Multivariate Optimization Applied to the Synthesis of Meglumine Antimoniate with Low Levels of Trivalent Antimony

Meglumine antimoniate (MA) is a pentavalent antimony (Sb) drug recommended for the treatment of leishmaniasis. It is known that the trivalent antimony (Sb) present as a residue in MA contributes to the drug side effects. In this article, multivariate optimization was used in the synthesis of MA in order to obtain a drug with low levels of Sb. Four variables (source of antimony, temperature, water volume and pH) were preliminarily evaluated by 2 fractional factorial design. Central composite design (CCD) was used to determine the optimal synthesis conditions, using two different sources of Sb and the significant variables selected in a fractional factorial design. Response surface methodology obtained by CCD provided a model with non-significant regression (p = 0.05) for the synthetic route via KSb(OH)6. On the other hand, synthetic route via SbCl5 reached minimum value of Sb content of 0.172% and significant regression, and it was selected for further evaluations. The analysis of MA formulations synthesized with SbCl5 under optimized conditions revealed the efficiency of multivariate optimization to reduce Sb content. In addition, the monitoring of some physicochemical parameters of these formulations maintained at 40 oC for 90 days, showed that stability was not altered at 95% confidence level.


Introduction
Leishmaniasis is a neglected tropical disease caused by protozoa parasites from over 20 Leishmania species (Trypanosomatidae), transmitted by the bite of over 90 infected female sandfly species (Phlebotomine). 1 There are three main forms of the disease: cutaneous leishmaniasis (CL), the most common form, which causes skin lesions; mucocutaneous leishmaniasis (MCL), leading to partial or total destruction of mucous membranes; and visceral leishmaniasis (VL), a fatal disease if not treated appropriately. It is estimated that 700,000 to 1 million new cases occur annually worldwide. 1 In the 1940s, two pentavalent antimonials started being used to treat the disease, and they are still the most widely used drugs, particularly antimony sodium stibogluconate, and meglumine antimoniate (MA). [2][3][4][5] These compounds should be administered parenterally daily (typically 20 mg Sb per kg per day for 20-30 days, not exceeding 850 mg of Sb). 2,4,5 Antimonial therapy is frequently accompanied by local pain during parenteral injections and systemic side effects, requiring very close medical supervision. Typical side effects include nausea, vomiting, weakness, myalgia, abdominal cramps, diarrhea, rash, hepatotoxicity and cardiotoxicity. 2,4,6 It is generally accepted that Sb III present as a residue in pentavalent antimonials or produced in tissues through the reduction of Sb V into Sb III , [7][8][9][10][11] is responsible for their side effects and antileishmanial action. 4,12 Residual Sb III may also be related to the development of antimonial drug resistance. Studies on the mechanism related to Sb III action suggested that it compromises thiol homeostasis by depleting intracellular glutathione and inhibiting glutathione reductase. 11,13,14 Trivalent antimony increases oxidative stress and leads to apoptosis by increasing reactive oxygen species (ROS). 11,[14][15][16] As an attempt to reduce the side effects of antimonial therapy, important efforts have been devoted to the preparation of less toxic compounds, 4,17 as well as the development of oral and topical formulations. 18,19 Demicheli et al. 17 proposed two new synthetic processes to produce MA, using KSb(OH) 6 or SbCl 5 as antimony sources. The MA obtained from the former was found to be less cytotoxic in vitro than the MA obtained from the latter and by the commercial drug. 20 Moreover, KSb(OH) 6 led to a lower level of apoptosis in the liver after parenteral administration in mice with VL, when compared to commercial drug. 12 It was then proposed that the lower toxicity of MA prepared from KSb(OH) 6 may be due to the lower amount of Sb III residue. 12,20 Comparative tests in a murine model of VL showed that oral synthetic formulations (300 mg Sb per kg of body weight per 12 h for 30 days) were as efficacious as commercial drug used parenterally (80 mg Sb per kg of body weight per 24 h for 30 days). 21 The efficiency of MA could be associated to the concentration of Sb V and its toxicity related to the presence of Sb III as a contaminant. Thus, it is important to determine the concentration of these two species as minimum quality parameters for drugs used in the treatment of leishmaniasis. For this purpose, a method for inorganic Sb speciation at concentrations in the µg L -1 range using hydride generation associated with atomic absorption spectrometry (HG-AAS) was recently developed by our group. 22 The traditional univariate approach has been consistently used to optimize the parameters that affect the performance of reactions, requiring a large number of experiments and, consequently, increasing reagent and time consumption. Furthermore, possible interactions between variables are ignored, so that the true ideal conditions may not be achieved. On the other hand, the multivariate optimization technique uses experimental design to efficiently obtain the best desirable characteristics, using a smaller number of experiments, understanding the interactions between variables and providing statistical models.
In this study, the experimental conditions for the synthesis of meglumine antimoniate were developed, using multivariate optimization, with the purpose of obtaining a drug with lower Sb III contents for a safer and effective leishmaniasis treatment. Among experimental design methodologies, fractional factorial design 23,24 allows to understand the effects of a greater number of factors with fewer observations. Central composite design (CCD), 25 as a response surface methodology (RSM), was useful in modeling and optimizing the effective parameters of the synthesis. This study also focused on the development of candidate formulations for oral and topical administration, establishing minimum quality parameters for drugs used globally in leishmaniasis therapy, evaluating not only the concentration of Sb III generated, but also the stability of the product.

Instrumentation
The temperature of the MA synthesis reaction was controlled with a Polyscience (71, Niles, USA) immersion recirculating water bath heater pump and the solution was mixed using a Fisatom apparatus (752a, São Paulo, Brazil) with magnetic stirring. The pH measurements were performed using a Hanna pH meter (HI 2221, Woonsocket, USA). The hydrated SbCl 5 was centrifuged for 7 min at 4,600 × g in an Eppendorf equipment (5430 R, Hauppauge, USA) for the precipitation of Sb pentoxide. A Büchi Labortechnik AG rotary evaporator (R-114, Flawil, Switzerland), equipped with a B-480 water bath, was used to concentrate MA formulations. An Eletrolab climate chamber (EL 101/1, São Paulo, Brazil) was used for accelerated drug stability tests. For mass measurements, an AUX220 Shimadzu balance (Tokyo, Japan) with a precision of ± 0.0001 g was used.
HG-AAS measurements were performed using a SpectrAA-240 atomic absorption spectrometer (Agilent Technologies, Santa Clara, USA), equipped with a VGA-77 continuous flow hydride generator accessory. Antimony cathode lamps (Agilent Technologies) were used at wavelengths of 217.6 nm, with a 0.5 nm spectral bandpass and a current of 7 mA. High purity argon was used as a purge gas at a flow rate of 90 mL min -1 . A quartz tube cell was heated under the flame and used for Sb atomization.
In the preparation of MA formulations for topical use, the gelling agent hydroxyethyl cellulose (HEC), or Natrosol © , was obtained from Merck KGaA, (Darmstadt, Germany) and propylene glycol was obtained from Vetec Química Fina (Duque de Caxias, Brazil).
High-purity reagents were used in all experiments and, in the preparation of aqueous solutions, ultrapure water was used, obtained with a Direct-Q 3 system (Millipore, Burlington, USA) with resistivity of 18.2 MW cm.

Synthesis procedure for meglumine antimoniate
The MA was synthesized from an equimolar mixture in water of NMG and Sb V . 17 Two different synthetic routes were evaluated as a function of the Sb V source: (i) SbCl 5 : the hydration of SbCl 5 was carried out with ultra-pure water in an exothermic reaction, yielding hydrochloric acid and antimony pentoxide. 26 The mixture was centrifuged for 7 min at 4,600 × g to precipitate Sb 2 O 5 . The supernatant was discarded and the precipitate was transferred to a round-bottom flask immersed in the water bath under magnetic stirring. After the solution reached the bath temperature, NMG was added.
(ii) KSb(OH) 6 : KSb(OH) 6 was dissolved in ultra-pure water in a round-bottom flask dipped in the water bath under magnetic stirring at 80 ºC. After complete dissolution, the solution was cooled to the specific temperature of the synthesis and NMG was then added.
In both synthetic routes, after the addition of NMG, the solution turned whitish color. The reaction immediately began with a sudden decrease in pH. If the beginning of the reaction was not observed and the pH of the solution became basic due to NMG solubilization, 1.0 mol L -1 HCl would be added until pH reached a value close to 7. Under this experimental condition, the reaction immediately began. The pH was increased with the addition of 1.0 mol L -1 KOH, keeping the value recommended for synthesis throughout the reaction. The pH was controlled until the solution cleared and the pH stabilized. After the end of the reaction, heating and stirring were switched off and the solution was allowed to cool down. Acetone was added in the amount of 3 times the volume of the resulting solution. After precipitation of MA, supernatant was discarded and the precipitate was dried at room temperature.

Optimization strategy for the synthesis of meglumine antimoniate
The optimization of the synthesis was aimed at the obtention of MA compounds with low levels of Sb III . As several variables can influence the result, a screening experiment was performed according to fractional factorial design, 23,24 to determine the experimental variables that significantly influence the result using a reduced number of experiments.
The significance of the effects was estimated by the Pareto chart that shows the absolute values of the standardized effects. Standardized effects are t-statistics that evaluate the null hypothesis that the effect is 0 through the p-value. This value represents the probability that the effect of a variable is caused exclusively by random error. Thus, considering the maximum α probability that would be subject to risk that the value of the effect is confused with its error, if p ≤ α or |t calc | ≥ t α , the effect of the corresponding variable is significant.
After the screening experiments, RSM was performed to determine the optimal condition of MA synthesis, using two different sources of Sb V (SbCl 5 and KSb(OH) 6 ). The factor that presented no significant effect on the response was fixed, and the factors which affected the response were optimized simultaneously by application of CCD. 25 The exploration of each synthetic route included a total of 2 k + 2k + n experiments , where k is the number of factors studied, 2 k are the points of the factorial experiments carried out in the corners of the cube, 2k are the points carried out on the star centered on the face and n is the number of experiments performed at the central point of the experimental domain. The repetition of the central points was used to estimate the variance. The adequacy and significance of the quadratic model were assessed by analysis of variance (ANOVA), using Fisher F-test. The significant terms for each response were those that presented a p-value greater than F at 95% confidence level.
All experiments were performed randomly. The software Statistica version 10 was used to model the experimental designs. 27

Meglumine antimoniate formulations for oral and topical treatment
After the optimization of the experimental conditions of MA synthesis to achieve a low level of Sb III residue, formulations were designed for oral and topical treatment, by choosing simple and scalable processes. The MA compounds were synthesized according to the protocol described in the "Synthesis procedure for meglumine antimoniate" subsection, replacing the precipitation step by adjusting the concentration in a rotary evaporator equipment in the values of 300 and 850 mg Sb mL -1 for oral and topical formulations, respectively. To prepare the topical formulations, a 1:1 v/v mixture of propylene glycol and water containing 2% m/v hydroxyethyl cellulose was added to the MA solution (850 mg Sb mL -1 ) at a proportion of 1:1 v/v and the mixture was kept under agitation at 60 ºC for 30 min.
To assess stability in terms of the formation of toxic sub-products (derived from Sb III ) and physicochemical properties, such as pH, clarity of solutions and osmolarity (only for oral formulations), different formulations were prepared and monitored for a period of 45 and 90 days with a climatic chamber maintained at 40 ºC, after initial analysis. Multiple comparison tests using analysis of variance (ANOVA) were performed using the Statistica 10 software. 27

Optimization of meglumine antimoniate synthesis
The optimization of MA synthesis was based on the study developed by Demicheli et al., 17 using compounds from 0.004 mol of Sb V (SbCl 5 or KSb(OH) 6 ) and NMG. In the optimization of MA synthesis conditions, four experimental variables and their interactions were investigated on the formation of Sb III , namely: source of Sb V (1); temperature (2); water volume (3); and pH (4). In this sense, a 2 4-1 fractional factorial design was performed and each assay was run in duplicate (2 4-1 = 8 assays × duplicate = 16 experiments). The experimental design matrix developed and the results obtained as %Sb III related to the amount of total antimony present in the synthesized MA compounds in each experiment are shown in Table 1. The MA compounds obtained in these experiments showed average levels of Sb III , ranging from 0.1415 to 0.8955%.
The Pareto chart (Figure 1) shows the standardized effects (t-statistics) represented by bars and the limit of significance at t 0.05 with 8 degrees of freedom. According to the Pareto chart, only the pH variable (4) was not significant for the response at 95% confidence. The Sb III residue was higher when SbCl 5 , the lowest level (−), was used as an antimony source. The step that contributed to the high level of Sb III in MA compound prepared with SbCl 5 was probably the hydration of this reagent, in which SbCl 5 was added to a volume of ultra-pure water in an exothermically reaction. Additionally, the highest levels (+) of the volume and temperature variables contributed to the increased Sb III as a residue.

CCD design for MA synthesis
Although the screening experiment indicated the use of KSb(OH) 6 was the most significant factor for obtaining MA compounds with lower levels of Sb III , two considerations were relevant to keep the synthetic route using SbCl 5 . The first was related to the evidences that MA compounds obtained from SbCl 5 showed better results in the leishmaniasis treatment tests for the VL murine model, orally. 21 The latter was due to changing in the color of the aqueous solutions of the MA compounds obtained via KSb(OH) 6 after a few weeks, indicating a possible degradation process. To decrease the contribution to the formation of Sb III during the SbCl 5 hydration step, the procedure was changed to the dripping of ultra-pure water to a volume of SbCl 5 , generating a smaller amount of heat.
Response surface models were developed for each synthetic route, considering only the significant factors in 2 4-1 fractional factorial design. The significant variables, volume (V) and temperature (T), were used for the construction of a CCD matrix with four experiments  in the factorial design (2 k ), four experiments in the axial points (2k) and with three replicates in the central point (n), resulting in a total of eleven experiments. The pH parameter was not significant on the response, so it was maintained between 6 and 7 during all CCD experiments. The experimental design matrix and the results obtained as %Sb III related to the presence of antimony in the synthesized MA compounds in each experiment of the synthetic routes via KSb(OH) 6 and via SbCl 5 are shown in Table 2.
Applying the CCD method, it was observed that the relationship between the response obtained and the variables followed the quadratic model expressed by the polynomial equations for the synthetic route via SbCl 5 (equation 1).
Sb III (%) = 2.257 -0.0748V + 1.48 × 10 -3 V 2 -0.0596T + 4.98 × 10 -4 T 2 + 5.86 × 10 -4 VT (1) Figure 2 shows the response surface of the model obtained for the %Sb III property in MA compounds by synthetic routes via SbCl 5 . Analyzing that surface, it can be concluded that the Sb III content in MA compounds reached a minimum value close to the central point, more specifically in the volume coordinate equal to 15.2 mL and in the temperature coordinate equal to 50.9 ºC. It is worth mentioning that it is impracticable to keep the temperature unchanged throughout the synthesis, but it is important to use a bath with temperature control with a resolution of ± 0.1 ºC. Likewise, a volume meter with an uncertainty of 0.1 mL should be used.
The ANOVA analysis showed that the regression was not significant for the model obtained in the synthetic route via KSb(OH) 6 , which can be explained by the low dispersion of the results for the analyzed levels. Thus, it was selected the central point as an optimal condition since lower Sb III content was obtained in this condition ( Table 2). The quality of the models can be seen in Table 3. The evaluation of the model obtained via SbCl 5 synthetic route showed that the regression was significant while the lack-of-fit was nonsignificant. In addition, the explained variance values were above 0.90, showed that the quadratic model is valid for this study. The signal-to-noise ratio was measured by the adequate precision, which comprises the value predicted at the points of the project and the average forecast error. In this study, the adequate precision values obtained were satisfactory, due to the signal-to-noise ratio higher than 4.

Stability studies of oral and topical candidate formulations
During the optimization of MA synthesis, it was observed that aqueous solutions changed their color for

Conclusions
This study demonstrated the applicability of multivariate optimization for the synthesis of meglumine antimoniate to obtain a formulation with low Sb III contents for a more efficient and safe leishmaniasis treatment.
Candidate formulations for oral and topical administration prepared with SbCl 5 were obtained under optimized conditions and presented Sb III contents close to the theoretical value. Multiple comparison tests using ANOVA for each studied parameter (Sb III content, pH and osmolarity) showed that no significant changes in the formulations during 90 days were observed, confirming the stability of the monitored MA formulations at 40 ºC. On the other hand, MA formulations prepared with KSb(OH) 6 showed changing in the color after a few days of preparation, which means possible indication of instability and degradation process.