AIM: Inhibition of cyclooxygenase-2 (COX-2) is proposed as a treatment option in several cancer types. However, in non-small cell lung cancer (NSCLC), phase III trials have failed to demonstrate a benefit of adding COX-2 inhibitors to standard chemotherapy. The aim of this study was to analyze COX-2 expression in tumor and stromal cells as predictive biomarker for COX-2 inhibition.

METHODS: In a multicenter phase III trial, 316 patients with advanced NSCLC were randomized to receive celecoxib (400 mg b.i.d.) or placebo up to one year in addition to a two-drug platinum-based chemotherapy combination. In a subset of 122 patients, archived tumor tissue was available for immunohistochemical analysis of COX-2 expression in tumor and stromal cells. For each compartment, COX-2 expression was graded as high or low, based on a product score of extension and intensity of positively stained cells.

RESULTS: An updated analysis of all 316 patients included in the original trial, and of the 122 patients with available tumor tissue, showed no survival differences between the celecoxib and placebo arms (HR 1.01; 95% CI 0.81-1.27 and HR 1.12; 95% CI 0.78-1.61, respectively). High COX-2 scores in tumor (n = 71) or stromal cells (n = 55) was not associated with a superior survival outcome with celecoxib vs. placebo (HR =0.96, 95% CI 0.60-1.54; and HR =1.51; 95% CI 0.86-2.66), and no significant interaction effect between COX-2 score in tumor or stromal cells and celecoxib effect on survival was detected (p = .48 and .25, respectively).

CONCLUSIONS: In this subgroup analysis of patients with advanced NSCLC treated within the context of a randomized trial, we could not detect any interaction effect of COX-2 expression in tumor or stromal cells and the outcome of celecoxib treatment in addition to standard chemotherapy.