Development, Comparison, and Qualification of Predictive Pharmaceutical Stability Models Using Independent Statistical Tools

Abstract

Purpose

The predictive modeling approach to assess long-term stability performance of pharmaceuticals by using short-term accelerated stability is of significant value to accelerate development timelines, enhance stability confidence, and improve product quality and regulatory compliance. Herein, the head-to-head quantitative comparison of predictive stability models developed by two independent statistical tools was conducted as a unique approach to assess and qualify the model parameters, the statistical tools, and stability predictions.

Methods

The moisture-modified Arrhenius equation and two independent statistical tools including ASAPprime^® and JMP^® software were utilized to develop predictive pharmaceutical stability models for a humidity independent case study and a humidity dependent case study.

Results

Various temperature and humidity stress conditions were utilized to develop stability models with ASAPprime^® and JMP^® softwares to provide a reasonably accurate fit as the coefficient of determination R² was not less than 0.99. Many statistical tools including leverage plot, p value, three-dimension plot in JMP^® models were employed to provide unique visual extrapolation. ASAPprime^® model offered database of packaging and excipient to enable assessing package protection, which JMP^® model lacked. The prediction outcomes of the stability models were later confirmed by the independent long-term stability data.

Conclusion

Both humidity independent and humidity dependent cases were investigated in the predictive stability modeling approach with success. This approach is applicable and is aligned with global regulatory agency expectations of using science, data, and statistical tools to de-risk stability concerns, enable early and fast decision making, and enhance product quality in pharmaceutical development.