MET immunolabelling is a useful predictive tool for MET gene amplification in glioblastoma

Aims MET gene amplification is rare in glioblastoma ( GBM ) and represents a potential target for MET inhibitors. An immunohistochemical screening may be useful to identify MET amplification. The aim of our study was to establish how MET immunolabelling correlates with MET amplification. Methods Three cohorts including 108 GBM (cohort 1, prospective), 104 GBM (cohort 2, retrospective) and 52 GBM (cohort 3, prospective) were investigated for MET expression by immunohistochemistry. MET amplification was assessed by comparative genomic hybridization on microarray ( CGH ‐array) in all cohorts and by fluorescent in situ hybridization ( FISH ) in cohorts 2 and 3. Active form of MET was assessed using p‐ MET (Y1349) immunohistochemistry. Results Diffuse MET amplification detectable by CGH ‐array was associated with diffuse, strong MET immunolabelling (four cases in cohort 1 and one case in cohort 2). Focal MET amplification detectable only by FISH was observed in small foci of strongly immunopositive cells in two GBM (cohort 2). In both cohorts, MET amplification was never detected in GBM devoid of strongly immunopositive cells. MET overexpression, observed in 23% of unamplified GBM , was associated with a predominant weak‐to‐moderate staining intensity and with necrosis ( P < 0.005). p‐ MET was detected in all MET ‐amplified GBM and in perinecrotic areas of nonamplified GBM . A strong MET immunostaining intensity, at least focal and distant from necrosis, showed 100% sensitivity and 84% specificity for predicting MET amplification in cohort 3. Conclusions MET amplification is characterized by strongly immunopositive cells. Only GBM showing strong MET immunostaining is appropriate for the assessment of MET amplification.