Percutaneous Delivery of Oncogel for Targeted Liver Tumor Ablation and Controlled Release of Therapeutics

Abstract

Advanced-stage liver cancers are associated with poor prognosis and have limited treatment options, often leading the patient to hospice care. Percutaneous intratumoral injection of anticancer agents has emerged as a potential alternative to systemic therapy to overcome tumor barriers, increase bioavailability, potentiate immunotherapy, and avoid systemic toxicity, which advanced-stage cancer patients cannot tolerate. Here, we developed an injectable OncoGel (OG) comprised of a nanocomposite hydrogel loaded with an ionic liquid (IL) for achieving a predictable and uniform tumor ablation and long-term slow release of anticancer agents into the ablation zone. Rigorous mechanical, physiochemical, drug release, cytotoxicity experiments, and ex vivo human tissue testing identified an injectable version of the OG with bactericidal properties against highly resistant bacteria. Intratumoral injection of OG loaded with Nivolumab (Nivo) and doxorubicin (Dox) into highly malignant tumor models in mice, rats, and rabbits demonstrated enhanced survival and tumor regression associated with robust tissue ablation and drug distribution throughout the tumor. Mass cytometry and proteomic studies in a mouse model of colorectal cancer that often metastasizes to the liver indicated an enhanced anticancer immune response following the intratumoral injection of OG. OG may augment immunotherapy and potentially improve outcomes in liver cancer patients.

Graphical Abstract

We developed an injectable Oncogel (OG) comprised of a nanocomposite hydrogel loaded with an ionic liquid, chemotherapy and a checkpoint inhibitor for achieving a predictable and uniform tumor ablation with long-term slow release of anticancer agents into the ablation zone. Intratumoral injection of OG can potentially augment immunotherapy and improve outcomes in liver cancer patients.

1. Introduction

Liver cancer is one of the most common malignancies worldwide.^[1] Cirrhosis in the setting of alcoholism, obesity, and hepatitis are the major risk factors; recent estimates suggest more than 270 million living with viral hepatitis are at risk for developing liver cancer.^[2] On presentation, most patients are in the advanced stages, which limits treatment options mostly to palliative care including chemotherapy or embolotherapy.^[3] Despite decades of research, liver cancer in advanced stages remains a death sentence with dismal 5-year survival rates; 3% if metastatic and 9% if locoregional.^[4] Today, chemotherapy and immunotherapy represent the standard of care for most cancers; however, they have been largely ineffective in liver cancer.^{[3b, 5]} Local factors have been implicated in the failure of systemic therapy including heterogeneity of the tumor microenvironment and vascularity, high interstitial pressures, as well as low pH and hypoxia within the tumor. These factors, coupled with challenges in achieving sufficient delivery of therapeutics to the tumor with limited infiltration of immune cells, in part account for the poor responses observed to systemic therapy.^[4] In patients with advanced stages of liver cancer, poor liver function further limits treatment options, often leading the patient to hospice care with median survival measured in months.

Achieving high bioavailability and uniform distribution of intravenously delivered drugs within liver tumors is a major challenge in oncology today. The dose of the drug is typically increased to encourage tumor penetration leading to predictable drug toxicity often making systemic therapies prohibitive.^[6] To bypass vascular barriers to effective drug delivery, direct intratumoral injection represents the most desirable approach to allow high local concentrations of the anti-cancer therapy avoiding systemic toxicity.^{[6a, 7]} Over 30 ongoing clinical trials are investigating the technique of intra-tumoral injection such as using needles with multiple side holes to increase the delivery of various therapeutics including oncoviruses, chemotherapy, and immunotherapy.^[7b] However, despite these significant efforts, uniform distribution and retention of directly injected drugs within the tumor bed continue to be a significant bottleneck in oncology;^{[3a, 3c]} upon injection, the therapeutic often rapidly wash away from the target site.^[7a] A recent study involving 327 image-guided intratumoral injections of anti-cancer therapies into solid tumors and 113 image-guided intra-tumoral injections of talimogene laherparepvec (TVEC) using a variety of injection methods showed that the method that maximizes drug distribution and retention is yet to be achieved and continues to be challenging.^[7a]

Here, we developed a hydrogel-based platform to circumvent the limitations of drug delivery by bringing a needle directly to the tumor under image guidance and injecting a shear-thinning biomaterial mixed with an IL strategically loaded with various anticancer and imaging agents. This gel, referred to as OG, is comprised of nanosilicate (NS) mixed with an IL (1:1 molar ratio of choline and geranic acid), chemotherapy (i.e., Dox), and an immune checkpoint inhibitor (ICI; i.e., Nivo) (Figure 1a). The OG platform creates a nanostructure of dual-charged nanodiscs complexed with IL-containing cations and anions, aiming to develop an injectable drug depot with high loading capacity and sustained-release properties. NS-based hydrogels exhibit shear-thinning properties, facilitating injection, and are biocompatible.^[8] ILs effectively solubilize and stabilize various drugs, including Dox, and can chemically ablate liver cancer tissues while enhancing drug diffusion across tissue barriers.^[9] Combining these components in optimized ratios yields an injectable formulation capable of local tumor ablation, sustained release, and diffusion of payloads (i.e. Dox and ICI). This sustained release aims to synergize with IL-induced tumor ablation and create an optimal environment for immunotherapy by concentrating ICIs at the interface of tumor cell death and inflammation, potentially enhancing cancer treatment efficacy. We show that OG can be delivered under image guidance for prolonged and sustained release of IL and ICI into the tumor bed of murine, rodent, and rabbit models of liver cancer. High-dimensional analysis of infiltrating immune cells using CyTOF and proteomic studies in ablated tumor tissues demonstrated an active anti-cancer immune response. We show that OG produces a consistent and predictable tissue ablation through the action of IL; this tumor cell death provokes an intense immunostimulatory response. With the arrival of T-cells to the margin of the ablation zone, they encounter a sustained release of immunotherapy drugs locally from the OG thereby potentially circumventing the tumor suppression of the immune system often seen with liver tumors. This novel approach has the potential to change the standard of treatment for liver cancer due to the simplicity of its use and delivery, low cost to prepare, and the flexibility to repeat the procedure; more importantly, this approach may enhance immunotherapy in liver tumors and reduce the systemic toxicity of these drugs due to local release. This approach, which may lead to immune activation, may also treat satellite and remote lesions, especially in highly malignant liver tumors, which would represent a breakthrough in oncology.

Figure 1. Characterization of the mechanical and drug delivery properties of OG formulations.

a. Schematic illustration of OG formulation and injection process from a syringe needle. b. Viscosity flow curves of 3–9% NS hydrogels displaying concentration-dependent changes. c. Graph depicting a concentration-dependent increase in G’ generated by 3%, 4.5%, 6%, and 9% NS. d. Representative fluorescence images and graph illustrating Dox diffusion from hydrogels containing 3–9% NS + 25% IL + Dox, demonstrating the impact of increasing NS concentration on Dox diffusion through agarose matrix. e,f. Representative viscosity flow curves and graph displaying G’ values generated by hydrogels containing 3% NS + 6.25–50% IL. g. Fluorescence images and graph showing Dox diffusion in hydrogels containing 3% NS + 6.25–50% IL + Dox. h,i. Plots of conductivity, and zeta potential measurements of 6.25–60% IL. j. Graph displaying zeta potential levels of NS or OG with and without Dox, or Nivo. k. Cumulative release profile of free Dox at pH 7.4 and pH 5. l. Representative image of a dialysis membrane system used for Dox release tests and plot showing cumulative release profile of Dox mixed in OG or NS at pH 7.4 and pH 5. m. FTIR spectra of NS and OG displaying a distinct vibrational peak for Si-O highlighted in blue at ~1000 cm⁻¹ wavenumber. n. Scanning electron microscopy images of lyophilized NS and OG aliquots displaying porous microstructure in NS compared to the less porous mesh-like microstructure of OG. Data in graphs represent means ± s.e.m., analyzed by one-way ANOVA with Tukey’s multiple comparison post hoc test. ns; not significant, * P< 0.05, ** P< 0.01, *** P< 0.001, and **** P< 0.0001.

2. Results and Discussion

2.1. OG preparation and characterization

We first examined the mechanical properties of a series of hydrogels containing 3–9 wt% of NS and assessed their viscosity, shear-thinning behavior, and injectability. Rheometry studies demonstrated the shear-thinning behavior of all hydrogels and concentration-dependent increase in viscosity and storage modulus (G’) at 37 °C (Figure 1b,c and Figure S1a,b). These results indicated that all NS levels tested could be candidates for the OG formulation. To test whether NS levels impacted the release of the IL and therapeutics from the OG formulation, the NS amount was varied, and the IL level was kept constant at 25%. Using a tissue-mimicking agarose matrix, the diffusivity of Dox was measured; 3% NS demonstrated the greatest level of diffusion at 4 and 24 hours while keeping IL constant at 25% (Figure 1d). Next, NS was kept constant at 3%, and the IL levels were varied; these formulations also demonstrated shear thinning properties with a higher rate of Dox diffusion at 4 and 24 hours with 25% and 50% IL (Figure 1e–g and Figure S1c,d). To better understand the role of NS in the OG formulation, the diffusivity of Dox was measured; as expected, Dox mixed with 3% NS and 25% IL demonstrated slow release, whereas control samples of free Dox showed a more rapid, burst-like release (Figure S1e). To determine the IL amount to use in the OG formulation, conductivity and the Zeta potentials were measured. Based on these results, 25% IL demonstrated ideal properties for drug delivery (Figure 1h,i);^[10] thus, 3% NS and 25% IL were selected as the components of OG. Consistent with Figure 1d and g, Figure 1j validated the observation that NS alone leads to a more negative Zeta potential value hindering diffusion. When 3% NS is mixed with 25% IL, Zeta potential becomes less negative enhancing diffusivity;^[11] this value does not substantially change when mixed with drugs such as Dox or Nivo, which is the most commonly used immunotherapy drug worldwide. Based on these findings, we compared the long-term drug diffusion profile of free Dox, OG+Dox, and NS+Dox through a membrane barrier at pH 7.4 or 5 conditions representing physiologic or acidic cancer tissue-like microenvironment. Free dox reached 100% release within 24 hours suggesting burst release behavior (Figure 1k). In contrast, NS demonstrated markedly lower Dox release at both pH conditions; however, OG mixed with Dox showed a progressive and sustained Dox release that persisted up to 56 days with a higher release rate observed under acidic conditions (Figure 1l). Based on these results, we hypothesized that the intermolecular interactions between drugs and ILs may weaken over time, enabling gradual long-term drug release. The highly negatively charged NS, as measured in Figure 1h, did not release positively charged Dox due to strong electrostatic interactions (Figure 1d,g). However, when IL was introduced to form OG, Dox was readily released, indicating that IL incorporation altered the intramolecular interactions between NS and Dox.

Following direct injection of OG into liver tumors, ablation of the tumor tissue and its microvessels may increase the risk for infection; to confirm the sterility and antibacterial properties of OG, bacterial studies showed no growth at 1 day and at 2 months incubation at 37° C (Figure S1f). Next, Fourier transform infrared (FTIR) spectroscopy analysis of lyophilized NS and OG was performed; Figure 1m shows a unique Si–O bond in both preparations. The preservation of this distinct Si-O bond in OG is important to allow tracking of its spatial distribution following injection into tissues. Additionally, scanning electron microscopy (SEM) of lyophilized NS and OG demonstrated less porous and smoother morphology in OG suggestive of enhanced molecular interactions (Figure 1n). Rheometry studies were repeated after incorporating ICG, Dox, and Nivo showing no significant change in the material viscosity, injectability, or modulus (G’) (Figure S1g–i).

To examine the effect of OG on drug stability, the fluorescent signal of Dox, or the near-infrared fluorescent (NIRF) agent ICG, which is used here as a surrogate for a drug, was measured over time. Aliquots of OG+Dox, NS+Dox, NS+ICG, and OG+ICG were loaded into a multiwell plate and incubated at 37 °C then serially imaged over eight weeks to assess fluorescence stability (Figure 2a,b). Fluorescence imaging showed that the fluorescence intensity of OG loaded with Dox and ICG remained stable for up to 56 days. Conversely, Dox and ICG fluorescence loaded in NS steadily declined, suggesting degradation (Figure 2a,b). ILs are known to stabilize small molecule drugs such as Dox through a complex interplay of chemical interactions. We speculate that among the potential interactions, Van der Waals/Coulomb forces and electrostatic interactions between the drug and the IL likely represent the potential mechanisms that contribute to the stability of the drugs within the IL solvent environment.^{[9a, 12]} To evaluate the impact of OG on a large protein-based drug such as Nivo, which is the most commonly used immune checkpoint inhibitor worldwide, dynamic light scattering (DLS) of Nivo mixed with IL was performed. Figure S1j shows that the hydrodynamic size of native Nivo is identical to Nivo + IL. In contrast, DLS of Nivo alone or Nivo+IL incubated with HepG2 subjected to thermal treatment induced by microwave (MWA) showed a larger hydrodynamic size with a variable radius (Figure S1j). Furthermore, protein fractionation of non-denatured Nivo alone or Nivo containing OG (IL+Nivo) extract exhibited a single band consistent with intact IgG4 molecule at 150 kDa, whereas MWA treatment of Nivo had no visible band on western blotting (Figure S1k). Similar findings were obtained when HepG2 cells were incubated with Nivo alone, or IL+Nivo showed preserved IgG4 on western blotting, whereas the Nivo signal could not be detected when treated with MWA (Figure S1l,m). These findings imply that OG may also provide stability to larger molecular weight drugs such as Nivo; a functional Nivo is critical for achieving liver tumor immunotherapy. The loss of Nivo signal on Western blots when treated with MWA suggests that the thermal energy likely denatured the protein drug, potentially rendering it inactive.

Figure 2. In vitro characterization of OG on drug stability, and cancer cell viability, adhesion, and migration.

a,b. Representative fluorescence images of wells containing OG or NS aliquots mixed with Dox or ICG respectively, alongside plots demonstrating the stability of Dox and ICG fluorescence in OG for up to 56 days, in contrast to diminished fluorescence intensity in NS. c. A plot of PD-1/PD-L1 bioactivity test demonstrates reproducible dose response of PD-1 inhibition in serially diluted aliquots of Nivo mixed with IL (IL+Nivo) and OG+ containing Nivo (OG+Nivo) compared to Nivo (Control) (n=3). d. Live/dead cell viability assay of HepG2 cells incubated with OG leachable extract dilution of 0.097%, 0.78%, or 1.56% IL for 4 hours, revealing a dose-dependent increase in cell death (green) compared to living cells in the untreated control (blue). e. Sigmoidal plots and IC₅₀ values demonstrating enhanced cytotoxicity against HepG2 cells following 24 hours of incubation with serially diluted OG+Dox leachable extract compared to OG alone or NS+Dox hydrogel. f. Graph depicting augmented HepG2 cytotoxicity after treatment with OG+Dox extract containing 0.049%, 0.097%, or 0.195% IL compared to OG or OG+Dox. g. Fractional viability blots of SNU478, Panc-1, or MC38 cancer cells treated with serially diluted OG extract containing 0.097–25% IL and the calculated respective IC₅₀ values for each cell line of 0.21%, 0.05%, or 0.018% dilutions. h. Schematic of the transwell setup assessing relative cytotoxicity and drug diffusion of NS and OG hydrogels containing Dox and ICG through a 400 nm pore size barrier. HepG2 cells seeded in the lower chamber, after 1 day (D1), show darker red coloration of the medium in the OG group due to Dox diffusion, while the NS group indicates inefficient Dox diffusion. i. Graphs showing complete HepG2 cell death at D1 in the OG+Dox+ICG group, compared to preserved viability in the NS+Dox+ICG group. j, k. Graphs displaying markedly higher Dox and ICG fluorescence intensity respectively at D1, indicating active diffusion from the transwell containing OG+Dox+ICG into the lower chamber, unlike the NS+Dox+ICG group where no diffusion occurs. l. Graph illustrating relative cell viability of HepG2 cells treated with NS, NS+Dox, or OG+Dox extracts, demonstrating effective cytotoxicity in OG+Dox-treated cells. m. Quantitative analysis of HepG2 cell adhesion at 5 hours after treatment with serially diluted OG extract. n. Representative images and quantitative analysis of cell migration of HepG2 cells treated with OG extract containing 0.04% or 0.097% IL, revealing decreased cell migration at 72 hours. Data are means ± s.e.m.. Statistical significance was assessed using Student’s t-test for two-group comparisons and one-way ANOVA with post hoc analyses for multiple groups. ** P< 0.01, *** P< 0.001, and **** P< 0.0001. Scale bar, 100 μm.

To confirm that OG will preserve the bioactivity of Nivo, we performed a PD-1/PD-L1 blockade assay. Three sample types were prepared and analyzed: 1 mg/mL Nivo alone (Control), OG hydrogel containing 1 mg/mL Nivo (OG+Nivo), and 25% IL containing 1 mg/mL Nivo (IL+Nivo). Samples were serially diluted to obtain Nivo concentrations ranging from 0.003 μg/ml to 5 μg/ml and applied to treat two genetically engineered cocultured cell lines: PD-1 effector Jurkat T cells expressing human PD-1 and a luciferase reporter, and PD-L1 aAPC/CHO-K1 cells expressing PD-L1. The assay demonstrated an equivalent dose-response PD-1 blockade in OG+Nivo and IL+Nivo samples, similar to the Nivo control, with IC₅₀ values of 0.2743, 0.2745, and 0.4996, respectively, suggesting the preservation of Nivo functionality in OG and IL (Figure 2c).

Based on these results, we speculated that OG provides stabilization of Nivo activity. ILs have been found to enhance the solubility and stability of biomolecules, such as proteins, through what’s known as the Hofmeister effect.^[13] Proteins like the immune checkpoint inhibitor Nivo (IgG4) rely on a delicate interplay of covalent (like hydrogen bonding) and non-covalent interactions (such as electrostatic and hydrophobic interactions) to maintain their structural and functional integrity. In OG, both choline and geranate, which are constituents of the IL, demonstrate Hofmeister behavior. This behavior likely contributes to maintaining the stability of Nivo by preserving the balance between intramolecular interactions within the protein and interactions with the surrounding solvent environment.^[14]

Next, the diffusion capability of IL in OG was examined using fluorescence and FTIR imaging to track Dox and IL, respectively, in a tissue-like agarose matrix (Figure S2a–d). Fluorescence imaging followed by chemical bond mapping using FTIR-ATR microscopy indicated that Dox and IL disperse radially from the centrally injected OG. However, most of the Dox remained centrally with IL demonstrating maximal diffusion reaching the edges of the matrix by 24 hours. Both IL and Dox are detected in an overlapping manner at 24 hours implying that IL through ionic interactions is likely co-delivering and depositing the Dox to peripheral regions (Figure S2e).

Since discerning IL in vivo using FTIR-ATR microscopy can be challenging because it contains biological chemical bonds, the data here implies that the detection of Dox using fluorescence also likely represents the location of IL. IL plays an additional role here in that free Dox when injected into tissue immediately washes away; and when injected with NS, minimal Dox diffuses (i.e., Figure 1d, l); however, when Dox is co-delivered with OG, Dox is detected radially through its interactions with the IL. Based on these results, we speculate that sustained drug release by OG is facilitated by reversible electrostatic interactions among NS, Dox, and IL. NS, highly negatively charged, robustly encapsulates positively charged molecules such as Dox via electrostatic interactions.^[15] Collectively, our in vitro studies confirm minimal Dox release from NS alone. However, introducing IL alters NS-Dox interactions, enhancing Dox release. Choline, abundant in IL, binds NS due to cation exchange capability.^[16] Continuous cation exchanges among choline, Dox, and other cations in physiological conditions likely drive gradual IL and Dox release from OG. Additionally, the osmolarity difference between OG and tissues promotes passive diffusion of IL, facilitating drug release.

