Chapter 13

Evaluation of Stability Data

Nanda Subbarao and Kim Huynh-Ba

Contents
13.1 Data Evaluation and Trending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
13.1.1 Evaluation of Raw Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
13.1.2 Evaluation of Stability Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
13.2 Investigation of Out-of-Specification (OOS) Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
13.2.1 Phase I – Laboratory Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
13.2.2 Phase II – Full-Scale OOS Investigation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
13.2.3 Outlier Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
13.2.4 When the OOS Result Is Confirmed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
13.2.5 Trending OOS Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
13.3 Setting Specifications and Stability Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
13.3.1 Refinement of Specifications Using Data from Stability Studies . . . . . . . . . . . . 275
13.3.2 Expiry Dating of Clinical Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
13.3.3 Commercial Specifications and Extension of Expiration Dating . . . . . . . . . . . . 277
13.4 Preparation of Stability Reports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
13.4.1 GMP Requirements for Records and Reports . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
13.4.2 Elements of a Stability Data Sheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
13.4.3 Anatomy of a Stability Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
13.4.4 Requirements for Stability Section in the CMC . . . . . . . . . . . . . . . . . . . . . . . . . . 280
13.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283

Abstract This chapter discusses the evaluation of stability data. It follows the
stability study information from the point that raw data is generated in the lab,
calculations are performed to give test results, and test results are entered in the sta-
bility summary sheets, until data is finally entered into a stability report for submis-
sion purposes. This chapter also includes a summary of data evaluation addressed
in ICH Q1E and a discussion of Out-of-Specification (OOS) and Out-of-Trend

N. Subbarao (B)
Biologics Consulting Group, Inc., 3 Serina Drive, Plainsboro, NJ 08536, USA
e-mail: nsubbarao@bcg-usa.com

K. Huynh-Ba (ed.), Handbook of Stability Testing in Pharmaceutical Development, 263

DOI 10.1007/978-0-387-85627-8 13,  C Springer Science+Business Media, LLC 2009
264 N. Subbarao and K. Huynh-Ba

(OOT) investigations. Specification setting and shelf-life extrapolation, which are

performed after evaluating stability data, are also described in this chapter.

13.1 Data Evaluation and Trending

ICH Q1A(R2), Stability Testing of New Drug Substances and Products [1], for
drug substances and drug products intended for marketing in the ICH Tripartite
region includes sections on the evaluation of stability data. ICH Q1E, Evaluation
of Stability Data [2], provides further details for data evaluation and includes rec-
ommended procedures for statistical analysis. These ICH guidelines are applicable
to New Chemical Entities (NCEs) and associated drug products but do not apply to
generics, manufacturing variations, clinical trial batches or devices.
This chapter describes the data evaluation that is to be performed from the time
that data are generated until they are reported in a regulatory submission. Figure 13.1
provides a flow diagram for stability data evaluation.

Initiate
Intermediate condition
sample testing

Raw data generated Invalidate original test,

in lab repeat test per internal SOP

Yes

Yes
Aberrant data
Lab Investigation Lab error?
/OOS

No
No

Data of Accelerated
Add results to Stability Full Scale Investigation Condition?
Summary Table

Yes Re-design formulation,

OOT? No Yes storage condition,
Report Results Failed lot? label changes etc

No
Yes

Report Results
Issue FAR if
Commercial lot

Fig. 13.1 Flow diagram of stability data evaluation

13.1.1 Evaluation of Raw Data

Stability data evaluation must begin when raw data is generated in the labo-
ratory. cGMPs require that drug products and drug substances must meet their
13 Evaluation of Stability Data 265

specifications for identity, strength, quality, and purity. Results for tests such as
appearance and package integrity are evaluated directly against the specification.
ICH Q1A(R2) defines significant changes for stability samples and which can be
found in Chapter 3, Table 3.4. Additional information regarding physical testing
is discussed in Chapter 10. For other tests such as purity by chromatography, the
raw data must be examined for changes such as new or growing impurity peaks. It
is important that any significant changes or aberrant observations be noted imme-
diately, and investigated promptly, at a time when the original unexpired sample
solutions and reagents are still available.
The evaluation of the raw data can be effectively performed only if the analyst
has access to the stability specifications as well as to the results and chromatograms
from the previous time points of the stability study. Chromatograms of excipient lots
and of the drug substance lot used for manufacture of the drug product lot are also
useful. Designation of an appropriate person in the laboratory to evaluate data and
act promptly if an OOS is found is invaluable for prompt and meaningful laboratory
investigations of aberrant raw data. Any OOS results found must be investigated
promptly and the procedures for Laboratory Investigations and OOS Investigations
are described in Section 13.2.

