Investigating the complex interplay between fibroblast activation protein α-positive cancer associated fibroblasts and the tumor microenvironment in the context of cancer immunotherapy

Abstract

Introduction: This study investigates the role of Fibroblast Activation Protein (FAP)-positive cancer-associated fibroblasts (FAP+CAF) in shaping the tumor immune microenvironment, focusing on its association with immune cell functionality and cytokine expression patterns.

Methods: Utilizing immunohistochemistry, we observed elevated FAP+CAF density in metastatic versus primary renal cell carcinoma (RCC) tumors, with higher FAP+CAF correlating with increased T cell infiltration in RCC, a unique phenomenon illustrating the complex interplay between tumor progression, FAP+CAF density, and immune response.

Results: Analysis of immune cell subsets in FAP+CAF-rich stromal areas further revealed significant correlations between FAP+ stroma and various T cell types, particularly in RCC and non-small cell lung cancer (NSCLC). This was complemented by transcriptomic analyses, expanding the range of stromal and immune cell subsets interrogated, as well as to additional tumor types. This enabled evaluating the association of these subsets with tumor infiltration, tumor vascularization and other components of the tumor microenvironment. Our comprehensive study also encompassed cytokine, angiogenesis, and inflammation gene signatures across different cancer types, revealing heterogeneous cellular composition, cytokine expressions and angiogenic profiles. Through cytokine pathway profiling, we explored the relationship between FAP+CAF density and immune cell states, uncovering potential immunosuppressive circuits that limit anti-tumor activity in tumor-resident immune cells.

Conclusions: These findings underscore the complexity of tumor biology and the necessity for personalized therapeutic and patient enrichment approaches. The insights gathered from FAP+CAF prevalence, immune infiltration, and gene signatures provide valuable perspectives on tumor microenvironments, aiding in future research and clinical strategy development.

Keywords: T cell infiltration; cancer immuno-therapy; fibroblast activation protein (FAP); immune cell subsets; patient enrichment; tumor immune microenvironment.

Figure 1

(A) Representative FAP (purple) and KRT (yellow) staining pattern obtained through IHC and digital slide analysis from primary and metastatic NSCLC and RCC samples. (B) FAP content reported as a percentage of tumor area covered in primary tumors and metastatic lesions within the context of different tumor types, including BC, Squamous Cervical Cancer (CESC), CRC, HNSCC, NSCLC, and RCC (dark pink). The data shows the individual measurements (filed circles), median for each cohort (white line), the overall median value (black line) and summary statistics including the number of samples, the median coverage as well as minimum and maximum values observed. (C) Comparison of FAP stained tumor areas and CD3+ T cell infiltration in primary RCC tumors and patient-matched metastatic lesions in a cohort of seventeen patients. Level of statistically significant differences between primary tumors and metastasis indicated as ** (p<0.01) or * (p<0.05).

Figure 2

(A, B) Correlation Matrix of FAP+ Tumor Stroma Characteristics and Key Immune Cell Populations, Measured by IHC and digital slide analysis. The tumor stroma is defined as KRTnegative, and FAP staining is used to assess FAP-positive tumor stroma. Key immune cell populations are characterized by various markers, as indicated in the legend. Results are presented for all tumor types pooled (A) or representative tumor indications (B). The values presented in the matrix correspond to Spearman’s correlation coefficients. Colored scale bar indicates direction and magnitude of correlation. The Spearman correlation coefficient measures the strength and direction of a linear relationship between two variables. When the Spearman correlation coefficient is higher in magnitude (closer to 1 or -1), it indicates a stronger correlation between the two variables. A positive correlation coefficient close to 1 (yellow) suggests a strong positive linear relationship, while a negative correlation coefficient close to -1 (purple) suggests a strong negative linear relationship. On the other hand, a Spearman correlation coefficient close to 0 indicates a weak or no linear relationship between the variables. Asterisk within the squares highlight significant correlations, with significance determined at a level of ¾0.05. Inset panel presents immunohistochemistry staining of immune cells within the tumors. The immune cell populations and stromal characteristics assessed include CD3 T cells (CD3+), CD8 T cells (CD8+), proliferating CD8 T cells (MKI67+CD8+), Tregs (FOXP3+), cytotoxic T cells (PRF+CD3+), NK cells (PRF+CD3-), FAP stained stroma (FAP+), and tumor stroma (KRT-). Representative IHC images are provided in the inset.