2.2. Testing OG performance in cell culture

To assess toxicity against the HepG2 human liver cancer cells, a serially diluted leachable extract ranging from 25–0.048% of OG, OG loaded with Dox (OG+Dox), or NS hydrogel containing Dox (NS+Dox), were incubated for 24 hours. The results showed profound cytotoxicity of OG+Dox that was superior to OG alone with an IC₅₀ of 10⁻⁵% and 0.23%, respectively (Figure 2d,e). Conversely, NS+Dox resulted in moderate cytotoxicity with an IC₅₀ of 14% suggesting a much slower Dox release. Figure 2f demonstrates the synergistic cytotoxicity effect of Dox with IL in OG, augmenting their anticancer cell effect at remarkably low concentrations (0.049, 0.097%, and 0.195% of leachable extract). Additionally, serially diluted OG leachable extracts showed robust cytotoxicity against various human and cancer cells including SNU874 (human cholangiocarcinoma cells), Panc-1 (pancreatic adenocarcinoma), and the mouse colorectal adenocarcinoma cells, MC38, with respective IC₅₀ values of 0.21%, 0.05%, and 0.018% (Figure 2g). To evaluate the capability of OG to release IL and small molecule drugs from the composite formulation, Transwell experiments were performed. The top chamber of the Transwell received either OG or NS containing OG Dox and ICG; this chamber was separated from the lower chamber containing HepG2 cells by a 400 nm pore-size membrane, similar to the porosity of the blood vessels in tumors (Figure 2h).^[17] Following 24 hr incubation, cytotoxicity and trans-membrane drug release in the lower chamber were measured (Figure 2h–k). Complete HepG2 cell death in OG+Dox+ICG treated transwells was observed; in contrast, control transwells that received NS+Dox+ICG showed the absence of any cell death (Figure 2i). Consistent with these viability assays, fluorescence of the cell culture fluid in the lower chamber also demonstrated a significant signal in the OG+Dox+ICG treated transwells in stark contrast to the NS+Dox+ICG samples suggesting that OG is required for IL, Dox, and ICG to traverse the membrane and accumulate in the lower chamber (Figure 2j,k). Based on these cytotoxicity findings we hypothesize that the tumor ablation effect of OG is primarily driven by its IL component. The amphiphilic nature of ILs allow them to interact with the phospholipid bilayer of cell membranes, causing destabilization and permeabilization. This disruption can lead to the leakage of cellular contents, loss of membrane integrity, and ultimately cell death.^[18] ILs are also known to generate reactive oxygen species (ROS) within cells, leading to oxidative stress.^[19] ILs contain charged ions that can disrupt ion gradients across cell membranes, leading to ion imbalances within cells that interfere with essential cellular processes such as osmoregulation, and signaling, ultimately leading to cell dysfunction and death.^[20] Additionally, the inclusion of the chemotherapeutic component, Dox, in the OG formulation enhances the IL cytotoxic effect, facilitating intracellular uptake and creating a synergistic cytotoxic effect for more effective tumor ablation.

Adhesion and migration are two important cancer cell activities that orchestrate tumor invasion.^[21] Here, we examined the ability of OG to disrupt HepG2 cell adhesion and migration using sublethal doses of OG extract. Quantitative analysis of HepG2 cell adhesion incubated with OG extract containing 0.048%, 0.097%, 0.19%, 0.39%, or 0.78% IL showed a proportional decrease in adhesion with rising IL concentration (Figure 2l). Additionally, HepG2 migration was substantially reduced after incubation with OG extract containing 0.04%, or 0.097% IL at 72 hours (Figure 2m). These data suggest that at IL concentrations less than the IC₅₀ of HepG2 cells (i.e., IC₅₀ = 0.23%; Figure 2d and Figure 2e), the cell may appear viable, however, with reduced ability to migrate and adhere, which may have major implications for further decreasing tumor recurrence at the tumor margins.^[22]

2.3. Assessing OG in normal rat livers

Locoregional therapy today involves ablating a treatment margin of >1 cm of normal liver which made testing OG in normal liver clinically relevant. Intraparenchymal injection of selected OG formulations into non-tumor-bearing rat livers can provide valuable information regarding injectability, tissue ablation, imaging characteristics, long-term drug distribution, and retention of IL and Dox prior to testing in tumor models. Rats received 50 μl intraparenchymal injection of NS, OG, or OG+Dox at three sites in each lobe. Subgroups of rats were euthanized after receiving ExiTron 12000 contrast agent at 1, 14, and 28 days followed by micro-computed tomography (microCT) and fluorescence imaging, as we previously described.^[23] Real-time ultrasound imaging during liver injection demonstrated echogenic appearance allowing clear visibility of OG (Figure S3a,b). 3D rendering of contrast-enhanced reconstructed micro-CT scans of the injected livers allowed segmentation of the ablated site and 3D volume measurements (Figure S3c–e). Micro-CT analysis revealed a larger ablation volume in the OG+Dox livers at 1, 14, and 28 days after injection compared to OG, or NS alone (Figure S3c–e). At necropsy, the liver tissues were bisected at the midline of the ablation zone for ex vivo fluorescent imaging. Fluorescence imaging revealed a larger area of ICG fluorescence in the OG+Dox compared to the OG, or NS injection sites (Figure S3f–h). Additionally, OG+Dox sites showed a large region of Dox fluorescence with higher intensity up to 28 days after injection (Figure S3i–k). These data suggest that OG containing Dox widens the ablation margin, enhances uniform distribution, provides stability of ICG and Dox for a prolonged period, and, most importantly, maintains the Dox within the ablation zone for up to 28 days.

Consistent with the micro-CT imaging, H&E-stained histology sections showed a larger area of ablation in the OG+Dox at 28 days compared to NS or OG (Figure S4a–d). Quantitative analysis of MPO+ cell count within the ablated zone showed a marked increase of MPO+ cells on day-1 that decreased by 14 and 28 days in all injected sites (Figure S4a–c, e). This increase in MPO+ count was markedly higher in the OG+Dox injected site compared to NS suggesting enhanced inflammation. Furthermore, quantitative analysis of T lymphocyte (CD3) count showed a progressive increase in T cell infiltration that peaked at 14 days and remained higher in the OG and the OG+Dox injected sites (Figure S4a–c, f). Quantitative analysis of CD68+ infiltration showed a higher cell count suggesting higher monocyte/macrophage infiltration in the NS injected site at 14 and 28 days compared to OG and OG+Dox (Figure S4a–c, g). A marked increase of CD3-positive cells in the OG-treated livers may potentiate immunotherapy at the treatment margins.

2.4. OG increases survival and cytotoxic T-cells in MC38 ectopic tumor model

Next, we evaluated survival, tumor growth, local immune cell recruitment, and protein expression after intratumoral injection of OG in the ectopic MC38 mouse model of colorectal adenocarcinoma. MC38 cell line was chosen because the liver is a common metastatic site in patients with colorectal cancers.^[24] The highly malignant MC38 mouse tumor model has extensively been used to assess tumor response to immunotherapy.^[25] To assess the effect of intratumoral injection of OG on tumor progression and survival rate, two groups of C57BL6 mice bearing 100–150 mm³ MC38 tumors were divided into the OG treatment group (OG) and the untreated control group (control). These mice survived either for 12 days for flow cytometry and CyTOF experiments, or the experiments were continued to determine survival rate and tumor response (Figure 3a). The treatment group received an intratumoral injection of OG loaded with Dox and the clinically used Nivo, which is an anti-PD-1 IgG4, under ultrasound guidance (Figure 3b). Nivo is the most commonly used immunotherapy drug worldwide; the goal of this experiment was to assess the deliverability of Nivo in a murine cancerous tissue matrix. The tumors in the control group progressed within 24 days and were considered dead per IACUC criteria whereas mice in the OG group had a complete tumor response associated with a diminished tumor volume and 100% survival rate up to 49 days following injection (Figure 3c–e). At necropsy, H&E and PCNA stained MC38 tumor histology sections at 49 days showed cancer regression and minimal cell proliferation in the OG group. The control tumor sections, however, showed larger tumor areas and marked cellular proliferation (Figure 3f,g). These data suggest that intratumoral injection of OG induces a positive tumor response, markedly enhancing survival.

Figure 3. Ultrasound-guided intratumoral injection of OG into ectopic MC38 tumors in mice.

a. Schematic illustrating ultrasound (US)-guided injection of OG into MC38 tumors in mice for evaluating survival and the immune response. b. Gross views and US images of MC38 tumor-bearing mice displaying the tumor before and during intratumoral OG injection. The images show a high echogenic needle inside a hypoechogenic tumor lesion (dotted outline) and complete tumor response at 49 days after OG injection, leaving a visible skin scar on ultrasound (dotted outline), in contrast to tumor progression in the control-treated mouse (dotted outline). c. Individual MC38 tumor volumes measured serially in control and after OG injection, demonstrating consistently reduced tumor volume up to 49 days (n=7) compared to mostly progressed tumors in the control group by 24 days (n=7). d. Graph depicting the mean change in tumor volume in control and OG-injected MC38 tumors. e. Survival analysis of mice with MC38 tumors after OG injection, showing significantly enhanced survival up to 49 days compared to control (P=0.0003, log-rank analysis). f. H&E and PCNA stained histology sections of the control MC38 tumor, illustrating a substantially large tumor section (H&E, control) and active cellular proliferation (Control, PCNA) compared to a completely regressed tumor with no active proliferation 49 days after OG injection. g. Analysis of PCNA+ cell count in MC38 histology sections, showing a significantly higher number of proliferating cells in control compared to OG-injected tumors. h. Flow cytometric plots demonstrating ~13-fold higher CD45+CD3+ lymphocytes in a single cell suspension of MC38 tumors injected with OG compared to control (n=4). i. Flow cytometric plots and analysis indicating a ~4-fold higher proportion of CD8+ T lymphocytes in gated CD45+CD3+ lymphocyte population in OG-injected MC38 tumors compared to control (n=4). j. H&E, CD3+, and CD68+ stained histology sections from MC38 tumors obtained 12 days after intratumoral OG injection, showing colocalization of T lymphocytes (CD3+), macrophages (CD68+), and anti-PD-1 antibody in the same ablation zone. Scale bar in b, 5 mm. Statistical analyses were performed using unpaired t-tests. ** P< 0.01, *** P< 0.001, and **** P< 0.0001.

To evaluate the local immunomodulatory response and protein expression in response to intratumoral injection of OG, treated and untreated MC38 tumor tissues were harvested at 12 days for flow cytometry analysis and histology. Flow cytometry analysis of tumors that received intratumoral injection of OG showed lower cell viability and a marked increase in CD45⁺CD3⁺ T-lymphocyte population compared to control samples (Figures 3h and Figure S5a–e). On further gating analysis, the majority of the CD45⁺CD3⁺ T-lymphocyte population were of CD45+/CD3+/CD8⁺ subset (Figure 3h,i). Histologic analysis of OG-injected tumor sections showed infiltrating lymphocytes (CD3+) in the ablated zone that colocalized with antigen-presenting cells (CD68+) in locations with abundant Nivo detection (Figure 3j). These findings suggest that OG-mediated tumor ablation induces a robust immune response that favors cytotoxic CD8+ T-lymphocyte recruitment to the tumor site.

To identify the immune cells in the MC38 tumors in response to OG injection, we created a single cell suspension from treated or untreated control tumors harvested at 12 days (based on Figure S4b where CD 3 detection peaked at 2 weeks) and performed a high dimensional mass cytometry analysis using a panel of forty-two labeled antibodies (Fluidigm, CyTOF). Mass cytometry analysis of tumor-infiltrating CD90.2/Thy-1.2-positive lymphocytes revealed 20 RphenoGraph cell clusters. Based on the expression of cell surface molecules and intracellular transcription factors, seven subsets and 5 subsets of CD4-positive and CD8-positive T cells respectively, were identified. CD4-positive T cells included both effector and T regulatory cells (Figure 4a,b).

Figure 4. High-dimensional analysis of infiltrating immune cells and proteomic variation in response to OG injection of MC38 tumors.

a, b. Representative t-SNE 2-dimensional map (left) and heat map (right) obtained through RphenoGraph analysis on a Helios mass cytometry workstation, depicting CD90.2/Thy-1.2-positive cells in a single-cell suspension of OG-treated MC38 tumors. Antibodies targeting 43 surface and intracellular proteins, detailed in the heat map clusters (b), revealed distinct subsets, including natural killer (NK), natural killer T cells (NKT), CD8+, and CD4+ cells, marked with outlines for clarity in the CyTOF analysis (a). c. Volcano plot resulting from proteomic analysis of OG-treated MC38 tumors versus control, illustrating the number of differentially expressed proteins (DEPs). DEPs were determined using the limma-voom method with a fold-change cutoff logFC ≥1 and a false discovery rate (FDR) of 0.055. LogFC was computed as the difference between the mean of log2(OG) and the mean of log2(control). Blue dots represent downregulated DEPs, gray dots indicate non-significant DEPs, and red dots represent upregulated DEPs. d. Heat map displaying immune response-related protein expression in OG-treated tumors compared to controls. The heatmap showcases enriched biological pathways filtered for consistency. Each colored square represents the Z score for the expression of one protein, with red denoting the highest expression, white indicating the median, and blue representing the lowest expression. Immune-related pathways are shown on the left, while selected proteins are listed on the right. Sample size: n=4 in each group.

To gain insights into the impact of intratumoral injection of OG on the tumor microenvironment, including interactions between cancer cells and the immune proteome, we examined the proteome of murine MC38 tumors after injection with OG or controls. Differentially expressed proteins (DEPs) were identified using the Limma-Voom method. We found that among the 4,071 proteins that were differentially expressed, 187 DEPs were upregulated, and 132 DEPs were downregulated (2-fold change >2, P<0.05) (Figure 4c). Among the downregulated DEPS included many proteins known to be involved in cellular homeostasis and metabolic pathways (Ppp4r3b, Nom1, Rab9a); however, the upregulated DEPs included those involved in immune response and oxidative damage to the cell membrane (Apoa4, Hbb-y, Hbb-bs) (Figure 4c). Moreover, ingenuity pathway analysis (IPA) predicted that most activated signaling pathways in our dataset were immune-related networks, including acute phase response signaling, immune response to leukocytes, and phagosome formation (Figure 4d and Figure S6). Altogether the altered expression of the 4,071 DEPs and IPA analysis demonstrate that intratumoral injection of OG activates pathways involved in cellular immune response associated with loss of tumor cell homeostasis that enhances cancer cell cytotoxicity consistent with the flow cytometry and CyTOF analysis in Figure 4.

2.5. OG promotes efficient ablation and distribution of drugs in human tissues

Next, we performed ex vivo experiments to assess the efficacy of OG in inducing ablation and ensuring consistent drug distribution in various human cancer tissues, such as renal cancer, gastric cancer, and cirrhotic liver tissues. Freshly excised human tissues were injected with OG containing Dox, Indocyanine Green, and Nivo (OG+) using a needle and then incubated for 24 hours at 37 °C in a humidified chamber (Figure 5a and Figure S7a,d). Fluorescent imaging was used to serially evaluate ICG and Dox diffusion, whereas morphometric analyses of serially cut histological sections were used to quantify nuclei count and detect Nivo levels within the ablated zone. Quantitative analysis of ICG and Dox fluorescence demonstrated a time-dependent increase in the diffusion area at 1 and 24 hours after OG+ injection (Figure 5b,c and Figure S7d,e). Additionally, histological evaluation at 24 hours after OG+ injection into human tissues revealed a significant decrease in nuclei count, destruction of tissue architecture indicating ablation, and substantial Nivo detection throughout the ablation zone compared to the control (Figure 5d,e and Figure S7c,f). These results underscore the efficacy of OG in eradicating tumor cells and ensuring consistent drug distribution in diseased or cancerous human tissues. This study highlights the promising potential of percutaneous OG injection as a therapeutic approach for patients with liver cancer where many patients have underlying densely fibrotic cirrhosis.

Figure 5. Assessment of ex vivo ablation and drug diffusion capability in human cancer tissues and human derived bacteria.

a, Gross image depicting intratumoral injection of OG loaded with Dox, Indocyanine Green (ICG), and Nivo (OG+) using a syringe needle into transected human liver cirrhotic tissue. b, Fluorescence images depicting the time-dependent increase in Dox and ICG distribution at 1- and 24-hours post-OG+ injection. c, Plot illustrating the increase in fluorescence intensity of Dox and ICG at 1- and 24-hours post-OG+ injection into human tissues excised from patients with renal cancer, gastric cancer, and cirrhotic liver. d, Stained histology sections of human cirrhotic liver tissue demonstrating effective tissue ablation and uniform Nivo distribution in the ablation zone 24 hours after OG+ injection compared to control. e, Graphs demonstrating a decrease in nuclei counts associated with an increase in Nivo detection in all human tissues injected with OG+ (n=9). f, Representative blood agar plates and chart displaying bacterial growth and susceptibility after treatment with serially diluted OG or NS extracts for 24 hours. Data presented as mean ± s.e.m. Statistical analysis was performed using two-way repeated measures ANOVA in c and unpaired t-tests in e. * P< 0.05, *** P< 0.001, and **** P< 0.0001. Scale bars: 1 cm in a and 75 μm in d.

2.6. Evaluating the antibacterial effect of OG against human-derived pathogens

In clinical settings, infections in patients with liver tumors following an intervention can be a devastating complication arising from disruptions in tissue barriers and compromised immunity due to factors such as chemotherapy, radiation, and stem cell transplantation.^[26] Our investigation aimed to determine whether OG could also serve as a preventive measure against potential post-treatment infections. Susceptibility tests were carried out using patient-derived pathogens known to commonly cause infections in individuals with liver tumors following an ablation procedure. Our study revealed that OG displayed a potent antimicrobial effect against a spectrum of pathogens commonly found in cancer patients, including Enterococcus faecium (E. faecium), Enterococcus faecalis (E. faecalis), the gram-negative anaerobe Klebsiella pneumoniae (K. pneumoniae), the gram-positive Staphylococcus intermedius (S. intermedius), and Escherichia coli (E. coli) bacteria (Figure 5f). The robust antimicrobial properties observed with OG suggest its potential to inhibit tumor abscess formation following ablation. This implies that the intratumoral injection of OG could play a crucial role in safeguarding patients with cancer and liver cirrhosis from potential septic complications, especially in the context of infections that may be induced during interventions.

2.7. OG ablates and uniformly distributes Nivo in the N1S1 orthotopic liver tumor model

Sprague-Dawley rats bearing N1S1 tumors in the liver were randomly divided into two groups to receive an intratumoral injection of OG or NS loaded with Dox and Nivo (namely OG+, and NS+ respectively) and then survived for two weeks to monitor tumor growth. Ultrasound imaging showed marked lower tumor volumes in the OG+ injected tumors (Figure 6a–c). At necropsy, histology sections showed a markedly smaller tumor area with diminished cellularity in the OG+ injected tumors compared to NS+ (Figure 6d–f). To determine the distribution of NS within the tumor zone, FTIR microscopy mapping of silicate oxide (Si-O) was performed. These images showed a high-intensity Si-O region at the injected site in the ablated tumor sections compared to a diminished and sporadically detected Si-O in the NS-injected tumors (Figure 6g–i). To correlate the Si-O location on the FTIR integration map with the distribution of the co-administered Dox, the slides were imaged using fluorescence scanning (Figure 6j–l). Fluorescence imaging for Dox in OG+ samples showed uniform distribution and retention of the Dox within the ablation zone. Similar to observations in Figure S2, the Dox signal was concentrated in the injected site where high levels of Si-O were detected. In NS+ control samples, little Dox signal was detected consistent with the low Si-O detection by FTIR mapping. These data demonstrate that OG-injected tumors generate an extensive area of ablation outside the NS mass implying that IL diffused away from the OG causing ablation while co-delivering and retaining drugs such as Dox throughout the ablation zone. Next, tissue sections were immunostained for Nivo; these images showed intense, uniform detection of Nivo (brown color) in the OG+ samples compared to minimal detection in the NS+ samples (Figure 6m–o), mirroring Dox fluorescence signal and the ablation zone seen in H&E (Figure 6d, e, j, k). To examine if OG+ treatment incites a T lymphocyte response, sequential histology sections were immunostained for CD3. These slides showed an abundance of CD3+ cells at the tumor margins where Nivo was also detected at high levels (Figure 6p,q), consistent with the MC38 and the CyTOF data.

Figure 6. Intratumoral injection of OG into N1S1 orthotopic HCC model in rats.

a. Ultrasound (US) scan of N1S1 tumor two weeks after intratumoral injection with NS hydrogel (control) containing anti-PD-1 antibody (AB) and Dox (NS+) (n=6 in each group). b. US scan of ablated N1S1 tumor two weeks after intratumoral injection of OG containing anti-PD-1 AB and Dox (OG+). c. Graph depicting the change in tumor volume measured with US at 1 and 2 weeks after injection with OG+ or NS+. d, e. Corresponding H&E-stained histology sections of N1S1 tumors at 2 weeks after injection of NS+ or OG+ respectively. f. Morphometric analysis of N1S1 cross-sectional tumor area in OG+ and NS+ injected groups. g, h. FTIR integration map of Si-O peak displaying NS distribution within the explanted N1S1 tumor section two weeks after injection with NS+ or OG+. i. FTIR spectra were obtained from the Si-O positive (blue +) or negative (red x) loci on the map marked in g and h. j, k. Images of Dox fluorescence in explanted N1S1 tissue sections demonstrating detection throughout the OG+ injected tumor compared to diminished Dox fluorescence in NS+ injected tumor. l. Quantitative analysis of Dox fluorescence showing the diffusion area in N1S1 two weeks after injection with OG+ or NS+. m,n. Immunostained histology sections visualizing PD1 AB detection within N1S1 tumors (brown staining) two weeks after injection of NS+ or OG+. o. Quantitative analysis of PD-1 AB area of detection within the tumor section two weeks after intratumoral injection of NS+ or OG+. p, q. Immunostained histology image of OG+ injected N1S1 tumor displaying abundant CD3+ T cells cordoning the ablation zone. Sample size: n=6 in both groups. Scale bars in US images are 0.5 cm. Statistical analysis between the two groups was conducted using unpaired t-tests. ns; not significant, ** P< 0.01, *** P< 0.001, and **** P< 0.0001.

