The prevention of cross-contamination of drugs in
pharmaceutical production must be avoided, therefore
the cleaning of the manufacturing equipment is an
important aspect of good manufacturing practice. The process
of demonstrating the efficiency of the cleaning procedure
is known as cleaning validation. Of course, the analytical
method applied must be validated itself.
Validation of cleaning methods should follow the same rules as
any other procedure in pharmaceutical analysis, but some aspects
need special consideration, as regulated by the intended application.
Therefore, sensitivity and recovery are of importance.
It is crucial to realise that the sampling procedure is an integral,
if not the dominating, part of the cleaning method.
Cleaning validation and verification
Often, the efficiency of the cleaning procedure is investigated by
swabbing a defined area of the cleaned equipment surfaces with
an appropriate material. The residual substance(s) sampled from
the cleaned surface are then extracted and their amount analysed.
Another approach is to analyse the final rinse solvent, applicable
if the equipment surface is not accessible and the equipment is
difficult to dissemble. However, the risk of tightly absorbed residues
that are not dissolved by means of rinsing should be considered.
With respect to chemical substances, residues include primarily
active ingredients and cleaning agents, while degradants, raw
materials, intermediates, or excipients may also be of concern.
The maximum acceptable amount of residue is dependent on
the pharmacological or toxicological activity of the respective
substance – often considered as a safety factor – as well as the
batch sizes and dosages of the previous and next product, and
the equipment surface. Often, a maximum limit of 10ppm in
the next batch is established. This ‘residual cleaning limit’ isthen normalised with respect to the sampled equipment area,
as ‘specific residual cleaning limit’ (SRCL). Reported SRCL are
between 4ng/cm2 and 3μg/cm2. Usually, before any quantitation,
the surface must show no visually detectable traces of residues.
Analytical technique
The selection of the analytical technique is dependent on the
properties of the residue and sensitivity, as required by the cleaning
limit. Selective methods such as high performance liquid
chromatography, gas chromatography, high performance thin layer
chromatography, and immuno assays are preferred for active (and
related) residues. Non-selective methods, such as pH-measurement,
conductivity and total organic carbon, are mainly used for detergents.
As a rapid, sensitive and selective method, ion mobility spectrometry
was introduced, particularly for cleaning verification. This technique
(used in airports to detect explosives) is based on the ionisation of the
sample directly from the swab and measures the characteristic speed of
the ions moving through an electric field under atmospheric pressure.
Validation characteristics
Development and optimisation of the analytical procedure and its
validation is an iterative process in which the influence of the cleaning
solvent, swab material, swabbing solvent, sampling technique, and
extraction of the analyte on specificity and recovery is investigated.
In this explorative stage, the recovery at one single concentration,
preferably at the defined limit, is sufficient. During this stage, the
robustness of the analytical technique should be investigated.
With respect to specificity, interferences from the sampling procedure
must be taken into account. This should include blank extractions of
the swab material, as well as blank swabbings of the respective surface.
The target substance might be altered during the cleaning process and
the original substance replaced by a degradant. This can be
investigated by using aged samples. Specificity is usually demonstrated
by sufficient chromatographic resolution, or lack of interference.
In linearity, the intended calibration model is the target. The
corresponding requirements are: for a single-point calibration
(external standardisation), a linear response function and a zero
intercept have to be demonstrated; for a linear multiple-point
calibration, only a linear function.
A widespread misinterpretation is to report the correlation coefficient
as demonstration of a linear response function. However,
this parameter is no proof of linearity, but requires a linear response
function to have any meaning. Basically, it cannot distinguish
between random and systematic deviations from the linear function,but the latter is the primary objective. Therefore, the first step
should be to inspect the experimental data for systematic deviations
from the applied regression function. Numerical parameters are
only suitable afterwards. The relative standard error of slope is
recommended because such a normalised (percentage) parameter
allows a straightforward evaluation by comparison with an
acceptable precision.
After the conditions of sampling and sample preparation are
optimised, accuracy, precision, and quantitation limits can be validated
simultaneously in a range of at least 50 to 120 per cent of the defined
cleaning limit, using at least nine spikings. In cleaning validation, no
authentic, homogeneous samples are available. Therefore, precision of
the analytical procedure must be obtained from spiked samples.
Recovery is performed from the spiked surface(s) of identical
equipment material, often from spiked swabs. The latter may be
omitted if the former recovery is acceptable. If excipients can be
suspected to interfere, the spiking of the active should be performed in
their presence. Due to the potentially large influence of the sampling,
the robustness of the recovery, which will also include the swabbing,
should be investigated by repeating the recovery with another operator.
The contribution of other factors, such as analytical instrument and
reagents, may be investigated, but can be expected to be small
compared to the sampling. Dependent on surface properties and
material, the analyte can often be only partly recovered. Values of
larger than 80 per cent and 50 per cent are regarded as good and
reasonable, respectively, while less than 50 per cent is questionable.
If relevant (with respect to the precision), the recovery factor should
be used to correct the results of the cleaning method. To allow a
straightforward evaluation, the recoveries are preferably presented
graphically as percentages with respect to the spiked concentration.
This plot should be inspected for a concentration-dependent
behaviour. In case of no or only slight dependency, the average and
the relative standard deviation of all recoveries can be calculated. If
the spikings are analysed in an independent series, the calculation
can be performed as an analysis of variances. Here, the precision
contributions within and between the series are obtained separately,
and conclusions can be drawn on the robustness of the procedure.
The overall average from the intermediate recovery study is then
used as the recovery factor. In case of a concentration dependency,
either a concentration-dependent recovery factor is used, or it is
calculated at the cleaning limit as the relevant concentration. If
a sufficient number of determinations is available (at least five),
average and relative standard deviation can be calculated for each
concentration level. However, it is bad practice to calculate, say, a
standard deviation from three values only, because the upper limit of
its 95 per cent confidence interval is 4.4 times the calculated value.

No comments: