CENTRAL COMPOSITE DESIGN FOR FORMULATION AND OPTIMIZATION OF LONG-ACTING INJECTABLE (LAI) MICROSPHERES OF PALIPERIDONE PALMITATE

Objective: The aim of the present study was to optimize long-acting injectable (LAI) microspheres of Paliperidone palmitate (PP) for treatment of schizophrenia using face-centered central composite design (FC-CCD). Methods: In this study, poly lactic-co-glycolic acid (PLGA) based LAI microspheres of paliperidone palmitate (PP) were formulated by using FC- CCD. LAI microspheres were developed by using oil in water (O/W) emulsion solvent evaporation technique. On the basis of preliminary trials, FC-CCD was employed to check effect of independent variables such as drug polymer ratio (X 1 ), homogenization speed (X 2 ) and rate of addition (X 3 ). While mean particle size (Y 1 ), drug loading (Y 2 ), entrapment efficiency (Y 3 ), burst release (Y 4 ), and drug release on day 60 (Y 5 ) were considered as dependent variables and statistically evaluation performed by using design expert 12 software. Morphology of prepared microspheres was studied by using the scanning electron microscopy (SEM) technique, while particle size was analyzed by laser diffraction technique. In vitro drug release studies were performed using a controlled temperature shaking water bath apparatus. Fourier transforms infrared spectroscopy (FTIR) and differential scanning calorimetric (DSC) study were performed to analyze any changes in crystal behavior or to detect any chemical bonding between ingredients. 13 C NMR and 1 H NMR techniques were used to analyze end-capping and monomer ratio in developed microspheres. Results: The factorial batches mean particle size was found to be 38 µm to 104 µm and drug loading were found between 27.2 % to 47.2%. Mathematical modelling of drug release kinetics revealed that near zero-order drug release of checkpoint formulations. Endcap analysis and molar ratio of formulated microspheres were found to be ester end cap and ~75:25, respectively. Morphologically all the prepared samples were found to be spherical in shape and smooth surface. FTIR data showed no significant interactions occurred between drug and excipients. The actual responses of checkpoint formulations were observed within 5% variation of predicted values. Conclusion: The prepared microspheres showed promising results of morphology, particle size, drug loading, entrapment efficiency, burst release and drug release on day 60. The successful predictive designs models were achieved from employed FC-CCD.


INTRODUCTION
Long-acting injectable (LAI) dosage forms are becoming choice of medical practitioners to combat diseases which need long-term therapy for management and cure [1]. LAI's are being used for clinical management of various disease conditions like cancer, schizophrenia, opioid intolerance; diabetes, acromegaly, osteoarthritis etc [2]. Many platform technologies are employed in development of LAIs such as oil based suspensions [3,4], polymeric particulate systems [5,6], polymer gels [7], implant [8,9] and particulate suspensions [10]. Continuous research is ongoing to cater the unmet need of medical field for newer disease conditions and dosage designs.
Schizophrenia is mental disorder that adversely affects thoughts and daily life of patient. Schizophrenia is associated with impact on health, social, occupational, and economics. Sleep, hostility, depression and body mass index may be targeted to improve quality of life [11].
Numerous treatment options available for schizophrenia include oral dosages, injectable as well as long-acting injectable (LAI) as first generation and second-generation antipsychotics. Antipsychotic medications are also classified based on their mechanism of action as typical and atypical antipsychotics. Compared to typical antipsychotics, atypical induce fewer extrapyramidal side effects, but the exact neurobiological substrate of this difference is still unknown [12,13]. Antipsychotic drugs are available in market in various dosage designs researched based on need of patients being treated. Long-acting antipsychotic drugs has proven benefits in terms of patient compliance, guaranteed adherence to drug treatment, relapse prevention and clinical outcomes. Long-acting antipsychotics are becoming the preferred choice for maintenance therapy for schizophrenia. Long-acting injectable antipsychotics of aripiprazole, fluphenazine decanoate, haloperidol decanoate, risperidone, paliperidone, olanzapine pamoate etc. are available in market for intramuscular administrations as microspheres, suspensions, oily suspensions etc. [14][15][16].
Poly-lactic-co-glycolic acid (PLGA) is aliphatic polymer with a polyester backbone that is formed through the copolymerization of lactic and glycolic acid monomers. Engineering in lactic acid and glycolic acid ratio provides different release characteristics to the delivery system [17,18]. It is preferred because of its biocompatibility, biodegradability and mechanical strength and continues to be used to develop new controlled release systems [19].
Paliperidone palmitate is most recent second-generation atypical antipsychotic approved in the USA and the EU. Paliperidone palmitate is practically insoluble in water, has low partition coefficient (Log P= 1.8). It has limited oral bioavailability of 28% [20,21]. In order to achieve greater control over drug release from long acting release system, an attempt has been made to formulate paliperidone palmitate drug in a long-acting PLGA based polymeric platform using design of experiment methodology. Advantages of long-acting paliperidone palmitate microspheres are less frequent dosing, greater compliance to treatment, reduced relapse and improved bioavailability.

Materials
PLGA Resomer ® RG750S, inherent viscosity 0.8 to 1.2 dl/g) was obtained from Boehringer-Ingelheim. Paliperidone palmitate (PP) API was received from Neuland Pharma, India. Poly (vinyl alcohol) (PVA  was gift sample obtained from Merck KGgA, Germany. Methylene chloride (DCM), methanol (HPLC grade), ortho-phosphoric acid (HPLC grade), acetonitrile (HPLC grade) were of avantor performance materials, USA. Milli-Q water was used for all chromatography studies. Water for Injection used for all formulation activities.

Preparation of long-acting injectable (LAI) microspheres
Paliperidone palmitate-loaded LAI microspheres were developed by the oil-in-water (O/W) emulsion solvent evaporation technique [22,23]. In this method, weighed quantity of Paliperidone palmitate was dissolved in dichloromethane by vortexing (Vortex shaker, Remi, India) and cool to 5 °C. PLGA was added and dissolved in drug solution by vortexing to form organic Phase. Polyvinyl alcohol (Molecular weight: 30-70 KD) was added and dissolved in 2000 ml of water for injection by stirring (Eurostar 20, IKA Germany) at 750 rpm for 6 h and cool the PVA solution to 5 °C. Emulsification was carried out by in-line high shear homogenization (Megatron MT300, Kinematica AG, Switzerland). Oil phase was added to the aqueous phase through the addition port of homogenizer by a peristaltic pump (Remi, India) at specific rate of addition and homogenization speed. Then, Emulsion was transferred to 5 liter of aqueous PVA solution and stirred at 300 rpm under vacuum for 8 h to evaporate methylene chloride and hardening of microspheres. Microspheres were separated by means of filtration using Whatman paper. Microspheres were washed three times using 3-liter water for injection. Microspheres were dried by freeze-drying.

Experimental design
Face centred central composite design (FC-CCD) was used to optimize process parameters in this study. The details of independent and dependant variables used in this design are shown in table 1. In this, design three independent factors were evaluated, each at two levels and experimental trials were performed for seventeen formulations including 3 center points as shown in table 2. There were four major factors affecting the formulation drug: polymer ratio (X1), homogenization speed (X2) and rate of addition (X3). While mean particle size (Y1), drug loading (Y2), entrapment efficiency (Y3), burst release (Y4), drug release on day 60 (Y5) were considered as dependent variables. Design expert version 12 (Stat-Ease, USA). The statistical software was used for data treatment and generating visualization. Effects were considered statistically significant if value of P ≤ 0.05.

Morphology
The surface morphology of formulated microspheres was examined by scanning electron microscopy (SEM) (Hitachi S-3700N, Japan) at 15 KV with an appropriate magnification.

Particle size and size distribution
Particle size and Particle size distribution of the microspheres were determined using a laser diffraction technique (Malvern 3000, Malvern, UK). The particles were suspended in 0.1% Tween 20 and counted using a laser sensor. The average particle size was expressed as volume mean diameter in microns ( m).

Drug loading
Drug loading was determined by using HPLC technique. 62.4 mg of paliperidone palmiate-loaded PLGA microspheres added in 100 ml volumetric flask. Add the 80 ml diluent (Isopropyl alcohol, acetonitrile, water and ortho-phosphoric acid in the ratio of (50:25:25:0.1, %v/v) in a volumetric flask and sonicate for 5 min. Make up the volume upto 100 ml with diluent and mix well. Solution was filtered through PVDF 0.45 μm syringe filter and the drug concentration was determined using HPLC system equipped with UV absorbance detector (280 nm wavelength). Mobile phase consist of a mixture of buffer pH-2.2 and acetonitrile (25:75, %v/v). Intersil C18 (4.6 mm × 250 cm, 5µ) analytical column was used with flow rate of 1.5 ml/min and injection volume of 20μl. Chromatographs were analysed by empower chromatography system.

