SOLUBILITY OF A NEW ANTIRETROVIRAL DRUG ( CRS 74 ) IN AQUEOUS ETHANOL MIXTURES

This study concerns a new antiretroviral drug named CRS 74, which has a limited bioavailability because of its low aqueous solubility and dissolution rate. To improve these properties, CRS 74 can be recrystallized by using Liquid Anti-Solvent (LAS) crystallization. Ethanol is chosen as the solvent under study for the molecule and water as the anti-solvent. Since solubility data is limited, it is necessary to collect experimental data for the molecule in relation to ethanol and water-ethanol mixtures at different temperatures in order to select suitable mixture compositions and temperature for LAS process design. In this work, the CRS 74 solubility measured in the temperature range 288.15 303.15 K in pure ethanol and in 95% water 5% w/w ethanol mixtures, and the CRS 74 solubility measured at 303.15 K in water-ethanol mixtures containing from 30 to 70% w/w ethanol are presented. Measurements were performed using the shake-flask method for generating the saturated solutions followed by compositional analysis by HPLC of the solution. The experimental data showed that the solubility of CRS 74 in binary hydroalcoholic mixtures increases upon increasing the temperature and mass fraction of ethanol. In order to better understand the behavior of the system and to estimate supersaturation conditions for a larger range of CRS 74 crystallization conditions, two models have been chosen to describe the experimental data: UNIQUAC and Jouyban-Acree models. The modeling of experimental solid-liquid equilibrium data proved that both models could correlate satisfactorily the solubility of the studied drug. This study provided valuable data for the recrystallization of CRS 74 by using the Liquid Anti-Solvent (LAS) crystallization process.

CRS 74 has a high biological activity as disclosed in PCT document WO 2005/111006 and US 7763733 (Bockelmann et al., 2005;Bockelmann et al., 2010) but its bioavailability is limited because of its low aqueous solubility and dissolution rate.Such properties imply difficulties not only in the design of pharmaceutical formulations, but may also result in biovariability.
Drug dissolution is a prerequisite for drug absorption and clinical response for almost all drugs admin-istrated orally.Solid forms that have been investigated for drug dissolution enhancement include salts, polymorphs and amorphous, among others.Amorphous formulations can achieve improved solubility, but the system is at serious risk of crystallizing to the thermodynamically stable form, even in the solid state (Yu, 2001;Rodríguez-Spong et al., 2004).Such transformations can compromise the performance of the formulation.
The Liquid Anti-Solvent (LAS) crystallization process is an attractive method for CRS74 recrystallization, aiming at controlling its dissolution kinetics.It requires mild conditions (ambient temperature and atmospheric pressure) with no requirement for expensive equipment.In the LAS process, crystallization of the solute is achieved by decreasing the solubility of the solid in the system.This is done by addition of a non-solvent component for the solute, called the anti-solvent and miscible with the solvent.
The LAS crystallization process for production of ultra-fine particles has been widely researched over the last few decades (Thorat and Dalvi, 2012).One goal in this crystallization operation is the control of product properties, e.g., particle size and particle size distribution (Zhao et al., 2007;Zhang et al., 2009), purity, residual solvent content, crystallinity, polymorphic form (Balani et al., 2010) and dissolution rates in specific media during in vitro-tests (Viçosa et al., 2012;Plakkot et al., 2011) to a certain extent.The product properties can be manipulated by changing the process parameters and formulation.
The solvent selection is one of the essential parameters to envisage any crystallization process.
Therefore, the knowledge of the solubility of a target component in different solvents is required.In this work, the solubility of CRS 74 in pure ethanol and ethanol-water binary mixtures was measured in the temperature range of 288.15K to 303.15 K at atmospheric pressure.
Although experimental data on solubility are essential to provide information about a system and help to understand its behaviour, correlations and prediction models are also required for the correct design of crystallization processes.Solid-liquid equilibria of ternary mixtures containing ethanol (solvent), water (anti-solvent) and the new antiretroviral drug were studied and the obtained solubility data were represented using two models: the UNIQUACbased model (Walas, 1985) and the Jouyban-Acree model (Jouyban-Gharamaleki and Acree, 1998).

