Quality- by-design in pharmaceutical development - Computer-Aided Applications in Pharmaceutical Technology - page 8

properties of the starting materials or the manufacturing process that
affect the quality of the pharmaceutical product. Once the increased
process and product understanding is obtained, it is possible to identify
and appropriately manage critical sources of variability, and design
effective and effi cient manufacturing processes that allow quality
assurance in real time. EMA representatives (Korakianiti, 2009) point
out that it is preferable for a design space to be complemented by an
appropriate control strategy. An example of a QbD application in
pharmaceutical product development is presented in the Examplain
Mock P2 document, available online. 2 The review of variations regulations
and the revised Variations Classifi cations Guideline (2008) has taken into
account QbD submissions, to enable easier updates of the registration
dossier. EMA templates and guidance documents used for the assessment
of any new drug application in the centralized procedure include the
possibility of design space appointment (e.g. Day 80 Quality AR
Template).
EMA, FDA, and ICH working groups have appointed the ICH quality
implementation working group (Q-IWG), which prepared various
templates, workshop training materials, questions and answers, as well
as a points-to-consider document (issued in 2011) that covers ICH
Q8(R2), ICH Q9, and ICH Q10 guidelines. This document provides an
interesting overview on the use of different modeling techniques in QbD.
In a QbD context, the model is defi ned as a simplifi ed representation of a
system using mathematical terms. Models are expected to enhance
scientifi c understanding and possibly predict the behavior of a system
under a set of conditions. For the purposes of regulatory submissions, the
ICH Q-IWG document classifi es the models according to their relative
contribution in assuring the quality of a product (Table 1.2). Development
and implementation of models include defi nition of the model purpose,
decision on the type of modeling approach (e.g. mechanistic or empirical),
selection of variables for the model, understanding of the model
assumptions limitations, collection of experimental data, development of
model equations and parameters estimation, model validation, and
documentation of the outcome of the model development. It is also
recommended to set the acceptance criteria for the model relevant to the
purpose of the model and to its expected performance. Also, accuracy of
calibration and accuracy of prediction should be compared and the
model should be validated using an external data set.
The ICH Q-IWG document also suggests that a design space can be
updated over the product lifecycle, as additional knowledge is gained. It
also notes that in development of design spaces for existing products,

￿

￿ ￿