Approaching Microbiological Method Validation-1
Approaching Microbiological Method Validation-1
discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/287992312
READS
114
1 author:
Tim Sandle
The University of Manchester
279 PUBLICATIONS 228 CITATIONS
SEE PROFILE
All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Tim Sandle
letting you access and read them immediately. Retrieved on: 20 August 2016
Approaching Microbiological Method
Validation | IVT
By
Tim Sandle, Ph.D.
Dec 4, 2015 7:21 am PST
Peer Reviewed:
Microbiology
Introduction
While it is unlikely that microbiological methods will meet the more exacting demands of the
analytical ones. Nevertheless, the advances with rapid microbiological methods and
alternative techniques require microbiologists to become more familiar with assay
requirements and parameters. Many novel methods are automated and use forms of data
capture where the data is subject to statistical analysis.
These developments have impacted upon the approaches taken to the validation (or
qualification) of microbiological methods. Validation, in this context, can be defined as the
process by which it is established, by laboratory studies, that the performance characteristics
of a method meet the requirements for the intended application. In terms of microbiological
tests, one of the most important aims is in determining whether the sample to be examined
has any inherent anti-microbial properties and whether the incubation and growth conditions
can recover microorganism that may be present to an acceptable level. Appendix I provides a
generalized flow chart to assist with some of these considerations.
Microbiological methods fall into one of three major categories (although this categorization
is not always exclusive) (3):
a) Qualitative tests for the presence or absence of microorganisms, e.g. the pharmacopoeia
sterility test. Commonly, qualitative tests are assessed through the use of turbidity or other
growth related changes in a culture medium, as evidence of the presence of viable
microorganisms in a test sample.
b) Quantitative tests for the enumeration of microorganisms, e.g. a total viable count test or
flow cytometry.
c) Identification tests, e.g. the speciation of bacteria using a biochemical test. This includes
morphological and biochemical characterization such as biochemical reactions, carbon
substrate utilization, characterization of fatty acid composition, restriction endonuclease
banding patterns and use of 16S or DNA sequence analysis (4).
Methods that fall within the three described categories have a degree of variability, and with
these variations microbiological methods are inherently different from analytical ones (5).
Not least in the lack of agreed demonstrative criteria. The wide variability is acknowledged in
Ph. Eur. 5.1.6, the chapter relating to alternative microbiological methods (6). This is
particularly with regards to relative broad ranges.
Variability may result in difficulties in comparing two methods. In general, only a 50%
comparison is achievable for culture methods. With rapid and alternative microbiological
methods greater comparability can be achieved, however levels of precision remain in the
order of 15 to 35% relative standard deviation (obtained by multiplying the standard
deviation by 100 and dividing this product by the average).
Method recovery
Culture based microbial validation is limited by the ability of microorganisms to reproduce
under a set of conditions in relation to sample preparation, cultivation and incubation. Any
method is, therefore, a general indicator only. With cultural based methods, as described in
the pharmacopeia, it is recognized that the limit of detection has never been established
quantitatively.
Growth media,
The colony forming unit or CFU is not a true cell count (individual cells are rare in
nature, leading to the CFU being an underestimation of the number of
microorganisms present),
Incubation conditions (temperature and time),
Nutritional requirement of the organism,
Physical condition of the organism (stressed or sublethally damaged due to
temperature, humidity, high ionic strength, pH extremes, osmotic shock (relating to
liquid), residues of antimicrobial chemicals),
Dilution errors (7),
Environmental organisms are unlikely to be recovered when they are in the
exponential growth phase (exponential growth leads to better recovery),
Characteristic of the item under test,
The types of neutralizers used (8).
Some of these considerations are contained within Appendix I, which provides a flow chart
for a generalized approach relating to experimental design.
With non-culture based methods, these are unlikely to be affected by the above list.
Method comparison
The introduction of a new method into a laboratory, where one method is intended to replace
another, invariably requires a level of comparison. When comparing an older method with a
new method, new methods will most probably recover a greater number of microorganisms.
Given that a positive or higher result could be obtained with an alternative method and a
negative or lower result could be obtained with an established method, this does not
necessarily suggest that the positive result is a false positive.
In evaluating an alternative method a close match is not necessarily required, in terms of the
result obtained, between the alternative method and the current method. What is important
that the alternative method is capable of allowing the microbiologist to make an equivalent
decision in relation to sample or product quality in a consistent fashion.
Specific,
Measurable,
Achievable,
Realistic,
Time based.
Validation requirements,
Test controls,
Microorganisms,
Type of data/statistical analysis (this is an area requiring care, especially when
comparing compendial methods with alternative ones) (9),
How many times does the experiment need to be run?
How many samples are required?
This section lists several aspects of experiment design that need to be considered and
incorporated into the design stage.
Samples
The size of the test sample must be considered. Importantly, the number of samples must be
representative and of a sufficient number. A statistical technique may be used to set the
number of samples required. Care must be taken here since the statistical technique selected
may influence the sample size. The appropriate volume of sample may be a factor,
particularly with bioburden testing and ensuring that the sample tested is representative of the
final homogenous bulk.
For most validation exercises the number of batches tested is three (or more). With areas like
environmental monitoring, consideration should be given to the location of samples, such as
the number of locations within a cleanroom.
Testing samples at the end of the shelf-life or expiry time (this may include
assessment at interim time points);
Degrading samples stored in containers;
Holding samples under worst case conditions (such as upper or lower temperatures);
Testing samples at the end of any required process hold times.
The above points are applicable to many bioburden and bacterial endotoxin tests.
Test controls
Experiments should normally have duplicate positive controls, negative controls and, where
recovery needs to be demonstrated, positive product controls (direct product challenges or
spikes). The level of recovery must be defined at the outset; and the level of recovery must
be justifiable in terms of experimental aims and the test method employed.
Microorganisms
In selecting microorganisms it is often a good idea to draw these across a range of different
morphological types. Suitable categories include:
i) Gram positive rod and/or a Gram positive spore bearing rod, e.g. Bacillus sp.
In all instances either the specific culture collection reference must be quoted or the source of
the isolate identified. For certain Gram-positive bacteria, the protocol should specify if the
organisms should be in the endospore state.
The type of experiment must be considered when allocating microorganisms. For example, a
project involving the examination of water would most likely require a Gram negative rod
(such as Pseudomonas sp.) and a coliform (such as Escherichia coli). In contrast, an
experiment conducted at 55C (as with a test for thermophilic microorganisms) would
involve a thermophile such as Geobacillus stearothermophilus. Consideration should also be
given to the storage of cultures.
Sometimes microorganisms are not recovered as expected. This is a case for carrying method
development work in advance. An example of poor recovery can occur with Gram-negative
rods. With these organisms, desiccation can occur during aerolisation. This may occur when
using an active air sampler or following loss of moisture content in an exposed settle plate
surface. This effect tends to damage Gram-negative bacteria more greatly. This is because
soluble cell contents tend to leak and mechanisms to control the transfer of molecules and
ions in and out of Gram-negative cells in particular are considerably impaired. Damage to
the mucopeptide lipopolysaccharide center cell structure also causes cell damage and loss of
viability. Such damage occurs very shortly after weight loss to a medium or after
aerolisation. Such cell damage is typically irreversible (10).
Microorganisms used should normally be prepared from cultures which are no more than 24
hours only and no more than five passages from the seed lot. However, some microorganisms
require longer cultivation than 24 hours. In these instances the culture age must be
documented in advance to ensure that the culture used is as young as it can be. With passages
(or subcultures), this is to prevent phenotypic variations from occurring which might
influence the way the microorganism behaves in the presence of the sample (11).
It is good practice that the purity of cultures is confirmed in advance. This can either be by
acceptance of a certificate of analysis from the supplier of the cultures, or by confirmatory
identification conducted by the recipient laboratory. In some circumstances, such as when
conducting a challenge test, a post identification confirmation may also be deemed
necessary.
