Modification of Small Dissolution Chamber System for Long-acting Periodontal Drug Product Evaluation

Abstract

Conventional dissolution testing methods may not be suitable for long-acting periodontal drug products due to the small volume, slow fluid flow rate, and environment in the periodontal pocket. The objective of this study was to evaluate a 3D-printed small volume flow-through dissolution chamber system (modified from a previous study) for biorelevant and dose-discriminating testing. Three periodontal drug products with different dosage forms were tested: Atridox, Arestin, and PerioChip. Modifications were made to suit the specific characteristics of these dosage forms. No significant differences were observed between the % drug release profiles in vitro and in vivo except for Atridox. The differences observed with Atridox could be related to the exposing surface area of the drug product. Similar differences were observed from this effect in COMSOL model simulations. Overall, the drugs show reasonable in vitro-in vivo correlations (R² ≥ 0.91) with linear regression slopes close to unity. For dose discrimination between 75% and full dosing, significant differences were observed in the drug release data at specific time points of the products (p ≤ 0.05). The present results suggest that a small volume dissolution chamber with slow flow rate could potentially provide biologically relevant and dose-discriminating evaluations for periodontal drug products.

Graphical Abstract

1. Introduction

Periodontal drug delivery systems are prolonged release systems applied to the periodontal pocket to deliver antimicrobial drug directly into the pocket and inhibit bacterial growth and inflammation. The application sites of these drug products are 1–3 mm wide pockets where the gingiva forms the outer wall surrounding the tooth (Orban, 1948). Gingival crevicular fluid (GCF) flows into the periodontal pocket at a rate ranging from 20 to 137 μL/h, depending on the disease stage and the depth of the periodontal pockets (Goodson, 2003). Due to this small volume and slow fluid flow, the in vivo conditions for drug release of periodontal products are different from those in the commonly used USP dissolution apparatus. A dissolution testing method that can mimic the in vivo conditions is preferred.

Dissolution testing is commonly used to evaluate drug release from a dosage form into the surrounding solution under controlled conditions. This test provides useful information for drug performance evaluation and quality control (Dressman et al., 1998). The U.S. Food and Drug Administration (FDA) defines the relationship between the in vitro property of a dosage form and its in vivo response by using a predictive model as in vitro-in vivo correlation (IVIVC). Establishing IVIVC enables the use of the in vitro dissolution data to predict the in vivo pharmacokinetics of formulations (US FDA, 1997). Therefore, developing an in vitro dissolution method that can predict drug performance by mimicking the in vivo conditions is important, and the incorporation of a dissolution environment similar to the in vivo conditions in the in vitro dissolution method can allow researchers to obtain bio-predictive results (Tsume et al., 2014).

Dissolution methods are also important in the quality control (QC) of drug products to detect variations during the manufacturing process that may negatively impact the product performance. It is a general practice in the pharmaceutical industry to use these methods for batch release and monitoring stability. QC dissolution methods must demonstrate a high level of discriminatory power and be able to distinguish small changes in dose performance to ensure that the product has the required quality (Grady et al., 2018). Most QC dissolution methods use common standard testing procedures such as United States Pharmacopoeia (USP) dissolution apparatus 1–4 (USP, 2011) and apparatus 5–7 (USP, 2013). Modifications can also be made such as the addition of a dialysis adapter for the in vitro release testing of dispersed system dosage forms (Bhardwaj and Burgess, 2010). However, these dissolution methods have relatively large volume to provide a sink condition, which does not reflect the conditions in the periodontal pockets that are significantly smaller in volume and have slower fluid flow.

Commonly used commercial long-acting periodontal drug products include Atridox (doxycycline hyclate), Arestin (minocycline HCl), and PerioChip (chlorhexidine gluconate). These prolonged release drug products are usually applied to the periodontal pockets in combination with scaling and root planing procedures for the treatment of periodontitis. Atridox is an in situ forming gel of the Atrigel delivery system. The in situ forming gel contains 36.7% poly(DL-lactide) (PLA) and 63.3% N-methyl-2-pyrrolidone (NMP) that coagulates in the periodontal pocket when the formulation interacts with water to form a sustained delivery matrix of doxycycline. Arestin is a biodegradable microsphere of poly(lactic-co-glycolic acid) (PLGA) that is administered into the periodontal pocket as a dry powder. The hydrated polymer matrix then releases minocycline that is detectable in the saliva of some patients up to 14 days (Williams et al., 2001). PerioChip is a biodegradable hydrolyzed gelatin matrix chip that is inserted into the periodontal pocket and releases the drug over seven to ten days. To our knowledge, there are no recommended methods for dissolution testing or drug release evaluations of these long-acting periodontal drug products (US FDA, 2021).

The objective of the present study was to evaluate modifications to the 3D-printed flow-through dissolution chamber system that was fabricated in a previous study to mimic the periodontal pocket condition for drug release testing. While influencing factors such as dissolution chamber size, dissolution medium flow rate, and the use of an enzyme in the dissolution medium were already investigated in the previous study (Ren et al., 2019), the flow-through dissolution chamber system was not suitable for all dosage forms. For example, the dissolution chamber from Ren et al. did not contain a filtration system and was not suitable for particulate dosage form as the undissolved particles could leak into the dissolution samples (Patel et al., 2021). To expand the range of dosage forms that can utilize this flow-through dissolution chamber system, modifications were made to the dissolution system for the specific product characteristics as described in Table 1. Three periodontal drug products Atridox, Arestin, and PerioChip were tested in the dissolution system in the present study. These drugs are available in the market and have been used in clinical practice to treat chronic periodontal diseases. Each product represents a unique sustained release system and different dosage form. The first aspect of the present study was to examine the bio-relevance of this dissolution system by evaluating IVIVC after these modifications. The second aspect was to determine whether the dissolution system was adequate to detect the difference in applied dosage at 100% vs. 75% weight for QC purposes. A goal was to examine the applicability of the modified dissolution system for periodontal drug products of different dosage forms.

Table 1.

Rationale for the modifications according to the needs of each dosage form

Problem	Solution/modification	Example of commercial product
API stability	Temperature-controlled sampling box	Arestin, Atridox
Insert dosage form--breakdown into small pieces and sometimes clogged the outlet	Increase the outlet tubing size a	PerioChip, PerioCol CG b
Powder dosage form--leak through the outlet tube	Filter paper added at the outlet to confine the powder in the chamber	Arestin, Dentamycine c
Gel dosage form--sealant used in clinical practice and blocking of drug releasing surface	Release liner film added on top of the gel to block the open surface	Atridox, Elyzol d

^a(Ren et al., 2019)

^b(Kondreddy et al., 2012)

^c(Joshi et al., 2016)

^d(Sholapurkar et al., 2021)

2. Materials and Method

2.1. Materials

Atridox (doxycycline hyclate, 10%) was manufactured by Tolmar, Inc. (Fort Collins, CO). Arestin (minocycline HCl, 1 mg) was manufactured by Valeant Pharmaceuticals (Bridgewater, NJ). PerioChip (chlorhexidine gluconate, 2.5 mg) was manufactured by Drexel Pharma Technologies Ltd (Yokneam, Israel). These periodontal drug products were purchased from their respective vendors via a dental clinic. Doxycycline hyclate USP and minocycline HCl USP were purchased from PCCA (Houston, TX). Chlorhexidine was purchased from Sigma-Aldrich (St. Louis, MO). Ammonium hydroxide solution (30%–33%) was purchased from Honeywell (Muskegon, MI). Triethylamine and trypsin (from porcine pancreas, lyophilized powder, 1,000–2,000 BAEE units/mg solid) were purchased from Sigma-Aldrich (St. Louis, MO). Sodium phosphate monobasic was from Acros Organics (Janssen Pharmaceuticalaan, Geel, Belgium). Tetrabutylammonium hydrogen sulfate was from TCI America (Portland, OR). Sodium chloride, potassium phosphate monobasic, ethylene diamine tetraacetic acid, and o-phosphoric acid were purchased from Fisher Scientific (Fair Lawn, NJ). Acetonitrile (HPLC grade) was from Pharmaco-AAPER (Shelbyville, KY). Materials were used as received.

2.2. Flow-through dissolution chamber system

Flow-through dissolution apparatuses with chamber volume of 0.06 mL (dimensions: 5 mm × 4 mm × 3 mm) were constructed using photo-initiated acrylic polymer clear resin and stereolithography 3D printing process (Formlabs, MA) as described in a previous study (Ren et al., 2019). The dissolution chamber was connected to a syringe pump (Model NE-300, New Era Pump Systems, Farmingdale, NY) through tubing (Intramedic PE-50 as inlet and PE-60 as outlet, Becton Dickinson, Parsippany, NJ) and the temperature was maintained at 34°C with a circulating water bath system (Fig. 1). The temperatures of the system components were monitored using an IR thermal camera (FLIR-E63900, FLIR Systems, Sweden). The syringe infusion pump provided a continuous flow of approximately 0.63 μL/min fresh simulated saliva across the dissolution chamber, and samples were collected at the outlet of the chamber by an Eppendorf tube. For dosage forms containing microspheres (i.e., Arestin), a small piece of filter paper (Whatman cellulose filter no. 5, W&R Balston Limited, England) of approximately 3 mm × 4 mm was placed on the inner wall at the outlet side of the chamber (covering the outlet) to secure the microparticles in the chamber. Both the dissolution chamber and the sampling tube were covered with aluminum foil to protect the system from light. For drugs that could be temperature sensitive and unstable (i.e., doxycycline and minocycline), the sample collection compartment containing the Eppendorf tube at the outlet of the chamber was modified by using a Styrofoam box and ice packs, so the temperature was maintained at below 10°C (measured temperature over 24 h interval = 6.1°C ± 2.2°C, mean ± SD). Simulated saliva, pH 8.0, consisting of 0.137 M sodium chloride, 0.0014 M potassium phosphate monobasic, 0.017 M sodium phosphate dibasic, and 0.3% Trypsin was prepared in deionized water. The present dissolution chamber setup and parameters (dissolution chamber volume, outlet tubing size, dissolution medium flow rate, and enzyme type and concentration in the dissolution medium) were selected according to the results reported in the previous study with PerioChip (Ren et al., 2019).

Fig. 1.

Flow-through dissolution chamber system: general setup of the dissolution system with (optional) sampling temperature-controlled box for Atridox and Arestin, (optional) filter paper for Arestin, and (optional) release liner film for Atridox. The modifications (optional setups) are highlighted in the photo inserts.

2.3. Drug release study

In the drug release experiment, the drug product was placed in the dissolution chamber and the chamber was sealed with hot glue (Superpower, Arrow Fastener Corporation, Mayhill, NJ) and secured with a pinch clamp. The samples were then collected at the outlet of the chamber at predetermined time intervals. The Eppendorf tubes used to collect the samples were weighed before and after sample collection to check the sample volume. The samples were then quantified for the drug concentration with HPLC. After the last sample, the simulated saliva (dissolution medium) in the tubing and dissolution chamber was collected for assay. The residual content in the dissolution chamber was also collected. For Atridox and Arestin, the residues collected from the chamber were dissolved in acetone until a clear solution was achieved and then the polymers were precipitated by the addition of ethanol. The precipitated polymers were removed by centrifugation to collect the supernatant for drug assay with HPLC. This extraction method was examined in drug recovery experiments (in a preliminary study) using the original drug products and showed average of 92%–100% recovery. For PerioChip, the residues collected from the dissolution chamber were mixed with the HPLC mobile phase for drug extraction (Murawsky et al., 2019) and the content was analyzed with HPLC. A previous study using pepsin digestion showed less than 10% drug content remained in the gelatin matrix after 10 days of release experiments in an experimental setting similar to the present study (Ren et al., 2019).

2.4. Drug stability study

The stability of minocycline and chlorhexidine was evaluated by preparing a known concentration of the drug (32, 64, 100, and 200 μg/mL) in simulated saliva with and without trypsin at two temperature conditions. The samples were either stored in a water bath at 34°C or in a refrigerator at 4°C for 7 days before analyzing the drug with HPLC. For minocycline and doxycycline, drug stability was also assessed using the results in the drug release experiments (i.e., total drug recovery by mass balance).

2.5. Effect of gel exposure area and dosing volume of Atridox on drug release

For the in situ forming gel product Atridox, the gel formulation can conform to the shape of the periodontal pocket in vivo that the drug releasing surfaces of the gel can be blocked. In addition, the application of retentive material (Coe-Pak or Octyldent) along with the product to seal the periodontal pocket is recommended in clinical practice (CollaGenex Pharmaceuticals, 2003). To mimic these in vivo conditions and evaluate their effects, different blocking conditions were tested. For the condition of “no blocking” (all drug releasing surfaces of the gel were available), the Atridox gel formulation was coagulated by adding the simulated saliva to the formulation for 1 min and allowing the gel to form, before loading the gel piece into the dissolution chamber. To mimic the “partially blocked” condition (e.g., drug release is blocked at one side), the gel formulation was loaded directly into the dry dissolution chamber. The formulation attached to the bottom of the dissolution chamber effectively restricting this region from access to the simulated saliva, which emulated the “partially blocked” condition in the periodontal pocket, e.g., by an adjacent tooth surface. To mimic the “mostly blocked” (mostly sealed) condition, a dry plastic release liner film was placed on top of the formulation during coagulation in the dissolution chamber to block the top surface of the formulation after it was loaded into the dry dissolution chamber. The release liner film prevented drug release from the top of the gel into the simulated saliva, so both the larger surfaces (release liner film at the top and dissolution chamber wall attached to the bottom) of the gel were blocked.

The standard dosing volume of Atridox is not provided in its product insert. The difference in dosing volume of the gel applied to the periodontal pocket can affect the gel surface to volume ratio and therefore the drug release kinetics from the gel. To this end, drug release experiments were conducted with Atridox to examine the dosing volume effect at 5, 10, and 15 mg doses. In the experiments, the in situ forming gel formulation was placed in an Eppendorf tube and 50 μL of simulated saliva was added into the tube to determine drug release under a non-flow through condition without stirring. The tubes were place in a water bath and the temperature was controlled at 34°C. At the sampling time points, the entire solution in the tube was removed and fresh simulated saliva was added back to the tube. The samples were then analyzed by HPLC.

2.6. COMSOL model simulation of drug release from Atridox surfaces

COMSOL model simulations were performed to evaluate the effect of blocking the bottom or both top and bottom of Atridox gel (one or two of the larger drug releasing surfaces, respectively) on the drug release kinetics. Briefly, a 3D-structure (rectangular cuboid, 3 mm × 3 mm × 0.89 mm) was constructed in the COMSOL program as the Atridox gel using the dimensions estimated from the 10 mg Atridox droplet when loaded into the dissolution chamber. The program settings were: transport of diluted species with general physics, extremely fine mesh, and constant diffusion coefficient of 1.5 or 3 × 10⁻¹² m²/s. Time-dependent drug concentration profiles in the gel structure were generated for the three drug release conditions: no blocking, 3 mm × 3 mm bottom surface blocked (no flux through the bottom surface), and both 3 mm × 3 mm top and bottom surfaces blocked (no flux through the top and bottom surfaces). The volume integral of drug concentration in the structure was calculated and used to determine the percent of drug release over time under these conditions.

2.7. Sensitivity of the dissolution chamber for dosage comparison

Different doses of drug products were loaded into the dissolution chamber to examine whether the flow-through dissolution chamber system was able to distinguish small differences in the product dose. Atridox is a multi-dose delivery system. The dosing condition of 10 mg Atridox was compared with 7.5 mg dose representing the 75% dosing condition. Arestin is a single dose system that delivers 4 mg dry powder formulation. The delivery device was weighed before and after dosing to obtain the powder weight. For the comparison, 3 mg was loaded into the dissolution chamber to mimic the 75% dosing condition. PerioChip is a gelatin chip, and the chip was cut to a smaller piece with 3.5, 5, and 7 mg being the 50%, 75%, and full dosing conditions, loaded into the dissolution chamber, respectively.

2.8. HPLC assay

The drugs were assayed using a HPLC system (Shimadzu Scientific Instruments, Inc., Addison, IL). The HPLC system consisted of two pumps (LC-20 AT), a variable wavelength UV absorbance detector (SPD-20A), an auto injector (SIL-20A), and a Microsorb-MV100–5 C18 column (15 cm × 4.6 mm, 4.6 μm, Varian, Lake Forest, CA). The assay was performed at room temperature. The mobile phase, HPLC conditions, and sampling time interval for each drug are listed in Table 2. The samples were diluted in appropriate volume of the mobile phases before the assay. Standard solutions at appropriate concentration ranges were prepared to construct the calibration curves, as described in Table 2.

Table 2.

Experimental and HPLC method parameters

	Atridox	Arestin	PerioChip
API	Doxycycline hyclate	Minocycline HCl	Chlorhexidine gluconate
Delivery system	Multi-dose Atrigel system	Bioresorbable microsphere	Gelatin matrix
Experimental Setting
Tested dose	Single dose: 10 mg Partial dose: 7.5 mg	Single dose: 4 mg Partial dose: 3 mg	Single dose: 7 mg Partial dose: 3.5 and 5 mg
Sampling time points	2, 4, 8, 18, 24, 48, 72, 96, 120, 144, and 168 h	0.5, 1, 2, 3, 6, 9, 18, 24, 48, 72, and 96 h	4, 8, 12, 16, 24, 48, 72, 96, 120, and 144 h
Outlet sampling compartment temperature	Below 10°C	Room temperature and below 10°C	Room temperature
Other modification and experiment	Release liner film blocking the product; 5, 10, and 15 mg doses under non-flow through condition	Filter paper at the outlet	None
HPLC Method Parameters
Mobile phase	32.2 mL 0.2 M tetrabutylammonium hydrogen sulfate in water, mixed with 32.2 mL 0.38 M EDTA in water and 15.6 mL water, adjusted to pH 7.0 with ammonia solution, then mixed with 20 mL acetonitrile to a final mixture of 80:20 water:acetonitrile	0.6% (v/v) triethylamine in water, adjusted to pH 3.0 with o-phosphoric acid, and mixed with acetonitrile to a final mixture of 85:15 water:acetonitrile	0.23 moles sodium phosphate monobasic in 1.39 L water, mixed with 10 mL 0.5% triethylamine in water, adjusted to pH 3.0 with o-phosphoric acid and mixed with 0.6 L acetonitrile to a final mixture of 70:30 water:acetonitrile
Flow rate	1.2 mL/min	1.2 mL/min	1.5 mL/min
Injection volume	30 μL	30 μL	50 μL
Wavelength for detection	280 nm	273 nm	239 nm
Approximate retention time	4.0 min	3.5 min	4.4 min
Run time	10 min	5 min	5 min
Standard concentration	16–750 μg/mL in mobile phase	1–200 μg/mL in mobile phase	0.1–200 μg/mL in water
Method reference	(Hofsass and Dressman, 2020)	(Matos et al., 2017)	(Ren et al., 2019)

2.9. Data analysis

The data obtained in the present study are presented as means and standard errors of the means (SEM). Statistical analyses were performed using unpaired t-tests and two-way ANOVA with GraphPad (La Jolla, CA), and a difference of p < 0.05 was considered statistically significant.

3. Results

3.1. Effect of sampling outlet temperature on Arestin data

Initial release experiments with Arestin showed low drug recovery (61% ± 2%; mean ± SEM, n=3) according to mass balance of the loading dose. A change in color of Arestin solution (from yellow to brown) was also observed when the samples were exposed to 34°C during sample collection in the drug release experiments. As a result, experiments were performed to investigate the stability of minocycline. The stability study showed that minocycline HCl had good stability at 4°C for 7 days with an average of 8.4% change in the HPLC peak areas (concentrations) of the drug samples under this condition. For comparison, significant degradation was observed with storage at 34°C for 7 days (an average of 86% decrease in HPLC peak areas of the drug samples). These observations are consistent with the unstable nature of minocycline and potential drug degradation in the drug release experiments. Similar stability issues were reported in previous studies of minocycline that the drug was found to be unstable in PBS at 37°C and stable under the frozen condition (Matos et al., 2017). In addition, minocycline is similar to other drugs in the tetracycline group that easily degrades under conditions such as alkaline pH and light (Halling-Sorensen et al., 2002). To reduce the potential of drug degradation, the temperature of the compartment for sample collection was maintained at below 10°C in the drug release experiments. This modification increased the percent of drug recovery of Arestin to ≈93%. For Atridox, the percent of drug recovery was ≈96% when the sampling compartment was controlled under the same temperature condition. For PerioChip, chlorhexidine had good stability at both 4°C and 34°C for 7 days in simulated saliva with trypsin, with an average of 6.6% and 17.6% changes in HPLC peak areas, respectively. The modification of the sample compartment temperature was found to be satisfactory in increasing the recovery of the degradable drugs.

3.2. Drug release in dissolution chamber in vitro and comparison to in vivo data

Fig. 2 shows the drug release profiles of Atridox, Arestin, and PerioChip as the percent of drug release over time in the modified flow-through dissolution chamber. The percent drug release data in the figure were calculated based on the total drug recovered in the experiments. The total drug amounts recovered in the experiments, i.e., including the drug recovered in the residual particles/matrices collected from the dissolution chamber after the experiments, were 96% ± 4%, 93% ± 6%, and 66% ± 3% (mean ± SEM, n=6) for Atridox, Arestin, and PerioChip, respectively, based on their loading dose in the dissolution chamber. The drug release data indicated that Atridox and Arestin had similar trends of drug release in the first 24 h and approached the plateau value within 2–3 days whereas drug release from PerioChip extended to 6–7 days. Arestin had the fastest release profile and PerioChip had the slowest.

Fig. 2.

Comparison of the in vitro drug release (circles) with in vivo human data estimated using saliva or GCF concentration (diamonds and squares, respectively) of (a) Atridox, (b) Arestin, and (c) PerioChip. Mean ± SEM (n = 6 for the in vitro data, n = 12–18 for the in vivo data). No significant differences were observed between the in vitro and in vivo data except for Atridox in vitro and in vivo from GCF (unpaired t-test, p ≤ 0.05).

Fig. 2 also shows the in vivo human data in previous studies for comparison with the in vitro drug release profiles in the present study. The in vivo % drug release data were calculated using the drug concentration in the saliva or GCF reported in the human studies (Fig. S1, Supplementary Materials), based on the total drug recovered as described previously (Ren et al., 2019). For Arestin, the in vivo data were calculated from the drug concentration in the saliva (US FDA, 2003). For PerioChip, the data were from the GCF (Killoy, 1998; Soskolne et al., 1998; US FDA, 2002). The in vivo data of Atridox were from two sources, saliva and GCF (Stoller et al., 1998; US FDA, 2005). Briefly, the in vivo % normalized value at each time point was calculated by:

$$\%\text{ normalized drug release}=\frac{\text{AUC}\left(\text{from t}=0\text{ to t}=\text{time point}\right)\times \text{flow rate }}{\text{AUC}\left(\text{total}\right)\times \text{flow rate }}$$ (1)

where t is time, AUC is the area under the curve of the drug concentration vs. time profile of GCF or saliva, and flow rate is the volumetric flow rate of GCF into the saliva under the assumption of a constant fluid flow and volume over time. Using this approach, the “concentration × flow rate” represents the amount of drug per unit time and “AUC × flow rate” represents the cumulative amount of drug release over time. This method provides the advantage that the calculated % drug release values are independent of the flow rate and associated variability. When comparing the % drug release results in the figure, no significant differences were observed between the in vitro and in vivo data except for Atridox (p ≤ 0.05, unpaired t-test). The in vitro drug release and in vivo human data estimated from GCF concentration for Atridox showed significant differences. Possible explanations of the differences between the in vitro and in vivo drug release data of Atridox are its distinctive dosage form (in situ forming gel) and the variable dose in the human study in vivo, which are investigated in the next section. The PerioChip data were consistent with those previously reported (Ren et al., 2019).

3.3. Effect of gel exposure area and dosing volume of Atridox on drug release

The dosage form of Atridox, which is an in situ forming gel that coagulates in the periodontal pocket, is different from those of Arestin and PerioChip. A possible explanation of the difference between the in vitro and in vivo drug release data of Atridox is that the gel can conform to the shape of the periodontal pocket. The direct contact of the gel with a tooth and surrounding tissues can block the drug releasing surfaces of the gel and affect drug release in vivo. In addition, retentive material (Coe-Pak or Octyldent) is applied to seal the periodontal pocket after Atridox application in clinical settings that can affect drug release in vivo. Fig. 3 shows the influence of blocking the bottom surface (“partially blocked” condition) and both top and bottom surfaces (“mostly blocked” condition) of the gel formulation in the dissolution chamber on drug release from Atridox in vitro. The result of the drug release experiment with gel formation before it was loaded into the dissolution chamber (“no blocking” condition) was the reference for comparison. The in vitro release data indicated that blocking the drug releasing surfaces of Atridox reduced the release rate in the first 24 h by approximately 20% and 50% for the “partially blocked” and “mostly blocked” conditions, respectively, compared to the “no blocking” condition. However, drug release under the “mostly blocked” condition although approaching the in vivo % drug release values in the figure, was still faster than those in vivo. The modification of the dissolution chamber system by adding the release liner film was able to reduce the exposure area of the gel, which was found to better mimic the drug release condition in vivo.

Fig. 3.

Effect of surface blockage on drug release of Atridox. (a) Representative drug concentration profiles in the drug-releasing cuboid structure at 7 days under conditions of no blocking (left), partially blocked at the 3 mm × 3 mm bottom surface (middle), and mostly blocked at both the 3 mm × 3 mm top and bottom surfaces (right) and diffusion coefficient of 3 × 10⁻¹² m²/s from COMSOL model simulations. (b) Release profiles under the different blocking conditions; no blocking (triangles), partially blocked (squares), and mostly blocked (circles). Mean ± SEM (n = 3–6). The curves represent the results from COMSOL model simulations assuming diffusion coefficient of 3 × 10⁻¹² m²/s (dashed curves). Release profile results from in vivo drug concentration in saliva (open diamonds) and 10 mg non-flow through condition (closed diamonds) are included as references.

Another possible explanation of the difference between the in vitro and in vivo drug release data of Atridox is dose variability in vivo (variable dosing volume) because the amount of gel formulation placed in each periodontal pocket is not standardized in clinical practice. For a larger Atridox gel volume, the surface to volume ratio decreases and drug release kinetics can be different from those of the smaller gel volume. Therefore, another set of release experiments for Atridox was conducted to examine the dose effect at 5, 10, and 15 mg doses under the “non-flow through” condition (second part of Section 2.5). Essentially the same % drug release profiles were observed for the three different doses of Atridox (5, 10, and 15 mg doses; Fig. S2, Supplementary Materials). This result indicated that the dosing volume (5–15 mg) did not impact the drug release kinetics. The examined gel surface to volume ratios were not a significant factor affecting the drug release. For comparison with the data of the flow-through dissolution chamber, the representative % drug release vs. time profile of the “non-flow through” experiments (10 mg dosing) is presented in Fig. 3. The fast drug release under the “non-flow through” condition suggests that drug release from the formulation was not significantly hindered by the non-sink condition relative to the condition in the flow-through dissolution chamber, and drug release was mainly controlled by the diffusion in the gel polymer matrix.

3.4. COMSOL model simulation of drug release from Atridox surfaces

Fig. 3 also presents the model simulation results of drug release from the cuboid gel (as a model of Atridox gel) using COMSOL. The COMSOL results were compared with the in vitro drug release data of the flow-through dissolution chamber. From the COMSOL model simulations, blocking the 3 mm × 3 mm bottom surface (“partially blocked” condition) and both 3 mm × 3 mm top and bottom surfaces (“mostly blocked” condition) of the gel resulted in approximately 20% and 50% decrease in the release rate, respectively, compared with the “no blocking” condition. The same effect of surface blocking on drug release from Atridox was observed in the experiment in vitro and in the model simulations under the conditions studied; the drug release profiles from the in vitro drug release experiment essentially overlapped with those in the model simulations when the bottom surface was blocked under the “partially blocked” condition and both top and bottom surfaces were blocked under the “mostly blocked” condition.

In the model sensitivity analysis, changing (a) the geometry setting to a circular cylinder of the same volume and similar dimensions, (b) the dimension setting to a thinner cuboid of the same volume, (c) the mesh size setting, or (d) the diffusion coefficient (by 20%–50%) did not significantly affect the conclusion of the model simulation results (i.e., similar trends of decreasing drug release from “no blocking” to “partially blocked” and “mostly blocked” conditions). In addition, setting the diffusion coefficient at 1.5 × 10⁻¹² m²/s provided a drug release profile in the model simulation (Fig. S3, Supplementary Materials) overlapping the profile from the saliva drug concentration data in vivo.

3.5. In vitro-in vivo correlation (IVIVC)

The in vitro drug release data (% cumulative amount of release) of Atridox, Arestin, and PerioChip in the present study were compared with the in vivo values calculated from the data in previous human studies, and the relationships between drug release in vitro and in vivo were analyzed. For Atridox, the in vitro data of the “mostly blocked” condition were used in this analysis because this condition could be more representative to the environment in vivo. The correlation of the in vitro and in vivo % drug release data is shown in Fig. 4. All three drugs have reasonable linear regression correlations (R² ≥ 0.91) and regression slopes within 20% of unity (1.05, 0.86, and 1.13 for Atridox, Arestin, and PerioChip, respectively), albeit that the in vitro method overpredicted the % drug release of Atridox and PerioChip and underpredicted that of Arestin in the early time periods. Possible explanations of the discrepancies are provided in the Discussion section.

Fig. 4.

In vitro-in vivo correlation (IVIVC) of % drug release from Atridox (circles), Arestin (triangles), and PerioChip (squares) at different time points.

3.6. Sensitivity of the dissolution chamber for dosage comparison

Fig. 5 presents the % drug release profiles obtained from the 75% and full dosing conditions for Atridox and Arestin and 50%, 75%, and full dosing conditions for PerioChip using the flow-through dissolution chamber system. The % drug release values in the experiments were calculated after normalizing the cumulative drug release by the total drug recovered under the full dosing conditions of the formulations (i.e., % drug release was based on the total drug recovered under the full dosing conditions as 100%). The results demonstrated that the dissolution system could detect dosage differences of these drug products. Significant differences were observed between the 75% and full dosing starting at 18, 48, and 48 h for Atridox, Arestin, and PerioChip, respectively (two-way ANOVA, p ≤ 0.05).

Fig. 5.

Comparison of drug release using different dosage: full dosing (circles), 75% dose (diamonds), and 50% dose (triangles) in the in vitro drug release of (a) Atridox, (b) Arestin, and (c) PerioChip. All % drug release values were calculated using the amounts of full dosing as 100%. Mean ± SEM (n = 3–6). Two-way ANOVA with *p ≤ 0.05.

4. Discussion

A dissolution method that is biorelevant and suitable for QC testing of periodontal drug products likely requires the following characteristics: small dissolution chamber volume, slow dissolution medium flow rate, representative dissolution medium composition, and experimental setup that allows long drug release studies. To satisfy these requirements, a dissolution method using a 3D-printed small volume flow-through dissolution chamber was developed in a previous study (Ren et al., 2019). This method was found satisfactory for PerioChip, and the dissolution medium flow rate and enzymatic digestion were observed to be critical factors that significantly affected the results. The present study was a continuing investigation of the previous method and evaluated the modifications made to this dissolution chamber system to accommodate different long-acting periodontal drug products for adequate drug release testing. Experiments were first conducted with PerioChip using the conditions developed previously, and the results were found to be reproducible of those in the previous study. The method was then evaluated for two periodontal drug products of different dosage forms (Atridox and Arestin), but modifications were required in the development to suit these dosage form differences as described in Table 1. This study evaluated the ability of the modified dissolution system to provide biorelevant and discriminating testing that could predict and assess drug release from the periodontal products.

For active pharmaceutical ingredients (API) that are prone to degradation at higher than refrigeration temperature such as 34°C, i.e., minocycline and doxycycline, the sample receiving compartment was modified in the dissolution system to provide a low temperature condition for sampling. The temperature-controlled and light protected compartment can protect the API in the samples from degradation. The success in reducing drug degradation and achieving good total drug recovery for Atridox and Arestin suggests the utility of this approach for sustained release dosage forms that require long duration release experiments and drugs that can decompose at body temperature.

For particulate dosage forms, the flow-through dissolution chamber was modified in the drug release study of Arestin. A Whatman filter paper with 2.5 μm pore size was mounted on the wall of the outlet side in the chamber to retain the Arestin microparticles (particle size approximately 28–40 μm, (Patel et al., 2021)) in the chamber. A preliminary test showed that without the filter paper, the microspheres leaked through the outlet of the chamber and blocked the outlet tube. The results in the drug release study with Arestin demonstrated adequate IVIVC. For Atridox that could involve the application of retentive material to seal the periodontal pocket in clinical settings and potential reduction in drug releasing surface area due to the nature of the dosage form in vivo (in situ forming gel with direct contact and blockage by the surrounding tissues), the present study investigated the effect of blocking the gel surfaces on drug release in the dissolution chamber. For example, a release liner film was applied on top the formulation to simulate the effect of the “mostly blocked” condition. The results from both the in vitro release study and COMSOL model simulations indicated that obstruction of these surfaces had a significant impact on drug release from Atridox. The use of the blocking film to reduce the drug releasing surface area and mimic the in vivo condition for the in situ forming gel improved IVIVC.

In the present IVIVC analysis, reasonable linear regression slopes and correlations were observed for the periodontal drug products. There is little information on IVIVC of dissolution methods for long-acting periodontal products, so a comparison among different dissolution methods is difficult. Although most dissolution studies of oral dosage forms have Level A IVIVC correlations of R² > 0.95 and slopes close to unity, IVIVC studies with R² and slopes similar to those observed in the present study are not rare (e.g., Malewar et al., 2013; Mittapalli et al., 2017; Warnken et al., 2018). First, the R² of 0.91–0.96 observed for the studied periodontal products could be partly due to the assumptions used in the calculation of the in vivo data such as (a) constant GCF flow rate over the duration of drug application in vivo and (b) negligible drug absorption into the surrounding tissues relative to drug release into the periodontal pocket in vivo. Second, IVIVC studies can be highly dosage form specific. For example, Park et al. reported IVIVC regression slope of 0.689 and R² of 0.918 for a three-layered tablet containing tamsulosin HCl vs. the slope of 0.956 and R² of 0.981 for the reference (Park et al., 2011). Khaled et al. reported R² of 0.720–0.973 for tablets of encapsulated metoprolol tartrate and the influence of the dosage forms on Level A IVIVC (Khaled et al., 2013). Besides traditional IVIVC approaches, an in silico method of efavirenz tablets showed R² of 0.85 (Honorio et al., 2013). The deviations from perfect IVIVC for the periodontal dosage forms can be attributed to the environment surrounding the drug products and physiological conditions in vivo. Specifically, drug release in the present dissolution chamber in vitro was faster than those in vivo for two of the tested products (Atridox and PerioChip), particularly in the early time periods. This can be related to that (a) the drug formulation in the periodontal pocket could be obstructed by the surrounding tissues such as the adjacent tooth and sub-gingival tissue for drug release and/or (b) the flow-through condition in the periodontal pocket could be different from those in the dissolution chamber in vitro. Although the use of the small volume dissolution chamber and slow fluid flow rate was an attempt to mimic the in vivo condition, the dissolution chamber volume was likely larger than that of a periodontal pocket. The unhindered flow in the flow-through dissolution chamber system could lead to a sink condition for drug release. However, a sink condition might not be encountered in the small periodontal pocket, and for Atridox, the periodontal pocket can be sealed in clinical practice. For the third product (Arestin), drug release in the dissolution chamber in vitro was slower compared to the in vivo data. This can be attributed to the loss of microspheres from the periodontal pocket to the saliva as there is no barrier to secure the microspheres in the periodontal pocket in vivo. Unlike Atridox and PerioChip, it could be difficult to monitor the expulsion of the microspheres from the periodontal pocket in vivo.

A previous study has demonstrated the utility of a small volume flow-through dissolution chamber compared to the USP Apparatus 4 in the evaluation of drug release from Arestin (Patel et al., 2021). The major differences between the dissolution systems in the present and previous studies were the size of the dissolution chamber (0.06 mL total vs. 0.25 mL chamber plus 0.75 mL reservoir), flow rate (0.63 μL/min vs. 0.5 μL/min), dissolution medium (simulated saliva with an enzyme vs. simulated GCF), filtration system (Whatman filter vs. dialysis membrane), and sample collection method (temperature controlled vs. online analysis), respectively. With the method and experimental setup in the previous study, drug release of Arestin was slower than that in the present study. Interestingly, the release data of Arestin using USP Apparatus 4 and simulated GCF in this previous study (Patel et al., 2021) showed a similar trend of fast drug release as the in vivo data calculated in the present study that approximately 70%–80% of the drug was released in the first day. Both the previous study and the small volume flow-through chamber in the present study employed an open-flow method. Besides the open-flow setup, a close-loop flow-through dissolution chamber can be used. The open-flow method was selected to mimic the physiological condition in the periodontal pocket. The open-flow setting is also similar to the USP Apparatus 4. A close-loop flow-through system is different from the physiological condition. In addition, it can lead to a non-sink condition, which may not be preferred.

The flow-through dissolution chamber system in the present study also demonstrated its ability to discriminate drug release from different dosage amounts of Arestin, Atridox, and PerioChip. The dissolution system was shown to detect 25% difference in the dosage loaded into the dissolution chamber as the drug product. This could be beneficial in QC tests of a manufacturing batch to ensure that the products have the correct API content and that the delivery systems are able to perform as expected. The present study suggests the potential of the dissolution system for drug release testing of long-acting periodontal drug products. However, cautions must be exercised to extrapolate the results in the present study to other sustained delivery systems (different from these three drugs) because other delivery systems might require a more complex dissolution system for adequate testing, due to the difficulty in simulating the small periodontal pocket volume, complex fluid environment, and drug degradation under different conditions. For the dosage forms examined in the present study (gelatin chip, microsphere, and gel), the use of a filter membrane at the outlet of the dissolution chamber is a modification essential to drug release testing of particulate dosage forms. The incorporation of the blocking film to control the exposure of a gel is another modification to mimic the restrictive enclosure in the periodontal pocket to provide reliable drug release profiles. The modification of temperature-controlled sampling can prevent the API from degradation for drugs that are temperature sensitive and unstable in long drug release experiments, which are normally required for long-acting periodontal drug products. These modifications were intended to cover the different aspects of the marketed periodontal dosage forms. Future studies such as with other modifications and designs are required to tackle other challenges from dosage forms not included in the present study.

5. CONCLUSION

A dissolution method for long-acting periodontal drug products was developed previously to mimic the small volume and slow fluid flow rate in the periodontal pocket. This method was modified and evaluated in the present study for its ability to provide biorelevant and dose-discriminating testing with three drug products, Arestin, Atridox, and PerioChip, of three different dosage forms (microsphere, in situ forming gel, and gelatin chip, respectively). The original method was designed for the evaluation of PerioChip and had its limitations for other dosage forms. The main modifications to the flow-through dissolution chamber system were the (a) temperature-controlled sampling chamber, (b) filtration system, and (c) blocking film to reduce drug release. These modifications improved the utility of the dissolution system and allowed the drug release testing. The drug release results indicated (a) reasonable correlations between the present in vitro and previous in vivo data and (b) dose discriminations between 75% and full dosing for the drug products. The method has the potential to evaluate drug release from periodontal dosage forms under a biological relevant condition and provide sufficient dose discrimination for QC and manufacturing purposes.