Transcatheter Arterial Embolization for Gastrointestinal Bleeding Associated with Gastric Carcinoma: Prognostic Factors Predicting Successful Hemostasis and Survival

CLINICAL STUDY

Transcatheter Arterial Embolization for Gastrointestinal Bleeding Associated with Gastric Carcinoma: Prognostic Factors Predicting Successful Hemostasis and Survival Sangik Park, MD, Ji Hoon Shin, MD, Dong-Il Gwon, MD, Hyoung Jung Kim, MD, Kyu-Bo Sung, MD, Hyun-Ki Yoon, MD, Gi-Young Ko, MD, and Heung Kyu Ko, MD

ABSTRACT Purpose: To evaluate outcomes of transcatheter arterial embolization (TAE) for gastric cancer–related gastrointestinal (GI) bleeding and factors associated with successful TAE and improved survival after TAE. Materials and Methods: This retrospective study included 43 patients (34 men; age 60.6 y ± 13.6) with gastric cancer–related GI bleeding undergoing angiography between January 2000 and December 2015. Clinical course, laboratory findings, and TAE characteristics were reviewed. Technical success of TAE was defined as target area devascularization, and clinical success was defined as bleeding cessation with hemodynamic stability during 72 hours after TAE. Student t test was used for comparison of continuous variables, and Fisher exact test was used for categorical variables. Univariate and multivariate analysis were performed to identify predictors of successful TAE and 30-day survival after TAE. Results: TAE was performed in 40 patients. Technical and clinical success rates of TAE were 85.0% and 65.0%, respectively. Splenic infarction occurred in 2 patients as a minor complication. Rebleeding after TAE occurred in 7 patients. Death related to bleeding occurred in 5 patients. Active bleeding (P ¼ .044) and higher transfusion requirement (3.3 U ± 2.6 vs 1.8 U ± 1.7; P ¼ .039) were associated with TAE failure. Successful TAE predicted improved 30-day survival after TAE on univariate and multivariate analysis (P ¼ .018 and P ¼ .022; odds ratio, 0.132). Conclusion: TAE for gastric cancer–associated GI bleeding may be a lifesaving procedure. Severe bleeding with a higher transfusion requirement and active bleeding on angiography predicted TAE failure.

ABBREVIATIONS GI ¼ gastrointestinal, NBCA ¼ N-butyl cyanoacrylate, PVA ¼ polyvinyl alcohol, RBC ¼ red blood cell, TAE ¼ transcatheter arterial embolization

It is reported that gastric carcinoma accounts for 1%–8% of acute upper gastrointestinal (GI) bleeding (1–3). Endoscopic hemostasis is considered to be the first-line modality for evaluation and treatment of upper GI bleeding (4–6) and is effective in the management of gastric cancer–related GI bleeding; initial hemostasis is achieved in 67%–100%

of patients treated with endoscopic therapy (2,3,7–9). However, rebleeding occurs in 41%–80% of patients after initial endoscopic hemostasis (2,3,9), and tumors measuring  20 mm in maximal diameter are associated with failure of endoscopic treatment (10). Palliative radiotherapy is an option for GI bleeding related to unresectable gastric

From the Department of Radiology and Research Institute of Radiology, University of Ulsan College of Medicine, Asan Medical Center, 86 Asanbyeongwon-Gil, Songpa-Gu, Seoul 138-736, Korea. Received July 2, 2016; final revision received March 12, 2017; accepted March 13, 2017. Address correspondence to J.H.S.; E-mail: [email protected]

© SIR, 2017

None of the authors have identified a conflict of interest.

J Vasc Interv Radiol 2017; 28:1012–1021 http://dx.doi.org/10.1016/j.jvir.2017.03.017

Volume 28 ▪ Number 7 ▪ July ▪ 2017

carcinoma, with reported clinical success rates of 54%–80% (11–16). However, it is not suitable for patients with resectable tumors or patients presenting with hemodynamic instability (13). Transcatheter arterial embolization (TAE) is another option for treating gastric cancer–related GI bleeding (17,18). Previous retrospective studies of TAE for gastric cancer–related GI bleeding reported a poor outcome in patients requiring TAE, with clinically successful hemostasis achieved in 40%–52% of the patients (17,18), and the predictive factors of successful TAE are not reported in the literature. Identification of the prognostic factors may aid in treatment decision making. Therefore, this study evaluated the prognostic factors for successful TAE and improved survival after TAE in patients with gastric cancer–related GI bleeding.

MATERIALS AND METHODS Patients This retrospective study was approved by the institutional review board of the center, and informed consent was waived. The study included 43 consecutive patients (34 men; mean age, 60.6 y ± 13.6 [SD]; range, 37–94 y) with GI bleeding associated with pathology-proven gastric carcinoma undergoing angiography and/or TAE at a single tertiary center between January 2000 and December 2015. Electronic medical records, laboratory findings, and endoscopic and/or radiologic features were reviewed. Clinical data including symptoms and signs of GI bleeding; history of previous bleeding necessitating hospital admission, gastric surgery, chemotherapy, and use of antiplatelet or anticoagulant agents; units of packed red blood cells (RBCs) transfused during the 24 hours before angiography; hemodynamic instability; use of vasopressors; and patient survival time were collected. Platelet counts, prothrombin time, and activated partial thromboplastin time on patient presentation and the lowest hemoglobin levels during the 24 hours before angiography were ascertained. Computed tomography (CT) images were reviewed for measurement of the maximal tumor diameter and assessment of the tumor location and the presence of lymph node, liver, or peritoneal metastases. The maximal tumor diameter was measured on either axial or coronal planes, whichever were larger, for mass-forming lesions, and in diffuse-type tumors, the maximal extent of the wall thickening was measured. The 7th edition of the American Joint Committee on Cancer TNM staging system was applied for disease staging (19).

Clinical Presentation The characteristics of the 43 patients and their gastric cancers are summarized in Table 1. Symptoms of hemorrhage, such as hematemesis, hematochezia, or melena, were present in 40 patients (93.0%); the other 3 patients presented with abdominal pain (n ¼ 2) and

1013

Table 1. Patient and Cancer Characteristics (N ¼ 43 Patients) Characteristics Age, y, mean ± SD

Values 60.6 ± 13.6

Sex, male, n (%)

34 (79.1)

History of previous bleeding, n (%)

15 (34.9)

History of previous chemotherapy, n (%)

26 (60.5)

History of gastric surgery, n (%)

7 (16.3)

Use of antiplatelet or anticoagulant agents, n (%)

6 (14.0)

Hemodynamic instability, n (%)

16 (37.2)

PLT count < 50,000/μL, n (%) Hb level, g/dL, mean ± SD RBC transfusion, U, mean ± SD Use of vasopressors, n (%)

2 (4.7) 7.70 ± 1.87 2.3 ± 2.1 2 (4.7)

Gastric cancer diagnosis TNM classification (I/II/III/IV), n (%)

2 (4.9)/4 (9.8)/ 7 (17.1)/28 (68.3)*

Borrmann classification (I/II/III/IV), n (%)

2 (4.9)/13 (31.7)/ 21 (51.2)/5 (12.2)*

Tumor diameter, mm, mean ± SD

81.0 ± 35.1†

Hb ¼ hemoglobin; PLT ¼ platelet; RBC ¼ red blood cell. *n ¼ 41 (with 2 cases of anastomosis recurrence). † n ¼ 42 (1 patient did not undergo CT scan).

dyspnea on exertion (n ¼ 1). The median interval between the onset of hemorrhagic symptoms and presentation was 36 hours (range, 40 min to 90 d), and 50% of the 40 patients with bleeding symptoms had an acute onset (within 24 h). Of patients, 34 (79.1%) had previously been diagnosed with gastric carcinoma, and the median interval between diagnosis and presentation was 228 days (range, 2–1,311 d; interquartile range, 55–423.25 d). Among the 34 patients, 15 had a history of cancer-related GI bleeding requiring hospital admission. During the admission for the previous episodes, 9 of the 15 patients received endoscopic hemostasis, whereas the other 6 patients improved with conservative management. Seven patients with a previous diagnosis of gastric carcinoma had undergone gastric surgery, including palliative gastrojejunostomy (n ¼ 3), subtotal gastrectomy (n ¼ 2), gastrotomy with ligation of the left gastric artery (n ¼ 1), and primary closure with jejunostomy (n ¼ 1). Bleeding was the initial presentation in the 9 patients without a previous diagnosis of gastric carcinoma. In all 9 patients, gastric carcinoma was subsequently diagnosed on pathology.

Endoscopic Evaluation and Hemostasis Endoscopy was performed within 72 hours before angiography in 23 patients (53.5%) (Fig 1). For the other 20 patients, angiographic evaluation was done as the initial evaluation because of the expected technical challenges based on the tumor extent and location on CT or previous endoscopic features (n ¼ 15) or because of the patient’s hemodynamic instability (n ¼ 5). The decision regarding whether a patient should undergo either endoscopy or

1014 ▪ TAE for Gastric Cancer–Related GI Bleeding

Park et al ▪ JVIR

Figure 1. Initial management, angiographic findings, and outcomes of 43 patients. *Persistent bleeding after endoscopic hemostasis. †Endoscopic hemostasis was not attempted owing to technical difficulties.

angiography as the primary evaluation method was made by a gastroenterologist and an interventional radiologist. Endoscopic evaluation was done by 1 of 6 endoscopists at the hospital. The tumor morphology on endoscopy was classified using the Borrmann classification system (20), and the Forrest classification was applied for describing the bleeding foci (21). Active bleeding on endoscopy was noted in 20 patients; an oozing type of bleeding (Forrest 1b, n ¼ 16) was more frequently observed than spurting bleeding (Forrest 1a, n ¼ 4). The other 3 patients showed signs of recent bleeding with adherent clots (Forrest 2b). The decision whether or not to attempt endoscopic hemostasis was made by the gastroenterologist. In 9 of the 20 patients with active bleeding, endoscopic hemostasis was performed with clipping along with epinephrine spray or injection (n ¼ 2), argon plasma coagulation with epinephrine spray (n ¼ 3), epinephrine spray (n ¼ 2), epinephrine injection (n ¼ 1) or epinephrine and ethanol injection (n ¼ 1). All patients who had endoscopic treatment showed persistent bleeding after the procedures. In the other 11 patients with active bleeding and the 3 patients with adherent clots, hemostatic procedures were not undertaken owing to technical difficulties.

