Patient-derived tumor explant models of tumor immune microenvironment reveal distinct and reproducible immunotherapy responses

Abstract

Tumor-resident immune cells play a crucial role in eliciting anti-tumor immunity and immunomodulatory drug responses, yet these functions have been difficult to study without tractable models of the tumor immune microenvironment (TIME). Patient-derived ex vivo models contain authentic resident immune cells and therefore, could provide new mechanistic insights into how the TIME responds to tumor or immune cell-directed therapies. Here, we assessed the reproducibility and robustness of immunomodulatory drug responses across two different ex vivo models of breast cancer TIME and one of renal cell carcinoma. These independently developed TIME models were treated with a panel of clinically relevant immunomodulators, revealing remarkably similar changes in gene expression and cytokine profiles among the three models in response to T cell activation and STING-agonism, while still preserving individual patient-specific response patterns. Moreover, we found two common core signatures of adaptive or innate immune responses present across all three models and both types of cancer, potentially serving as benchmarks for drug-induced immune activation in ex vivo models of the TIME. The robust reproducibility of immunomodulatory drug responses observed across diverse ex vivo models of the TIME underscores the significance of human patient-derived models in elucidating the complexities of anti-tumor immunity and therapeutic interventions.

Keywords: Breast cancer; IO-treatment; ex vivo model; immune checkpoint; patient-derived explants; renal cell carcinoma; tumor immune microenvironment.

Figure 1.

Ex vivo tumor landscape of breast cancer and renal cell carcinoma. a, Schematic representation of the processing of the patient-derived explant models, n=6 for each model, ‘BC1’, ‘BC2’ and ‘RCC’ b, Comparison of the estimated cell type abundances (mean ± SD) based on gene expression profiling ‘BC1’ = light pink, ‘BC2’ = dark pink and ‘RCC’ = grey. Cell abundance score refers to mean log2 normalized counts of cell type specific markers. c, Estimated cell type score per patient normalized to tumor immune infiltration, depicting relative immune cell composition for each patient. Samples clustered with Ward’s minimum variance method. Scale is z-score. d-f, Box and whisker (min to max) plot of baseline cytokine secretion of each patient presented as log₂ of the raw pg/ml (+1 as a small constant value).

Figure 2.

Effect of adaptive and innate immune modulation on cell abundances. a, Schematic of treatments selected for the study b, Box and whisker (min to max) of total estimated cell abundances after immunomodulation. c, No significant changes in CD8+ T cell numbers, but significant changes in d, exhausted CD8+ T cells e, NK cells f, cytotoxic cells g, macrophages, and h, neutrophils. Statistical significance was tested with a two-way ANOVA with Fishers LSD. All data are presented as mean values ± SD. (p= **** = <0.0001; p= *** = <0.001; p= ** = <0.01; p= * = <0.05).

Figure 3.

Ex vivo cultures reflect biological changes in response to immunomodulation. a. BC1, BC2 and RCC cytokine responses of each treatment shown as log2FC (treated explants to untreated controls). Statistical significance was tested with a two-way ANOVA with Fishers LSD. (p= **** = <0.0001; p= *** = <0.001; p= ** = <0.01; p= * = <0.05) b, heatmap of average BC1, BC2 and RCC patient pathway score determined by gene expression profiling. The scale reflects the treated pathway score – the untreated pathway score.

Figure 4.

Core innate and adaptive response signatures as potential biomarkers. a, Differentially expressed genes of pooled explant samples following Immunocult treatment. Genes above the solid like are adjusted p-value <0.01, genes above dotted line are adjusted p-value <0.05. b, venn diagram of significantly (adjusted p-value < 0.5) upregulated genes following Immunocult, shared core gene signature of 19 genes is bolded, and core genes are listed below the plot, and venn diagram of significantly downregulated genes following Immunocult c, Metascape analysis of pathways associated with the core signature, circle size refers to the number of core genes (=count) matching the pathway, and the numerical value is the percentage of all of the user-provided genes that are found in the given ontology term. “Log10(P)” is the p-value in log base 10. d, the core “Adaptive signature” in relation to existing biomarkers for immunotherapies in a cohort of anti-PD-1 treated melanoma patients from Gide et al., 2018 e, overall survival of anti-PD-1-treated melanoma patients with higher (red) vs. lower (blue) core adaptive core signature expression. f, Differentially expressed genes of pooled explant samples following ADU-S100 treatment g, venn diagram of significantly upregulated genes following ADU-S100, and core genes are listed below the plot, and venn diagram of significantly downregulated genes following ADU-S100 h, Metascape analysis of pathways associated with the core signature, circle size refers to the amount of core genes matching the pathway, and the numerical value is the percentage of all of the user-provided genes that are found in the given ontology term. “Log10(P)” is the p-value in log base 10. i, the adaptive core “innate signature” in relation to existing biomarkers for immunotherapies in a set of anti-PD-1 treated melanoma patients j, overall survival of anti-PD-1-treated melanoma patients with higher (red) vs. lower than median core innate signature expression (blue) of the core innate signature. OS, overall survival.