Cyclodextrin polymer preserves sirolimus activity and local persistence for anti-fibrotic delivery over the time course of wound healing

Abstract

Fibrosis and dysphagic stricture of the esophagus is a major unaddressed problem often accompanying endoscopic removal of esophageal cancers and pre-cancerous lesions. While weekly injections of anti-proliferative agents show potential for improved healing, repeated injections are unlikely clinically, and may alternatively be replaced by creating an esophageal drug delivery system. Affinity-based polymers have previously shown success for continuous delivery of small molecules for weeks to months. Herein, we explored the potential of an affinity-based microparticle to provide long-term release of an anti-proliferative drug, sirolimus. In molecular docking simulations and surface plasmon resonance experiments sirolimus was found to have suitable affinity for beta-cyclodextrin while dextran, as a low affinity control, was validated. Polymerized beta-cyclodextrin microparticles exhibited 30 consecutive days of delivery of sirolimus during in vitro release studies. In total, the polymerized beta-cyclodextrin microparticles released 36.9 milligrams of sirolimus per milligram of polymer after one month of incubation in vitro. Taking daily drug release aliquots and applying them to PT-K75 porcine mucosal fibroblasts we observed that cyclodextrin microparticle delivery preserved bioactivity of sirolimus inhibiting proliferation by 27−67% and migration of fibroblasts by 28−100% of buffer treated controls in vitro. Testing for esophageal injection site losses, no significant loss was incurred under simulated saliva flow for 10 minutes, and 16.7% of fluorescently labeled polymerized cyclodextrin microparticle signal was retained at 28 days after submucosal injection in esophageal tissue ex vivo versus only 4% of the initial amount remaining for free dye molecules injected alone. By combining affinity-based drug delivery for continuous long-term release with a microparticle platform that is injectable yet remains localized in tissue interstitium, this combination platform demonstrates promise for preventing esophageal fibrosis and stricture.

Graphical Abstract

1. Introduction

Esophageal cancer is the 18th most common cancer in the United States with an estimated 17,290 new cases and 15,850 related deaths occurring in 2018 ¹. Improved survival outcomes correlate with early detection and early treatment. While critical changes in early detection methods are in development ², minimally invasive early treatment options still exhibit some drawbacks ^3–5. Endoscopic submucosal dissection (ESD) is one such treatment option for early stage esophageal cancers and cases of Barrett’s esophagus with high-grade dysplasia ^6–7. ESD is advantageous as complete removal of cancerous tissue can be confirmed via histology of en bloc resected tissue and the procedure has proven to increase disease-free survival rates ^8–11. However, fibrotic processes accompanying ESD often result in scarring and dysphagic stricture of the esophagus, especially in patients with extensive or circumferential ESD for which post-procedure stricture rates of 70% and higher have been reported ^12–15. Post-surgical attempts to relieve stricture, such as mechanical balloon dilation require many procedures due to the refractory nature of the stricture over months to years with risk of perforation, internal bleeding, and reduced patient quality of life. Other strategies include introducing a stent, applying surgical meshes, and steroid therapies, yet none have sufficiently addressed post-ESD stricture to improve patient quality of life ^{12, 16–19}.

Interestingly, repeated local injections of anti-scarring agents to underlying tissue after ESD was observed to reduce esophageal stricture by up to 50% at four weeks post-ESD ²⁰. However, multiple, weekly intra-esophageal injections over the time course of wound healing present a clinical and patient burden similar to repeated esophageal balloon dilation procedures and risk perforation and abscess formation due to multiple injections. Additionally, high local concentrations of mitomycin C were observed to exhibit deleterious effects ²⁰. A more tolerable agent, sirolimus (SRL; rapamycin), is explored in this study. SRL has indicated potential for anti-scarring therapy in glaucoma surgery and has been shown to reduce the rate of restenosis of coronary lesions ^21–22. However, SRL is a small molecule drug (MW=914.172 g/mol) with poor oral bioavailability and a limited tissue half-life of only 62 hours as compared to the four week timeline of wound healing ^23–24. Therefore, we explored the potential of using a drug delivery system to provide long-term, local release of SRL as an anti-fibrotic therapy.

In previous work, polymerized cyclodextrins (pCD) have demonstrated the ability to continuously deliver small molecule drugs on the order of weeks to months by leveraging affinity interactions ^25–31. Cyclodextrin monomers are polymerized into larger macrostructures such as disks and particles or even device coatings in order to achieve a high concentration of affinity binding sites to delay drug release ^{27, 32–36}. pCD-based materials are capable of loading high amounts of many classes of drugs while extending release for several months compared to low affinity interactions that typically release >95% of drug within a week, which is clinically insufficient for the timeline of wound healing expected in ESD ³⁷. Herein the potential of polymerized cyclodextrin microparticles (pCD-MP) to release SRL for an extended time, the efficacy of released SRL to regulate proliferation and migration, and the retention of pCD-MP in esophageal tissue in vitro under diffusion and flow conditions is evaluated.

2. Experimental Section

2.1. Materials

β-cyclodextrin (β-CD) prepolymer, lightly crosslinked with epichlorohydrin, and 6-deoxy-6-amino β-cyclodextrin heptahydrochloride and 6-deoxy-6-amino γ-cyclodextrin octohydrochloride were purchased from CycloLab (Budapest, Hungary). Ethylene glycol diglycidyl ether was purchased from Polysciences, Inc. (Warrington, PA). Dextran (15−25k molecular weight) was purchased from Sigma-Aldrich (St. Louis, MO). SRL (rapamycin) was purchased from Biotang (Lexington, MA). 3A-amino-3A-deoxy-alpha-cyclodextrin, amino-dextran, and all other reagents, solvents, and chemicals were purchased from Thermo Fisher Scientific (Hampton, NH) in the highest grade available.

2.2. Microparticle Synthesis

Epichlorohydrin-crosslinked β-cyclodextrin prepolymer (or dextran for non-affinity control) was solubilized in 0.2 M potassium hydroxide (25% w/v) and heated to 60°C for 10 minutes. Light mineral oil was warmed in a beaker with a Tween85/Span85 solution (24%/76%) and stirred at 500 RPM. Next, ethylene glycol diglycidyl ether was added drop-wise and the solution was vortexed for 2 minutes before pouring into the beaker with the oil/Span/Tween85 mixture, increasing the temperature to 70°C, and increasing the stir speed. The reaction was carried out for 3 hours to form polymerized cyclodextrin microparticles. The microparticles were then centrifuged at 200xg to be separated from the oil mixture, washed with excess hexanes twice, excess acetone twice, and finally de-ionized water (diH₂O) twice. The microparticles were resuspended, frozen, and lyophilized before further use.

2.3. Microparticle Characterization

Microparticles were resuspended (1 mg/ml) in Dulbecco’s phosphate buffered saline (DPBS) with 0.1% Tween80 (polysorbate 80). Particles sizes were measured from images taken with a Nikon Eclipse TE300 inverted microscope (Nikon Inc., Tokyo, Japan) after 1 and 30 days of incubation.

2.4. Molecular Docking Simulations

Molecular structure data files for SRL and β-CD were downloaded from the PubChem database. Structures were converted to PDBQT format and energy was minimized before loading SRL as a ligand and cyclodextrins or dextran as a host in PyRx (Molecular Graphics Laboratory, The Scripps Research Institute, La Jolla, CA). The Autodock Vina algorithm was used to predict the strength of the ligand/guest interaction ^38–39

