Embolization for delayed arterial bleeding after percutaneous self-expandable metallic stent placement in patients with malignant biliary obstruction

Abstract

Objectives:

To retrospectively evaluate the safety and efficacy of transcatheter arterial embolization (TAE) for delayed arterial bleeding secondary to percutaneous self-expandable metallic stent (SEMS) placement in patients with malignant biliary obstruction (MBO).

Methods:

From January 1997 to September 2017, 1858 patients underwent percutaneous SEMS placement for MBO at a single tertiary referral center. Among them, 19 patients (mean age, 70.2 [range, 52–82] years; 13 men) presented with delayed SEMS-associated arterial bleeding and underwent TAE.

Results:

The incidence of delayed arterial bleeding was 1.0% (19/1858) after SEMS placement, with a median time interval of 225 days (range, 22–2296). Digital subtraction angiography (DSA) showed pseudoaneurysm alone close to the stent mesh (n = 10), pseudoaneurysm close to the stent mesh with contrast extravasation to the duodenum (n = 3), pseudoaneurysm close to the stent mesh with arteriobiliary fistula (n = 1), in-stent pseudoaneurysm alone (n = 4) and in-stent pseudoaneurysm with arteriobiliary fistula (n = 1). Bleeding was stopped after the embolization in all patients. Overall clinical success rate was 94.7% (18/19). One patient with recurrent bleeding was successfully treated with a second embolization. Overall 30-day mortality rate was 26.3% (5/19). A major procedure-related complication was acute hepatic failure in one hilar bile duct cancer patient (5.3%), which was associated with an obliterated portal vein.

Conclusion:

TAE is safe and effective for the treatment of delayed arterial bleeding after percutaneous SEMS placement for MBO.

Advances in knowledge:

This study demonstrated TAE is safe and effective for arterial bleeding after SEMS placement after MBO through the largest case series so far.

Introduction

Self-expandable metallic stents (SEMSs) play a crucial role for malignant biliary obstructions (MBOs) because of their long patency.^1–11 However, several major complications have been reported after percutaneous biliary SEMS placement, including stent dysfunction, cholangitis and stent migration,^{1–3,8,9,12,13} while serious bleeding complications are considerably less frequent, with reported incidence rates of 0.3–3.6%.^8,12–14 The majority of the bleeding is transient hemobilia from hepatic parenchyma.¹⁵ However, several authors reported delayed SEMS-associated bleeding complications requiring further treatment.^16–23 The suggested mechanisms of delayed arterial bleeding in these reports were adhesion between the stent and tumor, duodenal ulceration caused by the distal end of the stent, and mechanical irritation of adjacent arterial wall by the stent mesh.^16–21 Recently, Hyun et al demonstrated the association between the bleeding and circumferential tumor encasement.²² Nevertheless, there remains a paucity of reports describing delayed SEMS-associated arterial bleeding in detail, the subsequent endovascular results, and the survival prognosis of the patients, with the reports being limited to several case reports and small case series.

Therefore, we aimed to evaluate the safety and efficacy of transcatheter arterial embolization (TAE) for the management of delayed arterial bleeding after percutaneous biliary SEMS placement in patients with MBO.

Methods

This single-center retrospective case series study was approved by our institutional review board, and the requirement for informed consent was waived. This study was written in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.²⁴

Patient population

A single-center retrospective review of 1858 patients who underwent percutaneous SEMS placement for MBO between January 1997 and September 2017 was performed. After thorough review of electronic medical records, 1839 of the 1858 patients were excluded for the following reasons: patients did not have a bleeding complication after SEMS placement (n = 1817); bleeding was far from the SEMS, defined as no overlap of pseudoaneurysm contour and the inserted stent mesh (n = 14) and iatrogenic bleeding or hemobilia (n = 8) after percutaneous transhepatic biliary drainage (PTBD) or endoscopic retrograde cholangiopancreatography (ERCP). Finally, 19 patients (13 males and 6 females; age range, 52–82 years; mean age, 70.2 years) who underwent TAE for the treatment of delayed SEMS-associated arterial bleeding were included (Figure 1).

