Injectable Thermosensitive Hydrogel Targeting STAT3 Reprograms Neutrophils to Amplify Anti‐Tumor Immunity Post‐Radiofrequency Ablation in HCC

ABSTRACT

Radiofrequency ablation (RFA) for hepatocellular carcinoma (HCC) carries a high risk of recurrence due to residual tumors and the immunosuppressive tumor microenvironment. This study identifies that incomplete RFA activates the JAK2/STAT3 pathway in tumor‐associated neutrophils (TANs), polarizing them toward immunosuppressive phenotypes. In vitro, icaritin (ICT) inhibits TAN survival and migration, and reverses the immunosuppressive phenotype by suppressing STAT3 phosphorylation. To enable localized delivery, we developed an injectable hydrogel (HP@ICT) by leveraging the interfacial complexation of hydroxypropyl‐β‐cyclodextrin and the thermosensitive gelation of poloxamer 407. It is injected via the ablation needle into the residual cavity and undergoes in situ gelation, overcoming post‐ablation drug delivery barriers to enable sustained release of ICT. In vivo, HP@ICT effectively suppresses immunosuppressive TANs, reduces the proportion of PMN‐MDSCs, and promotes the infiltration and activation of CD8⁺ T cells without systemic toxicity. The hydrogel co‐delivering ICT and PD‐L1 inhibitor synergistically eliminates residual tumors and liver metastases post‐RFA by inducing systemic anti‐tumor immunity. Multi‐omics analyses reveal complementary mechanisms of ICT, which inhibits PD‐L1 expression in TANs, disrupts TAN‐tumor adhesion, and suppresses unsaturated fatty acid metabolism. Overall, with accessible materials and seamless integration with RFA, this strategy optimizes RFA efficacy, providing a safe and translational solution for HCC recurrence.

A schematic of an injectable thermosensitive hydrogel platform for augmented immunotherapy against RFA‐induced residual HCC via TAN reprogramming. The hydrogel platform leverages the solubilization enhancement of HP‐β‐CD and thermosensitive gelation properties of P407, effectively reversing immunosuppressive TANs via inhibiting JAK2/STAT3 pathway. Co‐delivery of ICT and PD‐L1 inhibitor within the hydrogel synergistically enhances therapeutic efficacy and eliminates residual tumors and metastases post‐RFA.

1. Introduction

Liver cancer ranks sixth in global incidence and third in all‐cause cancer mortality, with hepatocellular carcinoma (HCC) accounting for 75%–85% of cases [1]. Radiofrequency ablation (RFA) is a first‐line curative treatment for early‐stage HCC due to its minimal complications and short hospital stays [2]. Although RFA achieves favorable outcomes in very‑early‑stage HCC (<2 cm) with 5‑year survival rates approaching 90%, incomplete RFA (iRFA) remains a major clinical challenge, particularly for larger tumors (>5 cm) where complete ablation is difficult to achieve [3, 4]. According to clinical data, iRFA treatment commonly results in rapid local recurrence and worsening health conditions, thereby leading to a poor prognosis in patients with advanced‑stage HCC [5]. More importantly, emerging evidence indicates that the RFA‐induced residual tumors foster an immunosuppressive tumor microenvironment characterized by aggressive residual tumor growth and metastasis, which poses a significant challenge in the management of HCC [6, 7, 8]. Thus, it is imperative to develop adjuvant strategies for bolstering the immune response and preventing the growth and metastasis of HCC after RFA.

Neutrophils constitute the predominant subpopulation of circulating leukocytes, also serving as critical modulators of the tumor microenvironment. The tumor‐associated neutrophils (TANs) have been identified as a prospective target for tumor immunotherapy due to its functional plasticity [9, 10]. The N1 subtype enhances cytotoxicity and immuno‐activating cytokines (e.g., TNF‐α, CCL3), while the immunosuppressive TANs, including N2 TANs and polymorphonuclear myeloid‐derived suppressor cells (PMN‐MDSCs), could directly suppress CD8⁺ T cell activity through mediators such as ROS, arginase 1, and PD‐L1 [11, 12]. Studies have shown that immunosuppressive TANs are associated with unfavorable prognosis in HCC patients, promoting immune evasion and tumor growth [13]. Our recent study indicated that mice treated with RFA exhibited increased accumulation of TANs and induced N2 TANs polarization within residual HCC [14]. Nevertheless, the mechanism by which RFA induces the immunosuppressive TANs in the residual tumor is still poorly understood.

Emerging research has demonstrated that the JAK2/STAT3 pathway activation serves as a pivotal driver in reprogramming TANs toward immunosuppressive phenotypes [15]. Icaritin (ICT), a bioactive prenylflavonoid derived from Chinese herbal Epimedium genus, has been reported to potently suppress the JAK2/STAT3 pathway and has been approved as the clinical medicine derivative for advanced HCC therapy [16, 17]. Thus, ICT can be expected to be a compelling candidate for reversing immunosuppressive TANs in residual HCC after RFA. Although ICT is clinically administered via oral capsules for HCC, its therapeutic efficacy is hindered by limitations including pronounced hepatic first‐pass metabolism, poor solubility, and lower bioavailability [18, 19]. Besides, the disruption of tumor‐associated vasculature and fibrotic barriers post‐RFA further compromise drug delivery to residual lesions [20]. It is crucial to develop a novel drug delivery system to enhance the aqueous solubility of ICT and prolong the retention time in residual HCC. Injectable and in situ‐forming hydrogels are regarded as promising carriers to solve these challenges [21]. Given that hydroxypropyl β‐cyclodextrin (HP‐β‐CD) is characterized by a hydrophobic central cavity, its unique structure facilitates the formation of inclusion complexes with low‐polarity drug molecules, thereby enhancing the aqueous solubility of drugs [22, 23]. As a biocompatible FDA‐approved polymer, poloxamer 407 (P407) maintains a flowable liquid state at room temperature but undergoes a phase transition upon reaching the critical temperature threshold [24]. This thermosensitive behavior contributes to forming an in‐situ gel after radiofrequency ablation. Thus, the delivery system of HP‐β‐CD and P407 is expected to achieve localized drug accumulation and sustained release specifically at residual tumors post‐RFA, representing an attractive strategy to improve the efficacy of ICT.

In our study, residual HCC post‐RFA was found to promote activation of the JAK2/STAT3 pathway in TANs and polarize them toward immunosuppressive phenotypes, which accelerated residual tumor progression. Herein, we developed a facile and injectable thermosensitive hydrogel by integrating HP‐β‐CD with P407 as carriers, which can change from a liquid state to a hydrogel state in situ once the temperature reaches the critical threshold post‐RFA (Scheme 1). The system prolonged the retention time of ICT and PD‐L1 inhibitor in the residual HCC and synergistically reversed the immunosuppression, ultimately resulting in notable regression of residual tumors and metastases. Overall, the localized delivery system is a safe and effective treatment, providing a promising strategy to alleviate residual HCC induced by RFA.

SCHEME 1.

A schematic of an injectable thermosensitive hydrogel platform for augmented immunotherapy against RFA‐induced residual HCC via TAN reprogramming. The hydrogel platform leverages the solubilization enhancement of HP‐β‐CD and thermosensitive gelation properties of P407, effectively reversing immunosuppressive TANs via inhibiting JAK2/STAT3 pathway. Co‐delivery of ICT and PD‐L1 inhibitor within the hydrogel synergistically enhances therapeutic efficacy and eliminates residual tumors and metastases post‐RFA.

2. Materials and Methods

2.1. Reagents and Materials

Icaritin (Catalog No. II0040) was obtained from Solarbio. P407 (Catalog No. 16758) and HP‐β‐CD (Catalog No. 332607) were purchased from Sigma‐Aldrich. BMS‐202 (Catalog No. S7912) was purchased from Selleck.

2.2. Cell Lines and Mice

H22 (Catalog# STCC20036G) and Hepa1‐6 (Catalog# STCC20016G) were obtained from Servicebio (Wuhan, China). H22‐Luc was generated in our laboratory by stable transfection of the parental H22 cell line with a luciferase reporter construct, followed by limiting dilution cloning and validation of luciferase expression. Hepa1‐6 was cultured in DMEM while H22 and H22‐Luc were cultured in RPMI‐1640 containing 10% fetal bovine serum (FBS) and 1% penicillin‐streptomycin at 37°C and 5% CO₂. Female BALB/c mice at the age of 3–4 weeks were purchased from Guangdong Yaokang Biological Company and raised in the SPF animal laboratory to over 8 weeks of age for modeling. All animal experimental procedures were performed in accordance with the animal protocol approved by the Laboratory Animal Center of the Fifth Affiliated Hospital of Sun Yat‐sen University (202000139 and 2024060401).

2.3. Isolation of Neutrophils

Neutrophils were isolated from mouse bone marrow using a Percoll density gradient [14]]. Tibias and femurs were collected from euthanized mice, and bone marrow cells were flushed out by centrifugation and resuspended. A discontinuous Percoll gradient (2 mL each of 78%, 65%, and 55%) was prepared, and 2 mL of the bone marrow cell suspension was layered on top. After centrifugation, neutrophils were collected from the interface between the 78% and 65% layers. Purity was assessed by flow cytometry using anti‐CD11b‐AF700 and anti‐Ly6G‐FITC antibodies.

2.4. Sub‐Lethally Heated Cell Model and Harvesting Tumor Conditioned Media

Incomplete radiofrequency ablation (iRFA) or sub‐lethally heated cell model was established as described previously [25]. Culture dishes sealed with parafilm were heated at 44°C for 15 min. Upon being heated, they were moved to the 37°C incubator with 5% CO₂ for 48 h recovery. Cells cultured at 37°C were selected as the control group. The tumor conditioned media (TCM) was then harvested and filtered through 0.22 µm membrane to remove dead cell debris, and frozen at ‐20°C for subsequent experimental steps.

2.5. Transwell Assay

TAN migration was assessed using Transwell chambers (3 µm pore size, Corning/Greiner BioOne). Freshly isolated neutrophils (200 µL in serumfree medium) were placed in the upper compartment, and the lower compartment received 600 µL of TCM from the indicated treatment groups (control, iRFA, iRFA + 5 µM ICT, or iRFA + 10 µM ICT). After 6 h of incubation at 37°C, cells adhering to the membrane were fixed with 4% paraformaldehyde (15 min), stained with 1% crystal violet (30 min), and visualized by light microscopy. The percentage of migrated TANs was determined by counting cells in the lower chamber with a hemocytometer.

2.6. Apoptosis Assay

TANs were seeded in 6well plates and exposed to different treatments (control, iRFA, iRFA + 5 µM ICT, or iRFA + 10 µM ICT). After 24 h of incubation, cells were collected and stained with Annexin VFITC and propidium iodide (PI). Apoptotic cells were quantified by flow cytometry (CytoFLEX LX, Beckman Coulter, USA), and data were analyzed using FlowJo software (Beckman Coulter, USA).

2.7. Morphology Analysis by Immunofluorescence

For the analysis of actin polymerization and polarized morphology, neutrophils were seeded in a 24‐well plate with 14 mm coverslip (Biosharp). The coverslips were pre‐coated with 1 mg/mL polylysine in PBS for 1 h at 37°C. Neutrophils were added dropwise to the bottom of the well plate and subsequently flattened into a coverslip, then co‐incubated with different TCM (Ctl, iRFA, iRFA + 5 µM ICT, iRFA + 10 µM ICT) for 24 h at 37°C. Then, the cells were washed twice with PBS, fixed in 4% PFA for 15 min. Washed cells were next permeabilized in 3% BSA‐containing PBS for 15 min and stained with phalloidin‐FITC (RM02836, ABclonal) and DAPI (Solarbio). The green fluorescence signal was captured by CLSM at an excitation wavelength of 488 nm.

2.8. Adhesion of TANs on Tumor Cells

To evaluate the adhesion of TANs on Hepa1‐6 cells, we labeled neutrophils with DiO dye and Hepa1‐6 cells with DiD dye for 20 min, and then cocultured them at a ratio of 10:1 for 2 h at 37°C with different groups (Ctl, iRFA, iRFA + 5 µM ICT, iRFA + 10 µM ICT). After that, unbound TANs were washed away with PBS three times and fixed with 4% PFA for 15 min. The nuclei were stained with DAPI and then the fluorescence signals of DiO and DiD were observed under confocal microscopy.

2.9. Preparation of HP@ICT

The poloxamer hydrogels were prepared by the simple cold method [26]. For example, to prepare an ICT‐based solution, 10 mg ICT was dissolved in 1 mL DMSO. The ICT DMSO solution was then added to the 1 mL 10% HP‐β‐CD solution at room temperature. P407 (18%, 20%, 22%, 25%, w/v) was slowly dissolved in ICT/HP‐β‐CD solution and the solution was left in a refrigerator overnight with a vortex mixer to ensure complete dissolution. The HP@ICT hydrogel was kept at 4°C minimum for 24 h before use. The HP@BMS and HP@ICT/BMS hydrogel was obtained by the same protocol.

2.10. Characterization Analysis of HP@ICT

After preparation, the gelation time of the hydrogels was determined using the vial‐tilting method [27]. Specifically, 10% (W/V) HP‐β‐CD with different P407 concentrations was placed in 2 mL vials, and the time until no flow was observed upon inversion was recorded as the gelation time. Each measurement was repeated three times. Dynamic light scattering (DLS) was performed on a Malvern Zetasizer (Nano ZS, Malvern Ltd., UK) at 20°C to measure size, polydispersity index (PDI), and Zaverage of HP solutions (2%, 4%, 6% w/v). The injectability of HP was exhibited using a digital camera. Put the hydrogel into a freeze dryer for dehydration and lyophilization. The surface morphology and scaffold structure of the freeze‐dried sample were observed using a scanning electron microscope (Phenom pure). Rheological measurements were conducted on a straincontrolled rheometer (Waters DHR2, USA) at 37°C using 1 mL of hydrogel between parallel plates (40 mm diameter, 1 mm gap). Kinematic viscosity was assessed over 0.1–100 Hz at constant strain (1%). One‐dimensional ¹H NMR spectra were recorded on a Bruker Avance NEO (400 MHz) in D₂O (spectral width 8000 Hz, acquisition time 4.00 s, relaxation delay 1 s). The residual solvent (HOD) signal at 6.2 ppm served as internal reference.

2.11. Release of ICT In Vitro

One milliliter of HP hydrogel containing 1 mg ICT was incubated at 37°C to form a gel in a Falcon tube. After gelation, 500 µL of prewarmed PBS (37°C) was gently layered onto the surface of the gel. At 0, 2, 4, 6, 8, 24, 48, 72, 120 and 144 h, 10 µL of sample was withdrawn and replaced with 10 µL PBS. ICT concentration was quantified by UV–vis spectrophotometry at 270 nm using a standard curve prepared from known ICT concentrations.

2.12. Release of ICT In Vivo

To evaluate sustained release and biodegradability, indocyanine green (ICG, 5 mg/mL) was used as a fluorescent tracer. Mice were subcutaneously injected with HP@ICG or free ICG, and fluorescence was monitored over time using an in vivo imaging system (IVIS).

2.13. Photoacoustic (PA) Imaging

PA imaging was performed on a Vevo LAZR‐X system (VisualSonics, CA). For in vivo studies, PA signals at tumor sites (n  =  3) were recorded three days after intratumoral injection of HP@ICG in control and postablation mice. Mice were anesthetized with 4% isoflurane, positioned on a heated stage, and 3D PA signal reconstruction was performed using Vevo LAB 3.1.1 software.

2.14. Cellular Drug Uptake

Neutrophils (4×10⁴cells/well) were seeded in 24‐well plates and incubated with FITC‐labeled HP@ICT for 2, 4 or 6 h, respectively. Cellular uptake was evaluated by confocal laser scanning microscopy (CLSM) and flow cytometry.

2.15. Mouse Tumor Models In Vivo

All animal procedures were approved by the Laboratory Animal Center of the Fifth Affiliated Hospital of Sun Yatsen University. All animal experiments were performed using female BALB/c mice (aged >8 weeks, body weight 18–22 g). For subcutaneous tumors, H22 cells were injected into the lower flank. When tumors reached 6–10 mm in diameter, incomplete RFA (iRFA) was performed under anesthesia using a 5 W, 20 s ablation. For the liver metastasis model, H22‐Luc cells (10⁶) were injected into the spleen, allowing cell migration to the liver via the splenic and portal veins. A bilateral subcutaneous model was established by implanting a second tumor on the opposite side three days after the first inoculation.

To explore the effect of antitumor growth with drug‐carrying hydrogel in vivo, mice were first randomized into 4 groups (n = 5): Control (Treated with iRFA + PBS), HP (Treated with iRFA+ HP, ICT (Treated with iRFA+ ICT), and HP@ICT (Treated with iRFA+ HP@ICT). Subsequently, in order to stay close to the clinic as well as to determine whether ICT further enhances its antitumor effects when combined with PD‐L1 inhibitor, we further set up the following subgroups: Control (Treated with iRFA + PBS), ICT (Treated with iRFA+ ICT gavage), HP@BMS (Treated with iRFA+ HP@BMS), HP@ICT (Treated with iRFA+ HP@ICT) and HP@ICT/BMS (Treated with iRFA+ HP@ICT/BMS) (n = 6). The PBS, HP hydrogel without drugs, ICT, HP@ICT, HP@BMS, or HP@ICT/BMS was injected into the residual cavity after iRFA treatment. The total volume of drug injected for each mouse was 50 µL, and the amount of ICT and BMS was injected at 50 mg/kg and 4 mg/kg, respectively. Tumor volume and weight were recorded with the following formula: volume = (width² × length)/2. In addition, tissues were isolated from tumor‐bearing mice for flow cytometry analysis on the eighth day after treatment. To determine the efficacy of inhibiting liver metastases and distant tumors, the mice were observed by IVIS after treatment. Twenty‐one days after observation, the mice were sacrificed and the main organs (heart, lung, liver, spleen, and kidney) were resected for pathological examination, HE staining, and blood was collected for biochemical analysis.

2.16. Flow Cytometric Analysis

Antibodies were purchased from BioLegend or BD. Tumors and spleens were dissociated, digested with Type IV collagenase, filtered through 40 µm nylon mesh, and fixed using a Fixation/Permeabilization Kit (BD). Subsequently, the cells were stained with the following specified antibodies according to the specification: PerCP anti‐mouse CD45 (BioLegend, 103130), AF700 anti‐mouse/human CD11b (BD, 557960), FITC anti‐mouse Ly6G (BioLegend, 127606), PE anti‐mouse CD86 (BioLegend, 105007), BV421 anti‐mouse CD206 (BioLegend, 141717), BV421 anti‐mouse Ly6C (BD, 562727), APC anti‐mouse CD3 (BioLegend, 100236), FITC anti‐mouse CD4 (BioLegend, 100406), PE/Cyanine7 anti‐mouse CD8a (BioLegend, 100722), BV605 anti‐mouse CD25 (BioLegend, 102035), PE anti‐mouse Foxp3 (BioLegend, 320008), PE anti‐human/mouse Granzyme B Recombinant (BioLegend, 372208), APC/Cyanine7 anti‐mouse IFN‐γ (BioLegend, 505850). Flow cytometric data were acquired on cytoFLEX LX and analyzed using FlowJo software for quantitative analysis.

2.17. Mass Spectrometry Analysis of Neutrophils Post‐iRFA

The neutrophils were sorted from the ablation marginal zone to obtain highly purified CD45⁺CD11b⁺Ly6G⁺ cells with more than 99% purity via FCM and then subjected to timsTOF Pro mass spectrometry. The timsTOF Pro was operated in data independent parallel accumulation serial fragmentation (dia‐PASEF) mode. Full MS scan range was 300–1500 m/z; MS/MS scans (20 PASEF per cycle) covered 400–850 m/z with a 7 m/z isolation window. Protein identification was performed by Jingjie PTM Biolabs (Hangzhou, China).

2.18. Transcriptome Sequencing and Metabolomics Analysis of TANs

TANs harvested after 24 h culture with TCM were lysed with TRIzol for RNA extraction. cDNA libraries were sequenced on an Illumina platform (IGE Biotechnology, Guangzhou, China). Differentially expressed genes (DEGs) were defined by FDR < 0.05 and |log₂FC| > 1. For metabolomics, cell samples were analyzed by UHPLC (1290 Infinity LC, Agilent) at the same facility.

2.19. Western Blot

Cells were lysed, and proteins were separated by 10% SDS‐PAGE, then transferred to PVDF membranes. Membranes were blocked and incubated overnight at 4°C with primary antibodies against JAK2 (CST, 3230S), p‐JAK2 (CST, 3771S), STAT3 (CST, 30835S), and p‐STAT3 (CST, 9145S). After washing, HRPconjugated secondary antibodies were applied, and signals were detected by ECL assays.

2.20. Immunocytochemistry and Immunofluorescence Staining

Harvested neutrophils co‐cultured with TCM were stained with p‐STAT3 (CST, 9145S). Prepared tumor tissue sections were stained with TUNEL (Servicebio, G1502). Primary antibodies were used at 1:100, secondary antibodies at 1:500. Images were acquired by fluorescence or confocal microscopy.

2.21. Tissue Immunohistochemistry (IHC)

Tumor tissues were fixed in 4% PFA and subsequently embedded in paraffin, and cut into 3.5 µm‐thick sections for histopathological analysis. After antigen retrieval, sections were incubated overnight at 4°C with anti‐Ly6G (1:100, sc53515, Santa Cruz) or anti‐CD8 (1:400, GB15068, Servicebio), followed by polymer‐HRP secondary antibody (MaxVision) for 2 h at room temperature. Slides were scanned with a 3D HISTECH Panoramic scanner and analyzed with CaseViewer software (PerkinElmer).

2.22. Statistical Analysis

Each in vitro and in vivo experiment was performed at least three times independently. All quantitative data are presented as mean ± SD. All statistical analyses were performed using GraphPad Prism 7.0. For statistical comparisons, twotailed Student's ttest was used for two groups with normally distributed data, and the Mann–Whitney U test for nonnormally distributed data. For multigroup comparisons, oneway ANOVA with Tukey's posthoc test was applied for normally distributed data, and the Kruskal–Wallis test for nonnormally distributed data. A P‐value < 0.05 was considered statistically significant.