13.1.2 Evaluation of Stability Results

The results obtained, after calculations, are compared to the Specifications and must
be evaluated for OOS and OOT results. The procedures for identification of OOS
or OOT incidences are described in this section and the procedures to follow if an
OOS or OOT incident occurs are described in Section 13.2.

13.1.2.1 Identification of OOS Results

Although specifications are applicable to products at the intended storage temper-
ature, any stability test result that does not meet specifications is said to be OOS.
The individual result, calculated according to the analytical method, rather than the
average or mean, must be compared to the specification limit. The result is also
considered to have failed the specification limit if it has to be rounded in order to
meet the limit. For example, a result of 9.99 units does not pass the specification
limit of ≥10.0 units.

13.1.2.2 Identification of OOT results

Identification of OOT results is often more complicated than a simple comparison of
the results to the specification limits. Guidance documents up to now have provided
little guidance on the subject. Yet, trending is a critical part of an effective stability
program.
In principle, any data which deviates significantly from the norm for that product,
packaging configuration or lot is considered to be OOT. The OOT identification
266 N. Subbarao and K. Huynh-Ba

procedures therefore depend on the availability of data to define the norm. During
early development stability studies, where little information about the product or
formulation is known, the test results from earlier time points are set as the norm
for later time points. Any significant deviation from this earlier result is identified as
an OOT incident and action is taken as appropriate. Where a significant amount of
stability data is available, a lot or packaging configuration is identified as behaving
OOT if its rate of degradation is different from the normal degradation rate for that
formulation or package type.
The trend identification can be qualitative and performed by graphing the sta-
bility data or could be performed by statistical analysis of the collated data. In
both cases, the site OOT Standard Operating Procedure (SOP) defines criteria for
designating a data deviation from the norm as an OOT incident. The OOT criteria
must be set in such a way that all significant OOT incidences are identified, ideally
without false positives.

Graphical OOT Evaluation

Graphing can be used to identify stability data which are OOT within lot or within
product/packaging type. Figure 13.2 refers to stability data at different time points
for a single lot, with the result at the 9-month time point that can be considered as
OOT when compared the other time points in the study. Such an identification of
OOT data is of value during development studies where formulation information is
limited, and the graph provides information about product degradation or analytical
method variability. The presence of an OOT may then be evaluated further by cal-
culating the change from original, change from original per unit time, change from
last test or by evaluating the observed value directly, without further calculations.
In Fig. 13.3, lot 4 data could be considered as OOT with respect to the other
four lots (lots 2, 3, 5, and 6). This type of plot allows evaluation of multiple lots of
different strengths or different pack sizes and can identify OOT of degradation rates
in a specific strength or package size and/or configuration.

101
100
99
98
97
96
Lot 1
95
94
93
92
91
90
0 3 6 9 12 15 18 21 24 27
Months

Fig. 13.2 Stability results for a single lot over time. The 9 months result can be considered as OOT
13 Evaluation of Stability Data 267

101
100
99
98 Lot 2
97
Months

Lot 3
96
Lot 4
95
Lot 5
94
Lot 6
93
92
91
90
0 5 10 15 20 25 30

Fig. 13.3 Stability results for 5 lots over time. Here, the rate of change of lot 4 is being compared
to that of the other lots and appears to be OOT

Statistical Evaluation of OOT

The data comparison described above in Figs. 13.2 and 13.3 may be performed sta-
tistically in several ways. The statistical approach takes data variability into account
when setting limits. Therefore, a single acceptance criterion for OOT identification
can be set for different types of assays. Three such procedures for normally dis-
tributed data are described in a review by the PhRMA CMC Statistics and Stability
Expert Teams [3]. Each of these approaches has its own advantages and disadvan-
tages, a summary of which is provided in the paragraphs below.

