*3.1. Materials*

Pyrazinamide, citric acid, sodium starch glycolate (Primojel®), xylitol, Dulbecco's modified Eagle's medium (DMEM), fetal bovine serum (FBS), L-glutamine, non-essential amino acids, penicillin/streptomycin, disodium hydrogen phosphate, potassium dihydrogen phosphate, sodium chloride, 3-(4,5-dimethylthiazol-2-yl) 2,5-diphenyl tetrazolium bromide (MTT) and neutral red (NR) cell viability assay were purchased from Sigma Aldrich (St. Louis, MO, USA). Copolymer polyvinyl alcohol-polyethylene glycol (Kollicoat® IR) was procured from BASF (Ludwigshafen, Germany). Hepatocyte cell line (HepG2) was purchased from American Tissue Culture Collection (ATCC) (Manassas, VA, USA). All other chemicals employed were of analytical grade and used as received.

### *3.2. Experimental Design*

### 3.2.1. Constructing the Box Behnken Design Template

The systematic preparation and optimization of the PZA loaded formulation was based on a 4-factor, 3-level Box Behnken experimental design template, a response surface methodology (RSM), constructed utilizing the Minitab® 18 Statistical Software (Minitab LLC, State College, PA, USA). The independent variables were the formulation excipients namely polyvinyl alcohol polyethylene glycol (X1), sodium starch glycolate (X2), citric acid (X3), xylitol (X4). 3-levels of the independent variables referred to as lower (−1), midpoint (0) and upper (+1) limits were selected for the construction of the design template as represented in Table 1. The dependent variables or responses were parameters key to the performance of the formulation and these included disintegration time (Y1), dissolution pH (Y2) and formulation weight (Y3). Factor level selection for each excipient was set on their ability to produce stable orodispersible formulations, which was based upon the one-variable-at-a-time approach (OVAT) [38,61]. The OVAT approach was implemented by changing one variable per time while keeping the others constant so as to determine the influence exhibited by each excipient. Accordingly, the Box Behnken design template generated 27 possible combinations (F1–F27) with 3 replicates at central points to minimize errors as presented in Table 4 [62]. Model estimation and significance level were executed using the analysis of variance (ANOVA) where *p*-values below 0.05 indicated statistical

significance and correlation coefficient (R) closest to one (>0.9) was selected because of complexities associated with quadratic experimental design templates (Table 3).


**Table 4.** Independent variables employed for the Box Behnken design template.

### 3.2.2. Preparation of Orodispersible Formulations

Pyrazinamide loaded orodispersible formulations were prepared using the solvent casting technique [27,40,41]. Each formulation consisted of different amounts of polyvinyl alcohol polyethylene glycol, sodium starch glycolate, citric acid and xylitol based on the design template detailed in Table 5. As a result, 27 orodispersible formulations were prepared with each, containing a fixed quantity of pyrazinamide which equaled 500 mg per formulation. Briefly, for every orodispersible formulation variants, all excipients (factor levels) and drug were carefully weighed on a calibrated analytical balance (AS220.R2 Radwag Wagi Electroniczone, Radwag, North Miami Beach, FL, USA) and added to 20 mL deionized water under continuous stirring (Digital Hotplate Stirrer, Model H3760-HSE; Lasec; Ndabeni, Cape Town, South Africa) at 500 rpm over 60 min at 37 ± 0.1 ◦C until a homogeneous slurry was formed. The homogeneous mixture was left to cure in an airtight and dark environment until all air bubbles were visibly absent. Next, specific amounts (required to produce 20 films per formulation variant) of the cured slurry was filled into specialized, hollow plastic molds and then placed into a Labcon forced air circulation incubator (Model FSIH4, Krugersdorp, Gauteng, South Africa) until dried to constant weight at 25 ± 0.5 ◦C over 24 h. The resulting drug loaded formulations were then appropriately stored away in airtight, opaque vials for further testing.


**Table 5.** Box Behnken design template.


**Table 5.** *Cont.*

**Note**: **\*** indicates the experimental design center points; **X1** = Polyvinyl alcohol polyethylene glycol (Kollicoat® IR); **X2** = Sodium starch glycolate; **X3** = Citric acid; **X4** = Xylitol. Each orodispersible film variant blend (i.e., F1–F27) contained 20 mL deionized water as solvent, 500 mg pyrazinamide as model drug and produced an average of 20 orodispersible film formulations per variant blend meaning that a single film formulation was loaded with approximately 25 mg pyrazinamide.

### 3.2.3. Weight Determination for the Matrices

Each prepared orodispersible formulation (F1–F27; Table 2) was weighed using a calibrated analytical balance (AS220.R2; Radwag Wagi Electroniczone, Radwag, North Miami Beach, FL, USA). For each measurement, three independent samples were weighed and mean weight ± standard deviation was calculated and recorded.

### 3.2.4. In Vitro Disintegration Time and Dissolution pH of the Matrices

The in vitro disintegration time of the 27 experimental design orodispersible formulations was measured utilizing a modified petri dish method [63,64]. Disintegration time represents the specific period when formulation matrix collapse begins [38]. The disintegration time was determined visually using a dual-display digital stopwatch (Fotronic Corporation, Melrose, MA, USA). In this case, each sample was placed in 5 mL of pH 6.8 simulated saliva solution contained in a glass vial and placed in the shaking water bath (ST 30, NÜVE, Akyurt, Ankara, Turkey) maintained at 37 ± 0.1 ◦C and 10 rpm to mimic the oral cavity [26]. The vial was swirled after every 10 seconds and physical appearance of the formulation was consistently observed for any dimensional changes [28]. The simulated saliva was prepared by dissolving 2.38 g disodium hydrogen phosphate, 0.19 g potassium dihydrogen phosphate and 8.00 g sodium chloride in a liter of distilled water [65]. In vitro disintegration time was recorded at the point when the sample started breaking apart. Thereafter, test samples were allowed to dissolve completely to form a homogenous solution and dissolution pH recorded using a pH meter (GmbH 8603, Mettler Toledo, Sonnenbergstrasse, Schwerzenbach, Switzerland) [40,66]. All the measurements were done in three replicates.

### *3.3. Formulation Optimization*

The main objective of the statistical design approach was to develop an optimal pyrazinamide loaded orodispersible formulation. After generating a full quadratic polynomial regression which connected dependent with independent variables from the Box-Behnken design template, experimental outputs were fitted within set limits for predicting the optimal orodispersible formulation. Selection and analyses of optimized levels were performed using the Minitab® 18 statistical software by simultaneously applying specific constraints on the dependent variables namely, disintegration time, dissolution pH and formulation weight, as presented in Table 6. Accuracy and efficiency of the statistical optimization process was measured using the desirability function in which case a value closest to one is indicative of precision. To validate the experimental design approach, the optimized orodispersible formulation was prepared in triplicate, dependent variables measured and obtained values were compared to the predicted values. Thereafter, more optimized drug loaded and placebo formulations were prepared for additional in vitro characterization and testing.


**Table 6.** Model summary of optimization constrains and statistical significance of the selected response parameters.

### *3.4. Physical Properties of the Optimized Orodispersible Formulation*

### 3.4.1. Weight Determination

The optimized formulation weight was measured in triplicate using a calibrated analytical balance as previously described.

### 3.4.2. Measurement of Inner and Outer Diameter

The inner and outer diameter of the optimized formulation was manually measured in triplicate using a centimeter calibrated precision ruler.

### 3.4.3. Disintegration Time and Dissolution pH

The time elapsed at the onset ofin vitro disintegration and the media pH after complete formulation dissolution was quantified using methods already detailed above.

