Tumor-infiltrating lymphocytes and the PD-L1/PD-1 signaling axis in breast cancer : biology and clinical implications

Abstract

Cancer immunotherapy and especially immune checkpoint blockade therapy has revolutionized cancer treatment in many tumor types including breast cancer (BC). The expression and specific role of Programmed Death-1 (PD-1)/Programmed Death-Ligand 1 (PD-L1) signaling axis, immune-related gene signatures and immune cell subpopulations needs to be fully elucidated while the identification of prognostic and predictive biomarkers in BC is of utmost importance. The overall aim of the thesis was to investigate the expression, prognostic/predictive implications and regulatory mechanisms of the PD-1/PD-L1 checkpoints and immune infiltrate in early and advanced BC.

In paper I, a comprehensive analysis of PD-L1 expression patterns and its prognostic implications was performed in early BC. PD-L1 protein expression was mostly expressed in immune cells and was most abundant in the TNBC subtype. PD-L1 expression in tumor cells was a poor prognostic factor whereas PD-L1 expression in immune cells was correlated with improved survival outcomes in TNBC, as revealed in a trial-level meta-analysis including 38 studies. PD-L1 gene expression was associated with improved prognosis in the entire population as revealed in a pooled transcriptomic analysis of 39 publicly available datasets. PD-L1 expression can therefore represent a promising clinically relevant biomarker of good prognosis which can also select appropriate candidates for immunotherapy.

In paper II, PD-L1 expression was evaluated both at the protein and mRNA level in the same retrospective early BC patient cohort. Both protein (stained with the antibody clone SP263) and gene expression predicted improved outcome in early BC and also correlated with enhanced T-cell infiltration (CD3+ IHC and in silico estimation of CD8+/CD4+ T-cells). Of note, PD-L1 mRNA expression can add significant prognostic value to the prospectively validated 21- and 70-gene gene signatures in ER+/HER2- disease. Upon validation, this finding might improve prognostication capacity of the current gene signatures.

In paper III, we investigated the role of STAT3 as a regulator of PD-L1 expression and immune response in BC in vitro, in vivo and in BC patient samples. A positive correlation between STAT3 and PD-L1 was observed at the protein and gene expression level while a transcriptional STAT3-mediated regulation of PD-L1 was demonstrated in BC cells. Of note, STAT3 modulated antitumor immune response mainly through macrophage phenotype shift and accumulation of NK cells rather than via cytotoxic T-cell infiltration in a murine BC model. Furthermore, pro-tumoral macrophages were correlated with PD-L1 expression in BC patient tumors, thus providing insights on the tumor-immune cell interactions and potential clinical implications.

In paper IV, a multi-level study of the prognostic capacity of PD-1 expression in early BC was performed including: a) a study cohort with IHC and GEP data, b) systematic review and trial- level IHC-based meta-analysis of 15 studies and c) pooled analysis of publicly available transcriptomic datasets). PD-1 protein and gene expression were correlated with improved OS in the entire population of the study cohort. In the pooled analyses PD-1 protein and gene expressions were correlated with better survival outcomes in TNBC/basal-like patients and therefore is a promising biomarker which merits further validation.

In paper V the role of relevant gene signatures as predictors of response to chemotherapy was investigated in patients (n=109) with advanced BC participating in the translational sub-study of the phase III TEX trial. Fine-needle aspiration biopsies were used for gene expression profiling and for TILs enumeration. Immune-related gene signatures predicted better response to chemotherapy in ER+ and luminal BC patients which was further confirmed through an independent gene set enrichment and other in silico analyses. The lymphocytic abundance was low and predicted no effect to chemotherapy. These results may pave the way for the development of immune-based drivers of chemosensitivity in the least immunogenic luminal tumors.