2.5. Surface Plasmon Resonance

The binding strength between a variation of cyclodextrin monomers and SRL was measured experimentally through surface plasmon resonance (SPR) with a Biacore X100 system (GE Healthcare Bio-Sciences, Pittsburgh, PA). Conditions used were based upon previous optimization for small molecule drugs binding to cyclodextrins ^{33, 36}. The surface of each sensor chip CM-3 was conjugated with EDC (0.4 M) and NHS (0.1 M) followed by either 10 mM 6-amino-6-deoxy-β-cyclodextrin (Cyclolab) or 6-deoxy-6-amino γ-cyclodextrin (CycloLab) or 3A-amino-3A-deoxy-alpha-cyclodextrin (ThermoFisher Scientific) suspended in HBS-N buffer (a HEPES balanced salt solution with pH 7.4). The other channel was conjugated similarly with amino-dextran (Thermo Fisher Scientific) to determine specific versus nonspecific interactions with a chemically similar but non-affinity substrate. The remaining functional groups were capped by running ethanolamine through both channels. A multi-cycle kinetic experiment was performed on each chip individually with SRL dissolved in a 1% dimethylsulfoxide MilliQ water solution. The surface was regenerated with 50 mM sodium hydroxide between samples to fully dissociate any remaining bound drug. The differential responses between the channels were fit to both steady state affinity and a 1:1 kinetics binding model using Biacore evaluation software. Goodness of fit was verified by U-value <25 and Chi-squared <10 as specified in the manufacturer’s instructions.

2.6. Drug Loading and Release

Microparticles were solubilized in filtered diH₂O and SRL in dimethyl sulfoxide (DMSO). SRL was loaded in a 50:1 (drug:polymer) solution with a final solvent concentration of 50% DMSO and 50% diH₂O for 72 hours on a rotary shaker. Loading solutions were removed and particles were next mixed with release buffer (1x DPBS with 0.1% Tween80 which acts as a physiological buffer providing a hydrophobic sink ^{33, 40}) and incubated at 37°C on a rotary shaker. At pre-determined time points of 1 hour, 3 hours, and daily thereafter, the particles were centrifuged and release buffer was exchanged to monitor drug release and maintain sink conditions. SRL concentration was determined via a Synergy H1 Hybrid Multi-Mode Microplate Reader (BioTek Instruments, Inc., Winooski, VT) by measuring absorbance of the aliquots at 278 nm in a quartz microplate (Hellma, Plainview, NY) and comparing to a SRL standard curve in release buffer. The release aliquots were stored at −20°C prior to further analysis.

2.7. Cell Proliferation

PT-K75 porcine mucosal fibroblasts (ATCC.org, Manassas, VA) were grown in Dulbecco’s Modified Eagle’s Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin using standard culture methods at 37°C with 5% CO₂. Cells were plated at 6,000 cells per well in a 96 well tissue culture treated plate and allowed to adhere in media for 1.5 hours. Next, a fraction of SRL release aliquots or buffer control aliquots were added to the media and incubated for 24 hours. Aliquot treatments were 100 ul each, volume matched with drug only controls and diluted with 100 ul culture media. 10 µl of 0.15 mg/ml resazurin (Thermo Fisher Scientific), which is used to evaluate metabolic activity in the alamarBlue assay for quantifying fibroblast viability and proliferation ⁴¹, was added per treatment well. After incubation for an additional 20 hours, fluorescence at 530/590 excitation/emission was measured with a Synergy H1 Microplate Reader (BioTek Instruments, Inc., Winooski, VT) and fluorescent signals were normalized to buffer controls.

2.8. Cell Migration - Scratch Assay

Porcine mucosal fibroblast cells were plated at 25,000 cells per well in a 96 well plate in serum starvation media (only 0.5% fetal bovine serum) and incubated overnight. Plates were then scratched uniformly using a 200 µl sterile pipette tip and a custom guide, washed with serum starvation media, and treated with aliquots of SRL or release buffer for 24 hours with microscope images taken before and at the end of treatment. Aliquot treatments were 100 ul each, volume matched with drug only controls and diluted with 100 ul culture media. ImageJ analysis tools were used to quantify the scratch closure and the results show the average change in scratch width normalized to buffer only controls.

2.9. In Vitro Tissue Retention Analysis - Diffusion

Amino-β-CD MPs were synthesized by pre-mixing 6-amino-6-deoxy-β-cyclodextrin (0.5 wt %) with epichlorohydrin-crosslinked β-cyclodextrin prepolymer before synthesizing microparticles as stated in methods above. Alexa Fluor 647 (AF647) NHS-ester (Thermo Fisher Scientific) was mixed with amino-β-CD MPs in 0.1M sodium bicarbonate buffer (pH 8.3) according to manufacturer’s instructions for conjugation. The particles were then washed five times in MilliQ water and dialyzed against buffer for three days to remove excess free dye molecules before use. Porcine tissue was acquired for secondary use from animals on protocols approved by the Case Western Reserve University Institutional Animal Care and Use Committee. Esophageal tissue was dissected post-mortem and bisected into 2 cm wide strips to expose the lumen. 20 µl of near-infrared fluorescent pCD-β-MPs in DPBS were then injected ex vivo into the esophageal submucosa using a 25G needle and syringe. Tissue samples were protected from light and incubated in excess DPBS to simulate diffusion from tissue. Although this does not fully recapitulate particle retention in vivo, it does give a preliminary estimate for particle retention after injection as the large size of the particles inhibits clearance via blood capillaries in healthy tissue ^42–43. Remaining fluorescence in tissue samples was measured at each time point using an IVIS Spectrum imaging system and quantified using spectral unmixing and region of interest (ROI) selection tools (PerkinElmer, Waltham, MA). After 28 days, tissues were frozen in optimal cutting temperature (OCT) compound (Thermo Fisher Scientific), cryosectioned, and mounted on slides for confocal imaging using a Leica DMI4000 B TCS SPE confocal microscope (Leica Camera Inc., Allendale, NJ). A 405 nm laser with a PMT capturing 420−550 nm light was used to image the tissue autofluorescence while a 647 laser with a PMT capturing 670−750 nm light was used for imaging the AF647 conjugated fluorescent pCD-β-MPs.

2.10. Microparticle Persistence with Simulated Saliva Flow

To measure persistence of the polymer microparticles under flow conditions, esophageal tissue was dissected post-mortem and bisected into 2 cm wide strips to expose the lumen. 20 µl of near-infrared fluorescent pCD-β-MPs or Dex-MPs (conjugated to AF647 NHS-ester via same technique as pCD-β-MPs, but with amino-dextran (Thermo Fisher Scientific)) in DPBS were injected at an angle of 45° and depth of 5mm ex vivo into the esophageal submucosa using a 25G needle. To recapitulate in vivo washout and explore the potential of leakage from injection site, simulated saliva flow was introduced at 1 mL/min via syringe pump with artificial saliva (0.1 wt% albumin in prepared saline ⁴⁴) over the esophageal tissue lumen. Flow was stopped at the indicated volumes to measure remaining fluorescence in tissue samples with an IVIS Spectrum imaging system and quantified using spectral unmixing and region of interest (ROI) selection tools (PerkinElmer, Waltham, MA).

2.11. Statistical Analysis

Data are represented as the mean with standard deviation or standard error of the mean where specified. Statistics were calculated using GraphPad Prism 7 software (GraphPad Software, La Jolla, CA). Statistical significance was defined as p < 0.05 following statistical tests with further specifications noted in the figure captions.

3. Results

3.1. Affinity-based Delivery Predicted by Micromolar Affinity Between SRL and β-CD

Molecular docking was performed in silico to determine the strength of SRL interactions with potential hosts for affinity-based drug delivery. A set of common cyclodextrins were screened as host macromolecules and SRL was simulated as the ligand with binding energy minimizations simulated in Autodock Vina software (Figure 1A). The simulations indicated the strongest average affinity (lower K_D) for β-CD (22.51 µM) versus γ-CD (49.31 µM) or α-CD (103.52 µM). Dextran was simulated as a chemically similar control, but without a central cavity capable of forming an inclusion complex resulting in a much lower binding strength (1671.7 µM) comparable to previous studies ^{33, 45}.

