Addition of losartan to FOLFIRINOX and chemoradiation reduces immunosuppression-associated genes, Tregs and FOXP3+ cancer cells in locally advanced pancreatic cancer

Abstract

Purpose:

Adding losartan (LOS) to FOLFIRINOX (FFX) chemotherapy followed by chemoradiation (CRT) resulted in 61% R0 surgical resection in our phase II trial in patients with locally advanced pancreatic cancer (LAPC). Here we identify potential mechanisms of benefit by assessing the effects of neoadjuvant losartan on the tumor microenvironment.

Experimental Design:

We performed a gene expression and immunofluorescence (IF) analysis using archived surgical samples from patients treated with LOS+FFX+CRT (NCT01821729), FFX+CRT (NCT01591733) or surgery upfront, without any neoadjuvant therapy. We also conducted a longitudinal analysis of multiple biomarkers in the plasma of treated patients.

Results:

In comparison to FFX+CRT, LOS+FFX+CRT downregulated immunosuppression and pro-invasion genes. Overall survival (OS) was associated with dendritic cell (DC) and antigen presentation genes for patients treated with FFX+CRT, and with immunosuppression and invasion genes or DC– and blood vessel-related genes for those treated with LOS+FFX+CRT. Furthermore, losartan induced specific changes in circulating levels of IL-8, sTie2 and TGF-β. IF revealed significantly less residual disease in lesions treated with LOS+FFX+CRT. Lastly, patients with a complete/near complete pathological response in the LOS+FFX+CRT-treated group had reduced CD4+FOXP3+ regulatory T cells (Tregs), fewer immunosuppressive FOXP3+ cancer cells (C-FOXP3) and increased CD8+ T cells in PDAC lesions.

Conclusions:

Adding losartan to FFX+CRT reduced pro-invasion and immunosuppression related genes which were associated with improved OS in patients with LAPC. Lesions from responders in the LOS+FFX+CRT-treated group had reduced Tregs, decreased C-FOXP3 and increased CD8+ T cells. These findings suggest that losartan may potentiate the benefit of FFX+CRT by reducing immunosuppression.

Introduction

The poor survival of patients with locally advanced and metastatic pancreatic ductal adenocarcinoma (PDAC) is due to its aggressive biology, limited effectiveness of cytotoxic agents, and an immune-suppressive tumor microenvironment (TME) [1, 2]. The abnormal PDAC TME promotes the recruitment of M2 macrophages, myeloid-derived suppressor cells (MDSCs), neutrophils, and regulatory T cells (Tregs), which secrete inflammatory cytokines (e.g. IL-6, IL-10, GM-CSF, TGF-β) that promote PDAC progression and metastasis [3, 4], and impair the infiltration and activity of T cells, NK cells and dendritic cells (DCs) [4–7]. In PDAC, cancer cells expressing FOXP3 (cancer-FOXP3 or C-FOXP3) have been shown to be a crucial component of the TME and may serve as a biomarker of poor prognosis in patients [8].

We have shown that angiotensin system inhibitors (ASIs), including the angiotensin II receptor blocker (ARB) losartan, can enhance the delivery and efficacy of cytotoxic agents in (PDAC) models [9]. The mechanisms underlying this benefit include “remodeling” of cancer-associated fibroblasts (CAFs) and extracellular matrix (ECM), resulting in blood vessel decompression, improved perfusion, and decreased hypoxia [9, 10]. Based on our experimental results with losartan, we initiated a clinical trial at Massachusetts General Hospital (MGH) to determine whether losartan improves the effectiveness of the drug cocktail FOLFIRINOX (FFX) and chemoradiation (CRT) in patients with locally advanced PDAC (LAPC). The results of this trial indicated that adding losartan to FFX+CRT was associated with high rates of surgical resection (69.4%) and R0 resection (61%) in LAPC [11].

In a retrospective analysis, we examined the effect of long-term ASI use on the survival of patients with PDAC and explored its potential mechanisms [12]. Our findings indicated that chronic ASI use is independently associated with longer overall survival (OS) in non-metastatic PDAC patients [12]. Furthermore, unbiased analysis of the transcriptome suggested that the improved survival benefit associated with ASI (lisinopril) therapy could be due to remodeling of the extracellular matrix, improved oxidative phosphorylation, inhibition of tumor progression (down-regulation of cell cycle, NOTCH, and WNT pathways) and enhanced anti-tumor immunity (increased activity of pathways linked to T cells and antigen-presenting cells) [12]. ASIs are known to alleviate hypoxia – which suppresses DC activation [13] – and inhibit the activity of tumor-promoting macrophages in other cancer types [14, 15]. Cytotoxic agents can also induce anti-tumor effects by reprogramming immune cells, including CD8+ T cells, DCs, tumor associated macrophages (TAMs) and MDSCs [16, 17]. Recent studies have shown that neoadjuvant FFX can increase effector T cells and decrease immunosuppressive cells in patients with PDAC. Neoadjuvant FFX increased the intratumoral recruitment of CD4+ T cells, CD8+ T cells and MHC-1 expression, and reduced the infiltration of FOXP3+Tregs and CD163+TAMs in LAPC [18]. A recent study using CyTOF revealed that FFX reduced inflammatory monocytes and Tregs, increased Th1 cells and decreased Th2 cells in the peripheral circulation of PDAC patients [19]. Neoadjuvant chemoradiotherapy has been shown to enhance both CD4+ and CD8+ T cells in PDAC [20].

We currently have an incomplete understanding of how ASIs, alone or combined with cytotoxic agents, modulate PDAC-associated immune cells. We postulate that losartan in combination with cytotoxic therapy will preferentially enhance the accumulation of immune cells with anti-tumor phenotypes, thus reprograming the immunosuppressive PDAC TME to an immunostimulatory milieu. To test this hypothesis, we obtained resected PDAC samples from patients treated with neoadjuvant FFX+CRT or losartan+FFX+CRT from two different phase II trials (NCT01591733 and NCT01821729), and from patients that underwent surgery upfront, without any neoadjuvant therapy. We performed a transcriptome analysis of RNA extracted from surgical specimens and used immunofluorescence (IF) to assess the infiltration by immune cells and quantify FOXP3 expressing cancer cells in PDAC lesions. Lastly, we carried out a longitudinal analysis of multiple biomarkers in the plasma of treated patients.

Materials and Methods

Patient Population and Treatment Plan

The study included PDAC patients enrolled in two completed trials at MGH who had blood and tissue prospectively collected on the trial protocol. All patients provided a signed written informed consent, and the Dana Farber Harvard Cancer Center Institutional Review Board approved both studies. Eligibility criteria and trial details were included in published reports on the clinical outcomes from these trials (“Total neoadjuvant therapy with FOLFIRINOX followed by individualized chemoradiotherapy for borderline resectable pancreatic adenocarcinoma: a phase 2 clinical trial [21]” for trial NCT01591733 and “Total neoadjuvant therapy with FOLFIRINOX in combination with losartan followed by chemoradiotherapy for locally advanced pancreatic cancer: a phase 2 clinical trial [11]” for trial NCT01821729). For the retrospective analysis of resected tissue samples from these two trials, we obtained primary PDAC tissue from 17 patients with locally advanced PDAC treated with losartan+FFX+CRT (NCT01821729), and from 19 patients with borderline resectable PDAC treated with FFX+CRT (NCT01591733) (Fig. 1A). We also included in the gene analysis resected PDAC samples (N=9) from patients with stage I or II resectable PDAC who did not receive any neoadjuvant therapy, as a control for the effects of cytotoxic therapy and losartan on an IRB-approved protocol for use of discarded tissue (Fig. 1A). The analysis of resected PDAC samples was approved by the MGH Institutional Review Board (protocol #: 2022P001372), which waived the requirement to obtain informed consent. No informed consent was required since this was a retrospective study in which excess tissue was used from otherwise consented procedures as part of a clinical trial or routine clinical care. The studies were performed in accordance with ethical guidelines of the Declaration of Helsinki, Belmont Report and U.S. Common Rule.

Figure 1. Differential gene expression between the 3 treatment groups (losartan+FFX+CRT n=17; FFX+CRT n=19; and untreated, n=9).

(A) Study design showing the sources of human PDAC tissues. (B) PCA plots showing the clustering of all 3 groups. (C) Volcano plots of highly upregulated and downregulated genes between losartan+FFX+CRT versus untreated and FFX+CRT versus untreated. P values generated by DESeq2 (Wald test) and adjusted p values based on FDR set at 0.05 (BH method).

Analysis of gene expression in the immune TME in resected PDAC samples.

To characterize the effects and identify potential molecular mechanisms of cytotoxic agents in combination with losartan on the immune TME, we used the nCounter PanCancer Immune Profiling Panel of 730 genes (NanoString) to analyze the RNA extracted from formalin-fixed paraffin embedded (FFPE) PDAC tissue sections (performed by the Center for Advanced Molecular Diagnostics (CAMD) at the Brigham and Women’s Hospital).

Differential gene expression analysis and differential gene correlation analysis

The NanoString software package (nSolver) was used to import, process, export and analyze the nCounter raw data. The DESeq2 R package was used to determine differentially expressed genes (DEGs) between treatment groups. We used Benjamini & Hochberg method to control the False Discovery Rate (FDR) at 0.05. For principal component analysis (PCA), prcomp and autoplot functions were used from stats and ggplot2 packages, respectively. We made volcano plots and gene plots from the DEG list using the EnhancedVolcano [22] and ggplot2 packages, respectively in R. For gene set enrichment analysis (GSEA), we used the GSEA application v4.2.3 [23]. We ran GSEA against the following MsigDB gene sets: Hallmark, KEGG, C3 and C5 gene sets. To correlate genes with one another and with respect to OS we used the Spearman Rank correlation (Prism Version 9 Software GraphPad). To determine whether gene signatures could stratify patient OS (< 36 months versus > 36 months), the expression of individual genes was transformed into Z-scores.

Immunofluorescence

IF staining was performed on surgically resected FFPE PDAC samples. Paraffin sections (5 μM thick) were baked for 3 hours at 60°C then loaded into the BOND RX. Slides were deparaffinized in xylene and hydrated through graded alcohols. Antigen retrieval (Vector Citrate pH6 retrieval solution, #H-3300–250) was performed at pH 6.0 for 20 minutes at 98°C. Slides were incubated in CuSO₄ for 90 minutes to block autofluorescence and blocked with 5% normal donkey serum. For the triple FOXP3, CD4 and cytokeratin-19 stain, slides were incubated with FOXP3 (BioLegend #320102, RRID:AB_430881, 1:25) and CD4 (Abcam #ab133616, RRID:AB_2750883, 1:25) antibodies overnight at 4°C, followed by incubation with secondary antibodies (Jackson Immunoresearch) for 2 hours and cytokeratin 19-Alexa Fluor 488 antibody (Abcam #ab192643, RRID: RRID:AB_2927708, 1:50) overnight. Residual disease and tumor bed were evaluated by H&E staining on an adjacent section and quantified by cytokeratin-19 IF stain. We also performed double staining for CD31 (Agilent #M0823, RRID:AB_2114471, 1:60) and αSMA-Cy3 (Sigma-Aldrich #C6198, RRID:AB_476856, 1:100), CEACAM6 (Thermo Fisher Scientific #MA5–37801, RRID:AB_2897725, 1:800) and cytokeratin-19, and FAP (Abcam #ab28244, RRID:AB_732312, 1:100) and CD68 (Agilent #M0718, RRID:AB_2687454, 1:200). All slides were counterstained with DAPI (Invitrogen NucBlue Fixed Cell ReadyProbes Reagent, #R37606), washed with deionized water, air dried, and mounted with ProLong Diamond Anti-fade Mountant (Invitrogen, #P36961). The antibody staining for CD8 (Agilent #M7103, RRID:AB_2075537, 1:400) and CD11c (Cell Signaling Technology #45581, RRID:AB_2799286, 1:400) was performed by the Dana-Farber/Harvard Cancer Center Specialized Histopathology Core.

Imaging was performed with the Axio Scan.Z1 slide scanner (Zeiss) at 20x magnification. Immunofluorescence images were analyzed using QuPath [24] and Python. Cells were identified based on a positive DAPI signal, and each of the cell populations were classified as positive or negative based on a single intensity threshold on mean expression within the cell. Only cells located within the tumor bed (as confirmed by H&E stain of adjacent section) were included in the quantitative analysis. The mean number of positive or negative cells per mm² of tissue was subsequently calculated and reported.

The spatial analyses were conducted by calculating the Euclidean distance between two cell types of interest for all cells on a given slide, before taking the minima of the set, giving the shortest distance for a given cell to the target cell population. Histograms were generated on these shortest distances with bin sizes of 100 μm, and the mean number of cells were reported per distance bin across all patients in a particular group.

The number of a given cell population was compared between losartan+FFX+CRT and FFX+CRT-treated patients and patients were further stratified based on the assigned College of American Pathologists (CAP) tumor regression grade in PDAC after neoadjuvant therapy [25]. We grouped patients based on the CAP score into the following categories: responders (complete/near complete response, CAP 0/1) compared to non-responders (partial/poor or no response, CAP 2/3). The Wilcoxon test was conducted for each cell population on a per patient basis for each group. An alpha value of 0.05 was considered statistically significant. All analyses were performed using Prism Version 9 Software (GraphPad) and R Statistical Software (Foundation for Statistical Computing, Vienna, Austria).

Circulating Biomarkers

Baseline and serial plasma draws were completed on day 1 and day 8 and at the beginning of FOLFIRINOX cycles 3, 4, 7, 8 and post-CRT. All biomarker assays were performed according to manufacturer’s instructions and levels were measured with enzyme-linked immunosorbent assays (ELISA) (Promega, R&D Systems) and multiplex arrays from Meso-Scale Discovery. To determine effect of treatment on cytokine levels, we fit random effects models of cytokine levels to treatment group and serum sample collection time (fixed effects) and a random effect for each patient. For serum sample collection time for each patient, we considered the first sample collected as Day 0. The following formula was used in the lme4 package [26] in R:

1. 1. For modeling Treatment group (FFX+CRT as reference):

   $$Cytokine\sim Treatment\ast Day+(1\mid Patient)$$
2. 2. For modeling Treatment group and response, we created a new variable where the losartan+FFX+CRT group was divided into pathological responders (complete/near complete response, CAP 0/1) versus non-responders (partial/poor or no response, CAP 2/3) with FFX+CRT as reference:

   $$Cytokine\sim TreatmentResponse\ast Day+(1\mid Patient)$$

Separate models were created for each of the 16 cytokines (dependent variable).

Data Availability

The transcriptome data generated in this study is publicly available in NCBI Gene Expression Omnibus (GEO) at GSE222788. The remaining data generated in this study are available upon request from the corresponding authors.

Statistical Analysis

Statistical analysis was performed using R and GraphPad prism. Details of statistical tests are mentioned in their respective method subsections.

Results

Principal component analysis revealed that samples of patients not treated with neoadjuvant therapy clustered distinctly in comparison to losartan+FFX+CRT and/or FFX+CRT (Fig. 1B). Among the 730 genes tested, we detected 313 and 242 differentially expressed genes (DEGs) in losartan+FFX+CRT and FFX+CRT, respectively, compared to untreated (Supplementary Table 1). Upon comparing the upregulated and downregulated genes in losartan+FFX+CRT and FFX+CRT with respect to PDAC patients with no neoadjuvant treatment, we found that there were more genes overlapping between these two cohorts (Fisher exact test p < 0.0001): 124 and 79 genes were commonly up- and downregulated, respectively, in losartan+FFX+CRT and FFX+CRT treated samples (Supplementary Fig. 1A). The top 30 upregulated DEGs (fold-difference) in losartan+FFX+CRT and FFX+CRT versus untreated samples included cytokines / chemokines, complement factors, proliferation inhibition and blood vessel related genes (Fig. 1C, Supplementary Table 2). The top 30 downregulated DEGs in losartan+FFX+CRT versus untreated samples included immune checkpoints, pro-inflammatory genes, and B cell related genes (Fig. 1C, Supplementary Table 2). The top 30 downregulated DEGs in FFX+CRT versus untreated samples included cytokines, interferon and RIG-I-like receptor signaling pathway (Fig. 1C, Supplementary Table 2). FFX+CRT enriched for upregulation of the apical junction pathway and enriched for downregulation of cytosolic DNA sensing and RIG I like receptor pathways (Supplementary Fig. 1B). GSEA comparison of losartan+FFX+CRT versus FFX+CRT enriched for pathways involved in transcription factor activity, response to UV, response to radiation, cell cycle, regulation of autophagy and PI3K AKT MTOR signaling (Supplementary Fig. 1B). The direct comparison of losartan+FFX+CRT and FFX+CRT, revealed that 29 genes were downregulated, and 25 genes were upregulated (Supplementary Fig. 2, Supplementary Table 1). While these findings reveal commonalities, transcriptomic responses in PDAC after losartan+FFX+CRT as compared to FFX+CRT treatment, revealed some different signatures, which may explain the impacts of losartan’s addition to this therapeutic regimen.

FFX+CRT and losartan+FFX+CRT increased the expression of genes involved in the development and differentiation of blood vessels, and the transendothelial migration of leukocytes.

FFX+CRT and losartan+FFX+CRT increased the expression of genes linked to blood vessel maturation (CDH5, THBS1, THBD, PDGFRB, ENG) (Fig. 2A) and the transendothelial migration of leukocytes (PECAM1, MCAM, JAM3, ICAM2, SELL, SELPLG) (Fig. 2B). Losartan+FFX+CRT significantly reduced the expression of TNFSF15, while FFX did not affect the expression of TNFSF15 (Supplementary Fig. 3A). The autocrine production of the TNFSF15 protein by endothelial cells inhibits endothelial cell proliferation, angiogenesis and tumor growth [27]. Taken together, these results suggest that FFX+CRT and losartan+FFX+CRT can stimulate angiogenesis and the normalization of the PDAC vasculature, as well as the transendothelial migration of leukocytes.

Figure 2. FFX+CRT and losartan+FFX+CRT enhance the expression of genes linked to the maturation of blood vessels, leukocyte transendothelial migration, T cell activation, cytolytic activity of T cells and NK cells, and dendritic cells.

Effect of FFX+CRT and losartan+FFX+CRT on the expression of genes associated with (A) blood vessel maturation and integrity, (B) leukocyte transendothelial migration, (C) T cell activation, (D) cytolytic activity of T cells and NK cells and (E) dendritic cells. P values generated by DESeq2 (Wald test) and adjusted p values based on FDR set at 0.05 (BH method).

FFX+CRT and losartan+FFX+CRT enhanced the expression of genes that play critical roles in T cell activation and immune cytolytic activity.

FFX+CRT and losartan+FFX+CRT significantly reduced the expression of T helper (Th) Th1 (IFNG, IL12A) Th2 (IL4, IL13) and Th17 (IL23A) genes (Supplementary Table 1) but increased the expression of CD6, ALCAM, NFATC1, NFATC2, TNFSF8 (CD30 ligand), and CD4 (Fig. 2C). The cell adhesion molecule ALCAM binds to CD6 at the surface of T cells and promotes their activation and proliferation [28]. TNFSF8 is expressed by macrophages, DCs, and B cells and can interact with the CD30 receptor to promote the activation and proliferation of central memory T cells [29]. NFATC1 and NFATC2 are transcription factors involved in several biological processes including the formation of blood vessels, regulation of interaction of lymphocytes with endothelial cells and the activation of T cells. Losartan+FFX+CRT and FFX+CRT also increased the expression of cytolytic genes expressed by NK cells and T cells (GZMA, GZMB, KLRB1) (Fig. 2D), and losartan+FFX+CRT increased the expression of KLRK1 (Fig. 2D).

FFX+CRT and losartan+FFX+CRT increased the expression of dendritic cell-selective genes.

FFX+CRT and losartan+FFX+CRT significantly enhanced the expression of DC-associated genes [30], including CD1C, which is mostly found in MoDCs and conventional DC type 2 (cDC2s), and IL3RA (CD123), which is enriched in plasmacytoid DCs (pDCs) (Fig. 2E). Conversely, the expression of CD1A, typically over-expressed by cDC2s, was significantly down-regulated in both FFX+CRT and losartan+FFX+CRT-treated patients (Supplementary Table 1). CD86, a co-stimulatory factor that can be broadly expressed by various DC and macrophage types, was upregulated following either FFX+CRT or losartan+FFX+CRT treatment, whereas the expression of both CD80 and CD83 was lower in the losartan+FFX+CRT group (Fig. 2E). The expression of genes involved in the migration and maturation of DCs were also modulated by treatment. In comparison to untreated samples, both FFX+CRT and losartan+FFX+CRT increased the expression of CCL17 and CSF1 (Supplementary Fig. 3B), which are respectively involved in the migration and maturation of DCs [31–33]. Interestingly, the expression of chemokine genes (CCL3, CCL4) which play roles in DC recruitment [34, 35] was higher in tumors treated with losartan+FFX+CRT (Supplementary Fig. 3B). The expression of HMGB1 was also higher in samples treated with losartan+FFX+CRT (Supplementary Fig. 3B). The extracellular form of HMGB1 stimulates the maturation of DCs. These results suggest that FFX+CRT with or without losartan can stimulate the infiltration of specific DC subsets and maturation of DCs in PDAC.

The addition of losartan to FFX+CRT reduces the expression of pro-invasion and upregulates genes and pathways involved in tumor suppression.

Losartan+FFX+CRT compared to FFX+CRT significantly decreased the expression of pro-invasion genes (CEACAM6, CXCL16, CCR1, PRAME, ELK1) (Fig. 3A) and increased the expression of tumor suppressor genes (RORA, EP300) (Fig. 3B). The retinoic acid receptor-related orphan receptor alpha (RORα) the protein product of RORA is involved in tumor suppression, and plays pro- and anti-inflammatory roles, which are context-dependent [36–38]. RORA in tumors treated with losartan+FFX+CRT correlated strongly with the transcription factor Nuclear Factor of Activated T-Cells 4 (NFATC4) (Spearman R=0.918), a gene also upregulated by losartan+FFX+CRT (Supplementary Figure 2). NFATC4 is involved in the activation of T cells and endothelial cells [39, 40]. RORA and NFATC4 correlated with tumor suppressor genes as well as blood vessels, DCs, MHCII, T cell activation and inflammation inhibition related genes, and were inversely associated with epithelial/cancer cell biomarkers (EPCAM, CDH1) (Fig. 3C). In contrast, in tumors treated with FFX+CRT there was no or weak correlations between the same gene sets (Fig. 3D). The expression of tumor suppressor genes and the inverse correlation between cancer cell biomarkers and tumor suppressor genes, suggest that the addition of losartan to FFX+CRT reduces cancer cell proliferation and tumor progression. This is also supported by our GSEA results and IF analysis of residual disease. GSEA revealed an upregulation of the tumor suppressor p53 pathway (Supplementary Fig. 3C) and circadian pathway (Supplementary Fig. 3D). Circadian pathway genes regulate cell cycle check points as well as cell cycle progression [41]. Furthermore, we found significantly less residual disease (cytokeratin-19-positive cells) in the tumor bed of lesions treated with losartan+FFX+CRT than FFX+CRT (p=0.014, Fig. 3E).

Figure 3. The addition of losartan to FFX+CRT decreased the expression of pro-invasion-associated and immunosuppression genes and increased the expression of tumor suppressor genes.

(A) Pro-invasion genes. (B) Tumor suppressor genes. P values generated by DESeq2 (Wald test) and adjusted p values based on FDR set at 0.05 (BH method). (C-D) Heatmaps of Spearman rank correlation of genes associated with RORA and NFATC4 in tumor samples treated with (C) losartan+FFX+CRT and (D) FFX+CRT. Gene sets include the following: tumor suppressors (RORA, CYLD, FEZ1), blood vessels (NFATC4, AKT3, PECAM1, CDH5, JAM3), dendritic cells (IL3RA, FLT3LG, CSF1), MHC II (HLA-DRB3, LAMP2), T cell activation (STAT4, CD6, TNFSF8), inflammation inhibition (NCF4, SERPING1, NFKBIA, A2M) and epithelial (EPCAM, CDH1). (E) Quantitative analysis of residual disease in resected PDAC lesions (immunofluorescence of cytokeratin-19+ cells located in the tumor bed). (F) Losartan+FFX+CRT compared to FFX+CRT induced a greater down-regulation of macrophage related genes. (G) Losartan+FFX+CRT and FFX+CRT induced a significant down-regulation CEACAM1, while the addition of losartan to FFX+CRT induced a greater down-regulation of TIGIT and FOXP3. P values generated by DESeq2 (Wald test) and adjusted p values based on FDR set at 0.05 (BH method).

The addition of losartan to FFX+CRT reduces the expression of immunosuppressive genes.

The addition of losartan to FFX+CRT reduced the expression of M2 macrophage (CCR1, MARCO, APOE) related genes (Fig. 3A and 3F). The gene products of CCR1, MARCO and APOE are immunosuppressive in PDAC and other tumor types (see Discussion). FFX+CRT and losartan+FFX+CRT both reduced the expression of CEACAM1, a reported ligand of immune checkpoint molecules [42–44] (Fig. 3G). In comparison to untreated and FFX+CRT treated samples, losartan+FFX+CRT induced a significantly lower expression of the immune checkpoint TIGIT (Fig. 3G). The expression of the FOXP3 gene—a transcription factor which regulates the activity of CD4+ Tregs which express TIGIT—was significantly lower in losartan+FFX+CRT-treated tumors than untreated tumors (Fig. 3G).

In patients treated with losartan+FFX+CRT, the lower expression of immunosuppression and invasion / proliferation genes is associated with improved overall survival.

We assessed whether gene sets could identify differences in the immune and non-immune TME between low (< 36 months) versus high (> 36 months) OS. To perform this analysis, we included genes positively (Spearman R ≥ 0.5) or negatively correlated (Spearman R ≥ −0.5) with OS (Supplementary Table 3). For samples treated with losartan+FFX+CRT, a gene set which included immunosuppression and invasion / proliferation genes stratified OS (Fig 4A, Supplementary Table 4). The gene sets which included only immunosuppression or invasion / proliferation genes could also stratify OS (Fig. 4B and 4C). The gene set positively associated with OS – which included 30 genes – did not stratify with OS, but smaller grouping of genes which included DCs, blood vessel maturation, transendothelial migration and tumor suppression related genes stratified OS (Fig 4D, Supplementary Table 4). For samples treated with FFX+CRT, the gene sets negatively or positively associated with OS stratified OS (Fig 4E – 4H). DC and antigen presentation related genes (CD1C, CD209) provided a better stratification of OS (Fig. 4H) than the 16 gene set which included DC, B cell and T cell genes (Fig. 4G). Thus, for patients treated with losartan+FFX+CRT the lower expression of immunosuppression and invasion related genes, and higher expression of DC and tumor suppressor genes is associated with improved OS.

Figure 4. Gene set stratification of overall survival (OS) for patients treated with losartan+FFX+CRT and FFX+CRT.

(A-D) Patients treated with losartan+FFX+CRT. Gene sets composed of (A-C) immunosuppression and invasion / proliferation, (B) immunosuppression, (C) invasion / proliferation genes, (D) DCs, blood vessel maturation, transendothelial migration and tumor suppression related genes stratified OS. (E-H) For patients treated with FFX+CRT, gene sets (E, F) negatively or (G, H) positively associated with OS. P values based on t-test. For the complete list of gene sets refer to Supplementary Table 4.

The addition of losartan to FFX+CRT reduces the number of Tregs and increases the infiltration of CD8+ T cells in PDAC lesions with less residual disease.

We used IF to determine the intratumoral infiltration of multiple cell types in PDAC lesions treated with losartan+FFX+CRT compared to FFX+CRT. Quantification revealed no differences between the two treatment groups in the number of Tregs (CD4+FOXP3+ cells), CD4+ T cells, CD8+ T cells, CD11c+ cells, CD68+ macrophages, FAP+ cells, CD31+αSMA+ blood vessels and CEACAM6 staining (Supplementary Fig. 4A–4H). We observed that patients in both FFX+CRT and losartan+FFX+CRT showed a wide range of tumor regression grades as measured by the CAP tumor regression grading system in PDAC after neoadjuvant therapy [25] (Supplementary Fig. 5). Given the range of tumor regression grades, we stratified patients based on CAP score into the following groups: pathological responders (complete/near complete response, CAP 0/1) and non-responders (partial/poor or no response, CAP 2/3). We then compared the cell type populations in responders with non-responders in PDAC lesions treated with losartan+FFX+CRT or FFX+CRT. In FFX+CRT treated PDAC lesions, there were no significant differences in FFX+CRT responders as compared to non-responders for CD11c+ cells, FOXP3+Tregs, CD4+ T cells, CD8+ T cells, CD68+ macrophages, FAP+ cells, α−SMA+CD31+ blood vessels and CEACAM6 staining (Supplementary Fig. 6A–6H). Conversely, in resected samples from patients treated with losartan+FFX+CRT, we found significantly fewer FOXP3+Tregs in PDAC lesions in pathological responders as compared to non-responders (p = 0.029, Fig. 5A–5D). Furthermore, responders in the losartan+FFX+CRT group had significantly higher infiltration of CD8+ T cells than non-responders (p = 0.024, Fig. 5E–5F). These results suggest that the addition of losartan to FFX+CRT may be impacting both the FOXP3+Treg and CD8+ T cell populations in patients with a better pathological response. Lastly, in PDAC lesions treated with losartan+FFX+CRT, there were no significant differences in number of CD11c+ cells, CD68+ macrophages, FAP+ cells, α−SMA+CD31+ blood vessels or CEACAM6 staining (Supplementary Fig. 7A–7E).

Figure 5. Immunofluorescence (IF) staining and quantitative analysis in PDAC lesions in losartan+FFX+CRT-treated patients.

(A, C) Representative IF staining of cytokeratin-19+ cancer cells (cyan), CD4+ T cells (green), FOXP3+ cells (red) and CD4+FOXP3+ Tregs (yellow merge, bottom right) in (A) losartan+FFX+CRT non-responders (NR, CAP score 2/3) and (C) responders (R, CAP score 0/1). White box shows inset area. White arrows indicate FOXP3+Tregs. Nuclei are stained with DAPI (blue). Scale bar is 60 μM. (B) Corresponding quantitative analysis of FOXP3+ Tregs in PDAC lesions from panels A and C (p = 0.029). Non-responders in light blue and responders in dark blue. (D) Corresponding distance analysis of FOXP3+Tregs to cancer cells in losartan+FFX+CRT-treated patients from panels A and C (*p = 0.0058). (E) Representative IF staining of CD8+ T cells (green) in losartan+FFX+CRT non-responders (NR) and responders (R). Scale bar is 50 μM. (F) Corresponding quantitative analysis in PDAC lesions from panel E showing the number of CD8+ T cells in non-responders (NR, light blue) compared to responders (R, dark blue) (p = 0.024). (G) Corresponding distance analysis of CD8+ T cells to residual cancer cells from panel E. (H) Representative IF staining of CD4+ T cells (red) and cytokeratin-19+ cancer cells (green) in losartan+FFX+CRT non-responders (NR) and responders (R). Scale bar is 50 μM. (I) Corresponding quantitative analysis in PDAC lesions from H of the number of CD4+ T cells. (J) Corresponding distance analysis of CD4+ T cells to cancer cells from panel H (*p = 0.016). For barplots, ANOVA test followed by Fisher’s Least Significant Difference (LSD) test was performed and for distance metrics, Kruskal-Wallis test was done (alpha = 0.05).

Patients with a partial/poor pathological response in the losartan+FFX+CRT group had a higher number of Tregs and CD4+ cells in closer proximity to residual cancer cells.

Based on the relevant differences in FOXP3+Tregs and CD8+ T cells between pathological responders and non-responders in the losartan+FFX+CRT group, we then conducted a spatial analysis of the IF images to examine the distances of various immune cell populations in relation to residual cancer cells. We found that in the lesions of non-responders from the losartan+FFX+CRT group, not only was there an increased number of Tregs (p = 0.029, Fig. 5A–5C), but Tregs existed in closer proximity to residual cancer cells as compared to responders (p = 0.0058, Fig. 5D). Conversely, there was no significant difference in the distance metric analysis of CD8+ T cells (Fig. 5G). While there was no difference in the number of CD4+ T cells per mm² in responders compared to non-responders in the losartan+FFX+CRT group (Fig. 5H–5I), distance metric analysis showed that non-responders had a significantly higher number of CD4+ cells in closer proximity to residual cancer cells (p = 0.016, Fig. 5J). Spatial IF analysis of the other cell types examined in this study showed no other significant differences in responders versus non-responders in either losartan+FFX+CRT (CD11c+ cells, CD68+ macrophages, FAP+ cells) (Supplementary Fig. 7F–7H) or FFX+CRT (FOXP3+Tregs, CD4+ T cells, CD8+ T cells, CD11c+ cells, CD68+ macrophages and FAP+ cells) (Supplementary Fig. 6I–N). In summary, these IF results indicate that losartan in combination with FFX+CRT may promote a more immunostimulatory TME in patients with a better pathological response.

Cancer-FOXP3 was increased in patients with a partial/poor pathological response in the losartan+FFX+CRT group.

In addition to being a key transcription factor in Treg biology, FOXP3 is expressed in several tumors, including PDAC [8]. In PDAC, expressed cancer-FOXP3 has been reported to be a biomarker of poor prognosis in patients and highly associated with the numbers of FOXP3+Treg cells [8]. Using IF, we analyzed the amount of cancer-FOXP3 present in PDAC lesions from both losartan+FFX+CRT and FFX+CRT-treated groups. There was no difference in cancer-FOXP3 between FFX+CRT and losartan+FFX+CRT (Supplementary Fig. 4I). Interestingly, in the losartan+FFX+CRT non-responder group, the number of cancer-FOXP3 cells were significantly increased compared to responders (p = 0.0377, Fig. 6A–6C). There was also a strong positive correlation of FOXP3+Tregs with cancer-FOXP3 in the losartan+FFX+CRT-treated patients (R = 0.84, p = 2.6e-05, Fig. 6D). While there was no difference in cancer-FOXP3 in responders as compared to non-responders in the FFX+CRT group (Fig. 6E), there was a weak positive correlation of FOXP3+Tregs with cancer-FOXP3 in the FFX+CRT group (R = 0.6, p = 0.010, Fig. 6F).

Figure 6. Immunofluorescence (IF) staining and quantitative analysis of FOXP3+ cancer cells (C-FOXP3) in PDAC lesions.

(A-B) Representative images of IF staining of FOXP3+ cells (red), CD4+ T cells (white), cytokeratin-19+ cancer cells (CK, green), and C-FOXP3 (merged yellow, far right panels) in losartan+FFX+CRT-treated (A) non-responders (NR, CAP score 2/3) and (B) responders (R, CAP score 0/1). White arrows indicate FOXP3+Tregs. Nuclei are stained with DAPI (blue). Scale bar is 50 μM. (C, E) Corresponding quantitative analysis in PDAC lesions of number of FOXP3+Tregs in (C) losartan+FFX+CRT non-responders (NR, light blue) and responders (R, dark blue) (p = 0.0377) and (E) FFX+CRT non-responders (NR, light green) and responders (R, dark green). P-value based on the Fisher’s Least Significant Difference (LSD) test. (D, F) Correlation analysis of the number of C-FOXP3+ cells and FOXP3+Tregs in (D) losartan+FFX+CRT (R = 0.84, p = 2.6e-05) and (F) FFX+CRT (R = 0.6, p = 0.1). R correlation and p-value generated by fitting linear model using lm function in R. (G-I) Cytokine levels of serial plasma samples from FFX+CRT (green) and losartan+FFX+CRT (blue) for (G) IL-8, (H) TGF-β and (I) sTie2. (I) For sTie2, losartan+FFX+CRT group stratified based on pathological response for non-responders (NR, light blue) and responders (R, dark blue). In IL-8 and TGF-β, *slope indicates statistical significance in treatment x time interaction term in mixed effect model and in Tie2 plot, * indicates overall significant difference between FFX+CRT versus losartan+FFX+CRT responders (independent of time).

Losartan induces specific changes in circulating IL-8, HGF, sTie2 and TGF-β.

