Multi-omic cross-sectional cohort study of pre-malignant Barrett’s esophagus reveals structural variation and retrotransposon activity occur early in cancer evolution

An integrated study of frozen tissue from a large cohort of patients at the different stages of progression to oesophageal cancer. We describe how complex structural changes are already present in pre-cancerous lesions of the oesophagus. But creating a good clinical cohort is the hardest part.

Like Comment
Read the paper

Five years ago, when we were first discussing sequencing a few samples of Barrett’s oesophagus, a pre-malignant lesion which can progress to oesophageal adenocarcinoma, this seemed like a big step. At that time the focus was on cancer genomes and there had been minimal venture into pre-cancer whole genome sequencing. Yet Barrett’s oesophagus lends itself to the study of progression to cancer because patients with the disease undergo routine endoscopic surveillance with biopsies, monitoring for malignant transformation. Despite all the technological advances of the 21stcentury, oesophageal adenocarcinoma incidence continues to rise 1. In the UK, approximately 9000 cases of oesophageal cancer are diagnosed each year, of which just over half are adenocarcinomas. Overall, approximately 60% of patients have advanced, palliative disease at presentation 2, and this failure to diagnose oesophageal cancer in its earlier stages results in an overall 5-year survival of only 20% in Western countries 3. We still do not fully understand what the key changes are that drive the malignant transformation. We hypothesised that an integrated analysis with multiple sequencing modalities might elucidate the biological processes. Whole genome sequencing permits the analysis of large-scale structural rearrangements, which are known to be prevalent in oesophageal adenocarcinoma but have never been considered at the pre-invasive stage.

What we didn’t realise at that point what a challenge it would be to find good tissue for sequencing at the standard depth of 50x. For tumour tissue, when we look microscopically, often the whole biopsy consists of tumour cells, with contamination only from some stroma and inflammatory cells, and is said to have a high cellularity. However, Barrett’s samples often have only a small focus of dysplasia within a field of morphologically-benign cells. Sequencing good quality tissue, prior to the timepoint of cancer, which hadn’t been subjected to any thermal ablation therapy was of paramount importance to us, but these samples proved difficult to find. We have been collecting pre-cancer tissue in our lab for more than 20 years, with multiple samples from individual patients snap-frozen and carefully stored, but not previously cut or reviewed. So, creating our cohort of pre-cancer patients across the grades of disease was a slow process and involved a lot of time searching for samples in the -80 freezers. For every biopsy we had a section cut and stained and each month hundreds of slides were reviewed by our expert pathologists to find the samples that had captured the disease with a good cellularity. The dropout rate for samples was huge: frozen samples from dysplastic cases rarely captured the highest grade which the patient was known to have at that timepoint. In total, 1160 frozen biopsies from 315 patients were selected, cut and reviewed for inclusion. Each potential biopsy was then reviewed by a further two Pathologists to ensure a consensus. Frustratingly, there was then further sample dropout at the DNA extraction stage, where some biopsies were too small and did not yield sufficient DNA for WGS, or did not have a matched germline control available.

In our final cohort we had a very well-defined set of high-cellularity samples, for which we were completely certain not only of what we were sequencing but for which we had long-term follow-up data and the clinical characteristics of each patient. To us, this meticulous curation and level of clinico-pathological QC was key to the potential success of the study, on which a robust bioinformatical analysis could be performed and trusted.

With all this careful curation of the cohort we showed that large scale rearrangements of chromosomes, including complex ones such as chromosome shattering (chromothripsis), and mobile element insertions can occur whilst the tissue still looks benign microscopically. These events have not been widely described in pre-malignant disease. Integrating the expression and methylation data allowed us to look at the downstream effects of these alterations further, overall mapping out the key alterations seen in the progressive stages of transformation. We saw increased expression of genes related to cell cycle checkpoint, DNA repair and chromosomal instability; and epigenetic silencing of genes in Wnt signalling and cell cycle pathways. We infer from these observations that molecular features accumulate over time until the resulting genomic instability begins to increase exponentially and tips the balance to a ‘free-fall’ towards progression.

Incorporating a more quantitative assessment of the molecular alterations to the standard subjective pathological assessment, and tracking these alterations over time, could facilitate patient management decisions, ultimately with an aim to detect patients at risk of progression to cancer early. Overall, we hope that this strong cohort which will be an invaluable resource for the research community and for the future integration into progression models.

  1. Cancer Research UK. Oesophageal cancer statistics | Cancer Research UK. Cancer Research UK (2021). Available at: (Accessed: 4th February 2021)
  2. Varagunam, M. et al. National Oesophago-Gastric Cancer Audit 2017. (2017). Available at: (Accessed: 4th February 2021)
  3. Coleman, H. G., Xie, S. H. & Lagergren, J. The Epidemiology of Esophageal Adenocarcinoma. Gastroenterology 154, 390–405 (2018).

Annalise Katz-Summercorn

Clinical Research Fellow, MRC Cancer Unit, University of Cambridge

I'm a surgical registrar in the UK with an interest in oesophago-gastric cancer and genomics. I completed my PhD in the Fitzgerald lab at the MRC Cancer Unit, University of Cambridge.