2.8. OG ablates and uniformly distributes Nivo in the rabbit VX2 orthotopic liver tumor model

The VX2 rabbit liver tumor model is the most common preclinical model for testing interventional therapies. This highly malignant VX2 tumor offers the opportunity to evaluate novel therapies and interventions in human-sized tumor lesions with similar vascularity allowing the use of clinical imaging platforms and procedural tools. Two groups of rabbits bearing VX2 liver cancer received intratumoral injections of OG+ or NS+ and survived for two weeks. Rabbit tumors that received NS+ showed significant tumor growth with persistent vascularity on angiography. In contrast, rabbit tumors that received OG+ were markedly smaller in size with reduced tumor vascularity implying significant tumor response to OG+ (Figure 7a–d). At necropsy, harvested tumors were transected along the sagittal plane showing markedly smaller tumors with enhanced ICG fluorescence on IVIS imaging in the OG+ group compared to the NS+ control group consistent with the ultrasound findings (Figure 7e). To colocalize ICG with nanosilicate distribution, tissue sections were mounted on highly reflective slides and imaged using FTIR mapping showing widely distributed Si-O in the tumor region in the OG+ injected group compared to the diminished signal in the NS+ control group (Figure 7f). Immunostained tumor sections showed a markedly lower count of proliferating cells in the ablation zone associated with an abundant area of Nivo detection in OG+ compared to limited Nivo detection in NS+ (Figure 7g–i). These data demonstrate that IL in OG+ is necessary and ablates the highly malignant VX2 tumors and uniformly distributes Nivo. Complete blood counts and blood chemistries revealed no differences at two weeks following intratumoral injection of OG+ or NS+ and all values were within normal limits (Table S1). These results further confirm biocompatibility and the safety of OG+ injection.

Figure 7. Intratumoral injection of OG into VX2 liver tumors in rabbits.

a. Ultrasound and DSA images of VX2 tumors at baseline and 2 weeks after injection with NS+ or OG+ hydrogels containing anti-PD-1 AB and ICG, showing comparable tumor size at baseline and smaller tumors at 2 weeks after OG+ injection (dashed outline). b. Plot displaying serial measurements of tumor volume at 0, 1, and 2 weeks after injection with OG+ or NS+. c. Graph showing the relative change in tumor volume at 2 weeks after injection with OG+ or NS+. d. DSA run of VX2 tumors from the hepatic artery, illustrating the lateral branches of the left hepatic artery and high vascularity of the tumor lesion at two weeks after injection with NS+ (left, yellow dashed outline) compared to diminished vascularity after OG+ injection (right, yellow dashed circle). e. Gross images (top) and corresponding fluorescence scans (bottom) of transected VX2 tumors at two weeks after injection with NS+ or OG+, visualizing ICG detection. f. FTIR integration maps and corresponding spectra from scanned VX2 tumors showing NS distribution within the ablated tumor at two weeks after OG+ or NS+ injection. The spectra in f correspond to the NS location indicated with + on the FTIR-mapped locus, consistent with the highlighted peak in the black spectra, whereas the NS-free locus in the ablated zone is marked with x on the map, corresponding to the red spectra. g. H&E and immunostained histology sections for PD-1 AB and PCNA from OG+ or NS+ injected VX2 tumors. h,i. Graphs showing PCNA+ proliferating cell count and PD-1 AB localization area within the tumor zone at two weeks after NS or OG+ injection. Scale bars = 5 mm in a, e, f, g, and 10 mm in d. Data are presented as mean ± s.e.m.. Statistical analysis was conducted using unpaired t-tests. ** P< 0.01, *** P< 0.001, and **** P< 0.0001.

3. Conclusion

Liver cancer remains a major worldwide public health problem, especially with 270 million living with viral hepatitis who are at risk for developing cirrhosis and liver malignancies. Most liver cancers that present to the clinic are in advanced stages, limiting treatment options often to palliative care. Poor liver function often further limits systemic palliative therapies, however, in those patients who can receive chemotherapy and/or immunotherapy, they have been largely ineffective with no impact on 5-year survival. Here we describe a potential approach to salvage therapy of advanced stage liver cancer patients that may achieve immunotherapy. We show for the first time that direct intratumoral injection of a hydrogel can achieve complete tumor ablation and uniform drug delivery of both small compounds (i.e., Dox) and large compounds (i.e., Nivo), in addition to retaining therapeutics within the ablation zone. Consequently, OG treatment in vivo models led to a strong recruitment of T-cells to the ablation zone as part of the intense inflammatory response to OG. In a setting of tumor ablation, inflammation and high levels of local ICI delivery suggest the optimal timing to achieve immunotherapy and potentially overcome the limitations of immunotherapy in liver cancer.

4. Methods

4.1. In vitro studies

Preparation of IL-based OG:

The OG components include a specified ratio of NS hydrogel and IL. Neat IL was prepared from choline (Sigma) and Geranic acid (Sigma) at 1:1 molar ratio and incorporated into OG at a predetermined weight percent (wt%). To prepare the basic OG formulation, NS hydrogel was initially prepared by hydrating a 3–9 wt% of nano silicate powder (Laponite-XLG, G-6302889, BYK USA INC.) in an appropriate volume of ice-cold ultrapure water followed by 5 minutes efficient mixing at 3500 rpm using a SpeedMixer (FlackTek, Inc., Landrum, SC, USA). Selected wt% of NS hydrogel was combined with an appropriate amount of IL and mixed using the same mixing protocol to generate OG containing 6.25, 25, or 50 wt% IL. To prepare OG loaded with anticancer therapeutics and/or imaging agents, the anticancer drug, Dox (1.25 mg/mL, Cayman Chemical Company, Ann Arbor, MI, USA), the immunotherapeutic agent, anti-PD-1 IgG4 (1 mg/mL, Nivo, Bristol Myers Squibb, Princeton, New Jersey), and the near-infrared fluorescent agent, indocyanine green (ICG, 0.25 mg/mL, 340009, USP, Rockville, MD) were added and mixed during the initial NS hydrogel preparation step then mixed with IL. NS containing different therapeutics and imaging agents without IL was used as a control.

Rheometry studies:

We assessed the mechanical properties of various NS and OG formulations using an Anton Paar MCR 302 rheometer (Anton Paar USA Inc., Torrance, CA) at 37 °C, following established methods.^[27] A sandblasted aluminum upper plate (25-mm diameter) and an aluminum lower plate were used with a consistent 1 mm gap. Shear rate sweep tests (10⁻¹ s⁻¹ to 10³ s⁻¹) characterized shear-thinning behavior, while LAOS tests were performed at 10 rad s⁻¹. Frequency sweep tests (0.1% strain, 0.1 to 100 rad s⁻¹) were performed in the linear viscoelastic region. A solvent trap with water ensured a humidified environment to prevent drying. All measurements were performed in a minimum of triplicates.

Doxorubicin diffusion assay:

To assess enhanced drug diffusion of OG loaded with chemotherapy within a tissue-mimicking matrix, we used circular agarose disks (2%) with central wells as reservoirs for various hydrogel samples. Low electroendosmosis agarose (Roche, Germany) was mixed in PBS (pH 7.4) to create a 2% suspension, melted, and cooled to 60 °C. Using 3D-printed inserts, 5-mm central wells were formed in agarose poured into 6-well plates and solidified at room temperature. Wells were loaded with 3% NS mixed with Dox and various IL concentrations (6.25%, 12.5%, 25%, and 50% wt%) to study IL impact on Dox diffusion. Additionally, 25% IL with Dox and varying NS concentrations (3%, 4.5%, 6%, and 9% wt%) were used to explore NS concentration effects. Serial fluorescent images were taken at intervals using the IVIS Spectrum in vivo imaging system (PerkinElmer Inc., Waltham, MA) at 460/560 nm with a 1-second exposure, f/stop = 2, and uniform intensity thresholds. Fluorescence intensities were quantified as radiance values normalized to photons per second per square centimeter per steradian (p/s/cm2/sr). The enhanced fluorescence area within each disk was calculated using a standardized threshold.

Zeta potential and conductivity:

The zeta potential and conductivity of both NS and OG preparations were determined at 25 °C using a Zetasizer Ultra Red Label (Malvern Instruments Ltd., UK) following the manufacturer’s guidelines. In summary, NS or OG samples were dispersed in ultra-pure water using a Speed Mixer and transferred to dialysis tubes, ensuring a consistent nano silicate level of 1 mg/mL in each sample before measurement. These dispersed samples were then loaded into disposable folded capillary cells (DTS1070, Malvern Instruments Ltd.). To measure zeta potential, the electrophoretic mobility of the sample was assessed and subsequently converted into zeta potential values using the Henry equation. Both conductivity and zeta potential data were collected and analyzed utilizing the ZS Xplorer Software (Malvern Instruments Ltd.). These steps ensured an accurate assessment of the zeta potential and conductivity of the NS and OG preparations, providing valuable insights into their electrokinetic properties.

In vitro drug release:

The in vitro release profile of Dox from 3% NS+Dox and OG+Dox was assessed in PBS under both physiologic (pH 7.4) and acidic (pH 5) conditions using a modified dialysis method.^[4] A Spectra/Por Float-A-LyzerG2 dialysis device (MWCO: 8–10 kDa, G235031, Repligen Corporation, Rancho Dominguez, CA) was loaded with 0.5 mL of OG + Dox or 3% NS + Dox and placed in a 50 mL conical tube submerged in 20 mL of PBS. The tube was continuously agitated on a nutating platform at 60 rpm inside a 37 °C humidified chamber. At specified intervals, 1 mL of PBS was withdrawn from the tube and replaced with an equal volume of fresh PBS. The concentration of released Dox in PBS was determined by measuring its fluorescence, with Dox concentration extrapolated from a standard curve obtained using serially diluted Dox solution (Pfizer). Fluorescence measurements were performed using a SpectraMax iD5 plate reader (Molecular Devices, San Jose, CA). ^[23] The cumulative Dox release (CR%) over time was calculated using the formula: CR%(n) = (C_n ×20 + ∑(i=1)^(n-1) ci × 1)/(total amount of Dox loaded in the gel) × 100%. Where: $c_n$ is the concentration of Dox in the PBS outside of the dialysis tube at a given time point, ci is the concentration of Dox in the PBS outside of the dialysis tube measured at a previous time point, 20 is the volume of release medium in mL, and 1 is the collected sample volume at each time point. Additionally, Dox release from a tube containing free Dox in PBS was evaluated as a control for comparative analysis.

Injection force testing:

The injectability of various OG formulations through a 21-gauge standard micropuncture 7 cm needle (Cook Co, IN) was assessed by measuring the compression force applied to a syringe. This evaluation was conducted using a mechanical tester equipped with a 100 N load cell (Instron 5942, Instron Corporation, Norwood, MA), following a previously established protocol.^[28] This method enabled the determination of the injectability of different OG formulations, providing valuable insights into their practical application through standard medical needles. Briefly, OG materials were loaded into 1 cc syringes (Medallion, Merit Medical, South Jordan, UT) and injected at a controlled flow rate of 1 mL min⁻¹. The force required for injection was recorded using the system software (Bluehill version 3 Software, Instron Corporation).

