PBPK modelling of ceftriaxone Na-loaded starch-sodium alginate polymeric blend prepared by water-in-oil emulsification for oral delivery

Ceftriaxone is a third-generation cephalosporin antibiotic effective against many bacterial infections. However, owing to its instability in the gastrointestinal tract (GIT), it is administered by injections, which is an unfavourable route of administration. Therefore, the aim of this study was to formulate ceftriaxone into biodegradable and thermally stable polymeric blend microparticles that are suitable for oral delivery. The drug-loaded microparticles were prepared by the water-in-oil (W/O) emulsion method and consisted of starch and sodium alginate (NaAlg) as a polymeric matrix and glutaraldehyde (GA) as a cross-linking agent. Characterization of these particles using scanning electron microscopy (SEM) showed that the particles were spherical in shape with a smooth surface. Differential scanning calorimetry (DSC) and X-ray diffraction (XRD) of these particles showed no drug-polymer interactions. The highest percentage yield of particles was obtained at 3% polymer concentration. The particle size increased slightly after drug loading. The drug loading and entrapment efficiency appeared to increase with increasing polymer concentration. In vitro drug release at pH 1.2 and pH 7.4, revealed that drug release was below 20% at the acidic pH, while at pH 7.4, drug release of up to 85% was observed. The release mechanism followed first-order and Fickian diffusion patterns. Plasma concentration-time profiles were simulated for subcontinental Asian populations using commercial PBPK software, and the results suggest that microencapsulation of ceftriaxone sodium in a polymeric blend could represent a promising approach for controlled oral delivery of the drug, with enhanced absorption and bioavailability of the drug.


Introduction
Controlled drug delivery, which delivers the drug systemically or locally at a predetermined time for a prolonged period, ranging from days to months, has been described as one of the most effective ways of delivering poorly absorbed or gastric-unstable drugs (Debjit Bhowmik 1 * , Harish Gopinath 1, B. Pragati Kumar 1, S. Duraivel 1, 2018) [1]. It comprises a drug carrier, shielding the enzyme and acid-labile drugs, and tailoring the drug release rates. A broad range of carriers for oral drug delivery has been developed in recent years to provide a controlled release of drugs and maintain steady plasma levels for a prolonged period, with minimal adverse effects (Chien, 1992) [2].
containing tween-80 (2 ml) and magnesium stearate (0.5% w/v) to form H 2 O in oil (w/o) emulsion. The resulting emulsion was stirred at 600 rpm for 45 min, then 2 ml 25% glutaraldehyde solution containing 1.5 ml distilled water was added, followed by the addition of 1-2 drops of concentrated HCl. The mixture was stirred for 3 h to produce microparticles. The formed microparticles were filtered, using a 0.45 μm filter, and washed frequently with methanol and acetone to remove any excess surfactant and unreacted glutaraldehyde. Finally, the particles were dried overnight at 45°C in an oven and stored in a desiccator until further analysis. The details of the prepared microparticle formulations and their compositions are listed in table 1.

Characterization of starch-sodium alginate microparticles
An FT-IR study was performed to estimate the drug-polymer interactions and crosslinking to form a polymeric network. The spectra were recorded for pure starch, NaAlg, pure drug (ceftriaxone Na), blank, and ceftriaxoneloaded microparticles. Using 2 mg of the sample, KBR discs were prepared, and scanning was performed in the range of 4000-500 cm −1 at a resolution of 4 cm −1 .
XRD studies were performed using an XRD machine (Model JDX 3532, Japan) to evaluate the effect of microencapsulation on the crystallinity of the drug. XRD patterns of all the samples, including ceftriaxone sodium crystals, NaAlg, blank microparticles, and ceftriaxone sodium loaded microparticles, were recorded using CuKa radiation, Ni-filtered, 60 kV voltage, 50 mA current and 1°per minute scanning rate over a 10°to 40°diffraction angle.
The surface characteristics of the unloaded and ceftriaxone sodium-loaded microparticles were examined using scanning electron microscopy (SEM) (JEOL; Japan). Microparticles were deposited on carbon tape and then coated with gold using an SPI sputter module in a high-vacuum evaporator, and the SEM images were recorded at 20 kV.
To study the phase change and weight loss of ceftriaxone sodium, starch, NaAlg, blank microparticles, and ceftriaxone sodium-loaded microparticles were studied using thermal analysis (TGA, USA model Q600 Series). Before loading the samples, both the reference and sample pans were tarred. After sample loading, the oven was heated to 450°C at a rate of 15°C min −1 under a nitrogen atmosphere.
