Abstract P6-10-09: Multispectral immunofluorescence (mIF) to detect dynamic changes in PD-L1 expression, immune cell (IC) infiltration, and tumor-IC interactions in primary breast cancer (BC) immuno-oncology (I-O) clinical trials

Abstract

Abstract Introduction: Anti-PD-1/L1 (atezolizumab) is effective in first-line, PD-L1-positive triple negative breast cancer, however it is becoming increasingly apparent that the majority of breast cancers will require novel I-O combination strategies. We describe a framework for use of multispectral immunofluorescence (mIF) as a high-dimensional biomarker to facilitate discovery in BC I-O clinical trials. Specifically, we propose a method for recapitulating the prognostic stromal tumor infiltrating lymphocyte (sTIL) score and PD-L1 status using mIF, but with higher resolution to detect dynamic changes. Furthermore, mIF provides spatial and phenotypic data on the cellular level, allowing for comprehensive characterization of tumor-IC interactions, which may provide insight into the mechanisms of action of I-O therapy. Methods: In a recent ESBC I-O pre-operative clinical trial, a regimen combining low-dose systemic chemotherapy (cyclophosphamide) with intralymphatic cytokine injections (IRX-2) was evaluated for pharmacodynamic immune activity, as measured by dynamic changes in ICs and PD-L1 status. Pre-treatment and resection tissues were analyzed for sTIL score (H&E) and PD-L1 status (Ventana SP142). mIF was conducted (PerkinElmer Vectra) using a validated antibody panel that quantifies and provides geospatial coordinates of tumor cells (cytokeratin/CK+), ICs (CD3, CD8, CD163, FOXP3), and per-cell PD-L1 expression (QIF score). Multiple regions of interest (ROI, 20x) were captured per specimen (mean 18, range: 9-32 per specimen). Hierarchical regression employed to correct for heterogeneity across ROIs and patients. Results: By the H&E sTIL score, the IRX-2 regimen was associated with increases in sTILs of 139% (p=0.001). Using mIF, we found highly concordant increases in sTILs. mIF allowed for changes in cell counts across different IC phenotypes to be accurately quantified, allowing for lymphocytes to be highlighted while excluding myeloid mononuclear cells in the sTIL quantification. After adjusting for intratumoral and intrapatient heterogeneity with mixed effects modeling, effector T-cells (CD3+CD8+) increased by +173% (95% CI: +93% to +387%) and helper T-cells (CD3+CD8-FOXP3-) increased by +120% (95% CI: +21% to +401%). Without addressing heterogeneity with mixed- effects modeling both CD3+CD8+ and CD3+ CD8- T cells were underestimated (46% vs 40% raw cell avg). By the SP142 assay, IRX-2 regimen was associated with PD-L1 conversion in 36% (n=5/14) of cases, and PD-L1 up-classification in 79% (n=11/14) of cases. PD-L1-positive IC count by mIF was highly concordant with SP142 (Jonckheere-Terpstra test = 171, p<.001). By mIF, PD-L1 + IC increased by 337% with therapy (4.37-FC, p<0.001). Three spatial analysis metrics were evaluated: nearest neighbor distance (NND), stromal-tumor border density gradient, and clustering. Using these metrics we uncovered spatial pharmocodynamic effects including increases in clustering of helper cells to effector cells (z=25.2, p<.001), co-localization of effector cells to the tumor-stromal boundary (p=0.001), and penetration of effector cells into tumor nests (p=0.01). Conclusion: We describe a framework for use of mIF as a high-dimensional biomarker for use with I-O clinical trials to gain mechanistic insight into the biologic effects of I-O therapies in primary BCs. mIF can be used to detect dynamic changes in sTIL score and IC PD-L1 expression, and therefore may serve to complement the H&E sTIL score and SP142 PD-L1 clinical assay in the context of I-O clinical trials. The greater precision of regression-assisted mIF may facilitate discovery of treatment-related I-O effects, potentially with smaller sample size requirements. Citation Format: Katherine Sanchez, Isaac Kim, ShuChing Chang, Maritza Martel, Yaping Wu, Brady Bernard, Joanna Pucilowska, Zhaoyu Sun, William Redmond, Dottie Waddell, Walter Urba, David B Page. Multispectral immunofluorescence (mIF) to detect dynamic changes in PD-L1 expression, immune cell (IC) infiltration, and tumor-IC interactions in primary breast cancer (BC) immuno-oncology (I-O) clinical trials [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P6-10-09.