Gold Nanostars Obviate Limitations to Laser Interstitial Thermal Therapy (LITT) for the Treatment of Intracranial Tumors

Tumors of the intracranial compartment, whether primary or metastatic, present therapeutic challenges that are both persistent and stark. Features such as the blood–brain barrier (BBB) limit access to systemic and immune-based therapies, and the normal brain's intolerance for injury from imprecise surgical or medical therapies make tumor specificity obligatory (1, 2). Although the intracranial confines are often treatment-limiting, the calvarium has proved permissive to fixed-trajectory, minimally invasive technologies for surgical cytoreduction. Laser interstitial thermal therapy (LITT) has emerged as a leading FDA-approved example of such technology with its validated use of stereotactic hyperthermal ablation in the treatment of primary and metastatic intracranial tumors, in addition to inflammatory after radiation treatment effect and epilepsy (3–12). To date, thousands of patients with brain tumor in the United States have undergone the LITT procedure, including more than 200 at our institution alone.

LITT avails of imaging-derived stereotactic guidance to direct the precise placement of a laser diode-tipped catheter within a brain lesion, followed by delivery of thermally ablative energy via either directional or full-firing laser probe (13). The procedure is performed under real-time MR-guidance, and MR phase change data are converted into real-time temperature change to guide the surgeon. Thermal damage and cell killing are calculated in real-time by a software package (example: M-Vision, Monteris Medical, Plymouth, MN) that uses a modified Arrhenius equation. Intraoperatively, the extent of thermal ablation is displayed by software as thermal-damage-threshold (TDT) lines, such as those depicted in Fig. 1A (14). In this example, the external yellow line designates tissue that has received thermal energy equivalent to 43°C for at least 2 minutes (no permanent damage), the middle blue line to 43°C for 10 minutes (irreversibly damaged), and the internal white line to 43°C for 60 minutes (coagulative necrosis; ref. 15).

Beyond the direct cytotoxic effects of thermal ablation, there is significant evidence that LITT sensitizes tumors to further treatment by both opening the BBB and triggering local and systemic anticancer immune responses (16–21). Despite its increasing clinical use, limitations are tied to: (i) the non-uniformity of specific heat across various intracranial tissues, leading to differential rates of conduction within the tumor; (ii) non-specific damage to surrounding normal white and gray matter; and (iii) regional heat sinks, such as blood vessels and cerebrospinal fluid spaces. These differences generally limit the size of candidate lesions to a 3-cm diameter and can lead to an ablated tissue volume that does not perfectly conform to tumor margins. Figure 1B illustrates this latter challenge in 4 selected patients, with the red outline, indicating the boundaries of the contrast-enhancing lesion and the blue TDT line delineating the area of irreversibly damaged tissue. Such non-conforming treatment errs toward either incomplete penetration of the targeted lesion or collateral damage to healthy tissues beyond its margins. Identifying strategies to increase the coverage range and lesion specificity of LITT, as well as to protect surrounding healthy structures, is vital to the evolution, safety, and broader applicability of this rapidly expanding treatment paradigm.

The use of gold nanoparticles has generated interest in diagnostic and therapeutic studies across a range of pathologies, including cancer and even autoimmune disorders (22–24). Tumors tend to have a high uptake of nanoparticles via the enhanced permeability and retention (EPR) effect due to the typically leaky vasculature characterizing tumor sites (25). Gold nanoparticles, such as gold nanoshells, have been used for photothermal ablation of prostate tumors in clinical trials without significant toxicities (26). Mechanistically, under incident electromagnetic field irradiation (i.e., with a 1,064 nm near-infrared laser), conduction electrons in metallic nanoparticles are displaced into an oscillation frequency equal to that of the incident light. These oscillating electrons are termed surface plasmons and produce a secondary electric field that adds to the incident light field to produce intense localized energy fields. These high-energy fields are concentrated at curvature points on the nanoparticles and can be exploited for photothermal tissue heating. Gold is a favored material due to its greater thermal conductivity (310 W/m/K) than the tissue in which the nanoparticles collect (brain tissue = 0.51 W/m/K). The nanostar structure is of particular interest because it possesses multiple sharp branches to proffer the numerous curvatures referenced above responsible for the “lightning rod” effect that strongly enhances the local electromagnetic field when subject to light stimulation (27, 28). Exciting gold nanostars (GNS) with near-infrared light (700–1,100 nm) is ideal for biological applications because the wavelengths are within the “tissue optical window” where tissue absorption is minimized, permitting photons to travel through the tissues to be captured and converted into heat by GNS accumulating within a tumor. Even small amounts of light that reach GNS accumulated in the tumor edges due to the EPR effect are amplified by the plasmons to produce heat to treat tumors. Healthy cells that do not contain GNS are not sufficiently heated (thus minimizing collateral damage). GNS therefore serve as exceptional light “nano-enhancers” to extend ablation zones where heat from laser alone would be insufficient to treat tumors.

Our team has pioneered GNS development using a novel surfactant-free synthesis method for safe and effective star-shaped nanoparticle generation (29). We herein present, then, a novel platform using plasmonics-active GNS as an effective means for both extending LITT lesional coverage and for more specifically sculpting thermal energy to lesion borders. Previous studies have shown that GNS both in vitro and in vivo can amplify the effects of extracorporeal laser-mediated photothermal ablation for extracranial tumors (27, 30, 31). Although these previous extracranial applications of GNS have been successful, the externally administered near-infrared radiation (NIR) used is incapable of penetrating the cranial compartment (32–34). Here, we pair GNS with a clinically used 1,064-nm LITT laser system (NeuroBlate) and use rationally designed recapitulative brain tumor phantoms in a clinical neurosurgical intraoperative MRI suite (IMRIS) to demonstrate that GNS amplify the thermal conductivity profile of LITT in a manner that conforms to GNS distribution, reduces procedure duration, and protects surrounding structures. Likewise, we demonstrate that GNS administered systemically will selectively accumulate within both heterotopic and orthotopic tumors using an in vivo murine model system, with corresponding amplification of tumor ablation. Thus, GNS represent a translatable technology that may be systemically administered to improve both the range and tumor specificity of the increasingly prevalent LITT platform.

All chemicals were purchased from Sigma-Aldrich and used directly without further purification. GNS particles were synthesized using the surfactant-free method developed by our laboratory (29). Au_15 GNS was synthesized with 0.1 nmol/L seeds, 30 μmol/L AgNO₃, 0.15 mmol/L HAuCl₄, and 0.3 mmol/L ascorbic acid. Au_25 GNS was synthesized with 0.1 nmol/L seeds, 30 μmol/L AgNO₃, 0.25 mmol/L HAuCl₄, and 0.5 mmol/L ascorbic acid. Au_35 GNS was synthesized with 0.1 nmol/L seeds, 30 μmol/L AgNO₃, 0.35 mmol/L HAuCl₄, and 0.7 mmol/L ascorbic acid. The synthesized GNS nanoparticles were coated with thiolated polyethylene glycol (Thiol-PEG, M.W. 6000) by incubating at room temperature for 24 hours to improve in vivo stability and circulation time. After functionalization with Thiol-PEG, GNS nanoparticles were condensed in PBS solution by centrifugation for following phantom and in vivo applications. Gold mass concentration was measured using inductively coupled plasma-mass spectroscopy (ICP-MS) with a Varian 820 mass spectrometer (Varian). Nanoparticles’ hydrodynamic size, polydispersity index, and zeta potential were measured in 1X PBS solution with a Nanosight NS 300 instrument and a Zetasizer Nanoseries instrument (Malvern Panalytical). Transmission electron microscopy (TEM) image was obtained using a Tecnai G² Twin (FEI) under 160 kV voltage. Vis-NIR extinction spectrum was measured with a Cary 6000i spectrometer (Agilent).

Female C57BL/6 (RRID: MGI:2159769) mice were used at 6–12 weeks of age. C57BL/6 mice were purchased from Charles River Laboratories; CT2A (RRID: CVCL_ZJ44) is a syngeneic murine glioma cell line on C57BL/6 background. B16F0 (RRID: CVCL_0604) is a syngeneic murine melanoma cell line on C57BL/6 background. The CT2A murine glioma model is considered to accurately represent several glioblastoma (GBM) characteristics, including intratumoral heterogeneity, in vivo migratory patterns, radio-resistance, chemo-resistance and different modes of immune dysfunction observed in GBM. Both cell lines have been authenticated using National Institute of Standards and Technology–published nine species-specific short tandem repeat markers to establish genetic profiles. Interspecies contamination check for human, mouse, rat, African green monkey, and Chinese hamster was also performed. The cell-working stocks have also been tested negative for Mycoplasma spp. and karyotyped. Neither is among the ICLAC database of commonly misidentified cell lines. The CellCheck Mouse Plus cell line authentication and Mycoplasma spp. testing services were provided by IDEXX Laboratories. Tumor cells were grown in vitro in DMEM with 2 mmol/L 1-glutamine and 4.5 mg /mL glucose (Thermo Fisher Scientific) containing 10% FBS (Gemini Bio-Products). Cells were harvested in the logarithmic growth phase and passaged once before use in the previously described experiments. For intracranial implantation, tumor cells in PBS were then mixed 1:1 with 3% methylcellulose and loaded into a 250-μL syringe (Hamilton). The needle was positioned 2 mm to the right of the bregma and 4 mm below the surface of the skull at the coronal suture using a stereotactic frame. Then, a total of 1 × 10⁴ CT2A or 1×10³ B16F0 tumor cells were delivered in a total volume of 5 μL per mouse. Animals were maintained under specific pathogen-free conditions at the Cancer Center Isolation Facility of Duke University Medical Center. All experimental procedures were approved by the Institutional Animal Care and Use Committee.

