233 General discussion, conclusion, and future perspective FF tumor tissues during antibody validation, we do not known whether the observed unspecific 90 kDa band using the cell lines would have been detected by SP44 examining tumor material. Besides repositioning western blotting as an orthogonal validation method, MacNeil et al. (24) also propose to correlate the results observed using antibodies that are raised against different non-overlapping epitopes of the target of interest. This recommendation finds its origin in the work from Uhlen et al. (25) that states that a high correlation is expected between staining patterns generated by two different antibodies able to bind to different regions of the protein of interest. However, we would like to stress that caution is required when interpreting results obtained using antibodies that are raised against different non-overlapping epitopes, as observed differences can also be due to biology (e.g. ectodomain shedding) rather than nonspecificity (24). All the C-terminal immunohistochemical results presented in this thesis were obtained in a research setting. However, if the results of this thesis are to be implemented in patient care, efforts are needed to validate and implement the obtained results in a diagnostic setting. In contrast to Gruver et al. (26), who was able to detect comparable C- and N-terminal MET staining intensities (SP44 versus A2H23) using automated staining platforms (Leica and Dako) across non-small cell lung cancer WTSs, we were not able to reproduce our obtained OSCC research results using D1C2 and SP44 on the diagnostic platform available within the Erasmus MC (Ventana, unpublished results). This leads us to believe that not all diagnostic solutions are optimal for all antibodies and tumor types. Therefore, we either foresee a place for classical immunohistochemistry in routine diagnostics, or the outsourcing of immunohistochemistry to large centralized facilities having access to various automatic staining platforms. Ultimately, the most sensible choice in which antibody to use, will depend on the investigated tumor type and available lab facilities. Are TMA studies valuable for biomarker research? In this thesis, a TMA was initially used to explore the behavior of MET immunoreactivity across OSCC (chapters 3 and 4). During TMA construction, six cores were sampled from each selected cancer; three from the tumor center and three from the periphery. Although core-to-core correlations of average MET immunoreactivity in the center versus the periphery are well above 0.6 (0.710, p-value < 0.001 for D1C2 in chapter 3; 0.723, p-value < 0.001 for A2H2-3 in chapter 4; unpublished results), 8
RkJQdWJsaXNoZXIy MTk4NDMw