| Theme | How to Manage Distant Metastases from Gastrointestinal Cancer? | |
|---|---|---|
| Title | Assessment for Metastatic Potential of Colorectal Cancer | |
| Publish Date | 2005/02 | |
| Author | Yoshiaki Arimura | 1st Department of Internal Medicine, Sapporo Medical Scholl |
| Author | Kohzoh Imai | 1st Department of Internal Medicine, Sapporo Medical Scholl |
| [ Summary ] | Purpose : Using classical histologic criteria, many attempts to predict nodal metastasis have failed causing considerable controversy concerning the adequate management of cancer. We investigated the feasibility of individualizing treatment based on the assessment of nodal metastasis in colon and rectal cancer. Methods : The gene signature associated with nodal metastasis was investigated by cDNA array. We also analyzed a total of 424 colon and rectal cancer patients to identify possible causality for nodal metastasis. We then evaluated the artificial neural network model built-in matrilysin for its potential to predict nodal metastasis in colon and rectal cancer. Results : cDNA array revealed that matrilysin was maximally up-regulated in the metastasis signature identified. Tumor matrilysin expression was revealed as a stage-independent risk factor for nodal metastasis. The ROC analysis indicated that the artificial neural network model was superior to the logistic regression model for predictive ability. Sensitivity analysis of the neural model identified matrilysin as the most relevant predictor examined. Conclusion : We provide evidence that tumor matrilysin expression is a promising biomarker to predict nodal metastasis of colon and rectal cancer. Artificial neural network model built-in matrilysin would help clinicians move closer towards the ultimate goal of cancer treatment based on the individualized metastatic potential of colon and rectal cancer. |
|