TAE Procedure Angiography and/or TAE procedures were performed by 1 of the hospital’s 4 interventionists (J.H.S., D.I.G., G.Y.K., H.K.K.) with 6–27 years of clinical experience in interventional radiology. Via the right femoral approach, a 5-F

catheter (RH catheter; Cook, Inc, Bloomington, Indiana) was introduced over a 0.035-inch guide wire (Radifocus; Terumo, Tokyo, Japan). Celiac arteriography was performed, and further interrogation, including left gastric arteriography, was done based on the abnormalities on celiac arteriography. The operator decided whether or not to obtain superior mesenteric arteriograms. For patients with active bleeding or tumor staining on angiography, TAE was performed after selection of the target vessel using 2.0- to 2.4-Fr microcatheters (Progreat; Terumo or Renegade; Boston Scientific, Marlborough, Massachusetts). Embolic agents included absorbable gelatin sponge alone (n ¼ 26); absorbable gelatin sponge in combination with microcoils (n ¼ 6), microcoils and N-butyl cyanoacrylate (NBCA) (n ¼ 1), or polyvinyl alcohol (PVA) (n ¼ 3); microcoils and PVA (n ¼ 1); microcoils and NBCA (n ¼ 1); NBCA and ethiodized oil (Lipiodol; Guerbet, Villepinte, France) (n ¼ 1); and PVA alone (n ¼ 1). The choice was dictated by the operator’s preference and familiarity. Proximal and distal embolization was performed for a pseudoaneurysm. In the absence of active bleeding or tumor staining, prophylactic embolization was performed targeting the left gastric, right gastric, or right gastroepiploic arteries based on the tumor location. Final angiograms were obtained for determination of the disappearance of bleeding foci. The electronic medical records and follow-up CT scans after TAE were reviewed to determine the presence or absence of procedure-related complications, such as gastric ischemia, perforation, or splenic infarction. Complications

Volume 28 ▪ Number 7 ▪ July ▪ 2017

Table 2. Vessels with Abnormalities According to Tumor Location Tumor Location (n ¼ 35*) Body (n ¼ 10)

Abnormal Vessel (n ¼ 35*) LGA (n ¼ 6) LGA and RGEA (n ¼ 2) RGA (n ¼ 1) LGA, RGA, and RGEA (n ¼ 1)

Fundus (n ¼ 1)

LGA and RGEA (n ¼ 1)

Antrum (n ¼ 7)

GDA (n ¼ 4) LGA (n ¼ 1) LGA and GDA (n ¼ 1)

Body and fundus (n ¼ 4)

LGA and RGEA (n ¼ 1) LGA (n ¼ 4)

Body and antrum (n ¼ 9)

LGA (n ¼ 2) GDA (n ¼ 1) CHA (n ¼ 1) LGA and RGA (n ¼ 2) LGA and GDA (n ¼ 1) LGA and RGEA (n ¼ 1)

Body, fundus, and antrum (n ¼ 3) Anastomosis site (n ¼ 1)

1015

Table 3. Clinical Success of TAE According to Angiographic Findings and TAE Characteristics Angiographic Findings and TAE Characteristics

Number of Patients (n ¼ 40)*

Clinical Success, n (%)

Angiographic findings Active bleeding

9

3 (33.3)

Tumor staining

25

18 (72.0)

6

5 (83.3)

Negative Embolized arteries LGA

17

12 (70.6)

15

12 (80.0)

LGA and RGEA

6

5 (83.3)

LGA and GDA

4

4 (100)

LGA and RGA

3

2 (66.7)

LGA, RGEA, and splenic artery

1

0

LGA and LHA†

1

1 (100)

8 2

2 (25.0) 0

LGA and other arteries

Arteries other than the LGA RGA

LGA, RGA, and RGEA (n ¼ 1) RGEA (n ¼ 1)

GDA

1

0

GDA and RGEA

1

1 (100)

RGEA

1

1 (100)

LGA and RGEA (n ¼ 2)

RGEA and SGA

1

0

LGA and GDA (n ¼ 1)

CHA

2

0

Absorbable gelatin sponge Absorbable gelatin sponge and other embolic materials

26 10

16 (61.5) 8 (80.0)

Absorbable gelatin sponge and microcoils

6

4 (66.7)

Absorbable gelatin sponge, microcoils, and NBCA

1

1 (100)

Absorbable gelatin sponge and PVA

3

3 (100)

Materials other than absorbable gelatin sponge Microcoils and PVA

4

2 (50.0)

CHA ¼ common hepatic artery; GDA ¼ gastroduodenal artery; LGA ¼ left gastric artery; RGEA ¼ right gastroepiploic artery; RGA ¼ right gastric artery. *Includes tumor staining (n ¼ 26) and active bleeding (n ¼ 9).

were classified according to the criteria proposed by the Society of Interventional Radiology (SIR) (22).

Definitions Hemodynamic instability was defined as systolic blood pressure < 90 mm Hg. Active bleeding was defined as extravasation of contrast medium or the presence of a pseudoaneurysm, and tumor staining was defined as the presence of abnormal vessels with a mucosal blush. Technical success was defined as target area devascularization in cases of active extravasation or tumor staining and as stasis of flow in the vessel presumed to supply the tumor, based on its location on CT and endoscopic findings (23). Clinical success was defined as cessation of hematemesis, hematochezia, or melena or clearing of nasogastric aspirate with hemodynamic stability during the 72 hours following the embolization procedures. Rebleeding was defined as the recurrence of bleeding symptoms with a change in vital signs or a decrease in hemoglobin levels of > 2 g/dL after TAE in patients with initial clinical success.

Statistical Analysis Means and SDs were used for the continuous variables. Student t test was used for comparison of the continuous variables regarding clinical success and 30-day survival. Fisher exact test was used for univariate analysis of the

Embolic materials

1

1 (100)

Microcoils and NBCA

1

0

NBCA and Lipiodol

1

0

PVA

1

1 (100)

CHA ¼ common hepatic artery; GDA ¼ gastroduodenal artery; LGA ¼ left gastric artery; LHA ¼ left hepatic artery; NBCA ¼ N-butyl cyanoacrylate; PVA ¼ polyvinyl alcohol; RGA ¼ right gastric artery; RGEA ¼ right gastroepiploic artery; SGA ¼ short gastric artery; TAE ¼ transcatheter arterial embolization. *The 3 patients who did not undergo TAE are not included. † The patient had gastric cancer with direct invasion of the left liver.

categorical variables associated with clinical success and 30-day patient survival after TAE. Multivariate logistic regression analysis was performed to identify the predictive factors for 30-day survival after TAE. The Kaplan-Meier method was used for survival analysis. All statistical analyses were performed using IBM SPSS Statistics for Windows, Version 20.0 (IBM Corp, Armonk, New York). P values < 0.05 were considered statistically significant.

1016 ▪ TAE for Gastric Cancer–Related GI Bleeding

RESULTS Characteristics and Short-Term Outcomes of TAE The vessels with angiographic abnormalities according to the tumor location are summarized in Table 2, and the clinical success rates of TAE according to angiographic findings and TAE characteristics are shown in Table 3. The most common angiographic abnormality was tumor staining (n ¼ 26). Active bleeding was noted in 9 patients, consisting of 8 with active extravasation of contrast medium and 1 with a pseudoaneurysm (Fig 2a–d). The other 8 patients showed no angiographic abnormalities. TAE was performed in 40 of 43 patients (93.0%). For the other 3 patients, TAE was not done at the discretion of the interventionists in the absence of active bleeding (n ¼ 2) and owing to failure of target vessel selection (n ¼ 1). Most commonly, only the left gastric artery was chosen for selective embolization (n ¼ 17), whereas in 15 patients, the left gastric artery combined with other vessels, such as the right gastroepiploic artery (n ¼ 6), gastroduodenal artery (n ¼ 4), right gastric artery (n ¼ 3), left hepatic artery (n ¼ 1), or right gastroepiploic artery plus the

Park et al ▪ JVIR

splenic artery (n ¼ 1) received embolization for successful hemostasis. In the remaining 8 patients, arteries other than the left gastric artery received embolization (Table 3). After TAE, technical success was achieved in 34 patients (85.0%; 34 of 40), and clinical success was achieved in 26 patients (65.0%; 26 of 40). Causes of technical failure in the 6 patients were failure to select target vessels (n ¼ 3), inappropriate selection of target vessels (n ¼ 2), and missed active extravasation (n ¼ 1). Of the 14 patients with clinical failure, 13 had persistent bleeding after TAE, whereas the other patient had unstable vital signs without evidence of further bleeding. The patients were managed with volume replacement or transfusion of packed RBCs (n ¼ 9), reattempted endoscopic hemostasis (n ¼ 1), emergency surgery (n ¼ 3), or radiotherapy (n ¼ 1).

Characteristics of Rebleeding after Initially Successful TAE During a median follow-up period of 115 days (range, 10–2,879 d), rebleeding occurred in 7 patients (26.9%; 7 of 26) with initial clinical success. The median interval between TAE and rebleeding was 16 days (range, 5–318 d).

Figure 2. Images of a 49-year-old man with a history of gastric cancer–related pyloric obstruction treated with stent placement and palliative gastrojejunostomy, presenting with hematemesis. (a) Endoscopy shows tumor ingrowth with oozing bleeding. Part of the impacted stent is seen (arrow). (b) CT scan shows an ill-defined mass in the gastric antrum (arrowheads) surrounding the stent. The palliative gastrojejunostomy site is seen (white arrow). There is peritoneal infiltration in the left upper abdomen (black arrow), which is related to peritoneal carcinomatosis. Also noted is left hydronephrosis related to periureteral metastasis. (c) Celiac trunk arteriography shows active bleeding from the common hepatic artery (arrowheads) in the superior aspect of the stent. Embolization was done using microcoils and NBCA. (d) Completion angiogram shows no residual contrast extravasation. The patient died 4 days after treatment owing to persistent bleeding.

