Abstract
The transmembrane protein, c-Met, is thought to be overexpressed and activated in colorectal cancer (CRC). This study explored its potential as a diagnostic tissue biomarker for CRC in a large human CRC tissue collection obtained from a randomized clinical trial. Tissue microarrays of matched normal colorectal epithelium and primary cancer were prepared from specimens obtained from 280 patients recruited to the MRC CLASICC trial (ISRCTN 74883561) and interrogated using immunohistochemistry for c-Met expression. The distribution and intensity of immunopositivity was graded using a validated, semi-quantifiable score, and differences in median scores analysed using the Wilcoxon signed-rank test. A receiver operating characteristic (ROC) curve was plotted to measure the diagnostic accuracy of c-Met as a biomarker in CRC. Epithelial cell membrane expression of c-Met differed significantly between CRC and normal colorectal tissue: median 12.00 (Interquartile range (IQR) 6-15) versus median 6.00 (IQR 2.70-12.00) respectively (P = <.0001). ROC-AUC analysis of c-Met expression yielded a CRC diagnostic probability of 0.66 (95% CI: 0.61 to 0.70; P < .0001). A score of ≥14.50 showed high specificity at 85.32% (95% CI 80.33%-89.45%) but sensitivity of only 30.92% (CI 25.37%-36.90%). Thus c-Met is consistently overexpressed in human CRC as compared to normal colorectal epithelium tissue. c-Met expression may have a role in diagnosis and prognostication if combined with other biomarkers.