APPLICATION OF VALIDATION

细履平沙

Beginning with its first association with LVPs in the early
1970s, the application of validation spread quickly to
other sterilization processes. It was also applied for the
validation of other pharmaceutical processes, albeit
with mixed success. In sterilization and, to a slightly
lesser extent, in processes supporting the production of
sterile products using aseptic processing, there is little
difficulty applying validation concepts. The apparent
reasons for this are the common and predominantly
quantitative criteria for acceptance of the quality attributes
of sterile products. Building consensus on validation of
sterile products has been achievedbut not without debate.
There are numerous excellent guidance documents
outlining validation expectations on the various sterilization
processes, as well as numerous publications from
individuals and suppliers. The only relatively deficient
areas in sterile product validation are elements unrelated
to sterility, e.g., endotoxin and particulate matter.
Validation of non-sterile products and their related
processes is less certain. Despite the obvious importance
of cleaning procedures, cleaning validation was not
publicly discussed until the early 1990s. To this day
there is still confusion regarding the requirements for
validation of this important process. The difficulties with
validation are even more complicated for pharmaceutical
dosage forms. There are no widely accepted validation
requirements for the important quality attributes of drug
products. While the key elements are known (dissolution,
content uniformity, and potency), there are no objective
standards upon which to define a validation program.
The compendial standards of the various pharmacopeia
are poorly suited to validation. The small sample size
and absolute nature of the acceptance criteria are extremely
problematic for direct application to large scale
commercial production. After more than 30 years, the
absence of universal criteria for dosage forms is unfortunate
and problematic.
Applying validation requirements to water and
other utility systems is somewhat easier than for pharmaceutical
products. Equipment qualification of utility
systems is relatively easy to perform, and samples of
the supplied utility (water, steam, environmentally
controlled air, compressed gas, solvent, etc.) taken
across the system can directly support the acceptability
of the preparation, storage (where present), and delivery
system. Classified and other controlled environments
have proven relatively easy to validate. Their physical
elements readily lend themselves to equipment qualification,
and sampling affords confirmation of their
operational capabilities directly.
Biotechnology first came of age in the late 1980s into
a regulatory environment that expected validation of
important processes. Since the first biotech products
were injectable drugs, it was quite natural for these
firms to validate their processes from the onset. As a
consequence, cell culture and purification processes of all
types have always been subject to validation expectations.
There is a substantial body of validation
knowledge on these processes available. In marked
contrast, the bulk pharmaceutical chemical segment of
the industry has been relatively slow to embrace validation
concepts. While the rigorous environmental
expectations associated with many dosage forms and
virtually all biotechnology processes are not present, the
important considerations of impurity levels, byproduct
levels, racemic mixtures, crystal morphology and trace
solvents all suggest that there are important quality
attributes to be controlled (and thus validated) as well.
Computerized systems became subject to validation
requirements when they were first applied for cGMP
functions in the 1980s. For ease of understanding, the
parallels between computerized systems and physical
systems are utilized. The computer hardware can be
qualified like the process equipment to which it is often
connected, while computer software has some similarities
to the operating procedures utilized to operate the equipment.
This approach may be an over-simplification of the
required activities for the software, but it provides some
clarity to the uninitiated. Computerized system validation
is still a subject of substantial interest, but is no longer the
misunderstood behemoth task it appeared to be when first
encountered. The early efforts of PMA’s Computerized
Systems Validation Committee and the later development
of GAMP have reduced the uncertainty associated with
the use of computerized systems substantially (6).
One useful concept taken from the validation of
computerized systems as it evolved was the “life cycle
model” (7). Originally utilized for computer software,
it was later applied to the entire computerized system. It
suggests that considerations of system qualification,
maintenance and improvement be incorporated at the
onset of the design process. Its utility for computerized
systems is substantial; however it may have even greater
functionality for pharmaceutical processes. In the early
1990s, the FDA launched an initiative related to the demonstration
of consistency of processes and data from
clinical lots through to commercial manufacture (8). They
mandated the conduct of Pre-Approval Inspections to
affirm that commercial materials had their basis in the
pivotal clinical trial materials. The utility of the “life cycle
model” in this context is clear. Its application to pharmaceutical
development, scale-up, and commercial
production allows for a coordination of supportive
information in the same manner as software and computerized
systems validation. A landmark publication
in this area was Kenneth Chapman’s paper entitled
“The PAR Approach to Process Validation” (9). It addressed
the developmental influence on the ability to successfully
validate commercial operations, a message that has been
somewhat forgotten until just recently. Ajaz Hussain, then
of the FDA, voiced concerns relative to the lack of process
knowledge on the part of many pharmaceutical firms
(10). That the FDA believed that such a missive was
necessary supports the lack of appreciation for Chapman’s
earlier effort:
The goal of development is to identify the process
variables necessary to ensure the consistent production
of a product or intermediate (11).
Application of the “life cycle model” to pharmaceutical
operations addresses the compliance and quality
expectations of the industry in an appropriate manner
and should be a near universal goal.
Another regulatory development of some importance
is that of PAT (12). The concept was well articulated
by Dr Hussain while he was with the FDA. To many in the
industry, PAT seems like an advance of some magnitude
that could seemingly replace validation. To those well
versed in automation, PAT is nothing more than the
extension of long-standing control practices into pharmaceutical
batch production. Engineers familiar with
process control will recognize PAT as the installation of
feedback control relying on sensors in the process equipment.
This is by no means startling, except to those
unfamiliar with control loops. PAT has its utility and
will improve the quality of products produced by it—of
this there can be little doubt. It will not, however, replace
validation. In order to use a PAT system, the designer
must assure that the installed sensor accurately reflects
the process conditions throughout the batch otherwise it
will provide no benefit. The need for that assurance
means that the PAT system, rather than replacing validation,
will actually have to be validated itself!