Volume 28 ▪ Number 7 ▪ July ▪ 2017

Of the patients, 6 had undergone TAE with absorbable gelatin sponge, whereas microcoils and absorbable gelatin sponge were used as embolic agents in the other patient. Repeat angiography performed in 6 of the 7 patients showed recanalization of the left gastric artery (n ¼ 4) (Fig 3a–d), collateral tumor supply from the superior mesenteric artery (n ¼ 1) (Fig 4a–d), and irregularity of the left gastric and right gastroepiploic arteries that previously received embolization (n ¼ 1). Repeat TAE of the left gastric artery alone (n ¼ 3), both left and right gastric arteries (n ¼ 1), both left and right gastric and left and right gastroepiploic arteries (n ¼ 1), or the inferior pancreaticoduodenal branch of the superior mesenteric artery (n ¼ 1) was performed using absorbable gelatin sponge alone (n ¼ 2), absorbable gelatin sponge with microcoils (n ¼ 2), absorbable gelatin sponge with PVA particles (n ¼ 1), or PVA particles alone (n ¼ 1). TAE for rebleeding was clinically successful in 4 of the 6 patients. The 2 patients with TAE failure and the patient who did not undergo angiographic evaluation for rebleeding were managed conservatively. One of the 2 patients with failed

1017

repeat TAE died as a result of rebleeding 329 days after the initial TAE.

Complications of TAE Follow-up endoscopies and CT scans were performed within 1 month after TAE in 7 (17.5%) and 12 (30%) patients, respectively. No major procedure-related complications (SIR classification C or higher) were noted, although 2 cases of splenic infarction were detected on CT scans as minor complications (SIR classification B). One patient had undergone selective TAE of the left gastric, right gastroepiploic, and splenic arteries using absorbable gelatin sponge and microcoils, and a CT scan performed 44 days after TAE showed splenic infarction with rim enhancement and air densities, suggestive of combined infection or inflammation (Fig 5a–d). However, the patient did not have signs and symptoms related to splenic infarction. The other patient had undergone embolization of the left gastric artery using absorbable gelatin sponge and complained of abdominal pain

Figure 3. Images of a 43-year-old man with rebleeding after initially successful TAE. (a) CT scan with coronal reconstruction shows a 62-mm, ulcerofungating mass arising in the gastric lower body (arrows) with multiple hepatic metastases. There is a metastatic lymph node in the hepatoduodenal ligament (arrowhead). (b) Left gastric arteriogram shows positive tumor staining in the gastric body (arrowheads). Embolization was performed using absorbable gelatin sponge. (c) Completion arteriogram shows no residual tumor staining. Initial clinical success was achieved. (d) Repeat left gastric arteriogram performed 7 days after the initial embolization for evaluation of rebleeding shows the recanalized left gastric artery with tumor staining (arrowheads). Successful repeat TAE was done using absorbable gelatin sponge and PVA particles.

1018 ▪ TAE for Gastric Cancer–Related GI Bleeding

Park et al ▪ JVIR

Figure 4. Images of a 52-year-old man with rebleeding at 54 days after successful TAE. (a) Gastroduodenal arteriography shows tumor staining (arrowheads) supplied by the gastroduodenal and right gastroepiploic arteries. Embolization of the vessels was done using microcoils. (b) Completion angiogram shows no residual tumor staining. The microcoils deployed in the gastroduodenal (arrowheads) and right gastroepiploic arteries (arrow) are shown. (c) Superior mesenteric arteriography shows no collateral flow to the region. (d) Repeat angiography performed 54 days after initial TAE for evaluation of rebleeding shows reappearance of tumor staining (arrowheads) supplied by the inferior pancreaticoduodenal branch (arrows) of the superior mesenteric artery. Embolization of the branch was performed using absorbable gelatin sponge and microcoils (not shown). The repeat TAE procedure was clinically successful.

immediately after the procedure. Partial splenic infarction was noted on a follow-up CT scan performed 12 days after TAE.

Predictors of Successful TAE and Improved 30-Day Survival after TAE The results of univariate analysis of the predictive factors for clinical success of TAE and 30-day survival after TAE are presented in Table 4 and Table 5, respectively. Patients with successful TAE had a significantly lower requirement for packed RBCs during the 24 hours before angiography (1.8 U ± 1.7 vs 3.3 U ± 2.6; P ¼ .039). The clinical success rates of TAE for tumor staining (72.0%; 18 of 25) and negative angiographic findings (83.3%; 5 of 6) were higher than clinical success rates of TAE for active bleeding (33.3%; 3 of 9) as shown in Table 3, and the difference was statistically significant (P ¼ .044) (Table 4). The clinical success of TAE procedures using absorbable gelatin sponge alone (61.5%; 16 of 26) was

not significantly different from that using other materials along with absorbable gelatin sponge (80.0%; 8 of 10) or agents other than absorbable gelatin sponge (50.0%; 2 of 4) (P ¼ .730). On univariate analysis, clinical success of TAE was associated with increased rates of 30-day survival after TAE (P ¼ .018) (Table 5) and the association was independent of the TNM stage of the tumors and the hemodynamic status of the patients (P ¼ .022; adjusted odds ratio, 0.132; 95% confidence interval, 0.023–0.746) (Table 6).

