| Title: | Additive pre-diagnostic and diagnostic value of routine blood-based biomarkers in the detection of colorectal cancer in the UK Biobank cohort |
| Journal: | Scientific Reports |
| Published: | 24 Jan 2023 |
| Pubmed: | https://pubmed.ncbi.nlm.nih.gov/36693981/ |
| DOI: | https://doi.org/10.1038/s41598-023-28631-y |
| URL: | https://www.nature.com/articles/s41598-023-28631-y.pdf |
Publication 7976
WARNING: the interactive features of this website use CSS3, which your browser does not support. To use the full features of this website, please update your browser.
Abstract
Survival rates from colorectal cancer (CRC) are drastically higher if the disease is detected and treated earlier. Current screening guidelines involve stool-based tests and colonoscopies, whose acceptability and uptake remains low. Routinely collected blood-based biomarkers may offer a low-cost alternative or aid for detecting CRC. Here we aimed to evaluate the pre-diagnostic and diagnostic value of a wide-range of multimodal biomarkers in the UK Biobank dataset, including sociodemographic, lifestyle, medical, physical, and blood and urine-based measures in detecting CRC. We performed a Cox proportional hazard and a tree-boosting model alongside feature selection methods to determine optimal combination of biomarkers. In addition to the modifiable lifestyle factors of obesity, alcohol consumption and cardiovascular health, we showed that blood-based biomarkers that capture the immune response, lipid profile, liver and kidney function are associated with CRC risk. Following feature selection, the final Cox and tree-boosting models achieved a C-index of 0.67 and an AUC of 0.76 respectively. We show that blood-based biomarkers collected in routine examinations are sensitive to preclinical and clinical CRC. They may provide an additive value and improve diagnostic accuracy of current screening tools at no additional cost and help reduce burden on the healthcare system.2 Authors
- Gizem Tanriver
- Ece Kocagoncu
1 Application
| Application ID | Title |
|---|---|
| 87991 | Validation of an AI-powered online search strategy for finding optimal biomarker combinations |
Enabling scientific discoveries that improve human health