Impact of Insoluble Separation Layer Mechanical Properties on Disintegration and Dissolution Kinetics of Multilayer Tablets

Dissolution and disintegration of solid dosage forms such as multiple-layer tablet with different active ingredients depend on formulation and properties used in the formulations, and it may sometimes result in counterintuitive release kinetics. In this manuscript, we investigate the behavior of combined acetylsalicylic acid and mefenamic acid bi- and triple-layer formulations. We show that the simulation model with a cellular automata predicted the impact of the inert layer between the different active ingredients on each drug release and provide a good agreement with the experimental results. Also, it is shown that the analysis based on the Noyes–Whitney equation in combination with a cellular automata-supported dissolution and disintegration numerical solutions explain the nature of the unexpected effects. We conclude that the proposed simulation approach is valuable to predict the influence of material attributes and process parameters on drug release from multicomponent and multiple-layer pharmaceutical tablets and to help us develop the drug product formulation.


Introduction
In the past decades, mostly fixed dose combination products (FDC) have been developed and introduced on the market [1]. In the development of FDC, the dosage form of the multi-layer tablets is selected since the multi-layer tablets have several advantages as compared to the monolithic conventional tablets. For example, one of the benefits of a multi-layer tablet geometries is a possibility to reduce chemical incompatibilities between active pharmaceutical ingredients (APIs) or between other formulation components by physical separation. Additional benefit of a multi-layer approach is a possibility to design dedicated release kinetics for each API in the dosage form [2,3] to help prolonging the product's life cycle [4]. As an approach controlling and modifying the release rate of the API, the multilayer tablet technologies which combine the immediate release layers with the barrier or swellable/erodible properties was reported [5][6][7]. In those cases as well as in the case of multilayered pellet formulations, the drug release of the API was modified depending on the amount of hydrogel polymers such as hydroxy propyl methyl cellulose acetate succinate (HPMCAS)  The bi-layer tablet consists of MA and ASA layers, and the triple-layer tablet consists of MA, polycaprolactone, and ASA layers. The formulations of the bi-layer tablet and triple-layer tablet are shown in Table 2. The 1st layer, middle layer, and 3rd layer of the triple-layer tablet were ASA, polycaprolactone, and MA, respectively. The same ASA granulate and the MA granulate, which was used for each single layer tablet, were used in the multilayer tablets and compacted with the same tooling, which was used for single API tablets. The dwell times for pre-compression and main compression were set to 25 ms. The compaction parameters are presented in Table 2.

Determination of Tablet Porosity, Effective Compressive Stress, and Surface Area
Mean tablet weight was evaluated with an electronic balance (AX204 Delta Range, Mettler Toledo, Greifensee, Switzerland). In addition, tablet diameter and tablet thickness were evaluated with a digital caliper (CD-15CPX, Mitutoyo, Kanagawa, Japan). All obtained values of tablet weight, diameter, and thickness were within 1% deviation. True densities of all raw materials were evaluated using Pharmaceutics 2020, 12, 495 4 of 13 helium pycnometry (AccuPyc 1330, Micrometrics, Norcross, GA, USA). The values of the ASA and polycaprolactone are 1.3928 g/cm 3 and 1.1600 g/cm 3 , respectively. The values for other ingredients were reported previously [33]. Table 2. Formulation compositions and tablet compaction parameters. MA: mefenamic acid.

Formulation Composition
Bi-Layer Tablet  Triple-Layer Tablet  7.4 (ASA layer) 11.1 (MA layer) Compressive force (kN) 21 21 The true density of each layer in the tablets was calculated according to Equation (1).
where ρ layer and ρ i are the true densities (g/cm 3 ) of the layer and each raw material in the layer, respectively, and X i is the weight fraction of each component. The porosity ε of the layers was determined according to Equation (2).
where V is the layer volume calculated from the experimental layer thickness and F-CAD software and where m is the weight of the layer. The horizontal cross-sectional area and the surface area of the tablets were calculated using the computer-aided design software T-CAD (T-CAD, Tokyo, Japan), and the effective compressive stress P comp (MPa) was calculated according to Equation (3).
where S cross section (mm 2 ) is the horizontal cross section area of the tablets and F comp (kN) is the resultant compressive force in the compression of the tablets.

Disintegration Test
The disintegration times were measured using a disintegration tester (Sotax DT3, Sotax AG, Allschwil, Switzerland), according to the United States Pharmacopeia (USP) 24 method. Tests were carried out in 900 mL of 50 mM sodium phosphate buffer (pH 6.8) containing 1% CTAB at 37 • C ± 0.5 • C (n = 3). All tests were done in triplicate using six tablets for each test.

Dissolution Test
Dissolution tests of the ASA tablets, the bilayer tablets, and the triple layer tablets were carried out using the USP dissolution apparatus II with a paddle rotation of 75 rpm (n = 6) (AT7smart, Sotax, Allschwil, Switzerland). The dissolution media and its volume is 900 mL of 50 mM sodium phosphate buffer (pH 6.8) containing 1% CTAB at 37 • C ± 0.5 • C. Drug rate were analyzed by using an HPLC system (Agilent 1100 LC, Santa Clara, CA, United States) equipped with a reverse phase T3 Atlantis C18 column (3 um, 3.0 × 20 mm, Waters, MA, USA) every 5 min. The wavelength of the UV detector was set to 230 nm and 237 nm for MA and ASA, respectively. The column temperature was maintained at 40 • C. The mobile phase used for MA and ASA was acetonitrile/10 mM ammonium formate buffer solution (pH 3.5). The bilayer tablet and triple layer tablet were inserted into the dissolution vessels to ensure that the MA layer is the upper side in the bottom of the vessels. (to avoid the variation of dissolution due to difference in the layer position)

Water Sorption Test
Liquid sorption was measured for MA and ASA granulates in order to compare liquid imbibition rates of both layer components. Measurements were carried out with Krüss Tensiometer K100 (Krüss GmbH, Hamburg, Germany), and data were analyzed with Laboratory Desktop, Ver. 3.2.2.3068 (Krüss GmbH, Germany). The weight of the granulates used for experiments was approx. 700 mg; the inner diameter of the measurement cell was 11.33 mm. The liquid medium was distilled water. Time for liquid uptake experiment was taken enough to reach a plateau region, which indicates no significant further weight increase. The slope and capillary constants were calculated with the Laboratory Desktop software after manual selection of linear sections on the squared mass vs. time plots. Every sorption experiment was carried out in triplicates.

Simulation of Drug Release with Cellular Automata
The software package F-CAD v.2.0 was used for the simulations of the drug release of ASA tablets, the bilayer tablets, and the triple-layer tablets. For the simulation of the experimental tablet, the virtual tablets created in the software, for which geometrical information (i.e., size, shape, and layer order) is identical to the experimental tablets (i.e., 21.5 × 10-mm oval tablet), were generated. The virtual tablet was discretized into a cubic grid using a voxel side length of 74 µm (with 293 3 elements). In our previous work [33], the molecular dynamics approach was adopted to calculate the simulation constants of MA particle, which expresses the dissolution rate in the simulation model with a cellular automata; thus, the same method of the molecular dynamics approach was adopted for the calculation of ASA simulation constants as well.

Comparison of Drug Release Pattern between Experimental and Simulated Profiles
To calculate the similarity factor (f 2 ) of the drug release between experimental data and simulated data, Equation (4) was used [36].
where n is the number of time points, R t is the dissolution rate of the experimental tablet at time t, and T t is the dissolution rate of the simulated tablet at time t. A similarity factor (f 2 ) greater than 50 indicates a close correlation between simulated and experimental data.

Fitting of Noyes-Whitney Dissolution Model with Experimental Data
Fitting of the experimental dissolution results were made with Mathematica 11.0 (Wolfram Research Inc., Champaign, IL, USA). The screenshot is introduced in Supplementary Figure S1. The source code is available in the supplement. The data were fitted with the Noyes-Whitney equation: where C is the fraction of released drug (%, w/v), Cs is the drug solubility (%, w/v), D is the diffusivity coefficient (cm 2 /s), h is the length of the unstirred layer (cm), and S is the specific surface of the sample (cm 2 /g).
Unlike the dissolution and disintegration simulations with three-dimensional cellular automata, the model of Noyes-Whitney does not consider the spatial differentials and material fluxes, which mainly govern the dynamics of dissolution and disintegration of the tablets. The main use and purpose of the Noyes-Whitney equation in this manuscript was to obtain a single coefficient fitting model for comparative analysis of the experimental data. Specific surface was the sole fitting coefficient selected for this purpose. Fittings were made for release data of ASA layer for both triple-and bi-layer tablets. The obtained values for S were interpreted as surface available for dissolution contact in both tablet geometries.

In Vitro Evaluation of Drug Release
The properties of the experimental tablets and their compaction condition are summarized in Table 3. Also, with respect to the effective surface area contributing the disintegration and dissolution of the tablets in the dissolution tests, it is considered in the triple-layer tablet that the effective surface area for MA layer and the ASA layer decreased by the contacting polycaprolactone layer, which does not penetrate the water into these adjacent layers. Based on the theory, the effective surface area was calculated and given in Table 4.  In vitro drug release testing of the MA tablets, the ASA tablets, the bilayer tablets, and the triple-layer tablets were carried out, and the results of MA and ASA are shown in Figures 1  and 2, respectively.  The obtained release profiles for MA in Figure 1 are identical for all study cases, i.e., single tablet, bilayer tablet, and triple layer tablet. An unexpectedly different result is seen for ASA (Figure 2), with significantly slower release rates in the case of a triple-layer tablet formulation. As shown in Table 3, the disintegration time of the bilayer tablet and the triple-layer tablet are longer than that of the single component tablets, corresponding to the rank-order of the in vitro dissolution data.
Results of the fitting of the experimental data with differential form of the Noyes-Whitney equation yields 1.8 times reduction in contact surface for ASA in triple-layered tablet (R 2 = 0.99).
The results of the liquid uptake measurements for the granulates used to compact the layer of bi-and triple-layered tablet are show in Table 5. The total amount of the adsorbed liquid for both granulates was not significantly different. On the other hand, the rates of the liquid imbibition, i.e., a  The obtained release profiles for MA in Figure 1 are identical for all study cases, i.e., single tablet, bilayer tablet, and triple layer tablet. An unexpectedly different result is seen for ASA (Figure 2), with significantly slower release rates in the case of a triple-layer tablet formulation. As shown in Table 3, the disintegration time of the bilayer tablet and the triple-layer tablet are longer than that of the single component tablets, corresponding to the rank-order of the in vitro dissolution data.
Results of the fitting of the experimental data with differential form of the Noyes-Whitney equation yields 1.8 times reduction in contact surface for ASA in triple-layered tablet (R 2 = 0.99).
The results of the liquid uptake measurements for the granulates used to compact the layer of bi-and triple-layered tablet are show in Table 5. The total amount of the adsorbed liquid for both granulates was not significantly different. On the other hand, the rates of the liquid imbibition, i.e., a slope and the capillary constant, show differences by factors of approx. 6 and 4, respectively. The obtained release profiles for MA in Figure 1 are identical for all study cases, i.e., single tablet, bilayer tablet, and triple layer tablet. An unexpectedly different result is seen for ASA (Figure 2), with significantly slower release rates in the case of a triple-layer tablet formulation. As shown in Table 3, the disintegration time of the bilayer tablet and the triple-layer tablet are longer than that of the single component tablets, corresponding to the rank-order of the in vitro dissolution data.
Results of the fitting of the experimental data with differential form of the Noyes-Whitney equation yields 1.8 times reduction in contact surface for ASA in triple-layered tablet (R 2 = 0.99). The results of the liquid uptake measurements for the granulates used to compact the layer of bi-and triple-layered tablet are show in Table 5. The total amount of the adsorbed liquid for both granulates was not significantly different. On the other hand, the rates of the liquid imbibition, i.e., a slope and the capillary constant, show differences by factors of approx. 6 and 4, respectively.

Comparison of in Silico Drug Release Profiles between Bilayer Tablet and Triple-Layer Tablet
The in silico drug release of MA and ASA from the bilayer tablets and the triple layer tablets, which was simulated with cellular automata-based numeric models, are shown in Figure 3, respectively. Also, the comparison between in silico and in vitro release profiles from the bilayer tablets and triple-layer tablets are shown in Figures 4 and 5, respectively.

Comparison of in Silico Drug Release Profiles between Bilayer Tablet and Triple-Layer Tablet
The in silico drug release of MA and ASA from the bilayer tablets and the triple layer tablets, which was simulated with cellular automata-based numeric models, are shown in Figure 3, respectively. Also, the comparison between in silico and in vitro release profiles from the bilayer tablets and triple-layer tablets are shown in Figures 4 and 5, respectively.   Pharmaceutics 2020, 12, x FOR PEER REVIEW 8 of 12

Comparison of in Silico Drug Release Profiles between Bilayer Tablet and Triple-Layer Tablet
The in silico drug release of MA and ASA from the bilayer tablets and the triple layer tablets, which was simulated with cellular automata-based numeric models, are shown in Figure 3, respectively. Also, the comparison between in silico and in vitro release profiles from the bilayer tablets and triple-layer tablets are shown in Figures 4 and 5, respectively.      An attempt to simulate the dissolution profiles of individual layers and the tablet as a whole resulted in the simulated curves of high similarity to the experimentally obtained data, therefore, suggesting similarity between real physical processes and their simulated counterparts. In Table 6, the corresponding f 2 factors are given as a measure of similarity between simulated and experimental release profiles of MA and ASA bi-and triple-layered tablets. Table 6. Summary of similarity factors (f 2 ) between in vitro and in silico drug release profiles.

Discussion
The experimental results shown in Figures 1 and 2 were not expected in the case of ASA triple-layered tablet. The release profiles suggest significant reduction in the release rate for highly soluble compound as compared to the unchanged release rate of the low-soluble MA component. The change is only observed if an inert layer of the polycaprolactone is introduced between layers, i.e., in the statistical analysis (one-way ANOVA) of dissolution rates at 10 min, 15 min, and 30 min, p-value < 0.05 is only observed when comparing the dissolution rates between ASA tablets and triple-layer tablets. It is important to say that this effect cannot be simply explained by a geometrical contact surface reduction for ASA layer through MA or the inert separation layer as, in such a case, a similar effect would be observed for bi-layer tablets, i.e., tablets without separation component. On the contrary, the same behavior cannot be repeated for identically layered tablet composition without an intermediate separation. The results of fitting analysis suggest that ASA layer in a triple-layered tablet has reduced contact surface (by 1.8 times, according to Noyes-Whitney model-based analysis) exposed towards the dissolution medium. At the same time, the ASA layer is a highly soluble formulation and should have shown little or no dependency on the drug release from the MA layer. The results of the liquid uptake measurements corroborate the latter assumption. As shown in Table 5, the rate of liquid uptake for ASA layer is approx. 6 times faster than the capillary uplift registered for MA granules. The analysis of the simulated and experimental data of a bi-layer tablets, i.e., the tablets without inert separation layer, confirms that the upper layer with MA formulation is disintegrated faster due to its reduced thickness as compared to ASA layer and, hence, introduces no more obstacle for a liquid front to ingress into ASA layer. The Polycaprolactone (PCL) material apparently blocks this action for triple-layered tablets, as it remains attached to the surface of the ASA layer long after complete disintegration of the MA layer. The sensitivity of the ASA formulation to the geometry of the tablet was not expected, as it is normally not observed for fast disintegrating immediate release tablet formulations with highly soluble active substances. However, neither the three-dimensional cellular algorithms simulation of the drug release nor fitting of experimental data with the Noyes-Whitney model suggest an alternative explanation of the role of the inert layer. Similar findings reported in Reference [37] were also explained by different surface-to-volume ratios for single-and multi-layer formulations. The disintegration mechanism of ASA layer is different to the one of MA layer and can be best characterized as being erosion-driven. The difference in the disintegration mechanisms for almost identical formulations, i.e., not taking into account the geometrical differences, can be best explained by different diffusivities and aqueous solubilities of the active substances. Interestingly, the visual observation of the tablets' behavior during dissolution suggests rather limited liquid imbibition despite sufficient porosity; liquid transport in the porous meshwork is, apparently, not intensive enough to cause an immediate disintegration; therefore, an erosion-driven disintegration occurs. An effect of porous transport passivation may be caused by an increased viscosity of intra-porous liquid due to dissolved tablet components, such as ASA and mannitol. On the other hand, porosity of the ASA layer was lower as compared to MA (Table 3), which would allow only limited liquid imbibition into tablet body. In order to increase liquid sorption rate and, therefore, to induce disintegration and consequently dissolution speed, the porosity of the pharmaceutical compacts should be greater than the critical value, known as a percolation threshold [38]. However, the higher porosities of tablets lead to reduced tensile strength of the compact; therefore, such an approach to increase liquid diffusivity should be taken with care.
Another important aspect which was identified in this study is suitability of the two dissolution models. The Noyes-Whitney model, which accounts for basic characteristics of dissolving solid, e.g., diffusion and solubility parameters, can be applied for general description of the dissolution behavior under the assumption that the dissolving solid is a homogeneous solid. Significantly better modeling results can be obtained with models based on numerical methods, such as cellular automata method, used in this study. In the left side of Figure 5, the experimental release data of ASA are modelled with cellular automata (solid line). It is possible to spot a slightly sigmoidal shape of the modelled curve generated by the simulation. The models based solely on the Noyes-Whitney kinetic will not produce similar curves without significant extension of the model equations and introduction of further assumptions or simplifications.
The results of the study suggest that, for practical design of the multiple layer tablets, the intermediate layer must be designed as either a soluble or fast disintegrating component to avoid reduction of the dissolution rate, unless otherwise desired.
Inert layer material selection is not a trivial task, especially if this material must be compressed into fast disintegrating layers during tablet production. The layer must maintain separation integrity in order to effectively separate the active ingredients for enhanced stability of the final medicinal product.

Conclusions
In this study, the drug release from triple-layer multicomponent tablet formulation of mefenamic and acetylsalicylic acids was evaluated in vitro and in silico. Numerical simulations were applied to reveal the role of inert separation layer on the changing release behavior of the bi-and tri-layer tablets. For in silico simulation, the model with the three-dimensional cellular automata algorithm showed good agreements with the experimental dissolution data and predicted the impact of the insoluble layer on the dissolution of the APIs in both layers. The unmodified Noyes-Whitney dissolution model does not take disintegration into account and, therefore, was used only as a fitting model for comparative purposes. It is concluded that the simulation based on numerical methods, such as cellular automata, can help to understand the influence of different materials and their attributes, geometry, and process parameters on drug release from multicomponent and multiple-layer pharmaceutical tablet and can aid in rapid drug product formulation.
The results of this study identify the impact of the insoluble layer on the formulation performance of multiple-layer tablets. Introduction of the insoluble and inert layers may significantly reduce the dissolution speed due to reduced surface of both layers. This effect is present even if the bi-layer tablets composed of the same materials do not show dissolution rate reduction. This often-undesired effect will disappear if the separation layer can be composed of either fast soluble or fast disintegrating material.