Regression Control Chart Method

In this approach, a least squares regression line is fit to stability data either from a
single batch or from several batches. The expected result for any time point is given
by the expression:

Expected Result = intercept + (Slope × Time) (13.1)

The control limits at a given time point is given by

Expected Result ± (k × s) (13.2)

Where k is a multiplier chosen from a table of normal quantiles to give the desired
protection level and s is the square root of the mean square error from the regres-
sion. The choice of k value allows the control of the confidence level and thus the
rate of false alarms. This approach depends on the data being normally distributed
and independent and is applicable only to data with a common linear slope for all
batches.
268 N. Subbarao and K. Huynh-Ba

By-Time-Point Method
In this approach, historical data is used to compute a tolerance interval for each
stability time point. The tolerance interval can be calculated for the stability results
themselves or for the difference between the result and the initial stability result.
The interval at a certain time point can be calculated as:

(Mean of result at time point) ± k × s (13.3)

where k is the multiplier chosen from a table of normal quantiles to give the desired
protection and s is the standard deviation at the time point.
Any result outside the tolerance interval is considered OOT. This approach
depends on the data being normally distributed and independent and does not require
any assumptions about the shape of the degradation curve.

Slope Control Chart Method

In this approach, a control chart for the slope at each time point is constructed. At
each time point, a least squares regression is fit that includes all the earlier time
points. The slope estimate for each batch is used to obtain the overall slope estimate
for several lots. The OOT limit for the slopes at each time point are obtained from
the tolerance interval, in which k is chosen to obtain the desired protection from
false negative or false positive values.

13.1.2.3 Special Case – Reviewing Impurity Assay Results

The specific case of impurity assay results requires a more detailed discussion
because of its impact on the proposed drug substance and drug product specifi-
cations during development studies.
Often, the impurity acceptance criteria in the product specification are rounded
as required by the ICH Q3A(R2) guidance document. The impurity data is therefore
reported by the laboratory to match the rounding in the product specification, often
to one digit past the decimal, while the significant figures for a typical HPLC impu-
rity test are given to two digits past the decimal place. This practice of reporting to
match the number of decimals in the ICH guidance limits the power of OOT tools by
decreasing the information in the data set. It is therefore advisable to report stability
data per the significant figures appropriate to the analytical method, although it may
be reported to match the specification for lot release purposes.
The Limit of Quantitation (LOQ) and the reportable limit for impurities also
impact impurity test OOT procedures. It is common practice to report impurity
peaks below the Limit of Quantitation as < LOQ, which is set to the reporting limit
for that impurity per ICH Q3A(R2). Therefore, the purity result for peaks just below
the LOQ cannot be used for trending although the analytical method variance may
be satisfactory at that level. This reporting practice again decreases the information
presented in the stability data tables, limits the OOT tools for impurities close to the
13 Evaluation of Stability Data 269

LOQ levels, and could be a concern for those impurities with specification limits
close to the LOQ.
Some common procedures for evaluating OOT close to the LOQ are given below:
r If all test result values are above the LOQ, the distribution is normal, and the
variance is constant, the regression control chart method or the slope control
chart method may be applied to the data.
r If all test results are below LOQ, any test result which appears above the LOQ
may be considered as OOT. However, this OOT identification procedure will
result in false positives if the impurity peak is normally just below the LOQ and
its appearance above the LOQ is due to test method variability.
r If some test values are below LOQ, one strategy would be to consider all results
which are <LOQ as either the LOQ level or 1/2 (LOQ) for purposes of the statis-
tical calculation leading to the identification of the OOT. This OOT Identification
approach is impacted by the distortion of information due to the approximation
for peaks <LOQ.
OOT identification for Impurity test results may also point to the need for iden-
tification of unknowns which may be increasing in levels. A growing unknown
impurity should trigger identification of the impurity and validation of the impurity
analytical method, before the peak reaches levels where the guidance documents
require its identification.

13.1.2.4 Importance of Prompt Investigations

OOT result identification during development stability studies provides early warn-
ing about possible changes required to the formulation or packaging. OOT obser-
vations during annual commitment lot stability studies can provide early signals of
possible future lot failure.

13.2 Investigation of Out-of-Specification (OOS) Results

21 CFR 211.192 requires that all OOS occurrences be investigated. The FDA issued
a guide to inspection of Quality Control laboratories in 1993 and a draft OOS
guidance in 1998 [4], following the Barr case. The final OOS guidance document,
issued in 2006 [5], provides guidance on the procedures to be followed when OOS
or OOT results are observed during stability studies. Where the 1998 guidelines
were only applicable to the finished product, the 2006 guidelines apply to APIs,
excipients, other components, in-process materials, and finished products. The doc-
ument describes the laboratory phase of the investigation as well as the full-scale
investigation and lists the responsibilities of analysts, supervisors, and the Quality
Assurance unit. Performing inadequate investigations for OOS results continues to
be a leading cause for Warning Letters in the past 5 years.
It is important that the investigation is timely, unbiased, well documented, and
scientifically sound. Typically, an investigation should be closed within 30 days of
270 N. Subbarao and K. Huynh-Ba

OOS discovery. For stability testing handled by a contract research organization

(CRO), the sponsor is ultimately responsible for any investigations. Therefore, the
responsibilities of the sponsor and CRO must be clearly defined, and the sponsor
must be familiar with the CRO’s OOS procedure.
An OOS result could be due to errors in the measurement process or in the man-
ufacturing process. Therefore, an investigation must be performed to determine the
root cause of the batch failure even if a batch is rejected.
Every pharmaceutical company must have a written procedure for Laboratory
Investigation of OOS or OOT events for GMP stability studies. The procedures to
be followed and the responsibilities of various personnel as described in the 2006
guidelines are outlined below.

13.2.1 Phase I – Laboratory Investigation

The first phase of the investigation occurs in the lab and is focused on the possible
identification of assignable laboratory errors. The responsibilities of the supervisor
and the analyst during this phase are listed below.
Analysts are responsible for:
r Ensuring that the equipment used is calibrated and meets the required acceptance
criteria.
r Reporting data only if the required system suitability tests pass acceptance
criteria.
r Checking the data for compliance to specifications before discarding any test
solutions.
r Informing the supervisor if any unexpected results are obtained.
r Stopping testing if an obvious error occurs; they should not knowingly continue
testing when they expect to invalidate the data at a later time for an assignable
cause, except when the sole purpose is to see what results are obtained when
obvious errors are known.
The supervisor is responsible for:
r Performing an objective and timely assessment.
r Confirming the analyst’s knowledge and performance of correct procedures.
r Examining the raw data and identifying anomalous or suspect information.
r Confirming the performance of the instruments.
r Examining the solutions, reagents, and standards to confirm that they were appro-
priate for use during testing.
r Evaluating the performance of the test method.
r Documenting and preserving evidence of the assessment.
Prompt initiation of the investigation is essential for several reasons. Test solu-
tions, reagents, and standard solutions will still be available and may be re-analyzed
if necessary. The analyst’s memory of all stages of the testing will be clearest on
the day of the test, and equipment is more likely to be in the configuration used for
testing and can therefore be checked for errors.
13 Evaluation of Stability Data 271

Check list for Laboratory Investigations

LIR Number _________ Issued by/Date _______________

Product Name:__________ Product Lot Number:___________ Sample ID Number:__________

Stability Study Number:________ Stability Storage Condition:_______Stability Time Point:___________

Test type Test Date Analytical Method Number: _____

Observation leading to Investigation: ___________________________________________

_______________________________________________________

Equipment ID (s) Analyst (s):

System Suitability Passed?

Sample
Sample ID and condition satisfactory? y/n
Packaging satisfactory? y/n
Reagent
Correct reagent used? y/n
Within expiry Date? y/n
Glassware/supplies
Correct glassware type used? y/n
Clean glassware used? y/n
Solvent washed/dried Glassware used? y/n
Correct Volume (volumetric) glassware used? y/n
Equipment
Equipment qualified for intended purpose? y/n
Equipment within calibration period? y/n
Equipment setting appropriate? y/n
Chromatography Column:
Correct column used as per analytical method? y/n
Column wash steps completed prior to injection? y/n
Analyst Training
Trained on use of equipment? y/n
Trained on Analytical Method? y/n
SOP steps
Weights in correct range? y/n
Dilutions performed per analytical method? y/n
All steps performed as per Analytical method y/n
Calculations
Software qualified? y/n
All calculations checked and found correct? y/n
Other y/n

