154 Chapter 5 between D1C2 uniform positivity and NCH-38 uniform negativity. Besides EMT, the developed scoring system also allows investigation of MET protein status (no, decoy, TM C-terminal positive) and ECD shedding in TM C-terminal MET-positive cancers by aligning D1C2 and A2H2-3 staining patterns. Assuming that TM C-terminal MET-positive cancers are eligible for treatment with biologicals directed against MET, it was examined within this specific group whether the defined MET staining patterns and ECD shedding show a relation with survival using ROC curve analyses. As these staining patterns are highly related to one another, it was decided beforehand that only the most informative pattern per marker would be used in a final – multivariable – survival model to avoid colinearity. This approach resulted in the thresholds of ≥10% for D1C2 uniform positivity and ≥10% of ECD shedding within uniform positive patches of D1C2. The absence of an association between N-terminal MET immunoreactivity and survival is consistent with prior results (18). Using the same methodology, a relation was established between ≥10% of NCH-38 uniform negativity and survival. Considering the sample size, uniform positivity of D1C2, ECD shedding, and uniform negativity of NCH-38 were corrected using the same six variables used to correct for ECD shedding in the TMA study (18), more specifically, age at diagnosis, pT, pN, extranodal growth and degree of differentiation, all of which are known to be associated with survival (25-28). Because of the earlier observed interaction between vasoinvasive growth and the D1C2 uniform staining patterns (17), interaction between vasoinvasive growth and D1C2 uniform positivity or ECD shedding was excluded (results not shown). In addition, overcorrection by including both degree of differentiation and uniform negativity of E-cadherin in a single multivariable model was also excluded (results not shown). There are some discrepancies between the results observed using the TMA and WTSs. Using WTSs, there is no association between absence of C-terminal MET immunoreactivity and survival. In contrast to the TMA study (17), wherein each cancer was scored for one staining pattern, the WTSs can show combinations of staining patterns. This implies that although such combinations are not observed while scoring the TMA, they could be represented in the sampling tissue. Therefore, the earlier observed association of absence of C-terminal MET with poor survival is probably due to another – uniform positive – staining pattern not sampled during TMA production. Moreover, there is a difference in the thresholds set for D1C2 uniform positivity. This might be so because the TMA threshold was set including all cancers evaluated for D1C2 immunoreactivity (negative and positive). Although including only C-terminal MET-positive OSCC lowers the TMA threshold for uniform
RkJQdWJsaXNoZXIy MTk4NDMw