(558a) Design and Implementation of a Control Strategy Using Risk-Based Probabilistic Design Principles: A Case Study in the Manufacturing an API with a Critical Impurity

In the quality by design (QbD) paradigm proposed by the FDA, the concept of risk analysis and risk
management is a fundamental part of the development of the design space and control strategy for critical
quality attributes (CQAs) in the manufacture of active pharmaceutical ingredients (API’s). However, there
is not a standardized or well-adopted method for the quantification of risk in the development of
processes for the manufacture of API. Herein, we present a case study highlighting the use of probabilistic
(Bayesian) modeling techniques as (1) a tool to quantify the relative risk between different processing
options, and (2), as a method for developing the control strategy for a specific CQA. We describe the
process for manufacture of an API molecule that contains one critical impurity that is controlled as a CQA;
the impurity is formed in the final chemical step and is primarily purged in a complex crystallization
involving the use of a rotor-stator wet mill. In this case study, using probabilistic models to quantify
variability, estimates of expected failure rates for different scenarios were tested, enabling the evaluation
of process parameters and different options for control strategies. Lab-scale and pilot plant scale
experiments were conducted to generate, test, and validate a series of models, which were used to
develop a risk-based probabilistic design space and a control strategy for the CQA. Ultimately, we present
the use of probabilistic modeling as a risk-mitigation and optimization tool for the engineer in the
pharmaceutical industry.