The utility of a multi-gene methylation panel to risk-stratify patients with Barrett's oesophagus after radiofrequency ablation

Introduction

Radiofrequency ablation (RFA) is the treatment of choice for dysplastic Barrett’s oesophagus (BO), however recurrence occurs in up to 25% of cases, usually at the gastro-oesophageal junction (GOJ). We have shown that a multi-gene methylation panel (ZNF345, TFPI2 and ZNF569) can differentiate BO from normal GOJ epithelium. This study aims to determine whether this methylation panel can predict risk of relapse post-RFA.

Methods

This is a prospective cohort study on patients who achieved endoscopic remission post-RFA. Patients were eligible if they had 1) no endoscopic evidence of BO or islands ≥5mm, 2) no suspicious dysplastic lesion at GOJ, and 3) no histological evidence of oesophageal intestinal metaplasia (IM) at first post-RFA follow-up (GOJ IM allowed). Patients received endoscopy at baseline, 6 and 12 months. Methylation score (Meth-score) was assessed by Methylight assay on random or targeted GOJ biopsies. The primary outcome was the correlation of Meth-score and histological outcome. The secondary outcome was the correlation of total baseline methylation level and dysplastic recurrence at follow-up.

Results

We included 56 patients with a total of 90 endoscopies at various timepoints of follow-up. Mean age, circumferential and maximum length of BO pre-RFA was 68.6 years (standard deviation, SD 7.5), 2.2cm (SD 3.0) and 4.4cm (SD 3.0), respectively. Seven (13%), 28 (50%) and 21 (38%) patients had pre-RFA low-grade dysplasia (LGD), high-grade dysplasia (HGD) and intramucosal carcinoma (IMC), respectively. Meth-score comparing patients with gastric metaplasia (GOJ-GM, n=60) and intestinal metaplasia (GOJ-IM, n=25) was 1.2% vs 36.9%, p<0.001, and between GOJ-GM and dysplasia (n=5) was 1.2% vs 43.4%, p<0.001. When assessing baseline Meth-score stratified by highest histology at any time during follow-up, patients with non-dysplastic recurrence (n=50) had a lower total baseline methylation compared to dysplastic recurrence (n=6), with Meth-score of 8.5% vs 39.0%, p=0.03.

Conclusion

A multi-gene methylation panel can discriminate patients with GM, IM and dysplasia on GOJ biopsies. Patients with stable histological remission (no IM or dysplasia at follow-up) had low baseline methylation levels. This biomarker panel can risk-stratify patients so that stricter surveillance is offered only to those with higher total methylation levels who are more likely to develop recurrence.