Method
Validation Protocol
USP
Chapter 1225 on validation of analytical methods
specifically addresses terms and definitions,
but leaves protocol and methodology open for
interpretation. Some may argue that to some
degree this was intentional, allowing
flexibility in method validation. Laboratory
personnel were given latitude to do "good
science," however, the term "good science" means
different things to different people. This is
where the guidelines drafted as a result of the
ICH process have begun to play a vital role. The
ICH Guideline on Method Validation Methodology
(now at step four of the ICH process as of this
writing) begins to hint at experimental design
and protocol. Before outlining an experimental
design or protocol, however, it is necessary to
make some basic assumptions. These assumptions
include:
-
Selectivity has been previously
demonstrated, or is measured and documented
during the course of the validation
protocol.
- The
method has been developed and optimized to
the point where it makes sense investing
time and effort in validating the method.
Indeed, robustness should be the first
parameter investigated.
- Once
data is generated, statistically valid
approaches are used to evaluate it and make
decisions, thus removing some of the
subjectivity of method validation.
|
Given
that the above three steps are addressed in one
fashion or another, and bearing in mind the ICH
methodology guidelines, and the requirements for
validation depending on the type of analytical
method (Tables 2 and 3), the following stepwise
protocol can be proposed, as illustrated in
Figure 9:
h hi
- On
day one, a linearity test over five levels
for both the drug substance and dosage form
is performed. Comparison of the results
between the drug substance and dosage form
fulfills the accuracy requirement.
- At
the end of day one, six repetitions are
performed at 100% of the drug substance for
repeatability.
- Steps
one and two are repeated over two additional
days for intermediate precision.
- LOQ
is evaluated (as needed) by analyzing the
drug substance over five levels, plus six
repetitions for precision.
-
Baseline noise is evaluated over six
repetitions of blank injections for the
determination of LOD (if required).
|
|
Figure
9: Example Method Validation Protocol
In this
manner, a logical stepwise approach to
method validation using ICH methodology can
be performed. As stated previously, the use
of the appropriate statistical tests
(Students-T, Cochran, Dixon and Fisher
Tests) allows for less subjective decisions
to be made regarding the data, reducing
method validation to a much more objective
undertaking. A rigorous use of statistics
may in turn lend itself to some degree of
automation in the future.
|
|
|
|
|