Abstract
This is an early clinical analysis of the DEEPGENTM platform for cancer detection. Newly diagnosed cancer patients and individuals with no known malignancy were included in a prospective open-label case-controlled study (NCT03517332). Plasma cfDNA that was extracted from peripheral blood was sequenced and data were processed using machine-learning algorithms to derive cancer prediction scores. A total of 260 cancer patients and 415 controls were included in the study. Overall, sensitivity for all cancers was 57% (95% CI: 52, 64) at 95% specificity, and 43% (95% CI: 37, 49) at 99% specificity. With 51% sensitivity and 95% specificity for all stage 1 cancers, the stage-specific sensitivities trended to improve with higher stages. Early results from this preliminary clinical, prospective evaluation of the DEEPGENTM liquid biopsy platform suggests the platform offers a clinically relevant ability to differentiate individuals with and without known cancer, even at early stages of cancer.