With some experiments, an attempt may be made to induce a stressed state to the microbial
population. This will need to be decided at the time of writing the test protocol. The reason
for attempting this is based on the unstressed batch culture grown organisms being are an
artificial creation that rarely exists outside the laboratory (12). Stress factors faced by
microorganisms in the environment include:
Desiccation,
Nutrient deprivation,
Nutrient limitation,
Cold shock,
Heat shock,
Exposure to ultra-violet light,
Other types of radiation leading to sublethal damage,
Responses to disinfectant or detergent residues,
Responses to preservative residues,
Osmolality.
Creating a stressed state is difficult in itself and difficult to verify, given the unknown effects
of causing damage to cells or with suppressing cellular growth (13). In addition, the age of a
culture will affect its recovery and will be important for certain identification methods.
Temperature
Consideration needs to be given over the temperature ranges the experiment needs to be
performed at. Here there is little value in restricting validation to one temperature range for
testing that takes place over multiple ranges. An example here would be testing culture media
at 20-25oC and then using it across the range 20-40oC, and expecting its growth promoting
properties to be consistent. Commonly used ranges include:
2C - 8C,
20C - 25C,
30C - 35C,
36C - 38C
55C - 60C
In addition to the above, dual incubations across two or more ranges may be required (such as
for the enumeration of bacteria and fungi with a single-use culture medium for environmental
monitoring) (14).
Time
As with temperature, consideration needs to be given to the time period over which the study
is to be conducted. Getting this right is important since the selected time becomes the
maximum run-time for the test. It is important to ask if incubation time is set long enough to
show the limit of detection? The time of the validation read must never exceed the time used
by the testing laboratory for the reading of samples.
Moreover, in specifying incubation times the minimum time must be clearly stated. For
example, incubate plate at 20-25oC and read at 24 and 48 hours rather than incubate plate
at 20-25oC and read after 48 hours. Consideration of acceptable tolerance should be stated.
Is read at 48 hours, for instance, reading within 1 hour of 48 hours or will a wider
tolerance be used, and with what justification? Depending on the test method, it may be
appropriate to time samples going into and out of incubation.
Incubation times should be realistic in terms of loss of viability or loss of growth supporting
properties of media (e.g. the time set should not be so extensive that plate media desiccation
would occur).
Growth Promotion
A key question here is: What type of growth promoting conditions are required for the
cultivation of microorganisms? This would include different types of agars, broths, dilution
reagents and so on. These may require a separate (or first phase) validation.
Atmosphere
Antimicrobial Activity/Interference
Note should be taken of whether any of the experimental conditions produce an antimicrobial
effect or interfere with the test in way of enhancement or inhibition. Attention should be paid
to the best method for neutralizing any anti-microbial properties. Common methods for
neutralization include dilution, rinsing, filtration or the use of general or specific neutralizers.
pH
Growth Phase
The growth phase of a microbial population can impact upon the accuracy of a method, for
example, a turbidity method. Ordinarily, in culture, the following growth dynamics are
observed (15):
Cells initially adjust to the new medium (lag phase). Cells maybe growing by mass,
but not in number. Lag phase is influenced by size of the inoculum; time to recover
from physical damage; time required for synthesis of essential coenzymes; and time
required for synthesis of new enzymes necessary to metabolize the substrates present
in the medium.
Cells then start dividing regularly by the process of binary fission (exponential
phase). Here cells divide at a constant rate, expressed as generational or doubling
time. Ideal generation times wider according to different microbial species.
Escherichia coli, for example, will double every 17-20 minutes, whereas
Mycobacterium tuberculosis doubles every 790-930 minutes.
When their growth becomes limited, the cells stop dividing (stationary phase).
Eventually cells show loss of viability (death phase).
With other methods growth is not a requirement and instead cell count is of importance. Such
methods may or may not be able to distinguish between viable and non-viable
microorganisms. A method like flow cytometry, for instance, will count all cells irrespective
of whether they are viable.
Number of Microorganisms
Some methods will require a minimum number of microorganisms in order to detect them.
For example, a method linked to microbial catabolism; adenosine triphosphate (ATP) growth;
bioluminescence etc. A pre-growth preparatory step may be required in order to maximize
recovery.
Single Cultures
Most methods require a single (pure) culture in order to obtain a valid result. Mixed cultures
can be employed in certain circumstances, for example, container closure integrity. Here,
care must be taken that one microorganism does not significantly out-grow the other. In
general, mixed cultures should be avoided.
Enzymes
Validation Parameters
Specificity
The microbial challenge should be set above the limit of detection or quantification, while
also being at level that provides a measure of the efficacy of the method. Here:
For a growth based method, a low number of <100 CFU is appropriate. All challenge
microorganisms should be recovered. Where atypical colony morphology is observed,
supporting identification should be considered.
For a non-growth based method, suitable positive and negative controls should be
used to show that any extraneous matter does not interfere with the detection of the
microorganisms.
Appropriateness
A new method or test must demonstrate that it is appropriate for its intended use. If a method
or test is intended to replace an established method, parallel testing of both methods must
take place and the collected data compared if possible, ideally by statistical tests of
significance. In some cases, a direct comparison is not possible (for example, two different
models of particle counter cannot be directly compared because they will not be sampling the
same volume of air).
Accuracy
Accuracy is the closeness of agreement between the measured value and the true or
expected measure or reaction across the range of the test. This can be assessed by
determining the recovery of known quantities of a microorganism that has been added to a
sample.
For enumeration methods, the level of recovery should reflect the test method. This is
normally by the percentage of microorganisms recovered by the method.
For qualitative methods it is recognized that many hundreds of comparisons may be required
if a negative result is the expected outcome, such as with the sterility test. A limitation must
be established in relation to the number of samples; this is because testing samples until a
positive result is obtained will be impractical. When comparing two methods, the relative
rates of positive and negative results should be compared.
Precision
Precision is the closeness of agreement between a series of test results or the variation in a
series of test results, when a method is applied repeatedly to multiple samples.
a) Standard deviation,
b) Coefficient of variation (relative standard deviation),
c) Confidence interval of the mean.
d) With specific tests, such as microbial identification systems, other criteria will be used to
determine the similarity of the recovered organisms.
e) Other statistical techniques like Chi-squared, maybe more appropriate (18).
Normally at least 3 replicates are required. Depending on the types of sample, more than one
determination may be required (for example different dilutions or a range of
microorganisms). Because the testing technician and consumables are the same, this approach
shows variation in sample as assessed against the method.
Intermediate Precision
This is the variation in results obtained on the sample when assayed on several separate
occasions by different technicians using different reagents and equipment etc. This shows
reproducibility. This may be expressed as:
a) Standard deviation,
b) Coefficient of variation (relative standard deviation),
c) Confidence interval of the mean.
d) With specific tests, such as microbial identification systems, other criteria will be used to
determine the similarity of the recovered organisms.
e) Other statistical techniques like Chi-squared, maybe more appropriate.
Range
Range is the interval between the upper and lower levels of microbial count, for which the
procedure has been established as suitable with accuracy, linearity (if appropriate, where
there is requirement to construct a curve), and precision.
For example, the commonly used range for microorganism recovery is less than 100 CFU for
total count techniques. Where a range is required, in order to assess the range of the test, this
is covered by diluting a microbial population; for example, 100 to 106 cells. With some
methods, a regression analysis can be considered to compare two methods.
Robustness is the reliability of a method or test to withstand small (but deliberate) variations
due to external influence. For example, different technicians, instruments, incubation time,
ambient temperature, and reagents. With rapid methods, robustness can be undertaken by the
method supplier. However, once in the laboratory, long term performance must be
considered and, if possible, checked by internal control samples. For example, the
distribution of microorganisms on a membrane can affect robustness.
Ruggedness is the degree of reproducibility by testing samples using different testers and
equipment; this is assessed by coefficient of variation.
Limit of Detection
This is the lowest number of microorganisms which can be detected, but not necessarily
quantified (such as a low level challenge) under the stated experimental conditions. This test
is generally used for rapid and alternative methods.
Often the amount of sample tested the initial dilution of the sample and any subsequent
dilution of the sample may determine the limit of detection. The challenge microorganisms
selected should be of an appropriate range as indicated above). The challenge can consist of
taking each microorganism and making a serial dilution range.