Figure 1.

Flow chart in this study.

The causes of MBO were hilar bile duct cancer (n = 7), distal bile duct cancer (n = 4), pancreatic cancer (n = 4), advanced gastric cancer (n = 2), intrahepatic bile duct cancer (n = 1) and gallbladder cancer (n = 1). In all the 19 patients, percutaneous SEMS placement was successful without immediate complications.

The stent deployment techniques, stent types and configurations were decided by the operators according to cholangiographic findings and initial catheter drainage: unilateral stenting with a single stent placement through a single percutaneous route; unilateral stenting with two-stent placement through different percutaneous routes; bilateral stenting with stent-in-stent technique through a single percutaneous route (T-configuration) or through different percutaneous routes (Y-configuration); bilateral stenting with side-by-side technique through bilateral percutaneous routes; and bilateral stenting in a crisscross configuration through bilateral percutaneous routes. All SEMS deployments were performed using commercially available uncovered stents (Zilver, Cook Medical, Bloomington, Indiana; Epic, Boston Scientific, Natick, Massachusetts; or Niti-S stent, TaeWoog Medical, Gimpo, Korea) and/or covered stents (ComVi^™, TaeWoog Medical, Gimpo, Korea; Hercules, S&G Biotech, Seongnam, Korea). The covered stents were partially covered with polytetrafluoroethylene (PTFE), with two or 3 cm bare extensions at the proximal end to prevent tumor overgrowth, stent migration, and intrahepatic duct occlusion.

After stenting, one or two temporary PTBD tubes were inserted just proximal to the stent(s). Those were removed after 2–3 days of clamping after confirmation of patent contrast flow through the stent(s) into the duodenum or jejunum on cholangiography. The detailed demographics and clinical characteristics are summarized in Table 1.

Table 1.

