Development of enhanced ethanol ablation as an alternative to surgery in treatment of superficial solid tumors

Ethyl Cellulose-Ethanol Solution Preparation and Physical Characterization

Mixtures of ethyl cellulose (Sigma Aldrich, St. Louis, MO)-ethanol (200 proof, Koptec, King of Prussia, PA) and ethanol were prepared by stirring at room temperature. Ethyl cellulose concentration ranged from 0–6% (ethyl cellulose to ethanol, weight:weight). Concentrations greater than 6% ethyl cellulose were not evaluated because they did not fully dissolve in ethanol at room temperature. Solution viscosity was measured with a Brookfield Model RV-DVIII Ultra Programmable Rheometer (Brookfield Engineering, Middleboro, MA) at room temperature. The cone number was a CP-40. The range of shear rates tested spanned two decades to simulate the range of shear rates during injection. Data were only accepted for torques between 10 and 100% (this falls within the published sensitivity of the rheometer). To obtain a viscosity value, three viscosity measurements at each shear rate were averaged. Given that rheological behavior was Newtonian (viscosity was independent of applied shear rate), values across all shear rates were then averaged to characterize each ethyl cellulose concentration.

Upon contact with an aqueous environment, ethyl cellulose-ethanol solutions underwent a phase change, resulting in a stiffer material with higher viscoelasticity (rheological data not shown). We refer to this as “gel” for the sake of brevity, but emphasize that its full rheological characterization is not yet complete and will be explored, as needed, in future work. Gel formation was evaluated in relation to the ethyl cellulose concentration within and relative to the amount of water added to the original ethyl cellulose-ethanol solution. Gel formation rates were determined by adding 2 mL of ethyl cellulose-ethanol solutions to 10% fetal bovine serum (FBS, Sigma, St. Louis, MO) in phosphate buffered saline (PBS, Corning, Manassas, VA) in 15 mL or 50 mL centrifuge tubes. The FBS-PBS solution was added in volumes ranging from 0.5 to 35 mL. The centrifuge tubes were placed in a water bath at 37 °C (Branson 2510, Danbury, CT) and allowed to reach equilibrium for 10 minutes. To determine the amounts of ethyl cellulose within gels, they were placed in glass vials on a 120 °C hot plate for 6 hours until no liquid remained. Then the gels were removed from the vials and weighed. To calculate gel density, 3% ethyl cellulose-ethanol was added to an FBS-PBS solution (1:4 weight:weight), the gel was collected and weighed, and then placed in centrifuge tubes filled with water to determine the gel volume from the displaced fluid volume.

Agarose-based mechanical phantoms were used to evaluate distribution of injection volume in vitro. Such phantoms have been employed previously to optimize brain infusion protocols30,31,32. The phantoms were composed of 0.2% agarose (weight:weight, UltraPure Agarose, Invitrogen, Carlsbad, CA), which was stirred into deionized water over a hot plate for 3 hours until the solution was clear. This was then poured into 75 mL vials (20 dram polystyrene containers, Fisher Scientific, Hampton, NH), and allowed to cool at 4 °C for 24 hours to solidify. The phantoms were injected with test media, using 1 mL syringes affixed with 27 gauge needles using microtubing (Tygon Microtube Tubing, 0.25 mm inner diameter), and connected to a syringe pump (NE-300 Just Infusion Syringe Pump). Thus, these phantoms contained a poroelastic microstructure filled with aqueous medium (akin to tumors). Injections consisted of 50 µL of either pure ethanol or 3% ethyl cellulose-ethanol solutions, injected at rates ranging from 0.1 to 10 mL/hr, which were controlled by a syringe pump (NE-300 Just Infusion Syringe Pump). Higher injection rates of 100 mL/hr injections were performed manually, and measured with a stopwatch; this rate exceeded the capacity of the pump. Food dye was added to the injection media to enable visualization of distribution volume. This was defined as the volume which the injected solution occupied within the phantom. The depth for all injections was approximately 25 mm.

Thirty minutes after the onset of injections, images within the phantoms were obtained of the widest cross-sectional area of dye with a ruler in-plane. The phantom was then rotated 90 degrees and imaged again. MATLAB (version 2016a, Mathworks, Inc., Natick, Massachusetts) was used to optically segment the space occupied by the blue dye, and its volume was calculated using Equation (1):

where \(Are{a}_{cross-section}\) is the widest cross-sectional area and \(Radiu{s}_{orthogonal}\) is the widest radius from the orthogonal perspective. Only dye 3.6 mm above or below the tip of the needle was measured, to mimic distribution within a spherical tumor 200 mm3 in volume; this is the average volume expected in our follow up in vivo experiments. Liquid extending above or below this tumor volume is likely to leak out of the tumor in vivo either into surrounding tissue or out along the injection pathway. Experiments for combinations of different injection rate and ethyl cellulose concentration were repeated seven times with injection volume held constant at 50 µL. Experiments for different combinations of injection volume and ethyl cellulose concentration were repeated 5–7 times, with the injection rate held constant at 10 mL/hr. Based on the resulting injection parameter space, we could deduce putative optimum values for ethyl cellulose concentration and injection rate.

An initial evaluation of the cytotoxicity of the ethyl cellulose-ethanol injection mixture was performed using low passage HeLa human cervical carcinoma cells. Cells (obtained from the Duke University Cell Culture Facility, Durham, North Carolina) were maintained with Eagle’s minimum essential medium (MEM, Gibco, Carlsbad, California) supplemented with 10% (vol.) fetal bovine serum, 0.5% (vol.) penicillin, and 0.5% (vol.) streptomycin. Cells were passaged twice per week and maintained at 37 °C and 5% CO 2 . They were cultured in 12-well plates to 80% confluence.

Immediately before each experiment, cell medium was removed and 0.5 mL of fresh medium was added to each well. Next, 0.5 mL of either ethanol, 3% (weight:weight) ethyl cellulose (USP, Sigma Aldrich, Rockville, MD) in ethanol, or PBS (control) was added. The plates were then incubated at room temperature for 15 seconds; this exposure time to ethanol has been shown to induce substantial cell death33. The medium was then removed, and each well was rinsed twice with 1 mL of PBS and given 1 mL of fresh medium.

After all wells had been treated, each was rinsed one time with 1 mL of PBS, given 250 µL of 0.5% trypsin (Gibco, Carlsbad, CA), and returned to the incubator for 5 minutes. Once cells had lifted, 750 µL of medium was added, and the contents of the well were placed in individual vials and vortexed. Viability was then assessed with a trypan blue exclusion assay (Gibco, Carlsbad, CA) using a Countess Automated Cell Counter (Invitrogen, Carlsbad, CA). Two viability measurements (viability = live cell count/ total cell count) were obtained for each well and averaged. There were eight wells for each treatment group.

Chemical Induction of Squamous Cell Carcinoma in the Hamster Cheek Pouch

Effects of injections on tumor regression were evaluated in the hamster cheek pouch model34. The animal study protocol was approved by the Duke University Institutional Animal Care and Use Committee and all studies were performed in accordance with relevant guidelines and regulations (Protocol Number A216-15-08). All procedures were performed under isoflurane anesthesia. Hamsters were female Golden Syrian Hamsters between 100 and 150 grams. Tumors were induced through topical application of 7,12-dimethylbenz[a]anthracene (DMBA, Sigma-Aldrich)34. Three times a week, the buccal mucosa of each cheek pouch was everted and then stretched from the mouth. An area of approximately 5 cm3 was painted with a cotton swab dipped in the DMBA-mineral oil solution. The cheek pouches were painted until tumors developed, typically at around 22 weeks.

Control Injections in Hamster Cheek Pouch Model: High-Volume Pure Ethanol

Pure ethanol injections at high volume were performed manually. The average initial tumor volume was 42 ± 34 µL (s.d.). 192 ± 106 µL (s.d) of pure ethanol mixed with food dye was injected into the center of the tumor. These injection volumes were controlled to achieve a total volume that was 3 to 4 times the tumor volume. Tumor volumes were measured with digital calipers before each injection and daily for 7 days thereafter. For tumors that did not respond and were still present after 7 days, repeat ablations were performed. These were treated as independent ablations (i.e., as though they were new tumors receiving their first ablations). Justification for treating repeat injections as independent injections is shown in Supplementary Figure 1 (tumors receiving a repeat injection behaved similarly to tumors receiving a first-time injection). Such repeats were only performed if the tumor volume had increased for two consecutive days after the 7-day observation period. Volumes were calculated by measuring the longest axis and the orthogonal axis and using Equation (2):

$$Volume=\frac{4}{3}\ast \pi \ast {(Radiu{s}_{long})}^{2}\ast Radiu{s}_{orthogonal}$$ (2)

where \(Radiu{s}_{long}\) is the longest radius and \(Radiu{s}_{orthogonal}\) is the orthogonal radius. Complete tumor regression was defined as the absence of any gross evidence of a tumor or raised lesion by visual examination. Twelve injections were performed in 6 hamsters. Six injections were repeat injections, and only one tumor was treated per hamster.

Injection with Ethanol-Ethyl Cellulose: Varying Injection Rate and Ethyl Cellulose Concentration

Figure 1 illustrates the design of these experiments. For evaluation of varying injection rate and ethyl cellulose concentration, the average tumor volume was 195 ± 140 µL (mean ± s.d.). 50 µL of solution (either pure ethanol or 3% ethyl cellulose-ethanol) was injected into the center of the tumor. Injection volume was always less than tumor volume (about 25% of tumor volume, as compared to 400% in the control studies with pure ethanol that simulated current clinical practice). Injection rates were 0.1, 1.0 or 10 mL/hr (achieved using the syringe pump) or ~100 mL/hr (achieved manually). Notably, the rate of 10 mL/hr has been suggested as optimal for gene delivery into tumors27. Tumor volume was measured before injections, and at 1, 2, 4, and 7 days thereafter. For tumors that did not respond completely and were still present and growing after day 7, repeat ablations were performed (as for controls, above). These were treated as independent ablations (i.e., as though they were new tumors, as described above; justification for treating repeat injections as independent is shown in Supplementary Figure 1). Repeats were only performed if the tumor volume had increased for two consecutive days after the 7-day observation period. 36 total ablations were performed on 8 animals. 15 injections were repeat injections, and multiple tumors from each hamster were treated. The study design is illustrated in Fig. 1.

Figure 1 Study design for assessing effects of injection rate and ethyl cellulose concentration on therapeutic efficacy in vivo. Squamous cell carcinomas were induced in the hamster cheek pouch through topical application of DMBA 3X per week until tumors formed and reached a volume of 100 mm3 (approximately 20 weeks). Tumors were then injected with 50 µL of either ethanol or 3% ethyl cellulose-ethanol solution at a rate of 0.1, 1.0, 10 or 100 mL/hr. After ablation, tumor volume was measured at 1, 2, 4, and 7 days after treatment. For tumors that were still present after 7 days, repeat ablations were performed; these were treated as independent ablations if tumor volume had increased for two consecutive days after day 7. Full size image

All statistical analysis was performed using R software (R Foundation for Statistical Computing, Vienna, Austria)35. For cell viability analysis, a parametric one-way analysis of variance (ANOVA) was used followed by a Tukey post-hoc test, to determine the relative cytotoxicities of ethanol, 3% ethyl cellulose-ethanol, and PBS. For both phantom distribution volume and in vivo normalized tumor volume analyses, non-parametric ANOVAs were performed (Kruskal-Wallis) followed by a non-parametric multiple comparisons test (Dunn’s test) because of the extent of variability. For injection volume, two-way parametric ANOVAs were performed, and the relationship between injection volume and distribution volume was fit to a linear model. The Pearson’s product moment correlation coefficient was calculated to assess the relationship between the phantom distribution volume and in vivo normalized tumor volume. Normalized tumor volume was calculated by dividing the volume at a given time point by the initial volume before treatment. Repeat injections were considered as independent injections. They did not perform significantly differently from first-time injections (Supplementary Figure 1). For in vivo experiments, tumors were randomly assigned to experimental groups. A significance level of p = 0.05 was applied to reject the null hypotheses in all analyses.

The datasets generated during the current study are available in the Open Science Framework repository (found at: https://osf.io/582gd/).