It is recognized that the act of dilution may result in a greater loss due to lack of homogeneity
and the typical Poisson distribution of microorganisms in liquid; such as in the evaluation of
a raw material by using the pour plate method and where the limit of detection is <10cfu/g.
This is further complicated by the impossibility of obtaining reliable samples containing a
single microorganism. Therefore, microbiologically limits of detection must sometimes be
considered as theoretical rather than practically demonstrable. For this reason, many
pharmacopeial` tests require the use of a low level challenge (<100 CFU) and this is normally
considered sufficient. For comparing methods Chi-squared is often the statistic of choice.
Limit of Determination
This is the lowest level of the sample where the microbial content can be quantitatively
determined with defined precision and accuracy. Again, further complication arises by the
impossibility of obtaining reliable samples containing a set number of microorganisms.
Linearity
This is the ability to elicit results which are proportional to the concentration of
microorganisms within a given range. This is measured by correlation coefficient or a
goodness of fit test (such as Chi-Square). This will only be applicable to enumeration
methods using an analytical system.
When comparing two methods, non-linearity will occur if one method is superior to the other
(in terms of microbial recovery). Here Spearmans rank may be appropriate as a statistical
tool to use (a non-parametric measure of statistical dependence between two variables).
Predictive Value
For qualitative tests such as the sterility test or growth of selective media, the use of positive
or negative predictive values may be appropriate. This can be expressed as a percentage of
the observed test results against the total or expected test results.
Rapid and alternative microbiological methods embrace those methods that are distinct from
compendial methods. Rapid indicates that the method gives a faster time-to-result and
alternative indicates that the method differs from one presented in a recognized pharmacopeia
(19). In many cases such methods are more accurate (although this is not always the case)
and they are invariably automated (20). Given the pace of technological development and
commercial gains, such methods are becoming more commonplace.
With alternative methods the aim is to verify the detection capability of the alternative
method.
In terms of parameters to review, the following acts as guidance (based on USP chapter
<1223>) (23):
Validation parameter Qualitative test Quantitative test
Accuracy (a method No Yes
suitability test)
Precision (a method No Yes
suitability test)
Specificity (a method Yes Yes
suitability test)
Limit of detection Yes Yes
Limit of quantification No Yes
Linearity No Yes
Operational (dynamic) No Yes
range
Robustness Yes Yes
Repeatability Yes Yes
Ruggedness Yes Yes
Equivalency Yes Yes
a. Approach #1: Each organism is prepared as a serial dilution, with the inoculum adjusted to
a target of 50% that shows growth in an existing method. Both methods should be run over
several replicates. An appropriate statistical comparison method is Chi- squared.
b. Approach #2: A Most Probable Number (MPN) method is used. For a ten-fold series a
range of 10-1 to 10-2 microorganisms is used; or for a two-fold series, using the range 5 to 10-1
microorganisms. The alternative and established method should be run five times using 3
dilutions (the dilutions should provide at least one positive and one negative dilution). Chi-
squared is an appropriate statistical tool for method comparison.
Where the aim is to show equivalence between two methods, there are four possible ways to
assess this (again as referenced in USP <1223>). The appropriate category should be selected
in advance of conducting the validation and referenced in the validation protocol. The options
are:
Acceptable procedure
2. Equivalent Yes Results Multiple
Performance equivalence
3. Equivalent Yes Results Singular
Results equivalence
4. Equivalent Yes Conclusions Singular
Decision equivalence
Discussion
This paper has outlined the key criteria to be considered when undertaking microbiological
method validation, either in relation to the qualification of the method itself or in relation to
testing samples against a specific method. The approach outlined is partly applicable to
established method and partly applicable to rapid and alternative methods. Here not all of the
criteria outlined are applicable to all methods.
In drawing up the list of possible criteria, examples have been provided in order to help the
reader consider the applicability of each item. With this it is important to plan and to devise a
suitable validation protocol, outlining acceptance criteria and the way that the validation will
be executed.
References
Appendix