Investigation by/Date: _________________________________

Laboratory Error Identified? Y/N: __________________________________________

Fig. 13.4 Example of a laboratory investigation report checklist

The investigation must be documented and a checklist (see Fig. 13.4 for an exam-
ple) is often used to aid in reviewing all the relevant facts and serves to speed up the
review process.
If the review does not reveal the root cause of the anomalous results, there may
be a need to test the final prepared solution, retained samples from earlier steps of
the sample preparation or tablet grinds to identify the root cause. The procedures
for such testing must be defined in an SOP and the testing must be supervised and
272 N. Subbarao and K. Huynh-Ba

approved by a supervisor, with a review of the results at each stage before proceed-
ing to the next.
If the anomalous result can be unequivocally assigned to laboratory error, the
result may be invalidated. Marking the notebook entry as invalid and retaining all
related instrument outputs will be invaluable during future audits, to account for the
raw data and results which are retained in the instrument electronic database.
The OOS guidance document indicates that laboratory or analyst errors should be
relatively rare, and frequent occurrence can be an indication of inadequate training
of analysts, poorly calibrated/maintained equipment, or careless work. It should not
be assumed that the failing result is attributable to analytical error without perform-
ing and documenting an investigation. When a laboratory error is confirmed, the
company must determine the source of error, take appropriate corrective actions,
and prevent reoccurrence of the incident.

13.2.2 Phase II – Full-Scale OOS Investigation

When the laboratory phase of the investigation does not identify an assignable
cause, a full-scale investigation must be initiated. The functional groups involved,
in addition to the quality unit, should be included in the investigation team. The
investigation should be initiated and completed promptly.
A standard form which will aid in documentation of investigations is provided in
Fig. 13.5.
A critical part of the investigation is a review of other related documents to
identify the root cause of the OOS result. Some of the documents to be checked
include stability data of other time points of the same lot, other lots of the same
product, other pack sizes/pack configurations of the same lot or the same product,
and the batch production record for other investigations on the same lot/same prod-
uct. The data can reveal if the anomalous data was developing at earlier time points
or whether the root cause discovered as a result of this investigation could impact
other lots, other pack sizes, and other time point data.
The investigation may also include experimental work to determine the root
cause. Such experimental work must either be described in the OOS SOP or be
pre-approved and supervised by a responsible person.

13.2.2.1 Retesting
Retesting is performed using the same homogenous material as the original sample.
The concept of retesting as described in the OOS guidance does not apply to some
tests such as content uniformity and dissolution.
Companies must have a written procedure that specifies the maximum number of
retests. The SOP must define how retesting will be performed. It is understood that
the investigation procedure cannot be fully pre-defined and depends on the prob-
lem and product. Instead, each testing step must be approved and supervised by a
responsible person in the Quality unit. It is important that the retesting be performed
13 Evaluation of Stability Data 273

OOS Investigation Form

OOS Number ____ Issued by/Date ______ Close out date __

Product Name:__________ Product Lot Number:________________ Sample ID Number:_______

Stability Study Number:________ Stability Storage Condition:___________ Stability Time Point:_______

Test type Test Date Analytical Method Number: _______

Observation leading to Investigation: ________________________________________

___________________________________________________
Equipment ID (s)

Analyst (s):

Repeated Injection authorized by:______________________

Repeated Grinds testing authorized by:_______________________________________
Repeated testing of sample authorized by:____________________________________

Lab Error identified? Y/N

(Attach Lab Investigation report including Checklist and summary of all repeated testing listed above)

Re-sampling authorization (specify lot number, stability study number, condition, time point:

Other products/lots/packaging configurations that may be afffected

Investigation report including Root Cause and Corrective/Preventive action attached? Y/N

Comments/Recommended follow-up activities:

Investigation Close-out Approval

QA Management Signature/date Laboratory Management Signature/date

Fig. 13.5 Example of an OOS investigation form

by a second analyst if available. Repeating testing until a passing result is obtained

and then discarding the originally obtained data is commonly referred to as testing
into compliance and is objectionable under the cGMPs.
Where retesting of the original sample does not lead to the discovery of the root
cause, there may be a need to re-sample the lot. For stability studies, where the
original time point often cannot be resampled, due to the passage of time, a later
274 N. Subbarao and K. Huynh-Ba

time point sample is pulled and the results are designated as such. For example,
if in a study the 6-month sample test results are under investigation, and additional
containers at the 7-month time point are tested as part of the investigation, the results
are reported as belonging to the 7-month time point. The investigation may conclude
that either the original test result or the original sample tested was not representative
of the lot and may therefore be invalidated.
When faced with insufficient samples for testing of stability OOS investigations,
some companies may consider taking samples from other programs such as reten-
tion programs. However, such practices are not advisable as the storage conditions
of the stability and retention programs may differ significantly.

