Preparation of x-ray-imageable and nonimageable poloxamer-based hydrogels
X-ray-imageable Poloxamer 407 (POL) (Sigma Aldrich, St. Louis, MO, USA) formulations were prepared according to the cold method previously described by Huang et al.27, with modifications to generate five concentrations of POL, 17, 18, 20, 21, and 22% (w/v) (POL17, POL18, POL20, POL21, and POL22, respectively), in normal saline (Quality Biological, Gaithersburg, MD, USA) with iodixanol (Visipaque 320 mg I/mL, GE Healthcare (Boston, MA, USA) added to a final concentration of 40 mg I/mL. Briefly, the respective amounts of poloxamer, iodixanol, and normal saline were added to a total volume of 50 mL. The solution was maintained at 4°C under magnetic stirring for at least 12 h. A similar procedure generated nonimageable POL formulations at the same poloxamer concentrations without iodixanol.
Selection of iodine concentration for CBCT and fluoroscopy imaging.
Serial dilutions of iodixanol in saline were prepared at 0.25, 0.5, 1, 2, 4, 8, 16, 32, 40, and 64 mg I/mL in 2 mL vials44. The samples were suspended inside a cylindrical water phantom to mimic x-ray attenuation in patients and imaged with CBCT and fluoroscopy (Allura Xper FD20, Philips, Best, the Netherlands)44. CBCT images were acquired with clinical imaging protocols at 120 kVp, 148 mA and 100 kVp, 184 mA. Cylindrical regions of interest (ROI) of 1mm radius and 8.265 mm height were used and segmented for each dilution in the CBCT, and attenuation in Hounsfield units (HU) was measured using open-source software (3D Slicer, URL https://www.slicer.org/). A linear regression analysis was performed to determine the relationship between CT numbers (HU) and mg I/mL, where a higher slope represents greater sensitivity to the detection of iodine45. The contrast-to-noise ratio (CNR) was calculated with a detectability threshold defined by the Rose criterion as CNR of 2.5 (Eq. 1)46 where µvial and µbackground are the HU values in the test solutions and background, respectively while σviaL and σbackground are the standard deviations. The iodine detectability threshold was used to determine the iodine concentration range that will be visible44,47. Fluoroscopic images of the vials were acquired using the same cylindrical phantom.
$$\left(\text{E}\text{q}. 1\right) \text{C}\text{N}\text{R}=\frac{{\mu }_{vial}-{\mu }_{\text{B}\text{a}\text{c}\text{k}\text{g}\text{r}\text{o}\text{u}\text{n}\text{d}}}{{\sigma }_{Background}}$$
Gelation times
The gelation times of the imageable and non-imageable gels were qualitatively estimated via the tube inversion method20,48. Briefly, 400 µl of the POL preparation at 4°C was added to 8mL glass vials (Duran Wheaton Kimble, Millville, NJ, USA). The material was allowed to equilibrate at room temperature for 5 min and then the tube was submerged in a 37°C water bath and periodically inverted. The time after water bath submersion at which the formed gel remained at the bottom of the glass vial was recorded.
Oscillatory Rheology
Gelation temperature and viscoelastic properties of the materials were characterized using a Discover HR20 rheometer (TA Instruments, New Castle, DE) equipped with a 25.0 mm stainless steel parallel plate geometry with 500 µm gap. To prevent drying artifacts, mineral oil (ASTM oil standard) was applied to the geometry and stage with all the samples prior analysis.
Gelation temperature. The effects of gel concentration (%, w/v) and iodixanol on the gelation temperature (Tgel) were determined. POL formulations (17, 18, and 22%, w/v) with and without 40 mg I/mL were subjected to a temperature ramp from 5 to 37°C according to experimental parameters adapted from Baloglu, et a49l. The samples were subjected to 0.2% strain and 6.0 rad/s angular frequency. Herein, we report the gelation temperature according to literature convention50,51. The gelation temperature is the midpoint storage modulus value taken from temperature-sweep rheological data that measures a free-flowing poloxamer colloid transitioning to the gel state. Tgel was also determined for POL22 containing 2, 5, and 10 mg/mL of DOX and iodine.
