Issue link: https://www.e-digitaleditions.com/i/1302985
Pharmaceutical Technology Regulatory Sourcebook October 2020 23 • Qualification activities are performed by suit- ably trained and qualified personnel. • The drug product or formulation, which is subjected to process qualification, has been fully characterized. • The process risk assessment, including manu- facturing method, CQAs, CPPs, and control strategy, has been generated. • The analytical methods defined in the con- trol strategy are validated and appropriate for the intended use. Another essential prerequisite for the process performance qualification is the successful quali- fication of facilities, equipment, utilities, and ana- lytical instruments involved in the process. If the traditional approach of the European Medi- cines Agency (EMA) guideline on process validation is followed (7), a minimum of three consecutive batches, normally with the same batch size as the intended com- mercial batches, should be manufactured under routine conditions to confirm reproducibility. The process vali- dation protocol should be prepared and should define the CPPs, CQAs, and associated acceptance criteria. At the end of the activities, data must be col- lected and reviewed against predetermined accep- tance criteria of the validation protocol and fully documented in process validation reports. Suitable statistical tools for traditional PPQ data analysis have been covered in Part 1. Continued process verification (CPV) The aim of this stage (stage 3) is to assure the con- trol of CPP and CQA through the entire lifecycle. The key aspects and the statistical tools of CPV have been addressed in Part 1. Table V shows the correlation of a CQA failure with the CPPs specific for a solid oral dosage form. Statistical/chemometric tools (applicable among stages) The approach to process validation suggested by modern quality concepts requires the generation of experimental evidences where a multitude of properties should be measured, which generates a huge amount of data. The intrinsic nature of pharmaceutical process data implies that values, trends, and drifts are frequently expressed by a group of correlated variables and cannot be high- lighted simply by monitoring variables one at a time. Moreover, process design, investigation, and optimization are often laborious, time consuming, and may lead to unsaleable products; then, the adoption of a methodology that provides weighty information by means of a reduced number of tests is recommended to save time and money. Strategies for process understanding, monitor- ing, and control should hence implement statistical methodology capable to deal with data dimension- ality, collinearity, noise, and missing values. Design of experiments and multivariate data analysis have to be considered as suitable statistical/chemometric tools that might be used for process investigation, even if many others are available and might be used. The choice of a specific tool depends on the process/ product to be studied and the development/valida- tion stage and is usually defined, and suitably justi- fied, by the company strategy. A specific discussion of this topic is beyond the scope of this work. Additional information on this subject is presented in the reference section of this article (8–11). Conclusion Parts 1 and 2 of this work have illustrated the ap- proaches and the tools available to implement the