Gene-expression profiles correlate with the efficacy of anti-EGFR therapy and chemotherapy for colorectal cancer.

Comprehensive gene-expression analysis is very useful for classifying specific cancers into subgroups on the basis of their biological characteristics; it is used both prognostically and predictively. The purpose of this study was to classify unresectable advanced or recurrent colorectal cancer (CRC) by gene-expression profiling of formalin-fixed paraffin-embedded tissues and to correlate CRC subgroups with clinicopathological and molecular features and clinical outcomes. One hundred patients with advanced or recurrent CRC were enrolled. RNA extracted from FFPE tissues was subjected to gene-expression microarray analysis. The patients were stratified into four subgroups (subtypes A1, A2, B1, and B2) by unsupervised hierarchical clustering. By use of principle-components analysis (PCA), the patients were divided into subtypes A and B on the basis of component 1 and into subtypes 1 and 2 on the basis of component 2. Subtype A was significantly enriched among patients without the KRAS mutation and with an earlier clinical stage at diagnosis. With regard to anti-EGFR therapy, progression-free survival (PFS) was better for patients in subtype A without the KRAS mutation than for those with the KRAS mutation (P = 0.047). PFS for patients without the KRAS mutation in subtype B was comparable with that for patients with the KRAS mutation (P = 0.55). Similar results were observed in a validation set. We found that gene-expression profiles enabled stratification of CRC patients into four subgroups. The efficacy of anti-EGFR therapy was correlated with component 1 from PCA. This comprehensive study may explain the heterogeneity of unresectable advanced or recurrent CRC and could be useful for identifying novel biomarkers for CRC treatment.