Viscoelastic properties. The viscoelastic properties and thixotropic behaviour of POL gel formulations (17,18, and 22%, w/v) with and without 40 mg I/mL were determined with a time sweep experiment monitoring the G’ and G’’ for 1 h at 0.2% strain and 6 rad/s at 37°C. Then, a 1000% oscillation strain was applied for 1 min followed by 60 min at 0.2% strain to evaluate the recovery of the materials. The percent recovery was calculated from the plateau G’ value before high shear strain and the recorded G’ value at 0, 10, and 56 min post-shearing. Frequency sweeps were performed at 0.2% oscillation strain from 0.1 to 100 rad/s, and amplitude sweeps at 6 rad/s were performed from 0.1 to 1000% oscillation strain to determine the linear viscoelastic regions and flow points of the materials. The flow point was calculated as the modulus crossover point (G’ = G”, tan δ = 1) with the software TRIOS (TA instruments) using the cubic/linear method. The same methodology was used to evaluate the viscoelastic behaviour of POL22 containing 40 mg I/mL iodine and 2, 5, and 10 mg/mL of DOX.
In vitro elution profiles of x-ray contrast agent and doxorubicin
In vitro iodine release kinetics of POL formulations (17, 18, and 22% (w/v)) with 40 mg/mL were obtained by filling a dialysis cassette (Pur-A-Lyzer Midi 3500, Sigma Aldrich) (3.5 kDa molecular weight cut-off) with the POL formulation and incubation in a shaker (Roto-Therm Plus, Ward’s Science, Rochester, NY, USA) under constant shaking (rocking mode, 10) at 37°C in 40 mL of normal saline. Aliquots were taken at 0, 1, 2, 3, 4, 5, 6, 7, 21, 45, and 69 h with volume replacement. Absorbance was measured at λ = 281 nm52 with a Cell Imaging Multimode Reader (Cytation 5 Bio Tek, Agilent) and compared to an iodixanol standard calibration curve. Following the same procedure, the DOX concentration from POL22 with 2, 5 and 10 mg/mL of drug was calculated from absorbance measurement at λ = 483 nm with a calibration curve.
Ex vivo gel injection
Injection needles. Three needle types were studied: an 18G, 10 cm needle with a single-end hole (single-end hole needle, SEHN, Chiba biopsy needle, Cook Regentec, Bloomington, IN, USA); a 19G, 7.5 cm needle with multiple holes on the side of the needle shaft (multiple side hole needle, MSHN, ProFusion Therapeutic Injection Needle, Cook Regentec, Bloomington, IN, USA); and, two variants of a device with an 18G needle capable of deploying three curved injection prongs at variable distances. The first could deploy a maximum of 2 cm (short tip, ST) while the second had a deployment range of up to 5 cm (regular). These are collectively referred to as the multi-pronged injection needle (MPIN, Quadra-Fuse, Rex Medical, Conshohocken, PA, USA).
Temperature control. Bovine livers were submerged in 10 L of 1X phosphate buffered saline at 37°C and allowed to equilibrate. The temperature was controlled by a calibrated Ink Bird Tech thermocouple (Ink Bird Tech Shenzhen, China) and a Ktopnob heating coil (Ktopnob Silver Spring, MD, USA). To ensure the tissue reached 37°C, two thermocouples were inserted at different positions within the liver as previously reported 53.
CBCT and fluoroscopic imageability and injectability assessment of poloxamer 407 gels. POL22 with 40 mg I/mL, was injected into an ex vivo bovine liver previously heated to 37°C under submersion in PBS 10x, 7.4 pH (Gibco Thermo Fisher Scientific, Waltham, MA, USA). Before each injection, a CBCT scan was conducted to locate potential injection sites in the liver that were relatively devoid of hepatic vessels. Once these sites were identified, the needle was connected to silicon tubing (Smith Medical ASD, Dublin, USA), and a syringe (Monoject, Dublin, OH, USA) which was preloaded with the gel and maintained on ice. A 12 mL syringe filled with gel was used for SEHN and MPIN, while a 6 mL high pressure syringe (Medallion, Merit Medical, South Jordan, UT, USA) was employed for the MPIN needles. Prior to attachment, the gel was advanced to the needle tip, after which the syringe was connected to a programmable injection pump (Harvard Apparatus PhD Ultra Syringe Pump, Holliston, MA, USA) prior to needle insertion.
