Development of a Multivariate Predictive Dissolution Model for Tablets Coated with Cellulose Ester Blends

Abstract

The focus of the present investigation was to develop a predictive dissolution model for tablets coated with blends of cellulose acetate butyrate (CAB) 171-15 and cellulose acetate phthalate (C-A-P) using the design of experiment and chemometric approaches. Diclofenac sodium was used as a model drug. Coating weight gain (X1, 5, 7.5 and 10%) and CAB 171-15 percentage (X2, 33.3, 50 and 66.7%) in the coating composition relative to C-A-P and were selected as independent variables by full factorial experimental design. The responses monitored were dissolution at 1 (Y1), 8 (Y2), and 24 (Y3) h. Statistically significant (p < 0.05) effects of X1 on Y1 and X2 on Y1, Y2, and Y3 were observed. The models showed a good correlation between actual and predicted values as indicated by the correlation coefficients of 0.964, 0.914, and 0.932 for Y1, Y2, and Y3, respectively. For the chemometric model development, the near infrared spectra of the coated tablets were collected, and partial least square regression (PLSR) was performed. PLSR also showed a good correlation between actual and model predicted values as indicated by correlation coefficients of 0.916, 0.964, and 0.974 for Y1, Y2, and Y3, respectively. Y1, Y2, and Y3 predicted values of the independent sample by both approaches were close to the actual values. In conclusion, it is possible to predict the dissolution of tablets coated with blends of cellulose esters by both approaches.