| Background:Colorectal cancer(CRC)and Gastric cancer(GC)are the five most common and lethal cancers worldwide.The 5-year survival rate of advanced colorectal cancer and gastric cancer is less than 20%.Early detection and timely treatment can significantly improve the survival prognosis of colorectal cancer and gastric cancer.However,the current clinical screening methods,such as invasive endoscopy,noninvasive fecal occult blood test screening test,fecal immunochemical test and serum tumor markers,are limited in clinical application due to lack of patient compliance,low specificity and sensitivity.The free gene fragments in the blood are referred to as circulating cell-free DNA(cfDNA),circulating tumor DNA(ctDNA)is a part of cfDNA which is genetic material that is shed by apoptotic and necrotic tumor cells or actively released into the blood by tumor cells,carrying the genetic and epigenomic characteristics of released tumor cells.Methylation markers developed using ctDNA have shown great potential in non-invasive liquid biopsy for multi-cancer detection.In this study,we hope to explore new cfDNA methylation biomarkers by high-throughput targeted DNA methylation sequencing and other technologies,and verify their role in the detection and differentiation of gastrointestinal cancer.Method:In order to obtain the methylation markers of gastrointestinal cancer,construct and validate the gastrointestinal cancer diagnostic model,we prospectively collected plasma samples from 407 subjects from multiple centers,including 121 patients with gastric cancer,70 patients with colorectal cancer,and 216 patients with non-gastrointestinal cancer.All participants were randomly assigned in a 3:1 ratio to the diagnostic model development cohort and the independent validation cohort by a bioinformaticians blinded to the methylation sequencing results.In the model development cohort,the samples were further randomly divided into training and testing sets in a 2:1 ratio.An additional set of plasma samples from 90 lung cancer patients and 70 breast cancer patients were also included in the independent validation cohort to further verify the cross-reactivity of the experiments.Targeted methylation sequencing was performed on 407 plasma samples from patients diagnosed with CRC versus gastric cancer and non-Gastrointestinal(GI)diseases.Using 203 samples from the training set of the model development cohort,we identified 40110 GI cancer-specific methylated CpG sites and 63 predictive markers of tissue origin.Based on these methylation signatures,we further established and validated a diagnostic model for gastrointestinal cancers detection and a tissue of origin(TOO)prediction model for discriminating between CRC and GC.Results:The model achieved sensitivities of 83%(95%CIs:72.3-93.6%)and 81.2%(95%CIs:68.8-91.7%),and specificities of 81.5%(95%CIs:69.2-89.6%)and 80%(95%CIs:67.6-88.4%)in test set of the model development cohort and validation cohort respectively,for gastrointestinal cancer detection.In addition,the model showed accuracies of 95.8%(95%CIs:79.8-99.3%)and 95.8%(95%CIs:79.8-99.3%)for identification of GC and 86.7%(95%CIs:62.1-96.3%)and 93.3%(95%CIs:70.2-98.8%)for identification of CRC in test set of the model development cohort and validation cohort respectively.Conclusion:Collectively,we identified novel circulating tumor DNA methylation biomarkers for CRC and GC detection and show the promising potential as a noninvasive gastrointestinal cancer detection tool. |
PDF Full Text Request