13.2.3 Outlier Test

Outlier testing is a statistical procedure to determine if a value obtained is different
than others in a series. Discarding outliers may be appropriate for biological assays
that exhibit a high variability. In such cases, the outlier test must be described in
advance in a written procedure. For validated chemical tests, the guideline does not
recommend the use of outlier tests to invalidate suspect results. Furthermore, the
outlier test cannot be applied to data when the variability in the product is being
assessed, such as dissolution or content uniformity testing.

13.2.4 When the OOS Result Is Confirmed

If the investigation described above does not identify a laboratory error as a root
cause, then the OOS result is considered representative of the lot being tested.

13.2.4.1 Commercial Lots

For those products which are the subject of regulatory applications, regulations
require submittal within 3 working days of a Field Alert Report (FAR) concerning
the failure of a distributed batch to meet any of the specifications established in the
application.

13.2.4.2 Development Lots

OOS at Accelerated Conditions
For registration stability studies, for products intended for long-term storage at room
temperature, when stability study result shows significant change as defined in ICH
Q1A(R2), testing on the intermediate condition samples must be initiated immedi-
ately. Failure at accelerated conditions for registration lots may also trigger changes
to the labeling of the product. If necessary, the proposed shelf-life can be shortened
until data of long-term storage conditions is available.
13 Evaluation of Stability Data 275

OOS at Long-Term Storage Conditions

OOS in registration stability results at the long-term storage condition may trigger
changes to the product packaging, formulation, storage condition, or proposed shelf-
life of the product. The need to remove the lot from ongoing clinical studies should
be evaluated.

13.2.5 Trending OOS Results

Trending initial laboratory investigations is a convenient way to identify opportuni-

ties for process improvements in the lab. After completion of the lab investigation,
the key elements of the investigation, such as product, storage condition, equipment,
analytical method, and root cause are entered in a database. The database is period-
ically queried for occurrence rate per time period. Pareto charts are prepared for the
root cause categories. Each root cause, starting with the leading cause is addressed
as part of the lab’s continuous improvement program.
Trending of stability OOS investigations is considered best practice and is usually
included as part of the site OOS monitoring procedure.

13.3 Setting Specifications and Stability Data

The guidance for preparing specifications for drug substances and drug products is
provided in ICH Q6A [6] with additional guidance in ICH Q6B [7] for biologics.
The discussion below for considering stability data in specifications is applicable
only to drug products. The upper and lower acceptance criteria limits in the regu-
latory specification (shelf-life specification in the EU) are usually set based on the
potency and/or impurity levels of the clinical lots and safety and efficacy consid-
erations. The extent of degradation or change in the attributes during the shelf-life
of the product is factored in to determine the in-house release acceptance criteria
(lot release specification in the EU) to ensure that the product meets the regulatory
specification at the end of shelf-life.
The acceptance criteria for some attributes such as package integrity or sterility
must not differ between lot release limits and regulatory acceptance criteria, and test
results for these attributes must not change over the shelf-life of the product. How-
ever, results for other attributes such as potency and impurity profile could change
significantly over the shelf-life of the product. Stability data are used in deriving the
regulatory specification limits for these attributes.

13.3.1 Refinement of Specifications Using Data from Stability

Studies
ICH Q6A provides decision trees given in Figs. 13.6 and 13.7, which address the
extrapolation of meaningful limits on degradation products for drug substances and
drug products.
276 N. Subbarao and K. Huynh-Ba

Determine impurity level

in relevant batches
(development, pilot and
scale up batches)

Estimate maximum increase in

Determine mean +upper Is impurity Yes impurity at retest date
confidence limit also using data from relevant
for this impurity (Let this =A) a degradation accelerated and long
product term stability studies

No

No Determine maximum likely level as

Acceptance criterion = A or B Is A or B greater A + increase in degradation product
than the
(as appropriate) at appropriate storage condition
qualified level?
(Let this be =B)

Yes

Acceptance criterion = qualified level

or establish new qualified level

Fig. 13.6 Establishing acceptance criterion for a specified impurity in a new drug substance

Does degradation No Estimate maximum increase in impurity

occur during at retest date using data from relevant
product manufacture accelerated and long term stability studies
(Let this =D)

Yes

Estimate maximum increase in Determine maximum likely level as

degradation product during Drug substance acceptance criterion
manufacture from relevant batches (A or B) + C + D
(let this = C)

