Esophageal cancer bleeding is one of the most life-threatening acute complications in esophageal cancer patients, often presenting as melena or hematemesis (1, 2). Bleeding may result from tumor ulceration or the formation of an arterial-esophageal fistula (AEF), a condition in which tumor invasion or treatment erodes adjacent arterial walls, creating an abnormal passage between the artery and esophagus that can result in massive hemorrhage (3). In particular, AEF is associated with a high mortality rate ranging from 46.6% to 63% and requires urgent thoracic endovascular aortic repair or surgical intervention (2, 4–6). Currently, upper gastrointestinal endoscopy is the first-line treatment for non-variceal upper gastrointestinal hemorrhage (7, 8). Nevertheless, no available endoscopic treatment demonstrates long-term efficacy (9) and approximately 10% of patients experience rebleeding after endoscopic treatment, with reduced long-term hemostasis success and increased mortality (10). Additionally, endoscopy poses a risk of esophageal perforation (11), underscoring the need for alternative treatment options.

Transcatheter arterial embolization (TAE) has emerged as an effective alternative to endoscopic treatment (12–14), particularly for non-variceal upper gastrointestinal bleeding that is unmanageable with endoscopy (15). However, the vascular anatomy associated with esophageal cancer bleeding is complex, involving both tumor-feeding arteries and potential AEFs. Thus, accurately identifying bleeding vessels is crucial for TAE success. With advancements in imaging and interventional techniques, preprocedural computed tomography angiography (CTA) has become a valuable tool for evaluating tumor extent and identifying bleeding-related arteries, providing crucial anatomical guidance for TAE (16–18). This study aimed to evaluate the therapeutic effect of CTA-guided TAE for esophageal cancer bleeding.

A total of eight patients with esophageal squamous cell carcinoma who experienced upper gastrointestinal bleeding were included in the analysis. The cohort consisted of seven males and one female, with a mean age of 67 ± 12.9 years. Six patients had received prior antitumor therapy, including chemotherapy, chemoradiotherapy, immunotherapy, or radiotherapy. The primary tumors were located in the cervical, upper, middle, or lower esophagus, with varying TNM stages. CTA identified tumor-related vascular abnormalities in several cases, such as AEFs. No cases of aortic arterioesophageal fistulas were identified in this cohort. Detailed patient demographics, clinical characteristics, CTA findings, embolic agents, and TAE outcomes are summarized in Table 1 .

Guided by preprocedural CTA, the interventional radiologists located the likely culprit bleeding arteries and performed angiography and embolization. Preprocedural CTA identified three patients with non-aortic AEFs, manifesting as bronchial artery pseudoaneurysm (No. 2, Figure 1 ; No. 4, and No. 5, Figure 2 ) and superior thyroid artery pseudoaneurysm (No. 3). Angiography could further identify bleeding from multiple arteries that led to tumor staining, such as the left gastric artery, left hepatic artery, inferior phrenic artery, proper esophageal artery and posterior intercostal artery (No. 8, Figure 3 ). Seven of the eight patients achieved both technical success and clinical success following TAE. One patient failed to achieve embolization due to complex vascular anatomy and missed detection of a pseudoaneurysm in the left superior thyroid artery, resulting in death from hemorrhage (No. 3). Two patients experienced mild complications, including neck (No. 1) and chest discomfort (No. 6), which resolved spontaneously with conservative management.

During follow-up, five patients had rebleeding, four of whom died due to the recurrent hemorrhage (No. 1, 2, 4, 7), while one received a second TAE that successfully controlled the bleeding (No. 8). One patient died from tumor progression (No. 5), and one patient had no further episodes of bleeding (No. 6).

In this case series study, we reported that preprocedural CTA-guided TAE achieved technical success and clinical success in seven of eight patients with esophageal cancer bleeding, whereas only two patients experienced mild complications. These findings suggest that TAE may be a promising treatment option for esophageal cancer bleeding.

While endoscopic hemostasis remains the primary treatment for non-variceal upper gastrointestinal bleeding, its role is limited in initial bleeding control and rebleeding prevention, and lacks long-term efficacy (7, 9). Furthermore, evidence suggests that ‘emergent’ endoscopy (within 6 hours from the time of patient presentation) is associated with a higher in-hospital rate and 30-day mortality, especially in hemodynamically unstable patients (20); whereas ‘urgent’ endoscopy (within 12 hours from the time of patient presentation) could not prevent further bleeding or reduce mortality rate (21). In contrast, TAE offers a potential alternative for rapid hemostasis, while reducing blood transfusion requirements and avoiding complications associated with emergency surgery in patients who have failed endoscopic treatment or are unsuitable for endoscopy. Previous reports show that early clinical success rates for TAE in non-variceal upper gastrointestinal bleeding range from 44% to 94% (22). For arterial bleeding from esophageal lesions, early TAE clinical success rates are reported to be 77.8%-100% (12, 14, 23), aligning with the success rate observed in our study.

Esophageal cancer bleeding is often complex, with approximately 53.8% of cases due to tumor ulceration and 12.0% related to AEF. In tumor ulcer bleeding, the culprit bleeding vessel is typically the tumor-feeding artery; however, the tumor blood supply may be segmental and have multiple resources. Depending on the location of the tumor, feeding arteries may include the superior or inferior thyroid artery, bronchial artery, proper esophageal artery, posterior intercostal artery, left gastric artery, inferior phrenic artery, short gastric artery, or left hepatic artery, with a high degree of variability especially noted in the bronchial arteries (12, 24, 25). With the spread of the tumor, abnormal anastomoses can appear, further increasing both the number and variability of culprit blood vessels. Meanwhile, AEFs can be categorized into aortic AEF and non-aortic AEF (26), which can be managed with endovascular stent-graft placement and TAE, respectively (4). These complex, yet clinically important, variabilities in esophageal cancer bleeding underscore the need for accurate and comprehensive identification of culprit arteries and potential AEFs, thereby guiding interventional radiologists in devising intervention approaches and in choosing appropriate embolic materials (27, 28), ultimately improving the likelihood of successful hemostasis while minimizing complications. In addition, our findings suggest that recurrent bleeding may not be solely determined by the choice of embolic agents: the two patients treated with gelatin sponge alone experienced very late recurrence well beyond the absorption period, whereas three others treated with gelatin sponge combined with permanent agents still had early recurrence. Notably, patients without prior antitumor therapy tended to have longer recurrence-free intervals, suggesting that tumor necrosis and ulceration following such therapy may play a more important role in rebleeding than the use of temporary versus permanent embolic agents. However, this observation requires further validation in larger studies.

This study has several limitations. Firstly, the small sample size of only eight patients and the single-center design limits the generalizability of the findings and reduces the statistical power to draw robust conclusions. Secondly, the absence of a control group makes it difficult to directly compare CTA-guided TAE with alternative treatments, such as surgery or other endovascular interventions, thereby limiting the ability to evaluate its relative efficacy. Thirdly, as four out of eight patients died from rebleeding, the long-term effectiveness of TAE in preventing recurrent hemorrhage remains uncertain and warrants further investigation. Addressing these limitations in future research, such as conducting multi-center, prospective studies with larger cohorts and comparative analyses, would provide stronger evidence to support the use of this intervention in clinical practice.

CTA-guided TAE is a potentially effective option for achieving hemostasis in esophageal cancer bleeding, with a high rate of technical and clinical success and mild, manageable complications observed in this case series. Further studies with larger cohorts and longer follow-up periods are needed to evaluate its long-term efficacy and role in preventing recurrent bleeding.