BACKGROUNDIn this study the expression of COX-2 both in lung cancer tissues and tissues of benign disease of the lung is compared to explore the effect of COX-2 on occurrence and development of non-small cell lung cancer (NSCLC), and the relationship between COX-2 expression and clinical characteristics of patients with NSCLC.

METHODSThe COX-2 expression, vascular endothelial growth factor (VEGF) expression and microvessel density (MVD) were detected in 78 NSCLC lung cancer tissues and 14 benign tissues of the lung by immunohistochemical method.

RESULTSExpression of COX-2 protein was significantly higher in cancer tissues (69.2%, 54/78) than paired noncancerous tissues (14.3%, 2/14)(Chi-square=15.044, P < 0.01). There was significant difference of MVD between COX-2 positive lung cancers (19.41±8.56) and negative ones(12.00±5.37)(t=3.906, P < 0.01). There was no significant relationship between COX-2 protein expression and patients' age, sex, TNM stages and smoking status (P > 0.05); for different pathological types, COX-2 protein expression was remarkably higher in adenocarcinoma (88.9%, 32/36) than that in squamous cell carcinoma (50.0%, 20/40) (Chi-square=13.262, P < 0.01); and for squamous cell carcimoma it was significantly higher in low differentiated ones (81.0%, 17/21) than that in high differentiated ones (15.8%, 3/19) (Chi-square=16.942,P < 0.01); for adenocarcinoma it was also remarkably higher in low differentiated ones (100.0%, 18/18) than that in high differentiated ones ( 77.8% , 14/18) (Chi-square=6.046, P < 0.05).The expressions of COX-2 and VEGF in NSCLC were significantly positively correlated (r=0.509, P < 0.01), so were the expressions of VEGF and MVD (r=0.324, P < 0.01), and the expressions of COX-2 and MVD (r=0.477, P < 0.01). Controlling for VEGF, there still was significant correlation between COX-2 and MVD (r=0.383, P < 0.01).

CONCLUSIONSThe present study suggests that there is overexpression of COX-2 in lung cancer, which is closely related to angiogenesis of lung cancer. Further studies, therefore, are needed to find the exact mechanisms.