In vivo endomicroscopy improves detection of Barrett's esophagus-related neoplasia: a multicenter international randomized controlled trial (with video)

Abstract

Background: Confocal laser endomicroscopy (CLE) enables in vivo microscopic imaging of the GI tract mucosa. However, there are limited data on endoscope-based CLE (eCLE) for imaging Barrett's esophagus (BE).

Objective: To compare high-definition white-light endoscopy (HDWLE) alone with random biopsy (RB) and HDWLE + eCLE and targeted biopsy (TB) for diagnosis of BE neoplasia.

Design: Multicenter, randomized, controlled trial.

Setting: Academic medical centers.

Patients: Adult patients with BE undergoing routine surveillance or referred for early neoplasia.

Intervention: Patients were randomized to HDWLE + RB (group 1) or HDWLE + eCLE + TB (group 2). Real-time diagnoses and management plans were recorded after HDWLE in both groups and after eCLE in group 2. Blinded expert pathology diagnosis was the reference standard.

Main outcome measurements: Diagnostic yield, performance characteristics, clinical impact.

Results: A total of 192 patients with BE were studied. HDWLE + eCLE + TB led to a lower number of mucosal biopsies and higher diagnostic yield for neoplasia (34% vs 7%; P < .0001), compared with HDWLE + RB but with comparable accuracy. HDWLE + eCLE + TB tripled the diagnostic yield for neoplasia (22% vs 6%; P = .002) and would have obviated the need for any biopsy in 65% of patients. The addition of eCLE to HDWLE increased the sensitivity for neoplasia detection to 96% from 40% (P < .0001) without significant reduction in specificity. In vivo CLE changed the treatment plan in 36% of patients.

Limitations: Tertiary-care referral centers and expert endoscopists limit generalizability.

Conclusion: Real-time eCLE and TB after HDWLE can improve the diagnostic yield and accuracy for neoplasia and significantly impact in vivo decision making by altering the diagnosis and guiding therapy. (

Clinical trial registration number: NCT01124214.).

Keywords: BE; Barrett's esophagus; CLE; CPT; Current Procedural Terminology; ECA; HDWLE; HGD; NBI; RB; TB; WLE; confocal laser endomicroscopy; eCLE; endoscope-based CLE; esophageal adenocarcinoma; high-definition white-light endoscopy; high-grade dysplasia; narrow-band imaging; pCLE; probe-based CLE; random biopsy; targeted biopsy; white-light endoscopy.

Figure 1

Study Flow Diagram

Figure 2

A: Comparison of the total number of mucosal biopsies obtained per patient from Barrett’s esophagus patients (HDWLE alone versus HDWLE + CLE). HDWLE = high definition white light endoscopy, CLE= confocal laser endomicroscopy. P values shown for comparisons between groups were obtained using Student’s t test. B: Boxplots comparing the average number (with interquartile ranges) of mucosal biopsies obtained per patient from Barrett’s esophagus patients (HDWLE alone versus HDWLE + CLE). P values obtained for all comparisons were significant (p<0.0001 for all patients, p=0.0001 for patients with neoplasia, p<0.0001 for patients undergoing routine surveillance).

Figure 3

A: Comparison of the diagnostic yield for neoplasia using a per biopsy analysis for HDWLE with random biopsy approach (n=580 specimens) versus HDWLE + CLE and targeted biopsy approach (110 specimens). HDWLE + RB = high definition white light endoscopy with random biopsy; HDWLE+CLE+TB= high definition white light endoscopy with confocal laser endomicroscopy and targeted biopsy. P values for comparisons between groups were obtained using Fisher’s exact test. B: Comparison of the diagnostic yield for neoplasia using a per patient analysis for HDWLE with random biopsy approach (n=580 specimens) versus HDWLE + CLE and targeted biopsy approach (110 specimens). HDWLE + RB = high definition white light endoscopy with random biopsy; HDWLE+CLE+TB= high definition white light endoscopy with confocal laser endomicroscopy and targeted biopsy. P values for comparisons between groups were obtained using Fisher’s exact test.

Figure 4

High resolution white light endoscopy (HDWLE) images of Barrett’s esophagus with high grade dysplasia not detected by 2 prior endoscopies showing no suspicious neoplastic lesion (A and B). Multiple optical biopsy sites in BE imaged with confocal laser endomicroscopy (CLE) are shown with numbers (B). CLE images from sites 6 and 7 in image B after showing intestinal metaplastic glands with loss of normal mucosal pattern, darkening of epithelial cells, lack of goblet cells, glandular distortion, and a cribriform pattern, consistent with high grade dysplasia (C and D). Note the whitish squamous islands seen in the endoscopic and CLE images (arrows), with corresponding CLE image C suggesting partially buried dysplastic BE glands. In vivo diagnosis of neoplasia led to a decision to perform endoscopic mucosal resection (EMR) at the baseline procedure. Pathological examination of the EMR specimen confirmed the diagnosis of high grade dysplasia with partial pseudoregression.