Dynamic single-cell systemic immune responses in immunotherapy-treated early-stage HR+ breast cancer patients

Abstract

The limited added benefit of immune checkpoint inhibitors in breast cancer indicates the pressing need to identify biomarkers of response to minimize risk and maximize benefit. We used single cell RNA sequencing and T cell receptor (TCR) sequencing of peripheral blood mononuclear cells (for 28 samples comprising 79,284 cells) to monitor the peripheral immune dynamic of an exploratory cohort of hormone receptor positive breast cancer patients treated with neoadjuvant nab-paclitaxel+pembrolizumab with the ultimate goal of identifying potential peripheral blood predictive biomarkers. In responsive patients, Granzyme B positive (GZMB+) cytotoxic CD8 T cells expanded post-nab-paclitaxel+pembrolizumab, accompanied by rapid changes in TCR clones. In contrast, non-responders' peripheral T cells may experience terminal exhaustion and are not significantly altered by treatment. In addition, B cell and monocyte-specific interferon response signatures were also associated with response. Our data suggests that peripheral immunological signatures may represent a facile way to monitor dynamic antitumor immune response.

Fig. 1. Chemotherapy induced a decrease in monocytes and inflammatory response signatures that were rescued by ICI.

A Single-cell RNA sequencing was performed on longitudinal blood samples collected from seven patients, with nine major cell clusters identified. B) ssGSEA score for Hallmark Interferon response and Complement response calculated after aggregating the expression profile as a pseudobulk sample, a paired Wilcoxon test was performed. C The percentage of classical monocyte (left) and non-classical monocyte (right) out of the number of PBMC detected was compared across timepoints using a paired Wilcoxon test.

Fig. 2. Abundance of GZMB+ late-activation/EM CD8 T cells is negatively associated with response.

A Umap visualization of major CD8 T cell clusters, separated by the patient’s residual cancer burden after immunotherapy treatment. Responders (RCB 0/I) are depicted at the top and non-responders (RCB II/III) are at the bottom. B Representative gene marker for each major CD8 T cell cluster depicted in the dotplot. C The percentage of clonal CD8 T cells (left panel) and total CD8 T cells carrying tumor-associated TCRs (right panel) were identified. Naïve T cell does not carry any tumor-associated TCRs, and thus are not presented on the right panel. A paired Wilcoxon test was performed between T cell subtypes. D Abundance of CD8 T cells subtypes was compared between RCB0/I and RCB II/III patients, anova was performed. E Venn diagram summarizing clonal TCR overlaps across four sampling timepoints, with clonal defined as >3T cells sharing the same TCR CDR3 sequence. F Percentage of clonal CD8 T cells compared between RCB0/I and RCB II/III patients, anova was performed.

Fig. 3. Temporal dynamics of monocytes and B cell during treatment.

A Umap visualization of major monocyte clusters, with representative markers depicted by dotplot. B Differential gene expression analysis performed on total monocytes between RCB0/I and RCB II/III patients at baseline. C ssGSEA interferon response score calculated for individual monocytes, t-test was performed. D B cell abundance compared longitudinally, a paired T-test was performed. E Differential gene expression analysis was performed on total B cells between RCB0/I and RCB II/III patients at baseline. F Differential gene expression analysis performed on total B cells between RCB0/I and RCB II/III patients post-chemotherapy. For all volcano plots, Log2 fold change cutoff at 0.5, padj cutoff at 1e-5.

Fig. 4

Summary of peripheral immunological features associated with Nab-paclitaxel + pembrolizumab response in HR+ breast cancer.