(M1030-04-25) Predictive Model to Determine the Aqueous Solubility of BCS Class IV Drugs in Amorphous Solid Dispersions

Purpose: Solubility of a compound is a crucial parameter for drug absorption, and poor aqueous solubility is a major challenge for the oral delivery of drugs[1]. To improve the oral absorption of drugs with poor solubility, various solubility enhancement techniques are used. Amorphous solid dispersions (ASDs) in which the drug is molecularly dispersed in a polymer are being used to improve the solubility of drugs. Experimentally determining the enhanced solubility is time consuming, laborious and might end up having no significant results. Predictive models could improve the efficiency of this effort [2]. An empirical model was investigated that could predict the solubility of BCS (Biopharmaceutical classification system) class IV drugs from their ASDs.

Methods: Preparation of ASDs: The ASDs of the five BCS class IV drugs (acetazolamide, chlorothiazide, furosemide, hydrochlorothiazide and sulfamethoxazole) and four water-soluble polymers (polyvinylpyrrolidone, copovidone, hydroxypropyl methyl cellulose and Soluplus) were prepared by hotmelt method. Each drug-polymer combination has a varying polymer ratio (1:1, 1:5.5, 1:10) yielding a total of 60 compositions. ASDs were prepared by taking the desired ratio of the drug and polymer in a porcelain crucible, thoroughly mixed and heated 5°C above the drug melting point. After the mixture is molten it is cooled to room temperature. The obtained mass was pulverized and sieved. Characterization of ASDs: The thermal behavior of ASDs was characterized using differential scanning calorimetry (SII EXSTAR 6000 DSC). Drug polymer interactions were studied by using Thermo Scientific Nicolet Summit Pro Fourier Transform Infra Red spectroscopy. Equilibrium aqueous solubility of all the ASDs were determined using Shake flask method at 25°C. Model Development: Multiple linear regression was used to develop an empirical predictive model for solubility. The molecular properties of drug and polymer were used as the predictive variables and the experimentally obtained solubility was the response variable. To develop the model, statistical software SPSS 28.0.1.1(15) version was used. The developed model was validated by leave one out method.

Results: ASDs of all compounds have exhibited significant (P < 0.05) increase in equilibrium aqueous solubility. The DSC thermograms confirmed the existence of drug in amorphous form and FTIR analysis showed no significant interaction between the drug and polymer (fig 1). The empirical model suggested that the solubility of drugs in ASD depends on the drug to polymer ratio, the difference in the solubility parameter of the drug and polymer and the molecular weight of the drug, and resulted in the following predictive: Log S = 3.008 + (0.054*PR) + (0.047*Δ δ) – (0.006*M.W_D); Where S = Experimental solubility, PR is polymer ratio,Δ δ is solubility parameter, M.W_D is Molecular weight of drug. The present model has an R value of 0.79 and a P value less than 0.001. The regression equation obtained was used to calculate the predicted solubility of the ASDs and the average percentage error in prediction was 9%. The model was validated using the leave-one-out method. A second validation was also performed using training set data and the mean square error (MSE) of the training set was 0.09, which indicates the goodness of fit and prediction with an accuracy greater than 90%.

Conclusion: The empirical model developed for the BCS class IV drugs predicts their solubility in ASDs with an accuracy greater than 90%.

References: 1. Lipinski CA. Drug-like properties and the causes of poor solubility and poor permeability. Journal of Pharmacological and Toxicological Methods. 2000; 44(1):235-49.
2. Jain P, Yalkowsky SH. Prediction of aqueous solubility from SCRATCH. International Journal of Pharmaceutics. 2010; 385(1-2):1-5.

Fig-1: FTIR spectra (left) and DSC thermograms (right) of Furosemide-HPMC ASDs

Normal Q-Q Plot of Log Experimental Solubility