Detection of dysplasia in Barrett's esophagus with in vivo depth-resolved nuclear morphology measurements

Abstract

Background & aims: Patients with Barrett's esophagus (BE) show increased risk of developing esophageal adenocarcinoma and are routinely examined using upper endoscopy with biopsy to detect neoplastic changes. Angle-resolved low coherence interferometry (a/LCI) uses in vivo depth-resolved nuclear morphology measurements to detect dysplasia. We assessed the clinical utility of a/LCI in the endoscopic surveillance of patients with BE.

Methods: Consecutive patients undergoing routine surveillance upper endoscopy for BE were recruited at 2 endoscopy centers. A novel, endoscope-compatible a/LCI system measured the mean diameter and refractive index of cell nuclei in esophageal epithelium at 172 biopsy sites in 46 patients. At each site, an a/LCI measurement was correlated with a concurrent endoscopic biopsy specimen. Each biopsy specimen was assessed histologically and classified as normal, nondysplastic BE, indeterminate for dysplasia, low-grade dysplasia (LGD), or high-grade dysplasia (HGD). The a/LCI data from multiple depths were analyzed to evaluate its ability to differentiate dysplastic from nondysplastic tissue.

Results: Pathology characterized 5 of the scanned sites as HGD, 8 as LGD, 75 as nondysplastic BE, 70 as normal tissue types, and 14 as indeterminate for dysplasia. The a/LCI nuclear size measurements separated dysplastic from nondysplastic tissue at a statistically significant (P < .001) level for the tissue segment 200 to 300 μm beneath the surface with an accuracy of 86% (147/172). A receiver operator characteristic analysis indicated an area under the curve of 0.91, and an optimized decision point gave 100% (13/13) sensitivity and 84% (134/159) specificity.

Conclusions: These preliminary data suggest a/LCI is accurate in detecting dysplasia in vivo in patients with BE.

Figure 1

A) High-level a/LCI system diagram. Inset shows scale of probe tip compared to a U.S. dime. B) Image showing characteristic mark left by a/LCI probe following deployment, indicated by the white arrow. C) Detail of the a/LCI probe tip. Light is delivered as a collimated beam to the tissue. Scattered light is collected across the face of the fiber bundle for transport back to the a/LCI system.

Figure 2

Typical a/LCI data. A) Angle-resolved depth scan of light scattered from tissue. Lighter shades of gray indicate increased amount of scattered light. B) A-scan indicating depth increments used for processing. Tissue layers are labeled and gray bar indicates basal layer. C) Example angular scans for three tissue types pictured (solid line) with best fit Mie theory solutions (dashed line) and size indicated.

Figure 3

Scatter plot with each biopsy plotted as a function of its nuclear size and density, as measured by the a/LCI system, and categorized by its pathological diagnosis. Dashed black line indicates an optimized decision line between the two populations, resulting in 100% sensitivity and 84% specificity.

Figure 4

The receiver operator characteristic (ROC) curve for the depth segment between 200-300μm indicating relationship between sensitivity and specificity for varied decision lines using nuclear diameter as a discriminator. The gray area indicates the area under the curve (AUC = 0.91). ROC curves for depth segments from 0-100μm and 100-200μm (not shown) have an AUC of 0.58 and 0.52 respectively.

Figure 5

Nuclear size for each of the tissue layers segregated by pathological diagnosis.