3. Results

3.1. IRFA Induced the Infiltration of Immunosuppressive TANs and Promoted the Growth of Residual HCC

The immunosuppressive mechanism of residual HCC induced by RFA remains unclear. To elucidate the role of TANs in this process, we established tumor‐bearing mice inoculated with H22 cells and performed iRFA treatment (Figure 1A). It was observed that TANs rapidly infiltrated the residual tumor border on the first day after iRFA, and the count of TANs significantly increased by the third and eighth day through immunohistochemistry (IHC) staining (Figure 1B). The gating strategy for key Flow cytometry (FCM) analysis is detailed in Figure S1. The FCM results also revealed the same changes in residual HCC with time‐dependent escalation after iRFA (Figure S2A). Besides, we further evaluated the dynamic changes in the phenotype of TANs at different time points after iRFA by FCM. The results showed that the proportion of N1 TANs with tumor inhibitory effects remarkably decreased at the first, third and eighth day after iRFA (Figure 1C). In contrast, the percentages of N2 TANs with tumor‐promoting effects in residual HCC increased at the first day after iRFA, with the most significant rise observed on the third and eighth day (Figure 1C). Accordingly, a reduction of the N1/N2 TANs ratio was observed in residual HCC at the first, third, and eighth days after iRFA (Figure S2B). Moreover, the percentages of PMN‐MDSCs, a kind of subtype in TAN populations with potent immunosuppressive characteristics, markedly increased in residual HCC in a time‐dependent manner after iRFA (Figure 1C). It is notably observed that splenic neutrophil proportions increased after iRFA (Figure S3A), with PMN‐MDSCs being the predominant subtype (Figure S3B), indicating a potential immunosuppressive effect on systemic immunity induced by iRFA. To evaluate the impact of TANs on residual HCC progression, we employed a co‐injection model in vivo as depicted in Figure 1D. The results indicated that the co‐injection of H22 cells with neutrophils which co‐cultured with heated (iRFA) tumor‐conditioned medium (TCM) significantly accelerated tumor growth, compared with inoculation of H22 cells alone (Figure 1E,F). Overall, these findings indicated that iRFA induced the infiltration of immunosuppressive TANs and promoted the growth of residual HCC.

FIGURE 1.

iRFA induced the infiltration of immunosuppressive TANs which promoted the growth of residual HCC in vivo. (A) Schematic diagram of the in vivo radiofrequency ablation model. (B) Representative IHC images showing the expression of Ly6G in mice tumor on the different day after iRFA (n = 3). Dotted line = A dense brown in residual tumor border, denoting staining of Ly6G neutrophils marker. N = Ablation of coagulated necrotic zone. T = Tumor zone. (scale bar: 100 µm). (C) Typical FCM and statistical plots illustrating N1 TANs (Ly6G⁺CD86⁺), N2 TANs (Ly6G⁺CD206⁺), and PMN‐MDSCs (Ly6G^highLy6C^low) within TME (n = 3). (D) Scheme illustration of the subcutaneous seeding of mice with TANs mixed with H22 cells. TANs derived from bone marrow after co‐incubation with tumor conditioned medium (TCM) from Ctl and iRFA, respectively. (E) Curves of subcutaneous tumor volumes in each group after co‐injection (n = 3). (F) Subcutaneous tumor weight in each group after co‐injection (n = 3). (Calculated by One‐way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

3.2. ICT Reversed Immunosuppressive TANs Induced by iRFA Through Inhibiting JAK2/STAT3 Signaling Pathway

To elucidate the mechanism underlying immunosuppressive polarization of TANs after iRFA, we sorted TANs from the tumor tissue by FCM and then conducted protein mass spectrometry (Figure 2A). Compared to the control group, we identified 415 upregulated and 1090 downregulated differentially expressed proteins in TANs after iRFA using a volcano plot (Figure 2B). The Kyoto Encyclopedia of Genes and Genomes (KEGG) of these differentially expressed proteins revealed significant enrichment in the “JAK/STAT signaling pathway” (Figure 2C). The Gene Set Enrichment Analysis (GSEA) further indicated iRFA‐induced TANs are significantly associated with activation of the JAK/STAT pathway (Figure 2D). Moreover, we found a significant elevation of IL‐6 in residual tumor by ELISA (Figure S4), which is an upstream cytokine of the JAK2/STAT3 pathway. It's reported that activation of the JAK2/STAT3 pathway polarizes TANs toward an immunosuppressive subtype [28]. Thus, inhibiting the pathway holds promise as a potential target for reversing immunosuppressive TANs in residual HCC after iRFA. Interestingly, ICT could effectively inhibit this pathway along with reducing MDSCs infiltration [29]. To explore the effects of ICT on the phenotype of TANs induced by iRFA, we co‐cultured bone marrow‐derived neutrophils with TCM of H22 cells treated with sublethal heat exposure (iRFA‐TCM) under different treatments in vitro. Our results revealed that iRFA‐TCM conspicuously enhanced the chemotaxis of TANs versus control, while the effect was dose‐dependently suppressed by pretreatment with ICT (Figure 2E). Meanwhile, in the iRFA‐TCM group, the apoptosis of TANs was significantly reduced and their survival increased, while the 10 µM ICT treatment further significantly reversed these effects (Figure S5). Morphologically, TANs co‐cultured in iRFA‐TCM exhibited an elongated spindle shape compared to those in the control group, which could be a signal promoting the pro‐tumoral phenotypic polarization of TANs. As expected, this effect was significantly reversed by ICT treatment in a dose‐dependent manner, which maintained a rounded shape (Figure 2F). Moreover, FCM further supported that iRFA‐TCM induced polarization of TANs toward the tumor‐promoting phenotype, with a significant increase in N2 TANs and PMN‐MDSCs and a decrease in N1 TANs compared with control. Remarkably, ICT could reverse the above effect of iRFA‐TCM in a dose‐dependent manner (Figure 2G). Finally, we analyzed the phosphorylation of JAK2 and STAT3 in TANs. The results verified that the levels of p‐JAK2 and p‐STAT3 were upregulated in TANs induced by iRFA and can be effectively reversed by ICT (Figure 2H–J). Collectively, these findings demonstrated that ICT favorably reversed iRFA‐induced immunosuppressive TANs by inhibiting the JAK2/STAT3 pathway.

FIGURE 2.

ICT reversed immunosuppressive TANs induced by iRFA through inhibiting JAK2/STAT3 signaling pathway in vitro. (A) Schematic diagram of sorting TANs and performing proteomics after iRFA of tumor tissue. (B) Volcano plot of sorted TANs from tumor tissue after proteomics of indicated groups (n = 3). (C) KEGG pathway enrichment of differential protein involved in TANs after proteomics of indicated groups (n = 3). (D) GSEA analysis showing JAK/STAT pathways involved in TANs after proteomics of indicated groups (n = 3). (E) Typical diagram and statistical plots of transwell assay for TANs stimulated with TCM (n = 3) (scale bar: 50 µm). (F) Fluorescence images and statistical plots showing the morphology of TANs stimulated with TCM, and stained for F‐actin with phalloidin‐FITC (green) and nuclei with DAPI (blue) (n = 3) (scale bar: 20 µm). (G) Typical FCM and statistical plots illustrating N1 TANs (Ly6G⁺CD86⁺), N2 TANs (Ly6G⁺CD206⁺), and PMN‐MDSCs (Ly6G^highLy6C^low) (n = 3). (H) Protein expression of t‐JAK2 (total JAK2), p‐JAK2 (phosphorylated JAK2), t‐STAT3 (total STAT3) and p‐STAT3 (phosphorylated STAT3) in TANs after 24 h co‐cultured with different TCM (n = 3). (I) Representative fluorescence images and statistical plots of p‐STAT3 (red) and DAPI (blue) staining in TANs after 24 h with different TCM (n = 3) (scale bar: 20 µm). (J) Schematic diagram illustrating the inhibition of the JAK2/STAT3 signaling pathway by ICT to reprogram TAN into the N1 phenotype. (Calculated by One‐way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

