A plasma metabolite-based machine-learning model diagnosed colorectal cancer relapse with 62% sensitivity and 80% specificity, with a median lead time of 471 days before clinical diagnosis.
Cohort (n=160)
Does a plasma metabolite-based machine-learning model detect minimal residual disease and predict relapse in post-surgery patients with colorectal cancer?
A novel plasma metabolite-based machine-learning model can detect colorectal cancer relapse with moderate sensitivity and high specificity, offering significant lead time before clinical diagnosis.
Colorectal cancer (CRC) is among the leading causes of cancer-related deaths worldwide, predominantly caused by recurrence, underscoring the need for novel biomarkers for the early detection of relapse. We hypothesize that patients' response to cancer is closely linked to metabolic changes and can be detected in blood as an indicator for recurrence. Full metabolomics analysis was performed on the longitudinal samples collected from 160 patients with CRC before and after surgery for 24 months or more. Key blood metabolic biomarkers that distinguish relapse and non-relapse patients were identified, and a minimal residual disease (MRD) detection machine-learning model was constructed based on the discovered signatures. The model diagnosed CRC relapse with a sensitivity of 62% and specificity of 80%, with median and maximum lead times of 471 and 1056 days before diagnosed clinical relapse, suggesting it could be used as a novel diagnostic tool for the earlier detection of cancer relapse.
Zhou et al. (Mon,) conducted a cohort in Colorectal cancer (n=160). Plasma metabolite-based machine-learning model was evaluated on Detection of colorectal cancer relapse. A plasma metabolite-based machine-learning model diagnosed colorectal cancer relapse with 62% sensitivity and 80% specificity, with a median lead time of 471 days before clinical diagnosis.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: