Regorafenib combined with PD1 blockade increases CD8 T-cell infiltration by inducing CXCL10 expression in hepatocellular carcinoma

Abstract

Background and purpose: Combining inhibitors of vascular endothelial growth factor and the programmed cell death protein 1 (PD1) pathway has shown efficacy in multiple cancers, but the disease-specific and agent-specific mechanisms of benefit remain unclear. We examined the efficacy and defined the mechanisms of benefit when combining regorafenib (a multikinase antivascular endothelial growth factor receptor inhibitor) with PD1 blockade in murine hepatocellular carcinoma (HCC) models.

Basic procedures: We used orthotopic models of HCC in mice with liver damage to test the effects of regorafenib-dosed orally at 5, 10 or 20 mg/kg daily-combined with anti-PD1 antibodies (10 mg/kg intraperitoneally thrice weekly). We evaluated the effects of therapy on tumor vasculature and immune microenvironment using immunofluorescence, flow cytometry, RNA-sequencing, ELISA and pharmacokinetic/pharmacodynamic studies in mice and in tissue and blood samples from patients with cancer.

Main findings: Regorafenib/anti-PD1 combination therapy increased survival compared with regofarenib or anti-PD1 alone in a regorafenib dose-dependent manner. Combination therapy increased regorafenib uptake into the tumor tissues by normalizing the HCC vasculature and increasing CD8 T-cell infiltration and activation at an intermediate regorafenib dose. The efficacy of regorafenib/anti-PD1 therapy was compromised in mice lacking functional T cells (Rag1-deficient mice). Regorafenib treatment increased the transcription and protein expression of CXCL10-a ligand for CXCR3 expressed on tumor-infiltrating lymphocytes-in murine HCC and in blood of patients with HCC. Using Cxcr3-deficient mice, we demonstrate that CXCR3 mediated the increased intratumoral CD8 T-cell infiltration and the added survival benefit when regorafenib was combined with anti-PD1 therapy.

Principal conclusions: Judicious regorafenib/anti-PD1 combination therapy can inhibit tumor growth and increase survival by normalizing tumor vasculature and increasing intratumoral CXCR3+CD8 T-cell infiltration through elevated CXCL10 expression in HCC cells.

Keywords: combination; drug therapy; liver neoplasms; programmed cell death 1 receptor.

Conflict of interest statement

Competing interests: LG reports personal fees from Agios Pharmaceuticals, Alentis Therapeutics, QED Therapeutics, H3 Biomedicine, Taiho Pharmaceuticals, Debiopharm, Incyte Corporation, SIRTEX and AstraZeneca. AXZ is a consultant/advisory board member for Bayer. RKJ received honorarium from Amgen and consultant fees from Chugai, Ophthotech, Merck, SPARC, SynDevRx and XTuit. Dr Jain owns equity in XTuit, Enlight, SPARC, SynDevRx and Accurius Therapeutics and serves as a paid member of the boards of trustees of Tekla Healthcare Investors, Tekla Life Sciences Investors, Tekla Healthcare Opportunities Fund and Tekla World Healthcare Fund. He is a member of the scientific advisory board of Accurius Therapeutics. DZ, LF and the spouse of LLM are Bayer employees. MC is an AstraZeneca employee. DGD received consultant fees from Bayer, Simcere, Surface Oncology and Bristol Myers Squibb, and research grants from Bayer, Exelixis and Bristol Myers Squibb. No potential conflicts of interest were disclosed by other authors.

© Author(s) (or their employer(s)) 2020. Re-use permitted under CC BY. Published by BMJ.

Figures

Figure 1

Dose-dependent effects of regorafenib on vascular normalization when used in combination with programmed cell death protein 1 (PD1) blockade in RIL-175 hepatocellular carcinoma (HCC) model. (A, B) Mice with established (4–5 mm in diameter) HCCs were treated with anti-PD1 therapy (P) alone or in combination with 5 mg/kg (R5), 10 mg/kg (R10) or 20 mg/kg (R20) regorafenib versus control (C) (n=8 mice per group, 1-week treatment). We used immunofluorescence (IF) for evaluation of tumor vessels by CD31 and α-smooth muscle actin (α-SMA) immunostaining. Quantitative analysis of IF data showed increased pericyte coverage only after R10+P treatment (plots in B). Representative IF of tumor sections after staining for CD31 (for endothelial cells in green) and α-SMA (for pericytes in red) and DAPI counterstaining (in blue). Five random fields from the tumor center (defined as no inclusion of the edge of the tumor) were evaluated for each sample. Data represent mean values. Scale bars=500 µm. (C) Representative IF confocal microscopy images of tumor vessels using and CD31 staining (green) and tissue hypoxia using CA-IX staining (red) of whole liver sections; scale bars=2 mm. (D) Tissue hypoxia was significantly reduced by 10 mg/kg regorafenib treatment alone or with anti-PD1 blockade (n=5 mice, 12 days of treatment, 5 images/tumor sample). (E) Representative IF confocal microscopy images of mature tumor vessels using CD31 staining (green) and α-SMA staining (red) of RIL-175 tumors; scale bars, 100 µm. (F) CD31+α-SMA+ mature vessel density was significantly increased in the combination treatment group (n=5 mice, 12 days of treatment, images/tumor sample). Data represent mean±SEM. *P

Figure 2

Intermediate dose of regorafenib with antiprogrammed cell death protein 1 (anti-PD1) therapy increases intratumoral CD8+ cytotoxic T lymphocyte (CTL) infiltration and activation in RIL-175 model. (A) Mice with established (4–5 mm in diameter) hepatocellular carcinomas (HCCs) were treated with anti-PD1 therapy (P) alone or in combination with 5 mg/kg (R5), 10 mg/kg (R10) or 20 mg/kg (R20) regorafenib versus control (C) (n=8 mice per group, 1 week treatment). We used immunofluorescence (IF) for evaluation of T-cell infiltration by CD8 immunostaining. Representative IF of tumor sections after staining for CD8 (for T cells in red) and DAPI counterstaining (in blue). Five random fields from the tumor center (defined as no inclusion of the edge of the tumor) were evaluated for each sample. Data represent mean values. Scale bars=500 µm. (B) Quantitative analysis of IF data showed a significant increase in CD8+ T cells after R10+ P treatment compared with all other groups. (C) Regorafenib (10 mg/kg daily, R10) combined with anti-PD-1 therapy (P) significantly increased the number of CD8+ T cells per gram RIL-175 tumor tissue measured by flow cytometry versus control or R10-treated mice (n=11–13 mice, 12 days of treatment). (D) Representative IF confocal microscopy images of CD8 T-cell distribution (in red) in whole RIL-175 tumor tissue sections; CD31 staining of vessels in green and DAPI counterstaining in blue; scale bars=2 mm. (E) Quantification of intratumoral CTL infiltration in RIL-175 HCCs showing a more substantial increase in intratumoral CTL infiltration in R10/P-treated group, which was significantly higher than each treatment alone or control (n=5 mice, 12 days of treatment, 5 random fields from the tumor center (defined as no inclusion of the edge of the tumor) per sample). (F) R10/P combination therapy significantly increased the number of CD8+ interferon gamma (IFN-γ)+CTLs per gram of RIL-175 tumor tissue measured by flow cytometry versus each treatment alone or control (n=10–13 samples, 12 days of treatment). *P

Figure 3

Combined treatment with regorafenib with antiprogrammed cell death protein 1 (anti-PD1) antibodies inhibits signal transducer and activator of transcription (STAT) activation and increases CXCL10 production by hepatocellular carcinoma (HCC) cells. (A) Changes induced by combination of regorafenib (10 mg/kg) with anti-PD1 antibody in immune-related pathways from the gene set enrichment analysis (GSEA) using curated gene sets in RIL-175 HCC model (see also online supplemental dataset S1 and figure S2E). (B, C) Changes induced by combination of regorafenib (10 mg/kg) with anti-PD1 antibody in STAT3 and STAT1 activity (B), showing an increased ratio of phosphorylated (p)-STAT1/p-STAT3 (normalized to total STAT1 and STAT3, respectively) by western blot analysis of whole RIL-175 HCC tissue (C) (n=4 mice). (D) Treatment with regorafenib increases CXCL10 expression by RIL-175 HCC cells in vitro after 36 hours, in a dose-dependent manner (n=6). Culture medium alone was used as a negative control (NC). (E, F) Changes induced by regorafenib (1 µM) in STAT3 and STAT1 activity (E), showing an increased ratio of p-STAT1/p-STAT3 (normalized to total STAT1 and STAT3, respectively) by western blot analysis of RIL-175 HCC cells (F). (G, H) Representative in situ hybridization (ISH) pictures of CXCL10 (red dots, arrow heads) and CXCR3 (green dots) on RIL-175 HCC tissue from the control group (G) versus combination therapy (RP) group (H). Scale bars=50 µm (G, H). (I) Combination therapy significantly increases in the number of HCC cells expressing CXCL10 compared with control. Data represent mean±SEM. (J, K) Combination therapy increases CXCL10 protein expression in whole RIL-175 tumor tissue 6 hours and 12 hours after in vivo when regorafenib treatment was used at doses of 10 mg/kg (R10) and 20 mg/kg (R20) (n=3) by western blot analysis (J); quantitative data of CXCL10/β-actin ratios are shown in (K). C, control; P, anti-PD1 antibody; R, regorafenib 10 mg/kg; R10-P, regorafenib 10 mg/kg+anti-PD1 antibody; n=3–4 per group. *P

Figure 4

Intermediate dose of regorafenib shows efficacy when combined with antiprogrammed cell death protein 1 (anti-PD1) blockade in the RIL-175 mouse model of advanced hepatocellular carcinoma (HCC). (A) Example of tumor implantation and treatment schedule for survival studies in HCC models. (B, C) Survival experiment data from RIL-175 orthotopic mouse model. Tumor growth was significantly delayed in the group treated with a combination of regorafenib 10 mg/kg and anti-PD1 antibody (aPD1) compared with the other groups (n=10–13 mice) (B); moreover, this combination significantly prolonged overall survival (C). (D) Survival experiment data from HCA-1 highly metastatic orthotopic mouse model. Mice treated with combination with R10+P show a non-significant trend for increased survival versus control (HR=0.48, p=0.075) (n=11–13 mice). (E) Tumor growth is delayed after combination of R10+P in a spontaneous HCC model using Mst1–/–Mst2F/– mice (n=6–9 mice). (F) Treatment with regorafenib (10 mg/kg daily, R10) alone or in combination with anti-PD-1 therapy shows no benefit in survival (Kaplan-Meier survival distributions) compared with control. *P<0.05, **p<0.01, ***p<0.001. C, control; P, anti-PD1; R10, regorafenib 10 mg/kg daily; R10+P, regorafenib 10 mg/kg+anti-PD1. Error bars, SEM.

Figure 5

Increased intratumoral infiltration by CD8+ cytotoxic T lymphocytes (CTLs) after regorafenib plus antiprogrammed cell death protein 1 (anti-PD1) treatment is mediated by CXCL10/CXCR3 axis. (A, B) Representative immunofluorescence (IF) confocal microscopy images of whole liver sections (upper panel, low magnification, scale bars=2 mm; lower panel, high magnification, scale bars=50 µm) showing increased intratumoral localization of CD8+ CTLs co-localized with areas of cell apoptosis in orthotopic RIL-175 tumors in Cxcr3+/+/C57Bl/6 mice (upper panel) compared with orthotopic tumors grown in Cxcr3–/–/C57Bl/6 mice (lower panel) (A) after 1 week of regorafenib/anti-PD1 combination treatment. (B) Quantification of IF data showing that intratumoral CTL infiltration after regorafenib/anti-PD1 combination treatment is prevented when tumors are grown in Cxcr3–/–/C57Bl/6 mice (n=5 samples, 5 random areas/samples): C, control; R10+P, regorafenib 10 mg/kg+anti-PD1 antibody. (C) The added survival benefit of regorafenib with anti-PD1 blockade over anti-PD1 therapy alone is compromised in Cxcr3–/–/C57Bl/6 mice bearing hepatocellular carcinoma (HCC) (n=9–10 mice). C, control; P, anti-PD1 antibody; R10, regorafenib 10 mg/kg; R10+P, regorafenib 10 mg/kg+anti-PD1 antibody. Data represent mean±SEM. *P<0.05; **p<0.01; ***p<0.001.

Figure 6

CXCL10 expression of in hepatocellular carcinoma (HCC) tissues and concentration in blood samples from patients with cancer. (A–C) CXCL10 expression by RNA-in situ hybridization (ISH) in human HCC samples; staining was scores ‘high’ (13%–17%) (A), medium (1%–11%) (B) or low (