PET/CT scans were performed by using an Inveon small animal PET/CT instrument (Siemens) for four mice with CT2A intracranial tumors injected as above and one without CT2A intracranial tumor. GNS nanoparticles were labeled with ¹²⁴I by incubation at room temperature for 10 minutes and then purified by centrifugation wash. Two-percent isoflurane in oxygen was used to anesthetize mice for PET/CT scans. Images from a PET scan (5 minutes) and a CT scan (5 minutes) were acquired for mice immediately after intravenous administration of ¹²⁴I-radiolabled GNS nanoparticles (3.7 MBq) in 100 μL 1X PBS. Follow-up PET/CT scans were obtained at 24 and 72 hours after intravenous injection of ¹²⁴I-radiolabled GNS nanoparticles. PET/CT scan images were analyzed using Inveon Research Workplace software (Siemens).

Simulating the transport of the laser photons through tissue is important for characterizing the attenuation and penetration depth of the LITT system. Different tissue types, including the tumor and surround white and gray matter, have vastly different optical characteristics and will affect the effectiveness of the LITT system to ablate the intended targets. The photon propagation through the phantom was simulated using a 3D Monte Carlo photon propagation software (35). To match the experimental parameters of the LITT experiments, the optical properties of the tissue were set to values close to white and gray brain matter and light source as a single point source set at the interface between water and the tissue. A second simulation with 0.1-nmol/L concentration of GNS within the tissue is created to show the increased absorption of photons around a volume of GNS. The laser source was set to be a point source centered at the interface of water and brain tissue phantom to model the contact between the optical fiber of the LITT system and the brain tissue of the LITT procedure.

Optical phantoms were created to simulate the diffusion of light as it travels through tissue. An agarose-based gel (2–3 w/v%) was chosen as it can serve as a solid scaffold for the nanoparticles as well as mimic the heat transfer in tissue due to its high water composition. For the initial LITT-heating experiments, we produced a 12×12×10 cm solid gel phantom containing a smaller 2.5-cm gel cylinder infused with GNS at a 0.1 nmol/L concentration to simulate tumor within normal tissue. An identical phantom without GNS was also produced and a third model was generated with a cylindrical shape (radius of 2 cm) containing GNS in half of the phantom. All are shown in figures below and were maintained at room temperature immediately before LITT administration. The experiments were conducted in the clinical MRI suite at Duke University Medical Center equipped with the LITT system described in the following sections.

The NeuroBlate System (Monteris Medical Corporation) was used for all MRI-guided LITT phantom procedures. With this system, heating is provided via a CO₂, gas-cooled side-firing (directional) 1,064-nm diode-pulsed laser probe with a sapphire tip and thermo-couple to thermally ablate target tissue in situ agar phantom with or without the addition of GNS. A 3.2-mm diameter probe measuring 10–30 cm in length was used for each experiment. Heating information was measured using M-Vision, the accompanied proprietary software for planning, executing, and measuring the controlled heating with the NeuroBlate System. This software was used to determine the time and temperature of randomly selected “pick points” throughout the ablative field for both groups. As the MRI system measures relative temperature changes, temperatures are reported as the change from baseline. Temperatures were measured and recorded every 7 seconds and the laser was activated for a total time of at least 12 minutes for each experiment (15).

An IMRIS intra-operative MRI system with a 3.0 Tesla Siemens (Erlangen, Germany) magnet was used for imaging. For LITT, a volumetric rapid gradient-echo (MP RAGE) T1-weighted sequence was used throughout the experiment. During treatment, the M-Vision software displays three, 5-mm-thick MR slices that are perpendicular to the laser probe trajectory, including the current treatment slice, one slice deeper, and one slice more superficial with no gap between slices, providing an overall visual coverage of 15 mm in thickness perpendicular to the probe (36). This provides an estimate of thermal expansion in the 3D volume. The software also displays a single coronal image and a single sagittal image, which are updated in real-time throughout the procedure to show the cumulative treatment effect as heating progresses. After the probe is set to the chosen starting depth, quantitative MRI-based temperature mapping based on the proton resonant frequency shift sequences (i.e., MR thermography) are started. At least eight cycles (each lasting 7 seconds) of baseline scanning were completed before firing the laser per the NeuroBlate System protocol.

