| Summary: | <p>Despite immense efforts, the prognosis of pancreatic ductal adenocarcinoma (PDAC) remained very poor. The main reasons include the lack of early-stage symptoms and effective screening programmes, leading to delayed clinical presentations, with only 15-20% of patients being diagnosed with localised resectable tumours. However, ~40% of these patients develop distant recurrence within 12 months, suggesting that while clinically identified as localised, a subset of the patients likely had micro-metastatic lesions and neoadjuvant chemotherapies might have been the better option. There are strong demands for improving prognostic accuracy and this project aims to distinguish PDAC patients with metastasis-prone tumours from those with truly localised resectable tumours via DNA methylation (5mC) and hydroxymethylation (5hmC) signatures.</p>
<p>Starting with raw 5hmC & 5mC data, pairwise differential analyses were performed to identify significantly differentially hydroxy/methylated CpGs - D(h)MPs and significantly differentially hydroxy/methylated regions - D(h)MRs. ChAMP detected 3 DhMPs being significantly hypo-hydroxymethylated in patients with distant recurrence, whereas 21 DhMRs were significantly differentially hydroxymethylated. Similarly, 35459 DMPs and 551 DMRs were detected. Scorings were performed for subsequent unsupervised hierarchical clustering analyses to check how well the identified signatures distinguish between the two groups. The results showed that 5(h)mC profiles indeed have the power to separate patients into their prognostic groups. To gain a better understanding of the underlying mechanisms, functional enrichment analysis was performed for the top signatures. As validation, RT qPCR of selected genes in classical & squamous PDAC cell lines were performed; in which 6 of the candidate genes have a profound transcription pattern, suggesting an overlap between our hydroxy/methylomic-driven stratification and the transcriptomic-driven classical/squamous stratification.</p>
<p>In conclusion, this preliminary study showed that 5(h)mC signatures have the potential to improve prognostic accuracy. Most importantly, this approach can be adapted for other scientific questions, bringing new opportunities to combat PDAC and different diseases.</p>
|
|---|