Background: Gastrointestinal (GI) cancers accounts for one-third of the world’s cancer burden. Liquid biopsies such as SEPT9 methylation and Cologuard have been approved by the Food and Drug Administration for early detection of colorectal cancer (CRC), but not for other GI cancers. In addition, although CancerSEEK can be employed in the detecting of multiple GI cancers, it has low sensitivity and without localization. Stool is a promising sample for GI cancer detection because it contains the host’s exfoliated cells and circulating free DNA derived from GI cancer cells. Thus, we aimed to develop a sensitive and non-invasive multi-target stool DNA methylation test for early detection and localization of GI cancers. Methods: We prospectively enrolled 124 patients histologically diagnosed with GI cancer without treatment and 92 healthy controls who underwent GI cancer-related examinations every 3 years. Stool DNA was extracted from the supernatant of a mixture containing 5 g fresh stool and 25 ml preservative solution using the stool DNA extraction kit. The samples were subjected to bisulfite conversion using the EZ DNA Methylation Kit, and the treated with stool DNA multi-target methylation assay kit to detect the methylation levels of 3 target regions of stool DNA (PDX1, COL4A2-1, and FGF5) with qPCR. Based on the guidelines of the protocol, positive result was defined as ΔCt value (ΔCt = Ct (target gene) - Ct (GAPDH))＜10 for any three genes. Multiple logistic regression (MLR) models for single GI cancers were constructed including the ΔCt value of each gene to adjust for tissue of origin (TOO) of GI cancers. This study was approved by the Clinical Research Ethics Committee of Guangdong Second Provincial General Hospital (2021-KZ-138-03). Results: Overall, 98 of 124 patients (79%) tested positive, and 88 of 92 healthy controls (96%) tested negative. For each GI cancer type, the positivity rates were 71%, 83%, 75%, 81% and 91% for CRC (24/34), gastric cancer (GC, 19/23), esophageal cancer (EC, 18/24), pancreatic cancer (PC, 17/21) and ampullary cancer (AC, 20/22), respectively. Interestingly, among 4 positive healthy controls, 3 were diagnosed as advanced adenomas and one case was a pregnant woman. For the MLR model, the sensitivity of CRC was 88% (30/34), that of GC was 91% (21/23), EC was 88% (21/24), PC was 90% (19/21), and AC was 95% (21/22). Conclusions: The multi-target stool DNA methylation assay allowed early and accurate detection and identification of tissue of origin of GI cancers.

Abbreviations: GI cancers, gastrointestinal cancers; CRC, Colorectal cancer; GC, Gastric cancer; EC, Esophageal cancer; PC, Pancreatic cancer; AC, Ampullary cancer; MLR, Multiple logistic regression.