Detailed Abstract
[Poster Presentation 14 - Basic Research (Basic Research)]
[PP 14-3] NOVEL BLOOD-BASED DNA METHYLATION MARKER PANEL FOR EARLY DIAGNOSIS OF PANCREATIC CANCER
Yeshong PARK 1, Kwangrok JUNG 2, Harin CHOI 3, Sol-Yi KIM 3, Hyojeong KOO 3, Hyejin OH 3, Jong Hoo LEE 3, Yoo-Seok YOON 1, Jin-Hyeok HWANG 2
1 Surgery, Seoul National University Bundang Hospital, Korea, 2 Internal Medicine, Seoul National University Bundang Hospital, Korea, 3 ., GENINUS, Inc., Korea
Background : Pancreatic ductal adenocarcinoma (PDAC) is a lethal malignancy typically diagnosed at advanced stage, and early detection of pre-symptomatic patients could offer a greater chance of curative treatment. The current study aimed to identify molecular markers based on tissue and liquid biopsy approaches to facilitate diagnosis of PDAC at resectable stage.
Methods : DNA methylation analysis was performed on 100 tissue samples including 70 PDAC patients and 30 healthy controls using methyl-seq panel. Based on hierarchical clustering of differentially methylated regions (DMR) by tissue data analysis, candidate DMRs were selected and further validated in 144 cell-free DNA (cfDNA) samples (97 PDAC, 47 control).
Results : Comparative tissue methylome analysis identified 9,294 DMRs encompassing 147,256 CpG sites in PDAC. In cfDNA samples, hierarchical clustering with 2Mb targeted panels selected 197 CpG sites comprising 46 DMRs as potential biomarkers. Machine learning-based feature selection was performed via two independent models, and the superior Random Forest model was selected. With a cut-off of feature importance ≥ 0.01, 190 CpG sites including 14 (38%) located in promoter regions were selected as final markers. Selected markers successfully discriminated PDAC from healthy controls (specificity 0.78, sensitivity 0.95, AUC 0.98). The selected markers, located on 6 genes (TRIM46, HEPN1, HNF4A, MIR3649, EFNA2, NR1I2), showed significantly different methylation level between normal and PDAC samples.
Conclusions : The currently identified panel of methylated cfDNA markers shows promising results to discriminate PDAC from healthy controls. Diagnostic power in combination with other screening biomarkers including carbohydrate antigen 19-9 should be further evaluated.
Methods : DNA methylation analysis was performed on 100 tissue samples including 70 PDAC patients and 30 healthy controls using methyl-seq panel. Based on hierarchical clustering of differentially methylated regions (DMR) by tissue data analysis, candidate DMRs were selected and further validated in 144 cell-free DNA (cfDNA) samples (97 PDAC, 47 control).
Results : Comparative tissue methylome analysis identified 9,294 DMRs encompassing 147,256 CpG sites in PDAC. In cfDNA samples, hierarchical clustering with 2Mb targeted panels selected 197 CpG sites comprising 46 DMRs as potential biomarkers. Machine learning-based feature selection was performed via two independent models, and the superior Random Forest model was selected. With a cut-off of feature importance ≥ 0.01, 190 CpG sites including 14 (38%) located in promoter regions were selected as final markers. Selected markers successfully discriminated PDAC from healthy controls (specificity 0.78, sensitivity 0.95, AUC 0.98). The selected markers, located on 6 genes (TRIM46, HEPN1, HNF4A, MIR3649, EFNA2, NR1I2), showed significantly different methylation level between normal and PDAC samples.
Conclusions : The currently identified panel of methylated cfDNA markers shows promising results to discriminate PDAC from healthy controls. Diagnostic power in combination with other screening biomarkers including carbohydrate antigen 19-9 should be further evaluated.
SESSION
Poster Presentation 14
Exhibition Hall 3/28/2025 2:20 PM - 3:00 PM