December 17 2025
A recently published study in npj Precision Oncology adds further depth to the growing body of scientific evidence around multi-cancer early detection and the performance of OncoSeek®. OncoSeek® is a machine learning algorithm that uses a blood test to detect nine high-mortality cancer types by analysing the concentration of a selected number of protein tumour markers.
From feasibility to robustness
Over the past years, research into multi-cancer early detection has steadily evolved. Earlier publications focused on the potential population level impact of adding a multi-cancer test alongside existing screening programmes, as well as on the clinical and technical validation of individual assays. The new study builds on this foundation by further examining the robustness of OncoSeek across multiple cohorts and settings, using updated performance analyses.
Updated performance insights
The study reports an average sensitivity of 58.4 percent with a specificity of 92.0 percent across cancers included in the analysis, with performance remaining consistent across cohorts. In addition, cancer specific sensitivities are described for nine common cancer types, breast, colorectal, liver, lung, lymphoma, oesophagus, ovary, pancreas and stomach cancer, together accounting for a substantial proportion of global cancer related mortality. The publication also includes stage specific analyses, showing good sensitivity in early stages whilst sensitivity increases with advancing clinical stage.
Tissue of origin and clinical pathways
Beyond overall detection performance, the study further explores the ability of OncoSeek® to provide an indication of tissue of origin, supporting more focused follow-up and diagnostic pathways. Together, these findings contribute to a more detailed understanding of how blood based multi-cancer early detection can function in practice.
Implementation in healthcare settings
OncoSeek® is scientifically validated, CE marked and designed to be easily implemented using existing medical infrastructure. The test can be used by medical organisations providing health services to the public, supporting applicability across a wide range of healthcare settings worldwide.
Positioning within ongoing research
For OncoInv, this publication fits within a broader trajectory of ongoing research and validation. As the field matures, the emphasis is increasingly shifting from initial feasibility to reproducibility, refinement, and practical use across different clinical and laboratory contexts. Each new study adds another layer of insight, strengthening the foundation on which professionals assess the role of multi-cancer early detection within healthcare systems.
Cumulative evidence
Together with earlier peer reviewed studies, this new publication underlines the continued development of multi-cancer early detection as a field grounded in cumulative research and ongoing evaluation.
Links
Shen, Y., Xia, Y., Chang, Y. et al. A large-scale, multi-centre validation study of an AI-empowered blood-based test for multi-cancer early detection. npj Precis. Onc. 9, 321 (2025).https://doi.org/10.1038/s41698-025-01105-2
More information about OncoSeek®
