Biological classification of colorectal cancer (CRC) by gene expression analysis can elucidate the heterogeneous background of the disease. But most of previous studies used fresh frozen samples that were difficult to obtain, store and handle. The purpose of this study was to classify CRC based on gene expression profiles using formalin-fixed, paraffin-embedded tissues (FFPE) and to correlate subgroups of CRC with biological features and clinical outcomes.Methods
One hundred FFPE samples were obtained from CRC patients with unresectable metastasis or local recurrence who underwent surgical resection of primary tumor previously. RNA extracted from FFPE was subjected to gene expression microarray. Mutation analysis was carried out by direct DNA sequencing.Results
CRC was stratified into four subgroups by unsupervised hierarchical clustering. By principle component analysis (PCA), CRC was divided into clusters A and B by component 1 and was divided into clusters 1 and 2 by component 2. Cluster B was significantly enriched in patients with KRAS mutation. After anti-EGFR therapy, cluster A patients showed better PFS than cluster B patients although the difference was not significant (P = 0.12). Cluster 2 patients exhibited worse PFS after first line chemotherapy with l-OHP or CPT-11 (P = 0.04).Conclusions
We showed that gene expression profiles of CRC allowed stratification into four subgroups that had different biological features and clinical outcomes. Furthermore, we showed that the efficacies of anti-EGFR therapy and chemotherapy correlate with two distinct axes. Gene expression profile using FFPE may explain the heterogeneity of CRC and aid the identification of new biomarkers for anticancer therapy of metastatic CRC.