An abstract presented at the American Society of Clinical Oncology’s 2020 Gastrointestinal Cancers Symposium reported that a blood-based test predicted with high accuracy tissue of origin (TOO) for gastrointestinal (GI) and other cancers by identifying methylation signatures in cell-free DNA (cfDNA). This approach shows promise for detecting these cancers earlier, which often are not identified until later stages.

Methylation has a role in the body’s development of cancer. “The data show that evaluating methylation of cell-free DNA within a blood sample, may detect a variety of gastrointestinal cancers with good sensitivity and with a low rate of false positives,” lead investigator Brian M. Wolpin, MD, MPH, director of the Gastrointestinal Cancer Center and director of the Hale Family Center for Pancreatic Cancer Research at Dana-Farber Cancer Institute in Boston, summarized in a statement.

The study represents the second phase of the Circulating Cell-free Genome Atlas (CCGA) trial, a prospective, longitudinal study aimed at discovering, developing, and validating a blood test for the early detection of multiple types of cancer. The study’s sponsor, GRAIL, a diagnostics company, developed a technique that identifies methylation patterns in cfDNA. This technology “preferentially targets the most informative regions of the genome,” Wolpin told CLN Stat. “Through the company’s clinical study program, they found that methylation is a more pervasive signal compared to mutation sites, which are typically interrogated in traditional liquid biopsy approaches.”

Patients with more than 20 tumor types at all stages of cancer and a control group took part in the CCGA substudy. Using a cross-validated targeted methylation sequencing assay to analyze plasma cfDNA, investigators constructed an algorithm to decipher which patients had cancer and where it originated in the body. “Methylation fragments were combined across targeted genomic regions and assigned a probability of cancer and a predicted TOO,” Wolpin and his colleagues wrote.

Cancer classes included those of the esophagus/stomach, pancreas/gallbladder/extrahepatic bile duct, liver/intrahepatic bile duct, and colon/rectum.

Wolpin and his colleagues were able to detect multiple cancer types at an overall specificity rate >99% and a sensitivity rate of 82% on a training set of 312 patients and 81% on a validation set of 135 patients with established GI cancers. The researchers also found an overall predicted TOO accuracy of 92%. Among the specific cancer groups, the test yielded the highest sensitivity results for liver/intrahepatic bile duct cancers (86%), and highest predicted TOO accuracy for colon/rectum cancers (98%).

Sensitivity and predicted TOO accuracy also improved by disease stage among all cancers. For upper GI and liver/intrahepatic bile duct cancers, the method yielded a sensitivity of 75% and 79% for stages 1-3, respectively, which rose to 100% in both categories for stage 4 cancer.

Wolpin anticipates that this data will help further refine the methylation-based test to allow for improved detection of GI malignancies among the more than 20 cancer types he and his colleagues investigated.

“If further validated with additional testing, this approach has the potential to allow us to diagnose gastrointestinal cancers earlier, when they’re more treatable,” said Wolpin. Next steps are to validate this test’s screening capabilities in two large, separate studies targeting women undergoing screening mammograms, and men and women without a known cancer diagnosis.