RSNA 2009 

Abstract Archives of the RSNA, 2009


SSC06-02

Computed Tomography Evaluation of Imaging Features That Predict Variceal Hemorrhage

Scientific Papers

Presented on November 30, 2009
Presented as part of SSC06: Gastrointestinal (Interpretation of Hepatobiliary Findings)

 Research and Education Foundation Support

Participants

Katherine To'o MD, Presenter: Nothing to Disclose
Rizwan Aslam MBChB, Abstract Co-Author: Nothing to Disclose
Benjamin M. Yeh MD, Abstract Co-Author: Nothing to Disclose
Mujtaba Ali BS, Abstract Co-Author: Nothing to Disclose
Judy Yee MD, Abstract Co-Author: Research grant, General Electric Company
Ma Somsouk MD, Abstract Co-Author: Nothing to Disclose

PURPOSE

To identify computed tomography (CT) findings of portal hypertension that best predict variceal hemorrhage (VH).

METHOD AND MATERIALS

We retrospectively identified 28 patients with advanced liver disease who had an outpatient CT scan prior to sustaining a VH and no history of variceal band ligation or suspected hepatocellular carcinoma (HCC).  A control group was comprised of similar patients with advanced liver disease (n=56) who underwent a CT and upper endoscopy within 90 days of each other but had no history of VH.  Two readers (R1 and R2) independently reviewed each CT scan to evaluate liver nodularity and record findings of portal venous hypertension.

RESULTS

The three most predictive CT findings for VH were the presence of visible esophageal varices (EV) (OR 73.8 and 25.3 for R1 and R2, respectively, p<0.001 for both), any EV tortuosity (OR ∞ and 17.3, respectively, p<0.001 for both), and the presence of an EV with a diameter > 3 mm (OR 275 and 25.3, respectively, p<0.0001 for both).  The mean maximal EV diameter was significantly larger in patients with VH than without (5.0 vs. 0.6 mm for R1, 7.0 vs. 1.7 mm for R2, p<0.0001 for both).  The presence of paraesophageal varices, recanalized paraumbilical veins, liver nodularity, and coronary vein enlargement were less strongly associated with VH.  Portal vein diameter, ascites, and spleen size were not statistically different between cases and controls.  In the multivariate analysis, the maximal EV diameter was the only independent predictor of VH (OR 2.19 and 1.48 per millimeter increase in diameter for R1 and R2, respectively, p<0.05 for both).  Good interobserver agreement existed for number of EV (κω 0.73) and EV tortuosity (κω 0.79) and moderate agreement was present for largest esophageal varix diameter (κω 0.65).  Additionally, endoscopic grading and CT measurement of maximal EV diameter showed moderate agreement (κω 0.68).  

CONCLUSION

Of the CT findings of portal hypertension in patients with advanced liver disease, a maximal esophageal varix diameter of >3 mm was most strongly associated with the incidence of variceal hemorrhage.

CLINICAL RELEVANCE/APPLICATION

In addition to screening for HCC, CT may help identify patients with advanced liver disease who are at increased risk for VH and may benefit from prophylactic beta blockade.

Cite This Abstract

To'o, K, Aslam, R, Yeh, B, Ali, M, Yee, J, Somsouk, M, Computed Tomography Evaluation of Imaging Features That Predict Variceal Hemorrhage.  Radiological Society of North America 2009 Scientific Assembly and Annual Meeting, November 29 - December 4, 2009 ,Chicago IL. http://archive.rsna.org/2009/8006563.html