3.3. Preparation and Characterization of HP@ICT Hydrogel

To improve ICT's water solubility and prolong its retention time in residual tumor post‐iRFA, we designed a hydrogel drug delivery system. HP‐β‐CD was chosen as the co‐solvent and P407 as the carrier to synthesize HP@ICT by a simple physical mixing method (Figure 3A). We measured the gelation time of different concentrations of P407 hydrogel under 10% HP‐β‐CD conditions at physiological temperature (37°C). The results indicated that the gelation time of HP significantly decreased at 37°C as the concentration of P407 increased (124 s for 18%, 90 s for 20%, 56 s for 22%, and 26 s for 25%), and the addition of ICT did not affect gelation time (Figure 3B). Considering the gelation time, a 22% P407 concentration for HP@ICT should be reasonable to allow sufficient time for drug injection. Due to the amphiphilicity of P407 in water, it could self‐assemble into micelles [26]. As we measured by Dynamic Light Scattering, P407 micelles began to form from 2% to 6% concentration and the average micelle was 54 nm with a PDI of 0.49 at 6% concentration (Figure 3C,D). These micelles were small at room temperature (RT). When temperature rises and P407's concentration increases, the micelles spontaneously increase in size and dramatically aggregate to form a micellar network, which eventually forms a thermo‐responsive hydrogel. As shown in Figure 3E, HP@ICT appeared as a lightly yellow micellar solution at RT and underwent sol‐gel transition when the temperature rose to 37°C. Moreover, we detected the injectability and found the HP@ICT could be injected continuously into PBS solution through a 24G syringe needle at 37°C (Figure 3F). The scanning electron microscope (SEM) indicated the abundant porous microstructure of the freeze‐dried HP@ICT hydrogel (Figure 3G). Then, we observed the sol‐gel transition and viscoelastic curves of HP@ICT through dynamic rheological measurements in the temperature range of 15°C–40°C. The results showed that HP@ICT underwent temperature‐responsive gelation, characterized by demonstrating low storage modulus (G′) and loss modulus (G″) at a low temperature and G′ and G″ enhanced notably as the temperature rose to 27°C with an intersection between G′ and G″, which means gelation. And the viscosity η varied with the temperature, revealing the temperature sensitivity of the gel (Figure 3H). The strain sweep of HP@ICT demonstrated both G′ and G′′ remained constant with G' higher than G'' at 37°C, indicating that the hydrogel can stably maintain its gel state in the body (Figure 3I). Meanwhile, the viscosity of HP@ICT decreased when the shear rate was increased from 0.1 to 100 s⁻¹, which indicated its shear‐thinning and injectability (Figure 3J). In conclusion, we successfully developed an injectable thermoresponsive HP@ICT delivery system.

FIGURE 3.

Characterization of HP@ICT. (A) A schematic representation of the thermosensitive micelles–hydrogel hybrid system based on P407 and HP‐β‐CD. (B) Time to gelation of HP and HP@ICT at 37°C (10 wt.% HP‐β‐CD) (n = 3). (C) The PDI and (D) average size (z‐ave) of HP micelle (n = 3). (E) Photographs of HP hydrogel formation. (F) The injectability and gelation/micelle of HP@ICT to PBS at 37°C. (G) SEM images of HP@ICT hydrogel (scale bar: left: 20 µm; right: 5 µm). (H) Variation of viscoelastic parameters (η*, G′, G″) of HP@ICT at different temperatures. (I) G′ and G′′ of the HP@ICT in the strain amplitude sweep. (J) Viscosity with increasing shear rates (0.1–100 s−1) showing shear‐thinning feature of the HP@ICT.

3.4. Cellular Uptake, Sustained‐Release Capacity, and Distribution of HP@ICT

In order to explore the uptake of the drug released from the hydrogel system by neutrophils in vitro, we synthesized HP@FITC. The immunofluorescence assay showed that most of neutrophils co‐cultured with the hydrogel had efficiently taken up the FITC after 6 h (Figure 4A). These findings were also supported by the FCM results (Figure 4B). Subsequently, we performed a drug release assay in vitro, which exhibited a cumulative release of ICT from the gel and attained 90% by day 6 (Figure 4C,D). Furthermore, one‐dimensional 1H NMR indicated that, in comparison to HP‐β‐CD@ICT (b), the typical signal peaks of ICT (2.72 ppm) were attenuated in HP@ICT (d) (Figure 4E). This phenomenon may arise from the competition between P407 and ICT for the CD cavity when P407 is introduced into the aqueous solution of HP‐β‐CD@ICT, ultimately displacing ICT [26, 30]. Additionally, the in vivo drug release behavior of this hydrogel system was observed through local intratumoral injection. In vivo imaging system (IVIS) demonstrated that HP@ICG remained at the subcutaneous tumor site for over 8 days, while the ICG group decreased rapidly within 2 days (Figure 4F). Next, the degradation properties of the hydrogel in vivo were evaluated. The HP@ICT injected subcutaneously into the mice was also largely degraded within 8 days in vivo and no significant residue was observed at day 14 (Figure 4G), which indicated an excellent biodegradability of the hydrogel. Considering that the ablated zone undergoes coagulative necrosis and volumetric reduction compared to solid tumors with sham ablation, we hypothesize that this process effectively alleviates intra‐tumoral interstitial pressure. To validate this, we assessed hydrogel distribution dynamics in iRFA‐treated models using photoacoustic imaging. Consistent with our hypothesis, the iRFA group exhibited localized, high‐intensity signals confined to the ablation zone, in contrast to the control group where hydrogel demonstrated significant extravasation (Figure 4H). This observation showed that iRFA enhances the retention capacity of hydrogels within the residual tumor area. Collectively, these findings demonstrated that HP@ICT is efficiently taken up by neutrophils and exhibits good sustained‐release capacity along with favorable distribution within the ablated zone.

FIGURE 4.

Cellular uptake, sustained‐release capacity, and distribution of HP@ICT. (A) The cellular uptake and fluorescence image of neutrophils incubated with HP@ICT for different periods of time (n = 3) (scale bar: 20 µm). (B) Assessment of the ratio and statistical plots of neutrophils taken in by FITC via FCM (n = 3). (C) ICT release from HP@ICT. ICT loading amount: 1 mg/ml. (D) Degradation of ICT loaded HP@ICT in C. (E) ¹H NMR spectra data of HP‐β‐CD (a), the mixture HP‐β‐CD: ICT in 10:1 ratio (b), P407 (c), and the mixture P407: HP‐β‐CD: ICT in 22:10:1 (d) with D₂O as deuterated solvent. (F) Representative IVIS images and statistical plots showing the metabolism of ICG and HP@ICG in BALB/C mice (n = 3). (G) The degradation ability of the HP@ICT system to form a hydrogel after subcutaneous injection in vivo (scale bar: 2 mm). (H) Representative photoacoustic images and statistical plots showing the HP@ICG aggregation in Ctl and iRFA group within BALB/C mice subcutaneous tumors (n = 3) (scale bar: 2 mm). (Calculated by Student's t‐test or One‐way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

3.5. HP@ICT Inhibited Residual HCC Progression Through Reversing Immunosuppressive TANs Induced by iRFA

Motivated by the in vitro results, we further explored the efficiency of HP@ICT in reversing TANs polarization and activating the immune response on residual HCC in vivo. An iRFA subcutaneous tumor model was established and randomly divided into 4 treatment groups, including Ctl, HP, ICT, and HP@ICT group (Figure 5A). After iRFA in the subcutaneous tumor, a single gel injection was administered into the postoperative residual cavity through the ablation needle. As depicted in Figure 5B, differing from Ctl and HP treatment, ICT and HP@ICT significantly inhibited the growth of tumors, and HP@ICT demonstrated superior efficacy to ICT with a remarkable tumor inhibition rate of 90%, and 20% of these mice exhibited complete regression of residual tumors (Figure 5C,D, Figure S6A). Next, we utilized FCM to analyze the immune cells in residual tumors after different treatment. As expected, compared to Ctl and HP groups, the ICT and HP@ICT groups increased the percentages of N1 TANs, decreased the percentages of N2 TANs and PMN‐MDSCs, and this effect was more pronounced in the HP@ICT group, indicating that HP@ICT effectively promoted the polarization of TANs toward the immune activation phenotype (Figure 5E). Furthermore, the proportion of CD8⁺ T cells after ICT and HP@ICT treatments was higher than Ctl and HP groups, with a stronger effect of HP@ICT (Figure 5E, Figure S6B). In addition, we evaluated the influence of different treatment on tumor apoptosis. H&E and TUNEL staining showed that the residual tumor cells in the ICT group and HP@ICT group had extensive necrosis and apoptosis compared to the Ctl and HP groups. These effects were more pronounced in the HP@ICT group (Figure S6C). The above findings suggested that HP@ICT showed the ability to enhance the efficacy of ICT and suppress residual HCC growth after iRFA by reversing immunosuppressive TANs and promoting the infiltration of cytotoxic CD8⁺ T cells.

FIGURE 5.

HP@ICT inhibited residual HCC progression through reversing immunosuppressive TANs induced by iRFA in vivo. (A) Schematic diagram of the therapeutic mouse model in subcutaneous tumor. Mice were treated with I) Ctl, II) HP, III) ICT, IV) HP@ICT (n = 5). (B) Growth curve of the HCC residual tumor by caliper after different treatments (n = 5). (C) The residual tumor inhibition rate and (D) tumor weight of different treatments (n = 5). (E) Typical FCM and statistical plots illustrating N1 TANs (Ly6G⁺CD86⁺), N2 TANs (Ly6G⁺CD206⁺), PMN‐MDSCs (Ly6G^highLy6C^low), and CD8⁺ T cells (CD3⁺CD8⁺) (n = 4). (Calculated by One‐way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

3.6. Hydrogel Co‐Delivering ICT and PD‐L1 Inhibitor Synergistically Ameliorated Residual HCC After iRFA

The above results indicated that HP@ICT effectively reduced the immunosuppressive TANs and enhanced the infiltration of cytotoxic CD8⁺ T cells, which may potentiate immunotherapy efficacy. We hypothesized that the combination of ICT and PD‐L1/PD‐1 immune checkpoint inhibitors would maximize the anti‐tumor effect. To implement this strategy, we encapsulated the small‐molecule PD‐L1 inhibitor BMS202 into HP@ICT, developing a co‐delivery system designated HP@ICT/BMS. An iRFA mouse model of HCC was constructed and randomly divided into 5 groups including Ctl, ICT (gavage), HP@ICT, HP@BMS, and HP@ICT/BMS group (Figure 6A). The gavage group mimics clinical administration, while the hydrogel groups are administered by local injection as previously described, ensuring that the total dose is consistent for each mouse. Compared to Ctl group, treatment with ICT (gavage) showed relatively minor levels of effectiveness, while notably stronger inhibition of residual HCC progression was observed after HP@ICT, HP@BMS, and HP@ICT/BMS treatment (Figure 6B). Encouragingly, the therapeutic effect of HP@ICT/BMS was superior to that of HP@ICT or HP@BMS group, achieving complete remission with 5 out of 6 tumors after treatment (Figure S7A,B). Subsequently, immune responses were tracked by FCM after different treatments. In contrast to the other 4 groups, our results indicated that mice treated with HP@ICT/BMS had a significant increase in the percentages of N1 TANs, while N2 TANs and PMN‐MDSCs markedly decreased (Figure S7C–E). Simultaneously, HP@ICT/BMS effectively enhanced the ratio of CD8⁺ T cells, whereas correspondingly reduced the infiltration of regulatory T cells (Treg cells) (Figure 6C, Figure S7F). These CD8⁺ T cells exhibited a higher expression of Granzyme B (GrzmB) and interferon‐γ (IFN‐γ) than those treated with HP@ICT or HP@BMS alone (Figure 6C). Beyond that, H&E and TUNEL staining exhibited significant pathological damage on residual tumor in HP@ICT/BMS treatment (Figure 6D). No significant body weight loss was observed in any treatment group (Figure S8A). And we further verified the biosafety in vivo, the results revealed no significant change in the heart, spleen, lungs, kidneys, liver, and serum parameters on day 21 after different treatments (Figure S8B‐C, Figure S9). These inspiring results suggest that HP@ICT significantly complements anti‐PD‐L1 treatment, leading to regression of residual HCC after iRFA with good biosafety.

FIGURE 6.

HP@ICT/BMS synergistically ameliorated residual tumors after iRFA. (A) Schematic diagram of the therapeutic mouse model in subcutaneous tumor. Mice were treated with I) Ctl, II) ICT (gavage), III) HP@BMS, IV) HP@ICT, IV) HP@ICT/BMS (n = 6). (B) Growth curve of the HCC residual tumor by caliper after different treatments (n = 6). (C) Typical FCM and statistical plots illustrating CD8⁺ T cells (CD3⁺CD8⁺), GrzmB⁺CD8⁺ T Cells (CD8⁺GrzmB⁺) and IFN‐γ⁺CD8⁺ T Cells (CD8⁺IFN‐γ⁺) (n = 5). (D) Representative IHC images showing the expression of CD8 in mice tumor, and representative image of TUNEL fluorescent staining and H&E of residual tumor harvested from mice in different groups (n = 3) (scale bar: 100 µm). (Calculated by One‐way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

3.7. Hydrogel Co‐Delivering ICT and PD‐L1 Inhibitor Synergistically Inhibited Metastasis of Residual HCC After iRFA

The metastasis of residual HCC after iRFA presents a significant clinical challenge, warranting further investigation and intervention [31]. To further explore the inhibitory ability of HP@ICT/BMS for residual tumor metastasis, we established an iRFA subcutaneous tumor model and the liver metastasis model. Different treatments (Ctl, ICT gavage, HP@ICT, HP@BMS, and HP@ICT/BMS) were performed on iRFA‐treated mice and the growth of liver metastases was continuously captured by bioluminescent imaging, as depicted in Figure 7A. Monitored by IVIS imaging, our results showed that treatment with local HP@ICT or HP@BMS monotherapy exhibited a certain inhibitory effect on liver metastases after iRFA compared to the Ctl and ICT (gavage) group (Figure 7B,C). However, local HP@ICT/BMS treatment demonstrated the best efficacy in suppressing the growth of liver metastases, with 75% (3/4) of mice achieving complete remission (Figure 7D). Similar results were also confirmed in H&E staining of isolated liver tissue and the results showed more metastatic lesions in the Ctl group. HP@ICT or HP@BMS group had fewer metastases than the Ctl and ICT (gavage) group, while the HP@ICT/BMS showed no metastasis (Figure 7E). Moreover, not only local drug stimulation triggers anti‐tumor immune response in the tumor, but also generates systemic effects based on the stimulated anti‐tumor immune cells in the spleen [32]. Thus, given that the spleen is closely related to the accumulation and function of PMN‐MDSCs, we extracted spleens for FCM analysis after different treatment. As anticipated, HP@ICT/BMS showed a marked reduction in PMN‐MDSCs compared to the other groups (Figure 7F). Interestingly, the proportion of CD8⁺ T cells, CD4⁺ T cells and Tregs in the spleen did not change (Figure 7G and Figure S10A,B), while the immune stimulatory molecules from CD8⁺ T cells (including GrzmB and IFN‐γ) remarkably increased in the HP@ICT/BMS group (Figure 7H,I), suggesting the activation of systemic immunity. Inspired by the remarkable therapeutic effect of HP@ICT/BMS in treating liver metastasis after iRFA, we further investigated its inhibitory effect in distant tumors (Figure S11A). Consistent with the above results, HP@ICT/BMS exhibited the highest tumor inhibition rate, with 2 out of 5 mice achieving complete remission compared to the other groups (Figure S11B–F). Importantly, HP@ICT/BMS showed a significantly increased ratio of CD8⁺ T cells in distant tumor tissues (Figure S12A). Meanwhile, H&E and TUNEL staining (Figure S12B) revealed significant pathological damage in distant tumor tissues in HP@ICT/BMS group. Altogether, these results indicated that the local HP@ICT/BMS therapy could induce systemic immunity to inhibit the progression of liver metastasis and distant tumor in HCC after iRFA.

FIGURE 7.

HP@ICT/BMS synergistically inhibited the progression of liver metastasis after iRFA. (A) Schematic diagram of the therapeutic mouse model in liver metastasis. Mice were treated with I) Ctl, II) ICT (gavage), III) HP@BMS, IV) HP@ICT, IV) HP@ICT/BMS (n = 4). (B) The representative IVIS images of liver metastases taken at different time (n = 3). (C) Statistical plots of the intensity of the bioluminescence in the IVIS images (n = 3). (D) The representative IVIS images of isolated liver metastases at the end of treatment (n = 4). (E) H&E staining of liver metastasis in different treatment groups (n = 3) (scale bar: 100 µm). Quantitative analysis of (F) PMN‐MDSCs, (G) CD8⁺ T cells (CD3⁺CD8⁺), (H) GrzmB (CD8⁺GrzmB⁺), and (I) IFN‐γ (CD8⁺IFN‐γ⁺) in spleens from different groups (n = 4). (Calculated by One‐way ANOVA. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

3.8. Transcriptomics and Metabolite Alterations of ICT on TANs After iRFA

To delve deeper into the antitumor mechanisms by which ICT induces reprogramming of TANs after iRFA, we harvested TANs from both the control group (treated with iRFA TCM) and the ICT treatment group (treated with the iRFA TCM and ICT) to conduct transcriptomic analyses. The volcano plot analysis identified 3,255 differentially expressed genes (DEGs) as up‐regulated and 2999 as down‐regulated in the experimental group (Figure 8A). KEGG pathway analyses of the DEGs indicated that apart from the JAK/STAT pathway, these genes were predominantly related to the PD‐L1 checkpoint signaling, cell adhesion pathways, and fatty acid metabolism (Figure 8B). Our results indicated that ICT treatment caused a negative regulation of the “PD‐L1 checkpoint signaling”, which means partial relief of immunosuppression (Figure 8C). FCM analysis substantiated that the ICT treatment group showed a significant downregulation of PD‐L1 expression on TANs and an increase in T‐cell effector molecules including GrzmB and IFNγ compared to the control group (Figure 8D and Figure S13). Notably, our study also revealed ICT treatment downregulated adhesion molecules in TANs, such as Itga1, Icam2, and Vcam1, indicating the process of interaction between neutrophils and tumor cells may be weakened (Figure 8E). To confirm this interaction, we established a co‐culture model of neutrophils and tumor cells in vitro and observed a significant adhesion of DiO‐labeled TANs to DiD‐labeled Hepa1‐6 cells in the iRFA group. In contrast, ICT treatment could effectively inhibit this phenomenon (Figure 8F). It is reported that the metabolic properties of TANs are associated with their functional phenotype [33]. Given that transcriptomics revealed that ICT has an inhibitory impact on fatty acid metabolism in TANs (Figure 8A), we conducted metabolomics analysis for further verification to identify alterations in fatty acid metabolites. The results indicated that this effect primarily focused on the biosynthesis of unsaturated fatty acids, as evidenced by the reduction of linoleic acid and arachidonic acid after ICT treatment (Figure 8G,H). In summary, these findings implied that ICT inhibits the PD‐L1 expression in TANs, disrupts TANs‐tumor interaction and inhibits fatty acid metabolism. This mechanism further explains how ICT effectively reverses the immunosuppressive function of TANs and synergistically enhances the efficacy of anti‐PD‐L1 therapy, thereby alleviating the growth and metastasis of residual HCC after iRFA.

FIGURE 8.

Transcriptomics and metabolite alterations of ICT treatment on TANs after iRFA. (A) Volcano plot of treated neutrophils after mRNA sequencing of indicated groups (n = 3). (B) KEGG pathway enrichment of different genes involved in TANs cultured with tumor tissue supernatant post‐iRFA with or without ICT for 24 h after mRNA sequencing of indicated groups (n = 3). (C) Fold change of gene expressing in PD‐L1 expression and PD‐1 checkpoint pathway (n = 3). (D) Statistical plots of FCM illustrating PD‐L1 TANs (Ly6G⁺PD‐L1⁺), GrzmB⁺CD8⁺ T Cells (CD8⁺GrzmB⁺) and IFN‐γ⁺CD8⁺ T Cells (CD8⁺IFN‐γ⁺) in residual tumor from different groups (n = 4). (E) Fold change of gene expressing in cell adhesion signaling pathway. (F) Representative fluorescence images of tumors‐TANs clusters after coincubation of DiD‐labeled Hepa1‐6 cells and DiO‐labeled TANs in vitro (n = 3) (scale bar: 20 µm). (G) KEGG pathway enrichment of different metabolites involved in TANs after metabolism sequencing of indicated groups (n = 6). (H) Heat map of different metabolites involved in TANs on fatty acids of indicated groups (n = 6). (Calculated by Student's t‐test.) *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

4. Discussion

In this study, we found that iRFA promoted activation of the JAK2/STAT3 pathway in TANs and polarized them toward immunosuppressive phenotypes, which accelerated residual HCC progression. To address the above problem, we developed an injectable thermosensitive hydrogel platform, which forms a gel in situ once the temperature reaches the critical threshold post‐RFA, prolonging the retention time of the JAK2/STAT3 inhibitor and the PD‐L1 inhibitor within the residual HCC. Our experiments demonstrated that the delivery system resulted in significant regression of residual tumors and metastases with favorable biosafety in vivo. This localized delivery system offers an effective strategy to prevent residual HCC progression, holding enormous potential for optimizing RFA treatment and clinical translation.

Our previous research preliminarily demonstrated that recruitment of abundant TANs occurred in the residual HCC, which predominantly localized within the ablation transition zone following iRFA [14]. In this study, we further indicated that these TANs polarized toward an immunosuppressive phenotype (N2 TANs / PMN‐MDSCs), thereby facilitating the progression of residual HCC. The possible explanation for the early infiltration of TANs after iRFA is that the local thermal injury induced by ablation energies triggers rapid necrotic tumor cell death, releasing substantial endogenous damage molecules within a short time frame [34]. These molecules synergize with pre‐existing tumor‐derived chemokines to collectively promote TANs recruitment in the residual tumor microenvironment [9]. Our proteomic profiling identified JAK2/STAT3 pathway activation in TANs isolated from residual tumor, coinciding with elevated secretion of its upstream stimulator IL‐6. Activation of the JAK2/STAT3 pathway serves as a key driver for the shift of TANs toward an immunosuppressive phenotype [35]. Thus, targeted inhibition of this signaling to reduce immunosuppressive TANs represents a promising therapeutic strategy for inhibiting iRFA‐induced residual HCC. Intriguingly, prior studies suggest that ICT, a clinically approved traditional Chinese medicine, inhibits the proliferation of tumor cells via JAK/STAT pathway suppression [36]. Our research demonstrated that ICT could inhibit JAK2/STAT3 pathway in TANs, which not only reduces the proportion of immunosuppressive TANs but also downregulates PD‑L1 expression on TANs. Moreover, RNA‑seq data indicated that ICT treatment decreases the expression of adhesion molecules in TANs, which may weaken TAN‑tumor cell clustering and further enhance T cell access. Therefore, ICT may not directly chemoattract CD8⁺ T cells but rather removes the physical and chemical barriers imposed by immunosuppressive TANs, thereby indirectly promoting T cell infiltration. Collectively, this study revealed that iRFA induced the polarization of TANs immunosuppressive phenotype within residual HCC, whereas ICT effectively reversed this process by inhibiting the JAK2/STAT3 pathway, which holds potential to enhance the therapeutic efficacy of RFA for HCC.

The challenges to the clinically oral administration of ICT include conspicuous water insolubility and low bioavailability [18]. To address these problems, we designed an injectable thermosensitive and sustained TAN reprogramming hydrogel system. The drug carrier HP hydrogel mainly synergistically integrates HP‐β‐CD's superior inclusion complexes capacity with P407's thermoresponsive behavior and micellar self‐assembly characteristics. Due to the advantages of both, such as biocompatibility, non‐toxicity and ease of gelation, HP hydrogel is a reasonable choice for enhancing the efficacy of ICT and can achieve dual optimization of ICT solubility and slow‐release properties. Besides, the distinctive cyclic structure of HP‐β‐CD not only enhances the aqueous solubility of the nonpolar ICT but reinforces the mechanical strength of the hydrogel through hydrogen bonding with polyethylene oxide chains in P407 [26, 37]. This may explain the excellent characteristics of the hydrogel system, which has long‐term sustained release of ICT both in vitro and in vivo, and is effectively taken up by neutrophils. The platform was injected through the ablation needle into the ablation cavity immediately after RFA to facilitate local accumulation of ICT in residual and occult cancer nests surrounding the zone. Compared with oral administration, it reduces the drug toxicity to the normal tissues. Benefiting from the excellent clinical application prospects and sustained‐release properties of hydrogel delivery system, our results indicated the system remodeled the residual HCC immune microenvironment by reversing immunosuppressive TANs and promoting CD8⁺ T cell infiltration and activation. Maximal synergistic suppression of post‐iRFA residual tumors and metastatic foci is achieved when the hydrogel platform co‐delivers ICT with PD‐L1 inhibitors. An important finding of this study is the superior effect of localized hydrogel delivery in suppressing post‐iRFA residual HCC. Whereas elevated interstitial pressure in solid tumors triggers rapid drug efflux post‐injection [38, 39], the system exhibits markedly attenuated leakage in iRFA‐treated models. This is mechanistically linked to iRFA‐induced coagulation necrosis and tissue dehydration processes that alleviate tissue tension within the ablation zone, thereby enhancing hydrogel distribution at residual HCC sites. This strategy ingeniously converts the ablation zone into a continuous drug “reservoir”, which promotes prolonged immune modulation to eradicate occult cancer, ultimately achieving the suppression of local recurrence in HCC. Of importance, the system exhibits exceptional translational potential due to clinically approved raw materials, simple synthesis process, low cost, and favorable biosafety. Collectively, these benefits propel the hydrogel platform as a promising therapeutic candidate for residual HCC with excellent clinical translation.

Given the paucity of mechanistic data on ICT modulation of TANs, we further employed multi‐omics profiling to delineate its regulatory networks. Our analyses revealed that ICT treatment downregulated the PD‐L1 checkpoint signaling, cell adhesion pathways, and fatty acid metabolism. Previous studies discovered that activation of the JAK2/STAT3 pathway upregulates PD‐L1 expression on TANs, which then interacts with PD‐1 on CD8⁺ T cells, thereby establishing an inhibitory checkpoint that suppresses T cell activation [35, 40]. Our RNA‐seq and cellular experiments confirmed that the JAK2/STAT3 inhibitor ICT downregulates TAN PD‐L1 expression, thereby blocking the PD‐L1/PD‐1 checkpoint and consequently promoting T cell activation. Animal studies further demonstrated that ICT‐loaded hydrogel induced T cell activation. Significantly, hydrogels co‐loaded with ICT and PD‐L1 inhibitor, which simultaneously blocks the immune checkpoint between tumor cells and T cells, synergistically enhanced T cell activation, leading to superior antitumor efficacy. Additionally, studies have shown that STAT3 signaling activation promotes the expression of adhesion molecules, such as Vcam1 in leukocytes, which could facilitate the formation of pro‐metastatic tumor cell clusters [41, 42]. In this study, our results confirmed that ICT treatment downregulated adhesion molecules in TANs, including Itga1, Icam2, and Vcam1, potentially attenuating the formation of TAN‐tumor cell clusters. These results elucidate a novel mechanistic insight for the excellent anti‐metastatic efficacy of hydrogel drug delivery system against residual tumor in the iRFA mouse model. Besides, the metabolism of TANs is closely linked to their function. Our multi‐omics analyses demonstrated that ICT effectively inhibits fatty acid metabolism in TANs, primarily reducing the synthesis of unsaturated fatty acids, including linoleic acid and arachidonic acid. Prior evidence has indicated that activation of the JAK/STAT3 pathway has been shown to promote the fatty acid metabolism in breast cancer stem cells [43]. Nevertheless, the observed changes are consistent with STAT3 inhibition, although potential off‐target contributions cannot be entirely excluded.

Overall, our research proposes a promising synergistic strategy combining RFA with local hydrogel drug delivery system to prevent residual HCC progression. Further engineering with inherent imaging capabilities (e.g., ultrasound microbubbles) would achieve real‐time visualization of gel distribution during delivery. The hydrogel system opens avenues for integration with real‐time ultrasound‐guided intervention to ensure precise gel deposition and drug release within high‐risk regions of peripheral ablation zones in HCC after RFA. Considering that RFA is also used for treating breast cancer, lung cancer and kidney cancer, this approach may be further extended to these types of tumors as well.

5. Conclusions

Collectively, we found that iRFA activated the JAK2/STAT3 pathway in TANs, polarizing them toward immunosuppressive phenotypes that accelerated residual HCC progression. The thermosensitive hydrogel platform effectively transformed the ablation zone into a sustained‐release drug reservoir and reversed immunosuppressive TANs by inhibiting the JAK2/STAT3 pathway, thereby eliciting a robust immune response. Furthermore, hydrogel co‐delivering ICT and PD‐L1 inhibitor synergistically eliminated residual tumors and metastases post‐RFA. Mechanistically, ICT inhibited the PD‐L1 expression in TANs, disrupted TANs‐tumor cell interaction and suppressed fatty acid metabolism. This study presents a safe, effective and clinically translational strategy for inhibiting the residual tumor progression and enhancing RFA efficacy in HCC management.

Author Contributions

H.C.: writing –original draft, validation, project administration, methodology, investigation, data curation, software, visualization, conceptualization. S.Z.: writing – review & editing, visualization, methodology, funding acquisition, resources. D.Y.: project administration, validation, software, data curation. S.P.: methodology, investigation, formal analysis. Y.Z.: software, visualization. C.M.: software, methodology. Y.L.: supervision. Z.S.: writing – review & editing, resources, project administration, formal analysis, supervision, funding acquisition. D.L.: writing – review & editing, resources, project administration, supervision, investigation, methodology, formal analysis. Y.H.: writing – review & editing, conceptualization, supervision, project administration, investigation, methodology, funding acquisition, formal analysis.

Funding

This work was supported by the National Natural Science Foundation of China [grant number 82371975]; Guangdong Basic and Applied Basic Research Project [grant number 2023A1515010620, 2024A1515013242]; Core Talent Fund of the Fifth Affiliated Hospital of Sun Yat‐sen University [grant number 310103050302–220904094228]; Medical Science and Technology Research Foundation of Guangdong Province [grant number A2023100]; the Excellent Young Researchers Program of the Fifth Affiliated Hospital of SYSU [grant number WYYXQN‐202517]; Guangdong Provincial Medical Science and Technology Research Project [grant number 20231111195130817]; Guangdong Provincial Bureau of Traditional Chinese Medicine Research Project [grant number 202405081740049570]; Zhuhai Basic and Applied Basic Research Project [grant number ZH2420004000265].

Ethics Approval and Consent to Participate

All animal experimental procedures were performed under the context of the animal protocol approved by the Laboratory Animal Center of the Fifth Affiliated Hospital of Sun Yat‐sen University (202000139 and 2024060401), and were carried out in compliance with all relevant ethical regulations.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: adhm71386‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.