Gastrointestinal Imaging

Why Is CTA the Next Step for Upper GI Bleeding When Endoscopy Is Nondiagnostic?

·
Why Is CTA the Next Step for Upper GI Bleeding When Endoscopy Is Nondiagnostic?

It’s 2 AM, and the patient in the intensive care unit has had another episode of hematemesis. The gastroenterology team just finished an emergent esophagogastroduodenoscopy (EGD), and their post-procedure call is direct: “We confirmed it’s a nonvariceal upper gastrointestinal bleed, but we can’t localize the source. It’s obscure. What imaging do you want next?” This is a critical decision point where the right study can guide life-saving intervention, while the wrong one can delay care. This article provides a clinical workflow for this exact scenario: an adult with endoscopically confirmed but unlocalized nonvariceal upper gastrointestinal bleeding (UGIB). For this presentation, the American College of Radiology (ACR) Appropriateness Criteria rate CTA abdomen and pelvis without and with IV contrast as Usually Appropriate.

Who Fits This Clinical Scenario?

This guidance is specifically for an adult patient where two key conditions have been met: first, there is definitive evidence of a nonvariceal upper gastrointestinal bleed (e.g., observed melena, hematemesis, or a positive nasogastric lavage). Second, a standard upper endoscopy has been performed and confirmed ongoing or recent bleeding but failed to identify a discrete source. The bleeding is often described as “obscure” or “occult” in this context, meaning the cause is not apparent on initial endoscopic evaluation.

This workflow is distinct from several related but different clinical situations. This guidance does not apply if:

  • Endoscopy has not yet been performed. For patients with suspected but endoscopically unconfirmed UGIB, the initial workup is different.
  • A bleeding source was identified but endoscopic treatment failed. If, for example, a duodenal ulcer was seen but could not be controlled endoscopically, the next step is often direct intervention, not necessarily diagnostic imaging.
  • The patient has a negative endoscopy and is hemodynamically stable with no active bleeding. This represents a different diagnostic challenge, often managed with outpatient testing.
  • The bleeding is known or strongly suspected to be from varices due to portal hypertension, as this follows a separate management pathway.

What Diagnoses Are You Working Up in This Scenario?

When a standard EGD fails to find a source, the differential diagnosis shifts toward lesions that are small, intermittent, or located just beyond the endoscope’s reach. The goal of imaging is to uncover these challenging pathologies.

Aortoenteric Fistula: While uncommon, this is a life-threatening diagnosis that must be considered, especially in patients with a history of aortic surgery or an aortic aneurysm. It represents an abnormal connection between the aorta and the gastrointestinal tract, most often the duodenum. Patients may present with a “herald bleed” hours or days before catastrophic hemorrhage, making timely diagnosis critical.

Dieulafoy’s Lesion: This is a classic cause of obscure UGIB. It consists of an abnormally large, tortuous submucosal artery that erodes the overlying mucosa without a primary ulcer. Because the surrounding tissue is normal, the lesion can be nearly impossible to see on endoscopy unless it is actively bleeding at the exact moment of visualization. It can cause massive, intermittent hemorrhage.

Small Bowel Source: The bleeding may originate from the distal duodenum or proximal jejunum, an area just beyond the reach of a standard EGD. Common causes in this region include angiodysplasia (arteriovenous malformations), small bowel tumors like gastrointestinal stromal tumors (GISTs), or inflammatory conditions such as Crohn’s disease. Although technically not “upper GI,” bleeding from the proximal small bowel is a primary consideration when EGD is negative.

Hemosuccus Pancreaticus: A rare cause of UGIB, this refers to bleeding from the pancreatic duct into the duodenum. It is typically a complication of pancreatitis, a pancreatic pseudocyst, or a tumor that has eroded into a nearby artery (e.g., the splenic artery). The bleeding is often intermittent, making it difficult to diagnose.

Why CTA abdomen and pelvis without and with IV contrast Is the Recommended Study for This Presentation

For a patient with an endoscopically unlocalized upper GI bleed, the ACR designates CTA abdomen and pelvis without and with IV contrast as Usually Appropriate. This multiphase study is the cornerstone of modern evaluation because it is fast, widely available, and highly effective at both identifying the bleeding site and diagnosing the underlying cause.

The power of this study lies in its multiphase acquisition, which provides distinct diagnostic information at each stage:

  • Non-contrast phase: This initial scan is essential for establishing a baseline. It can identify high-density material that could be mistaken for contrast extravasation, such as surgical clips, ingested pills, or intramural hematoma.
  • Arterial phase: Timed to capture peak arterial enhancement, this is the most critical phase for identifying active bleeding. A “blush” of high-density contrast material actively extravasating into the bowel lumen is the hallmark finding, directly localizing the bleed.
  • Portal-venous phase: This later phase helps characterize underlying masses (like GISTs), evaluate for less common venous bleeding, and assess for complications like pseudoaneurysms or organ injury.

This comprehensive approach provides a vascular roadmap that is invaluable for guiding subsequent treatment, typically catheter-based embolization by an interventional radiologist. The radiation dose is a key consideration (ACR Relative Radiation Level ☢☢☢☢, corresponding to 10-30 mSv), but in the setting of acute, life-threatening hemorrhage, the diagnostic benefit substantially outweighs the risk.

Alternative studies are rated lower for this specific scenario. For instance, a tagged red blood cell (RBC) scan is rated May be appropriate. While highly sensitive for very slow bleeding, its poor spatial resolution often fails to pinpoint the exact location needed for intervention. Conventional arteriography is also rated May be appropriate but is now typically reserved as a therapeutic procedure performed after a positive CTA, rather than a primary diagnostic tool, due to its invasive nature.

The initial non-contrast phase is crucial for a correct interpretation. For more on the fundamentals of non-contrast abdominal CT technique, see our protocol guide: CT Abdomen/Pelvis Without Contrast (Renal Stone).

What’s Next After CTA abdomen and pelvis without and with IV contrast? Downstream Workflow

The results of the CTA will dictate the immediate next steps in patient management. The workflow typically branches into one of three paths.

If the CTA is positive for active extravasation: This is a clear indication for intervention. The patient should be immediately referred to Interventional Radiology for visceral arteriography and embolization. The CTA provides the interventionalist with a precise map of the bleeding vessel, which significantly increases the speed and success rate of the procedure.

If the CTA is negative for active extravasation but identifies a potential source: The study may not show an active blush but could reveal an underlying cause like a small bowel tumor, an aortic graft with adjacent fluid concerning for an aortoenteric fistula, or a pancreatic pseudoaneurysm. In these cases, the next step is tailored to the finding. A suspected tumor may require further evaluation with enterography or biopsy, while a suspected aortoenteric fistula is a surgical emergency requiring immediate vascular surgery consultation.

If the CTA is completely negative: This is a common and challenging outcome, often indicating that the bleeding was intermittent and had temporarily stopped during the scan. If the patient remains hemodynamically unstable or has further bleeding, the next step may be a tagged RBC scan to detect a very slow bleed. If the patient stabilizes, the workup may proceed to video capsule endoscopy or deep enteroscopy to evaluate the small bowel more thoroughly. This situation aligns more closely with the ACR scenario for “Nonvariceal upper gastrointestinal bleeding; negative endoscopy.”

Pitfalls to Avoid (and When to Get Help)

In this high-stakes clinical scenario, several common pitfalls can compromise diagnostic accuracy and delay treatment. Be mindful to:

  • Avoid single-phase studies. Ordering a “CT abdomen with contrast” without specifying a multiphase bleeding protocol is a frequent error. The non-contrast and arterial phases are essential and must be explicitly requested.
  • Image during active bleeding. The sensitivity of CTA for detecting extravasation drops significantly if the patient is not actively bleeding. Coordinate with the clinical team to time the scan, if possible, when the patient shows signs of active hemorrhage (e.g., drop in blood pressure, fresh hematemesis).
  • Communicate with the radiologist. A brief call to the reading radiologist before the scan to convey the specific clinical question—”looking for an obscure UGIB source post-negative EGD”—ensures the correct protocol is used and the images are interpreted with high suspicion.

If the CTA is positive for a complex vascular anomaly or an aortoenteric fistula, immediate consultation with both Interventional Radiology and Vascular Surgery is warranted.

Related ACR Topics and Tools

Navigating imaging choices for gastrointestinal bleeding requires familiarity with the complete ACR Appropriateness Criteria and the technical aspects of the recommended studies. The following resources can help refine your decision-making for this and related scenarios.

Frequently Asked Questions

Why not go straight to conventional arteriography instead of CTA?

Conventional arteriography is invasive and carries risks such as vessel dissection and hematoma. Modern CTA is non-invasive, faster, and has excellent sensitivity for detecting arterial bleeding rates. CTA also provides a comprehensive view of the entire abdomen, which can identify non-vascular causes of bleeding (like tumors) that would be missed on arteriography. Arteriography is now primarily used as a therapeutic tool after the bleeding site has been localized by CTA.

What is the minimum bleeding rate that a multiphase CTA can detect?

Multiphase CTA can reliably detect active arterial bleeding at rates of approximately 0.3 to 0.5 mL/minute. This is significantly more sensitive than conventional arteriography but less sensitive than a tagged red blood cell (RBC) scan, which can detect rates as low as 0.1 mL/minute. However, the superior spatial resolution of CTA makes it the preferred initial imaging test.

What if my patient has renal insufficiency and cannot receive IV contrast?

This is a significant challenge. In a patient with severe renal dysfunction where IV contrast is contraindicated, the diagnostic algorithm must change. A tagged RBC scan becomes a more viable option as it does not require iodinated contrast. In some cases, a non-contrast CT may be performed to look for secondary signs of pathology, but its utility is very limited. This complex situation requires a multidisciplinary discussion between the primary team, nephrology, and radiology.

Does this guidance apply to lower gastrointestinal bleeding (LGIB)?

While the recommended imaging modality (multiphase CTA) is the same for acute LGIB, the clinical context and preceding workup are different. For LGIB, colonoscopy is the primary diagnostic tool, analogous to EGD for UGIB. This article’s workflow is specific to the scenario where an upper endoscopy has been performed and was nondiagnostic.

Is there a role for MR enterography in this acute setting?

No. MR enterography is rated as Usually Not Appropriate by the ACR for this acute scenario. While it is an excellent tool for evaluating the small bowel for inflammation or tumors in a stable, non-bleeding patient, it is too slow and less sensitive for detecting active arterial extravasation compared to CTA. Its use is reserved for the outpatient workup of obscure GI bleeding after acute causes have been ruled out.

Reviewed by Pouyan Golshani, MD, Interventional Radiologist — May 26, 2026