Abstract Background Pancreatic cancer (PAC) lacks methods for screening, patient stratification, resulting in late diagnosis and inadequate treatment of patients. Cell-free DNA (cfDNA) methylation patterns are signals representing real-time cancer characteristics such as early presence, tissue, and cell of origin. The aim of this study is to provide a better patient stratification based on differential patterns on the combined methylation and gene expression results in PAC and normal samples and provide proof-of-concept for early detection. Method Differential methylation region (DMR) analysis was performed on 180 TCGA PAAD and normal tissue samples using paired methylation and expression data. Validation was done on 143 Infiltrating duct carcinomas (IDC), (stage I, II, II), and 4 normal samples and the 4 probe clusters (FPC) were defined based on their methylation status: HM = high methylation, IM = intermediate methylation, LM = low methylation and MM = minimum methylation. 3-year and 5-year survival analysis with Kaplan-Meier was done. Gene ontology (GO) and Biological Pathway analyses (PA) were performed on DMRs of FPC. Integration analysis on paired methylation and expression data and the limma analysis was performed for significant genes in 5000 bp vicinities of the selected DMRs. Finally, selected 23 DMRs, were targeted in plasma cfDNA from 9 PAC patients (stage I-III) and 20 matching control samples. Results The survival analysis of FPC, shows the highest probability difference between HM and MM groups with p-values of 0.018 and 0.029 for 3-year and 5-year survival, respectively. Variability analysis shows 5740 significant CpG sites and 1135 significant DMRs comparing IDC tumor samples to controls. GO and PA on DRMs of FPC, indicated the most PAC specific pathways in the group of probes with MM status. Integration analysis on the methylation and mutation data of the selected genes of GO analysis shows potential to discriminate between normal and PAC patients and in between HM and MM groups and showed smaller mutation burden in MM and LM sample groups. Comparing MM to HM groups of patients, shows the highest significance in 5 downregulated genes in 5000 bp vicinity from Hypermethylated probes and 13 upregulated genes and 425 regions in 5000 base pairs vicinity from 170 hypomethylated probes. Survival analysis on the selected genes showed a very high correlation between the expression of the genes like SIX2, DNMT3A, ZFP91 and POR and the survival probability on FPC. Additionally, it has been shown that the selected DMRs, show good individual performance of AUC >0.95 in the in-house analysis of plasma cfDNA targeted data from 9 PAC patients (stage I-III) and 20 matching control samples. Conclusion Methylation markers carry distinct biological signals and combined with gene expression differentiate the PAC patient groups, showing potential of the methylation signals for better patient stratification outcomes for improved survival. High accuracy for cancer detection was observed for early-stage cancers when compared to cfDNA from matching controls. Citation Format: Pardis Saadatmand, Urban Kunej, Kristi Kruusmaa. Combining the methylation and gene expression contiguous signals leads to better pancreatic cancer patient stratification and early detection [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Pancreatic Cancer; 2023 Sep 27-30; Boston, Massachusetts. Philadelphia (PA): AACR; Cancer Res 2024;84(2 Suppl):Abstract nr C006.
Read full abstract