Long-Term Outcomes after TAE The overall 30-day mortality rate after TAE was 25% (10 of 40). During a median follow-up period of 85 days, the median survival time of patients was 85 days (95% confidence interval, 23.0–147.1 d). At the end of the study period, 3 patients remained alive, and all had undergone curative resection for the tumors after TAE. None of the patients without subsequent curative resection

Volume 28 ▪ Number 7 ▪ July ▪ 2017

1019

Figure 5. Images of a 67-year-old man with partial splenic infarction after embolization targeting the splenic artery. (a) Endoscopy shows a hematoma filling the gastric body (arrows). The primary lesion could not be evaluated owing to the hematoma. (b) Splenic arteriogram shows active contrast extravasation (arrowheads), probably from the short gastric artery. Microcoils that were deployed in the right gastroepiploic artery as part of the empirical embolization are visible (arrows). (c) Arteriogram obtained after embolization shows no further extravasation. (d) Follow-up CT scan obtained 44 days after embolization shows microcoils in the splenic artery (arrow). There is partial splenic infarction with rim enhancement and internal air densities, suggestive of combined infection or inflammation (arrowheads). The patient did not have symptoms related to the splenic lesion.

Table 4. Univariate Analysis of Clinical Outcomes after TAE Variables

Clinical Success (n ¼ 26)

Clinical Failure (n ¼ 14)

59.7 ± 13.4

62.1 ± 13.7

.587

21 (80.8)

13 (92.9)

.399

Age, y, mean ± SD Sex, male, n (%)

P Value

History of previous bleeding, n (%)

8 (30.8)

7 (50.0)

.310

History of previous chemotherapy, n (%)

17 (65.4)

8 (57.1)

.736

History of gastric surgery, n (%)

3 (11.5)

3 (21.4)

.646

Use of antiplatelet or anticoagulant agents, n (%)

5 (19.2)

1 (7.1)

.399

Hemodynamic instability, n (%)

9 (34.6)

6 (42.9)

.736

Hb, g/dL, mean ± SD RBC transfusion, U, mean ± SD

7.74 ± 1.81 1.8 ± 1.7

7.90 ± 2.12 3.3 ± 2.6

.801 .039*

16 (64.0)†

11 (78.6)

.477

84.5 ± 35.6

78.4 ± 36.1

.612

3 (11.5)

6 (42.9)

.044*

TNM stage 4, n (%) Tumor diameter, mm, mean ± SD Active bleeding, n (%) Negative angiography, n (%)

5 (19.2)

1 (7.1)

.399

Absorbable gelatin sponge–only embolization, n (%)

16 (61.5)

10 (71.4)

.730

Hb ¼ hemoglobin levels; RBC ¼ red blood cell; TAE ¼ transcatheter arterial embolization. *Indicates statistical significance. † n ¼ 25 (1 patient did not undergo CT scan).

survived > 462 days after TAE. Bleeding-related mortality occurred in 5 patients (12.5%; 5 of 40), with 3 patient deaths occurring within 4 days after TAE as a result of failure of hemostasis. Another patient died 107 days after

TAE, owing to chronic, persistent bleeding after TAE and subsequent palliative radiotherapy. The other patient died 329 days after successful TAE as a result of recurrent bleeding.

1020 ▪ TAE for Gastric Cancer–Related GI Bleeding

Park et al ▪ JVIR

Table 5. Univariate Analysis of Overall 30-Day Patient Survival after TAE Survival (n ¼ 30)

Mortality (n ¼ 10)

P Value

61.4 ± 13.2

58.1 ± 14.2

.511

Sex, male, n (%)

25 (83.3)

9 (90.0)

1

History of previous bleeding, n (%)

11 (36.7)

4 (40.0)

1

History of previous chemotherapy, n (%)

17 (56.7)

8 (80.0)

.269

History of gastric surgery, n (%)

4 (13.3)

2 (20.0)

.629

Use of antiplatelet or anticoagulant agents, n (%)

5 (16.7)

1 (10.0)

1

Hemodynamic instability, n (%)

9 (30.0)

6 (60.0)

.135

Hb, g/dL, mean ± SD RBC transfusion, U, mean ± SD

7.74 ± 2.00 2.4 ± 2.3

7.97 ± 1.68 2.3 ± 1.8

.741 .923 .131

Variable Age, y, mean ± SD

TNM stage 4, n (%)

18 (62.1)†

9 (90.0)

78.7 ± 32.5

93.4 ± 43.2

.261

Active bleeding, n (%)

6 (20.0)

3 (30.0)

.665

Clinical success, n (%)

23 (76.7)

3 (30.0)

.018*

Subsequent curative resection, n (%)

4 (13.3)

0

.556

Tumor diameter, mm, mean ± SD

Hb ¼ hemoglobin levels; RBC ¼ red blood cells; TAE ¼ transcatheter arterial embolization. *Indicates statistical significance. † n ¼ 29 (1 patient did not undergo CT scan).

