PEG-NRs were prepared from stock CTAB-coated gold nanorods to minimize toxicity and extend particle circulation time as previously described. (, left)16,17
Transmission electron microscopy (TEM) images revealed the PEG-NR dimensions to be 41 nm in length and 10 nm in width for an aspect ratio of 4.1 (, left). The surface plasmon resonance (SPR) peak absorbance of PEG-NRs was 808 nm in accordance with previously observed measurements (, left). Implanted NIR illumination devices were composed of silica rods and fiber optic meshes (, center). Silica rods were designed with varying geometric parameters such as diameter, length, and tip tapering, treated to have uniform surface diffusivity along the rod, and joined with an SMA fiber-optic connector to interface with a continuous 810 nm NIR laser source. To provide a flexible alternative to the silica rods, fiber optic meshes consisting of submillimeter fibers encased in adhesive medical-grade silicone were prepared. Fibers were treated by an optomechanical scoring process prior to encasement to alter the total internal reflection at distinct locations and thereby permit NIR light emission along the length of the fibers. For each device, fluence rate–distance relationships were computed for input currents ranging from 9.0 to 13.0 A (, center). The combination of systemic PEG-NRs and implanted NIR devices was investigated as a strategy to achieve photothermal heating in deep tumor nodules (, right).
Figure 1 First component for plasmonic photothermal heating of intraperitoneal tumors consists of polyethylene glycol-coated gold nanorods (PEG-NRs), which are characterized by TEM imaging and optical absorbance. (Left panel) PEG-NRs with an aspect ratio of 4.1 (more ...)
The abdominal cavity consists of organs and biological tissues with varying optical and geometric properties contained within a confined anatomical compartment. To develop a quantitative framework for understanding NIR illumination in the abdominal cavity and thereby assess the feasibility of the implanted NIR devices, a three-dimensional solid model incorporating the liver, spleen, stomach, kidney, large and small intestine, and skin was developed (Supporting Information
, Figure 1, a). For modeling simulations, abdominal organs were assigned approximate indices of refraction, scattering coefficients, and absorption coefficients based on values obtained from the literature (Supporting Information
, Table 1). Previous computational models developed in our lab applied finite element simulations to approximate photothermal heating of gold nanorod-containing subcutaneous tumors using the bioheat transfer equation.16
While the current model describes NIR light distribution in nontumor tissues within the abdominal cavity, future models incorporating tumor tissue elements, absorption coefficients dependent on gold nanorod tissue content, and biophysical heat transfer equations will be useful for providing a more detailed description of plasmonic photothermal heating in the abdominal cavity. The current deterministic model predicts that most of the emitted NIR light focuses on regions containing the intestine, liver, and spleen within the abdominal cavity with lambertian, or diffusely reflected, illumination along the silica rod (a). In comparison, with a lambertian profile along the flexible mesh, the model predicts an enhanced, more uniform transmittance within the abdominal cavity due to increased surface area relative to the silica rod (a). Importantly, simulations using this anatomically based model predicted the limitations of NIR light penetration due to tissue absorption and scattering. For example, at the interface of the silica rod and abdominal organs with an input power of 1 W along the device, the model predicts a peak fluence rate of 1.35 × 103
, while at a distance 8 mm deep to the device, the peak fluence rate decays to 5.9 × 102
(a). Similarly, at the interface of the flexible mesh and abdominal organs, the peak fluence rate is estimated to be 3.0 × 103
, and the peak fluence rate decays to 7.3 × 102
at a depth of 8 mm (a). The depth-dependent decay observed is consistent with tissue absorption and scattering and suggests that the model parameters approximate to a first order the in vivo
Figure 2 (a) Fluence rate simulation in the abdominal cavity with two implanted NIR devices at a length up to 8 mm depth from the skin surface. (b) Photographs of implanted NIR devices investigated include silica rods (devices 1–5) and mesh. (c) Fluence (more ...)
To validate and refine model predictions for the silica rods and flexible mesh, fluence rate was calculated at multiple distances and input currents for six distinct implanted NIR devices (b). Consistent with the model, fluence rate decreased approximately as the square of the distance between the device and detector plate when measured in an air medium (c). As the input current from the laser was increased, a linear increase in fluence rate was observed at the surface of each implanted device, highlighting the ability to precisely tune the NIR emission from the devices to achieve the desired level of photothermal heating in vivo (c). In contrast to the model predictions, the peak fluence rate measured for the flexible mesh surface (~500 W/m2) did not reach the maximal surface fluence rate of 3.0 × 103 W/m2 predicted by the model. Limitations related to the material properties and device fabrication likely contributed to the difference and suggest that future models could be refined by incorporating specific physical properties of the device in addition to tissue properties. Finally, surface fluence rate was highly dependent on the geometry of the device and spanned several orders of magnitude (4 × 103 W/m2 to 2 × 10–1 W/m2) (d).
On the basis of the surface fluence rate measurements, we next assessed whether our devices possessed differing capacities to photothermally heat PEG-NRs. In an ex vivo
setting, multiple concentrations of PEG-NRs in PBS were subjected to continuous NIR laser irradiation (810 nm) dispersed by each implanted device, and the temperature of PEG-NR solutions was continuously monitored via
infrared thermography (a,b). Several distinct patterns of photothermal heating were observed, which are denoted as (1) nonspecific, (2) PEG-NR-dependent, and (3) inefficient. Owing to elements of device construction that resulted in a greater localized concentration of laser energy at the device tip (device 3), nonspecific devices led to temperature elevations in both PEG-NR- and PBS control solutions (c,d, Supporting Information
, Figure 2). Conversely, devices with inefficient heating failed to yield an appreciable rise in temperature of PEG-NR solutions at approximately physiologic PEG-NR concentrations (c,d, Supporting Information
, Figure 2). A subset of implanted devices investigated, however, was observed to selectively elevate the temperature of PEG-NR solutions in a concentration-dependent manner while failing to significantly heat PBS control solutions (c,d, Supporting Information
, Figure 2). After 3 min of NIR irradiation, one PEG-NR-dependent heater (device 2) elevated a 100 μg/mL PEG-NR solution by 20.53 °C ± 0.68 °C, a 10 μg/mL PEG-NR solution by 11.90 °C ± 0.44 °C, a 1 μg/mL PEG-NR solution by 4.43 °C ± 0.47 °C, and a PBS control solution by 2.57 °C ± 0.79 °C. Collectively, this ex vivo
characterization helped identify a subset of implanted NIR light sources to evaluate in an in vivo
Figure 3 (a) Images of implanted NIR devices 2, 3, 5, and 6 (top to bottom). Infrared thermographic images of PEG-NR and PBS solutions after NIR exposure with implanted NIR devices. Scale bar: 1 cm. (b) Representative thermographic timecourse for device 2, illustrating (more ...)
The ex vivo fluence rate and PEG-NR photothermal heating measurements, coupled with computational models for two generic device geometries (i.e., silica rod and flexible mesh sheet) allowed for progressive down-selection of candidate devices prior to attempting plasmonic photothermal therapy in an animal model. Despite comparable fluence rate modeling predictions between the rod and mesh sheet, the various silica rod designs were pursued further because their material properties and construction proved able to sustain input powers in the anticipated range necessary for in vivo photothermal heating. The 6-fold lower measured surface fluence rate compared to the model prediction illustrated that incorporation of physical properties of the device would be advantageous for future simulations, and also that future iterations of the flexible mesh should include design parameters that can sustain the necessary power requirements for plasmonic photothermal heating.
Of the silica rod devices with the highest measured fluence rates, only a subset was able to heat PEG-NR solutions in a selective and dose-dependent manner. Those devices which failed to achieve PEG-NR-specific heating likely did so due to a combination of suboptimal geometric design and the incorporation of highly absorptive elements. Suboptimal geometries such as the long, narrow cylindrical design may have resulted in poorly diffused incoming NIR light with inhomogeneous emission to the surrounding environment. Such geometric constraints have been previously observed in the related field of laser-induced interstitial therapy, where NIR scattering from implanted laser applicators in part determines the volume of tissue susceptible to heating therapy.32,33
While the power required for this ablative therapeutic modality is substantially higher than that required for plasmonic photothermal therapy, the ability of the implanted device to scatter NIR light isotropically and minimize heat conduction at the tissue-device interface is highly dependent on geometric and material design parameters and is important for ensuring broad tissue coverage. Incorporating a metallic cap at the device tip led to heating of both PEG-NR and control solutions nonselectively, suggesting that the metallic cap absorbed the incident NIR light to generate heat within the device itself. Such designs are not suitable for implantation into the abdominal cavity due to reduced, inhomogeneous light emission, as well as local coagulation of tissues in the vicinity of the device tip.33
Limiting the incorporation of absorptive elements and improving light scattering from the device surface would help reduce the nonspecific heating observed here. The optimal implanted NIR device based on fluence rate and selective PEG-NR heating had a tapered tip design and a length of ~3 cm, which was among the shortest of the candidate devices investigated. Future iterations on silica- or fiber optic mesh-based implanted NIR devices will focus on improved flexibility, isotropic NIR emission from the device surface, and improved materials able to sustain a broader range of powers and thereby reduce the time required to reach the desired temperature range for plasmonic photothermal therapy.
