Amorphous Solid Dispersion of a Binary Formulation with Felodipine and HPMC for 3D Printed Floating Tablets

International Journal of Pharmaceutics 658 (2024) 124215 Available online 11 May 2024 0378-5173/© 2024 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/bync-nd/4.0/). Amorphous solid dispersion of a binary formulation with felodipine and HPMC for 3D printed floating tablets Gloria Mora-Casta˜ no a , M´ onica Mill´ an-Jim´ enez a , * , Andreas Niederquell b , Monica Sch¨ onenberger c , Fatemeh Shojaie a , Martin Kuentz b , Isidoro Caraballo a a Department of Pharmacy and Pharmaceutical Technology, Faculty of Pharmacy, Universidad de Sevilla, 41012 Seville, Spain b School of Life Sciences, University of Applied Sciences and Arts Northwestern Switzerland, CH 4132 Muttenz, Switzerland c University of Basel, Swiss Nanoscience Institute, Nano Imaging Lab, Klingelbergstrasse 82, 4056 Basel, Switzerland ARTICLE INFO Keywords: Amorphous solid dispersion 3D printing Fused deposition modeling Gastroretentive floating tablets Drug-loaded filaments Zero-order release ABSTRACT This study focuses on the combination of three-dimensional printing (3DP) and amorphous solid dispersion (ASD) technologies for the manufacturing of gastroretentive floating tablets. Employing hot melt extrusion (HME) and fused deposition modeling (FDM), the study investigates the development of drug-loaded filaments and 3D printed (3DP) tablets containing felodipine as model drug and hydroxypropyl methylcellulose (HPMC) as the polymeric carrier. Prior to fabrication, solubility parameter estimation and molecular dynamics simulations were applied to predict drug-polymer interactions, which are crucial for ASD formation. Physical bulk and surface characterization complemented the quality control of both drug-loaded filaments and 3DP tablets. The analysis confirmed a successful amorphous dispersion of felodipine within the polymeric matrix. Furthermore, the low infill percentage and enclosed design of the 3DP tablet allowed for obtaining low-density systems. This structure resulted in buoyancy during the entire drug release process until a complete dissolution of the 3DP tablets (more than 8 h) was attained. The particular design made it possible for a single polymer to achieve a zero-order controlled release of the drug, which is considered the ideal kinetics for a gastroretentive system. Accordingly, this study can be seen as an advancement in ASD formulation for 3DP technology within pharmaceutics. 1. Introduction Three-dimensional printing (3DP) is an innovative additive manufacturing technique capable of converting 3D computer models into real objects through the sequential deposition of material layer by layer. The ability to manufacture complex structures has opened new possibilities in the design of pharmaceutical dosage forms with different shapes, sizes, dosages, as well as drug release characteristics and multiple drug combinations (Ayyoubi et al., 2021; dos Santos et al., 2023; Parulski et al., 2022; Sadia et al., 2018; Shojaie et al., 2023; Verstraete et al., 2018; Zhang et al., 2017). Three-dimensional printing offers a solution to a major drawback of conventional gastroretentive drug delivery systems (GDDS), which is the limitation of the tablet design. With 3D printing, it is possible to design personalized dosage forms with complex geometries and specific characteristics to improve gastric retention and controlled release of the drug in the stomach. It allows the density and composition of the formulation to be precisely adjusted, which can be beneficial in achieving controlled flotation in the gastric environment (Dumpa et al., 2020; Huanbutta and Sangnim, 2019; Khizer et al., 2023; Melocchi et al., 2021; Mora-Casta˜ no et al., 2023). 3DP is introducing a new approach to personalized treatment, as it enables pharmaceutical forms to be manufactured adapted to the individual needs of patients. Currently, one of the most evaluated 3DP techniques in the pharmaceutical area is fused deposition modeling (FDM), due to the low cost of the printer, the good quality of the final product, high reproducibility, and the potential for innovative drug management strategies. FDM is based on the extrusion of a filament from a heated extrusion head through a nozzle. In this process, materials are melted and deposited layer by layer on a platform that moves in the x and y axes. As the plate descends, the object is built from the bottom up. The ability to precisely control processing parameters allows FDM to have enormous potential and utility for the preparation of personalized medicine (Cailleaux et al., 2021; Dumpa et al., 2021; Melocchi et al., 2021; Pereira and Figueiredo, * Corresponding author. E-mail address: [email protected] (M. Mill´ an-Jim´ enez). Contents lists available at ScienceDirect International Journal of Pharmaceutics journal homepage: www.elsevier.com/locate/ijpharm https://doi.org/10.1016/j.ijpharm.2024.124215 Received 1 March 2024; Received in revised form 19 April 2024; Accepted 7 May 2024 International Journal of Pharmaceutics 658 (2024) 124215 2 2020). However, one of the main drawbacks of FDM is that it requires prior preparation of drug-loaded filaments, usually by hot melt extrusion (HME) (Bandari et al., 2021; Mora-Casta˜ no et al., 2022; Zhao et al., 2022). The high dependence on the physical and mechanical properties of the filaments for the viability of printing and the difficulty of filament preparation are the main drawbacks of this technology (Bandari et al., 2021; Mora-Casta˜ no et al., 2022). HME is a process in which a blend of materials, mainly polymers, drugs, and eventual additives, is melted or softened under elevated temperature and pressure to pass under force along a barrel containing rotating screws. The final product emerges from the barrel through a die that shapes the extruded product (Tan et al., 2018). HME offers many advantages, such as the absence of organic solvents and a low number of processing steps. The possibility of continuous processing and scalability allows its use for pharmaceutical development (Alzahrani et al., 2022; Bandari et al., 2021; Jennotte et al., 2022; Tambe et al., 2021). HME and 3D extrusion-based printing technologies are also innovative tools for formulating amorphous solid dispersions (ASDs), which consist of amorphous drugs dispersed and stabilized on a polymeric support (Bhujbal et al., 2021). Obtaining ASD formulations is a strategy to improve the apparent solubility of drugs and potentially increase their absorption and bioavailability (Alzahrani et al., 2022; Yani et al., 2017). Among the drawbacks that ASDs present, we can highlight the risk of degradation due to hydrolysis or oxidation, as the amorphous drug is more hygroscopic than its crystalline form, and the possibility of drug recrystallization. These drawbacks can be overcome by stabilizing the amorphous drug with a carrier, usually a polymer (Jennotte et al., 2022). The polymer must have the ability to raise the energy threshold necessary for the nucleation and crystallization of the drug and decrease the mobility of the molecules. These mechanisms inhibit drug crystallization during dissolution tests, maintaining the supersaturation state (Jennotte et al., 2022; Vo et al., 2017; Xiang and Anderson, 2017). Factors such as drug-polymer miscibility, solubility of the drug in the polymer, residual crystallinity, molecular mobility, drug-polymer interaction, and the manufacturing process also influence the stability of ASDs, as well as the temperature and humidity of storage. (Alzahrani et al., 2022). A rational approach is preferred to select suitable drugs and excipients for manufacturing ASD, thereby reducing time to market and minimizing costs associated with development (DeBoyace and Wildfong, 2018; Han et al., 2019; Tambe et al., 2022; Zhang et al., 2023). Understanding and predicting the miscibility between the carrier and the drug is a crucial aspect of ASDs to ensure the physical stability of the drug (Xiang and Anderson, 2017; Zhang et al., 2023). Different analytical technologies used to characterize the amorphous solid state (Deon et al., 2022) and verify the results obtained in theoretical predictions made in previous stages are valuable tools. Confirming the compatibility and miscibility of the components of the amorphous dispersion is essential to ensure the safety and efficacy of 3D printed dosage forms formulated with amorphous solid dispersion (Kim et al., 2021; Skowyra et al., 2015; Zhao et al., 2022). Felodipine (FEL) is a class II hydrophobic drug in the Biopharmaceutics Classification System, with a poor water solubility of 0.58 μ g/mL at 25◦C, a melting point of 145 ◦C at crystalline state and a glass transition temperature (Tg) of 47 ◦C at amorphous state (Karavas et al., 2006; Lu et al., 2019). ASD technology has emerged as a promising strategy to improve the solubility and dissolution rate of felodipine (FEL) (Lu et al., 2019; Marsac et al., 2009, 2006; Palazi et al., 2018; Vo et al., 2017; Yi et al., 2019). In conjunction with this technique, GDDS can be used to improve drug release and absorption by maintaining a low concentration around the dosage forms, avoiding in situ recrystallization and allowing gradual drug absorption. GDDS have the ability to maximize the absorption area of drug molecules on the surface of the gastrointestinal tract, ensuring optimal absorption. In contrast, conventional controlled-release pharmaceutical forms can rapidly pass through the small intestine, which restricts their effectiveness (H. Blaesi and Saka, 2024; Vo et al., 2017). The aim of this work was to manufacture FDM 3D printed (3DP) floating tablets composed of an ASD of felodipine with a hydrophilic polymer. For this purpose, the polymer Affinisol™ 15 LV (AFF) has been used as the carrier. Theoretical methods were employed to predict the drug-polymer interactions. Subsequently, the drug-loaded filaments and the 3DP tablets were studied using different analytical techniques for physical characterization and to complement the modeling predictions of the drug-excipient interactions. In addition, the buoyancy and release kinetics of the 3DP tablets were studied. 2. Materials and methods 2.1. Materials Felodipine (Carbosynth Ltd., Compton, Berkshire, UK) was used as the model drug. Hydroxypropyl methylcellulose (HPMC) Affinisol™ 15 LV, a hydrophilic, amorphous polymer, was kindly donated by The Dow Chemical Company (Midland, MI, USA). 2.2. Methods 2.2.1. Preparation of physical mixtures The binary physical mixtures used for the extrusion process were made by weighing out the samples. Previously, felodipine powder was sieved, and the size fraction used was smaller than 180 µm. The raw materials were manually mixed by a “geometric dilution” protocol. A mortar and a pestle were used until the homogenous physical mixtures were obtained. Physical mixtures contained 5, 10, and 15 % (w/w for these and following percentages) of FEL. The drug loading of 5 %, 10 % and 15 % were selected to obtain filaments for printing systems with therapeutic doses of felodipine (2.5–10 mg) as the target dose. The physical mixtures were placed in a vacuum desiccator for 24 h prior to extrusion. 