed with catheter tip in origin of dorsal pancreatic artery (black arrow) demonstrating pseudoaneurysm of transverse pancreatic artery (white arrows). b. Coeliac axis angiogram post embolization demonstrating pseudoaneurysm to be excluded from the circulation by multiple platinum coils (white arrows).
On retrospective review, of 317 head of pancreas resections between 1993 and 2007, five cases of delayed arterial hemorrhage were identified. Details of patient characteristics, clinical presentation, management and outcome are summarised in Table 1. Delayed arterial hemorrhage post PD occurs in 2-4% of cases and our experience is comparable with a frequency of 1.6%. Based on the timing of bleed, hemorrhagic complications after PD maybe divided into early or delayed. Early hemorrhage occurs within the first 48 hours after surgery and is generally caused by technical failure which normally requires immediate laparotomy for resolution [6], whereas delayed arterial hemorrhage is any bleeding after this time period. The timing of delayed arterial hemorrhage presentation is variable. In our series, delayed arterial hemorrhage on average occurred on the 18th post operative day (range: 4-38 days), with one case presenting after discharge (Case 3), highlighting the need for awareness of the clinical presentations of delayed arterial hemorrhage in the post operative PD patient, in order to facilitate prompt and appropriate transfer, to a specialist unit where all facilities, interventional radiology and surgical, are available as required.
Multiple factors have been suggested to contribute or predispose to the development of delayed arterial hemorrhage post PD, such as in the regional lymphadenectomy and skeletonization of the vasculature leaving the vessels vulnerable to damage from local sepsis arising from pancreatic fistula, anastomotic leakage or abscess formation [7]. In the published delayed arterial hemorrhage post PD case series, approximately two thirds of reported delayed arterial hemorrhage cases are found to have an underlying collection or anastomotic leak, with pseudoaneurysm formation identified in approximately a third of cases. In this series, local complications potentially contributed to the development of delayed arterial hemorrhage in 4 patients: one (Case 1) from a disruption of the pancreaticojejunal anastomosis; another one (Case 2) from a disruption of duodenojejunal anastomosis secondary to ulceration through tissue ischemia; the third (Case 5) from intraabdominal sepsis and the fourth (Case 3) remained unclear but most likely related to an extensive surgical dissection weakening the advential wall of the artery. Emphasising that when present, contributing local complications of pancreatic resection should be identified and managed appropriately, in addition to targeted delayed arterial hemorrhage therapy.
Delayed arterial hemorrhage post PD has been reported to occur into the gastrointestinal tract or more commonly, the abdominal cavity [8]. Though in our clinical experience, the most common presentation of delayed arterial hemorrhage was as an upper gastrointestinal bleed (Table 1). Endoscopy, which is the standard first line of investigation for upper gastrointestinal bleeding, may fail to be therapeutic in the post PD scenario through difficulties in localizing the site of hemorrhage. Our experience illustrates this point and other reports in the literature have given accounts of “positive” endoscopic findings being misleading and as a consequence delay definitive intervention or lead to death in the worst case [9, 10, 11]. Therefore, endoscopy has no role in the initial management of delayed arterial hemorrhage post PD.
Historically, laparotomy has been the first choice of treatment for delayed arterial hemorrhage post PD [9, 12]. Laparotomy allows hemostasis by ligation or under running of vessels/pseudoaneurysms to be combined with treatment of contributing factors to delayed arterial hemorrhage such as anastomotic disruption or intra-abdominal collections. From the published cases, laparotomy and interventional radiology appear to be equally effective in arresting hemorrhage. Though in our institute’s experience, the two cases where laparotomy was performed (Cases 1 and 4), the site of bleeding was not localised, with one patient continuing to bleed post operatively and the other rebleeding in the subsequent 24 hours. Whereas, in all the cases where interventional radiology was used as the first line management of delayed arterial hemorrhage post PD, definitive hemostasis was achieved in all cases.
A higher morbidity and mortality is reported in the literature for patients managed by laparotomy compared to interventional radiology. In our experience, admission to intensive care was required for the two patients (Cases 1 and 4) initially managed by laparotomy and both of them eventually died (Table 1). In contrast, when delayed arterial hemorrhage post PD was treated initially by interventional radiology, no intensive care admission was required and the patients were discharged, on average 9 days later. The higher morbidity and mortality of laparotomy is not unexpected on two counts. Firstly, patients selected for immediate laparotomy often have a greater degree of hemodynamic compromise, compared with patients where interventional radiology has been used as first line management. Secondly, an operative approach is an invasive strategy, in patients potentially debilitated by complications relating to the original PD, such as sepsis or anastomotic breakdown which may have predisposed to the development of delayed arterial hemorrhage initially.
Selective visceral angiography of the celiac and superior mesenteric arteries is probably the most valuable initial examination to identify the site of bleed post PD when clinical parameters allow for a more measured evaluation [13]. However, angiography also has its limitations like surgery and endoscopy, when the bleed is venous, diffuse or intermittent [14], which may then lead to interventional radiology diagnostic failure which happened in one patient (Case 4). Trans-catheter arterial embolization is emerging, as a safe and effective option for the first line treatment of delayed arterial hemorrhage post PD. In addition, our institute’s experience of interventional radiology also found it to be efficacious, with minimal morbidity and no mortality, though recognised drawbacks of interventional radiology can be the accidental or intentional occlusion of the hepatic artery, which can lead to cholangitis, hepatic abscess or fatal liver failure. The recent utilization of covered stents may prove to be a solution to this issue, establishing further the role of interventional radiology in the management of delayed arterial hemorrhage post PD [15, 16].
Deciding on the optimal management of this life threatening complication is difficult. Currently, there is no high quality evidence defining the standard management of delayed arterial hemorrhage post PD. As interventional radiology becomes more sophisticated, it will increasingly be used as the first line of diagnosis and therapy in this complex group of patients. Though ultimately, the final decision of which delayed arterial hemorrhage management pathway is selected will be dictated by the clinical status of the patient and institution preference. A meta-analysis or a multi-institutional analysis would provide more data to provide more information on this difficult problem, as a prospective study is impractical and no single institution will acquire enough patients to reach any conclusion.
The authors have no potential conflicts of interest