A multi-cancer detection test could detect cancer with high specificity and predict its origin with accuracy among symptomatic individuals in a new analysis in JCO Precision Oncology. As part of the Circulating Cell-free Genome Atlas study, researchers from Grail and elsewhere examined the performance of a multi-cancer detection test that combines a targeted methylation assay and machine learning classifiers to detect abnormal methylation patterns within cell-free DNA samples. Their analysis included individuals who presented clinically with cancer or with a high suspicion of cancer and individuals without cancer but with underlying medical conditions. The researchers reported that the test had high specificity across the noncancer groups, with an aggregate specificity of 99.5 percent. Meanwhile, among the overall sensitivity to detect cancer in the cancer groups was 64.3 percent, and a high cancer signal origin prediction accuracy of 90.3 percent. The researchers further note that test performance was high among individuals with gastrointestinal cancers. "A specific and sensitive multi-cancer detection test represents a novel, noninvasive tool that can support appropriate triaging of patients in whom clinical symptoms and signs raise a suspicion of cancer," they write.
Specificity, Accuracy of Multi-Cancer Detection Test in Symptomatic Individuals Assessed
Jul 24, 2023