Understanding the molecular age of Barrett's oesophagus in a population-representative sample of patients

Abstract Background The incidence of oesophageal adenocarcinoma (OAC) increases dramatically with patient age but only a small proportion of patients with diagnosed Barrett’s oesophagus (BO), the precursor to OAC, will develop dysplasia and/or cancer. Beyond chronological age, biomarkers of progression that capture biological aging offer largely untapped potential for objectively identifying BO patients at highest risk of progression, who could undergo personalised surveillance at shorter intervals. We have developed computational tools to determine tissue-specific aging using genome-wide methylation data as a “molecular clock” for estimating patient-specific BO dwell times at the time of incident diagnosis that cannot be clinically measured by other means. Methods Using the population-based Northern Ireland BO register in a retrospective study, we have identified 46 non-dysplastic BO patients who have 2-4 serial endoscopic biopsies each, and have not progressed to OAC (age range 29-77 years). FFPE biopsies for 10 age-matched patients who had prevalent HGD/OAC at index BO diagnosis were also retrieved. DNA has been extracted, quantified using fluorescence, quality checked through qPCR, and prepared for Illumina EPIC methylation arrays. We created a Python package called “MethylDrift” to determine genome-wide aging rates in patient data. Model outputs are used in the molecular clock for BO tissue age. Results We used MethylDrift to quantify aging rates in both cross-sectional data (population-level epigenetic drift) and longitudinal data within the same patients to obtain individual aging rates. Computational analyses using our previously developed Bayesian framework for the BO molecular clock will be applied to estimate the molecular age of BO in patients, i.e., how long the patient has been living with BO since onset of metaplasia. Results will be compared between age groups, birth cohorts, sex, and importantly between dysplastic and non-dysplastic BO to evaluate biomarker potential. Data analysis is ongoing, and the final results will be presented at the meeting. Conclusions Our results from this nested case-control study demonstrate feasibility and generate pilot data on molecular age as a proxy of BO duration at the time of incident diagnosis, in a large population-based registry of patients with BO. This will inform our computational tools for determining biological aging and can be applied in future work to investigate progression risk according to molecular age. Ultimately, this biomarker could inform surveillance frequency for BO patients, enable earlier detection of neoplastic progression, leading to improved patient outcomes and optimal distribution of limited endoscopy capacity for surveillance.