235 General discussion, conclusion, and future perspective only labelled with one staining pattern per tumor. Therefore, it was assumed that association observed for MET negativity with poor survival in chapters 3 and 4 is probably due to another – uniform positive – staining pattern not sampled during TMA production (chapter 5). Another discrepancy between the TMA and WTS studies are the differences in thresholds set for D1C2 uniform positivity and MET ECD shedding, which were assumed to be explained by inclusion criteria, TMA sampling, and tissue heterogeneity (chapter 5). The fact that the uncorrected HRs for D1C2 uniform positivity and ECD shedding are of the same order of magnitude for transmembranous C-terminal MET positive cancers for both the TMA and WTS studies, convinced us that the developed scoring system provides valid results (chapters 4 and 5). Since MET facilitates invasive growth by orchestrating EMT (30), a process during which epithelial cells obtain a more mesenchymal phenotype by down-regulation of epithelial proteins such as E-cadherin (31, 32), it was investigated whether the developed scoring system was biologically informative through comparing the scores obtained for C-terminal MET and E-cadherin immunoreactivity. The fact that both the gradient towards periphery and gradient towards center, as well as the uniform positive versus negative staining patterns were inversely correlated for C-terminal MET and E-cadherin immunoreactivity, illustrated that this was indeed the case. The fact that for N-terminal MET immunoreactivity (A2H2-3), uniform negative areas were significantly larger compared to those observed for C-terminal MET immunoreactivity (D1C2), and the opposite was true for uniform positivity and the gradient towards periphery, further corroborated the biological value of the scoring system as this result is indicative for ectodomain shedding. The obtained results presented in this thesis illustrate that use of TMAs can be extremely valuable in grasping the behavior of biomarker immunoreactivity across cancers. In our case, the mere assignment of cancers to a C-terminal MET staining pattern based on the visual examination of the distribution of events across a scatter plot (chapter 3) led to a prognostically informative scoring system for WTSs (chapter 5). Moreover, alignment of C- and N-terminal MET immunoreactivity, again using a scatter plot, led to the assignment of each cancer to one of three categories based on MET protein status (chapter 4), and illustrated the occurrence of MET ectodomain shedding in OSCC (chapter 4), which was also found to be prognostically relevant (chapter 4). However, it is important to realize that although the TMAs provided us the basis for the translational insights obtained in this thesis, it was the vali8
RkJQdWJsaXNoZXIy MTk4NDMw