Figure 3

(A, B) Enrichment analysis of cell types was conducted on transcriptomic datasets using the xCell gene signatures-based approach, as previously detailed (29). Highlighted are specific stromal and immune cell types in correlation with elevated FAP mRNA expression: a comprehensive view across (A) stromal, and (B) immune cells (tow row showing on top all tumor indications pooled versus tumor indication specific breakdown in bottom rows). Spearman correlation coefficients were assessed for each subset and plotted. Colored scale bar indicates direction and magnitude of correlation. A positive correlation coefficient close to 1 (yellow) suggests a strong positive linear relationship, while a negative correlation coefficient close to -1 (purple) suggests a strong negative linear relationship. On the other hand, a Spearman correlation coefficient close to 0 indicates a weak or no linear relationship between the variables. Asterisk within the squares highlight significant correlations, with significance determined at a level of FDR ¾0.05. Data from both primary tumors and metastatic lesions were aggregated to increase the overall dataset for each indication. DC - Dendritic cells and MEL, Melanoma.

Figure 4

(A, B) Comprehensive transcriptomic analysis analyses to evaluate the relationship of FAP mRNA expression with various biological states of the Tumor microenvironment (TME). These states are characterized by distinct gene signature sets, including (A) cytokine signaling pathways (sourced from REACTOME database), (B) Tumor angiogenesis indicators, infiltration markers, and inflammation indicators (derived from the IMmotion150 study) (31). Spearman correlation coefficients were assessed for each pathway and plotted. Colored scale bar indicate direction and magnitude of correlation. A positive correlation coefficient close to 1 (yellow) suggests a strong positive linear relationship, while a negative correlation coefficient close to -1 (purple) suggests a strong negative linear relationship. On the other hand, a Spearman correlation coefficient close to 0 indicates a weak or no linear relationship between the variables. Asterisk within the squares highlight significant correlations, with significance determined at a level of FDR ¾0.05. Angiogenesis Signatures: VEGFA, KDR, ESM1, PECAM1, ANGPTL4, CD34. T effector Signatures: CD8A, EOMES, PRF1, IFNG, and CD274. Myeloid inflammation Signatures: IL-6, CXCL1, CXCL2, CXCL3, CXCL8, and PTGS2. REACTOME cytokine signaling pathways: Interleukin/IL-1/2/4/6/10/12/13/17, Interferon-alpha/beta/gamma/IFN-a/b/g, Tumor necrosis factor/TNF, Transforming growth factor-beta/TGF-b.

Figure 5

(A–C) Kaplan-Meier survival curves demonstrating the impact of high FAP expression on patient outcomes across different cancer types and specific patient subgroups characterized by the presence of prognostic gene signatures (McDermott et al) (A) Renal Cell Carcinoma (RCC) data from the IMmotion151 study (Atezo in combination with Bevacizumab), comparing progression free survival rates for high versus low FAP expression across T effector signature, angiogenesis signature, and myeloid inflammation signature. (B) Squamous Non-Small Cell Lung Cancer (NSCLC) data from the IMpower131 study (Atezo in combination with chemotherapy) and (C) Non-squamous NSCLC data from the IMpower150 study (Atezo in combination with chemotherapy). Each panel displays survival probability (progression-free survival/PFS) on the y-axis against time (months) on the x-axis, with hazard ratios (HR) and p-values provided for each comparison. High FAP expression consistently correlates with poorer outcomes, particularly in RCC and squamous NSCLC across various TME-related pathways. Supplementary: Overview of Roche clinical studies included in this analysis.