No
Acceptance criterion = A or B Is maximum likely level greater
(as appropriate) than the qualified level

Yes

Acceptance criterion = qualified level

or establish new qualified level or new
storage conditions or reduce shelf life

Fig. 13.7 Establishing acceptance criterion for a degradation product in a new drug product

A mathematical model for deriving specifications based on the manufacturing

capability and stability data is given below. The parameters for the shelf-life limits
are estimated by the equation
13 Evaluation of Stability Data 277

2
STOT
LRL = LR − EAC + t0.95; DF ST2 + (13.4)
n
where,

LRL = Lower release limit

LR = Shelf-life limit or Lower registration limit
STOT = Standard deviation (total) for the analytical method
DF = Degrees of freedom for S
t = Percentile in the t-distribution
n = Number of repeated and independent assay determinations for release
EAC = Average slope for degradation∗ shelf-life
ST = Standard error of slope∗ shelf-life

The mathematical model thus provides the procedure for including the degrada-
tion slope in the calculation of the specification acceptance criteria.

13.3.2 Expiry Dating of Clinical Materials

Expiry dating for clinical lots is required for clinical trials conducted in Europe.
Acceptable procedures for extrapolating expiry dates are described in ICH
Q1E.

13.3.3 Commercial Specifications and Extension of Expiration

Dating

The drug product and drug substance shelf-life and expiry periods may be extended
after product approval when satisfactory data from three stability lots has been
obtained. It may also be possible to propose excluding or replacing certain specifica-
tion tests originally included in the new drug application from the commercial drug
product specification. For example, degradation product testing may be reduced or
eliminated if it has been conclusively proven that a certain impurity is not formed
in the specific formulation and under the specific storage condition proposed in the
new drug application. Any testing reduction must be approved if the product has
been filed with regulatory authorities.

13.4 Preparation of Stability Reports

13.4.1 GMP Requirements for Records and Reports

21 CFR Part 211 Subpart J indicates that records and reports must reviewed at least
annually and be available for inspections at any time. Laboratory records include
278 N. Subbarao and K. Huynh-Ba

a complete record of data and description of samples such as storage, location,

quantity, lot, date received, etc. A complete record of instrumentation, reagents,
and standards must also be available. Stability data must be well documented in a
timely manner. Data must be traceable and defendable during inspection.
Raw data is defined as any record that is the result of original observations and
activities of a laboratory study and is necessary for the reconstruction and evalu-
ation of the report of that study. Raw data could be in laboratory notebooks, lab
data recording sheets, a laboratory information management system (LIMS), or a
combination of these means. Documentation is critical in day-to-day operations;
improper documentation continues to be a leading cause of warning letters. Under-
documentation leads to the risk of insufficient information and can contribute to
filing delays and 483’s. However, over-documentation with non-value-added infor-
mation can be confusing, time consuming and wastes valuable resources.

13.4.2 Elements of a Stability Data Sheet

Stability results from all the time points are collated into tables called stability data
sheets. Figure 13.8 presents a typical example of a stability data sheet. The data
sheet usually comprises three main sections.

13.4.2.1 Lot Information

This section contains information for identification of the study, such as product
name, strength, lot number, batch size, package, formulation identification, storage
condition, and sample orientation. This section should also include all relevant dates

STABILITY ANALYTICAL REPORT

Sample Name: Manufacturing Date: Storage condition:
Lot#: Manufacturing Site: Sample Orientation (if applicable):
Study #: Expiration Date:
Protocol #: Packaging Information:
Study Start Date: Testing Site: Packaging Date:
Study Purpose: Packaging Site:
Test Name Method Acceptance Criteria Time Zero 1 Mo 2 Mo 3 Mo 6 Mo 9 Mo 12 Mo
Test Date
Pull Date
Test Date
LIMS ID
Appearance
Assay
Impurities
Individual
Total
Dissolution
Average
% RSD
Range
Moisture

Completed By: Date:

Reviewed By: Date:
Approved By: Date:

Fig. 13.8 Example of a stability data sheet

13 Evaluation of Stability Data 279

such as manufacturing date, packaging date, and expiration date, and site informa-
tion such as manufacturing site, packaging site, and testing site.

13.4.2.2 Study Information

This section includes study number, study start date, and time points. Protocol infor-
mation and purpose of the study should also be included.

