Microarray Gene Expression Profiling for Predicting Complete Response to Preoperative Chemoradiotherapy in Patients with Advanced Rectal Cancer

Purpose: Preoperative chemoradiotherapy is widely used to improve local control and sphincter preservation in patients with locally advanced rectal cancer. In the present study, we investigated whether microarray gene expression analysis could predict complete response to preoperative chemoradiotherapy in rectal cancer. Methods: Tumor tissues were obtained from 46 patients with rectal cancer (31 for training and 15 for validation testing). All patients underwent preoperative chemoradiotherapy involving 50.4 gray radiotherapy, followed by surgical excision 6 weeks later. Response to chemoradiotherapy was evaluated according to Dworak's tumor regression grade. Tumor regression Grades 1, 2, and 3 were considered partial responses, and tumor regression Grade 4 was considered a complete response. By using the 31 training samples, genes differentially expressed between partial response and complete response were identified, and clustering analysis was performed. Prediction analysis of response to chemoradiotherapy was performed on the 31 training samples by using a selected set of 95 “predictor” genes. Those findings were validated by independent analysis of the 15 test samples. Results: The 31 training samples comprised 20 partial response and 11 complete response cases. A primary set of 261 genes was identified as differentiating between partial response and complete response. By supervised clustering using these 261 genes, 30 of 31 training samples were clustered correctly according to tumor response. A gene set comprising the top-ranked 95 genes displaying differential expression between partial response and complete response was applied to predict response to chemoradiotherapy. Complete response and partial response were accurately predicted in 84 percent (26/31) of training samples and 87 percent (13/15) of validation samples. Conclusions: Microarray gene expression analysis was successfully used to predict complete responses to preoperative chemoradiotherapy in patients with advanced rectal cancer.