Table 6. Multivariate Logistic Regression Analysis of Predictive Factors for 30-Day Patient Survival after TAE P Value

Odds Ratio

95% CI

Hemodynamic instability TNM stage 4

.169 .224

3.352 4.399

0.598–18.780 0.404–47.867

Clinical success

.022*

0.132

0.023–0.746

CI ¼ confidence interval; TAE ¼ transcatheter arterial embolization. *Indicates statistical significance.

DISCUSSION The technical and clinical success rates of TAE for gastric cancer–related GI bleeding in this study were 85.0% (34 of 40) and 65.0% (26 of 40), respectively. Patients requiring TAE had poor outcomes with a median survival of 85 days after the procedures. Active bleeding on angiography and a higher requirement of RBC transfusion during the 24 hours before TAE were associated with failure of TAE. Clinical success of TAE predicted improved 30-day survival after TAE. Patients who had subsequent curative resection after TAE for had excellent survival outcomes compared with patients who did not have subsequent resection. For patients with potentially resectable tumors, TAE may serve as a bridge to curative surgery, reducing the requirement for emergency surgery, which is associated with higher postoperative complication rates (24). The clinical success rate of TAE for active bleeding was lower than the success rates for tumor staining and negative angiography, although the difference was not statistically significant. Considering the higher frequency of tumor staining in patients with cancer-related gastric bleeding, also

noted by Meehan et al (17), and the effectiveness of TAE for tumor staining and negative angiography, it may be reasonable to perform TAE in the absence of active bleeding on angiography. Tandberg et al (25) reported higher rates of bleeding cessation when empirical embolization was performed for tumor-related GI bleeding than when embolization was not performed. As active extravasation may suggest more severe bleeding than tumor staining or negative findings, it could be associated with higher failure rates. The clinical success rate of 65.0% (26 of 40) in this study is higher than the clinical success rates of the previous studies (40%–52%) (17,18). This is partially due to the difference in the definition of clinical success and rebleeding and in patient selection. The study by Meehan et al (17) defined clinical success as persistent absence of bleeding for > 30 days after TAE compared with 72 hours in this study. The present definition of clinical success also differs from the study by Lee et al (18) in that it includes patients with resectable tumors. As there is no proven treatment with curative intent other than surgical resection for gastric carcinoma, inclusion of patients with resectable tumors may have led to improved survival in the patient group of this study. The left gastric artery alone and the left gastric artery with other vessels, such as the right gastroepiploic artery, were the most frequent target vessels for embolization in this study, which was in agreement with that seen in previous studies (17,18). Despite the absence of guidelines regarding the optimal target vessels for embolization, the choice of target vessels should be based on the tumor locations. For example, the right gastroepiploic artery could be considered as the target vessel for tumors located in the greater curvature side of the gastric body, and the short gastric artery could be considered for tumors located in the fundus, as the vessels are responsible for the blood supply in those areas.

Volume 28 ▪ Number 7 ▪ July ▪ 2017

Although there was a wide variation in the choice of embolic materials, temporary embolic materials, such as absorbable gelatin sponge, were most commonly used in this study. As recanalization of the left gastric artery with embolization with absorbable gelatin sponge was seen in the 4 patients with rebleeding, it is possible that this embolic material is associated with recurrent bleeding in patients with initial clinical success. Although there are no guidelines indicating the optimal choice of embolic materials, there is a potential benefit of adding permanent embolic materials to prevent recanalization of target vessels. Moreover, repeating embolization with absorbable gelatin sponge in patients with rebleeding may not be appropriate, although it was performed in 2 patients with rebleeding in this study. However, embolization using absorbable gelatin sponge particles alone was not associated with clinical failure compared with embolization using other embolic agents or absorbable gelatin sponge particles along with other agents. Schenker et al (23) also reported that the choice of embolic agents did not predict clinical success after TAE for acute upper GI bleeding. This study has several limitations. Owing to its retrospective, nonrandomized nature, there is heterogeneity in the patient group, tumor characteristics, and angiographic findings with potential confounding. The choice of the target vessel and embolic agent varied and depended on operator preference, which could have affected the outcomes. Some complications after the procedure might not have been detected because of the lack of a standardized follow-up protocol after TAE. The small number of the patients precludes robust statistical analysis. There is also a possible selection bias with the data coming from a single tertiary center, considering there was a significant proportion of known gastric carcinoma and previous chemotherapy in the patients. In addition, there were no patients with pathologic diagnosis for exact cause of bleeding, which might have affected the treatment choice. In conclusion, TAE for GI bleeding may be a lifesaving procedure. Severe bleeding with a higher transfusion requirement and active bleeding on angiography predicted TAE failure.