Figure 1.

PyRx simulated SRL complexation with cyclodextrin monomers and predicted binding affinity with standard deviation derived from different possible conformations (A). Experimental determination of SRL and β-CD affinity measured by SPR indicates lowest affinity (B).

Additionally, all cyclodextrin and SRL complex affinity interactions were measured directly by SPR (Figure 1B). We have observed previous success for affinity-based drug delivery for affinity constants in the micromolar range ³³. Interactions between flowing SRL and surface-bound 6-amino-6-deoxy-β-CD demonstrated a K_D in the micromolar range via steady state affinity (81.24 µM) and Langmuir 1:1 kinetics (97.11 µM) models. Alternatively, flowing SRL interactions with surface-bound α-CD (510 µM) and γ-CD (146 µM) resulted in less affinity binding, in agreement to the trend of the previous PyRx simulation results with slight differences due to estimations in the simulation. Reported K_D values were within the model confidence interval with Chi² values below 10% of the maximum SPR response ³³. Therefore, β-CD was chosen to maximize affinity-based delivery of SRL and dextran was used as a low-affinity control to determine the improvement in SRL delivery due to affinity interactions.

3.2. pCD-β-MP Delivery Increases the Amount and Timeframe of SRL Release

β-CD prepolymer lightly crosslinked with epichlorohydrin was further crosslinked with ethylene glycol diglycidyl ether using an inverse emulsion polymerization method to create insoluble microparticles with a high CD concentration for drug delivery ³⁴. Excess SRL was loaded at a 50:1 ratio of drug to polymer to maximize particle loading. After 72 hours, loading solution was removed and particles were suspended in release buffer in separate tubes with measurements taken at predetermined time points. The linear detection range was calculated to be from 125 to 1.9 µg/ml for SRL in release buffer. As expected pCD-β-MPs maintained a consistent daily release while non-affinity material, Dex-MPs demonstrated an initial burst release of 27 µg/ml SRL in the first hour followed by a drop in concentration to 8.8 µg/ml by day 3 (Figure 2A). The pCD-β-MPs not only provided a significantly longer delivery window but also released an average of 1314.9 µg SRL/mg polymer/day for the first 21 days of release which was twice the average daily amount released by the Dex-MPs.

Figure 2.

Released SRL in buffer was measured daily from pCD-β-MP and Dex-MP (A, left) and normalized to the amount of total polymeric particles weighed before loading (A, right). The normalized daily cumulative depicts the rate of SRL release (B). Total SRL released was calculated from daily release data (C) and loading efficiency was determined by mass released divided by mass of drug initially added to samples (D). n=3 separate samples per group; error bars represent standard deviation; * indicates p < 0.05 by two-way ANOVA with Bonferroni test post-hoc Days 2−19 for (A), Days 6−30 for (B) or by Student’s t-test for (C, D).

Considering the cumulative release data (Figure 2B), the Dex-MPs released SRL more rapidly in the first four days, while the pCD-β-MPs exhibited higher cumulative delivery during later time points on days 16−23 of release. The pCD-β-MPs exhibited relatively little initial burst release in the first day (9% of total released) versus the 18.8% burst release from Dex-MPs. This trend was observed in previous studies to be associated with affinity interactions which reduce burst release while extending the drug delivery profile versus control non-affinity polymers ³³.

Not only was the average daily amount and rate influenced by affinity, but the total amount of SRL released over the 30 day period was 1.5 times greater for the pCD-β-MPs. The pCD-β-MPs released an average of 36852 µg SRL per mg polymer while the Dex-MPs released 24406 µg SRL per mg polymer in solution (Figure 2C).