Patient demographics, clinical manifestations, treatment and prognosis

No./Age/Sex	Underlying malignancy	Time to bleeding (days)	Symptoms	Stent location	Stent type	CT	Angiography	Bleeding focus	Embolic material	Technical success	Clinical success	Prognosis	Survival period (days)
1/74/M	Pancreas head cancer	165	Hematemesis	CBD	Uncovered	Not performeda	Pseudoaneurysm close to the mesh +extravasation	GDA	Coils	Yes	Yes	Died of cancer	25
2/68/F	GB cancer	268	Hematochezia	LHD to CJ	Uncovered	Pseudoaneurysm close to the mesh	Pseudoaneurysm close to the mesh	LHA	Coils	Yes	Yes	Died of cancer	4320
3/79/M	Klatskin tumor IV	1168	Hemobilia	B3 to B6, B6 to CBD	Uncovered	In-stent pseudoaneurysm	In-stent pseudoaneurysm	PHA	NBCA	Yes	Yes	Died of cancer	4
4/82/F	Pancreas head cancer	2296	Hematemesis	CBD	Partially covered	Pseudoaneurysm close to the mesh	Pseudoaneurysm close to the mesh	RHA	Coils + NBCA	Yes	Yes	Died of cancer	398
5/56/M	Klatskin tumor IV	155	Hemobilia	RAHD to CBD, RPHD to CBD	Partially covered	Pseudoaneurysm close to the mesh	Pseudoaneurysm close to the mesh	RHA	Coils + NBCA	Yes	Yes	Died of cancer	98
6/78/F	Klatskin tumor II	225	Hematemesis	RMHD to CBD, LMHD to CBD	Uncovered	Pseudoaneurysm close to the mesh	Pseudoaneurysm close to the mesh +extravasation	GDA	NBCA	Yes	Yes	Died of cancer	10
7/67/F	Intrahepatic cholangiocarcinoma	1030	Hematemesis	B6 to CBD	Partially covered	Pseudoaneurysm close to the mesh	Pseudoaneurysm close to the mesh	RHA	NBCA	Yes	Yes	Died of cancer	203
8/64/M	Klatskin tumor IV	480	Hematochezia	B3 to RAHD, B3 to CBD, B6 to CBD	Uncovered	In-stent pseudoaneurysm	In-stent pseudoaneurysm	RASA	Coils	Yes	Yes	Died of cancer	61
9/79/M	Distal CBD cancer	114	Hemobilia	LMHD to HJ, RAD to HJ	Partially covered	Pseudoaneurysm close to the mesh, hemoperitoneum	Pseudoaneurysm close to the mesh	RPSA	Coils + gelfoam	Yes	Yes	Died of cancer	71
10/67/M	Pancreas head cancer	22	Hematemesis	CBD	Partially covered	Not performedb	Pseudoaneurysm close to the mesh	PSPDA	NBCA	Yes	Yes	Died of cancer	38
11/56/M	Klatskin tumor IIIB	29	Hematemesis	B2 to CBD, B3 to CBD	Uncovered	In-stent pseudoaneurysm	In-stent pseudoaneurysm	RHA	NBCA	Yes	Yes	Died of cancer	71
12/73/F	Klatskin tumor IIIA	25	Hematochezia	B3 to HJ	Partially covered	In-stent pseudoaneurysm	In-stent pseudoaneurysm	LHA	Coils + NBCA	Yes	Yes	Died of acute hepatic failure	4
13/74/M	Gastric cancer	306	Hematochezia	LMHD to CBD	Partially covered	In-stent pseudoaneurysm	In-stent pseudoaneurysm +ABF	RHA	Coils + NBCA	Yes	Yes	Died of cancer	61
14/76/M	Pancreas head cancer	284	Hemobilia	CBD	Partially covered	Pseudoaneurysm close to the mesh	Pseudoaneurysm close to the mesh	RASA	Coils + NBCA	Yes	Yes	Died of cancer	79
15/52/M	Distal CBD cancer	116	Hematemesis	RMHD to CBD, LMHD to CBD	Uncovered	Pseudoaneurysm close to the mesh	Pseudoaneurysm close to the mesh +extravasation	GDA	NBCA	Yes	Noc	Died of cancer	21
16/75/M	Gastric cancer	128	Hematochezia	CBD	Uncovered	Pseudoaneurysm close to the mesh	Pseudoaneurysm close to the mesh +ABF	RHA	Coils + NBCA	Yes	Yes	Died of cancer	508
17/76/M	Distal CBD cancer	113	Hemobilia	RPHD to CJ	Uncovered	Pseudoaneurysm close to the mesh	Pseudoaneurysm close to the mesh	RHA	Coils + NBCA	Yes	Yes	Died of cancer	56
18/71/M	Klatskin tumor IIIA	396	Hemobilia	B3 to HJ	Partially covered	Pseudoaneurysm close to the mesh	Pseudoaneurysm close to the mesh	RHA	NBCA	Yes	Yes	Died of cancer	33
19/67/F	Distal CBD cancer	240	Hematochezia	RPHD to B2, RAHD to CJ	Uncovered	Pseudoaneurysm close to the mesh	Pseudoaneurysm close to the mesh	RHA	Coils + NBCA	Yes	Yes	Died of cancer	118

GB, gallbladder; CBD, common bile duct; LMHD, left main hepatic duct; CJ, choledochojejunostomy; HJ, hepaticojejunostomy; RAHD, right anterior hepatic duct; RPHD, right posterior hepatic duct; RMHD, right main hepatic duct; MPV, main portal vein; NA, not available; ABF, arteriobiliary fistula; GDA, gastroduodenal artery; LHA, left hepatic artery; PHA, proper hepatic artery; RPSA, right posterior sectional artery; RHA, right hepatic artery; RASA, right anterior sectional artery; PSPDA, posterosuperior pancreaticoduodenal artery; NBCA, N-butyl cyanoacrylate

^aConventional angiography was performed in urgent setting because the patient showed massive hematemesis with profound decrease of hemoglobin level (10.2 to 6.4 g dl⁻¹), but hematoma obscured the sight of upper endoscopy.

