Multi-dimensional cell-free DNA-based liquid biopsy and early detection of gastric cancer.

Abstract

4060 Background: Gastric cancer is one of the most common cancer types. Most patients were diagnosed at advanced stages and experienced poor prognosis. A non-invasive assay for the detection of early-stage gastric cancer is highly desirable for reducing the gastric cancer-associated mortality. Methods: We prospectively collected a study cohort of 249 participants including 110 pathologically confirmed with stage I-II gastric cancer and 139 with non-cancerous conditions. A plasma sample was collected from each patient before conventional screening modalities for gastric cancer. We performed whole-genome sequencing with plasma samples and profiled four types of cell-free DNA (cfDNA) characteristics, fragment size pattern, copy number variation, nucleosome coverage pattern, and single nucleotide substitution. We used these differential profiles to develop an ensemble model to identify gastric cancer patients from non-cancer individuals. We further collected a validation cohort consisting of 73 gastric cancer patients and 94 non-cancer individuals to validate the cfDNA-based assay. Additionally, we compared the performance of our assay and conventional gastroscopy in a hypothetical 100,000 screening population by Monte Carlo simulation. Results: The liquid biopsy assay that incorporated four types of cfDNA characteristics was able to identify early-stage gastric cancer patients from non-cancer individuals at an AUROC of 0.962 in the study cohort and 0.972 in the validation cohort. At a specificity of 92.1% (128/139), it achieved a sensitivity of 88.2% (97/110) in the study cohort. With this threshold, in the validation cohort, 91.5% (86/94) of healthy individuals and 91.8% (67/73) of gastric cancer patients were correctly identified. In both study and validation cohorts, the inferred probabilities of cancer showed consistent trends to increase with pathological stages in the cancer group and disease statuses in the non-cancer group. Of note, our approach detected all gastric tumors located in the cardia and fundus (100.0%, 19/19), which could be challenging for gastroscopic examination. Additionally, through in-silico simulations, we showed that our cfDNA-based non-invasive assay may detect 96.3% more gastric cancer cases than conventional gastroscopy owing to higher sensitivity and anticipated better participant compliance in a large hypothetical screening population. Conclusions: We introduced a liquid biopsy assay using multiple dimensions of cfDNA characteristics that could accurately identify early-stage gastric cancer from non-cancerous conditions. As a cost-effective non-invasive approach, it may provide population-wide benefits for the early detection of gastric cancer.