Application of Box–Behnken design to prepare gentamicin-loaded calcium carbonate nanoparticles

Abstract

The aim of this research was to prepare and optimize calcium carbonate (CaCO₃) nanoparticles as carriers for gentamicin sulfate. A chemical precipitation method was used to prepare the gentamicin sulfate-loaded CaCO₃ nanoparticles. A 3-factor, 3-level Box–Behnken design was used for the optimization procedure, with the molar ratio of CaCl₂: Na₂CO₃ (X₁), the concentration of drug (X₂), and the speed of homogenization (X₃) as the independent variables. The particle size and entrapment efficiency were considered as response variables. Mathematical equations and response surface plots were used, along with the counter plots, to relate the dependent and independent variables. The results indicated that the speed of homogenization was the main variable contributing to particle size and entrapment efficiency. The combined effect of all three independent variables was also evaluated. Using the response optimization design, the optimized Xl–X3 levels were predicted. An optimized formulation was then prepared according to these levels, resulting in a particle size of 80.23 nm and an entrapment efficiency of 30.80%. It was concluded that the chemical precipitation technique, together with the Box–Behnken experimental design methodology, could be successfully used to optimize the formulation of drug-incorporated calcium carbonate nanoparticles.

Keywords::

• Box–Behnken
• calcium carbonate
• gentamicin sulfate
• nanoparticles

Introduction

Pharmaceutical nanoparticles are fine drug-loaded particulates with sizes ranging between 10–1000 nm, and properties that are different from those of their original materials (Adibkia et al. 2007, Lu et al. 2009, Hallan et al. 2014, Daraee et al. 2014). These nanosystems can advantageously improve the therapeutic efficacy of a drug by enhancing its bioavailability (Mohammadi et al. 2011, Lotfipour et al. 2011, Adibkia et al. 2012, Chaudhary et al. 2015) and serum stability (Bowman and Leong 2006).

Gentamicin, an aminoglycoside, inhibits protein synthesis by binding to the 30 S ribosomal subunit of bacterial cells (Yoshizawa et al. 1998). Similar to other aminoglycosides, gentamicin has a relatively short half-life and low bioavailability, and may result in side effects such as nephrotoxicity (Tange et al. 1995). Different kinds of gentamicin-loaded nanoparticles have been previously prepared using some polymers such as poly(lactide-co-glycolide) (PLGA) (Abdelghany et al. 2012), chitosan (Ji et al. 2011), and caroboxymethyl dextran-b-poly(ethylene glycols) (Soliman et al. 2010).

Bone infectious disease—known as osteomyelitis—is initiated by a variety of pathogens, most commonly Staphylococcus aureus. This pathogen can penetrate endothelial, epithelial, and osteoblastic cells (Almeida et al. 1996, Wesson et al. 1998) and provide a reservoir of bacteria in the infected area. Considering this, targeting the pathogen using the antibiotic may be more important for treating chronic bone infection than physically eliminating only colonized pathogens (Bost et al. 1999). Local application of antibiotics could eradicate the microorganisms more effectively, without systemic toxicity (Richards et al. 2006).

Therapeutic applications of calcium carbonate (CaCO₃) nanoparticles in modern medical systems have attracted the attention of researchers, due to their great potential and capabilities. This material is low-cost, biocompatible, safe, and accessible. According to a research by Combes et al., CaCO₃-based cements showed no cytotoxic effect on the osteoprogenitor cells of human bone marrow (Combes et al. 2006). The use of CaCO₃ nanoparticles as a controlled release system for different drugs has been reported by different investigators (Qian et al. 2011, Wang et al. 2006). According to research studies, CaCO₃ materials can be suitable to increase the rate of resorption of biomedical cement and initiate its replacement by bone tissue (Fujita et al. 1991). Therefore, a dual therapeutic system (drug release and bone substitution) could be obtained by development of the antibiotic-loaded resorbable bone-filling materials using CaCO₃. A direct contact between the bone and calcite CaCO₃ without interposition of soft tissue at the interface has been reported as well (Ohgushi et al. 1992). Aragonite CaCO₃ has also been used for designing anticancer drug carriers and scaffolds for bone repair and tissue engineering (Islam et al. 2012).

Response surface design is a collection of statistical procedures to analyze the effects of several independent variables. It mostly involves the central composite design and the Box–Behnken design (Ghanbarzadeh et al. 2014, Thompson 1982). Compared to the central composite design, Box–Behnken designs contain fewer design points, and they can be less expensive to implement than central composite designs with the same number of factors (Thompson 1982).

