Integrating multi-omics features for blood-based multi-cancer early detection.

Abstract

156 Background: The current standard-of-care cancer screening paradigm is constrained to just a few cancer types and has challenges in patient compliance due to the invasive procedures endured from the tests. Recently studies have demonstrated that blood-based multi-cancer detection (MCED) approaches may hold promise for identifying asymptomatic cancer patients from the general population. However, most studies only exploit a single aspect of cancer hallmarks which is challenging for the biological reasons since cancer is a heterogenous disease with a wide spectrum of pathological and clinical behaviors. Methods: Here we report SeekInCare, a CE-IVD Mark MCED test, based on a novel multi-dimensional cancer risk score (CRS) model incorporating copy number aberrations, fragment size, end motifs and oncogenic viruses via shallow whole genome sequencing (sWGS) from cell-free DNA (cfDNA), and seven common tumor markers in a single 8ml blood draw. Results: Our research cohort consisted of 898 healthy subjects and 615 stage I-IV cancer patients that covered eight common cancers and 19 uncommon cancer types. The CRS model identified 427 cancer patients with 69.4% sensitivity at 98.0% specificity, resulting in an AUC (area under the curve) of 0.926. The sensitivities were 50.3%, 64.0%, 73.8% and 86.2% in stage I, II, III and IV cancers respectively. The sensitivities of eight common cancer types, breast, stomach, lung, colorectum, lymphoma, liver, pancreas and leukemia, were 45.1%, 50.0%, 63.4%, 69.4%, 70.5%, 81.4%, 82.4% and 90.9% respectively. We also prospectively evaluated SeekInCare in a real-world cohort consisting of 1212 subjects who received the test as a LDT (laboratory developed test) (median follow-up time: 753 days, range: 78~1669 days). 13 out of 18 cancer cases were detected while 46 subjects tested positive but without cancer. Thus, SeekInCare achieved 72.2% sensitivity, 96.1% specificity, 22.0% PPV and 99.6% NPV in the real-world cohort. Conclusions: In this study, we provided a non-invasive MCED test (SeekInCare) based on the multi-omics and multi-dimensional features. The case-control study demonstrated that SeekInCare could detect >20 cancer types with 69.4% sensitivity at 98.0% specificity. The outstanding real-world performance of SeekInCare warrants future investigation of its clinical utility and health economics as a mass cancer screening test in average-risk populations.