Hepatic Artery Embolization Reduces Tumor-infiltrating-CD8-positive Lymphocytes in an Orthotopic Rat Hepatocellular Carcinoma Model

Abstract

Purpose:

Hepatic artery embolization is a locoregional treatment for hepatocellular carcinoma that may modulate the tumor immune microenvironment. However, its effects on tumor-infiltrating cluster of differentiation 8-positive lymphocytes remain unclear. This study evaluates changes in tumor-infiltrating cluster of differentiation 8-positive lymphocytes following hepatic artery embolization in an orthotopic rat hepatocellular carcinoma model using flow cytometry and immunohistochemistry.

Material and Methods:

Orthotopic hepatocellular carcinoma was established in rats using N1S1 cells. Animals were divided into an hepatic artery embolization group (n = 11) and a control group (n = 11). One week after the procedures, liver tumor tissues were harvested, and flow cytometry and immunohistochemistry were performed to quantify total and exhausted cluster of differentiation 8-positive lymphocytes.

Results:

Tumor-infiltrating cluster of differentiation 8-positive lymphocytes significantly decreased in the hepatic artery embolization group compared to controls by both flow cytometry (0.13 ± 0.09 vs. 0.46 ± 0.27 × 10⁶/g, p < 0.01) and immunohistochemistry (14.67 ± 15.17 vs. 31.56 ± 20.98 cells/field of view, p < 0.05). However, the proportion of exhausted cluster of differentiation 8-positive lymphocytes by flow cytometry remained unchanged (8.24 ± 3.66% vs. 8.88 ± 2.60%, p = 0.72).

Conclusions:

Hepatic artery embolization reduces the number of tumor-infiltrating cluster of differentiation 8-positive lymphocytes without altering the proportion of exhausted T cells. These findings suggest that, to increase hepatic artery embolization efficacy in combination with immunotherapy, a strategy of recruiting more cluster of differentiation 8-positive lymphocytes into the tumor tissue may be necessary.

Introduction

Recently, combination strategies of hepatic artery embolization (HAE) or transcatheter arterial chemoembolization (TACE) and immune checkpoint inhibitors (ICIs) have gained attention for treating HCC. Currently, several clinical trials are ongoing [1, 2]. The efficacy of ICIs is strongly influenced by the population of tumor-infiltrating effector cytotoxic cluster of differentiation 8 (CD8)-positive lymphocytes [3, 4]. However, the effects of HAE/TACE on tumor-infiltrating CD8-positive lymphocytes have not been sufficiently evaluated. HAE induces ischemia and hypoxia within the tumor, which subsequently alters the tumor microenvironment by promoting angiogenic signaling, immunosuppressive cytokine release, and recruitment of immunosuppressive cells such as tumor-associated macrophages [5-7]. These changes may limit the infiltration and activity of effector CD8-positive lymphocytes. Thus, understanding how HAE modulates the immune microenvironment, especially CD8-positive lymphocyte dynamics, is essential to elucidate the biological rationale for combining HAE with ICI therapy and to establish the therapeutic significance of such a strategy.

Therefore, the effects of HAE on tumor-infiltrating CD8-positive lymphocytes were assessed in tumor tissues collected after HAE using flow cytometry and immunohistochemistry (IHC) in a rat orthotopic HCC model.

Material and Methods

All animal procedures complied with the study protocol approved by our institution's animal care and use committee. All procedures were conducted according to the Fundamental Guidelines for Proper Conduct of Animal Experiments and Related Activities in Academic Research Institutions in Japan. All surgical procedures used in this study were performed under anesthesia with intraperitoneal injection of 0.3 mg/kg of medetomidine (Domitol; Nippon Zenyaku Kogyo Co., Ltd. Fukushima, Japan), 4.0 mg/kg of midazolam (Dormicum injection; Astellas Pharma Inc., Tokyo, Japan), and 5.0 mg/kg of butorphanol (butorphanol tartrate; Fujifilm Wako Pure Chemical Corp., Osaka, Japan).

Establishment of the rat HCC model and HAE procedure

After 22 male Sprague-Dawley rats obtained at 10 weeks of age from a vendor (CLEA Japan, Inc., Tokyo, Japan; average body weight, 372 g) had been acclimatized for 1 week, orthotopic implantation of N1S1 rat HCC cells (DS Pharma Promo Co., Ltd., Osaka, Japan) was performed using mini-laparotomy. After anesthesia and a midline abdominal incision, 1 × 10⁷ of N1S1 cells in 100 μL of phosphate buffer solution were injected under the hepatic capsule of the left lateral lobe. The abdominal incisions were closed using a two-layer technique. One week after orthotopic implantation of N1S1 cells, the rats were randomly divided into an HAE group (n = 11) and a control group (n = 11). Hepatic tumor engraftment was confirmed by ultrasound before performing HAE; however, the engrafted tumors generally exhibited ill-defined margins on ultrasound, making accurate measurement of tumor diameter or volume infeasible. After anesthesia was administered in the same manner as for tumor implantation, a left paramedian incision was made in the neck, and the left common carotid artery was exposed. A 20-gauge intravenous catheter was used to cannulate the left common carotid artery. In the HAE group, a 1.6-Fr microcatheter (Carnelian Marvel; Tokai Medical Products, Inc., Aichi, Japan) was inserted through the 20-gauge intravenous catheter and advanced to the proper hepatic artery using a 0.014-inch guidewire (Asahi Chikai; Asahi INTECC Co., Ltd., Aichi, Japan) under fluoroscopy. Then, HAE was performed using porous gelatin (Serescue; Nippon Kayaku Co., Ltd., Tokyo, Japan) particles, which were prepared by pumping 20 times with two injection syringes and a three-way stopcock. After stasis of hepatic artery blood flow was determined by angiography using contrast medium, the catheters were removed, and the carotid artery was ligated. For the control group animals, cannulation of the common carotid artery was performed, but HAE was not performed. Then, the neck incision was sutured. The animals were allowed to recover.

Flow cytometry and immunohistochemistry

One week after the procedure, the animals were euthanized. Then, implanted tumor tissues were harvested and used for flow cytometry (n = 8 in each group) and immunohistochemistry (IHC) (n = 3 in each group) analysis.

Tumor tissues for flow cytometry were dissociated to obtain single-cell suspensions using a cell dissociation reagent (Dri Tumor & Tissue Dissociation Reagent; BD Biosciences, Franklin Lakes, NJ, USA) [8]. Then, tumor-infiltrating mononuclear cells (TIMCs), including lymphocytes, monocytes, and dendritic cells, were isolated by density gradient centrifugation using a density gradient medium (Ficoll-paque Premium 1.084; GE HealthCare, Chicago, IL, USA). After cell counting, isolated TIMCs were stained with the following fluorochrome-conjugated antibodies: allophycocyanin conjugated anti-CD8a antibody (200610; BioLegend, San Diego, CA, USA), fluorescein isothiocyanate conjugated anti-lymphocyte activation gene 3 (LAG-3) antibody (MBS9461124; MyBioSource Inc., San Diego, CA, USA), and phycoerythrin-conjugated anti-programmed cell death 1 (PD-1) antibody (bs-1867R; Bioss Inc., Woburn, MA, USA). After washing the cells using phosphate-buffered saline with 1% fetal bovine serum, they were analyzed with a flow cytometer (BD FACSCalibur; BD Biosciences, San Jose, CA, USA).

For IHC analysis, tumor tissues were fixed in 4% paraformaldehyde. After dehydration with ethanol, the tissues were embedded in paraffin. Then, using a microtome, 5 mm sections of the samples were prepared and collected directly onto the slide. After endogenous peroxidase was blocked, the tissue sections were incubated with an antibody against CD8 (550298; BD Pharmigen, San Diego, CA, USA), PD-1 (bs-1867R; Bioss Inc.), and LAG-3 (ab235907; Abcam plc., Cambridge, UK). Next, the sections were incubated with the associated secondary antibodies (Alexa Fluoro 594 goat anti-rabbit IgG and Alexa Fluoro 488 donkey anti-mouse IgG). The samples were then sealed with mounting medium containing 4',6-diamidino-2-phenylindole (DAPI) (SlowFade™ Gold Antifade Mountant with DAPI; Thermo Fisher Scientific K.K., Tokyo, Japan). After sealing, the samples were observed. Fluorescence-labeled images of each slide were captured using a slide scanner (Slideview VS200; Evident Scientific, Inc., Tokyo, Japan). Three regions on each slide were acquired at high magnification (200×). The cells positive for CD8, PD-1, and LAG-3 in each sample were counted manually. In total, three samples per group were analyzed, corresponding to nine regions per group.

The TIMC count per unit tumor weight was evaluated using an automated cell counter (Countess II FL; Thermo Fisher Scientific K.K., Tokyo, Japan). The CD8-positive cell count was evaluated using a flow cytometer. For this study, exhausted CD8-positive cells were defined as the subpopulation of CD8-positive cells co-expressing PD-1 and LAG-3. Co-expression of these inhibitory receptors has been widely recognized as a representative marker of T-cell exhaustion in previous studies, and was therefore adopted as the criterion in our analysis [9, 10]. The counts of CD8-positive cells and PD-1 or LAG-3-positive cells among CD8-positive cells in each FOV were also assessed using IHC.

Statistical analysis

All data are expressed as the mean and standard error. Student's t-test was used to compare differences between the control and HAE groups. Statistical analyses were conducted using software (PRISM 8, 2018; GraphPad Software Inc., San Diego, CA, USA). A p-value of less than 0.05 was inferred as representing a significant difference.

Results

The TMIC counts per unit tumor weight were significantly lower in the HAE group (2.16 ± 0.53 ×10⁶/g) than in the control group (7.08 ± 3.75 ×10⁶/g; p < 0.01) (Figure 1a). Results of flow cytometry analysis revealed that the CD8-positive cell count was significantly lower in the HAE group (0.13 ± 0.09 ×10⁶/g) than in the control group (0.46 ± 0.27 ×10⁶/g; p < 0.01) (Figure 1b). The proportions of exhausted CD8-positive cells in the HAE and control group were comparable (8.24 ± 3.66% vs. 8.88 ± 2.60%, p = 0.72) (Figures 1c and 2).

Figure 1.

Changes in TIMC counts, CD8-positive cell counts, and the exhausted CD8-positive cell population after HAE were evaluated using a cell counter and flow cytometer.

(a) TIMC counts per unit tumor weight decreased significantly after HAE (control group vs. HAE group, 7.08 ± 3.75 × 10⁶/g and 2.16 ± 0.53 × 10⁶/g; p < 0.01).

(b) CD8-positive cell counts per unit tumor weight decreased significantly after HAE (control group vs. HAE group, 0.46 ± 0.27 × 10⁶/g and 0.13 ± 0.09 × 10⁶/g; p < 0.01).

(c) Population of PD-1 and LAG-3 double-positive exhausted cells among all CD8-positive cells remained unchanged after HAE (control group vs. HAE group, 8.88 ± 2.60% and 8.24 ± 3.66%; p = 0.72).

CD8: cluster of differentiation 8; HAE: hepatic artery embolization; LAG-3: lymphocyte activation gene 3; PD-1: programmed cell death 1; TIMC: tumor-infiltrating mononuclear cell

Figure 2.

Representative flow cytometry images. The populations of PD-1 and LAG-3 double-positive cells among the CD8-positive cells were comparable between the control and the HAE groups.

CD8: cluster of differentiation 8; HAE: hepatic artery embolization; LAG-3: lymphocyte activation gene 3; PD-1: programmed cell death 1

The IHC analysis also showed a marked decrease in CD8-positive cells in the HAE group (14.67 ± 15.17 cells/FOV, p < 0.05) compared with the control group (31.56 ± 20.98 cells/FOV) (Figures 3a, 4, and 5). The populations of the PD-1-positive and the LAG-3-positive cells among the CD8-positive cells remained unchanged after HAE (p = 0.82 and p = 0.50) (Figures 3b, 3c, 4, and 5).

Figure 3.

Changes in CD8-positive, PD-1-positive, and LAG-3-positive cells after HAE on immunohistochemistry.

(a) The CD8-positive cell counts decreased significantly after HAE (control group vs. HAE group, 31.56 ± 20.98 cells/FOV and 14.67 ± 15.17 cells/FOV; p < 0.05).

(b) The population of PD-1-positive cells among all CD8-positive cells remained unchanged after HAE (control group vs. HAE group, 15.62 ± 11.84% and 14.25 ± 13.58%; p = 0.82).

(c) Population of LAG-3 positive cells among all CD8 positive cells remained unchanged after HAE (control group vs. HAE group, 6.89 ± 2.86% and 10.60 ± 15.78%; p = 0.50).

CD8: cluster of differentiation 8; HAE: hepatic artery embolization; LAG-3: lymphocyte activation gene 3; PD-1: programmed cell death 1

Figure 4.

Immunohistochemistry of CD8 and PD-1 in rat hepatocellular carcinoma. In the HAE group, the number of CD8-positive cells decreased.

CD8: cluster of differentiation 8; DAPI: 4’,6-diamidino-2-phenylindole; PD-1: programmed cell death 1

Figure 5.

Immunohistochemistry of CD8 and LAG-3 in rat hepatocellular carcinoma. In the HAE group, the number of CD8-positive cells decreased.