### *3.5. Drug Content Analysis*

Pyrazinamide loaded and placebo optimized formulations of about 12 × 10 mm dimension were separately dissolved in 100 mL of simulated saliva contained in an Erlenmeyer flask. The resulting aqueous mixture was placed on a digital hotplate magnetic stirrer (Model H3760-HSE; Lasec; Ndabeni, Cape Town, South Africa) set at 37 ± 0.1 ◦C and 500 rpm. The samples were visually monitored until a complete clear solution was formed. Subsequently, 1 mL of the clear solution was appropriately diluted in simulated saliva and passed through the 0.45 μm nylon syringe filter (Whatman®, GD/X syringe filters, Sigma Aldrich, Johannesburg, South Africa). The placebo formulation was also subjected to the same dilution and filtration processes as the drug loaded samples and used as blank measurements to nullify background absorbance associated with included excipients. Filtrates collected from both drug loaded and placebo samples were then separately analyzed by measuring absorbance using a UV/VIS spectrophotometer (Nanocolour® UV/VIS, Macherey Nagel, Separations, Bellville, Cape Town, South Africa) set at a λmax of 268 nm, specific for PZA [26]. The final absorbance measurements obtained from this differential computation were fitted into a linear calibration curve (*y* = *654.34 x; R<sup>2</sup>* = *0.96*) to obtain the actual and percentage PZA content of the optimized formulation. All quantifications were performed using three replicate samples.

### *3.6. Evaluation of In Vitro Drug Release Kinetics*

The in vitro drug release experiment was carried out on three separate optimized formulations. Each sample was separately enclosed in lidded glass vials containing 5 mL simulated saliva and the entire contrivance was immersed into a shaking water bath at 37 ± 0.1 ◦C under gentle agitation of 10 rpm, mimicking the buccal environment. Thereafter, 2 mL sample was collected and replaced with an equal volume of freshly prepared, temperature equilibrated simulated saliva (37 ± 0.1 ◦C) at different time intervals (10, 30, 60, 90 s and 2, 5, 10, 30, 60 min). The samples were then diluted, filtered using 0.45 μm Whatman® nylon syringe and analyzed with a Nanocolour® UV/VIS spectrophotometer at λmax = 268 nm to detect drug absorbance which was eventually translated into percentage drug release values employing a linear polynomial equation (*y* = *654.34 x; R<sup>2</sup>* = *0.96*). Furthermore, obtained drug release profile was analyzed employing model dependent methods namely zero-, first-, second-order as well as Higuchi and Korsmeyer–Peppas and Hixon–Crowell [67]. The model of best-fit optimally describing the mechanism of drug release from the optimized orodispersible formulation was selected based on the coe fficient of determination (R2) closest to one. All mathematical fitting was performed using the KinetDS, version 3.0 open source software.

### *3.7. Physicochemical Characterization*

### 3.7.1. Di fferential Scanning Calorimetry (DSC)

The thermal properties of PZA, all excipients used, optimized drug loaded and placebo were evaluated and compared using a di fferential scanning calorimeter (DSC, Q2000 DSC, TA Instruments, New Castle, DE, USA). Approximately 6 mg of each sample was placed into a flat bottomed standardized aluminum pan which was directly transferred into the calorimeter for testing purposes. For referencing, an empty aluminum pan was included for each measurement as needed. All test samples were analyzed three times at 10 ◦C/min−1, temperature range between −65 ◦C and 300 ◦C under an inert nitrogen gas flow rate of 25 mL/min. The thermograms obtained were recorded and analyzed.

### 3.7.2. Thermogravimetric Analysis (TGA)

The drug model PZA, excipients, optimized drug loaded and placebo formulations were assessed using a thermogravimetric analyzer (TGA Q500 V20.13 Build 39, TA Instruments, USA). About 8 mg of each sample was separately placed into platinum pans, heated at a temperature range of 10–400 ◦C, flow rate of 5 ◦C/min and maintained under constant nitrogen and air flow set at 40 mL/min and 60 mL/min respectively. The percentage weight loss during each heating cycle was recorded using the TGA universal analysis software. Measurements were completed in triplicate and results expressed as the mean of the three readings.

