Konda, Vani

Abstract Aberrant DNA methylation and copy-number alterations (CNA) drive Barrett’s esophagus progression to esophageal adenocarcinoma; however, their combined utility for early detection is unclear. We aimed to identify and validate methylated DNA markers (MDM) and CNAs to distinguish esophageal adenocarcinoma/high-grade dysplasia (HGD) from nondysplastic Barrett’s esophagus (NDBE). In this multiphase, multicenter study, we discovered and validated MDMs and quantified CNAs utilizing whole-genome methylation sequencing of esophageal brushings. DNA biomarkers identified from discovery were further validated in independent patients with paired esophageal brushing and swallowed capsule sponge samples. MDMs were filtered against a reduced representation bisulfite sequencing dataset obtained from independent tissue samples to advance only concordant candidates. CNA burden was quantified using ichorCNA-derived aneuploidy scores (AS). Two hundred MDMs discovered in HGD (N = 18) and esophageal adenocarcinoma (N = 18) versus NDBE brushing samples (N = 18) were tested in independent samples (N = 146). A 52-MDM panel achieved a cross-validated AUC of 0.88 [95% confidence interval (CI), 0.82–0.95]; the addition of AS improved discrimination of HGD/esophageal adenocarcinoma from NDBE to 0.91 (95% CI, 0.86–0.97) AUC. At 80% specificity, the combined model detected 93% of esophageal adenocarcinoma and 88% of HGD cases. In paired capsule sponge samples, a 58-MDM panel achieved a cross-validated AUC of 0.77 (95% CI, 0.66–0.88); a combined 58-MDM and AS model achieved AUC 0.80 (95% CI, 0.7–0.9). MDMs and AS discerned HGD/esophageal adenocarcinoma from normal esophagus/NDBE in endoscopic brushing and capsule sponge samples. This approach may improve Barrett’s esophagus surveillance. Prevention Relevance: This study demonstrates that combining epigenetic and genomic biomarkers across minimally invasive sampling methods can accurately distinguish HGD/esophageal adenocarcinoma from nondysplastic Barrett’s esophagus, offering promising, less invasive strategies to improve BE surveillance and enable endoscopic therapy for esophageal adenocarcinoma prevention and treatment.

INTRODUCTION: The risk of progression to esophageal adenocarcinoma in Barrett's esophagus (BE) increases with advancing degrees of dysplasia. There is a critical need to improve the diagnosis of BE dysplasia, given substantial interobserver variability and overcalls of dysplasia during manual community pathologist reads. We aimed to externally validate a previously cross-validated BE dysplasia diagnosis deep learning model (BEDDLM) that predicts dysplasia grade on whole slide images (WSIs). METHODS: We digitized nondysplastic BE (NDBE), low-grade (LGD), and high-grade dysplasia (HGD) histology slides from 3 external academic centers. A consensus read by 2 expert study pathologists was used as the criterion standard. Slide stain characteristics were normalized using cycle-generative adversarial networks. WSIs were assessed by BEDDLM using an ensemble approach, combining a “You Only Look Once” model followed by a ResNet101 classifier model. RESULTS: We included 489 WSIs. Consensus histopathology revealed 232 NDBE, 117 LGD, and 140 HGD WSIs. The mean age (SD) was 66.9 (11.4) years; 413 (84.7%) were men. Using the BEDDLM ensemble model, sensitivity and specificity for NDBE were 73.3% (95% confidence interval [CI]: 67.09%–78.85%) and 93.4% (95% CI: 89.62%–96.10%); for LGD, 84.6% (95% CI: 76.78%–90.62%) and 80.6% (95% CI: 76.26%–84.54%); and for HGD, 80.7% (95% CI: 73.19%–86.89%) and 94.8% (95% CI: 91.97%–96.91%), respectively. The F1 score was 0.81, 0 0.69, and 0.83 for NDBE, LGD, and HGD, respectively. DISCUSSION: Our externally validated deep learning model demonstrates substantial accuracy for the diagnosis of BE dysplasia grade on WSIs.