Material sterility testing:

The sterility of OG or OG loaded with Dox (OG+Dox) was assessed following a well-established protocol.^{[8, 29]} In this procedure, 200 μL aliquots of extracts from each material were diluted in a 2.5 wt% Miller LB broth solution (Fisher Scientific). These diluted samples were then incubated on an orbital shaker for either 24 hours or 2 months at 37 °C, with continuous agitation at 180 rpm. In the control setup, plain LB broth served as the negative control, while LB broth inoculated with Escherichia coli (E. coli) was used as the positive control. To measure the results, the optical density of the suspension was assessed at 600 nm utilizing a microplate reader (Molecular Devices). Each experiment was conducted in five replicates to ensure the accuracy and reliability of the findings. This method allowed for the thorough evaluation of the sterility of both OG and OG+Dox, confirming their suitability for various applications by ensuring the absence of contaminating microorganisms for in vitro and in vivo experiments.

Fourier Transform Infrared Spectroscopy (FTIR) mapping:

The surface chemistry of both NS and OG was analyzed using FTIR (Fourier Transform Infrared) spectroscopy. FTIR spectra were obtained using attenuated total internal reflectance Fourier transform infrared (ATR-FTIR) spectroscopy equipment (Alpha II platinum ATR Accessory, Bruker, Billerica, MA). Spectra were recorded in absorbance mode, ranging from 4000 to 500 cm⁻¹, with a resolution of 4 cm⁻¹ for each measurement. To ensure consistency, each material was measured at least three times, with samples taken randomly from the bulk.

For FTIR mapping and analysis of tumor tissue, 4 μm thick paraffin-embedded sections were mounted on Low-e reflective microscope slides (Kevley Technologies, OH). These sections were baked at 58 °C and briefly deparaffinized for 1 minute in xylene (Thermo Fisher Scientific, Waltham, MA) and 200-proof ethanol. Spectral mapping in reflection mode was performed using an FTIR microscopy system (Lumos II, Bruker, Billerica, MA), equipped with a thermoelectrically cooled mercury cadmium telluride (TE-MCT) detector. After selecting the region of interest, the tissue was scanned at 4 cm⁻¹ resolution within the spectral range from 550 cm⁻¹ to 4000 cm⁻¹. Chemical correlation maps were generated by integrating the area under the Si-O peak (1100–950 cm⁻¹) to visualize NS distribution within the tumor section. A background reference was obtained from a tissue-free surface on each slide. The resulting maps used a color scale, with red indicating the highest band intensity/correlation and blue indicating the lowest intensity/correlation. Individual FTIR spectra from locations containing NS within the tissue boundary were extracted using OPUS software (Bruker, Billerica, MA).

Scanning Electron Microscopy (SEM):

A scanning electron microscope (JCM-6000Plus, JEOL, Peabody, MA) was employed to examine the microstructures of NS and OG. Initially, specimens were frozen at −80 °C and subjected to freeze-drying in a lyophilizer (0.120 mBar, and −50 °C, Labconco, Kansas City, MO). Following this process, all prepared samples were sputter-coated with a 7 nm layer of gold/palladium using a Leica EM ACE200 sputter coater (Wentzler, Germany) before being imaged using SEM.

Fluorescence imaging:

Fluorescence imaging was conducted to monitor the changes in Dox and ICG fluorescence over time in both OG and NS samples. Samples containing NS or OG mixed with 1.25 mg/mL Dox were used to measure Dox fluorescence, while samples mixed with 0.25 mg/mL ICG were used to measure ICG fluorescence. Specifically, 300 μL of these samples were loaded into black 96-well plates (Greiner) (n=5) and centrifuged at 1200 rpm for 2 minutes. Fluorescence images were captured using the IVIS imaging system (PerkinElmer).

Evaluating Nivo bioactivity in vitro:

The bioactivity of Nivo was evaluated using a PD-1/PD-L1 Blockade Bioassay following manufacturer instructions (J1250, Promega, Madison, WI). Samples, including OG hydrogel (3%NS and 25% IL) containing 1 mg/mL Nivo (OG+Nivo), 25% IL mixed with 1 mg/mL Nivo (IL+Nivo), and 1 mg/mL Nivo alone (Control), were prepared and serially diluted at a 1:2.5 ratio with assay buffer to generate concentrations ranging between 0.003 μg/ml to 5 μg/ml Nivo. Briefly, 100 μl aliquots of aAPC/CHO-K1 cells expressing human PD-L1 were seeded into a solid white flat bottom 96-well plate (Corning) and incubated overnight at 37° C with 5% CO2. Following the removal of 95 μl of basic cell recovery medium from the wells, 40 μl aliquots of the prepared diluted sample containing Nivo were added to designated wells in the plate containing the seeded cells, excluding the assay control wells. Subsequently, 40 μl aliquots of PD-1 Effector Cells were added into each well containing the samples and the PD-1 expressing cells and incubated at 37° C for 6 hours. After the incubation period, 80 μl aliquots of Bio-Glo Reagent were added into the wells and incubated at room temperature for 10 minutes. Luminescence values were acquired using a SpectraMax iD5 plate reader (Molecular Devices, San Jose, CA). The relative Nivo bioactivity was calculated based on the luminescence values and expressed as a fold of PD-1 blockade relative to the luminescence values of the untreated assay control wells.

Cell culture:

SNU-478 (adenocarcinoma of the ampulla of Vater), PANC-1 (pancreatic carcinoma of ductal origin, CRL-1469, American Type Culture Collection, Manassas, VA), HepG2 (human liver cancer cell line, CRL10741, American Type Culture Collection), and MC38 (mouse colon carcinoma cell line, ENH204-FP, Kerafast, Boston, MA, USA) cells were cultured in 75-cm2 flasks using specific media. SNU-478 and PANC-1 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM, Thermo Fisher Scientific, Waltham, MA) supplemented with 10% heat-inactivated bovine serum (HyClone, Logan, UT), 100 IU penicillin, and 10 μg/mL streptomycin (Thermo Fisher Scientific). HepG2 and MC38 cells were maintained in Iscove’s Modified Dulbecco’s Medium (IMDM, HyClone) supplemented with 10% heat-inactivated bovine serum, 100 IU penicillin, 10 μg/mL streptomycin, 1 mM sodium pyruvate (Gibco, Grand Island, NY), and 0.1 mM nonessential amino acids (Lonza, Walkersville, MD). All cells were incubated in a humidified 5% CO2 incubator at 37 °C.

In vitro cytotoxicity assay:

SNU-478, PANC-1, HepG2, and MC38 cells were incubated at 37°C until they reached confluency. Cells were then detached using 0.25% Trypsin-EDTA solution (HyClone) and seeded into 96-multiwell plates at 5,000 cells/well density. After 24 h of incubation, the medium was removed and 200 μL of fresh growth medium containing serially diluted OG leachable extracts ranging from 25–0.048% was added into designated replicate wells. In a series of experiments, extracts were obtained from hydrogels containing different components, including OG without Dox, OG with Dox, and NS with Dox, each containing 1.25 mg/mL of Dox. Extracts were acquired by incubating hydrogel aliquots with an equivalent volume of growth medium at 37° C for 24 hours in sterile polypropylene centrifuge tubes. After incubation, extracts were serially diluted in a growth medium to generate concentrations ranging from 25% to 0.048% for each extract. Subsequently, 100 μL aliquots from each dilution were transferred to designated wells in a 96-well plate, which had been previously seeded with 5000 HepG2 cells per well. The plates were then incubated for an additional 24 hours to assess the effects of the hydrogel extracts on the HepG2 cells. After the 24-hour incubation, the medium was removed, and wells were washed thrice with Dulbecco’s modified phosphate-buffered saline (DPBS, Sigma-Aldrich, Saint Louis, MO). Then, 100 μL of fresh growth medium was added to each well. Cell viability was assessed using WST-1 reagents (Cayman Chemicals, Ann Arbor, MI) following the manufacturer’s protocol. Briefly, 10 μL of WST-1 solution was added to each well and incubated at 37 °C for 2 hours. Absorbance was measured at 450 nm using a microplate reader. Viability was calculated relative to untreated cells using the formula: Viability (%) = (1 - OD_treated/OD_control) × 100%. Prism Software ver. 9 (GraphPad, San Diego, CA) was used to determine the IC₅₀ concentrations of OG extracts for inducing 50% cytotoxicity in HepG2, SNU478, PANC-1, or MC38 cells after 24 hours incubation.

Transwell diffusion assay:

The HepG2 cells (1 × 10⁶ cells per well) were seeded in the lower chamber of a 12-well transwell plate with a 0.4 μm polycarbonate membrane (3401, Corning, Kennebunk, ME) and incubated overnight at 37 °C in a 5% CO₂ cell culture incubator. The next day, the growth medium was replaced with 1.5 mL of fresh growth medium. Transwell inserts were placed into each well and loaded with 0.5 mL of NS hydrogel, NS hydrogel containing Dox and ICG, or OG containing Dox and ICG (n=4). After 24 hours of incubation, the inserts were removed. The growth medium in the lower chamber was collected and used to measure Dox and ICG fluorescence using a microplate reader. Cell viability assessment in the lower chamber was conducted by directly adding WST-1 reagent to the cells. The resulting solution was transferred to a separate 96-well plate for absorbance measurements.

Cell migration assay:

We conducted a cell migration assay using Ibidi’s Culture-Insert 2-Well 35-mm μDish. HepG2 cells (68,000 cells per 100 μL) were seeded into the Ibidi chamber in growth media and allowed to reach confluence, creating two zones separated by a standardized distance. At 24 hours after seeding, the chamber inserts were removed, and the growth medium was replaced with either a fresh growth medium (control) or a growth medium containing 0.097% or 0.048% NG extract. The cells were imaged using an EVOS FL Auto-2 microscope at 48 and 72 hours of incubation to assess migration (n=4).

Cell adhesion assay:

HepG2 tumor cells (0.5×10⁶) were seeded in 6-well plates containing growth medium with varying OG extract concentrations (0.048, 0.097, 0.19, 0.39, or 0.78 wt% IL). The cells were then incubated at 37 °C in a 5% CO₂ humidified chamber for 5 hours. Non-adherent cells were removed by washing twice with DPBS, followed by staining with the Live/Dead cell viability imaging reagent (R37609, ThermoScientific) in a growth medium for 15 minutes. After the incubation period, live cells were visualized using an EVOS FL Auto-2 microscope (Thermo Fisher Scientific) and counted in ten fields per well.

