1445P - Novel DNA methylation markers for early detection of cardia gastric adenocarcinoma and esophageal squamous cell carcinoma

Background

Cardia gastric adenocarcinoma (CGA) and esophageal squamous cell carcinoma (ESCC) remain major health burdens in China. Most of cases are diagnosed at advanced stages and carry a dismal prognosis. However, biomarkers for early detection of CGA and ESCC are still lacking. Here, we integrated methylome and transcriptome data and identified tumor-specific and tumor-shared DNA methylation markers for early detection of CGA and ESCC.

Methods

Infinium MethylationEPIC array was performed on 36 paired CGA and non-tumor adjacent tissues (NAT) in the discovery stage and differentially methylated CpG sites (DMCs) were identified between CGA/ESCC and NAT by combined analyses of in-house data and public database. Targeted pyrosequencing and quantitative real-time RT-PCR were performed on paired tumor and NAT from 50 CGA and 50 ESCC patients in the validation stage. An independent cohort of 438 CGA, ESCC, high- and low-grade dysplasia (HGD/LGD), and normal control biopsies was tested for selected DMCs using pyrosequencing.

Results

We identified and validated three CGA-specific, two ESCC-specific, and one tumor-shared DMCs, which were significantly hypermethylated with lower expression of their located genes in tumor compared with NAT samples. Using these DMCs, we developed a CGA-specific 4-marker panel (cg27284428, cg11798358, cg07880787, and cg00585116) that discriminated cardia HGD/CGA patients from cardia LGD/normal controls with the area under ROC curve (AUC) of 0.917, and an ESCC-specific 3-marker panel (cg14633892, cg04415798, and cg00585116) distinguishing esophageal HGD/ESCC with AUC of 0.865. Integrating cg00585116, age, and alcohol drinking, the tumor-shared model showed good discrimination for two cancer/HGD in the training set with AUC of 0.740, which was confirmed in the test set with AUC of 0.841.

Conclusions

Collectively, novel DNA methylation markers could differentiate CGA/ESCC and HGD from LGD and normal controls with promising accuracy. Our findings pave the way for targeted DNA methylation assays in future minimally invasive cancer screening methods.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.