Abstract PD14-04: Contribution of tumor and immune cells to PD-L1 as a predictive biomarker in triple-negative breast cancer (TNBC): Analysis from KEYNOTE-119

Prognostic Significance of CSF-1R Expression in Early Invasive Breast Cancer

Colony-stimulating factor-1 receptor (CSF-1R) signaling promotes an immune suppressive microenvironment enriched in M2 macrophages. Given that CSF-1R inhibitors are under investigation in clinical trials, including in breast cancer, CSF-1R expression and association with immune biomarkers could identify patients who derive greater benefit from combination with immunotherapies. TIMER2.0 and bc-GenExMiner v4.7 were used to assess the correlation of CSF1R mRNA with immune infiltrates and prognosis. Following a prespecified training-validation approach, an optimized immunohistochemistry assay was applied to assess CSF-1R on carcinoma cells and macrophages on breast cancer tissue microarray series representing 2384 patients, coupled to comprehensive clinicopathological, biomarker, and outcome data. Significant positive correlations were observed between CSF1R mRNA and immune infiltrates. High carcinoma CSF-1R correlated with grade 3 tumors >2 cm, hormone receptor negativity, high Ki67, immune checkpoint biomarkers, and macrophages expressing CSF-1R and CD163. High carcinoma CSF-1R was significantly associated with poor survival in univariate and multivariate analyses. Adverse prognostic associations were retained in ER+ cases regardless of the presence of CD8+ T cells. CSF-1R+ macrophages were not prognostic. High carcinoma CSF-1R is associated with aggressive breast cancer biology and poor prognosis, particularly in ER+ cases, and identifies patients in whom biomarker-directed CSF-1R therapies may yield superior therapeutic responses.

Seeing the forest and the tree: TILs and PD-L1 as immune biomarkers

Here we will provide an immune-focussed overview of biomarkers in early and advanced stage breast cancer. It should be noted from the outset that all the biomarkers under discussion here have not been tested in prospective clinical trials to determine their predictive performance. Such trials require very large sample sizes due to the statistical burden of testing an interaction between a treatment and a biomarker, which is compounded by the heterogeneous biology of breast cancer (Polley et al. in J Natl Cancer Inst 105:1677-1683 2013 [1]). For a detailed discussion of the immunobiology of breast cancer, analytical aspects of these biomarkers, emerging biomarkers such as tumour mutation burden and detailed immunotherapy clinical trial data, see other articles in this issue.