2.2.2. Extrusion process The mixtures were extruded using a twin-screw filament extruder ZE 9 (Three-Tec, Seon, Switzerland) to obtain the drug-loaded filaments. The extruder was operated at 33 rpm, extruding batches of around 5 g of the mixtures at 150 ◦C through a die with a diameter of 1.75 mm. To establish a thermal equilibrium before processing, the extruder’s heat soak time was set to 15 min. The extruded filaments were stored in suitable packaging to avoid water uptake until printing. 2.2.3. Tablet design and 3D printing The design of the 3D printed tablets was created using BlocksCAD software V1.13.0 (BlocksCAD, Burlington, MA, USA), exported as a stereolithographic (.stl) file and sliced on IdeaMaker V3.1.7 software (Raise3D Technologies, Irvine, CA, USA). The drug-loaded filaments were printed by FDM technique using a Pro2 Dual Extruder Raise3D printer (Raise3D Technologies, Irvine, CA, USA). The size of the cylindrical 3D printed tablets was 2.5 mm height x 7.5 mm width for the filaments containing 5 %, 10 %, and 15 % of FEL, based on the therapeutical doses of felodipine (2.5–10 mg) as the target dose for our 3DP tablets (Alhijjaj et al., 2016; Chaturvedi et al., 2014; Ghamami et al., 2014; Govender et al., 2021; Vo et al., 2017; Yi et al., 2019). Additionally, filaments containing 5 % of drug were selected to print tablets of 4 mm height x 10 mm width, to maintain therapeutic doses. The selected range of drug loading on the filaments allows the effects of drug loading and tablet size to be studied while maintaining therapeutic doses. The printing settings were as follows: nozzle diameter of 0.5 mm, nozzle temperature 200 ◦C, build plate temperature 80 ◦C, first layer height 0.2 mm, layer height 0.1 mm, printing speed of 10 mm/s, flowrate 150 %, infill of 25 %, 1 sealing perimeter, 3 solid bottom layers and 4 solid top layers. G. Mora-Casta˜ no et al. International Journal of Pharmaceutics 658 (2024) 124215 3 2.2.4. Characterization of the tablets To assess the uniformity and reproducibility of the 3D printing process, the dimensions (height and diameter) and weights of the 3D printed tablets (n =6) were determined using an electronic micrometer (Comecta, SA, Barcelona, Spain) and an analytical balance (Shimadzu AUW120, Manila, Philippines). According to the European Pharmacopoeia, a hardness tester (Sotax HT1, Teknokroma, Spain) was used to assess the tablet crushing force (n =6). Subsequently, the tablet tensile strength, σ t, was calculated according to eq 1: σ t=2F/ π dh (1) where F is diametrical crushing force (N), d is tablet diameter (mm), and h is tablet height (mm). 2.2.5. Evaluation of amorphous solid dispersion •Molecular Modeling The molecular dynamics simulations (MD) were conducted using YASARA software version 20.12.24 (YASARA Biosciences GmbH, Vienna, Austria) (Krieger and Vriend, 2014). Accelerated calculations were making use of graphic processing units (GPUs) to determine nonbonded interactions (Van der Waals and real-space Coulomb forces) (Krieger and Vriend, 2014). A general AMBER-type force field GLYCAM06 (Kirschner et al., 2008) and GAFF2 were employed (Wang et al., 2004), wherein atomic charges were based on a semi-empirical quantum chemical estimation (AM1BCC) (Jakalian et al., 2002). A simplified HPMC as octamer was prepared in simulation box of 65x 65x 65 Å with randomly placed felodipine molecules corresponding to a concentration of 15 % (w/w) to match the experiments of the binary mixture. Following steepest decent and simulated annealing minimizations to remove clashes, a simulation run was conducted under periodic boundary conditions at 423 K for 15 ns to simulate an extrusion process. The second main simulation run was then conducted at 298 K (under periodic boundary conditions) and lasted another 15 ns. Equations of motion were integrated with a 2x 1 fs, which meant that intramolecular forces were calculated every 1 fs, while intermolecular forces were calculated every 2 fs to provide a full simulation step. An NTP ensemble was used, where pressure control was achieved by rescaling the simulation cell along the x, y, and z axes to reach a constant pressure of 1 bar. For temperature control, atom velocities were rescaled using a weakly coupling thermostat that kept the macroscopic temperature at the requested value. The software did not use the strongly fluctuating instantaneous microscopic temperature to rescale velocities at each simulation step (i.e., a classical Berendsen thermostat) but instead, a scaling factor was calculated according to Berendsen’s formula from the time average temperature to avoid artifacts occasionally observed with a classical Berendsen thermostat. A cut-off value of 8 Å was selected for the Van der Waals forces, and the particle mesh Ewald algorithm was applied to electrostatic forces (Essmann et al., 1995). Finally, the interactions were calculated in 200 ps step (NTP ensemble) following the 15 ns equilibration at 298 K, and mean values from 10 sampling points were used for hydrogen bonding and hydrophobic interaction energy. The entire simulation runs were repeated with n =4. •Surface area The determination of specific surface area employed the standard BET (Brunauer, Emmett, Teller) method using a Gemini VII 2390 Surface Area and Porosity Analyzer (Micromeritics Instrument Corporation, Norcross, GA, USA). The measurements were conducted at a relative pressure (p/p 0 ) within the range of 0.05–0.3. Prior to analysis, approximately 1200 mg of the samples were heated to 80 ◦C and kept for 24 h under a nitrogen atmosphere using the FlowPrep 060 instrument (Micromeritics Instrument Corporation, Norcross, GA, USA). The specific surface area of each sample was measured three times, and the mean values, as well as standard deviations, were calculated. Pure felodipine samples included particles of size >180 µm. Drug-loaded filament samples were previously cryomilled and sieved, using a particle size >250 µm for the cryomilled filament samples. The specific surface area data were needed to evaluate the following gas chromatographic experiments. The surface area mean values and standard deviations of physical mixture and cryomilled filament of 15 % FEL were 0.045 ±0.014 m 2 /g and 0.078 ±0.007 m 2 /g, respectively. •Inverse Gas Chromatography Inverse Gas Chromatography (IGC) was conducted using a SEASurface Analyzer (Surface Measurement Systems Ltd., Wembley, United Kingdom) to characterize surface energies of the samples. IGC is based on the adsorption of a vapor (probe molecule) with known physicochemical properties onto a stationary adsorbent for analysis. A practically infinite gas dilution is hereby targeted for subsequent calculations of the surface energy. Approximately 1.2 to 1.7 g of each sample was inserted into a silanized column with a 4 mm internal diameter. To fix the powdery sample, silanized glass wool was plugged at both ends of the column. Prior to analysis, each column was conditioned at 30 ◦C for 2 h at 0 % RH with helium as a carrier gas at 10 standard cubic centimeters per minute (sccm). This procedure facilitated the removal of any moisture uptake or impurities from the system. Moreover, the dead volume was determined by methane injection before and after the main experiments at 30 ◦C (303 K). The gas injections targeted a 5 % nominal surface coverage (in proximity to the Henry region), and analytics was based on a flame ionization detector (FID). Besides the sample weight, the measured BET-specific surface areas were necessary for the targeted injection concentrations. The net retention volumes (V N ) were then determined using the respective peak center of mass (Peak CoM). The principle of iGC is based on injecting various organic probe solvents with known characteristics into the flow of the carrier gas. The extent of interactions between the solid phase of interest and the probe gas is determined by the net retention volume V N (see eq 1): VN=j mF(tR−t0)T 273.15 (2) where T is the column temperature, F is the carrier gas flow rate at 1 atm and 273.15 K, m is the sample mass, t R is the retention time of the absorbed probe gas, t 0 is the mobile phase hold-up time, and finally, j represents the James −Martin correction (that adjusts retention time for the pressure drop effect in the column bed). A free adsorption/desorption energy is then obtained by eq 2: ΔG=RTln(VN) + K(3) where R is the gas constant and K is an experimental constant (Mohammadi-Jam and Waters, 2014). This total free energy is the sum of the dispersive and specific (acid −base) components of the free energy of adsorption (see eq 3) (Balard et al., 2000). ΔGtotal =ΔGdispersive +ΔGspecific (4) The dispersive surface energy was determined by a homologous series of n-alkanes (heptane, hexane, octane, and nonane), and the polar solvents dichloromethane, ethyl acetate, ethanol, acetone, and toluene were used for the determination of the specific part of the surface energy. The software Cirrus Plus (version 1.2.3.2 Surface Measurement Systems Ltd. (Wembley, United Kingdom)) was used for these energy calculations by selecting the “Della Volpe” scaling option and “Schultz” method for which more details can be inferred from the literature (MohammadiJam and Waters, 2014). Measuring conditions were the same as applied G. Mora-Casta˜ no et al. International Journal of Pharmaceutics 658 (2024) 124215 4 for sample conditioning and mean values and standard deviations are shown from duplicate measurements of three individual packed columns for each material. •Atomic Force Microscopy (AFM) and Laser Scanning Microscopy (LSM) Samples for AFM and LSM were previously cut and broken into small pieces. The AFM and LSM were performed at the Nano Imaging Lab in Basel, Switzerland. Atomic force microscopy (AFM) images were taken in the oscillation mode keeping the amplitude constant (AC tapping mode) using a NanoWizard 4 AFM instrument (JPK Instruments AG, Berlin, Germany). Height and phase images were collected simultaneously using a 160ACNA cantilever (Nanosensors AG, Neuchatel, Switzerland) with a resonance frequency of approximately 320 kHz and 26 Nm −1 spring constant. The resolution of the images was 512 pixels per line. Laser scanning micrographs of the sample surfaces were collected by means of a 3D laser scanning confocal microscopy (LSM) VK-X1100 (Keyence, Osaka, Japan) using a violet laser (408 nm) and a 150x objective lens (Nikon Plan CF Apo, 150x/0.95, WD 0.2 mm). The surface was scanned at high speed in X, Y and Z, allowing image capturing and height measurements with high lateral resolution. Reflected white light and laser light emitted from the focal point were reflected back through the objective lens. The intensity of the laser light that passes through a pinhole is determined by a very sensitive 16-bit photomultiplier. Since the pinhole blocks most of the returning light (except the light from the focal point), confocal LSM delivers much sharper images than conventional microscopy techniques. In addition, a true color image from the integrated second light source is overlaid. •Differential Scanning Calorimetry (DSC) Differential Scanning Calorimetry (DSC) was employed to investigate the thermal characteristics of the samples, and to confirm the compatibility between the drug and the polymer. A DSC 3 STARe system (Mettler Toledo, Greifensee, Switzerland) and a DSC Q20 V24.11 Build 124 were used for samples (about 5–12 mg) weighed into 40 µL aluminum pans covered with pierced aluminum lids. Nitrogen was set to 200 mL/min as the purge gas. The samples were heated at a ramp rate of 5 ◦C/min. Heating and cooling ramps at 10 ◦C/min rate were also used to find the glass transition temperature (T g ) of the samples. The temperature range was set from 40 to 280 ◦C. The thermograms were evaluated using the STARe Evaluation-Software version 16, and the TA Instruments Universal Analysis V4.7A, at the Functional Characterization Service of the CITIUS in the University of Seville. •X-ray Powder Diffraction (XRPD) The crystallinity of the samples was studied using a D2 Phaser diffractometer and a D8 Advance A25 diffractometer from Bruker AXS Ltd. (Karlsruhe, Germany), equipped with a copper tube anode (30 kV, 10 mA and 40 kV, 30 mA, respectively) and a Lynxeye® detector. The samples were automatically rotated at 15 rpm on a sample holder. The increment, number of steps, time per step and range 2θ range were set to 0.02◦, 2124 steps, 1.5 s, and 6–40◦, respectively. A filter of 0.02 Ni, a divergence slit scanning range of 0.5◦(from 3◦to 70◦on a 2θ scale), and a scanning rate of 9◦/min were set to the D8 Advance A25 diffractometer. The data were analyzed using DIFFRAC.SUITE EVA V7.3.1 software and DIFFRAC.SUITE EVA V5.2 software. •X-ray tomography X-ray tomography was performed using a Zeiss Xradia 610 Versa (Zeiss, Oberkochen, Germany) by the X-ray Laboratory Service of the CITIUS in the University of Seville, to evaluate the inner structure of the 3D printed tablets. The scan was conducted using no filter, at a peak voltage of 40 kV, using an optical magnification of 0.4X, and a pixel size of 0.88 µm. Reconstructor Scout-and-Scan v.16.0, 11, 592 software was used to perform the image reconstruction The images were exported as a 16-bit tiff file for visualization. •Scanning Electron Microscopy (SEM) The surface and inner portions of the samples were evaluated using Scanning Electron Microscopy (SEM) performed at the Microscopy Service of the CITIUS in the University of Seville with a FEI TENEO electronic microscope (FEI Company, Hillsboro, OR, USA) at 5 kV. Previously, a Leica EM SCD500 high vacuum sputter coater was used to apply a 10 nm-thin layer of Pt onto the samples. 2.2.6. In vitro release studies Drug release studies of the 3D printed tablets were carried out on the Agilent 708-DS (Agilent, CA, USA) using 900 mL of a pH 1.2 HCl dissolution medium at 37 ±0.5 ◦C, to simulate gastric conditions. 1 % sodium dodecyl sulphate (SDS) was added to the dissolution medium to ensure sink conditions, as previously reported (Brown et al., 2004; Mahmah et al., 2014; McDonagh et al., 2023; Vo et al., 2017). The study was performed in a basket apparatus in triplicate for each batch during 24 h at 50 rpm. The basket apparatus is better suited for preventing erosion of the surfaces of swelling hydrophilic tablets compared to the paddles. The buoyancy was also evaluated through the drug release studies, studying the float lag time and the overall duration of floating. The percentage of drug released was analyzed using a UV–vis spectrophotometer Agilent 8453 (Agilent, CA, USA) at a wavelength of 363 nm (Nimje et al., 2011). The drug release kinetics were also investigated according to Zero order (4), Higuchi (5), Korsmeyer (6) equations (Higuchi, 1963; Korsmeyer et al., 1983): (Mt/M∞) = k0.t(4) (Mt/M∞) = k.t0.5(5) (Mt/M∞) = kK.tn(6) where M t /M ∞ is the fractional drug release at time t (drug loading is considered as M∞); k is the Higuchi kinetic constant. k K is the Korsmeyer’s kinetic constant, t is the release time, n is a release exponent that depends on the release mechanism and the shape of the system (Ritger and Peppas, 1987). Finally, k 0 is the zero-order release rate constant. 2.2.7. Statistical analysis The data are expressed as mean ±standard deviation (SD). Statistical analysis was performed using IBM SPSS Statistics Software (Version 26), by one-way analysis of variance (ANOVA), followed by Scheffe’s post-hoc test. This analysis aimed to discern variations in drug release profiles attributable to the considered formulation factors, namely, drug load and the size of the 3D printed tablet. The significance level was determined at a 95 % confidence limit, with factors demonstrating p ≤ 0.05 deemed as statistically significant. 