Quantification of lower gastrointestinal bleed characteristics on computed tomography and relationship with findings during catheter angiography

Tuesday, March 7, 2023

3:54 PM – 4:03 PM

Location: Phoenix Convention Center, 222AB

Presenting Author(s)

RM

Robert R. Meyerhoff, MD, PhD

Resident, Diagnostic and Interventional Radiology
Duke University

Disclosure(s): No financial relationships to disclose

Author/Co-author(s)

CK

Colm Kelleher, MD

Assistant Professor, Radiology
Duke University
Brendan Cline, n/a

Assistant Professor of Radiology
Duke University Medical Center

Disclosure(s):

Robert R. Meyerhoff, MD, PhD: No financial relationships to disclose

Purpose: Many patients who present with lower gastrointestinal (LGI) bleeding and active contrast extravasation on computed tomography angiography (CTA) do not demonstrate active bleeds during catheter angiography (CA). Bleeding stops spontaneously in many patients, and CA represents an unnecessary procedural risk. The purpose of this study is to determine whether a subset of patients with slow extravasation on CTA are unlikely to have positive catheter angiography.

Materials and Methods: Retrospective review of 100 sequential patients from 11/3/2018 to 8/30/2022 who presented with LGI bleeding and positive CTA and underwent CA from was performed. Volume of extravasated contrast on arterial and portal venous phases was measured by a senior resident and attending abdominal radiologist using 3-dimensional segmentation in Visage Client version 7.1.17. Volume and time between imaging phases was used to calculate hemorrhage rate. Extravasation volume and rate were correlated with active contrast extravasation during CA. Patient age, bleed etiology, blood transfusion volume, hemoglobin, and time from CTA to CA were included for multivariate analysis.

Results: Majority of patients in our cohort were males presenting with diverticular bleeds. 55% of patients were found to have active extravasation on CA and of those 63% underwent coil embolization. Using a binary cutoff of bleed rate less than or greater than 0.5 mL/min, there was a significantly greater rate of positive CA with calculated hemorrhage rates greater than 0.5 mL/min than less than 0.5 mL/min (61% vs 25%, p=0.008). For absolute hemorrhage rate, there was a trend toward more likely positive CA with greater hemorrhage rate which was not statistically significant (p=0.1). Percent increase of bleed volume between arterial and portal venous phase for patients with positive CA was 297% versus 146% for patients with negative CA (p < 0.05). Additionally, decreased time between CTA and CA approached statistical significance in a multivariate analysis (p = 0.08).

Conclusion: While patients with a bleed rate less than 0.5 mL per minute were less likely to have positive CA, 25% of these patients still had a positive CA, suggesting that even patients with a slow bleed by CTA may warrant intervention. There was no correlation between absolute hemorrhage rate and positive CA. Interestingly, percentage increase in extravasated volume between arterial and venous phase was a predictor of CA. These findings may assist in identifying patients likely to have ceased bleeding by the time of CA and would benefit from conservative management rather than intervention.