In vitro release testing
In this study, weighed quantity of microspheres (i.e. 6 mg drug equivalent) was placed in flat bottom conical flasks filled with 200 ml phosphate buffer saline of pH 7.4 with 0.02% polysorbate 20. Conical flasks were placed in a temperature-controlled shaking water bath [24](SW23, Julabo, Germany) set at 70±2 rpm, 37±0.5 °C. At predefined sampling points, 1.5 ml of release sample was withdrawn and was replaced with fresh release medium. Testing was performed in triplicate (n = 3). Drug concentrations in release samples were quantified spectrophotometrically at 280 nm for Paliperidone (Shimadzu UV 1800, Shimadzu Japan).

Differential scanning calorimeter (DSC) analysis
Thermal behaviour and glass transition temperature of paliperidone palmitate API, PLGA 75:25, Physical Mixture (1:1), Placebo and formulated microspheres using differential scanning DSC (DSC60 Shimadzu, Japan). Samples were heated at a rate of 1 °C/min from 25 °C to 140 °C. The thermograms were analysed to determine the glass transition temperatures and melting point.

Fourier transform infrared (FT-IR)
FT-IR spectrum of Paliperidone palmitate API, PLGA 75:25, physical mixture (1:1), placebo and formulated microspheres were measured in the solid-state as potassium bromide dispersion using a FT-IR spectrometer-430 (Shimadzu 8400S, JAPAN). Each sample was properly grounded in mortar and pestle with potassium bromide in ratio of 1:100 (sample: Potassium Bromide). Powders were compressed into discs with the help of hydraulic compression press at 12 tonnes for 5 min. The FT-IR scan range was ranging from 4000 to 500 cm −1 .

Proton nuclear magnetic resonance ( 1 H-NMR)
1 H-NMR spectroscopy analysis was performed to determine the Lactide: Glycolide (L: G) molar ratio of the PLGA 75:25 and formulated microspheres. Sample was dissolved in deuterated chloroform (CDCl3) solvent (0.9 ml) at 500 MHz. Peak intensities at 5.2 ppm and 4.8 ppm was considered for proton counting. The integration at 5.2 ppm represents a single H of Lactide, while the integration at 4.8 ppm represents 2H of Glycolide. The mole fractions of Lactide (ML) and Glycolide (MG) were estimated from peak integration (P) of each component marked with 'L' for Lactide and 'G' for Glycolide, respectively [25].

Carbon-13 nuclear magnetic resonance ( 13 C NMR)
End group capping of PLGA and prepares microspheres were analysed using 13 C NMR [25]. Total of 12,000 scans were acquired over 12.5 h. Ester end-cap was confirmed by the existence of a peak at ~14 ppm, which correlates to an end-methyl unit on a long alkyl chain.

Mean particle size (Y1)
Particle size analysis was performed by the principle of laser diffraction and results are represented in table 2. Composition and manufacturing process parameters were observed to show an impact on the particle size and particle size distribution of prepared microspheres. ANOVA data in table 3 for response mean particle size (D50) depicts that factor X1 (Drug: Polymer ratio), X2 (Homogenization speed) and X3 (rate of addition) are statistically significant main effects at level 0.05. Two-way interaction of X2 (Homogenization speed) and X3 (rate of addition) found to significantly impacting mean particles size at level 0.05. The lack of fit F-value of 6.40 implies that the lack of fit is not significant relative to pure error. The predicted R 2 of 0.9316 is in reasonable agreement with an adjusted R 2 of 0.9609 i.e. the difference is less than 0.2. The signal-to-noise ratio of 29.536 indicates adequate signal. Regression analysis data for response mean particle size showed in Contour plot and three-dimensional surface plot of mean particle size (D50) shows the impact of variation of factor values of on mean particle size ( fig. 1). The homogenization speed (X2) and rate of addition (X3) of organic phase to aqueous phase found to significantly impacting the mean particle size of prepares microspheres. The increase in homogenization speed (X2) and rate of addition (X3) reduces the mean particle size of microspheres ( fig. 1).

Fig. 1: Graphical representation of data-mean particle size (Y1)
Batches with homogenization speed of 1000 rpm and 4000 rpm showed varied particle size of microspheres. Homogenization speed of 1000 rpm resulted in microspheres with the mean particle size of 92µm, 68µm, 104µm, 77µm and 92µm. While, homogenization speed of 4000 rpm yield microspheres with the mean particle size of 51µm, 46µm, 38µm and 48µm. Particle size reduction was achieved during manufacturing of microspheres by high-speed homogenization through rotor/stator generators with two teeth rows, each having tip speed of 10.0 to 31.0 m/s. Higher homogenization speed yield lower mean particle size. Larger the size of microspheres lower will be the surface area per unit volume and rate of encapsulated drug release is reduced [26]. Particle size distribution of prepared Paliperidone microspheres showed in fig. 2.    3). The increase drug: polymer ratio (X1 found to decrease the drug loading of prepares microspheres. The increase in homogenization speed (X2) and rate of addition (X3) reduces the drug loading of microspheres ( fig. 3).

Fig. 3: Graphical representation of data-drug loading (Y2) Entrapment efficiency (Y3)
Entrapment efficiency is a crucial parameter in the design of drug loaded PLGA microspheres. The obvious aim of dosage development is to entrap maximum drug in polymer matrix. Drug: polymer ratio was found to directly impacting the entrapment of drugs inside the PLGA matrix. As well, mechanical parameter such as homogenization speed (X2) was found to impact the entrapment efficiency of microspheres.
ANOVA data in table 3 of response entrapment efficiency depicts that X2 (Homogenization speed) and X3 (rate of addition) are statistically significant main effects at level 0.05. Two-way interaction of X1 (Drug: Polymer ratio) and X2 (Homogenization speed) found significantly impacting entrapment efficiency at level 0.05. The lack of fit F-value of 3.27 implies that the lack of fit is not significant relative to the pure error. The predicted R 2 of 0.8941 is in reasonable agreement with an adjusted R 2 of 0.9720 i.e. the difference is less than 0.

Fig. 4: Graphical representation of data-burst release (Y4)
Contour plot and three-dimensional surface plot of burst release vs. drug: polymer ratio, rate of addition ( fig. 4) shows that decrease in drug: polymer ratio and rate of addition decreases the burst release of resultant microspheres. Results of burst release of experimental run 1 to 17 showed that burst release was observed higher in high drug-loaded microspheres. Therefore, higher drug loading may not always be advantageous to get desired drug release profile. In the contrary, the lower mean particle size of microspheres increases the exposed surface area of microspheres and increases the burst release of drug from microspheres. It is important to achieve the optimum mean particle size and optimum drug loading to get a drug release profile suitable for long-acting injectable formulations. It can be achieved by properly optimizing various process parameters during the development of long-acting injectable microspheres. In these experimental trials, lower burst release was obtained with run

Fig. 5: Graphical representation of data-drug release on day 60 (Y5)
Contour plot of drug release on day 60 vs. drug: polymer ratio, homogenization speed ( fig. 5) shows that drug: polymer ratio and homogenization speed increases drug release on day 60 of resultant microspheres increases. Release from polymeric microspheres occurs due to diffusion and/or homogeneous bulk erosion of the biopolymer. Release rate impacted by physico-chemical parameters of encapsulated drug. Hydrophobicity, size of molecule, partition coefficient, and charge has greater influence on drug release of encapsulated drug from polymer matrix [27. 28]. Water uptake in the polymer matrix leads to hydrolysis of polymer. Monomers and oligomers diffuse out of polymer matrix through pores resulting in drug erosion from polymer matrix.