3. Results and discussion 3DP technology has gained much attention in the pharmaceutical sciences based on the potential to obtain patient-centric dosage forms that are tailored to individual medical needs. Due to the biopharmaceutical challenges associated with especially poorly water-soluble drugs, a combination of ASD and 3DP technology has become a topic of interest (Ayyoubi et al., 2021; Gala et al., 2020; Patil et al., 2016). In this study, formulations of FEL and AFF were developed to produce drug-loaded filaments through HME, which were then utilized in FDM G. Mora-Casta˜ no et al. International Journal of Pharmaceutics 658 (2024) 124215 5 3DP technology, with the aim of obtaining ASDs. In the manufacturing process, different steps were followed to ensure and evaluate the formation of ASDs. This evaluation consisting of two phases (theoretical and experimental studies) will reduce resource consumption and increase accuracy in optimizing a 3D printing process, and by extension, formulation, extrusion process, and final printed systems (Censi et al., 2018; Thakkar et al., 2020). For the theoretical study, solubility parameter estimation and molecular simulations were employed. On the experimental side, different bulk and surface analytics were conducted to physically characterize the obtained mixtures. HPMC AFF was used as the polymeric carrier, as HPMC grades are common hydrophilic polymers utilized in stabilizing amorphous solid dispersions. HPMC is widely used in extended-release systems (Aho et al., 2017; Mandal et al., 2016; Notario-P´ erez et al., 2018; Reynolds et al., 2002; Yi et al., 2019; Zhang et al., 2017). Moreover, cellulose derivatives such as HPMC are well known for their low toxicity (Select Committee on GRAS Substances (SCOGS), 1973). Affinisol™ 15LV is a modified HPMC polymer designed by Dow Chemical Company for utilization in HME applications. In addition, this polymer has shown good printability properties without additives (Gupta et al., 2016; Maˇ skov´ a et al., 2020; Prasad et al., 2019). 3.1. Theoretical approach for the evaluation of ASD 3.1.1. Solubility parameter Firstly, the miscibility of the FEL and AFF was studied. It is crucial to know the solubility/miscibility of drugs in polymers when developing ASDs. This knowledge provides useful information regarding prediction of the stability of solid dispersions (Greenhalgh et al., 1999; Lu et al., 2016; Page et al., 2016). Thus, close solubility parameter values of the mixture components mean that the cohesive energy between the drug and excipient(s) matches, which is a viable method to initially estimate the miscibility of the drug and matrix (Jankovic et al., 2019; Thakkar et al., 2020). Therefore, two substances are known to show good miscibility when the difference in their solubility parameters is less than 7 MPa 1/2 , especially when this difference is less than 2 MPa 1/2 (Greenhalgh et al., 1999); whereas the substances are not considered miscible when the difference is larger than 10 MPa 1/2 . According to the literature, the solubility parameter is 22.7 MPa 1/2 for FEL (Lu et al., 2016) and 24.0 MPa 1/2 for HPMC (Newman, 2015) according to the Hoftyzer/Van Krevelen method calculations (Van Krevelen and Te Nijenhuis, 2009). Therefore, the solubility parameters of FEL and AFF are very close to each other, which indicates a good miscibility of drug–polymer. 3.1.2. Molecular simulations The molecular modeling showed, following the simulated extrusion and subsequent equilibration cycle at room temperature, that felodipine was dispersed in the polymeric matrix without forming aggregated clusters, which could have otherwise pointed to a possible phase separation (see Fig. 1). This result was in line with the previously discussed consideration of solubility parameters in that felodipine was well dispersed in the HPMC carrier. Despite this molecular-level dispersion, there was still not an entirely uniform distribution of the drug in the matrix noted because some surface accumulation of felodipine was noted. This finding agreed with previous molecular dynamics (MD) simulations using a simulated annealing protocol where ibuprofen molecules were observed to stick out of polymeric coils in the solid dispersions (Ouyang, 2012). Regarding the molecular interactions, there was a notable contribution of hydrogen bonding energy that was on the average 40.2 %, whereas the remaining 59.8 % of the total interaction energy was due to hydrophobic interactions. The mean value of the ratio of total hydrophobic to H-bonding interaction energy was 1.5. The individual hydrophobic contacts were only 0.75 to 0.76 kJ per contact but appeared in great numbers as in these long-range interactions. Therefore, the net molecular drug-polymer interaction was more impacted by these Van der Waals forces than hydrogen bonding although individual hydrogen bonds were found to have on the average 19 kJ per bond. The drug was more accepting hydrogen bonds than donating, which was expected from the chemical structure of felodipine. Thus, the compound possesses a hydrogen bond donor N–H group and two hydrogen bond acceptor groups from the ester moieties that enter hydrogen bonding with HPMC. In summary, the simulation results indicated a molecular-level Fig. 1. Snapshot of a molecular dynamics (MD) simulation (after the cycle at 298 K, 15 ns) with felodipine and HPMC as tube model, where for the latter polymer, a semi-transparent molecular surface area is shown. G. Mora-Casta˜ no et al. International Journal of Pharmaceutics 658 (2024) 124215 6 distribution of the drug in the HPMC matrix, which was partly due to strong excipient interactions from both hydrophobic as well as hydrogen bonding interactions that collectively suggested good miscibility. The present findings are overall in good qualitative agreement with previous work on felodipine where the effect of water was studied, which indicated the tendency that hydrogen interactions being favorable to miscibility with the cellulosic polymer were partly disrupted in presence of water (Xiang and Anderson, 2017). Therefore, the good miscibility as indicated by the MD simulations could be potentially hampered by substantial water-uptake at relatively higher humidity conditions. Table 1 Means and standard deviations values (n =3) of the dispersive, specific, and total surface energy of physical mixture and filament (15 % of FEL) obtained from the IGC experiment. Sample name Mean dispersive surface energy (mJ/m 2 ) Mean specific surface energy (mJ/m 2 ) Mean total surface energy (mJ/m 2 ) Physical mixture 18.8 ±2.1 12.8 ±0.9 31.6 ±3.0 Cryomilled filament 23.1 ±0.9 14.1 ±0.2 37.2 ±1.0 Fig. 2. Images of the drug-loaded filaments containing 15% of felodipine (a) AFM topographical, (b) 3D-overlaid height/phase, and (c) confocal LSM; images of powder felodipine (d) AFM topographical, (e) 3D-overlaid height/phase, and f) LSM. Fig. 3. DSC thermograms of FEL, AFF, the physical mixture (PM 5, PM 10, PM15), extruded filament with 5%, 10%, and 15% drug loading and 3DP tablet with 15% drug loading. G. Mora-Casta˜ no et al. International Journal of Pharmaceutics 658 (2024) 124215 7 3.2. Experimental approach in evaluating ASD As the theoretical studies have shown positive results concerning solubility and potential for interactions leading to the formation of stable ASDs, the experimental phase was carried out. 3.2.1. IGC analysis The IGC analysis was used to obtain measurements for dispersive, specific, and total surface energy, as detailed in Table 1. In this work, IGC analysis was applied for the first time to extrudates, suggesting this analysis as part of the experimental approach to evaluate ASDs. Regarding the dispersive surface energy, higher values signify increased hydrophobic interactions, while a higher specific surface energy indicates the prevalence of hydrogen bonding interactions (Mohammadi-Jam and Waters, 2014; Voelkel et al., 2009). As shown in Table 1, both the physical mixture and the cryomilled filament exhibit dispersive surface energies that exceed their specific surface energy. The data reveal that 59.5 % of the total surface energy comes from hydrophobic interactions, while the remaining 40.5 % is attributed to hydrogen bonding interactions, for the physical mixture. The dispersive surface energy was higher for the filament than for the physical mixture. In this case, 62.1 % of the total surface energy comes from hydrophobic interactions, while the remaining 37.9 % is attributed to hydrogen bonding interactions. These findings of the combination of hydrogen bonding and hydrophobic interactions validate the predictions of the MD simulation. 3.2.2. AFM and LSM analysis AFM can be used to evaluate the drug-polymer miscibility, residual crystalline drug, phase separation and recrystallization (Alzahrani et al., 2022; Censi et al., 2018; Ditzinger et al., 2019; Saboo et al., 2021). AFM was employed in this work for detecting residual felodipine crystals within the amorphous solid dispersion. Samples of drug-loaded filament containing 15 % of FEL (the higher drug-load used in this work) and pure felodipine were analyzed. The representative images are visible in Fig. 2. As can be observed, crystalline areas cannot be distinguished in Fig. 2(a), nor in 2(c), where the sample is shown as a homogeneous matrix. Furthermore, as MD simulation results predicted, a phase separation cannot be detected in Fig. 2(b). Drug-loaded filament is shown as a continuous matrix (Fig. 2a-c). Fig. 2d-f illustrate the morphology of pure felodipine crystals within a powdered felodipine sample. Thus, these results confirm the formation of an amorphous solid dispersion between AFF and FEL through the extrusion process. 3.2.3. Thermal characterization DSC studies were performed to evaluate the physical state of FEL in the different steps of the manufacturing process. The thermal behavior of pure substances, physical mixtures, drug-loaded filaments, and 3D printed tablets is presented in Fig. 3. The DSC results showed that pure felodipine exhibited a sharp endothermic peak at 147 ◦C, which was a melting peak (Yi et al., 2019). This demonstrated the crystallinity of the drug. This peak can be also observed in the physical mixtures’ samples. However, this peak was no longer evident in drug-loaded filaments and 3D printed tablet samples. The absence of the melting point at this temperature in the thermographs of filaments and 3DP tablet, indicates that the drug is in amorphous state (Parulski et al., 2022). Regarding the degradation of the drug, no felodipine degradation event was observed in the felodipine thermogram or in the physical mixtures, filaments and 3DP tablet thermograms, which agrees with other works that reported that the degradation temperature of felodipine was higher than 230 ◦C (Alhijjaj et al., 2016; Govender et al., 2020; Guo et al., 2020). This would mean that no degradation of the drug would occur during the extrusion and printing processes. Moreover, it can be observed a T g at 167 ◦C in the samples of filament and 3DP tablet containing 15 % of FEL (Fig. S1 in Supplementary material). Thus, this suggested that amorphous solid dispersions of FEL and AFF were indeed obtained from the formulations used in this work, through the extrusion and 3D printing processes. AFF has demonstrated to be a suitable polymer to form ASD with felodipine but also with other drugs, such itraconazole as it was shown in a previous study (Parulski et al., 2022). 3.2.4. X-ray powder diffraction (XRPD) X-ray powder diffraction patterns of the pure substances, physical mixtures, drug-loaded filaments, and 3D printed tablet were collected (Fig. 4) to qualitatively examine the changes in the physical state of the materials. The XRPD pattern of AFF did not show intensity diffraction peaks suggesting amorphous samples (Solanki et al., 2018). In contrast, the XRPD pattern of pure felodipine exhibited sharp 2θ diffraction peaks at Fig. 4. XRPD patterns of FEL, AFF, the physical mixtures (PM 5, PM 10, PM15), extruded filament with 5%, 10%, and 15% drug loading and 3DP tablet with 15% drug loading. G. Mora-Casta˜ no et al. International Journal of Pharmaceutics 658 (2024) 124215 8 0.20◦, 16.15◦, 23.24◦, 23.33◦, 26.43◦, 32.65◦, confirming the crystallinity of the drug. As it can be observed in Fig. 4, the pattern of the physical mixtures showed the characteristics peaks of FEL. However, the XRPD patterns of the filaments and 3D printed tablet did not show diffractions peaks, confirming the complete amorphization of the FEL with AFF during the extrusion and 3D printing processes, in agreement with the DSC results (Fig. 3). The peak at ca. 32◦observed in the filaments and 3DP tablet belongs to AFF (polymer) as it can be seen in the AFF diffractogram. Overall, the XRPD and DSC data corroborated the prediction by the solubility parameter and molecular modeling study and were also useful for assessing the state of the material in the different steps of manufacturing 3DP tablets. These techniques of evaluation of ASD (XRD and DSC) are the most frequently used in other works dealing with extrudates and 3DP systems (Alhijjaj et al., 2016; Govender et al., 2020; Ily´ es et al., 2019; Solanki et al., 2018). 3.2.5. X-ray tomography 3D printed tablets containing 15 % of FEL were also analyzed by Xray tomography images. Fig. 5 shows a cross-sectional top view X-ray tomography image of the strands of the tablets. The image shows the perimeter as well as the strands that form a quadrilateral internal mesh. As it can be seen, no crystals are detected given that the image shows a homogeneous matrix. This suggests that the strands of 15 % of FEL with AFF are still an amorphous solid dispersion in the 3D printed tablets. This confirmed the predictions that the formulation with the maximum drug content used in this work provided indeed an amorphous solid dispersion. 3.2.6. SEM images Through SEM images, the surface of filaments and 3DP tablets was analyzed. As it can be observed in Fig. 6a-c, filaments present a smooth and homogeneous surface, with an absence of visible crystals. Again, Fig. 5. A cross-sectional top view X-ray tomography image of the strands of the 3DP tablet. Arrows indicate the perimeter and the internal mesh of the tablet. Fig. 6. Drug-loaded filaments surface from 5%, 10% and 15% of felodipine (a, b, c, respectively). Solid top layer of the 3DP tablets made with the filaments of 5%, 10% and 15% of felodipine (d, e, f, respectively). Internal mesh of the 3DP tablets made with the filaments of 5%, 10% and 15% of felodipine (g, h, i, respectively). G. Mora-Casta˜ no et al. International Journal of Pharmaceutics 658 (2024) 124215 9 this confirmed again that an amorphous solid dispersion was obtained with the formulations used in this work. As also occurred in the filaments, the SEM images of the 3DP tablets display homogeneous surfaces, both on the external surface (Fig. 6d-f) and internal mesh (g-i), confirming that the amorphous solid dispersion persists after the 3D printing process. However, the long-term physical stability of ASD could be affected by the relative humidity (RH). The impact of RH on stability increases with the rising hydrophilicity of the pure polymers used in the formulations. Other works using felodipine reported that the formulations with felodipine and HPMCAS remained miscible even at 94 % RH. Felodipine crystallization rates from dispersions were found to be not very sensitive to changes in storage RH (Lehmkemper et al., 2017; Rumondor et al., 2009; Rumondor and Taylor, 2010). Additionally, felodipine was found to be a so-called class III compound regarding its recrystallisation tendency (Baird et al., 2010). Accordingly, the inherent recrystallisation tendency of felodipine is lower than with many other compounds, especially those that are in class II or even I (so-called “fast crystallizing compounds”). In view of the foregoing, the release data showed in Section 3.4 can be viewed as in good agreement with the inherent low recrystallisation nature of felodipine. 3.3. Physical appearance and characterization of 3D printed tablets 3D printed tablets were designed (Fig. 7) and extruded (Fig. 8) layer by layer by FDM, onto a build plate covered by blue tape, according to the digitally designed object. First, FDM extrudes solid underlayers. In our design, these underlayers consist of parallel filaments without any gaps, forming solid layers rather than a network structure. Above the solid bottom layers, parallel lines with gaps between them are built in perpendicular direction from one layer to the next one. Thus, a quadrilateral internal mesh was created. Finally, the FDM extruder formed the solid top layers to complete the object. All these steps were automatically performed, building the systems without the need for drying Fig. 7. Images of the digital design of the (a) top solid layer, (b) internal structure of the 3D printed tablet. Fig. 8. Digital images of the 3D printed tablet: (a) top solid layer, (b) internal structure. X-ray tomography images of (c) top view of 3DP tablet; (d) external perimeter of 3DP tablet. G. Mora-Casta˜ no et al.