Intracranial tumors in the mouse model were produced as above with either B16F0 or CT2A tumor cell lines. After 18 days for CT2A and 11 days for B16F0, mice were administered 100 μL injection of 20 mg/kg GNS solution or 100 μL sterile PBS through the tail vein. Mice were then sacrificed and perfused with 4% paraformaldehyde in PBS at 10 minutes, 24 hours, and 72 hours after injection. Mouse brains were harvested for histopathology and two-photon imaging. For microscopic analyses, brains were fixed in 4% paraformaldehyde for 24 hours at 4°C, rinsed with water, then submerged in 70% ethanol for at least 24 hours. Brains were then paraffin-embedded and sectioned. Slides from each timepoint were stained with hematoxylin and eosin (H&E) or 4′,6-diamidino-2-phenylindole (DAPI; Prolong Gold, Invitrogen, P3693). H&E-stained slides were imaged at ×10 magnifications using a BZ-X800 fluorescence microscope (Keyence) and merged into whole brain images using BZ-X800 Analyzer software. DAPI-stained sections of tumor and adjacent normal brain were imaged using a SP8 DIVE multiphoton microscope (Leica Microsystems; Duke Light Microscopy Core Facility) to identify the presence of GNS. For mass spectrometry, additional brain samples were collected at 24 hours after GNS injection and dissected into tumor and surrounding normal tissue samples. Each sample was digested with aqua regia and Au mass was measured with ICP-MS with a Varian 820 mass spectrometer (Varian).

Once CT2A flank tumors had reached a size between 1 and 1.5 cm, animals were randomly placed into the GNS group and control groups. All animals in the GNS group received a 100 μL injection of 20 mg/kg GNS solution 24 hours before LITT treatment. All mice in the control group received a 100 μL injection of sterile PBS 24 hours before LITT treatment. A 1,064-nm laser (Laserglow Technologies) was used to deliver 0.4 W of energy via a 400-μm optical fiber (Monteris Medical Corporation), and the power was verified by a power meter (Thorlabs). The tip of the optical fiber was inserted 2–3-mm deep into the tumor of each animal. To monitor the temperature change over time, hypodermic K-type thermocouples and a TC-08 Omega thermocouple reader were used in conjunction with the Omega logging software. Thermocouples were placed 2 and 4 mm away from the laser fiber to record the change in temperature throughout LITT treatment as a function of time and distance from the power source. For the continuous heating study, 0.4 W of power was delivered via optical fiber for 10 minutes, and subsequent temperature changes were monitored. The laser was triggered on and off every 30 seconds for the ramped heating study. At the conclusion, mice were sacrificed and tumors were harvested for gold concentration measurement with ICP-MS as described above.

The data in this study were generated by the authors and are available upon request from the corresponding author.

GNS can be synthesized in a controlled fashion that permits exploitation as a photothermal “adjuvant” for LITT excitation within the organic tissue optical window. This optical window typically ranges from 600 to 1,300 nm, within which most tissues are sufficiently weak absorbers as to permit significant penetration and scattering of light. We therefore developed GNS that can produce maximal photothermal effects at the 1,064-nm laser wavelength used by the FDA-approved clinical NeuroBlate LITT system.

Importantly, our GNS synthesis involves a novel surfactant-free method that does not require toxic cetyltrimethylammonium bromide (29). In short, 12-nm gold sphere nanoparticles were synthesized by reducing chloroauric acid (HAuCl₄) with trisodium citrate seeds, which were then rapidly mixed with silver nitrate (AgNO₃), ascorbic acid, and HAuCl₄. The ratio between seeds and HAuCl₄ was varied and tuned to achieve maximal absorption near 1,064 nm. Three ratios were synthesized as described previously in the Materials and Methods section: Au_15, Au_25, and Au_35. The PEG-functionalized GNS nanoparticles were condensed and the gold mass concentration measured with ICP-MS. The hydrodynamic size, polydispersity index, and zeta potential for each were measured as follows: Au_15 (70 nm, 0.25, −11.2 mV), Au_25 (92 nm, 0.23, −12.2 mV), Au_35 (99 nm, 0.24, −13.9 mV). Representative TEM of the GNS is shown in Fig. 1C. The extinction spectra for Au_15, Au_25, and Au_35 GNS are shown in Fig. 1D. The Au_35 GNS were demonstrated to have the highest extinction near 1,064 nm and therefore selected for subsequent testing as the primary GNS.

Theoretical and numerical models of the photon migration and temperature evolution were generated using Monte Carlo simulations. In brief, the Monte Carlo Modeling of Photon Transport simulates randomly scattered and absorbed photons as they travel in optical media of specific absorption and scattering properties. The photons are launched and tracked as they deposit energy at different spatial points. A photon fluence map is produced describing the concentration of photons, which is subsequently converted to an absorption map correlated with energy and heat distribution inside the material.

The representation of brain matter was generated as a homogenous layer of specified optical properties at 1,064-nm excitation. Gray matter was designated to have an absorption coefficient of 0.19 cm⁻¹, scattering coefficient of 26.7 cm⁻¹, and anisotropy value g of 0.96. The control tumor was modeled as a spherical tissue volume as shown in Fig. 1E (top). A 1 W isotropic point photon source placed within the tumor model corresponded to the typical laser irradiation of a fiberoptic probe. An experimental tumor volume representing GNS added within the tissue to a radius 20 mm was modeled as shown in Fig. 1E (bottom). Accumulated GNS were simulated at a concentration of 0.1 nmol/L, equivalent to an attenuation coefficient of approximately 1.154 cm⁻¹, to match the experimentally observed in vivo concentration of 20 μg/g. To simplify modeling, the attenuation coefficient was taken as the absorption coefficient added to the underlying tissue. Each simulation ran for 30 minutes and evaluated 20 million generated photons.

The results of the Monte Carlo simulations are shown in Fig. 1. In Fig. 1F and G, the absorption values are displayed in a log normalized map and cross-section profile demonstrating the diffusion of heat radially from the point laser source without the addition of GNS. Figure 1H and I characterize the heat diffusion with the addition of GNS, depicting a highly concentrated absorption pattern centered tightly around the laser source. When GNS are added as an additional absorptive element in a sphere around the laser source, most of the energy is absorbed by the tissue containing the GNS. On the other hand, the simulation that does not contain GNS results in a more diffuse distribution of energy.

The current clinical applications of LITT are limited by moderate thermal damage range and lack of specificity at tumor margins. To characterize the impact of GNS on the thermal range, efficiency, and lesion conformation of a clinically used 1,064-nm LITT system, we prepared agarose brain tumor phantoms with and without GNS, as diagrammed in Fig. 2A. These phantoms were then subjected to LITT using our clinically employed NeuroBlate system and intra-operative MRI suite. Temperatures were measured at a point 2 cm from the laser tip, with representative thermal monitoring images shown in Fig. 2B after 12 minutes of heating. The GNS-infused phantom demonstrated a grossly apparent increased rate of heating. Likewise, linear regression models fitted to the dataset showed that the rate of temperature increase for the GNS phantom was nearly 5.5 times greater than that of the control (Fig. 2C). The regression lines both demonstrated a high degree of accuracy, with R² values of 0.9743 for the GNS phantom and 0.9152 for the control phantom, respectively.

We next tested whether GNS distribution can influence the shape and pattern of heating across the phantom model system to improve lesion conformation. The experiment above was repeated with an hourglass-shaped GNS phantom. Representative images from the Monteris M-Vision MRI-thermometry software comparing thermal distribution in the cylindrical and hourglass-shaped phantoms are shown in Fig. 2D. The heat conducted most rapidly and efficiently along the distribution pattern of the GNS within the tumor model, conforming to the GNS-containing lesion borders.

Next, we examined whether GNS would impact the thermal exposure of bordering “normal” tissues (not containing GNS) during the administration of LITT. Another tumor phantom was produced with an internal cylinder in which only one side of the cylinder contained GNS, as shown in Fig. 2E (“split” phantom model). LITT was administered per the methods described above, with temperatures monitored at the borders of the split tumor phantom, 2 cm from the laser probe (GNS/Ctrl_{2 cm}), and at a point 0.5 cm beyond the lesion border (2.5 cm total from the laser probe GNS/Ctrl_{2.5 cm}) in an area representing “normal” surrounding tissue. Figure 2F depicts a representative image of the experiment from the M-Vision MRI-thermometry software.