^bConventional angiography was performed without the CT scan, because bleeding from the stent was highly suspicious as the hematoma was detected at the distal end of the biliary stent.

^cRebleeding (3 days after the first embolization) was successfully treated with the second embolization using coils and NBCA.

Diagnosis of delayed SEMS-associated arterial bleeding

All 19 patients presented with hemodynamically unstable bleeding (i.e., hematemesis, hematochezia, or hemobilia) after a variable time interval from SEMS placement. CT was the initial diagnostic modality for delayed SEMS-associated arterial bleeding in 17 of the 19 patients. Two patients underwent emergency conventional angiography without CT because of rapid deterioration of their clinical status. Diagnosis of delayed SEMS-associated arterial bleeding was defined as the presence of active bleeding (i.e., pseudoaneurysm, extravasation of contrast media) within or close to the SEMS (overlap of the stent mesh with the pseudoaneurysm contour or origin of extravasation), on DSA and/or contrast-enhanced CT at least 2 weeks after the SEMS placement.²⁰

TAE procedures

Following the puncture of the common femoral artery, a 5 F sheath and a 5 F angiographic catheter (RH, Cook Medical, Bloomington, Indiana, USA) were introduced over a 0.035-inch guidewire. Angiograms of the superior mesenteric artery and common hepatic artery were obtained to identify the bleeding focus. Additionally, selective angiography was subsequently performed using a 2.0 to 2.4-F microcatheter (Progreat, Terumo, Tokyo, Japan) via a 5 F angiographic catheter, to reveal the bleeding site according to the location depicted on CT scan and endoscopy, if available. If a bleeding focus was identified, TAE was performed in all patients, with various embolic materials being used.

The embolic materials were chosen according to the operator’s preference on the basis of the angiographic findings, including platinum coils (MicroNester, Cook Medical, Bloomington, Indiana, USA; or Interlock-18, Boston Scientific, Natick, Massachusetts, USA), N-butyl cyanoacrylate (NBCA) (Histoacryl, B. Braun Melsungen AG, Melsungen, Germany) and gelatin sponge particles (SPONGOSTAN, Ethicon, Somerville, New Jersey, USA). NBCA was mixed with iodized oil (Lipiodol, Guerbet, Roissy, France) at ratios varying from 1:2 to 1:4, depending on the distance between the tip of the microcatheter and the target lesion, or the length of the bleeding segment. Completion angiography was performed to confirm target vessel occlusion or cessation of contrast extravasation.

Follow-up

All patients were monitored in the intensive care unit after TAE. Clinical examinations and laboratory findings (i.e., complete blood cell count, liver function tests) were followed up every 4–8 h to detect any complications or recurrent bleeding. If there were any clinical signs or symptoms indicating recurrent bleeding, CE-CT with or without subsequent angiography were performed. CE-CT was routinely obtained 3–5 days after embolization. After discharge, the patients were followed with physical examination, laboratory blood tests and CE abdominal CT at 2–3 month intervals.

Study endpoints and definitions

The main study endpoints were technical success, clinical success, major complications, and overall 30-day mortality rates. Technical success was defined as elimination of observable active bleeding and pseudoaneurysm on the angiogram obtained immediately after TAE. Clinical success was defined as successful cessation of bleeding without a need for additional management (including repeated DSA or surgery) during the same admission. Recurrent bleeding was considered as another episode of bleeding at the same site at least 24 h after the initial TAE requiring additional treatment. Major complications were defined according to the Society of Interventional Radiology clinical practice guidelines.²⁵ Hepatic failure was defined as jaundice (serum bilirubin ≥5 mg dl⁻¹) and coagulopathy (international normalized ratio [INR]>1.5 or prothrombin activity <40%) complicated by ascites and/or encephalopathy.²⁶ Overall survival was measured from the time from the first embolization procedure to death from any cause.

Statistical analysis

Cumulative survival curves after TAE were estimated using Kaplan-Meier analyses, and statistical analyses were performed using MedCalc version 18.11 (MedCalc Software).

Results

Delayed SEMS-associated arterial bleeding

The incidence of delayed SEMS-associated arterial bleeding was 1.0% (19/1858). The median time interval from SEMS placement to the onset of bleeding was 225 days (range, 22–2296 days). Six patients showed hemobilia detected on PTBD tube (n = 6), and the other patients showed hematemesis (n = 7) or hematochezia (n = 6) as clinical manifestations of the delayed SEMS-associated arterial bleeding. Endoscopy was performed in 11 patients who showed hematemesis (n = 6) or hematochezia (n = 5). However, endoscopic bleeding control was not performed in any of the patients because active bleeding was not evident (n = 5), or approach to the bleeding focus was impossible due to massive arterial bleeding within the duodenum (n = 2), or directly from the distal end of the stent (n = 4). In one patient who showed hematemesis (patient 1), endoscopy was impossible because of huge amount of hematemesis and poor general condition. Another one patient with hematochezia (patient 16) only underwent colonoscopy. Pre-procedural CT scans were available in 17 of the 19 patients. In 17 of the 19 patients, there were symptoms or signs (e.g., fever, high level of C-reactive protein, or increased serum bilirubin and liver enzymes) suggestive of cholangitis with biliary ductal dilatation on CT due to stent dysfunction. In 11 of the 17 patients with cholangitis, hepatic abscesses were observed on CT scans performed prior to the onset of bleeding.

CT images demonstrated pseudoaneurysm close to the stent mesh (n = 12) and in-stent pseudoaneurysm (n = 5) (Figures 2 and 3). Ancillary CT findings showed dilated bile ducts filled with high-attenuation soft tissue materials representing hemobilia in four patients, and hemoperitoneum in one patient. DSA was performed without CT scan in two patients (patient 1 and 10). In patient 1, DSA was performed in urgent setting because the patient showed massive hematemesis with profound decrease of hemoglobin level (10.2 to 6.4 g dl⁻¹). In patient 10, DSA was performed without the CT scan because bleeding from the stent was highly suspicious as the hematoma was detected at the distal end of the biliary stent on endoscopy. DSA of the celiac axis showed pseudoaneurysm alone close to the stent mesh (n = 10), pseudoaneurysm close to the stent mesh with contrast extravasation to the duodenum (n = 3), pseudoaneurysm close to the stent mesh with arteriobiliary fistula (n = 1), in-stent pseudoaneurysm alone (n = 4) and in-stent pseudoaneurysm with arteriobiliary fistula (n = 1). The origins of bleeding were the right hepatic artery (n = 9), the gastroduodenal artery (n = 3), the right anterior sectional artery (n = 2), the right posterior sectional artery (n = 1), the left hepatic artery (n = 2) and proper hepatic artery (n = 1) and the posterior superior pancreaticoduodenal artery (n = 1).

Figure 2.

A 67-year-old female (patient 19) who underwent pancreatoduodenectomy due to common bile duct cancer presenting with hemobilia and hematochezia (pseudoaneurysm close to the stent mesh). (a,b) Arterial phase axial and coronal reconstructed CT images show an enhancing round lesion (arrow) overlapping with the biliary stent in the hepatojejunostomy area. (c) Digital subtraction arteriography (DSA) of the common hepatic artery demonstrated a pseudoaneurysm (arrow) on the right hepatic artery. Note the left percutaneous transhepatic biliary drainage (PTBD) tube inserted into the stented bile duct (arrowheads) via segment III bile duct. (d) The pseudoaneurysm is not visible on a DSA image after successful embolization using microcoils and NBCA, and the distal hepatic arterial flow is preserved (arrowheads).

Figure 3.

A 74-year-old male (patient 13) who underwent subtotal gastrectomy because of advanced gastric cancer experienced hematochezia 306 days after percutaneous biliary stent placement (in-stent pseudoaneurysm with arteriobiliary fistula). (a,b) Non-enhanced (a) and contrast-enhanced (b) axial CT images show an enhancing round lesion (arrow) within the stent, suggesting an in-stent pseudoaneurysm. (c) DSA image of the common hepatic artery demonstrates a pseudoaneurysm (arrow) in the right hepatic artery with contrast extravasation into the stented common bile duct (arrowheads). (d) DSA image of the common hepatic artery after embolization shows complete exclusion of the pseudoaneurysm with cessation of bleeding. Note that the distal hepatic arterial flow is maintained (arrowheads).

