Knowledge Management and Process Monitoring of Pharmaceutical Processes in the Quality by Design Paradigm - Measurement, Monitoring, Modelling, and Control of Bioprocesses - page 235

Traditional Control Strategy (Fixed controls)

Knowledge Space

Variability in
characteristics
of feed material
and raw materials

Design Space

Variability in
product quality

Normal
Operating
Ranges

Failures

Dynamic Control Strategy

Knowledge Space

Variability in
characteristics
of feed material
and raw materials

Design Space

Consistency in
product quality

Adaptive
Operating
Ranges

Failures

Adaptive
Operating
Ranges

Fig. 14 Illustration of different control strategy approaches for a pharmaceutical process.
Adapted from Rathore [ 17 ]

and Quality Assurance initiated an effort that eventually evolved into QbD [ 15 ]. The
underlying principles of science and risk-based process and product development
and commercialization are also reflected in the contents of the International Con-
ference on Harmonization (ICH) guidelines: ICH Q8 Pharmaceutical Development,
ICH Q9 Quality Risk Management and ICH Q10 Pharmaceutical Quality System
(ICH Guideline Q8, Q9, Q10) [ 7 , 8 , 9 ]. The recently issued Guidance on Process
Validation from the US Food and Drug Administration (US FDA) also imbibes these
principles [ 5 ]. The last 5 years have seen QbD gaining widespread adoption in the
pharmaceutical industry, with several publications attempting to elucidate a path
forward for implementation of QbD and resolving the various issues that otherwise
serve as detriments to successful implementation [ 12 , 17 , 18 ].
In the traditional approach to pharmaceutical production (Fig. 14 ), manufac-
turers define a process and run it consistently such that the process parameters are
controlled within a narrow range to maintain consistent product quality. This has
been the approach adopted for the vast majority of pharmaceutical products on the
market today. The major downside of this approach is that, since the process
controls are fixed, variability in raw materials and process manifests as variability
in product quality and sometimes results in lot failures. Since variability is not
eliminated at any step, it accumulates as the process advances from one step to the
next, with the final product quality varying significantly. In contrast, in the QbD
paradigm, the control strategy is dynamic and allows the process to be run dif-
ferently to deal with incoming variability in characteristics of feed material and
raw materials. The net result of this is that the incoming variability is either
reduced or eliminated and the resulting product quality is very consistent. This is
best shown by the data presented in a recent publication [ 19 ]. As seen in Table 1 ,
when a process analytical technology (PAT)-based control scheme was used for