OP010 Shallow whole-genome sequencing predicts the future cancer risk of low-grade dysplastic lesions arising in ulcerative colitis

Abstract

Introduction: The management of low-grade dysplasia (LGD) in ulcerative colitis (UC) is uncertain due to the variable risk of progression to colorectal cancer (CRC). Chromosomal copy number alterations (CNAs) are known to occur in colonic epithelial cells of UC patients who have developed CRC. The burden of CNAs in precursor LGD relative to high-grade dysplasia (HGD) and CRC has not been defined, and the correlation between LGD CNA burden and future CRC risk is unknown. Shallow whole genome sequencing1 is a novel, high-throughput, cost-effective technique for high-resolution CNA assessment in formalin-fixed, paraffinembedded tissue. We have successfully utilised this technique using as little as 500 picograms of relatively degraded DNA that has been extracted from archival tissue as much as 20 years old, at a cost of approximately €75 per tissue sample. Aims and Methods: To define the genomic changes that differentiate LGD from both HGD and CRC, we identified 19 UC proctocolectomy specimens with HGD and/or CRC, and analysed 77 neoplastic regions of interest (36 LGD, 34 HGD and 7 CRC). To determine the utility of shallow whole genome sequencing in predicting future CRC, we then analysed dysplastic tissue from 35 patients: 13 'progressor' patients with 27 LGD lesions who subsequently developed HGD and/or CRC a median 427 days later (IQR 213-777 days), and 22 'non-progressor' patients with 26 LGD lesions who remained free of HGD and CRC at least 5 years later. The two patient groups are matched for age, gender, disease duration and LGD location. Histological diagnosis was confirmed by two blinded pathologists. Shallow whole-genome sequencing (0.1x) was performed using a standardised pipeline for epithelial cell enrichment, DNA extraction, library preparation, next generation sequencing and bioinformatic analysis. Results: A median 12% of the genome of LGD tissue from proctocolectomy specimens showed CNAs (IQR 4-32%), compared to 23% in HGD/CRC (IQR 19-42%, p=0.003). Similarly, the number of CNA events was greater in HGD and CRC compared to LGD (p<0.001). Multiple CNAs involving key driver genes were more frequent in HGD/CRC compared to LGD (adjusted p values<0.05), including 8q gain (MYC loss, OR 17.2), 4q loss (OR 4.59) and 18q loss (DCC/SMAD4 loss, OR 4.15). Both the maximal total CNA burden and number of CNA events are greater in the LGD lesions of progressor patients compared to the LGD lesions of nonprogressor patients (p<0.01). A Kaplan-Meier survival analysis of the 35 grouped 'progressor' and 'non-progressor' patients demonstrates that patients in this cohort bearing LGD with the 25% greatest number of copy number alteration events are significantly more likely to develop future CRC/HGD than the remaining 75% of patients (HR 7.1, p<0.001). Conclusion: LGD lesions demonstrate a surprising genomic diversity in copy number alteration burden, with some LGD lesions bearing CNA profiles indistinguishable from HGD and CRC. Shallow whole genome sequencing has potential translational utility, by stratifying patients with LGD according to their future risk of progression to HGD and/or CRC.