EGFR and COX-2 immunohistochemical assessment Tumor EGFR and COX-2 immuno-expression was determined from biopsies taken at baseline (archived paraffin-embedded samples were permitted). Biopsy samples (≥2 mm2) underwent fixation in 4% neutral buffered formalin for 8 to 16 hours at room temperature followed by routine specimen dehydration using graded ethanols to xylene (or chloroform). Samples were then embedded longitudinally in paraffin under vacuum at 60 °C. In the event that paraffin-embedded tumor biopsies could not be provided, 5 μm thick sections were cut from tumor biopsies and applied Inhibitors,research,lifescience,medical to ten positively charged glass slides. EGFR protein expression was assessed

at the central laboratory by immunohistochemistry using the EGFR pharmDx kit (DAKO, Glostrup, Denmark), Inhibitors,research,lifescience,medical and a staining intensity of 0 to 3+. For the purpose of statistical analyses, staining intensities of 0 or 1+ were considered negative, and scores of 2+ or 3+ were considered positive for EGFR protein expression. Immunohistochemistry for COX-2 was performed using a murine anti-COX-2 monoclonal antibody (clone 33, BD Transduction Laboratories, Lexington, KY, USA) at a dilution of 1:100. Samples were incubated for 16 hours at 4 °C, Inhibitors,research,lifescience,medical amplified using an avidin-biotin-peroxidase system, with OSI-906 in vitro antigen recovery performed under pressure (3.30 min) in sodium citrate solution (pH 6.0). The extension of stromal and tumoral COX-2 staining was assessed in a semiquantitative

manner from 0 to 3+, where 0 and 1+ were considered negative and 2+ or 3+ were considered positive. Statistical analysis This was a pilot feasibility

study and no formal statistical power calculations were performed. Nevertheless, a sample size of 30 patients was considered Inhibitors,research,lifescience,medical sufficient Inhibitors,research,lifescience,medical to examine the primary objective given that any event with an underlying incidence of 8% has a probability in excess of 90% of occurring in at least one patient out of 30. The intent-to-treat population (i.e., all patients who enrolled and received study medication) was used to analyze efficacy parameters. Median duration of response, TTP, and overall survival were summarized using Kaplan-Meier methods along with the appropriate 4-Aminobutyrate aminotransferase 95% confidence interval (CI). Tolerability outcomes were described using standard summary statistics. Results Patients In total, 30 patients were enrolled into the study between December 2002 and April 2003 and their demographic characteristics are summarized in Table 1. Colorectal carcinoma was the most common primary GI tumor (83% of patients). Twenty-nine patients had received prior chemotherapy, with the majority receiving at least two previous regimens. Nearly one quarter of patients had also received prior radiotherapy. ECOG performance status was 0 to 1 in 90% of patients. All enrolled patients received at least one dose of gefitinib and celecoxib, and the median duration of treatment throughout the study was 70 days (range, 13 to 290 days).