Purpose: Preoperative chemotherapy has demonstrated a survival benefit for patients with potentially resectable esophageal cancer; however, currently it is not possible to predict the benefit of this treatment for an individual patient. This prospective study was designed to correlate gene expression profiles with clinical outcome in this setting.
Patients and methods: Eligible patients were deemed to have resectable disease after staging by computed tomography, endoscopic ultrasound, and laparoscopy as indicated and following discussion at the multidisciplinary team meeting. All patients received neoadjuvant platinum and fluoropyrimidine-based chemotherapy; and clinical data were entered prospectively onto a study-specific database. Total RNA was isolated from pretreatment tumor biopsies obtained at baseline endoscopy and analyzed using a cDNA array consisting of 22,000 cDNA clones.
Results: Of the patients with adequate follow-up accrued between 2002 and 2005, 35 satisfied the quality control measures for the microarray profiling. Median follow-up was 938 days. Supervised hierarchical clustering of normalized data revealed 165 significantly differentially expressed genes based on overall survival (OS; P < .01) with 2 distinct clusters: a poor outcome group: N = 17 (1 year OS 46.2%) and a good outcome group: N = 18 (1 year OS 100%). Genes identified included those previously associated with esophageal cancer and, interestingly, a group of genes encoding proteins involved in the regulation of the TOLL receptor-signaling pathway.
Conclusion: This initial study has highlighted groups of tumors with distinct gene expression profiles based on survival and warrants further validation in a larger cohort. This approach may further our understanding of individual tumor biology and thus facilitate the development of tailored treatment.