The results of this experiment are shown in Fig. 2G and H, with the GNS-containing half demonstrating accelerated heating in line with previous trials. Simple linear regression models were fitted to each set of data points and plotted with equations as shown. The lines demonstrated a high degree of accuracy, with R² values for G_{2 cm} of 0.9949, G_{2.5 cm} 0.9641, C_{2 cm} 0.9965, and C_{2.5 cm} 0.9518. These regressions were used to interpolate and extrapolate heating times and temperatures for various target temperature changes at the tumor border, as shown in Table 1. At each target tumor border temperature change, the GNS-infused portion of the phantom required less time and energy, and therefore exposed surrounding structures to lower levels of heating. For example, to increase the temperature +8°C at the phantom border, the GNS-infused side required 14.18 minutes of heating compared with 27.41 minutes in the control, and the point 0.5 cm beyond the border experienced a temperature change of +1.7°C and +2.4°C, respectively. These differences indicate a protective effect of GNS on surrounding structures by reducing their exposure to potentially damaging thermal energy, likely a result of the lower levels of photothermal energy needed to raise the temperature of GNS-containing regions and reduced time for energy transfer to occur during the rise to maximum temperature.

The capacity for GNS to enhance the extent and specificity of LITT coverage for tumors in which they are present is only relevant if the GNS can be made to selectively collect within those tumors. To test whether GNS might be subject to EPR, we used murine models of intracranial CT2A glioma and B16F0 melanoma to represent primary and metastatic intracranial tumors, respectively. Both tumor models exhibit disruptions to the BBB that permit contrast agents to collect within the tumor when injected, and we hypothesized that the size of GNS might permit them to behave in the same manner, collecting selectively within tumor when in the intracranial compartment.

First, purified ¹²⁴I radiolabeled GNS nanoparticles were systemically administered into 4 CT2A tumor-bearing (TB-A, TB-B, TB-C, and TB-D) and 1 non–tumor-bearing (NTB) mice via tail vein injection. PET/CT scan was performed 10 minutes, 24 hours, and 72 hours after the ¹²⁴I-GNS intravenous injection and the imaging results (coronal slices) are shown in Fig. 3A. There was no uptake of GNS within the normal brain of any mouse at the 10-minute timepoint and all TB mice demonstrated tumor-selective uptake at 24 and 72 hours. The NTB mouse did not demonstrate intracranial GNS uptake at any timepoint by PET/CT.

To examine generalizability across primary and metastatic models, mice were also implanted with intracranial CT2A or B16F0 and either GNS nanoparticles or sterile PBS systemically administered via tail vein injection. GNS were administered on day 18 for CT2A and day 11 for B16F0 to accommodate differences in tumor growth rates. Mice were sacrificed at 10 minutes, 24 hours, or 72 hours after GNS injection and brains harvested, preserved, and stained with H&E and DAPI for further analysis. Representative H&E staining for the CT2A and B16F0 tumors are shown in Fig. 3B. Two-photon photoluminescence imaging of the DAPI-stained samples was conducted to visualize GNS localization within both tumor and normal brain samples, with results as shown in Fig. 3C and D. Both tumor and normal brain tissues demonstrated the presence of some GNS at the 10-minute timepoint, although at 24 and 72 hours, GNS were visualized within tumors, exclusively.

This experiment was then repeated with mice sacrificed preferentially at 24 hours. Tumor and surrounding brain samples were harvested and underwent ICP-MS to quantify the concentration of Au in each, as shown in Table 2. The unit depicted (%ID/g) is defined as the percentage of the injected dose per gram of tissue. The concentration of GNS in tumor ranged between 22 and 52 times greater than that found in normal brain.