Technical and Clinical Outcomes and Complications of TAE

In all 19 patients, immediate hemostasis was achieved after TAE. Recurrent bleeding 3 days after initial direct NBCA embolization for pseudoaneurysm of the gastroduodenal artery occurred in one patient (5.3%, 1/19, patient 15), who was treated with a second TAE. During the second TAE, regions both proximal and distal to the pseudoaneurysm and the pseudoaneurysm itself were embolized using coils and NBCA. Thus, the overall clinical success rate was 94.7% (18/19).

The rate of major procedure-related complications was 5.3% (1/19): acute hepatic failure occurred in one patient with a history of right hepatectomy due to hilar bile duct cancer (patient 12). Of note, the patient had main portal vein occlusion due to tumor involvement prior to TAE.

Survival

Follow-up until death was available for all 19 patients. The mean survival period after TAE was 325 days (range, 4–4320 days; Figure 4). The overall 30-day mortality rate was 26.3% (5/19). One patient (patient 12) with post-procedural hepatic failure died 4 days after the TAE, and four patients died due to tumor progression, a mean of 12.8 days (range, 4–25 days) after the TAE (patients 1, 3, 6, and 15).

Figure 4.

Kaplan-Meier curve for cumulative survival after transcatheter arterial embolization.

Discussion

SEMS-associated bleeding rarely occurs, with the reported incidence rates being only 0.3–3.6%,^8,12,13 although it is potentially fatal.^{17,18,20–23} Therefore, awareness of the clinical manifestations of delayed SEMS-associated arterial bleeding in patients with MBO is imperative for its prompt detection and management. However, to our knowledge, descriptions of delayed SEMS-associated arterial bleeding have been limited.^{17,18,20–23} Thus, in this largest study, we sought to analyze the clinical manifestations of 19 patients who presented with delayed SEMS-associated arterial bleeding, and to describe the clinical and technical outcomes after TAE.

The frequency of SEMS-associated delayed arterial bleeding is unclear. Nezu et al demonstrated that the incidence of delayed arterial bleeding after SEMS placement by ERCP was 1.2%.²⁰ To the best of our knowledge, the present study is the first report describing the frequency of delayed arterial bleeding after percutaneous biliary SEMS placement. We found that only 1.0% (19/1858) of the patients with MBO experienced delayed arterial bleeding after percutaneous biliary SEMS placement, and this low incidence is consistent with the results reported by Nezu et al.²⁰

Previous reports stated that bleeding occurred from 2 weeks to 1 year after SEMS placement,^{17,18,21–23} although Nezu et al reported that it could occur after only 5 days.²⁰ In the present study, delayed SEMS-associated arterial bleeding occurred at a median interval of 225 days (range, 22–2296 days), and none of the patients showed hemobilia or bleeding immediately after SEMS placement.

The precise mechanisms of delayed SEMS-associated arterial bleeding are not well-known. The suggested mechanisms are arterial wall necrosis and ductal inflammation caused by continuous irritation from the stent mesh, tumor invasion of the adjacent artery, adhesion between the stent and tumor and mucosal damage around the distal end of the stent^{18,20–22,27}: all of these present as a chronic process, and this can partially explain why the bleeding occurred in a delayed fashion.