The loading efficiency was 73.7 ± 2.3% for pCD-β-MPs and 48.8±1.4% for Dex-MPs (Figure 2D). The higher drug loading efficiency of the pCD-β-MPs is likely due to affinity interactions and contributes to the greater amount of drug release observed from the pCD-β-MPs over time.

3.3. Released SRL Reduces Fibroblast Proliferation In Vitro

A fraction of release aliquots were applied to porcine PT-K75 mucosal fibroblasts in vitro to determine if SRL maintains bioactivity via reducing proliferation following delivery from polymer microparticles. Optimum incubation time and cell plating density were determined for the resazurin reduction assay to measure metabolic activity as a proxy for proliferation ⁴¹. The volume matched SRL release aliquots from pCD-β-MPs and Dex-MP both maintained a bioactive effect in reducing cellular proliferation relative to buffer treated controls (Figure 3A). The standard curve indicated that free drug alone had a similar effect to the released SRL (Figure 3B). The trend for early time points correlated with SRL release in the 10−30 µg/ml range corresponding to proliferation of ~60−70% relative to buffer control. The two polymers’ release activity at days 21 and 28 gave similar results as expected for the similar release amounts at the indicated time points. Albeit released SRL maintained an anti-proliferative effect up to 28 days after the initial drug release study was began in vitro, indicating the potential for extended biological activity following long-term release in vivo.

Figure 3.

Proliferation of pig fibroblast cells (PT-K75) was measured by Alamarblue assay after 44 hour incubation with daily SRL release aliquots from pCD-β-MP and Dex-MP (A) and standard curve validating the drug’s activity alone (B). n=3 separate wells per group; * indicates p < 0.05, ** p < 0.01 by two-way ANOVA with Bonferroni test post-hoc versus 100% buffer controls; error bars represent standard error of the mean.

3.4. Porcine Fibroblast Migration Inhibited by Released SRL

SRL bioactivity was also measured against cell migration. PT-K75 porcine fibroblast monolayers were scratched with a pipette tip and a fraction of SRL release aliquots were applied in serum starvation media ⁴⁶. Culture conditions and plating density were optimized prior to testing release aliquots (volume matched to controls) and images of the scratch assay setup are presented in Supplementary Figure S1. Day 1 and 7 aliquots of pCD-β-MPs released SRL completely inhibited PT-K75 cellular migration as compared to Dex-MP (Figure 4A). At day 21 and 28, both pCD-β-MPs and Dex-MP released SRL were capable of partially inhibiting migration as compared to the buffer treated controls. Similar activity was observed for free SRL drug alone (Figure 4B) with 10 µg/ml inhibiting migration by approximately 50%, which is comparable to the day 21 and day 28 release aliquots. These results corroborate the potential for pCD-β-MPs delivery of bioactive SRL to inhibit migration for up to 28 days post-loading.

Figure 4.

PT-K75 porcine mucosal fibroblast migration into a scratched area after 24hr incubation with SRL aliquots from pCD-β-MP and Dex-MP in serum starvation media from indicated release times (A) and free drug applied at pre-determined concentrations for control (B). n=9 measurements per group; * indicates p < 0.05 by t-test versus controls incubated with buffer alone; error bars represent standard error of the mean.

3.5. Polymerized Particles Retain Their Size After 30 Day Incubation

Particle sizes were measured to ensure that pCD-β-MPs and Dex-MPs remained intact in physiological solution at 37°C over the four weeks required for wound healing resolution. Particles were imaged at 1 and 30 days post-incubation in release buffer (Figure 5A). Measured particle diameters and standard deviations at day 1 and day 30 were 14.68±5.28 µm and 13.81±4.31 µm for the pCD-β-MPs while the Dex-MP diameters were 23.87±8.97 µm and 23.16±9.31 µm, respectively (Figure 5B). Mean particle sizes were observed to remain constant over the 30 days of incubation in buffer with no indication of derivative products released in solution. As hypothesized the size of particles did not change; therefore results indicate that drug release from the synthesized particles was not due to particle degradation or erosion and that both polymer microparticles are able to maintain their size for extended time periods in physiological buffer.

