Comprehensive analysis of fibroblast activation protein expression across 23 tumor indications: insights for biomarker development in cancer immunotherapies

Introduction

Fibroblast activation protein (FAP) is predominantly upregulated in various tumor microenvironments and scarcely expressed in normal tissues.

Methods

We analyzed FAP across 1216 tissue samples covering 23 tumor types and 70 subtypes.

Results

Elevated FAP levels were notable in breast, pancreatic, esophageal, and lung cancers. Using immunohistochemistry and RNAseq, a correlation between FAP gene and protein expression was found. Evaluating FAP's clinical significance, we assessed 29 cohorts from 12 clinical trials, including both mono and combination therapies with the PD-L1 inhibitor atezolizumab and chemotherapy. A trend links higher FAP expression to poorer prognosis, particularly in RCC, across both treatment arms. However, four cohorts showed improved survival with high FAP, while in four others, FAP had no apparent survival impact.

Conclusions

Our results emphasize FAP's multifaceted role in therapy response, suggesting its potential as a cancer immunotherapy biomarker.

Predictive potential of angiopoietin-2 in a mCRC subpopulation treated with vanucizumab in the McCAVE trial

Introduction

Angiopoetin-2 (Ang-2) is a key mediator of tumour angiogenesis. When upregulated it is associated with tumour progression and poor prognosis. Anti-vascular endothelial growth factor (VEGF) therapy has been widely used in the treatment of metastatic colorectal cancer (mCRC). The potential benefit of combined inhibition of Ang-2 and VEGF-A in previously untreated patients with mCRC was evaluated in the phase II McCAVE study (NCT02141295), assessing vanucizumab versus bevacizumab (VEGF-A inhibitor), both in combination with mFOLFOX-6 (modified folinic acid [leucovorin], fluorouracil and oxaliplatin) chemotherapy. To date, there are no known predictors of outcome of anti-angiogenic treatment in patients with mCRC. In this exploratory analysis, we investigate potential predictive biomarkers in baseline samples from McCAVE participants.

Methods

Tumour tissue samples underwent immunohistochemistry staining for different biomarkers, including Ang-2. Biomarker densities were scored on the tissue images using dedicated machine learning algorithms. Ang-2 levels were additionally assessed in plasma. Patients were stratified by KRAS mutation status determined using next generation sequencing. Median progression-free survival (PFS) for each treatment group by biomarker and KRAS mutation was estimated using Kaplan-Meier plots. PFS hazard ratios (and 95% confidence intervals) were compared using Cox regression.

Results

Overall low tissue baseline levels of Ang-2 were associated with longer PFS, especially in patients with wild-type KRAS status. In addition, our analysis identified a new subgroup of patients with KRAS wild-type mCRC and high levels of Ang-2 in whom vanucizumab/mFOLFOX-6 prolonged PFS significantly (log-rank p=0.01) by ~5.5 months versus bevacizumab/mFOLFOX-6. Similar findings were seen in plasma samples.

Discussion

This analysis demonstrates that additional Ang-2 inhibition provided by vanucizumab shows a greater effect than single VEGF-A inhibition in this subpopulation. These data suggest that Ang-2 may be both a prognostic biomarker in mCRC and a predictive biomarker for vanucizumab in KRAS wild-type mCRC. Thus, this evidence can potentially support the establishment of more tailored treatment approaches for patients with mCRC.

86 Extensive FAP expression analysis in 23 tumor indications and potential application in defining the patient population in FAP-targeting cancer immunotherapies

Background

Fibroblast activation protein alpha (FAP) is frequently over-expressed in the tumor microenvironment (TME) while exhibiting limited expression in normal tissues. FAP expression was reported to be immunosuppressive in tumor mouse models and generally associated with worse prognosis in clinical studies. Therefore, it is important to understand the context in which FAP both exhibits immunosuppressive characteristics and be a useful target for immunotherapy.

Methods

Comprehensive immunohistochemistry (IHC) analyses on formalin-fixed paraffin-embedded tissue specimens with emphasis on lymph nodes and primary and metastatic tumor lesions spanning a wide range of indications were undertaken in this study. FAP staining of tumor tissues was performed with an optimized IHC robust-prototype-assay (RPA) and manually scored. The area (normal stroma & neoplastic) staining positively relative to the total tumor area at each intensity level was recorded and an H-score calculated (FAP-intensity score).These were supplemented by gene expression analysis using public as well as Roche phase 1, 2 and 3 cancer immunotherapy (CIT) clinical trial data sets.

Results

Analysing FAP expression on normal tissue confirmed the general absence of FAP apart from a subset of pancreatic islet cells. Unlike the more homogenous expression of typical protein targets on tumor cells, FAP expression in the TME is heterogeneous in both pattern and intensity, requiring the analysis of a large sample set. Therefore, we evaluated 1216 samples from 23 tumor indications and 70 sub-indications. FAP expression exhibited a significant spread ranging from indications with highly abundant expression to those with low coverage.Using data from matching IHC and gene expression samples we confirmed FAP mRNA expression to significantly correlate with RPA H-scores (Spearman correlation: 0.62) (N=289, P=1.2E-31). Gene expression data from 12 atezolizumab clinical studies, including standard of care (SOC) randomized studies, with more than 6000 samples from 4 major indications were interrogated for the association between FAP expression and clinical response as evaluated by overall and progression free survival. This analysis suggests that FAP expression is generally associated with higher hazard ratios across all atezolizumab-treated samples (OS: 95% CI 1.04–1.09; PFS: 1.04–1.08), with the highest effect observed in Renal Cell Carcinoma (OS: 95% CI 1.08–1.31; PFS: 1.05–1.21), indicating a potential role of FAP in limiting CIT.

Conclusions

Data from these analyses can tailor indication and patient enrichment strategies for achieving optimal FAP-targeting. We propose to select indications with FAP-levels that are high enough to enable drug accumulation, yet low enough to reduce immunosuppressive effects that can hamper successful immunotherapy.

930 Fibroblast activation protein alpha expression in tumor stroma and its association with immuno-regulatory circuits across epithelial tumors

Background

Carcinoma associated fibroblasts (CAFs) play important roles in modulating tumor development and prognosis through biochemical and biomechanical signals, but also through their immuno-modulatory characteristics. Fibroblast activation protein alpha (FAP), a serine protease with selectively high expression on CAFs, may be an ideal target for therapeutic intervention, including cancer immunotherapy. Therefore, a thorough understanding of FAP expression, but also immune cell composition and especially their interaction is key to optimally inform drug development and patient enrichment strategies.

Methods

Formalin-fixed paraffin embedded tissue specimens comprising 253 primary tumors and 277 metastatic lesions were included in this study. Tumor sections were analyzed by digital immunohistochemistry (IHC) to assess tumor-stroma composition, FAP content and immune cell infiltration, complemented by transcriptomic analyses.

Results

Across different types of epithelial tumors, FAP was detected by digital IHC in the tumor-associated stroma at a low to moderate proportion and with heterogeneous distribution patterns. Primary tumors in breast and lung cancer demonstrated a higher median FAP content (6.5% and 6.6% area coverage, respectively) compared to renal cell carcinoma (0.2% area coverage), which was confirmed on mRNA expression level. Median FAP levels were similar between primary and metastatic lesions in most tumor types except for renal cancer, for which FAP levels were significantly increased in metastasis lesions (3.3% area coverage). FAP content positively correlated with the density of FoxP3 positive regulatory T cells, but indication and tissue type specific differences were observed. Transcriptomic analysis revealed that both stroma-richness as well as higher FAP content were positively correlated with macrophage and dendritic cell gene signatures. However, while a higher stromal content was associated with signatures related to endothelial cells and preadipocytes, higher FAP content showed a stronger correlation with regulatory T cells. These findings are suggestive of a distinct biological role of FAP positive stroma in human tumors.

Conclusions

FAP-targeted therapy is a promising strategy to optimize accumulation and action of anti-cancer drugs in the tumor microenvironment, potentially leading to more specific and effective therapies. Our study further elucidates the role of FAP by providing a comprehensive and granular landscape of FAP content in primary and metastatic tumor lesions derived from the same patient population and its association with immune cell composition. Future studies aim to elucidate the complex and dynamic interplay between malignant, stromal and immune cell populations in both temporal and spatial contexts and how that contributes to outcome in cancer immunotherapy.

Vanucizumab mode of action: Serial biomarkers in plasma, tumor, and skin-wound-healing biopsies

Vanucizumab is a novel bispecific antibody inhibiting vascular endothelial growth factor (VEGF-A) and angiopoietin-2 (Ang-2) that demonstrated safety and anti-tumor activity in part I of a phase I study of 42 patients with advanced solid tumors. Part II evaluated the pharmacodynamic effects of vanucizumab 30 or 15 mg/kg every 2 weeks in 32 patients. Serial plasma samples, paired tumor, and skin-wound-healing biopsies were taken over 29 days to evaluate angiogenic markers. Vanucizumab was associated with marked post-infusion reductions in circulating unbound VEGF-A and Ang-2. By day 29, tumor samples revealed mean reductions in density of microvessels (-32.2%), proliferating vessels (-47.9%) and Ang-2 positive vessels (-62.5%). Skin biopsies showed a mean reduction in density of microvessels (-49.0%) and proliferating vessels (-25.7%). Gene expression profiling of tumor samples implied recruitment and potential activation of lymphocytes. Biopsies were safely conducted. Vanucizumab demonstrated a consistent biological effect on vascular-related biomarkers, confirming proof of concept. Skin-wound-healing biopsies were a valuable surrogate for studying angiogenesis-related mechanisms.

Abstract B47: Unraveling tumor metabolism with in silico IHC multiplexing supported by automated imaging analysis

The aim of our study is to understand and characterize tumor biology and its metabolism by evaluating some of the most relevant hallmarks of cancer like hypoxia, apoptosis, tumor proliferation, angiogenesis and immune status.

We selected from our tissue bank FFPE blocks from 3 main tumor indications (CRC, GC, RCC) analyzed with 9 different markers: CA9, CC3, Ki67, CD34, aSMA, Podoplanin, CD3, CD8 and FOXP3 in consecutive sections. We applied 2 triplex, one duplex and one single IHC assays. After quality check made by board certified pathologist, all slides were scanned with the iScan-HT scanner (VENTANA) at 20x magnification and the whole slide sections were analyzed with help of automated imaging algorithms developed in house.

Tumors harbor different metabolic states. Our cutting-edge approach enables comprehensive understanding of the interactions among different tumor components with respect to tumor metabolism and immune infiltration. Our automated image analysis tool is able to reconstruct and understand the architectural patterns of different tumor metabolic states and to show distribution of hypoxia (CA9), proliferation (Ki67), cell death (CC3), angiogenesis (CD34/aSMA/Podoplanin) and immune cells (CD8/CD3, FOXP3).

Deep understanding of tumor biology and metabolism is crucial for guiding drug development and overcoming challenges in clinical translation.

Citation Format: Suzana Vega Harring, Marta Canamero, Konstanty Korski, Georges Marchal, Oliver Grimm, Claudia Ferreira, Joerg Bredno, Christophe Chefd`hotel, Fabien Gaire. Unraveling tumor metabolism with in silico IHC multiplexing supported by automated imaging analysis. [abstract]. In: Proceedings of the AACR Special Conference: Metabolism and Cancer; Jun 7-10, 2015; Bellevue, WA. Philadelphia (PA): AACR; Mol Cancer Res 2016;14(1_Suppl):Abstract nr B47.

