Tu1074 Videocapsule Versus Enteroscopy in Patients With Obscure GI Bleeding: Results of a Randomized Controlled Trial

Abstract

Background: Videocapsule endoscopy (VCE), when compared to enteroscopy (E), is associated with increased detection of mucosal lesions in obscure gastrointestinal bleeding (OGB). However, few randomized trials have compared these two modalities, and data on clinical relevance are lacking. The aim of this study was to compare VCE with E in terms of diagnostic yield and subsequent clinical outcome. Methods: We randomly allocated adult patients with occult bleeding (iron deficiency anemia >6 months needing blood transfusion or iron supplementation with repeated positive stool occult blood testing) or overt bleeding (≥1episode of melena/hematochezia within 7 days of presentation and a hemoglobin drop ≥15 g/dL) to VCE or E after a negative initial work-up (single or repeated gastroscopy and colonoscopy examinations with or without small bowel radiology). The main outcomes were recurrence and persistence of bleeding using standardized criteria defined a priori over 1year. Crossovers were permitted after the main outcome was reached. Results: 79 patients (68.5±14.5yrs, 36.7% female) were included over a 5-year period. 40 subjects were allocated to VCE and 39 to E (of push type). Overall, 76.0% presented with overt bleeding. At baseline, 24.1% were taking ASA, 11.4% an anticoagulant, and 2.5% an NSAID; 76.7% of patients had a nuclear scan, and 37.0% an angiography. No significant differences were noted in baseline characteristics between the two groups. At least one lesion was detected in 72.5% of the VCE group compared to 48.7% in the E group (p=0.03); they were believed to be the source of bleeding in 79.3% and 35.0% (p=0.002), respectively. Although detection rates were similar for gastric and duodenal findings, VCE performed better in the first and second parts of the jejunum (50% vs. 10.3% P=0.0001, and 40% vs. 2.6% P<0.0001). No disparities were observed in terms of the nature of the lesions found between the two modalities. Mean duration of follow-up was 289.0±118.8 days, with 50.0% requiring a blood transfusion (42.5% VCE vs. 57.9% E, with a mean of 2.9 ± 5.0 units (VCE: 2.8±4.3 units vs. E: 3.0±5.6 units, P=NS)). Hospitalization for bleeding or continued bleeding occurred in 40.0% of VCE and in 53.9% of E patients (P=NS) (mean stay: 11.0±11.9 and 18.5±28.2 days respectively, P=NS). Crossovers occurred more frequently from E to VCE than from VCE to E (48.7% vs. 22.5%, P=0.015). Conclusions: VCE increases diagnostic yield in OGB when compared to E, especially when the bleeding source is in the jejunum. However, subsequent impact on clinical care was not significantly better in this trial although trends favored the VCE group. More crossovers were noted from E to VCE than conversely. Overall, our data support the use of VCE over E following an initial work-up in patients with OGB.