Figure 5.

Representative images of synthesized pCD-β-MP and dextran microparticles following 1 and 30 days incubation in physiological release buffer at 37°C (A). Particle diameters were calculated with ImageJ and presented as mean with standard deviation (B). n≥150 particles were measured per group; day 1 vs. day 30 by one-way ANOVA and Bonferroni test post-hoc were not significant as expected for pCD-β-MP (p=0.866) or Dex-MP (p=0.714).

3.6. In Vitro Tissue Retention of pCD-β-MPs

Our pCD particles not only extend the delivery window of SRL, but also improve localization of the released drug due to their limited degradation rate and macroscopic size. As a proof of concept test for particle retention in tissue, porcine esophageal tissue strips were injected ex vivo with near-infrared fluorescently labeled AF647-pCD-β-MPs and maintained in excess physiological buffer (Figure 6A). Fluorescence signal was measured longitudinally and it was observed that 16.7% of the initial signal remained after 28 days despite tissue degradation and less than ideal retention conditions in this mock scenario (Figure 6B). The half-life of retention was calculated to be 3.07±8.72 days by a one phase decay model (R²=0.8721). Confocal imaging of tissue slices confirmed that AF647-pCD-β-MPs were intact within tissue 28 days after injection and that the fluorescence signal was still co-localized with particles (Figure 6C). Contrastingly, after the injection of small molecule dye alone, similar in size to unbound SRL, only 4% of the initial amount remained after 28 days in buffer.

Figure 6.

Fluorescently conjugated pCD-β-MP were injected in excised esophageal tissue and incubated in DPBS solution with the remaining signal imaged at the indicated incubation times by IVIS (A) and quantified over time (B). Esophageal tissue was sectioned and imaged via confocal microscope to reveal fluorescently conjugated AF647-pCD-β-MPs remaining in tissue after 28 days incubation in buffer (C). n=2 replicates per time point.

3.7. In Vitro Tissue Retention of Polymeric Microparticles with Simulated Saliva Flow

To further test the persistence of our polymer microparticles after injection in esophageal tissue, we have adapted other models of drug adhesion in the esophagus with simulated saliva flow ^47–48. Esophageal tissue strips were injected ex vivo and the initial signal of fluorescent particles was compared to the signal remaining after 1 ml and 10 ml of washout (Figure 7A). With a 25G needle, which is typically used for endoscopic procedures, we did not observe any significant leakage of either pCD-β-MPs or Dex-MPs from the injection site. Microparticle persistence was also similar between both polymer particles, indicating that 15−25 micron diameter particles have a potential for high retention in esophageal tissue under physiological flow conditions (Figure 7B).

Figure 7.

Local persistence was tested after injection of both fluorescently labeled polymer microparticles in separate esophageal tissue samples (A) and exposed to simulated saliva flow conditions with remaining particles measured by fluorescence signal using IVIS (B). No significant particle leakage was observed from the injection site after flow. No significant differences were measured between the polymer microparticle types. n=3 replicates per time point; error bars represent standard error of the mean.

