258?Scientific correlatives from LCCC 1525: a phase II study of a priming dose of cyclophosphamide prior to pembrolizumab to treat metastatic triple negative breast cancer

摘要

Background In metastatic triple negative breast cancer (mTNBC), median progression-free survival (PFS) with chemotherapy alone is approximately 2–4 months 1 and improvements with single agent checkpoint inhibitors (CI) are limited by modest response rates. Murine breast cancer models have demonstrated a role for intratumoral regulatory T cells (Tregs) in modulating response to CIs. 2 A phase II clinical trial was conducted to test the hypothesis that a single, priming dose of cyclophosphamide prior to pembrolizumab would improve PFS in mTNBC. Here we present the correlative genomic and immunologic analyses from this study. Methods This trial (https://clinicaltrials.gov/ct2/show/ NCT02768701 ) recruited 40 patients with largely pretreated mTNBC. Response was defined as 30% decrease in imaging-assessed disease burden. Clinical benefit was defined as treatment response or stable disease. Tumor specimens were collected prior to enrollment, and peripheral blood mononuclear cell (PBMC) samples taken prior to cyclophosphamide and before each cycle of pembrolizumab. RNA sequencing was performed on tumor samples for gene expression and immune repertoire reconstruction. Targeted sequencing of the T-cell beta chain, IG kappa, lambda and heavy chain (TRB, IGK, IGL, and IGH, respectively) on PBMCs captured the peripheral immune repertoire. Whole exome sequencing was performed on tumor samples with PBMCs serving as a matched normal. Results Of 40 patients enrolled, 31 patients had tumor RNA-seq and at least 15 had matched PBMC-derived immune chains capturing both pre and post treatment. When preliminary RNA-seq samples (n=22) revealed upregulation in B-cell receptor pathways and related gene signatures (figure 1), we updated our planned analysis to exclude tumor specimens collected from lymph nodes. In our final analysis, response to therapy (4 of 25, 16%) was associated in tumor RNA-Seq with gene pathways involving programmed cell death and MAPK activation, while non-responding tumors were enriched in G-protein signaling and inhibition of insulin secretion (figure 2a,b, table 1). Immune gene signatures related to NK cells and B-cell activation, signaling and interaction with T follicular helper cells, 3–7 were associated with response (figure 2g). Pre-treatment immune repertoire measures demonstrated a significant association between increased peripheral IGH abundance and richness, and both future clinical benefit and response to therapy (figure 3a-d). Abstract 258 Figure 1 Gene set enrichment and immune gene signatures in preliminary RNA-Seq samples. Demonstrating pathways (A) and gene signatures (B) associated with B cell activation as significant in patients with clinical benefit on checkpoint inhibitor therapy Abstract 258 Figure 2 Tumor genomic and immune features. A-B: Differential gene expression (A) and gene set enrichment (B) results in non-nodal tumor samples, by treatment response. Genes in (A) passing FDR correction (Benjamini Hochberg) are labeled and in red. C: Frequently mutated genes implicated in breast cancer, samples sorted by response. Raw tumor mutational burden is noted at the top of each sample column. Treatment response and tumor PAM50 subtype for each sample is listed at bottom of each column. D-E: Sample PD-L1 was not significantly associated with either clinical benefit (D) or response (E) to therapy (T-test; proportion of sample staining with 22C3 antibody). F-G: Immune gene signatures significantly associated with clinical benefit (F) and response (G) in non-nodal tumor samples Abstract 258 Figure 3 Tumor and peripheral immune repertoire diversity. A-D: In tumor RNA-Seq, higher IGH chain abundance and richness was associated with both clinical benefit (A, C) and response (B, D) (n=31). E-F: Inter-group comparisons showed fewer TRB chain similarities between patients who derived clinical benefit (E) or response (F) versus those who did not, in pre-treatment PBMC samples. G-I: Univariate Cox proportional hazards models for PFS showing immune diversity measures derived from pre-treatment tumor RNA-Seq (G), PBMC-derived amplicon sequencing pre-pembrolizumab (H), and PBCM-derived amplicon sequencing post-pembrolizumab (I) Abstract 258 Table 1 Gene set pathways, C2CP set from MSigDB. To accompany gene sets in Figure 2b Conclusions Response to CI therapy was associated with immunogenomic features of programmed cell death and B-cell activation. Pre-treatment circulating immunoglobulin diversity measures (high IGH abundance and IGH richness) also correlated with future response to therapy. Taken together, these data suggest that B-cell activity contributes significantly to response to CI therapy in mTNBC. Acknowledgements UNC Office of Clinical and Translational Research (OCTR), High Throughput Sequencing Facility (HTSF), and UNC Bioinformatics Core. We also thank the patients in this study and their families, without whom this study would not have been possible. Trial Registration Clinical Tria