Dynamic light scattering (DLS):

The size measurement of dispersed Nivo molecules was assessed using Dynamic Light Scattering (DLS). Separate aliquots of Nivo were either incubated with OG extract containing 1.56% IL for 1 hour at 37 °C or subjected to microwave energy to investigate the impact of different ablation techniques on IgG₄ integrity and potential aggregation. The final concentration of Nivo was 100 μg per mL of medium. To simulate microwave ablation of liver cancer in patients receiving immunotherapy, Nivo was exposed to microwave energy at 110 Watts for 5 minutes using a benchtop microwave (Frigidaire FFMO1611LS). Following the respective treatments, the hydrodynamic size of each sample was examined using Multi-Angle Dynamic Light Scattering (MADLS) with a Zetasizer Ultra Red Label instrument from Malvern, UK.

Polyacrylamide gel electrophoresis and western blot analysis:

The molecular integrity of Nivolumab (Nivo) post-incorporation into OG or exposure to microwave treatment was analyzed via SDS-PAGE under non-reducing conditions. Samples were mixed with Laemmli buffer (#1610747, Bio-Rad, USA). Untreated Nivo aliquots served as controls. Equal amounts of protein (5 μg) were loaded into wells of a 4–15% Mini-PROTEAN TGX stain-free gel (#4568084, Bio-Rad) and electrophoresed at 100 volts for 75 minutes. Gel activation and protein visualization were performed using Gel Doc XR+ (BIO-RAD, Hercules, CA). Proteins were subsequently transferred onto a 0.2 μm PVDF membrane using Trans-Blot Turbo (BIO-RAD, Hercules, CA) and visualized with the Gel Doc XR+ imaging system. To assess OG’s impact on Nivo integrity in the presence of tumor cells, HepG2 cells were seeded in 6-well plates (#1610747, Bio-Rad, USA) at 1.5 × 106 cells per well and incubated for 48 hours. Selected wells were treated with a medium containing 1.56% IL and 100 μg/mL of Nivo. In separate experiments, HepG2 cells mixed with Nivo were microwave-treated at 110 W for 5 minutes. Protein extracts from OG and microwave-treated HepG2 cells were prepared using RIPA buffer, and protein concentrations were determined using the Pierce BCA protein assay (#23225, Thermo Scientific). Untreated HepG2 cell extracts served as controls. Protein aliquots (40 μg per lane) were loaded into 4–15% Mini-PROTEAN TGX stain-free gels and electrophoresed at 100 volts for 75 minutes. Trans-Blot Turbo was used to transfer proteins to 0.2 μm PVDF membranes for visualization with the Gel Doc XR+ system. GAPDH was probed for equal loading using rabbit anti-GAPDH IgG (ab70699, 1:2000, Abcam, Cambridge, MA), followed by HRP-conjugated anti-Rabbit IgG (ab97051, 1:5000, Abcam). Specific bands were visualized with Clarity Western ECL substrate (#170–5060, Bio-Rad) and analyzed using ImageJ software (NIH, Bethesda, MA, USA).

Bacteria susceptibility test

Bacteria were suspended in 0.45% saline solution tubes (Remel, San Diego, CA) to reach a 0.5 McFarland standard (1.5 × 10⁸ CFU/mL). In a 96-well plate, 100 μL of the bacterial suspension was mixed with 100 μL of serially diluted NS or OG in a 1:1 ratio. After a 24-hour incubation at 37 °C, the mixtures were plated directly on sheep blood agar (Remel) using cotton-tipped applicators (MEDLINE, Northfield, IL). Subsequently, the plates were further incubated for 24 hours to evaluate bacterial susceptibility. Interpretation of the results was based on a color-coded heatmap: Susceptible concentrations (depicted as green in the heatmap) indicated the absence of bacterial growth, Intermediate concentrations (depicted as orange in the heatmap) demonstrated reduced bacterial growth compared to the control, and resistant concentrations (depicted as red in the heatmap) showed bacterial growth equivalent to that of the control.

Evaluating ablation and drug delivery potential in ex vivo human tissues

Freshly harvested specimens including gastric cancer (n=1), renal cell carcinoma (n=1), and cirrhotic liver tissues (n=7) were utilized for ex vivo evaluation of both ablation and drug delivery potential in diverse human tissues. Deidentified discarded tissue samples were obtained from the Mayo Clinic Arizona pathology department. The Mayo Clinic Institutional Review Board (IRB) determined that this activity does not constitute Human Subjects Research, in accordance with the Code of Federal Regulations, 45 CFR 46.102, and therefore does not require IRB review. To evaluate the potential of ablation and drug delivery, approximately 100 μL of OG containing 1 mg/mL Nivo and 0.25 mg/mL Dox was precisely injected into the core of each tissue mass using a 25 G needle. Following injection, the tissues were incubated within a humidified chamber at 37 °C for up to 24 hours. Ex vivo fluorescence scans were conducted to examine differences in ICG or Dox diffusion after injection, employing the IVIS SPECTRUM system from PerkinElmer Inc., Waltham, MA. Near-infrared illumination at the excitation wavelength of 750 nm was used to capture gross images of each tissue, while fluorescent emission at 850 nm visualized ICG. For Dox detection, an excitation wavelength of 460 nm and an emission wavelength of 560 nm were employed. Fluorescent images across various experimental specimens were acquired using consistent settings, including a 1-second exposure time (f/stop= 2), and were displayed using identical scales. The area of fluorescence enhancement was calculated by applying a standardized threshold value through the Living Image 4.7.3 Software. Subsequently, tissues were fixed in 10% buffered formalin and processed for histological analysis. Serial sections, stained with H&E or immunostained for Nivo detection, underwent morphometric analysis for nuclei count and calculation of the Nivo detection area. This analysis was facilitated using morphometric analysis software, specifically the Qupath 0.4.4 software.

4.2. Animal Studies

The in vivo procedures conducted in this study were approved by the Institutional Animal Care and Use Committee (IACUC) at the Mayo Clinic and were performed in accordance with the applicable regulations and guidelines (protocol numbers: A00003613, A00003528, and A00002551). The endpoints for our animal studies using tumor models were determined according to our institutional policies. The animals were euthanized or considered deceased just prior to reaching the suggested tumor volumes (equal to or greater than 10% of body weight) or upon manifesting ulceration as observed through ultrasound imaging or gross observations. These procedures comply with ARRIVE and ARAC guidelines. The determination of the number of animals used in the study was calculated based on a power analysis utilizing statistical software.

Intraparenchymal injection of OG into normal rat liver:

NS+ICG, OG+ICG, and OG+Dox+ICG were intraparenchymally injected into the livers of 18 anesthetized Sprague Dawley rats weighing 350–400 grams (Envigo, Placentia, CA). The procedures were performed through a laparotomy in a supine position on a warming platform. A cotton-tipped applicator dipped in sterile saline and blunt tweezers was used to position the left lower liver lobe for injections of three different NS or OG preparations at sites approximately 0.5 cm apart, using a 25-gauge syringe needle. Specifically, the medial site received a 50 μL injection of OG containing ICG (0.25 mg/mL) and Dox (1.25 mg/mL); the middle injection site received a 50 μL injection of OG containing ICG (0.25 mg/mL); and the lateral injection site received a 50 μL injection of NS containing ICG (0.25 mg/mL). Following the injection procedure, the subcutaneous and dermis layers were closed using 5–0 Vicryl sutures (Ethicon, Somerville, NJ). Subgroups of six rats survived for predetermined periods of 1, 4, or 28 days after injection. At each time point, the rats received an intravenous injection of the ExiTron 12000 contrast agent 4 hours prior to euthanasia. Subsequently, the rats were euthanized, and the liver tissues were explanted and fixed for ex vivo micro-CT imaging. The liver tissues were then transected to expose the core of the treatment zone at each location for ex vivo fluorescent imaging. This allowed for the measurement of the ablation area and the fluorescence intensity of ICG and Dox at each injection site using the IVIS 200 system. Following this, the transected liver sections were processed for histology.

MC38 tumor model of ectopic colorectal cancer in mice:

In this study, an MC38 tumor model of ectopic colorectal cancer was established in twelve-week-old female C57BL6/J mice (n=14, Jackson’s Laboratories) (IACUC approval: A00003613–18-R24). The mice were housed in a vivarium with a 12-hour light/dark cycle and provided ad libitum access to food and water. For tumor cell inoculation, the mice were anesthetized using continuous inhalation of isoflurane, and 2×10⁶ MC38 mouse colorectal adenocarcinoma cells suspended in 0.1 mL of Hank’s balanced salt solution were injected into the subcutaneous space on the right flank. Tumor size was monitored by measuring perpendicular diameters using ultrasound (US), and the tumor volume was calculated using the formula: 0.523 × (length × width × depth). When the tumor volume reached approximately 150 mm³, the tumor-bearing mice were randomly divided into two groups. One group received an intratumoral injection of OG containing 0.25 mg/mL ICG (n=7), with the injection volume calculated to be 1.25 times the tumor volume. Following the treatment, tumor volume was assessed twice a week using ultrasound. In adherence to the Institutional Animal Care and Use Committee (IACUC) regulations, mice were euthanized when tumors reached approximately 2 cm³ or exhibited subcutaneous ulcers whichever came first. Survival rates between the two groups were compared using Prism Software 9 to calculate log-rank survival. At the endpoint, mice were euthanized, and tumors were harvested for histological examination.

Histological evaluation:

The harvested tissues were fixed in 10% buffered formalin, sectioned to reveal the treatment zone, and embedded in paraffin. Sections were cut at 4 μm, mounted on glass slides (Fisher Scientific, #12-550-15, Pittsburgh, PA), baked at 60 °C for 30 minutes, deparaffinized, and rehydrated. Sections were stained with hematoxylin and eosin (H&E, 7111 and 7221, ThermoFisher Scientific) to visualize tissue morphology. For immune cell infiltration, consecutive sections underwent immunostaining as described previously.^[8] Rat N1S1 tumor sections were incubated with mouse anti-Rat CD3 IgG3κ (550295, BD, 1:50) and mouse MC38 tumor sections with rabbit anti-mouse CD3 IgG (Ab16669, Abcam, 1:100). For granulocytes, monocytes, and macrophages, sections were incubated with rabbit anti-CD68 IgG (ab125212, Abcam, 1:250).^[23] Goat Anti-Rabbit IgG H&L (HRP) (ab97051, Abcam, 1:500) was the secondary antibody. Anti-PD-1 (Nivo) was visualized using biotin-conjugated rabbit anti-human IgG4 (ab238617, Abcam, 5 μg/mL). Images were obtained at 200x magnification using an EVOS FL Auto-2 microscope. Morphometric analysis counted MPO+, CD3+, and CD68+ cells at each injection site at 1, 14, and 28 days post-injection, across twelve fields per specimen. Collagen score and fibrous capsule thickness were measured in Mason’s trichrome-stained sections (22-110-648, ThermoFisher Scientific). All analyses were performed using Qu Path software.

