Conference 2021 Live Talk

 

Talk title

Architecture of the tumor microenvironment predicts response to cancer immunotherapy

 

Authors and Affiliations

Darci Phillips1,2,3,8, Magdalena Matusiak3,8, Belén Rivero Gutierrez3, Salil S. Bhate1,3 ,4, Graham L. Barlow1,3, Sizun Jiang1,3, Janos Demeter1, Kimberly S. Smythe5, Robert H. Pierce5, Steven P. Fling5, Nirasha Ramchurren5, Martin A. Cheever5, Yury Goltsev1,3, Robert B. West3, Michael S. Khodadoust6,8, Youn H. Kim2,6,8, Christian M. Schuerch1,3,7,9*, and Garry P. Nolan1,3,9*

1. Department of Microbiology & Immunology, Stanford University School of Medicine, Stanford, CA 94305, USA
2. Department of Dermatology, Stanford University School of Medicine, Stanford, CA 94305, USA
3. Department of Pathology, Stanford University School of Medicine, Stanford, CA 94305, USA
4. Department of Bioengineering, Stanford University Schools of Engineering and Medicine, Stanford, CA 94305, USA
5. Cancer Immunotherapy Trials Network, Fred Hutchinson Cancer Research Center, Seattle, WA, USA
6. Division of Oncology, Stanford University School of Medicine, Stanford, CA 94305, USA
7. Department of Pathology and Neuropathology, University Hospital and Comprehensive Cancer Center Tuebingen, Tuebingen, Germany
8. Co-first authors
9. * Co-senior / corresponding authors

 

Abstract

Background

Immunotherapies can induce long-lasting remissions in advanced-stage cancer patients, but many patients do not benefit. Novel predictive markers are needed to stratify patients before treatment to minimize adverse effects and high treatment costs. Characterizing the architecture of the tumor microenvironment (TME) at the single-cell level by highly multiplexed tissue imaging should reveal novel spatial biomarkers of immunotherapy response.

Methods

We used CODEX highly multiplexed fluorescence microscopy to investigate the TME of cutaneous T cell lymphoma (CTCL) in samples from patients treated with pembrolizumab anti-PD-1 immunotherapy. 55 protein markers were visualized simultaneously in matched pre- and post-treatment skin biopsies from 7 pembrolizumab responders and 7 non-responders. RNA sequencing was performed to extract cell-type specific gene expression profiles by CIBERSORTx analysis.

Results

CODEX enabled the identification and characterization tumor and reactive immune cells in the CTCL TME at the single-cell level, resulting in 21 different spatially resolved cell type clusters. Cluster frequencies were not significantly different between responders and non-responders. However, advanced computational analysis of the tumor architecture revealed cellular neighborhoods (CNs) that dynamically changed during pembrolizumab therapy and were correlated with response. Effector-type CNs enriched in tumor-infiltrating CD4+ T cells and dendritic cells were significantly increased after treatment in responders. In contrast, a regulatory T cell-enriched CN was significantly increased in non-responders before and after therapy. Furthermore, a spatial signature of cell-cell distances between tumor cells and effector/regulatory immune cells predicted therapy outcome. In addition, CIBERSORTx analysis revealed that tumor cells in responders, but not in non-responders, increased their expression of immune-activating genes.

Conclusions

Multidimensional analysis of CTCL tumors revealed a pre-existing immunosuppressive TME state that correlated with non-response upon pembrolizumab therapy. Thorough analysis of the TME therefore enables the discovery of novel spatial biomarkers in a concept that accounts for both cell type information and tumor architecture. These technologies will pave the way for future studies that functionally address these cell types and their interactions.