Figure 3.

Subgroup Analysis of Immune Cell Profile by Whole-Transcriptome and Immunofluorescence Analyses

(A) Subgroup-specific distribution of immune cell subsets shown after CIBERT calculation using RNA sequencing expression profiles, indicating heterogeneous frequencies for each subset. Prominence of CD8 T regulatory T cells in subgroup 2, and B cell naive and memory resting CD4 T cells in subgroup 3. Major features tabulated in Figure 3C.

(B) Comparison of three subgroup-specific scores, namely, stromal, immune, and cytolytic activity scores (Data S3). Subgroups 1, 2, and 3 were the lowest, intermediate, and highest, respectively. Subgroup 2 had higher levels of CD8 T and regulatory T cells, yet had lower cytolytic activity scores than subgroup 3, consisting mostly of memory resting CD4 T cells. (∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001, Student's t test.)

(C) Most immune cells were not activated in subgroup 1. NK cell activation and high immune score were common features between subgroups 2 and 3. Subgroup 2 had relatively more CD8 T cells, Treg cells, and lower stromal score, whereas subgroup 3 had significantly more memory resting CD4 T cells and higher stromal score. A bar graph represents the quantification of the immune cell profile in each subgroup. Each proportion or score was normalized by Z score across samples, and then averages were taken for each subgroup.

(D) Multiplexed immunofluorescence images showing distribution of T lymphocyte subsets, clearly distinguishable among subgroups. Very few CD8 T and CD4 T cells in subgroup 1. Subgroup 2 showing sparse CD8 T (arrow C) and CD4 T cells (arrow A) outside the tumor nests, and more CD4 T and memory T cells (CD45RO, arrow B) in subgroup 3 (cyan, pan-cytokeratin in tumor cells; green, CD4; red, CD8; yellow, FOXP3; orange, PD-L1; magenta, CD45RO; blue, DAPI).