Morphology
Surface morphology of microspheres was studied by SEM technique revealed that shape of prepared microspheres was found spherical with smooth surface as shown in fig. 6. No broken microspheres were observed in SEM analysis. Smooth surface of microspheres is important to get good flow properties of microspheres and have industrial importance during filling of microspheres in vials. SEM of all the formulation showed spherical shape of prepared. Particle shape was studied under different magnifications. Overall, the process technology and parameters were found capable to produce spherical shape of microspheres consistently in all the runs.    Lactide: glycolide ratio 1 H NMR spectroscopy technique was used for determination of lactic acid to glycolic acid ratio of PLGA polymer and microsphere formulation. Deuterated chloroform was used as a solvent for NMR testing. LA: GA ratio was analysed for PLGA and optimized microsphere formulation to check compatibility with excipients and solvents used and to study any impact of processing parameters during microspheres manufacturing. Fig. 9  End cap analysis 13 CNMR spectroscopy technique was used for end group analysis of PLGA polymer and microsphere formulation. Peak assignments are shown in fig. 10 (A, B). Peaks at 16 ppm, 20 to 30 ppm, 50 to 80 ppm and 160 to 180 ppm are ascribed to carbon. Peaks at 16 ppm and in between 160 to 180 ppm indicate the presence of ester carbons in 13 C NMR spectra. Based on 13 C NMR spectra of PLGA and prepared microspheres, it can be concluded that end group characteristics were remained unaltered in microsphere formulation.

Selection of optimum conditions
The optimum conditions for manufacturing of PP microspheres were obtained using software numerical and graphical optimization using Design Expert® software, version 12.0 [29]. The desirability function scale operates between 0 and 1. Desirability value of 1 represents a fully desired response. The desirability for each response was achieved by setting the goals in ranges [30]. To get microspheres with optimum characteristics, goals for dependent variables were fixed as 50 to 100 µm (for mean particle size), 24.1 to 47.2 % (for drug loading), 70.4 to 95.04% (Entrapment efficiency), 0.09 to 2.5 % (for burst release) and 80 to 90 % (for drug release on day 21). After treatment of responses of all seventeen formulations, the software has provided one hundred solutions having combinations of independent variables to achieve desirability function of 1. Fig. 11 represents overlay plot obtained graphical optimization respectively. Overlay plot was used to visually identify an area where the predicted means of one or more response variables are in an acceptable range for successful manufacturing of PP microspheres. A.

B.
C.   The observed values for mean particle size of CPF-1, CPF-2 and CPF-3 were found to be 61.04 µm, 58.44 µm, 53.18 µm against predicted values of 62.40 µm, 56.28 µm and 54.91 µm, respectively. Drug release from sample CPF-1, CPF-2 and CPF-3 was studied. All three check point formulations showed a consistent drug release profile from microspheres. Drug release from PLGA based microspheres mainly governed by diffusion and/or erosion. The cumulative in vitro drug release profile of PP from microspheres showed in fig. 12. It showed that burst release from prepared microspheres was minimal. Burst release was followed by lag phase and constant release phase. mo is the initial drug amount; m is the amount of drug remaining at a specific time; k is the rate constant; t is the time; r 2 is correlation coefficient; n is diffusion coefficient Mathematical modelling of drug release of CPF-1, CPF-2 and CPF-3 showed near zero-order drug release from prepared PP microspheres. The observed values of all responses were compared with predicted values and shown in table 5. The percentage error in observed values and predicted values were found to be less than 5 %. The samples of CPF-1, CPF-2 and CPF-3 showed spherical shape and smooth surface. The data of verification of model revealed that developed models were well suited for the manufacturing of PP microspheres.

CONCLUSION
In the presented experimental work, PLGA based biodegradable microspheres formulation of PP was optimized using face centred central composite design (FC-CCD) using desirability function. The developed model was confirmed as definitive for deriving optimal independent variable values and successfully predicting responses for the manufacturing of PP microspheres. The responses obtained from seventeen formulation runs were showed strong effect of independent process variables on formation of PP microspheres. The observed responses of checkpoint formulations were found to be in close agreement with predicted values. The PP microspheres showed predominant zero-order drug release kinetics. The prepared microspheres were found to be spherical in shape and smooth in surface. DSC and FTIR studies showed that the ingredients used in formulations were compatible. LA: GA ratio of prepared microspheres was determined to be 75:25 using 1 H-NMR. 13 C NMR studies confirmed ester end cap of prepared PP microspheres. Therefore, it can be concluded that a three-factor, three-level Box-Behnken design can be successfully used for the optimization of PP microspheres using emulsion solvent evaporation technique with minimum number of experimental runs.