We measured cytokine levels in the plasma of patients treated with FFX+CRT and losartan+FFX+CRT at multiple timepoints. To compare the effect of the two treatments, we performed random effect modeling where FFX+CRT patients were the reference group. We found that levels of interleukin-8 (IL-8) significantly decreased in patients treated with losartan over time (Fig. 6G; Supplementary Table 5; p < 0.001). The levels of HGF and tenascin C increased significantly over time in losartan+FFX+CRT group regardless of pathological response (Supplementary Tables 5 and 6; p for HGF <0.001, p for tenascin C <0.001). Similar to our previous results [11], the total level of TGF-β was high (Supplementary Table 5, p = 0.001) but significantly decreased over time in the losartan+FFX+CRT group (Fig. 6H, Supplementary Table 5; p = 0.003). Conversely, the overall level of sTie2 was significantly lower in the losartan+FFX+CRT group (Fig 6I; Supplementary Table 5, p = 0.007) but the levels increased over time (Supplementary Table 5; p<0.001). The analysis of the relationship between pathological response and cytokine levels showed that sTie2 levels significantly increased over time in non-responders (Fig. 6I, Supplementary Table 6; p < 0.001). However, sTie2 levels remained low among responders and did not increase over time (Supplementary Table 6; p = 0.004).

Discussion

Our results show that FFX+CRT and losartan+FFX+CRT induced several changes in borderline PDAC and LAPC samples, respectively. These included increases in gene expression levels for factors involved in 1) the maturation of blood vessels and transendothelial migration of leukocytes, 2) activation of T cells, 3) cytolytic activity of T cells and NK cells, and 4) DC activity. The addition of losartan to FFX+CRT treatment 1) further reduced the expression of genes associated with immunosuppression and invasion which correlated with improved OS, 2) reduced the number of FOXP3+Tregs and FOXP3+ cancer cells, and increased CD8+ T cells in resected PDAC lesions with a complete/near complete pathological response, and 3) induced specific changes in circulating levels of IL-8, sTie2 and TGF-β.

A detailed analysis showed that addition of losartan to FFX+CRT reduced the expression of pro-invasion genes (CEACAM6, CXCL16, ELK1, PRAME), as well as M2 macrophage-associated genes (APOE, MARCO, CCR1) involved in tumor progression and immunosuppression. In pancreatic cancer models CEACAM6, CXCL16, ELK1 can stimulate tumor cell proliferation and invasion [45–48]. In PDAC, APOE – which is expressed by fibroblasts, macrophages, and plasma cells – stimulates the infiltration of MDSCs which suppresses the infiltration of CD4+ and CD8+ T cells [3]. The expression of CCR1 by MDSCs and macrophages in PDAC has also been linked to immunosuppression [49]. In a PDAC model, MARCO promoted tumor progression and metastasis [50], while another report showed that high expression of MARCO is an independent indicator of poor prognosis in PDAC [51]. Targeting of MARCO in lung cancer models reduced the activity of Tregs and potentiated the cytolytic and anti-tumor activity of T cells and NK cells [52]. Our results also revealed a lower expression of pro-invasion and immunosuppression-related genes in resected samples of PDAC patients with superior OS (Fig 4A–4C, Supplementary Table 4). We have shown in a transcriptome analysis that the chronic administration of the ASI lisinopril in patients with PDAC reduces pro-invasive pathways (ECM / receptor interaction, WNT and Notch signaling) and expression of pro-invasion genes [12]. Thus, the decrease expression of pro-invasion and M2 macrophages genes suggest that the addition of losartan to FFX+CRT produces a PDAC TME that attenuates invasion and immunosuppression.

In PDAC the paucity of conventional DCs (cDC1 and cDC2) [5, 6, 53] in the TME impairs immune surveillance and accelerates tumor progression [5, 6]. The production of granulocyte colony stimulating factor and IL-6 in mouse models of PDAC and human PDAC impairs the development of cDC1s [6, 54]. The Fms-related Receptor Tyrosine Kinase 3 Ligand (Flt3L) combined with CD40 agonist antibody significantly increased the infiltration of cDC1s and cDC2s in PDAC, but while the combination inhibited tumor growth it did not induce tumor regressions [5]. In contrast, the addition of radiation to Flt3L and CD40 activation induced the regression of PDAC tumors. Here we show that FFX+CRT with and without losartan increased the expression of DC-related genes CD1C and IL3RA, and the expression of CD1C and CD209 – which can be expressed by DCs or macrophages – in FFX+CRT-treated samples was associated with improved OS. Gene sets that included IL3RA and CSF1 were also associated with improved OS in samples treated with losartan+FFX+CRT. In another study which analyzed resected samples from untreated PDAC patients, the higher gene expression of CD209 and CD1C was also correlated with improved survival [55]. While we found high levels of CD1C (cDC2) and IL3RA (pDCs) in PDAC patients with improved OS, it will be significant in mechanistic studies to assess the specific role of cDC2 and pDCs in PDAC immunosurveillance. Of significance, cDC2 vaccination in mice reduced intratumoral MDSCs, reprogrammed M2 macrophages and inhibited tumor growth in a colon carcinoma model [56]. Furthermore, in another study Treg depletion reduced the suppression of cDC2 and potentiated the anti-tumor activity of conventional CD4+ T cells [57].

Neoadjuvant FFX can increase the infiltration of effector T cells and decrease immunosuppressive cells in patients with PDAC [18]. The results of our transcriptome analysis showing an increase in expression of genes linked to blood vessel maturation and transendothelial migration of leukocytes is consistent with this improvement in T cell infiltration. Indeed, our IF analysis showed that patients with a better pathological response in the losartan+FFX+CRT group had significantly higher infiltration of CD8+ T cells than non-responders. The maturation of blood vessels has been associated with the increased infiltration of CD4+ and CD8+ T cells and the efficacy of immunotherapy [58–60]. PECAM-1, MCAM, I-CAM2, and VE-cadherin (the gene product of CDH5) play significant roles in the transendothelial migration of leukocytes [61]. For example, the targeting of endothelial VE-cadherin enhanced the intratumoral infiltration of T cells and T cell-mediated immunotherapy [62].

FFX+CRT reduced the expression of the immune checkpoint TIGIT, while the addition of losartan to FFX+CRT induced a significantly greater reduction in TIGIT and reduced the expression of FOXP3. In addition, resected PDAC lesions from responders in the losartan+FFX+CRT-treated group had decreased FOXP3+Tregs. TIGIT is expressed by a subset of Tregs that suppress the activity of Th1 and Th17 cells [63, 64], while in human PDAC TIGIT is expressed by Tregs and CD8+ T cells [7]. Inhibition of TIGIT combined with PD-1 blockade and CD40 activation potently induced tumor regression in orthotopic PDAC models [65]. Interestingly, TIGIT inhibition increased the production of IFNγ and proliferation of CD8+ T cells isolated from the blood of PDAC patients treated with FFX but did not affect the activity of CD8+ T cells isolated before the administration of FFX [66].