Abstract A59: A novel bispecific FAP-DR5 antibody inducing potent and tumor-specific death receptor 5 (DR5) activation by fibroblast activation protein (FAP) dependent crosslinking

Background: Activation of the extrinsic apoptosis pathway in tumor cells through agonistic death receptor 5 (DR5) antibodies has been evaluated in the clinic with limited success so far. In this context, several reports show that DR5 activation is strongly dependent on receptor hyperclustering on the cell surface. Therefore a therapeutic principle that induces DR5 hyperclustering specifically at the tumor site may provide superior efficacy, potency and safety compared to conventional DR5 agonistic antibodies. Fibroblast activation protein (FAP) is a marker for activated fibroblasts and abundantly expressed in cancer associated fibroblasts of various epithelial tumor indications and as a tumor antigen on tumors of mesenchymal origin. Due to its relative absence from normal tissues, FAP can be used as a tumor targeting antigen. Here, we are using the broad expression of FAP in tumor stroma for crosslinking of DR5 by a bispecific antibody.

Aim: In order to achieve superior tumor targeting and tumor located DR5 hyperclustering we have generated a bispecific antibody, RG7386, comprised of an agonistic DR5 binder and a FAP targeting moiety.

Results: RG7386 shows potent and selective binding to FAP and DR5 and can simultaneously bind to both targets. In in vitro co-culture assays, using human DLD1 colon tumor cells and FAP expressing fibroblasts, RG7386 induces potent, FAP dependent DR5 hyperclustering and apoptosis induction in DR5 positive tumor cells (IC50: 0.05 nM). In preclinical in vivo models with co-injection of DLD-1 tumor cells and fibroblasts as well as patient-derived colorectal cancer models, RG7386 shows FAP dependent efficacy and apoptosis induction superior to conventional DR5 antibodies. Furthermore the superior induction of apoptosis could be confirmed by in vivo and ex vivo analysis of cleaved Caspase-3 with imaging, Luminex and histopathology.

Conclusion: RG7386 is a promising novel therapeutic entity for the treatment of solid tumors with FAP positive tumor stroma inducing DR5 activation by FAP dependent DR5 hypercrosslinking which results in potent anti-tumor activity.

Citation Format: Katharina Wartha, Barbara Weiser, Thomas Friess, Meher Majety, Valeria Runza, Frank Herting, Thomas Weber, Werner Scheuer, Suzana Vega Harring, Hadassah Sade, Huifeng Niu. A novel bispecific FAP-DR5 antibody inducing potent and tumor-specific death receptor 5 (DR5) activation by fibroblast activation protein (FAP) dependent crosslinking. [abstract]. In: Abstracts: AACR Special Conference on Cellular Heterogeneity in the Tumor Microenvironment; 2014 Feb 26-Mar 1; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(1 Suppl):Abstract nr A59. doi:10.1158/1538-7445.CHTME14-A59

Determination of phosphorylated proteins in tissue specimens requires high-quality samples collected under stringent conditions

AIMS

For selection of patients who will benefit from targeted therapies, identification of biomarkers predictive of treatment response is desirable. Activation of the targeted pathway becomes apparent by protein phosphorylation. Determination of this phenomenon is therefore considered a promising biomarker approach. To date, however, it is unclear whether routinely collected tissue specimens allow determination of in-vivo phosphorylation states.

METHODS AND RESULTS

To investigate whether routinely collected tissue specimens retain the true phosphorylation states of a tumour's proteins, we compared protein phosphorylation states between matched tumour samples that were subjected to different ischaemic times by immunohistochemistry. The influence of formalin fixation and paraffin-embedding on phosphorylation states was investigated by comparison of matched fresh frozen and formalin-fixed paraffin-embedded surgical specimens as well as small biopsies. We show that ischaemia influences protein phosphorylation in a tumour-specific, unpredictable manner. Formalin fixation and paraffin-embedding lead to a decrease in detectable protein phosphorylation in larger surgical specimens, but not in small biopsies.

CONCLUSIONS

Determination of protein phosphorylation using routinely collected surgical specimens results in artefacts which do not reflect a tumour's true states of pathway activation. Valid measurement of phosphorylated biomarkers requires that tissue collection procedures are tightly controlled, avoiding ischaemia and large-specimen fixation.