Materials
The molecular formula of the active pharmaceutical ingredient (CRS 74) is C 46 H 66 N 8 O 5 S 2 and its molar weight 875.2 g/mol.Its chemical structure is presented in Figure 1.It was provided with 99% purity as a courtesy from Cristalia Ltda (Itapira, SP, Brazil).Ethanol (EtOH) (from Fluka Analytical, Sigma-Aldrich, France) and acetonitrile (ACN) high performance liquid chromatography (HPLC) grade (from Scharlau Chemie, Barcelona, Spain) were used as solvents with purity higher than 99%.

Solubility Measurements
The solubility of CRS 74 was determined by equilibrating an excess of CRS 74 in 5 g of water, ethanol and different ethanol/water combinations at a given temperature (from 298.15 to 303.15 K) in a temperature-controlled bath for 72 hours.Brazilian Journal of Chemical Engineering Vol. 33, No. 04, pp. 1073-1081, October -December, 2016 According to this method, the solid drug was added in excess to a certain medium, the flasks were sealed, and shaken at a predetermined time, 24, 48 and 72 hours.The saturation was confirmed by observation of the presence of un-dissolved material and measurements of dissolved drug concentration.The concentration was determined by removing the solid phase by filtration (0.22 µm pore size, Pall 4506, France) and injection of the filtered solution onto the HPLC system with analysis at a wavelength of 210 nm.The HPLC system consisted of an Agilent Chromatograph (Model1100 series) equipped with a UV-vis detector and a HPLC column ProntoSIL 300-5-ODSH 5 µm, 250x4 mm ID.The flow rate of the mobile phase (acetonitrile/water in the ratio of 70:30) was 1.0 mL/min.

Differential Scanning Calorimetric (DSC) Analysis
DSC measurement was carried out using a DSC-Q200 thermal analyzer (TA Instruments, France) in a temperature range of 293 to 483 K at a heating rate of 10 °C/min under nitrogen atmosphere (50 mL min -1 ) in order to determine the onset of the melting point and the fusion enthalpy with about 3 mg of solid.The molecule was characterized by a melting point of 461.75 K and a melting enthalpy of 86.6 J/g.

Modelling
Two models were chosen to describe experimental data: the UNIQUAC model (UNIversal QUASI Chemical) (Walas, 1985) and the Jouyban-Acree model (Jouyban-Gharamaleki and Acree, 1998).The first model is based on the calculation of activity coefficients and the second is based on the mathematical representation of the solubility of drugs in mixed solvents with respect to solvent composition and temperature.

UNIQUAC Model
In the UNIQUAC model, the activity coefficient is calculated by assuming a combinatorial contribution, essentially due to the differences in size and shape of the molecules, and a residual contribution essentially due to energetic interactions.The expression for the activity coefficient of the component i, i γ , is described by a combinatorial and a residual contribution: The combinatorial term C i γ is given by: ( ) where i x is the mole fraction of component i; z is the coordination number which is usually set to 10.The bulk factors i l are given by: ( ) ( ) i θ and i φ are the molecular surface area and mo- lecular volume of each component i, given by: For pure components, i r and i q are calculated as the sum of the group-volume and group-area parameters, k R and k Q : where k i υ (an integer) is the number of groups of type k in the molecule i.
The active molecule was decomposed into functional groups similarly to the group contribution UNIFAC method.The first parameters, volume and surface parameters ( k R and ) k Q , were estimated from the volume and surface area parameters of CRS 74 functional groups.The functional groups and subgroups are reported in Table 1.The thiazole heterocycle (C 3 HNS) was replaced by the thiophene one (C 4 H 2 S).The volume and surface area parameters r i and q i of the three components (water, ethanol and active molecule) are given in Table 2.The expression of the residual term is given by:

Brazilian Journal of Chemical Engineering
In Equation ( 8), the parameters ij τ are calculated from UNIQUAC binary interaction parameters ij u as follows: For a ternary system, twelve parameters are necessary: six parameters concerning the geometry of the molecules plus six binary interaction parameters.The binary interaction parameters ( ) ij u water/ CRS 74 and ethanol/CRS 74 were identified from solubility measurements.Binary interaction parameters water/ethanol are available in DIPPR (Data Compilation Tables of Properties of Pure Compounds) and reported in Table 3.
The binary interaction parameters of the UNIQUAC model were identified in two steps.Initially, solubility data for the binary ethanol/CRS 74 as a function of temperature were considered to estimate the two binary interaction parameters ethanol/CRS 74 from the activity coefficient of the solute calculated by Equation ( 10).
The temperature dependent experimental activity coefficient of the solute CRS 74, ( ) In a second stage, the binary interaction parameters of water/CRS 74 were calculated from experimental data for the ternary water/ethanol/CRS 74 with a mass ratio ethanol/(ethanol-water) of 5% and variable temperature.A ternary mixture was chosen because the solubility data in water as a function of temperature are not available.
In both cases, the objective function C of the curve fitting method used to estimate the binary interaction parameters was the weighted root mean square error, given by: The Jouyban-Acree model is represented by Equation ( 12):

RE CRS 74 Sol Temperature
First, the d erimentally f 288.15 to 3 From data oncluded tha ibrium after 2 ure ± 0.5 K. w was taken

Jouyban-Acree Model
The model parameter values were assessed by nonlinear curve fitting based on the least mean square method.The trained form of the Jouyban−Acree model for prediction of CRS 74 solubility in ethanol and water mixtures at specific temperature T is given in Equation ( 14 where * , m T w denotes the solute solubility in the binary solvent mixtures at temperature T (K) expressed in g/kg.This trained form of the model may be used to explore the predictability of the Jouyban−Acree model for the solubility of the tested drug in ethanol + water mixtures at various temperatures and solvent compositions.In addition, it may be used to predict the aqueous solubility of CRS 74 at various temperatures by considering 1 0.0 z = and 2 1.0 z = .Table 4 and Figure 2 and Table 5 and Figure 3 give respectively the solubilities calculated in pure ethanol and in 5% ethanol -95% water mixture as a function of temperature with the two models.In pure ethanol, the calculated concentrations are in good agreement with experimental values.In 5% ethanol -95% water mixture, the UNIQUAC model understimates the solubility of the active molecule.The Jouyban-Acree model seems to provide more accurate predictions.
Figure 4 and Table 6 present the calculated solubility at 303.15 K as a function of the ethanol mass proportion in the mixture ethanol-water with the two models.The data calculated with UNIQUAC model showed good agreement with experimental results.The two models revealed a maximum solubility of CRS 74 for an ethanol mass fraction in the mixture ethanol-water between 0.70 and 1.The maximum solubility for alcohol-water mixtures was predicted by the models but still not confirmed experimentally.A maximum solubility for a solute in a mixed solvent system has been observed experimentally for other systems (water/acetone/ketoprofen (Espitalier et al., 1995); water/ethanol/paracetamol and water/dioxane/ phenacetin (Ruckestein et al., 2003); water/ethanol/ hydrocortisone (Ali et al., 2009) and n-heptane/ ethanol/eflucimide (Teychene and Biscans, 2011)).Different studies with the Hidelbrand solubility approach have shown that the location and the height of the peaks could be linked with the polarity of the solute (Jouyban-Gharamaleki, 2000;Peña et al., 2006).Hydroxyl and amine groups of the solute give a polar character to this molecule that could explain the maximum of solubility calculated for a high ratio of ethanol in the mixture.
Indeed, the Jouyban-Acree model predicts a local minimum for the mixture 95% ethanol-5% water.To the best of our knowledge, this behaviour has not yet been described in the literature and should be treated with caution.

CONCLUSIONS
CRS 74 is soluble in ethanol (92.6 mg/g solution at 303.15 K).To the best of our knowledge, there are no other published data on the solubility of the new antiretroviral molecule (CRS 74) in pure ethanol and in ethanol-water mixtures.
The data generated in the present work represent a useful tool to define the mass proportion between solvent (ethanol) and anti-solvent (water) for our ongoing research on LAS recrystallization of CRS 74, in an attempt to improve its dissolution properties.The models demonstrate the capability to predict the solubility trend and magnitude, predicting a maximum in solubility in ethanol-water mixtures of around 115 mg/g solution as determined using the UNIQUAC-based model, or 112 mg/g solution using the Jouyban-Acree correlation.The data could be used for pharmaceutical purposes to explore a larger range of operating conditions for drug recrystallization, requiring less experimental measurements.
estimated from the melting enthalpy, Δ m h , its melt- ing temperature T m and the mole fraction at saturation, 74 CRS solutex , at a given solution temperature T: