Fig. 3. Pan-cancer assessment of gene-expressed based immune scores as predictive biomarkers for pembrolizumab.
A Distribution of immune (IM) score within each pembrolizumab sensitivity group. Scores derived from baseline and cycle 3 tumor samples are shown. Lines connect samples collected from the same patient. P-values were calculated by two-sided Wilcoxon rank-sum tests. B Comparison of IM, C Interferon gamma (IFNG), and D cytolytic (CYT) scores between LS and HS/CB groups at baseline and cycle 3 of treatment. For all violin plots in A–D, the median for each group is shown as a solid horizontal line dot and data points are shown as black dots. The distance between the third-quartile (Q3) and first-quartile (Q1), known as the interquartile range (IQR), is marked around the median by a black rectangle. Vertical lines extending from the top and bottom of the rectangle show the maximum (Q3 + 1.5-times IQR) and minimum (Q1 + 1.5-times IQR). Statistical significance was determined using two-sided Wilcoxon rank-sum tests between indicated sample groups. E Forest plot summaries of comparisons between HS/CB and LS baseline tumor gene-expression derived immune activity/infiltrating immune cell signatures. Two-sided Wilcoxon rank-sum tests were performed to assess statistical significance of the observed differences in signatures between groups. For each score, the difference between group means (MeanHS/CB−MeanLS) is shown as a solid dot with whiskers indicating the 95% confidence interval. P values are uncorrected for multiple testing. Source data are provided in SourceData_Fig3.xlsx. LS low sensitivity, HS/CB high-sensitivity/clinical benefit.