4. Discussion

Our work demonstrated that inclusion of affinity components (Figure 1) in a polymer microparticle has the capacity to both increase the total amount of SRL loaded in the polymer and mediate the release rate of SRL to a more physiologically relevant profile (Figure 2). Because particles were synthesized with only CD and a low-molecular weight crosslinker we were able to achieve a high concentration of molecular “pockets” within the particles. It is hypothesized that the drug release profile in our work (Figure 2) was due to a multitude of affinity-based interactions occurring as drug diffused through the particle structure ^{45, 49}. Herein we show that affinity can be leveraged to increase drug loading of pCD systems but needs to be balanced to provide an appropriate context-dependent release rate ³⁶. For example, using a derivate cyclodextrin, such as methyl-β-CD which exhibits even greater affinity for SRL ⁵⁰, may result in higher drug loading, but would further limit the release rate and effective release concentration of the system. The affinity-based delivery profile achieved in our work appears to balance sufficient loading with release of biologically active SRL for the four weeks required for stable wound healing following ESD (Figures 3 and 4) ^{20, 37}. The pCD-β-MPs retained their physical size during the incubation period (Figure 5) indicating that drug release was mediated mainly by diffusion and affinity-based mechanisms rather than degradation or erosion ^{26, 51}. Other techniques such as molecular imprinting could be utilized to further improve cyclodextrin polymer drug loading ⁵². Moreover, as the polymer persists at the injection site (Figure 6, 7), there is a potential for a booster effect in which the particles can be re-loaded with a local drug dose if needed ⁵³. This reloading can extend the release profile and local concentration which may aid in dealing with refractory strictures ⁵⁴.

Our drug/device platform fills a gap in the field as other materials and devices such as drug eluting stents, nanoparticles, and thin films have been tested for SRL delivery in other disease contexts, but have not been validated for esophageal drug delivery. Using nanoemulsion techniques, acetylated β-CD nanoparticles 250 nm in diameter were found capable of SRL delivery for 20 days in vitro, yet neither in vivo retention nor direct anti-fibrotic activity was reported ⁵⁵. SRL eluting cardiovascular stents were crafted to deliver 14–28 days of continuous release, but the initial burst release is high with the following continuous release at a much lower concentration ⁵⁶. Interestingly, insoluble pellets of SRL trapped in microporous polycaprolactone thin films provided the longest delivery of up to 16 weeks of SRL release, but this results in a daily concentration which may be below the effective dose and the application of such a film to the esophageal wall should be tested in future studies ⁵⁷.

We test injected our polymer microparticle suspensions through a 25G endoscopic needle. This is critical for translational potential using commercially available endoscopic needles. Additionally, while no adverse response has been observed in animals implanted with pCD disks and microparticles ^{34, 58}, exact immune system interactions still need to be determined, especially in an inflammatory context. There have been very few previous attempts at particle injections within the esophagus. Previous studies have focused on anti-cancer drug delivery or tissue bulking ^59–60, which leaves many unanswered questions to be explored in future studies of esophageal drug delivery. Injectability of microspheres in the esophagus and localization is hypothesized to be important as to inhibit fibroblast migration from the muscle layer yet allow for epithelial migration/proliferation and wound healing to occur laterally from the periphery of dissected tissue ⁶¹. Local delivery of SRL to the esophagus post-ESD is expected to promote wound healing by preventing excessive fibroblast proliferation similar to previous results with mitomycin C without the cytotoxic drawbacks ^{20, 62}. However, SRL may potentially lengthen the time to resolve healing in vivo due to anti-inflammatory and other reported effects, yet this may be a dose balancing issue ^63–64. Alternatively, other molecules promoting wound healing may be co-delivered, such as trefoil peptides, growth factors, polyamines, and more ^65–66 which could promote the healing of the epithelium and may be applied via a different set of drug-loaded pCD-MP injections for a greater combined effect than anti-fibrotic therapy alone.

5. Conclusions:

Cyclodextrin pre-polymers were formulated into microparticles capable of sustained SRL delivery over the course of four weeks. The released SRL was confirmed to be bioactive against fibrotic processes in assays against porcine fibroblast proliferation and migration in vitro. pCD-β-MPs retained their size after injection and were localized in esophageal tissue 28 days after injection and incubation ex vivo. This combination of an injectable microparticle platform which remains localized in tissue interstitium and affinity-based drug delivery for continuous long-term release regardless of physical properties exhibits potential for preventing esophageal stricture.