Fig. 5

Lymphocytic infiltration and PI3K pathway activation in breast tumors. a Heat map represents unsupervised hierarchical clustering of 215 breast carcinomas (columns) based on the expression of lymphocyte markers and the activation of PI3K pathway on tumor cells (rows). Information on percentage of ER positivity (from 10 to 100%) and tumor grade (1, 2, or 3) is indicated by gray scale: low as lighter gray and high as darker gray. Boxes of PIK3CA mutation status are filled with gray (wild-type), black (mutated), or white (unknown). CD4, CD8, and FOXP3 status was filled with color (high) or gray (low). The forest plots represent multivariable linear regression models. Each plot merges 6 multivariable models (described in Additional file 2: Tables S2A-F) calculating the association between the (phospho-) levels of PI3K downstream proteins and CD4 (b), CD8 (c), or FOXP3 (d) expression. Measurements of these associations are represented by the coefficients plotted on the x axes. The size of each dot is proportional to the number of samples used for each multivariable complete case analysis, and bars represent the standard deviation