Following insertion of the needle, its position within the liver was verified using CBCT. The threshold volume that could be injected without causing extravasation in the ex vivo livers was determined by evaluating injections of 4 mL, 8.6 mL, and 14 mL using the SEHN at an injection rate of 1000 mL/h. These target volumes correspond to the volume of spheres with diameters of 2 cm, 2.5 cm, and 3 cm, respectively.
Once the critical volume threshold preceding gel extravasation was identified, 4 mL injections at 10, 100, and 1000 mL/h were performed to examine the impact of injection parameters on the morphology of the gel deposit. A series of 4 mL control injections of saline (no gel) with 40 mg I/mL at 10 mL/h were also conducted. CBCT imaging was performed. The mean radiopacity and volume of the segmented iodinated injected gel were calculated. The percent of volume error was calculated using Eq. 2. For subsequent analysis, the gel morphologies were segmented and exported as stereolithography (STL) files.
$$\left(\text{E}\text{q}. 2\right) \text{%} \text{o}\text{f} \text{v}\text{o}\text{l}\text{u}\text{m}\text{e} \text{e}\text{r}\text{r}\text{o}\text{r}= \left|\frac{theoretical volume-experimental volume}{theoretical volume}\right|\times 100$$
Table 2 provides a summary of the ex vivo liver injection parameters investigated with CBCT imaging.
Table 2
Summary of injection parameters with different needle devices: single end hole needle (SEHN), multiple side hole needle (MSHN), and multi-pronged injection needle (MPIN).
Formulation | Needle type | Volume | Injection rate | Gauge |
Image-able Gel: POL22, 40 mg I/mL | SEHN | 4, 8.6, and 14 mL | 1000 mL/h | 18 |
SEHN | 4 mL | 10, 100, and 1000 mL/h | 18 |
MSHN | 10, 100, and 1000 mL/h | 19 |
MPIN (deployed 1 cm and 2 cm) | 10, 100, and 1000 mL/h | 18 |
No Gel: Normal saline, 40 mg I/mL | SEHN | 4 mL | 10 mL/h | 18 |
MSHN | 10 mL/h | 19 |
MPIN (deployed 1 cm and 2 cm) | 10 mL/h | 18 |
Morphology of injected gel. The sphericity and solidity of the gel injections, as outlined in Table 1, were calculated using 3D Slicer and Blender (Blender Foundation, Amsterdam, Netherlands). The equations employed to assess solidity and sphericity are as follows:
$$\left(\text{E}\text{q}. 3\right) \text{S}\text{o}\text{l}\text{i}\text{d}\text{i}\text{t}\text{y}= \frac{V}{Convex hull volume}$$
$$\left(\text{E}\text{q}. 4\right) Sphericity= \frac{{\pi }^{\frac{1}{3}}{\left(6V\right)}^{\frac{2}{3}}}{SA}$$
‘V’ represents the volume and ‘SA’ signifies the surface area for the sphericity calculation. 3D Slicer was used to determine the SA and V of the infused material. Blender software was utilized to calculate the convex hull volume, which is defined as the smallest convex volume set that encompasses the previously segmented volume of the gel injections, after importing the STL file to the Blender software.
Time course of gel Injection volume and distribution.
The spatiotemporal volumetric distribution of the gel was examined by injecting 4 mL of POL22 with 40 mg I/mL into an ex vivo liver at 10 mL/h. CBCT scans (100 kVp) were acquired at t = 0, and at 6, 12, 18, and 24 min during the injection, corresponding to 0, 1, 2, 3, and 4 mL of injected volume, respectively. The SEHN, MSHN, and MPIN devices were assessed, with the MPIN deployed at both 1 and 2 cm. Gel volumes were segmented at each time point and exported as a STL file. Afterwards, the volumes were displayed as 3D color-coded distributions per mL injected using MATLAB (R2020a version, Mathworks, Natick, MA).
Using the segmented volumes per mL injected, 2D cross-sections were created by importing the STL files to Meshmixer software (Autodesk Inc. (2019) Autodesk Meshmixer. http://www.meshmixer.com), generated at each timepoint for each needle device. For SEHN, using the cutting tool in Meshmixer, planes were generated longitudinal to the needle axis and the same procedure was applied for MSHN adding an orthogonal plane. For MPIN, an orthogonal plane was generated with respect to the tip of each of the three needles. Each 2D section was manually converted to grayscale using Paint software (Microsoft, Redmond, WA) for subsequent conversion to color contours with a customized code in MATLAB.