All formulations were further examined for their particle size distributions using a Zetasizer instrument (Zetasizer Nano-series ZEN3600, Malvern instrument Ltd with the software DTS-Nano, Kingdom).
Quantitative analysis of ceftriaxone sodium in starch -NaAlg microparticles The quantity of ceftriaxone sodium encapsulated within the microparticles was determined using an ultraviolet spectrophotometer (UV-1601, Shimadzu). Dry ceftriaxone sodium-loaded microparticles (50 mg) were ground, allowed to hydrate, and fully swell in 100 ml phosphate buffer (pH 7.4) for 24 h. After 24 h of soaking, the dispersion was filtered using a 0.45 μm filter, and suitable dilutions were made. The absorbance of ceftriaxone sodium was measured at 240 nm, and the concentrations were calculated using a calibration curve. To calculate percent yield of the microparticles, recovered dry drug-loaded microparticles were weighed and divided by the total amount of ingredients (drug and polymer) used initially in the preparation.
The formulas below were used for calculations of encapsulation efficiency, drug loading and percentage yield respectively. Rheological (Micrometric) properties of the microparticles The micrometric properties of the formed particles were evaluated using different techniques, such as angle of repose, bulk and tapped density, Hausner's ratio, and compressibility index. Details of the methods for conducting these tests are provided in the Supplementary Information.

In vitro drug release and release kinetic studies
The in vitro release pattern of ceftriaxone sodium from the loaded microparticles was determined using USP dissolution apparatus-2 (paddle apparatus) connected to an autosampler (Pharma Test PTFC 2, Germany). The specific weight of microparticles (corresponding to 250 mg of ceftriaxone sodium) was filled in a cellulose dialysis bag, tied to a paddle by a silk thread, and dipped in a dissolution flask containing 900 ml of dissolution medium to sink. The dissolution studies were performed for 12 h at pH 1. For the analysis of ceftriaxone sodium release kinetics, the zero-order, first-order, and Higuchi models were applied. The Korsmeyer-Peppas model was used to determine the drug release mechanism of the controlled drug delivery system. For the detailed mechanism of the drug release, preliminary 60% of release data were fitted to the Korsmeyer-Peppas equation below: where M 1 and M ¥ are the released drugs at times 0 and ¥. where, n is the diffusion constant. In spherical matrices, if n<0.43, diffusion is Fickian (case −1). If 0.43<n<0.85, it is a non-Fickian, and if n>0.85, the zero-order release mechanism dominates (case −2). Dissolution data were determined using DDSolver ® software.
Physiologically based pharmacokinetic studies Simulations for ceftriaxone Na pharmacokinetics were performed using PK-Sim ® and Mo-Bi ® version 9.1 (Bayer Technology Services GmbH, Leverkusen, Germany). Physicochemical properties of the drug that were entered into PK-Sim ® are molecular weight and solubility, which are 554.6 g mol −1 and 0.5 g l −1 respectively. The integration of these physicochemical properties helps creating a distribution model. First, PBPK model was developed in an individual Asian human and then extended to a population of 100 individuals. The demographic characteristics of the individuals are presented in table 2. An oral dose of 250 mg microparticles administered once was selected as the administration protocol. Particle dissolution was selected as the formulation, and the particle size distribution was monodisperse. The thickness of the unrestricted water layer of the particles was 20 μm, whereas the particle size was 10 μm. Parameter optimization was performed in MATLAB ® using the Levenberg Marquardt algorithm in the parameter identification toolbox. Clearance simulation for the drug was also conducted using a biliary clearance value of 0.070 ml min −1 kg −1 previously entered into the PK-Sim ® .

