Stage III colon cancer prognosis prediction by gene expression profiling.

B20 Purpose. This study aimed to assess the possibility to build a microarray-based prognosis predictor (PP) for stage III colon cancer that could be used to guide postoperative chemotherapy. Material and methods. Thirty-six patients, operated on for a stage III colon cancer, were included in this study. Eighteen patients have subsequently developed a liver metastasis, while the other 18 have remained disease-free for at least 5 years. Tumor mRNA samples were profiled using the Affymetrix HGU133A GeneChip. Patients were repeatedly and randomly divided into 10,000 training (TS) and validation sets (VS) of 10 different sizes. For each TS/VS split, a 30-gene prognosis predictor (PP), identified on the TS by selecting the 30 most differentially expressed genes and applying diagonal linear discriminant analysis, was used to predict the prognoses of VS patients. Results. The 10,000 30-gene PP yielded the following average prognosis prediction performance measures: 72.9% accuracy, 72.2% sensitivity, 73.6% specificity. Improvements in prognosis prediction were observed with increasing TS size (76.1% accuracy, 75.2% sensitivity, and 77.1% specificity for TS of size 32). The 30-gene PP were found to be highly-variable in composition across TS/VS splits. A total of 7,096 genes were included in the 10,000 PP; the higher number of selections for a gene was 5,896. Conclusions. Microarray gene expression profiling is able to predict the prognosis of stage III colon cancer patients and, thus, might be used to guide adjuvant chemotherapy.