REFERENCES 1. Webb WA, McDaniel L, Johnson RC, Haynes CD. Endoscopic evaluation of 125 cases of upper gastrointestinal bleeding. Ann Surg 1981; 193:624. 2. Savides T, Jensen D, Cohen J, et al. Severe upper gastrointestinal tumor bleeding: endoscopic findings, treatment, and outcome. Endoscopy 1996; 28:244–248.

1021

3. Loftus EV, Alexander GL, Ahlquist DA, Balm RK. Endoscopic treatment of major bleeding from advanced gastroduodenal malignant lesions. Mayo Clin Proc 1994; 69:736–740. 4. Sung JJ, Chan FK, Chen M, et al. Asia-Pacific Working Group consensus on non-variceal upper gastrointestinal bleeding. Gut 2011; 60:1170–1177. 5. Laine L, Jensen DM. Management of patients with ulcer bleeding. Am J Gastroenterol 2012; 107:345–360. 6. Hwang JH, Fisher DA, Ben-Menachem T, et al. The role of endoscopy in the management of acute non-variceal upper GI bleeding. Gastrointest Endosc 2012; 75:1132–1138. 7. Mathus-Vliegen EM, Tytgat GN. Analysis of failures and complications of neodymium: YAG laser photocoagulation in gastrointestinal tract tumors. Endoscopy 1990; 22:17–23. 8. Mathus-Vliegen EM, Tytgat GN. Laser photocoagulation in the palliative treatment of upper digestive tract tumors. Cancer 1986; 57:396–399. 9. Kim Y-I, Choi IJ, Cho S-J, et al. Outcome of endoscopic therapy for cancer bleeding in patients with unresectable gastric cancer. J Gastroenterol Hepatol 2013; 28:1489–1495. 10. Koh KH, Kim K, Kwon DH, et al. The successful endoscopic hemostasis factors in bleeding from advanced gastric cancer. Gastric Cancer 2013; 16:397–403. 11. Tey J, Choo BA, Leong CN, et al. Clinical outcome of palliative radiotherapy for locally advanced symptomatic gastric cancer in the modern era. Medicine 2014; 93:e118. 12. Tey J, Back MF, Shakespeare TP, et al. The role of palliative radiation therapy in symptomatic locally advanced gastric cancer. Int J Radiat Oncol Biol Phys 2007; 67:385–388. 13. Kondoh C, Shitara K, Nomura M, et al. Efficacy of palliative radiotherapy for gastric bleeding in patients with unresectable advanced gastric cancer: a retrospective cohort study. BMC Palliat Care 2015; 14:37. 14. Kim MM, Rana V, Janjan NA, et al. Clinical benefit of palliative radiation therapy in advanced gastric cancer. Acta Oncol 2008; 47:421–427. 15. Hashimoto K, Mayahara H, Takashima A, et al. Palliative radiation therapy for hemorrhage of unresectable gastric cancer: a single institute experience. J Cancer Res Clin Oncol 2009; 135:1117–1123. 16. Asakura H, Hashimoto T, Harada H, et al. Palliative radiotherapy for bleeding from advanced gastric cancer: is a schedule of 30 Gy in 10 fractions adequate? J Cancer Res Clin Oncol 2011; 137:125–130. 17. Meehan T, Stecker MS, Kalva SP, Oklu R, Walker TG, Ganguli S. Outcomes of transcatheter arterial embolization for acute hemorrhage originating from gastric adenocarcinoma. J Vasc Interv Radiol 2014; 25:847–851. 18. Lee HJ, Shin JH, Yoon HK, et al. Transcatheter arterial embolization in gastric cancer patients with acute bleeding. Eur Radiol 2009; 19:960–965. 19. Washington K. 7th edition of the AJCC cancer staging manual: stomach. Ann Surg Oncol 2010; 17:3077–3079. 20. Kajitani T. The general rules for the gastric cancer study in surgery and pathology. Jpn J Surg 1981; 11:127–139. 21. Forrest JH, Finlayson N, Shearman D. Endoscopy in gastrointestinal bleeding. Lancet 1974; 304:394–397. 22. Omary RA, Bettmann MA, Cardella JF, et al. Quality improvement guidelines for the reporting and archiving of interventional radiology procedures. J Vasc Interv Radiol 2003; 14:S293–S295. 23. Schenker MP, Duszak R, Soulen MC, et al. Upper gastrointestinal hemorrhage and transcatheter embolotherapy: clinical and technical factors impacting success and survival. J Vasc Interv Radiol 2001; 12:1263–1271. 24. Lee HJ, Park do J, Yang HK, Lee KU, Choe KJ. Outcome after emergency surgery in gastric cancer patients with free perforation or severe bleeding. Dig Surg 2006; 23:217–223. 25. Tandberg DJ, Smith TP, Suhocki PV, et al. Early outcomes of empiric embolization of tumor-related gastrointestinal hemorrhage in patients with advanced malignancy. J Vasc Interv Radiol 2012; 23: 1445–1452.