Defense Date


Graduation Date

Fall 12-21-2018


One-year Embargo

Submission Type


Degree Name





School of Pharmacy

Committee Chair

James Drennen

Committee Member

Carl Anderson

Committee Member

Frank D'Amico

Committee Member

Ira Buckner

Committee Member

David Good


The quality of a drug product may be characterized by the consistency with which its indicated clinical effect, and safety profile, is experienced by the patient. The concept that such quality should be built into a product is at the core of the United States Food and Drug Administration’s (FDA) quality by design (QbD) initiative. This vision for pharmaceutical product development emphasizes the risk-based identification of critical quality attributes (CQA) which summarize a product’s performance, the efficient refinement of critical product/process parameters (CPP) that can affect such attributes, and the systematic development of CPP limits, which assure appropriate performance of CQAs. For a tableted drug product, a cornerstone CQA is dissolution.

Often, a formulation and/or manufacturing process can change during a patient’s course of treatment, potentially jeopardizing the consistent performance of the drug product. Regulatory agencies typically require that sponsors demonstrate how the generic/post-change product is bioequivalent to the reference/pre-change product. While in vivo clinical trials are one strategy for demonstrating this, sponsors typically prefer in vitro dissolution tests as an alternative. During these in vitro test, the F2 metric is commonly used to assess dissolution profile similarity. This work sought to compare the F2 method with an alternative method, on the basis of errors in bioequivalence. The alternative method was based on the use of a physiologically based in vitro-in vivo correlation (PB-IVIVC) model that had been nested within a clinical trial simulation platform. The PB-IVIVC method provides a direct link from dissolution performance to clinical performance. Thus, when it is used to refine a CPP-vs-dissolution response surface, based on the performance of a reference product, the assurance of clinically defined bioequivalence can be directly built into a model tablet system. The model drug product for this work was an immediate release carbamazepine tablet. Carbamazepine was selected as the model active pharmaceutical ingredient because it has a narrow therapeutic index and is designated as a class II compound (i.e. high permeability, low solubility) according to the FDA’s biopharmaceutics classification system (BCS). As such, this compound is identified within the FDA’s scale-up and post-approval change guidance as possessing elevated risk for biononequivalence when changes are imposed to its formulation and/or manufacturing process.

After gathering single dose in vitro-in vivo data from the literature, the construction of the PB-IVIVC began according to a two-step process. Here, the respective parameters for the rate and extent of each product’s absorption were calculated using classical pharmacokinetic modeling and then regressed against each product’s rate and extent of dissolution. Next, the classically defined clearance parameter was replaced using a physiologically based clearance model. This allowed routinely available population pharmacokinetic data to be combined with first principles of human physiology, for the mechanistic prediction of intersubject variability via correlated Monte Carlo simulations. This PB-IVIVC was then used to not only define the CPP ranges for the model carbamazepine tablet system that would directly provide for bioequivalent performance but to perform a post hoc assessment of the CPP ranges conferred by the use of F2 statistic. Ultimately, the results showed that when the product’s CPPs were refined using the F2 statistic the was a higher risk of biononequivalence was higher when compared to a product that had been refined using the PB-IVIVC. It is intended that this work support the movement of product/process optimization practices away from methods that result in rigid factors of unknown clinical significance, and towards those that are focused on efficiently achieving specific clinical objectives.



Available for download on Saturday, December 21, 2019