Quality by design approach with multivariate analysis and artificial neural network models to understand and control excipient variability

Purpose

Although understanding the variability in the physicochemical properties of an excipient in a drug formulation is becoming an important aspect of the quality by design approach, few studies have reported the effect of excipient variability on design space. This study aimed to understand how the variability of excipient physicochemical properties caused by changes in manufacturer and grade influence the tablet quality.

Methods

In the quality by design approach, the formulation of the immediate-release tablet was optimized with a D-optimal mixture design. Subsequently, polyvinylpyrrolidone of different grades and manufacturers, which is used as a binder, was used to confirm the variability within the design space. The main cause of variability was identified by multivariate analysis, and a predictive model using an artificial neural network was developed to predict the dissolution profile.

Results

The design space greatly shifted the grade changes, mainly because of variabilities in the K values and particle size distribution that are strongly correlated with critical quality attributes. The predictive model accurately predicted the dissolution profile with low absolute and relative errors.

Conclusion

These findings highlight the importance of understanding and controlling excipient variability in the quality by design approach.

Read more

Kim, J.Y., Choi, D.H. Quality by design approach with multivariate analysis and artificial neural network models to understand and control excipient variability. J. Pharm. Investig. (2022).
https://doi.org/10.1007/s40005-022-00608-5