Flow Cytometry Analysis of T-Lymphocytes in MC38 tumors:

Female C57BL/6/J mice aged 6 to 8 weeks were anesthetized and injected subcutaneously with murine colorectal cancer cells (MC38) at a concentration of 1 × 10⁶ cells onto the right flank. When the tumor reached an established size of approximately 150 mm³, the mice were intratumorally injected with OG containing ICG, Dox, and Nivo. A parallel group of untreated mice with comparable-sized MC38 tumors served as the control. At 11–12 days after the intratumoral injection, the mice were euthanized, and tumor tissues were excised and minced into smaller pieces using scissors to create single-cell suspensions. This was achieved using a tumor cell isolation kit (130-096-730, Miltenyi Biotech, Bergisch Gladbach, Germany) following the manufacturer’s instructions. Cell suspension samples were incubated with ammonium-chloride-potassium (ACK) lysing buffer (KD Medical, RGC-3015) for 2 minutes to remove red blood cells. Subsequently, the samples were blocked with an anti-CD16/32 monoclonal antibody Fc blocker (553142, BD Biosciences, Franklin Lakes, New Jersey) for 10 minutes at 4°C. Aliquots of the cell suspension were stained using the Live/Dead viability reagent (65-0865-14, eBioscience, San Diego, California) and immunostained with 0.25 μg of the following fluorophore-conjugated mouse-specific antibodies mixture: PerCP/Cyanine5.5-conjugated anti-CD45 (103132, Biolegend, San Diego, CA), fluorescein isothiocyanate (FITC)-conjugated anti-CD3 (553062, BD Biosciences), and allophycocyanin (APC)-conjugated anti-CD8 (553035, BD Biosciences). The samples were further incubated for 60 minutes at 4 °C, protected from light. Following incubation, the samples were washed twice with phosphate-buffered saline (PBS) with 0.1% sodium azide and 1% bovine serum albumin (PAB buffer) and resuspended in 500 μL of fixation buffer (554655, BD Biosciences). Flow cytometry analysis was performed using an LSRII Fortessa (BD Biosciences), and the acquired data were analyzed using FlowJo software (BD Biosciences).

Mass cytometry analysis of MC38 tumor tissues:

To create single-cell suspensions, tumor tissues were minced in Roswell Park Memorial Institute (RPMI 1640) medium within a petri dish. They were then incubated with collagenase II and DNase I for 1 hour at 37 °C. The resulting cell mixture was filtered through a 40 μm strainer, and erythrocytes were lysed using an ACK lysis buffer. Cells were stained with cisplatin (Fluidigm, South San Francisco, CA), washed, and resuspended in Cell Staining Buffer (Fluidigm). Following this, cells were incubated with Fc Block (Biolegend) and an antibody cocktail designed to stain 34 cell surface molecules (refer to Table S2) for 45 minutes at room temperature. After washing, the cells were suspended and permeabilized in eBioscience Foxp3/Transcription Factor Staining Buffer Set (Thermo Fisher Scientific). They were then incubated with an antibody cocktail targeting 9 intracellular molecules (see Table S2) at room temperature for another 45 minutes. Subsequently, the cells were washed, fixed in 1.6% paraformaldehyde for 30 minutes, and resuspended and incubated in Maxpar Fix/Perm buffer containing Intercalator-Ir (Fluidigm) for 30 minutes at 4 °C. Before data acquisition, samples were spiked with equilibration beads (Fluidigm) for subsequent normalization. These samples were then analyzed on a Helios workstation (Fluidigm), generating FCS files. For data analysis, the integrated mass cytometry data analysis package Cytofkit (ver. 3.3) was employed. PhenoGraph analysis was used to identify cell subsets, and a t-distributed stochastic neighbor embedding (t-SNE) 2-dimensional map and a heat map were generated.

Proteomic analysis of MC38 tumor tissues using mass spectrometry:

Formalin-fixed MC38 tumor tissues were harvested from mice 10–12 days after intratumoral injection of OG or saline. The tissues were fixed in 10% buffered formalin for 24 hours, then dehydrated using increasing ethanol concentrations and embedded in paraffin, following our previously described method.^[30] Paraffin-embedded tissue sections were cut and then deparaffinized in xylene. Tissue lysis was performed in sodium dodecyl sulfate. Proteins were reduced, alkylated, and digested with sequencing grade-modified porcine trypsin (Promega, Madison, WI) via a filter-mediated process.^[31] A 60-minute gradient elution from 98:2 to 65:35 (buffer A: 0.1% formic acid, 0.5% acetonitrile; buffer B: 0.1% formic acid, 99.9% acetonitrile) was applied. Peptides were ionized by electrospray (2.4 kV) and analyzed on an Orbitrap Exploris 480 mass spectrometer (Thermo Fisher Scientific). MS data were acquired with a Fourier transform MS (FTMS) analyzer in profile mode at a resolution of 120,000 over a range of 375 to 1500 m/z. Higher energy collisional dissociation (HCD) activation was followed by MS/MS data acquisition in centroid mode at a resolution of 15,000 with a normalized collision energy of 30%. Protein identification was performed using MaxQuant (Max Planck Institute) with label-free quantification, a parent ion tolerance of 2.5 ppm, and a fragment ion tolerance of 20 ppm. Scaffold Q+S (Proteome Software) was utilized to validate MS/MS-based peptide and protein identifications, with proteins accepted if established with less than 1.0% false discovery rate and containing at least two peptides. Protein probabilities were assigned using the Protein Prophet algorithm.^[32]

Bioinformatics analysis of proteomic data:

Proteins were identified and quantified using MaxQuant (Max Planck Institute, version 2.0.3.0) against the UniprotKB Mus musculus plus MMR database (March 2021), with a parent ion tolerance of 3 ppm, a fragment ion tolerance of 0.5 Da, and a reporter ion tolerance of 0.003 Da. Scaffold Q+S (Proteome Software) was used to validate MS/MS-based peptide and protein identifications, accepting proteins with less than 1.0% false discovery rate and at least two identified peptides. Protein probabilities were determined using the Protein Prophet algorithm.^[32] MS1 iBAQ normalized intensity values were quality-assessed using ProteiNorm, with data normalization via cyclic loess. Statistical analysis was conducted using linear models for microarray data (limma) with empirical Bayes (eBayes) smoothing as previously described (PMID: 36457491). Proteins with an FDR-adjusted P value < 0.05 and a fold change > 2 were considered significant.

N1S1 rat model of hepatocellular carcinoma (HCC):

Male Sprague-Dawley rats weighing between 300 to 325 g were used to induce N1S1 HCC tumors, following established protocols (IACUC approval: A00003528–18-R23).^[33] N1S1 rat hepatoma cells (CRL-1604, ATCC) were cultured in IMDM with 10% heat-inactivated bovine calf serum. For inoculation, N1S1 cells were suspended in plain IMDM, yielding 100 μL aliquots with 2 × 10⁶ cells. Under anesthesia, the rat liver was exposed via midline laparotomy, and cells were subcapsularly injected into the left lower liver lobe using a 25-gauge needle. The wound was then sutured with a 5–0 Vicryl suture. Tumor growth was monitored with an ACUSON S2000 ultrasound system (Siemens Inc., Germany) using a multifrequency linear transducer (9L4, 9.0MHz). Ultrasound imaging in B-Mode was used to measure tumor diameters in the SI, LM, and AP planes, and volume was calculated using V = 4/3 × π × 1/2 × SI × 1/2 × LM × 1/2 × A.^{[33a, 34]} Once tumors reached ~0.15 cm³, rats received intratumoral injections of OG or NS with ICG (0.25 mg/mL) and Dox (1.25 mg/mL) at 125% of the tumor volume. This method is adapted from ethanol injection calculations for chemical ablation in humans.^[35] Rats were monitored for 2 weeks post-injection, and tumor volume was documented via ultrasound.

VX2 liver cancer model in rabbits:

A rabbit model of VX2 liver cancer was established following approved protocols (IACUC approval: A00002551–17-R22).^[36] VX2 tumor tissue slurry, preserved in liquid nitrogen, was thawed and suspended in 1 ml of DMEM (ThermoFisher Scientific). This mixture was injected into the calf muscle of a donor New Zealand white rabbit (Charles River Laboratories, Wilmington, MA) using a 16-gauge needle. After 2–3 weeks of tumor growth, the VX2 tumor tissue was harvested post-euthanasia, minced into 1 mm^3 pieces in cold DMEM, and surgically implanted into the left medial liver lobe of recipient rabbits using aseptic techniques. Hemostasis was achieved by compressing the liver incision with an absorbable gelatin sponge (Ethicon, Summerville, NJ) for 3 minutes. Tumor growth was confirmed via ultrasound, and 2 mL of OG containing 1.25 mg/mL Dox and 1 mg/mL Nivo was percutaneously injected into the tumors. Tumor dimensions in the SI, LM, and AP planes were measured, and volumes calculated using V = 4/3 × π × ½ SI × 1/2 LM × 1/2 AP. The injected volume was calculated to be 1.25 times the tumor volume.^{[33a, 34–36]} The injected volume was calculated to be 1.25 times the tumor volume.^[37] During percutaneous US-guided tumor injection, rabbits were positioned supine on a warming platform.^[38] Tumors were visualized using an ACUSON S2000 system and a multifrequency linear transducer (9L4; 9.0 MHz) in B-mode. The injection needle was inserted into the tumor, and the OG+ or NS+ material was administered slowly. Rabbits were euthanized at two weeks post-injection, followed by liver harvesting for fluorescence imaging, FTIR-mapping and histological analysis.

Blood collection and analysis in rabbits:

Whole blood samples were collected from both the OG and control groups through a needle puncture of the inferior vena cava using a 21-gauge needle before euthanasia. For complete blood counts, the blood was aliquoted into EDTA-vacuum tubes and analyzed. For serum chemistry analysis, blood was collected into Vacuette tubes (Greiner Bio-One, North American, Inc., Monroe, NC), allowed to clot for 30 minutes at room temperature, and centrifuged at 1,500 ×g for 5 minutes. Serum aliquots were stored and analyzed for the levels of alkaline phosphatase (ALP), alanine aminotransferase (ALT), creatinine (CRE), blood urea nitrogen (BUN), and glucose (Glu) using a DRI-CHEM 4000 analyzer (Heska, Loveland, CO).

4.3. Statistical analyses:

The data presented were expressed as mean ± s.e.m. and represent at least triplicate experiments unless stated otherwise. In vitro experiments were independently replicated at least three times unless otherwise noted. Statistical analyses were performed using two-tailed unpaired Student’s t-test and one-way ANOVA with Tukey’s or Dunnett’s multiple comparisons test, as appropriate. Survival analysis for mice in each group was conducted using the Kaplan–Meier method and compared using the two-sided log-rank (Mantel–Cox) test. A P value less than 0.05 was considered statistically significant. All statistical analyses were carried out using GraphPad Prism 9.