57 MET immunoreactivity and poor prognosis Association between MET staining patterns and prognosis To establish whether there is a relation between the pattern of MET immunoreactivity (uniform negative, uniform positive or variable staining) and survival, each data point in the scatter plot was labelled with the 5-year OS or DFS status of the corresponding patient (Supplementary Figures S5B & S5C). The result reveals that events (red dots) cluster in the lower left and upper right corners of both scatter plots. This nonrandom distribution of events provided the basis to assign one of the three defined staining patterns to each point in the scatter plot. Therefore, it was assumed that the cluster of points in the lower left corner of the scatter plot represents cancers with uniform negative MET staining patterns. Similarly, it was assumed that the cluster of points in the upper right corner of the scatter plot represent cancers with uniform positive MET staining patterns. Furthermore, it was assumed that the points residing outside the observed clusters represent cancers with variable MET staining patterns. Exact boundaries were set for the two observed clusters – representing uniform staining – in such a way that the relative number of events (i.e. OS or DFS) within them is higher compared to the relative number of events outside them. The boundary for the uniform negative staining cluster is < 10% MET immunoreactivity in the center and periphery. The boundary for the uniform positive staining cluster is > 75% MET immunoreactivity in the center and periphery (Supplementary Figures S5B & S5C). Univariable analysis revealed that patients showing the variable staining pattern perform significantly better than patients showing either uniform staining patterns (negative or positive) in terms of 5-year OS (HR = 2.188, p < 0.001) and DFS (HR = 1.974, p = 0.001) (Figures 3A & 3B; Supplementary Figures S5D & S5E). Besides MET staining pattern, clinical and histological T and N-stage, differentiation grade, vasoinvasive growth and extranodal growth are significantly associated with poor 5-year OS (Supplementary table S5). The same applies for: age at diagnosis, clinical and histological T and N-stage and extranodal growth with respect to 5-year DFS (Supplementary table S5). 3
RkJQdWJsaXNoZXIy MTk4NDMw