A recent meta-analysis indicated that high levels of intratumoral or peritumoral FOXP3+Treg cell infiltration in PDAC could be considered as a negative prognostic factor [67]. In PDAC, FOXP3+ cancer cells have been reported to be highly associated with the numbers of FOXP3+Treg cells [8]. We observed that patients with more residual disease in the losartan+FFX+CRT group had increased FOXP3+ cancer cells. Indeed, there was a strong positive correlation of FOXP3+Tregs to FOXP3+ cancer cells in the losartan+FFX+CRT-treated patients and a weak positive correlation in the FFX+CRT group. We also found that in the lesions of non-responders from the losartan+FFX+CRT group, not only was there an increased number of Tregs, but Tregs were in closer proximity to residual cancer cells than in responders. An increase in tumor-infiltrating FOXP3+Tregs correlated with disease progression and were observed more frequently in stroma immediately adjacent to tumor cells [68]. Given the extent of FOXP3+Tregs in close proximity to residual cancer cells in the losartan+FFX+CRT-treated patients with a partial/poor pathological response, these results suggest that a high proportion of Tregs in the residual tumor may reflect ongoing suppressed anti-tumor immunity thus preventing a more robust anti-tumor response.

It has been shown that cancer-FOXP3 recruits FOXP3+ Treg cells by directly activating the secretion of CCL5 in PDAC [8]. Recently, it was demonstrated that PD-L1 is a direct target of FOXP3+ cancer cells in PDAC and combined blockade of PD-L1 and CCL-5 may provide an effective therapy for patients with PDAC, especially those patients whose tumors have high cancer-FOXP3 levels [69]. Given that cancer-FOXP3 was shown to mediate the immune escape mechanism in PDAC by directly inhibiting CD8+ T cells via the PD-L1/PD-1 pathway [69], we speculate that if losartan does reduce cancer-FOXP3 and Tregs while simultaneously increasing the number of CD8+ T cells, losartan may enhance the anti-tumor effect of immune checkpoint blockade. Future studies in PDAC preclinical models will need to examine if losartan directly down-regulates cancer-FOXP3, FOXP3+Tregs and TIGIT to potentiate the anti-tumor activity of immunotherapy and cytotoxics in PDAC, thus providing new mechanistic insights into overcoming immune evasion in PDAC.

The analysis of cytokine levels in plasma revealed that losartan+FFX+CRT reduced the levels of circulating IL-8. This finding is consistent with a recent study, which showed that losartan combined with the tyrosine kinase inhibitor toceranib reduced the plasma levels of IL-8 in dogs with osteosarcoma [70]. Interestingly, our results also show that sTie2 levels increased as a function of time in patients treated with losartan+FFX+CRT and a partial/poor pathological response. The increase in plasma sTie2 in these patients could be an indication of vascular and tumor progression. High plasma levels of sTie2 – a receptor associated with angiogenic blood vessels – has been linked with vascular progression in human ovarian and colorectal cancer treated with chemotherapy and bevacizumab or other VEGF inhibitors [71, 72].Our study has several limitations. Patients selected for surgical resection upfront may not be the optimal control baseline for borderline or locally advanced disease. Aside from tumor size and vascular involvement, other factors such as age and comorbidities are also considered in the preoperative staging and greatly impact the decision to proceed with surgical resection [73]. PDAC is a complex disease and the TME in resectable tumors may differ from more invasive tumors. However, pre-clinical studies in mice as well as studies which compared benign lesions versus invasive PDAC have shown that the TME of PDAC is immunosuppressive even at early stages of the disease [6, 53]. Another limitation is that samples belonging to the losartan+FFX+CRT and FFX+CRT groups were not obtained from a randomized clinical trial. In examining the responders and non-responders from the losartan+FFX+CRT group, it is unknown what factors determined the pathological response to neoadjuvant therapy since this was a retrospective study with a limited number of samples and no simultaneous control arm. Hence, there might be confounding factors which could influence the transcriptomic and cellular differences observed here. Finally, the NanoString approach was limited to a set of genes included in the panel and we focused our IF on a select group of cell types. These limitations notwithstanding, our transcriptomic and IF analysis provides novel insights on the effects of FFX+CRT compared to losartan+FFX+CRT on the expression and functional activity of invasion- and immune-related genes and the differences in cell populations between pathological responders and non-responders in the PDAC TME.

In conclusion, the addition of losartan to FFX+CRT reduced pro-invasion and immunosuppression related genes in PDAC, which was associated with improved treatment outcomes in patients with LAPC. Furthermore, patients with a complete/near complete pathological response in the losartan+FFX+CRT-treated group had reduced FOXP3+Tregs, decreased FOXP3+ cancer cells and increased CD8+ T cells. Our findings suggest that the addition of losartan to FFX+CRT further reprograms the PDAC TME to become immunostimulatory. Our results offer a potential explanation for the OS benefit observed in a retrospective analysis of GI patients who received ASIs along with immune-checkpoint blockade [74]. Overall, our findings bear importance, as they would not only reveal how losartan may synergize with emerging cytotoxic regimens, but also provide valuable information for overcoming the resistance to immune checkpoint blockers in PDAC.

Translational Relevance.

Our phase II trial in patients with locally advanced pancreatic ductal adenocarcinoma (PDAC) – stemming from our preclinical findings in murine models – showed that the angiotensin II receptor blocker losartan plus FFX chemotherapy and chemoradiation (CRT) led to high rates of surgical resection. Here, we identified the potential mechanisms of benefit of neoadjuvant losartan+FFX+CRT. We show that FFX+CRT improved the expression of genes linked to vascular normalization, transendothelial migration of leukocytes, T cell activation and dendritic cell maturation, effects mediated by chemotherapy. Losartan+FFX+CRT downregulated immunosuppression and pro-invasion genes which were associated with improved overall survival. Longitudinal analysis of multiple biomarkers showed that losartan induced changes in circulating levels of IL-8, sTie2 and TGF-β. Lastly, PDAC lesions from pathologic responders in the losartan+FFX+CRT-treated group had decreased Tregs, fewer immunosuppressive FOXP3+ cancer cells and increased CD8+ T cells. These findings suggest that losartan may potentiate the benefit of FFX+CRT by reducing immunosuppression.