236 Chapter 8 dation using WTSs (chapter 5) that positioned the TMA results in the context of a routine diagnostic setting. As such, it is stated here that although highly relevant for biomarker exploration, TMA studies should always be validated using WTSs. The association between MET immunoreactivity and prognosis, room for improvement? Because both moAbs and TKIs can target MET when its located at the cell membrane (20), one of the major focus points in this thesis was to determine whether reliable membranous MET immunoreactivity is associated with prognosis in OSCC. Based on the results observed using TMAs (chapters 3 and 4) and WTSs (chapter 5), it can be stated that membranous D1C2 uniform positivity and MET-EC- are associated with poor OS and DFS. Yet, a growing awareness was developed regarding the use of regression models during the work performed to accomplish this thesis. In hindsight, the use of backward elimination (33) to define the most optimal multivariable survival models that assess the independent association of the D1C2 uniform staining pattern with OS and DFS in chapter 3 was suboptimal. In retrospect, there was enough statistical power to correct for all clinico-histopathological variables of interest, i.e. those significantly associated with OS and DFS (Supplementary table 3, chapter 3), and those predicted to be confounders (Supplementary tables 4 and 5, chapter 3), while excluding those variables presumed to have a high correlation with the ones included in the model (e.g. cT/pT and cN/pN). Redoing multivariable survival analyses, illustrated that the MET uniform staining pattern remained independently associated with both OS and DFS and that the accompanying HRs were in the same order of magnitude as those presented in chapter 3 (unpublished results). Moreover, extranodal growth remained independently associated with both survival measures, confirming the conclusions drawn in chapter 3 with respect to survival analyses. So, despite the fact that the survival analyses in chapter 3 were suboptimal, we dare to state that they had no major impact on the obtained results. In this thesis, it was assumed that only patients positive for transmembranous C-terminal MET are eligible for treatment with MET targeted therapies, as they develop cancers that are positive for the protein’s catalytic domain (chapter 4, 5, and 7). Therefore, in chapters 4 and 5, the prognostic value of D1C2 uniform positivity and MET-EC- was only examined for this subpopulation of OSCC patients. However, while assigning each cancer to one of three MET categories (no MET, decoy MET, or transmembranous C-terminal MET) in chapter 5, we did not consider that
RkJQdWJsaXNoZXIy MTk4NDMw