Background:
Prognostic information for colon cancer is largely derived from pathologic staging. There is, however, significant variability in individual patient outcomes that cannot be explained using histopathologic factors alone. Recent data has suggested that several molecular markers may be predictive of disease progression in colon cancer. Most of these studies have examined individual genes and markers, whose role in a broader prognostic system remains unclear. The aim of this study was to examine the association of several well-studied molecular markers with lymph node metastasis in colon cancer and to develop a broader multivariate predictive model incorporating multiple genes.

Methods:
The mRNA expression of COX-2, VEGF, Bcl-2, cyclin-E, EGF-R, DPD, MMP-9, MDR-1, thymidine phosphorylase (TP), and thymidylate synthase (TS) were determined using quantitative real-time reverse transcription PCR from archived formalin-fixed paraffin embedded specimens of 65 node-negative and 66 node-positive T3 colon cancers obtained from the USC Colon Cancer Family Registry. The microsatellite instability (MSI) status of these tumors was available through the registry. The association of MSI status and gene expression levels with lymph node status was examined using univariate and multivariate logistic regression. A stepwise model building technique was used in an attempt to identify a multivariate predictive model.

Results:
On univariate analysis, the expression of TS was statistically significantly associated with lymph node status, adjusting for age, gender, ethnicity, and location of tumor (OR 0.36; 95% CI: 0.16, 0.81), with a decreased level of expression predictive of greater risk of metastasis. MSI status and the expression of the other genes were not significantly associated with lymph node metastasis. Because only one gene was predictive, multivariate logistic regression did not identify a significant predictive model beyond that containing only TS.

Conclusion:
Decreased expression of thymidylate synthase is associated with a higher risk of lymph node metastasis in patients with T3 colon cancers. MSI status and the expression of other genes were not predictive of lymph node status. As a result, a multivariate predictive model incorporating several genes could not be identified for this sample population. As an individual marker, thymidylate synthase gene expression may help identify a subset of patients at greater risk for progression of disease.