Magnetic resonance angiography versus endoscopy for the assessment of gastroesophageal varices in biliary atresia.

This study was undertaken to compare magnetic resonance angiography (MRA) and gastrointestinal tract endoscopy (ENDO) for detecting varices in postoperative biliary atresia (BA) patients. Thirty-four BA patients were divided into groups according to age and liver function; group 1: 1-4 years old; group 2: 5-9 years old; group 3: over 10 years old; group A: normal liver function; and group B: moderate liver dysfunction. All subjects had MRA and ENDO. ENDO showed esophageal or gastric varices in 16 of 34 subjects. MRA depicted gastroesophageal vessels in 12 of 16 cases with varices on ENDO (sensitivity: 75%). Two subjects appeared to have varices on MRA that could not be confirmed on ENDO. Sixteen subjects had no varices on ENDO and normal MRA (specificity: 88.9%). Sensitivities and specificities of MRA for detecting varices were 20 (1/5) and 100% (6/6) in group 1 (n = 11), 100 (6/6) and 66.7% (4/6) in group 2 (n = 12), 100 (5/5) and 100% (6/6) in group 3 (n = 11), 100 (8/8) and 100% (9/9) in group A (n = 17), 50 (4/8) and 77.8% (7/9) in group B (n = 17), respectively. From these findings, MRA was not accurate in BA patients less than 10 years old or with moderate liver dysfunction. Therefore, periodic endoscopic examination is recommended for the accurate assessment and follow-up of varices in postoperative BA patients.