LCM enabled us to precisely dissect areas of immune cell infiltrates in FFPE tissue sections, for comparative gene expression profiling of immune cells infiltrating cancer vs non-cancerous inflamed tissues (Fig.
2A). We employed multiplex qRT-PCR, a robust assay for use with partially degraded RNAs from autopsy specimens, to query 122 candidate genes characterizing immune cell subsets, cytokines/chemokines, and immunomodulatory receptor-ligand pathways (Additional file
1: Table S1). An unsupervised heatmap analysis normalizing gene expression (Ct) values to immune cell content (
PTPRC, a.k.a. CD45, a pan-immune cell marker) revealed many phenotypic and functional similarities between inflammatory infiltrates in cancerous and non-cancerous tissues (Fig.
2B). This suggested similar immunologic mechanisms underlying anti-tumor and anti-self immunity in patient MA-6. For instance, genes associated with the major histocompatibility complex and antigen presentation (
B2M, CD74, HLA-DRA), interferon signaling (
CXCL9, STAT1), immunosuppression (
IL10RA, HAVCR2 [TIM3],
LGALS9 [galectin-9],
TGFB1, TIGIT), and innate immunity (
RIG1,
LYZ,) were abundantly expressed across all samples. Conversely, expression of genes related to neutrophils (
CEACAM8 [CD66b],
CSF2 [GM-CSF]), Th2 cells (
IL4, IL13) and Th17 cells (
IL17A, IL22, IL22RA2, IL23A) was low or undetectable across all samples, suggesting that these cell subsets were unlikely to be involved in irAEs or anti-tumor immunity in patient MA-6; the successful detection of these genes with identical assay conditions in other studies in our laboratory diminishes the possibility of technical failure. Consistent with our findings, a dominance of Th1 but not Th17-associated genes has been reported by others in irAE dermatitis and colitis specimens [
10].
The B-cell homing chemokine CXCL13, which resides in the IL-21/CXCL13 auto-antibody axis and is also produced by tumor mutation-specific CD8 + TILs in lung cancer [
11], exhibited moderate expression across specimens in our study, implying a role for B cells in both anti-tumor and anti-self immunity in this patient.
Sixteen genes were up-regulated in immune infiltrates from non-cancerous inflamed (n = 4) vs tumor (n = 3) tissues (Fig.
2C and Additional file
1: Table S2
; expression fold change magnitude ≥ 2.0 and p ≤ 0.10 normalized to
PTPRC). Notably, 12/16 up-regulated genes encode molecules associated with immunosuppressive signaling pathways and cellular subsets, including
CSF1R, IL10RA, IL27/EBI3 [IL-27 heterodimer]
, KLRG1, SOCS1; Treg hallmarks (
CD4, FOXP3, TGFB1); and the COX-2/PGE2 pathway (
IL-1B, PTGER1 [EP1, prostaglandin E2 receptor 1], and
PTGER4 [EP4, prostaglandin E2 receptor 4]). Other differentially expressed genes are associated with lymphocyte activation (
CD70, CD84, ICOSLG). COX-2/
PTGS2, with known immunosuppressive functions [
12], was up-regulated in tumor tissues when expression was normalized to
GUSB (fold change 4.16, p = 0.09) rather than
PTPRC, implicating non-immune cells (e.g., tumor and/or stromal cells) as an important source of this enzyme in the tumor microenvironment (TME) (Additional file
2: Figure S1, and Additional file
1: Table S2). Two additional immunosuppressive/pro-carcinogenic factors,
IL6 and
ENTPD1 (CD39), known to be expressed broadly by both immune and non-immune cells such as cancer-associated fibroblasts [
13] and endothelial cells [
14], were also upregulated in tumor vs inflamed non-cancerous tissues with
GUSB normalization. The predominantly immunosuppressive gene expression profile up-regulated in non-cancerous inflamed tissues after ICB likely reflects homeostatic feedback inhibition mechanisms that engage after acute immune activation, but may have also dampened anti-tumor immunity in this patient.