The sphericity (Eq. 3) and solidity (Eq. 4) for each injected volume were calculated. The circle-like shape (circularity) and determination of convex or concave shape (solidity) for the 2D cross-sections were also computed (MATLAB).
The centroid point [Pc = (xc, yc, zc)] of the 3D distribution after each milliliter injection was computed using Blender. The distance (d) between the centroid and the needle tip [Pn = (xn, yn, zn)] was calculated using the formula:
(Eq. 5) d (Pc, Pn) = \(\sqrt{({({x}_{n}-{x}_{c})}^{2}+{({y}_{n}-{y}_{c})}^{2}+{({z}_{n}-{z}_{c})}^{2} )}\)
MPIN injection technique variations.
The volumetric distribution and surface-area-to-volume (SA/V) ratio of MPIN injections were characterized for five variations in technique. All injections used POL22 with 40 mg I/mL at 100 mL/h.
-
Using a short tip needle, the prongs were deployed 1 cm, i.e., the distance between the tips was 1 cm (MPIN-1cm). After incremental injection of 2 mL, the prongs were advanced to 2 cm for the remaining volume of 2 mL, without rotation of the device.
-
Using a short tip needle, the prongs were deployed 2 cm (MPIN-2cm). During the injection of 4 mL, the prongs were retracted in four 0.5 cm increments from 2 cm to 0.5 cm, without rotation of the device.
-
Using a long tip needle (“regular”), the prongs were deployed 5 cm. During the injection of 4 mL, the prongs were retracted in five 1 cm increments from 5 cm to 1 cm. Approximately 0.8 mL were injected per deployment increment, without rotation of the device.
-
Using a long tip needle (“regular”), 4 mL were injected as defined in technique 3. The prongs were then withdrawn, the needle rotated 60°, and the prongs redeployed to 5 cm. Another 4 mL was then injected in the same manner for 8 mL total.
-
Using a long tip needle (“regular”), 2 mL were injected as defined in technique 3. The prongs were then withdrawn, the needle rotated 60°, and the prongs redeployed to 5 cm. Another 2 mL was then injected in the same manner for 4 ml total.
Gross pathology.
Liver specimens that included the imageable gel were excised utilizing fluoroscopy to localize the gel. Tissues were flash-frozen in 2-mercaptoethanol that had been previously cooled in liquid nitrogen, and subsequently stored at -80°C before further processing. The tissue samples were sectioned along the needle shaft axis using a 3D-printed mold equipped with cutting slots at 5 mm intervals54. These sections were then photographed and examined for any signs of tissue fractures. The lengths of the major and minor axes were subsequently measured to evaluate the gel distribution.
2D Drug dose distribution across three needle devices.
DOX distribution in tissue was determined after 4 mL x-ray imageable POL22 with 20 µg/mL of DOX was delivered with SEHN, MSHN, and MPIN-1 cm devices at 10 mL/h injection rates over 4 mL. In addition, with the same DOX dose, two MPIN techniques were performed as depicted in 2.7.7. The techniques were 2 and 5 without rotation at 100 mL/h with 4 mL and 2 mL respectively. The 5 mm tissue section (2.7.8) were imaged under an In vivo Imaging System (IVIS III, PerkinElmer, Whaltham, MA) (λex = 480 nm and λem = 620 nm) and CBCT (80 kVp). To determine DOX distribution profiles, the fluorescence intensity of DOX was determined along a line through the center of the injection using the Living Image software (PerkinElmer, Whaltham, MA). Drug concentration was estimated with fluorescence intensity compared to a calibration curve based on 5, 10, and 20 µg/mL of DOX in a black 96-well plate (Greiner Bio-one, Monroe, NC). The maximum DOX concentration was not expected to exceed 20 µg/mL. The average of CBCT and optical imaging circularities, calculated from MATLAB were divided to determine similarity.
Statistical analyses.
Statistical analyses were performed using GraphPad Prism 9 (GraphPad Software, Boston, MA, www.graphpad.com). One-way ANOVA complemented by Tukey's multiple comparison test was employed to compare gelation times, sphericities, and solidities. The t test was utilized to compare the sphericities and solidities of gel and non-gel specimens and to compare circularities of CBCT and optical images. Descriptive statistics were presented as mean ± standard deviation. Unless otherwise noted, all experiments were conducted in triplicate.