### 3.7.3. Evaluation of Structural Transitions

A Fourier transform infrared (FTIR) spectrophotometer (Perkin Elmer Spectrum 100 Series, Beaconsfield, UK) equipped with the Spectrum V 6.2.0 software was utilized for the characterization of PZA, all excipients, optimized drug loaded and placebo formulation samples. The FTIR spectra of each sample were recorded in the transmission mode at a frequency range of 550–4000 cm<sup>−</sup>1. Each spectrum was an average of 32 scans combined in order to achieve a satisfactory signal-to-noise ratio. In all cases, spectra resolution was maintained at 8 cm<sup>−</sup><sup>1</sup> and the gauge force at 150. The compatibility of the samples was checked and FTIR spectra documented for further analysis.

### 3.7.4. Surface Area and Porosity Analyses

The surface area and porosity of optimized drug loaded and placebo formulations were quantified utilizing the Brunauer–Emmett–Teller (BET) analyzer (Micromeritics TRISTAR II 3020, Micromeritics, Norcross, GA, USA) employing nitrogen adsorption mechanisms. About 0.3 g of each sample was degassed under a vacuum environment overnight at 40 ◦C. The specific surface area for each specimen was calculated using the BET method with experimental points fixed at a relative pressure of 0.01–1.

### 3.7.5. X-ray Di ffraction (XRD)

The di fferences in the crystalline structures of PZA, excipient, drug loaded and placebo formulations were identified using an X'Pert Pro Powder X-ray di ffractometer (PANalytical, Westborough, MA, USA). Anode material used was copper based, machine divergence slit was set at 0.38 mm and measurements were performed using a reflection-transmission spinner. Measurement operations were carried out using 1.54 Cu K-alpha (1 and 2) radiation, 45 kV generator voltage and 40 mA tube current. Continuous scanning was performed at 0.026 scan step size and 126.99 s/step between 5◦ and 90◦ (2θ).

### *3.8. Surface Conformational Transitions of Dry and Hydrated Formulations*

First, the surface morphology of pure PZA, drug loaded and placebo formulations were viewed using the Zeiss Supra 55 SM Scanning Electron Microscope (SEM) (Carl Zeiss, Germany) at a 2 kV accelerating voltage. The samples were cut into small pieces, mounted on aluminum stubs using double sided adhesive carbon tape and then sputter coated with approximately 15 nm chromium using a Quorum T150 ES coater (East Sussex, UK) before imaging.

Afterwards, the changes in the surface geometry of the optimized drug loaded formulation upon hydration under biorelevant conditions, similar to that earlier described for the disintegration analysis, were studied to further corroborate previously observed disintegration and drug release patterns. At predetermined time intervals (10, 30, 60, 90, 120 s), photographs of the observed physical changes were captured, then remnants of the disintegrating formulation were carefully collected and dried to constant weight with a Labcon forced air circulation incubator at 25 ± 0.5 ◦C. Subsequently, dried remnants collected at the di fferent time points were processed as described above and mounted for viewing on a Zeiss Supra 55 SM Scanning Electron Microscope. Photomicrographs for both dry (whole) and hydrated samples were taken at 500× magnification.

### *3.9. Preliminary Organoleptic Evaluation*

A single blinded approach was used to evaluate taste acceptability and physical appearance of optimized drug loaded orodispersible formulation (each containing 25 mg PZA) by human volunteers (*n* = 5) [68,69]. Each volunteer was requested to allow formulations disperse in their mouths and to record the taste of each formulation on the provided chart after some seconds (under a minute) before removing formulation remnants from their mouths without ingestion. All panelists were provided with potable water to thoroughly rinse their mouths of any formulation residue using drinkable water before evaluating another sample (each panelist assessed 3 samples). The bitterness and quality attributes were evaluated using a 4-point hedonic scale with 1 point = very bitter, 2 points = moderate to bitter, 3 points = slightly bitter and 4 points = tasteless/taste masked). An average numerical value indicating the overall acceptability of the formulation was computed [68,69].

### *3.10. In Vitro Cytotoxicity Assay*

The PZA, PZA loaded and placebo optimized formulation were employed as samples for investigating the cytobiocompatibility using Hepatocyte cell line (Hep G2 also referred to as ATCC ® HB-8065 ™) was obtained from the American Type Culture Collection (Manassas, VA, USA). Two colorimetric assays were employed to quantify cell viability of the samples namely 3-(4,5-dimethylthiazol-2-yl) 2,5-diphenyl tetrazolium bromide (MTT) and neutral red (NR) cell viability assay.

### 3.10.1. Cell Culturing and Sample Preparation

The hepatocyte cell line was cultured in Dulbecco's modified Eagle's medium (DMEM), 10% fetal bovine serum (FBS), 1% L-glutamine, 1% non-essential amino acids (NEAA), and 1% penicillin/streptomycin. Tissue culture flasks (75 cm2) were used to grow the cell in an incubator maintained at 37 ◦C in 5% of carbon dioxide. The cells were harvested and passaged when they were confluent. Assay samples were dissolved in Dulbecco's modified Eagle's medium (DMEM) with serial dilutions (5, 0.5, 0.005, 0.0005 mg/mL) and prepared samples evaluated using MTT and NR assay detailed below.

### 3.10.2. MTT Cell Viability Assay

A modified technique outlined by Mosmann (1983) and Vistica (1991) was used for MTT viability assay [57,70]. HepG2 cells were seeded at a density of 40,000 cells/mL in a 96-well plate. The cells were left to be attached overnight, then they were exposed at di fferent concentrations (μg/mL) of the

samples. The spent medium was aspirated after 24 h of incubation at 37 ◦C and substituted in simple DMEM by a 0.5 mg/mL MTT. After another 3 h of incubation at 37 ◦C, the medium was removed and 200 μL DMSO dissolved the purple formazan crystals. A microplate reader (SpectraMax ® Paradigm ® Multi-Mode Detection Platform, Molecular Devices LLC, San Jose, CA, USA) measured the absorbance at 540 nm.

### 3.10.3. Neutral Red Cell Viability Assay

The media was aspired after the 24 h of incubation with the sample products, then adding 20 μL of the neutral red solution (Sigma Aldrich) to each well. The culture was incubated in a humidified chamber at 37 ◦C for 3 h (5% carbon dioxide). The cells were washed with pre-warmed PBS after incubation, followed by 200 μL of neutral red solubilization solution and left at room temperature for 10 min. A micro plate reader (SpectraMax ® Paradigm ® multi-mode detection platform) was used to measure absorbance readings at 540 nm. The cytotoxicity of both the MTT and NR results are reported as a percentage according to the following calculation:

$$\% \text{Cell viability} = \text{Sample} - \text{Blank} (\text{Control} - \text{Blank}) \times 100. \tag{1}$$

### *3.11. Stability Studies*

Environmental stability studies were performed on drug loaded formulations over a period of 12 weeks utilizing selected settings that were intended to simulate everyday use. Generic protocols set by the International Conference on Harmonization (ICH) were considered for the environmental conditions used for this preliminary investigation [71,72]. Samples were kept in airtight, glass jars containing desiccant bags and stored: (a) in a dark enclosure (23 ± 3 ◦C/65 ± 5% RH), (b) refrigerator (4 ± 2 ◦C) and (c) under room conditions (24 ± 3 ◦C/70 ± 5% RH) and tested in triplicate. Formulation weight, disintegration time, drug content uniformity, dissolution pH, inner and outer diameter were selected as indicators for determining the influence of set storage conditions on the physical and chemical stability of these samples. All stability indicators quantified at the end of 12 weeks were compared to measurements conducted at the point when the formulations were freshly prepared (time = 0 weeks).