The optimization of preparation methods for drug-loaded CaCO₃ nanoparticles to be used as biodegradable and biocompatible material was investigated in this study. Considering the great potential of CaCO₃ in drug delivery, its bone-related advantages, as well as the benefits of nanoparticle-based drug delivery systems, this study may be extended to the delivery of gentamicin sulfate into bone infections such as osteomyelitis. The effect of three independent variables on the entrapment efficiency and particle size of gentamicin sulfate–CaCO₃ nanoparticles, that is, the molar ratio of CaCl₂:Na₂CO₃, drug concentration, and the speed of homogenization, was evaluated using a 3-level, 3-factor Box–Behnken experimental design.

Materials and methods

Materials

Gentamicin sulfate was supplied by Zahravi Co (Tabriz, Iran). Calcium chloride, sodium carbonate, and ninhydrin were purchased from Merck Company (Darmstadt, Germany).

Preparation of gentamicin sulfate–CaCO₃ nanoparticles

The chemical precipitation method was used to entrap gentamicin sulfate into CaCO₃ nanoparticles (Ueno et al. 2005). Concentrations were selected based on previous studies for the development of CaCO₃ nanoparticles (Ueno et al. 2005, Qian et al. 2011, Haruta et al. 2003) (Table I). To prepare the gentamicin sulfate–CaCO₃ nanoparticles by the precipitation method, 100 μl of aqueous solution of the drug and 100 μl of aqueous solution of CaCl₂ were moderately mixed for 20 min, followed by the addition of 100 μl of aqueous solution of Na₂CO₃, and the solutions were stirred together vigorously for 15 min. Next, 5 ml of distilled water was added into the system and the large particles were discarded. The final suspension was centrifuged (Eppendorf AG 5810R, Germany) at 12,000 rpm for 5 min, to separate the nanoparticles. Based on the experimental design, the responses for fifteen formulations were obtained (Table II).

Table I. Factor-setting for the experimental design.

Factors\Codes	− 1	0	1
Molar ratio of CaCl2: Na2CO3 (X1)	1:1	3:1.5	5:2
Drug concentration (X2)	1 g/dl	3 g/dl	5 g/dl
Speed of homogenization (X3)	2000 rpm	7000 rpm	12,000 rpm

Table II. Particle size (PS) and percent entrapment efficiency (EE %) values for fifteen formulations.

Formulation code	Molar ratio of CaCl2:Na2CO3 (X1)	Drug Concentration (X2)	Speed of Homogenization (X3)	PS ± SD (nm)	EE% ± SD
1	−1	0	1	280.00 ± 0.5	37.0 ± 1.0
2	−1	0	−1	269.30 ± 1.7	34.2 ± 4.7
3	0	−1	1	259.00 ± 2.3	33.2 ± 5.6
4	−1	1	0	300.12 ± 1.6	42.1 ± 1.8
5	0	1	−1	280.00 ± 0.9	35.0 ± 10.7
6	1	1	−1	80.00 ± 13.2	24.0 ± 15.7
7	0	1	1	280.20 ± 0.6	33.0 ± 1.0
8	0	−1	−1	210.00 ± 1.6	26.1 ± 9.0
9	0	0	0	120.30 ± 1.8	24.2 ± 1.7
10	0	0	0	109.00 ± 1.1	23.5 ± 0.7
11	−1	−1	0	200.00 ± 7.1	25.3 ± 0.2
12	1	0	1	109.63 ± 1.9	24.5 ± 4.7
13	1	1	0	89.00 ± 1.0	27.5 ± 0.7
14	1	−1	0	87.40 ± 0.1	20.0 ± 0.2
15	0	0	0	100.00 ± 1.1	23.0 ± 0.9

Characterization of the nanoparticles

Particle size

The values for mean particle size for all the formulations were determined by using a particle size analyzer (SALD 2101, Shimadzu, Japan). All measurements were performed in triplicate. Mean particle sizes for the samples are shown in Table II.

Entrapment efficiency

After dissolving 100 mg of the prepared gentamicin sulfate–CaCO₃ nanoparticles in EDTA 0.5 M (pH 7.5) (Ueno et al. 2005, Qian et al. 2011), the amount of gentamicin loaded was detected by the colorimetric method (using ninhydrin as the reagent, in 1.25% w/v aqueous ninhydrin solution) using the UV spectrophotometer (UV1800 Shimadzu, Japan) (Frutos et al. 2000). A previously prepared standard curve (linear in the range of 5–80 mcg/ml, y = 0.0121x + 0.0154, R² = 0.9992) was used to determine the entrapment efficiency, based on the following equation:

$$\text{Entrapment efficiency \% = }\left(\text{drug content}/\text{total drug}\text{added}\right)\times 100$$

Each set of data for entrapment efficiency was reported as an average of three calculations (Table II).

Experimental design

A 3-factor, 3-level Box–Behnken design was utilized in the present study (Table I). The molar ratio of CaCl₂:Na₂CO₃ (X₁), the concentration of drug (X₂), and the speed of homogenization (X₃) were chosen as the independent variables. The size of the nanoparticles and their entrapment efficiency were considered as dependent variables. The experimental design set for all fifteen formulations predicted is shown in Table II. The effects of the independent variables on responses were investigated by means of mathematical equations, counter plots, and surface plots. Analysis of variance (ANOVA) was used to evaluate the results, and the final equations were shortened to include only significant parameters (P < 0.05). The response optimization approach was also applied to obtain gentamicin-loaded CaCO₃ nanoparticles with desired particle size and entrapment efficiency. Response optimization helps to identify the combination of variables to optimize a single response or a set of responses.

Results and discussion

Effect of the factors on particle size

The particle size is one of the most important parameters that is strongly considered in the development of nanoparticulate drug delivery systems. The particle sizes of the different batches of nanoparticles prepared in this research were in the range of 80.0–300.12 nm. The relationship between the independent variables and particle size was clarified using Equation 1, contour plots, and response surface plots. Equation 1 shows the relationship between the different variables and particle size as the first response:

(1) $${\text{Y}}_{\text{PS}}\text{=109}\text{.76+11}{\text{.19X}}_{\text{1}}\text{+24}{\text{.11X}}_{\text{2}}\text{-85}{\text{.42X}}_3\text{+81}{\text{.56X}}_1{}^2\text{+65}{\text{.96X}}_2{}^2\text{-12}{\text{.2X}}_1{\text{X}}_2-24.{\text{63X}}_2{\text{X}}_3$$(1)

where PS denotes the particle size, and X₁, X₂, and X₃ represent the molar ratios of CaCl₂:Na₂CO₃, the drug concentration, and the homogenization speed respectively.

The equation clearly indicates that the particle size values are strongly dependent on the independent variables selected. The particle size is mostly affected by the homogenization speed (X₃). The negative sign for coefficient of homogenization speed (X₃) indicates that the particle size decreases noticeably following an increase of homogenization speed (p = 0.0001).

The graphs, which evaluate the effects of independent variables on the particle size, are shown in Figures 1–3. As seen in Figure 1, an increase in the homogenization speed from 2000 rpm up to 12,000 rpm caused a decrease in the particle size from 300.12 nm to 80.0 nm. Enhancing the homogenization speed increased both mechanical and hydraulic shear, which in turn could have effectively reduced the particle size (Nakashima et al. 2009). In a research study by Ueno et al., the particle size of betamethasone phosphate-loaded CaCO₃ nanoparticles was decreased when the stirring speed was multiplied (Ueno et al. 2005). Similar observations were also reported by other researchers (Maa and Hsu 1996).

Figure 1. Counter plot and response surface plot for the effects of the homogenization speed and drug concentration on particle size.

Figure 2. Counter plot and response surface plot for the effects of the homogenization speed and the molar ratio of CaCl₂:Na₂CO₃ on particle size.

Figure 3. Counter plot and response surface plot for the effects of the drug concentration and the molar ratio of CaCl₂:Na₂CO₃ on particle size.

The effect of drug concentration is also illustrated in Figure 1. The result shows that the size of the particles was enlarged when the drug concentration was enhanced. As reported in previous research studies, the viscosity of the media is enhanced by increasing the drug concentration, and the effect of viscosity on the particle size is obviously in contradiction to the effect of shear stress (Masson et al. 2011, Noori Koopaei et al. 2014, Kheradmandnia et al. 2010).

The ratio of Ca²⁺:CO₃²⁻ has been reported to affect the particle size significantly (Chen et al. 2011). Nevertheless, in the present research, the molar ratio of CaCl₂:Na₂CO₃ showed no reasonable effect on the particle size. However, according to both graphs in Figures 2 and 3, in addition to other factors such as homogenization speed and drug concentration, the minimum particle sizes were obtained at the intermediate molar ratios of CaCl₂:Na₂CO₃.

Effect of the variables on entrapment efficiency

The entrapment efficiency for all the formulations was in the range of 20.0–42.1% (Table II).

(2) $${\text{Y}}_{\text{EE\%}}\text{=23}\text{.50+4}{\text{.11X}}_2\text{-5}{\text{.35X}}_3\text{+4}{\text{.72X}}_1{}^2\text{+3}{\text{.57X}}_2{}^2$$(2)

Where EE stands for entrapment efficiency

According to Equation 2 and Figure 4, homogenization speed had the effect of lowering entrapment efficiency. Considering a number of reports, this effect could be attributed to drug diffusion out of the particles during the size reduction at the high homogenization speeds (Yoncheva et al. 2003, Patel et al. 2013). Conversely, the drug concentration had a positive effect on entrapment efficiency. At low homogenization speeds, increasing the drug concentration from 1 g/dl to 5 g/dl led to improvement in the entrapment efficiency from 20.0% to 42.1% (Figure 6). As revealed previously, an increase in the drug concentration leads to enhanced viscosity of the reaction media, thereby decreasing the shear stress. A decrease in the shear stress could lengthen the diffusion pathway, which in turn may cause an increase in the drug content (Görner et al. 1999).

Figure 4. Counter plot and response surface plot for the effects of the drug concentration and the homogenization speed on entrapment efficiency.

It can be seen from Equation 2 that the molar ratio of CaCl₂:Na₂CO₃ (X₁) had no significant effect on entrapment efficiency individually (p > 0.05). However, at low homogenization speeds and high values of the molar ratio of CaCl₂:Na₂CO₃, entrapment efficiency was increased to levels above 35% (as illustrated in Figures 5 and 6).

Figure 5. Counter plot and response surface plot for the effects of the drug concentration and the molar ratio of CaCl₂:Na₂CO₃ on entrapment efficiency.

Figure 6. Counter plot and response surface plot for the effects of the homogenization speed and the molar ratio of CaCl₂:Na₂CO₃ on entrapment efficiency.

The correlation coefficients (R²) and adjusted R² were calculated to be 0.95 and 0.87 for Y_PS, and 0.94 and 0.84 for Y_EE% respectively. The lack of fit was non-significant for Y_PS (p = 0.213) and Y_EE% (p = 0.412), representing that the models could properly explain the variability of the observed responses.

Response optimization of the nanoparticles

The optimization plot with appropriate particle size and acceptable entrapment efficiency is shown in Figure 7. Coded units for the optimum point were defined as − 0.52, 0.89, and 1, which gives the values of 2:1.25, 4.89 g/dl, and 12,000 rpm for the molar ratio of CaCl₂:Na₂CO₃, concentration of drug and the homogenization speed, respectively. Drug entrapment efficiency and particle size for the optimized point were predicted as 26.49% and 90.12 nm respectively.

Figure 7. Optimization plot for the preparation of gentamicin sulfate–CaCO3 nanoparticles. The optimum points (PS = 90.12 nm, EE% = 26.49%) were defined as − 0.52, 0.89, and 1, which gives the values of 2:1.25, 4.89 g/dl, and 12,000 rpm for the molar ratio of CaCl₂:Na₂CO₃, concentration of drug, and the homogenization speed respectively.

Validation of the model

To evaluate the validation of the model, three more experiments were carried out in duplicate as a test set. The percentage error (PE) between calculated and observed values for each response was calculated by means of Equation 3. Details of the three formulations as test sets are summarized in the Table III.

$$\text{PE=}\left(\left|\text{Calculated-Observed}/\text{observed}\right|\right)\times 100$$

Table III. Percent error (PE) and mean percent error (MPE) between calculated and observed values for each response.

Set number	Calculated *PS (nm)	Observed PS (nm)	PE for PS	Calculated *EE%	Observed EE%	PE for EE%
1	300.00	320.10	6.25	40.35	37.0	9.05
2	210.02	244.22	14.00	32.21	28.09	14.66
3	80.23	96.50	16.86	39.32	33.64	16.89
			MPE = 12.37			MPE = 13.53

*No statistical analysis has been performed for calculation of PS in the validation step and the reported values have been estimated according to the equation in “validation of the model” section.

Consequently, optimized formulations were prepared according to the optimized point (particle size: 90.12 nm, entrapment efficiency: 26.49), and the PE between observed and calculated amounts was determined. Particle size and entrapment efficiency for optimized formulation were 80.23 nm (PE = 12.32%) and 30.80% (PE = 13.99%) respectively, which are considered to be acceptable (Acton 2012). According to our results, the observed values were close to the calculated ones, and hence the fit of the model was confirmed. Particle size distribution for the optimized formulation actually prepared is shown in Figure 8.

Figure 8. Particle size distribution for the optimized formulation actually prepared.

Conclusion

Particle size and entrapment efficiency are important characteristics to be considered in the formulation of nanoparticles. The present study focused on the preparation of gentamicin sulfate–CaCO3 nanoparticles using the Box–Behnken statistical method. The optimized formulation was obtained with a relatively limited number of experimental runs, using the proper experimental design and an optimization approach. The combined effect of all factors considered led to the development of an optimized formulation, with a drug entrapment efficiency of 30.80% and a particle size of 80.23 nm. It was concluded that the chemical precipitation technique could be used along with the Box–Behnken experimental design methodology to determine the significant variables and optimum conditions for preparation of drug-incorporated calcium carbonate nanoparticles.

Acknowledgement

The authors would like to thank the Research Vice Chancellor of Tabriz University of Medical Sciences, for financial support provided for the study. This article is a part of a thesis (No. 96) submitted for the PhD degree in the Faculty of Pharmacy, Tabriz University of Medical Sciences.

Declaration of interest

The authors report no declarations of interest. The authors alone are responsible for the content and writing of the paper.