The function of a silica rod from the PEG-NR-dependent activator group was next assessed in an orthotopic mouse model of ovarian cancer. The device was implanted into the abdominal cavity through an approximately 1 cm midline incision on the ventral surface and connected to the laser source via
fiber optic cable (a). To visualize the device inside the abdominal cavity, computed tomography (CT) imaging was performed. CT images clearly delineated the implanted device and revealed its placement relative to abdominal structures including intestine, abdominal wall, and vertebral bodies (a, Supporting Information
, Figure 3). To investigate the capability and specificity of the device to photothermally heat various tissues, the device was implanted into the abdominal cavity of animals pretreated with PEG-NRs or control animals lacking PEG-NRs, and the temperature of ovarian tumor tissue, intestine, and liver was continuously monitored with thermocouples placed at approximately equal distances (0.3–0.5 cm) from the implanted device within each of these tissues (Supporting Information
, Figure 4). Thermal profiling revealed a more efficient heating process in ovarian tumors, intestines, and livers of PEG-NR-treated animals relative to control animals lacking PEG-NRs (b–d). Modest temperature changes between 2.4 and 3.0 °C observed in tissues lacking PEG-NRs may be attributed to heat nonspecifically conducted from the device itself. In contrast, temperature elevations observed within the first minute of NIR exposure rose significantly steeper for tumors and liver in animals pretreated with PEG-NRs, while the trend was also increased but not significant for the intestine (e–g). For tissues in the PEG-NR group, the maximum temperature changes for tumor, liver, and intestine were 6.27 °C ± 2.47 °C, 6.51 ± 1.47 °C, and 4.66 ± 2.49 °C, respectively (h). Following PEG-NR photothermal heating, tissue sections were prepared to assess tissue viability and any resulting histological changes consistent with inflammatory or necrotic tissue damage. Importantly, normal tissue histology was observed for intestine, tumor, and liver by H&E staining (i). Furthermore, Ki67-positive immunohistochemical staining confirmed that these tissues remained actively proliferative after photothermal heating (i). Evidence of mild bleeding and coagulation at the device incision site, however, was noted in some animals investigated. These studies therefore establish that an implanted NIR device shown to selectively heat PEG-NR solutions in an ex vivo
setting can likewise heat tissues containing PEG-NRs deep within the abdominal cavity while not inducing irreversible damage in these organs.
Figure 4 (a) Placement of implanted NIR device in a tumor-bearing animal shown macroscopically, by axial CT scan (R, rod; I, intestine; V, vertebrate; dashed line outlining body cavity), and 3D-rendered CT scanning. (b–d) Thermal profiling in the presence (more ...)
Plasmonic photothermal therapy and other modalities to heat tumors in the subablative range (ca
. 41–45 °C) have been shown in subcutaneous models to alter the flow and permeability of tumor vessels and enhance accumulation of diffusion-limited therapeutic cargoes.34−38
To investigate whether NIR irradiation from the implanted silica rod similarly drove accumulation of therapeutic cargoes in an orthotopic site, tumors from NIR-treated and control animals were studied following administration of doxorubicin-loaded liposomes and the diagnostic imaging agent AngioSPARK750 (AS750; core diameter, 20–50 nm). We observed from prior studies that the lower third of the silica rod dispersed NIR light with high intensity, so this region was used to induce heating in tumors in this proof-of-principle study. Intraperitoneal tumors from the quadrant receiving NIR illumination and harvested 3 h after receiving AS750 and PEG-NR/implanted NIR heating demonstrated elevated macroscopic accumulation of AS750 (j). Significantly elevated enhancements of 3.27 and 11.2 for doxorubicin-loaded liposomes and AS750, respectively, were measured in homogenized tumor samples from the PEG-NR/implanted NIR group relative to control tumors from animals receiving no PEG-NR/implanted NIR therapy (k). These data suggest that plasmonic-based enhancement of therapeutic cargoes previously observed in subcutaneous tumor models35
may also have utility for delivering cargoes to tumors in the abdominal cavity.
For in vivo
thermographic studies, the liver and intestine were studied in addition to tumors for several reasons. First, previous work on the biodistribution of PEG-NRs has demonstrated a significant accumulation of PEG-NRs in the liver (~14% ID/g).16
Because of its high PEG-NR concentration, large size, and proximity to orthotopic ovarian tumor nodules, the liver poses a challenge for tumor-specific photothermal heating in the abdominal cavity. The temperature elevations observed in the liver were anticipated, and, importantly, they relaxed to baseline temperatures within several minutes after the procedures. The intestine, while accumulating fewer PEG-NRs than the liver, has been shown to be sensitive to broad elevations in temperature. This has been observed most notably during investigations with heated chemotherapy administered directly into the abdominal cavity for ovarian cancer patients.39
In this study, elevations in intestinal temperature were detected, but no tissue damage or proliferative defects were observed, and the thermal effects were more modest than in ovarian tumors or the liver, which is perhaps representative of differing PEG-NR concentrations within each of these tissues. A key advantage of PEG-NRs and plasmonic nanomaterials more generally is the ability to “sensitize” tumor tissue to incident NIR light relative to surrounding tissues containing fewer PEG-NRs. The ability to locally deposit heat in desired tissues using plasmonic nanomaterials stands in marked contrast to heated chemotherapy approaches which cause nonspecific temperature elevations throughout the entire abdominal cavity and have been associated with high morbidity and even mortality.39
In the future, strategies to both shield nontumor, PEG-NR-rich tissues (e.g., liver) from implanted NIR light as well as spatially localize thermal gradients around PEG-NRs via
pulsed laser sources could prove fruitful for mitigating any off-target effects.