Similar to Monroe et al²¹ and Hyun et al,²² the case of delayed arterial bleeding presenting with a pseudoaneurysm within the stent or close to the stent mesh was associated with circumferential tumor encasement of the stented bile duct. Moreover, anatomical differences may also reflect different mechanisms responsible for delayed arterial bleeding.²⁰ The right hepatic artery (12/19, 63.1%) was often near the edge of the SEMS and could be easily damaged by stents that penetrated the wall of the bile duct, and bleeding from the gastroduodenal artery or one of its branches (4/19, 21.0%), which may be adjacent to the proximal to middle portions of the SEMS, can be caused by arterial wall necrosis due to pressure exerted on the tumor. Moreover, inflammation surrounding the bile duct and the adhesions between the metal stent and artery may contribute to the pseudoaneurysm formation.¹⁸ An animal experiment by Carrasco et al²⁸ showed that chronic inflammation and fibrosis of the submucosal layer were observed after metallic stent placement, with the degree of those being in proportion to the stent placement period. Furthermore, Tsuji et al reported that the degree of chronic inflammation and fibrosis in the submucosal layer was increased by a combination of metallic stent placement and external irradiation.²⁹ Considering that 10 out of 19 patients (50.2%) in the present study had a history of radiation therapy, radiation may also be one of the contributors to the formation of a pseudoaneurysm. Also worthy of note is that 17 (89.5%) of the 19 patients in the present study had cholangitis. In 11 of these patients, hepatic abscesses were observed on CT scans performed prior to the onset of bleeding. This finding potentially indicates that cholangitis may be one of the causes of bleeding.

Generally, TAE is the preferred method to treat arterial pseudoaneurysms or arteriobiliary fistulas.^30–32 Our study demonstrated that TAE for delayed SEMS-associated arterial bleeding showed technical and clinical success rates of 100 and 94.7%, indicating that it is an effective therapeutic option. Recurrent bleeding occurred in one patient after initial direct NBCA embolization for pseudoaneurysm of the gastroduodenal artery, and the patient was successfully treated with a second TAE with both proximal and distal sites of the pseudoaneurysm being embolized. Herein, it is noteworthy that a basic principle is to avoid treating the pseudoaneurysm sac like a true aneurysm would be treated, as the pseudoaneurysm does not have walls. If only the pseudoaneurysm sac is packed with embolic agent, there is a high tendency to recanalization and re-expansion of the pseudoaneusym sac as the feeding vessel can transmit pressure to the soft tissues surrounding the pseudoaneurysm. Thus, attention should be paid to embolize regions both proximal and distal to the pseudoaneurysm.^33–35 Likewise, bleeding from the gastroduodenal artery or pancreaticoduodenal arcade should be treated by embolizing the arteries both proximal and distal to the pseudoaneurysm, to prevent retrograde bleeding because of its dual supply from both the common hepatic artery and superior mesenteric artery.³⁶

Nevertheless, ischemia is a major concern after TAE. In this study, acute hepatic failure occurred after TAE in one patient who had undergone right hepatectomy and bile duct resection without hepatopetal portal flow because of tumor involvement of the main portal vein. Thus, TAE for hepatic artery bleeding after major hepatectomy and extrahepatic bile duct resection may be contraindicated, and poor portal venous flow makes hepatic artery embolization inadvisable. There has been a recent increase in the use of stent grafts for the treatment of bleeding from the gastroduodenal artery stump or extrahepatic artery, to preserve distal hepatic arterial flow and minimize the risk of hepatic injury,³⁴ but stent-graft placement was technically impossible in our advanced cancer patients with severe anatomical tortuosity of the artery or a small-caliber target artery.

In the present study, patient survival was not longer than expected despite the successful hemostasis because SEMS-associated bleeding was a delayed complication in patients with advanced malignancies. The mean survival period after TAE was 325 days (range, 4–4320 days). However, there was no patient who died of ongoing bleeding after TAE.

The present study has some limitations. First, it has the inherent limitations of a retrospective study. Second, the small number of patients with delayed SEMS-associated arterial bleeding prohibits robust statistical analysis. Nevertheless, this study is the largest case series so far reported. Finally, we did not explore the factors associated with the development of delayed arterial bleeding after SEMS placement. However, there would be not much clinical significance of exploring associated factors because incidence rate of the bleeding was only 1%.

In conclusion, TAE is a safe and effective therapeutic option for delayed SEMS-associated arterial bleeding, which is a rare but potentially life-threatening complication.

This study was written in accordance with STROBE guidelines.