CD8: cluster of differentiation 8; HAE: hepatic artery embolization; LAG-3: lymphocyte activation gene 3

Discussion

The results of this study demonstrated that the number of tumor-infiltrating CD8-positive cells decreased significantly after HAE. Recently published clinical studies also support the results obtained in this study [11, 12]. Tan et al. [11] reported that TACE decreased the number and antitumor function of tumor-infiltrating CD8-positive T cells. Similarly, Pinato et al. [12] reported that post-TACE intratumoral tissue samples had lower concentrations of tumor-infiltrating CD8-positive cells in patients with HCC. Tan et al. [11] performed TACE with doxorubicin and microbeads, whereas Pinato et al. [12] reported TACE using doxorubicin, lipiodol, and gelatin powder. Interestingly, both studies demonstrated a reduction in tumor-infiltrating CD8-positive T cells, consistent with our findings. These results suggest that the decrease in CD8-positive T cells occurs irrespective of the embolic material or the use of chemotherapeutic agents, and may be related to the hypoxic response induced by embolization. Nevertheless, this remains speculative and requires further investigation.

Although the mechanisms underlying the decrease in tumor-infiltrating CD8-positive lymphocytes after HAE/TACE are not adequately understood, recent evidence suggests that tumor-associated macrophages critically regulate the CD8-positive T-cell infiltration by suppressing C-X-C motif chemokine ligand 9 expression, thereby reducing chemotactic recruitment, and by upregulating galectin-1, which induces T-cell apoptosis and functional inhibition [11, 13]. These mechanisms may explain, at least in part, the reduction in tumor-infiltrating CD8-positive (AE-1) T-cells observed after HAE in our study.

These findings provide new perspectives for developing effective strategies for combining the HAE/TACE and immunotherapy. Currently ongoing clinical trials mostly involve combinations of HAE/TACE and ICIs intended to reinvigorate exhausted cells [1, 2, 14]. However, this study demonstrated that the population of exhausted CD8-positive cells did not change, whereas the number of CD8-positive cells decreased after HAE. Therefore, the mere combination of ICIs might not exert the expected effects of HAE if the number of tumor-infiltrating CD8-positive cells is reduced. The findings from this study suggest that, to increase HAE/TACE efficacy in combination with immunotherapy, a strategy of recruiting more CD8-positive lymphocytes into the tumor tissue (e.g., in combination with the tyrosine kinase inhibitors [TKIs]) will be necessary [15]. Recent evidence indicates that TKIs such as lenvatinib can modulate the tumor immune microenvironment by restoring interferon-γ signaling and promoting CD8-positive T-cell infiltration [15]. Furthermore, TKIs have been reported to normalize tumor vasculature and facilitate T-cell trafficking, thereby providing a mechanistic rationale for combining TKIs with locoregional therapies such as HAE/TACE [16, 17]. These findings support our hypothesis (AE-2) that HAE in combination with TKIs may enhance antitumor immunity.

It is noteworthy that in our rat HCC model, the proportion of exhausted CD8-positive lymphocytes among total CD8-positive lymphocytes was approximately 8%-9% in both the control and HAE groups. Previous clinical reports in human HCC have shown that exhausted CD8-positive lymphocytes typically comprise about 10%-30% of intratumoral CD8-positive lymphocytes (AE-3), although the exact proportion varies depending on tumor stage, heterogeneity, and analytical methods [3, 18]. Therefore, the baseline of exhausted CD8-positive lymphocytes observed in our model appears to be within the lower range of what has been described in human HCC, supporting the translational relevance of our results.

The use of a rat orthotopic HCC model is a readily apparent limitation of this study for extrapolating the results to human HCC. In addition, only gelatin sponge particles were used as the embolic material in this study, although the type of embolic material might influence changes in the tumor immune microenvironment [19]. Furthermore, although hepatic tumor engraftment was confirmed by ultrasound before HAE, the tumors typically showed ill-defined margins, preventing accurate tumor size measurement; therefore, baseline tumor volumes could not be compared between groups. (R#1-1) Despite these limitations, the results of this study provide a new perspective on HAE and the immune system. Given that HAE has the potential to reduce tumor-infiltrating CD8-positive lymphocytes, a strategy of recruiting more CD8-positive lymphocytes into the tumor tissue will be necessary to increase the efficacy of HAE when used in combination with immunotherapy.

Author Contributions

Research Planning: HT, TH, KY

Animal Experiments: HT, RW, EU, HK

Manuscript Preparation: HT

Manuscript Review and Editing: HT, RW, EU, HK, TH, KY

Conflicts of Interest

KY is the president of JSIR. This study is partially financially supported by Eisai Co., Ltd., Tokyo, Japan. TH is an employee of Eisai Co., Ltd.

Author Haruyuki Takaki is a member of the Editorial Board of this journal but was not involved in the editorial decision or peer review of this manuscript.