Soluble c-Met Levels Correlated With Tissue c-Met Protein Expression in Patients With Advanced Non–Small-Cell Lung Cancer

Background Immunohistochemistry (IHC) and fluorescent in situ hybridization are reliable methods for identifying c-Met protein expression or c-Met gene amplification. However, each technique requires a high-quality tissue sample, which might not be available. The aim of the present study was to investigate the correlation between the soluble c-Met level and tissue c-Met protein expression and the relationship between these markers and patient prognosis. Materials and Methods In 198 patients with advanced non–small-cell lung cancer, tumor tissue c-Met expression was determined using IHC according to the H score criteria. Positivity was defined as ≥ 50% of cells with strong staining (IHC 3+). The concentration of c-Met protein in paired plasma samples was measured using a human soluble c-Met quantitative enzyme-linked immunosorbent assay kit, and the predictive value was determined using receiver operating characteristic curve analysis. Results Of the 198 patients, 140 (70.7%) had tissue c-Met− findings and 58 (29.3%) tissue c-Met+ findings. Receiver operating characteristic curve analysis showed 67.2% specificity and 65.0% sensitivity for predicting tissue c-Met positivity at a plasma c-Met cutoff of 766 ng/mL. The correlation between the soluble c-Met level and tissue c-Met protein expression was significant (Pearson's r = 0.309; P < .001). Patients with high soluble c-Met levels (> 766 ng/mL) had poorer overall survival than patients with low soluble c-Met levels (9.5 vs. 22.2 months; P < .001). Multivariate analyses demonstrated the same result (hazard ratio, 2.15; 95% confidence interval, 1.334-3.446; P = .002). Conclusion A significant correlation was found between the plasma soluble c-Met levels and tissue c-Met protein expression in patients with advanced non–small-cell lung cancer. A high level of soluble c-Met was associated with a poor prognosis.