Results and discussion
Synthesis and characterization of starch-NaAlg microparticles A microparticulate system comprising starch and sodium alginate was developed for the sustained delivery of ceftriaxone sodium, combining two release approaches: biodegradation and time-dependent release. Microparticles were prepared using the emulsion cross-linking method. Both polymers are anionic, hydrophilic, and biodegradable, which means they can be mixed in all fractions to prepare microparticles by the emulsion crosslinking method. Under acidic conditions, the aldehydic groups of glutaraldehyde react with both the OH groups of starch and NaAlg to form hemiacetal and acetal rings during the crosslinking reaction. Hemiacetal and acetal ring formation between polymers and glutaraldehyde has been previously reported by El-Tahalway et al  [20,21]. Various factors, such as polymer and crosslinker concentrations, stirring speed, and stirring period, can significantly affect the morphology and particle size, in addition to the incorporation efficiency of the formed microparticles. The FT-IR spectra of the starch, sodium alginate, ceftriaxone sodium, blank starch-sodium alginate microparticles, and ceftriaxone-loaded microparticles are shown in figure 1(a). FT-IR spectrum of starch (black band) indicates two peaks at 673.21 cm −1 and 1007.29 cm −1 , which are for C-H bend and C=O bond stretching (Vedha Hari et al, 2012) [22]. Sodium alginate (red band) showed an intense peak at 1582.79 cm −1 owing to the bending vibration of the free carboxyl group. The peak at 1560 cm −1 is due to the asymmetric stretching of carboxylate ion intermolecular interaction through the formation of hydrogen bonds between COO 2 groups in the sodium alginate and OH groups in starch [23]. Peaks at 2925.04 cm −1 and 3500 cm −1 are due to the C-H stretching and OH groups of NaAlg (Çaykara et al, 2005) [24]. The FT-IR spectrum of the polymer microparticles without drug showed that the aldehyde groups of glutaraldehyde reacted with the OH groups of  [13,26]. In ceftriaxone FT-IR spectra peaks in the region of 1400 to 1450 are due to CH2 deformation and C=C stretching, but these are diminished in formulation FT-IR due to the formation of intramolecular chains of polymers in the same region [27]. The spectrum of ceftriaxone-loaded microparticles (orange band in figure 1(a)) showed that the starch-sodium alginate mixture did not change the chemical properties of the drug. Thus, the presence of distinct peaks other than starch and alginate confirmed the presence of the drug encapsulated in the polymeric particles in a stable form (Chattopadhyay et al, 2015; Saravanan & Rao, 2010) [28,29]. Peaks at 3495 and 3600 cm −1 in Alginate and starch are peaks of the natural polysaccharide hydroxyl groups, 3600-3200 cm −1 . The disappearance of these peaks in both drug-loaded and unloaded microparticles suggests the intermolecular interaction through the formation of hydrogen bonds between COO-2 groups in the sodium alginate and OH-groups in starch [30].
The PXRD patterns of starch, sodium alginate, ceftriaxone sodium, unloaded microparticles, and ceftriaxone Na-loaded microparticles are shown in figure 1(b). The diffraction peak of NaAlg at approximately 13.47°is the characteristic peak of the hydrated crystal-like structure, but its intensity decreased after the formation of the microparticles (Wang et al, 2010) [27]. The diffraction pattern of the unloaded microparticles indicated a wide hump in the region of 11°−24°. The consistency of the crystal structure might be due to the eggbox areas along the direction of the NaAlg chain, which is the aggregation direction. This shows the complex formation between the polymers and the decreased crystalline behaviour of NaAlg (Fontes et al, 2013) [31]. The XRD pattern of pure ceftriaxone sodium peaks was confirmed at diffraction angles of 13°, 15°, 24°, and 25°i ndicating that it was crystalline in nature (Mio Tangea [30,33]. The presented PXRD is for formulation A-15 that has a composition of 90% starch and 10% sodium alginate. From literature, we come to know that Starch has a semi-crystalline granular structure, so peaks at 30°and 32°in the X-ray analysis found in the microparticles can be assigned to the presence of starch crystals in low fractions as the peaks are not intense [34]. The scanning electron micrographs, Ci and Cii in figure 1, of the unloaded and drug-loaded microparticles, respectively, show that the synthesized starch-sodium alginate microparticles are spherical in shape, with smooth surfaces and no clusters. Ceftriaxone Na-loaded microparticles appeared to have larger particles than unloaded microparticles. This is because of the incorporation of the drug into the polymeric matrix, which limits the interior shrinkage of the polymers. In previous studies, microparticles with planar and non-porous surfaces were reported to be effective for extended dissolution (Das & Senapati, 2008) [35]. The NaAlg microparticles synthesized in this study were also discrete (figure. S1 in the Supplementary Information), which is a good flow criterion.
The DSC-TGA curves of the polymers, unloaded microparticles, drug-loaded microparticles, and pure drug are shown in figure 2. Thermal analysis of sodium alginate indicates that 1st endothermic peak at approximately 100°C was due to the evaporation of moisture ( figure 2(a)). The second and third exothermic peaks at 238.1°C and 368.3°C to 394.8°C belong to the degradation of the biopolymer and the formation of the corresponding carbonates (Devi & Kakati, 2013) [36]. DSC curve of pure starch is demonstrated in figure 2(b) where the first endothermic peak at around 27.21°C is attributed to moisture discharge, then weight loss (25 to 50% of the original weight) occurs in the second step within the temperature range of 317.26°C to 331.35°C, which is an indicator of the start of starch decomposition. The endothermic peak at 533.75°C shows that the decomposition of starch is rather slow, and 23% of its weight remains. This could be due to the partial replacement of the OH groups of the starch by bulkier groups, leading to a decrease in hydrogen bonds and the formation of stronger crystals (Biliaderis et al, 1986; Thiebaud et al, 1997) [37,38]. DSC-TGA thermogram of unloaded microparticles is shown in figure 2(c), which indicates that particles remained physically stable up to 533.77°C and weight remained as a residue (P41.5%). Comparing the results of the microparticles with those of the pure polymers, it can be concluded that the polymeric matrix is more thermally stable than the individual polymers (Baimark et al, 2010) [34]. The DSC-TGA curve of pure ceftriaxone sodium presented in figure 2(e) shows that the melting point of pure ceftriaxone was greater than 155°C. The first endothermic peak is at 25.15°C where the weight is 100%. Ceftriaxone sodium showed 27 to 30% weight loss during the 2nd stage temperature range (between 256 and 260°C), which is attributed to the dehydration process. The last endothermic peak appears at 533.79°C where weight loss was 46 .71% which is an indication of the complete decomposition of the drug (Manimekalai et al, 2017) [13]. The DSC-TGA curve of ceftriaxone sodium-loaded starch-NaAlg is shown in figure 2(d). In contrast to the pure drug, no sharp peak was observed for ceftriaxone sodium, indicating that the drug was uniformly distributed throughout the polymeric matrix. The endothermic peak at 335°C was attributed to the polymorphism and melting of ceftriaxone sodium. Additionally, this matrix remained stable up to 534°C with a total loss of weight of approximately 73.6% (Devi & Kakati, 2013; Fontes et al, 2013) [31,36].
Analysis of particle sizes using a Zetasizer showed that the average size of blank starch-sodium alginate microparticles was smaller than that of drug-loaded microparticles ( Table S.1 in the supplementary information). This means that drug loading slightly increased particle size, which can be associated with the attachment of ceftriaxone sodium molecules, which is suggested to reduce the open volume spaces in the polymer complex, hindering the interior shrinkage of microparticles (N. Patel et al, 2016a) [39]. Because the particle size can also be inversely affected by the stirring speed, the size of the particles decreases as the stirring speed increases. Various batches of formulations AS-1 to AS-15 (table 1) were fabricated at various stirring speeds (500 rpm, 600 rpm and 800 rpm), to establish the optimal stirring speed for preparing microparticles. Microparticles with optimal shape and size were formed at 600 rpm. When the agitation speed was maintained below 600 rpm, the polymer solution was not dispersed homogeneously, resulting in agglomeration

Drug incorporation, loading efficiency and percent yield of microparticles
The drug loading (%) and incorporation efficiency (%) of all formulations are shown in figure 3(a). A linear relationship was observed between the incorporation efficiency of ceftriaxone sodium and the concentration of the polymer. Drug loading (%) and incorporation efficiency (%) increased from 39% to 85% and 37% to 89%, respectively, as polymer concentration increased from 1% to 3% (table 1)  In-Vitro drug release and release kinetics The release of ceftriaxone Na from the microparticles was evaluated at two different pH values: acidic (pH 1.2) and basic (pH 7.4). The release profile from a representative formulation 'formulation AS-15' is shown in figure 4. A distinct difference in the drug release data between pH 1.2 and pH 7.4 is observed. At acidic pH, ceftriaxone sodium released was below 20%. This could be ascribed to the reduced swelling of the polymeric matrix at an acidic pH, which results in decreased matrix permeability, thus leading to limited drug diffusion. An increase in the drug release to more than 85% was observed at pH 7.4 in 12 h. These results indicate that the interpolymeric interaction that exists in the 'gel' form that retards the release of the drug in the dissolution medium at pH 1.2, while at pH 7.4, microparticles swell and the gel gradually breaks down to release the drug. Thus, the characteristic feature of ceftriaxone Na release from the polymeric matrix was a biphasic release pattern. The early fast release within the first 15 min is stated as the 'burst' effect, which is then followed by slower 1st order release kinetics. At a basic pH, deprotonation of sodium alginate causes a breakdown of the polymeric complex and completes the release of the drug as soluble ions. The in vitro release profile from all other formulations showed a similar pattern to that of AS-15 (figure S.2 in the supplementary information). As ceftriaxone Na is acid-labile, these results suggest that it can be successfully administered orally when incorporated into the polymeric matrix used in this study, with minimal degradation.
To evaluate the order of drug release from formulations AS-1 to AS15, various kinetic models were applied. The value of R 2 in zero-order and first-order for all tested formulations ranged from 0.85 to 0.91 and 0.95 to 0.988 respectively as demonstrated in table 3. The comparison of R 2 values for both the zero-and first-order models reveals that all the formulations follow first-order kinetics. R 2 for the Higuchi model and Hixson-Crowell ranged from 0.96 to 0.98 and 0.94 to 0.97 respectively. The AS-2 formulation exhibited the highest fit/ correlation for the first-order and Higuchi models. Based on these results, it can be predicted that all formulations follow first-order release kinetics and the Higuchi model, as suggested by the diffusion processes for the drug. The value of n in the Korsmeyer-Peppas model for all formulations was lower than 0.45, which means that all formulations followed Fickian diffusion (Desai, 2007; Raza et al, 2018) [47,48].

Pharmacokinetic analysis
Plasma concentration versus time profiles for ceftriaxone 250 mg microparticles administered orally (n = 1 and n=100) were created using the PBPK model and are shown in figure 5, whereas the pharmacokinetic parameters are displayed in table 4. Plasma concentration versus time profiles showed consistent absorption of the drug in microquantities, thus supporting the micro-release pattern of the microparticles.
In both, the individual and population simulation results, ceftriaxone plasma concentration reached about 1E-11 μmol l −1 (1.131E-12 μg ml −1 ) after 24 h and stayed steady at this level. This concentration is less than that achieved when the drug is administered to humans via i.v or i.m injections (  information). This could be attributed to a couple of reasons. First, the used ceftriaxone concentration in the simulation is less than that used in the i.v or i.m delivery reported. Second, it is well known that physiological barriers in the GIT can contribute to losing some of the drug during its passage, therefore, we suggest that the  amount of ceftriaxone given orally using the formulation proposed here to be increased in order to match i.v and i.m plasma levels reached in previous reports. Clearance simulation is also a part of the PBPK model, where specified values governing the clearance or metabolic aspects of drugs could be added, apart from built-in values known as the PK-Sim standard. The reported plasma clearance of ceftriaxone was 15.4 ml min −1 (at an IV dose of 500 mg (I. H. Patel et al, 1981) [44]), and the biliary clearance in PK-Sim ® was 0.070 ml min −1 kg −1 (Grime & Paine, 2013) [49]. According to the available mass balance information in literature, 30 to 60% of administered ceftriaxone is eliminated by biliary excretion, while the rest is through renal excretion in its unchanged form. Therefore, ceftriaxone is not an ideal metabolic candidate.

Conclusion
In this study, microparticles were fabricated using an emulsion cross-linking method. Microparticles were prepared using starch and sodium alginate as a polymeric blend and glutaraldehyde as a crosslinking agent. The highest % yield was obtained at 3% polymer concentration. FT-IR analysis of the microparticles indicated the formation of a polymeric network and successful entrapment of ceftriaxone sodium. XRD and DSC analyses of the microparticles indicated that the lower crystallinity of ceftriaxone Na resulted in the higher thermal stability of the prepared microparticles. A size study using a Zetasizer showed that all microparticles were in the micrometric range, and drug loading caused a slight increase in the particle size. Higher drug release in a basic medium (pH 7.4) was observed when compared to drug release in an acidic medium (pH 1.2). In vitro drug release studies indicated that microparticles can protect the therapeutic drug from the acidic environment of the stomach and can favourably deliver it at pH 7.4. The release kinetics of the ceftriaxone sodium-loaded microparticles followed first-order and Fickian diffusion. The reached plasma level through the simulated administration of ceftriaxone using PBPK modelling appeared to be less than that of the i.v and i.m administration reported in clinical research. This was attributed to different factors such as the low dose used here and the biological barriers that can aggravate drug loss when administered orally. Our results suggest that the PBPK software tool combined with biorelevant dissolution tests could be suitable for formulation development and analysis of unexpected in vivo results.