As the worldwide prevalence of colorectal cancer (CRC) is increasing, it is of vital importance to reduce its morbidity and mortality by early detection. Saliva is a noninvasively accessible fluid that potentially reflects both oral and systemic diseases. We report here the investigation and validation of salivary biomarkers to distinguish patients with CRC from those with polyps and healthy controls.
Saliva samples from subjects with CRC, polyps, and healthy controls were collected after 9 hours of fasting, and were split into training and validation data. Capillary electrophoresis-mass spectrometry-based metabolomics was used to quantify numerous hydrophilic metabolites.
A total of 2,602 unstimulated saliva samples were collected from 231 subjects with CRC, 99 subjects with polyps, and 2272 subjects with healthy controls. The data were randomly divided into training (n = 1,301) and validation data (n = 1301). Biomarkers for distinguishing subjects with CRC from the others, as well as metabolites that normalize whole saliva concentration were identified. Analysis of these metabolites using machine learning-based artificial intelligence showed a high area under the receiver operating characteristic curve (AUC = 0.876; P < 0.0001) in the training dataset. This combination also showed high AUC values using the validation dataset (AUC = 0.861; P < 0.0001). Saliva samples were also collected multiple times from identical subjects and the robustness of these biomarkers were confirmed.
Combinations of salivary metabolites show high potential as a screening tool for CRC.
Clinical trial identification
Legal entity responsible for the study
Has not received any funding.
All authors have declared no conflicts of interest.