The above results highlight the ability of GNS to expand LITT coverage and specificity in life-sized brain tumor phantoms using a clinical LITT system, as well as the ability of GNS to accumulate within brain tumors in vivo when administered systemically. To evaluate the impact of GNS on LITT-mediated tumor ablations in vivo, we used a heterotopic flank CT2A murine glioma model. As the clinical 1,064-nm laser cannot be adapted for use in mice, a miniature 1,064-nm laser fiber (400-μm diameter) was used (Monteris Medical), and thermal measurements were recorded using hypodermic K-type thermocouples and a TC-08 Omega thermocouple reader. Mice with tumors >1 cm diameter were obtained and GNS nanoparticles administered intravenously via tail vein injection whereas control tumor-bearing mice received PBS. The presence of GNS was confirmed by ICP-MS with a measured uptake of 4.1%ID/g. Twenty-four hours after injection, the laser fiber was inserted into the center of the tumor, 3 mm below the skin, with thermocouples placed 2 and 4 mm away from the laser fiber (within the tumor) to collect temperature measurements (Fig. 4A). All animals were treated with 400 mW for 10 minutes. Figure 4B shows representative heating versus time graphs for one animal from the GNS group and one animal from the control group over 10 minutes. Figure 4C shows the highest recorded temperature at 2 and 4 mm away from the laser fiber from all animals in both groups. Animals that had received GNS injection before photothermal therapy reached average temperatures of 58.0°C ± 5.5°C and 49.3°C ± 4.6°C at 2 and 4 mm, whereas animals in the control group reached 38.8°C ± 2.0°C and 37.2°C ± 2.5°C. At both distances from the laser fiber, animals that received GNS before treatment had significantly higher temperatures compared with the control group. Figure 4D compares heating curves between animals from each group when the laser was alternated on and off every 30 seconds. The average heating rates during that 30-second interval in the GNS group were 0.48°C ± 0.04°C per second and 0.17°C ± 0.02°C per second at 2 and 4 mm, respectively. In the control group, the heating rates during the same intervals were 0.07°C ± 0.004°C per second and 0.03°C ± 0.004°C per second at 2 and 4 mm. These results corroborated the Monte Carlo and tumor phantom data above, highlighting the dramatic amplification of LITT ablation by systemically administered GNS particles in vivo.

These results form a foundation for further translation of GNS as a means of optimizing LITT in the treatment of intracranial tumors. We have demonstrated preferential accumulation and retention of GNS within intracranial primary and metastatic tumors in murine models, as well as in vivo ablation amplification in a heterotopic murine glioma model. These data, combined with our mathematical and simulative evidence for the amplification of LITT photothermal propagation by GNS within tumor, present an opportunity to significantly enhance this treatment paradigm. Building off the results in our models and others is a valuable path forward to optimize the rising use of LITT in patients with intracranial tumors, portending potential applications more broadly across the field of neuro-oncology.

Administration of GNS by tail vein injection in mice with modeled primary and metastatic intracranial tumors resulted in a passive tumor–targeting process, as shown by the PET/CT and two-photon photoluminescence images in Fig. 3. These images demonstrate that at the 24- and 72-hour timepoints, the GNS had selectively accumulated within tumors relative to surrounding normal brain. The GNS signal in normal brain at 10 minutes on two-photon photoluminescence likely represents circulating GNS within the microvasculature of the tissue, as the GNS have disappeared by the 24-hour timepoint. The ICP-MS data in Table 2 quantify the uptake of GNS into tumor tissue at 24 hours and demonstrate a concentration roughly 22 to 52 times greater than that of surrounding normal brain. This suggests the 24-hour timepoint following GNS administration as an optimal time to conduct the LITT procedure, although further studies are needed. These results are also consistent with published literature on the accumulation of GNS within tumors, as previous work has demonstrated that this phenomenon is a function of increased vascular permeability within pathologic tissue (37). This effect is particularly relevant in the case of intracranial tumors, as the protective BBB often exhibits breakdown around the tumor, creating a mechanism for further selectivity within the intracranial compartments (38). Earlier studies by our group have demonstrated that remaining GNS are non-toxic and predominantly cleared by macrophages in the spleen and liver over time (27).

Importantly, pairing GNS with LITT creates a multiplicative photothermal effect, as shown in the GNS phantom experiments above using the clinical LITT equipment. Importantly, no charring or gas formation events were noted in any of the experimental trials. We demonstrated that the addition of GNS increased the rate of temperature rise by more than 500% at a point within the tumor phantom 2 cm from the laser probe tip. Accelerating the rate of tumor heating decreases the total operative time necessary for the procedure and limits the amount of thermal energy required for ablation, thus reducing the likelihood of off-tumor thermal damage. This is further demonstrated in the experiment with the hourglass shaped phantom, in which heat propagation was sculpted by the distribution pattern of GNS. Likewise, the split phantom studies also demonstrated that GNS reduce the heat exposure of peripheral normal structures. At each given temperature along the tumor border, the surrounding structures were heated to a lesser degree in the GNS-containing phantoms than they were in controls. These differences are thought to be a result of the lower levels of photothermal energy needed to raise the temperature of GNS-containing regions, GNS release of heat (flash) over only very short distances, and the reduced time for energy transfer to occur during the rise to the target maximum temperature. Therefore, the GNS permit LITT treatments to become faster, more tumor-specific, and safer.

