Related Experiment Video
Updated: Jan 12, 2026
Genome-Wide Analysis of DNA Methylation in Gastrointestinal Cancer
Published on: September 18, 2020
Clinical validation of a multi-model blood cfDNA methylation assay for early-stage gastrointestinal cancer screening
Lingqin Zhu1, Shuye Lin1, Junfeng Xu2
1Department of Gastroenterology, Beijing Chest Hospital, Capital Medical University, Beijing Tuberculosis and Thoracic Tumor Research Institute, Beijing 101149, China; Laboratory for Clinical Medicine, Capital Medical University, Beijing 101149, China.
Background:
Gastrointestinal (GI) tract cancers are the second leading cause of cancer-related mortality, often due to late detection. There is a critical need for non-invasive, highly sensitive biomarkers for early-stage cancer and precancerous lesion detection to enable timely intervention. This study aimed to develop a blood-based method specifically optimized for early GI cancer detection and population-level screening.
Methods:
Using large-scale public tissue methylation data and the Twist probe cfDNA profiles, we developed SPOGIT (Screening for the Presence of Gastrointestinal Tumors), a multi-algorithm model (Logistic Regression/Transformer/MLP/Random Forest/SGD/SVC) for early GI cancer detection. The model was rigorously validated through an internal (n = 83) and multicenter external validation (386 cancers/113 controls/580 precancers), with an interception model assessing its clinical potential.
Results:
SPOGIT demonstrated high accuracy in detecting GI cancers, with a sensitivity of 88.1 % and a specificity of 91.2 %. Notably, it effectively identified early-stage (0-II) cancers with 83.1 % sensitivity. The model also showed significant potential for intercepting premalignant progression, detecting advanced adenomas (AA) and gastric precancerous lesions with sensitivities of 56.5 % and up to 62.4 %, respectively. In the external independent validation cohort, a complementary model CSO (Cancer Signal Origin) demonstrated an accuracy of 83 % for colorectal cancer and 71 % for gastric cancer. Most importantly, simulation analyses projected that SPOGIT implementation could significantly reduce late-stage diagnoses and increase 5-year survival rate by 27.02 % through early interception.
Conclusions:
This study introduced a novel, dual-model blood architecture (SPOGIT/CSO) that enables highly accurate, early detection of GI cancers and their precursors. By facilitating timely clinical intervention, SPOGIT/CSO represented a paradigm-shifting strategy with the potential to significantly improve patient survival outcomes and transform GI cancer management.