Because of current restrictions for adapting the experimental laser and temperature measurement equipment to an intracranial murine tumor model, a heterotopic model with flank murine glioma tumors was used for in vivo validation. The gathered data demonstrated greater and more rapid heating of the target tumor at distances of 2 and 4 mm from the laser source after the administration of GNS. The control group did not experience a temperature rise that would cause tissue injury, whereas all animals that received GNS injection had an average maximum temperature several degrees over 43°C at the same power. These results support the theory that LITT can selectively ablate tumor tissues that have passively accumulated GNS particles with powers that would not harm healthy tissue. We believe that combining these technologies will extend the range of potential tumor ablation while limiting the damage to surrounding structures, due to the increased specificity and decreased power requirement.

Although other studies have examined the use of gold nanoparticles as photothermal transducers in tumors, these studies have used extracorporeal laser sources (27, 31, 39–43). Such approaches are susceptible to scattering and absorption at each tissue interface from the skin surface to potentially deep-seated tumors, with significant limitations on total penetration (32–34). Studies in cadaveric models have shown that the penetration of such laser sources is limited to more surface-level targets within the cranium (32). The integration of interstitially applied LITT permits targeting of tumors at any location within the brain, including deep-seated tumors that are inaccessible to surgical resection. To this end, our study presents early data supporting the combination of LITT and gold nanoparticles in the treatment of intracranial tumors. Combining the high level of precision and intracranial access permitted by LITT with the specificity and multiplicative effect of GNS is a novel strategy with promising results to support further investigation.

This study also presents some limitations. Our murine model does not perfectly re-create spontaneous glioma or metastatic spread, and the intervention of placing tumor cells within the CNS may disrupt the BBB in such a way that the accumulation of GNS is impacted. In addition, our in vivo models of laser ablation were performed on flank rather than intracranial tumors due to scale constraints; however, the thermal dynamics are unlikely to differ substantially within the tumor. Validation of these results in larger animal tumor models with clinical LITT equipment is a necessary next step to bringing this concept to real-world utility. Regarding such future in vivo experiments, our group has recently collaborated on a system to translate LITT for use in canines with gliomas. To do so, adapted cranial mini-bolts, fixation equipment, and navigation software were developed with promising early results in cadaveric and live-animal validation testing. The ongoing progression of this system will both permit effective treatment of canine patients as well as produce a novel model for further preclinical evaluation of GNS with LITT.

In conclusion, our study indicates a promising direction for evolving the current approach to nanoparticle-mediated photothermal therapy. By pairing the tumor specificity and thermal conductivity of GNS with the highly accurate stereotactic placement of LITT, the strengths of each approach are amplified to create a more effective and safe treatment model. Further investigation of this strategy with in vivo experiments will allow researchers to optimize parameters to validate this approach and push forward translational opportunities in this domain of primary and metastatic brain tumor therapy.

R.A. Odion reports a patent for Nanoplasmonics-Enhanced LITT Systems and Methods Thereof pending to T. Vo-Dinh, P. Fecci, Y. Liu, P. Chongsathidkiet, R. Odion. P. Chongsathidkiet reports a patent for U.S. Provisional Patent Application No. 63/355,711 pending. C.L. Mariani reports other support from Monteris Medical outside the submitted work. T. Vo-Dinh reports a patent for Nanoplasmonics-Enhanced LITT Systems and Methods Thereof pending. P.E. Fecci reports grants from Cancer Research Institute during the conduct of the study. No disclosures were reported by the other authors.

E.S. Srinivasan: Conceptualization, resources, software, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. Y. Liu: Conceptualization, resources, data curation, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. R.A. Odion: Conceptualization, resources, software, formal analysis, supervision, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. P. Chongsathidkiet: Conceptualization, formal analysis, supervision, validation, investigation, methodology, writing–original draft, writing–review and editing. L.P. Wachsmuth: Investigation, methodology, writing–review and editing. A.P. Haskell-Mendoza: Investigation, methodology, writing–review and editing. R.M. Edwards: Formal analysis, investigation, visualization, methodology, writing–review and editing. A.J. Canning: Software, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. G. Willoughby: Resources, supervision, validation, investigation, writing–review and editing. J. Hinton: Conceptualization, resources, data curation, supervision, investigation, writing–review and editing. S.J. Norton: Conceptualization, resources, data curation, supervision, validation, methodology, writing–review and editing. C.D. Lascola: Conceptualization, resources, data curation, supervision, validation, investigation, project administration, writing–review and editing. P.F. Maccarini: Conceptualization, resources, supervision. C.L. Mariani: Conceptualization, methodology, writing–review and editing. T. Vo-Dinh: Conceptualization, resources, supervision, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. P.E. Fecci: Conceptualization, resources, data curation, supervision, funding acquisition, validation, investigation, visualization, methodology, project administration, writing–review and editing.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.