13.4.2.3 Testing Information

Testing information should include the current validated stability-indicating meth-
ods and corresponding specifications. Any modification of methods must be
recorded and justified. This section records any additional information on the study.
Most companies have an SOP describing recording of results. Consistent represen-
tation of data is required, and data rounding practices usually align the significant
figures reported with the specifications for the test. Analysis test dates must also be
included in this section.

13.4.3 Anatomy of a Stability Report

The stability data generated from the stability study, the data analysis, interpreta-
tions, and conclusions are reported at the end of the study. The stability report
contents are an important component in any regulatory submission and the report
is one of the documents reviewed in most audits and/or inspections. Figure 13.9
shows sections that contribute to a stability report.

STABILITY REPORT

Data
Primary Secondary
Stability Stability Statistical Summary
Stability Stability
Protocols Commitment Evaluation And
Data Data
Evaluation

Fig. 13.9 Content of a stability report

280 N. Subbarao and K. Huynh-Ba

13.4.3.1 Stability Commitment

The stability commitment section could include the protocol to be followed for
future stability studies. These studies can be for either the first three production
batches or for representative batches for annual product monitoring. Commitment
should be clearly made and followed through. For a new submission, a commitment
is made for the first three production batches, which follows a protocol similar to
the submission batches, and also for the annual product monitoring, which generally
contains only stability conditions at long-term storage.

13.4.3.2 Stability Summary

The stability summary is a brief section describing the stability profile of the drug
substance or drug product. This section should indicate whether all the results meet
specifications and support the proposed expiry period. Any differences among the
packages, storage conditions, etc. are also discussed.

13.4.3.3 Statistical Analysis

Statistical analysis could be performed via LIMS or by a stability statistician. More
information on the prediction of shelf-life can be found in Chapter 6.

13.4.3.4 Stability Protocols

Stability protocols attached to the stability report allow the reviewer to understand
the procedures followed for the study. Additional information on this topic is found
in Chapter 15.

13.4.3.5 Primary and Secondary Stability Data

Primary stability studies are those used to directly support the expiry dating or shelf-
life of the drug substance or drug product, while secondary stability studies are
those that provide supporting information. These could be from lab-scale batches,
development batches, or experimental batches. The contents of the stability data
tables are described above in Section 13.4.2.

13.4.4 Requirements for Stability Section in the CMC

The stability portion of the Chemistry and Manufacturing Controls (CMC) dossier
contains the sections from the stability report described above. The requirements
for the CMC sections can be found in 21 CFR Part 312 for IND application and
Part 314 for NDA and Abbreviated New Drug Application (ANDA). Tables 13.1
and 13.2 provide the requirements for the CMC, and the location of the stability
related documents within the CMC are highlighted in Table 13.2.
13 Evaluation of Stability Data 281

Table 13.1 Requirements of IND content based on 21 CFR Part 312

Drug substance r Acceptance limits and analytical methods.

r Sufficient information to support the stability of the drug substance during
the toxicological studies and the planned clinical studies.

Drug product r Brief description of the manufacturing and packaging procedure

r Acceptance limits and analytical methods
r Information sufficient to assure the product’s stability during the planned
clinical studies

Table 13.2 Structure of CTD

Drug substance 3.2.S.4 Control of Drug Substance
3.2.S.4.1 Specification
3.2.S.4.2 Analytical Procedures
3.2.S.4.3 Validation of Analytical Procedures
3.2.S.4.4 Batch Analysis
3.2.S.4.5 Justification of Specification
3.2.S.5 Reference Standards or Materials
3.2.S.6 Container Closure System
3.2.S.7 Stability
3.2.S.7.1 Stability Summary and Conclusions
3.2.S.7.2 Post-approval Stability Protocol and Stability Commitment
3.2.S.7.3 Stability Data
Drug product 3.2.P.4 Control of Excipients
3.2.P.4.1 Specifications
3.2.P.4.2 Analytical Procedures
3.2.P.4.3 Validation of Analytical Procedures
3.2.P.4.4 Justification of Specification
3.2.P.4.5 Excipients of Human or Animal Origin
3.2.P.4.6 Novel Excipients
3.2.P.5 Control of Drug Product
3.2.P.5.1 Specifications
3.2.P.5.2 Analytical Procedures
3.2.P.5.3 Validation of Analytical Procedures
3.2.P.5.4 Batch Analyses
3.2.P.5.5 Characterization of Impurities
3.2.P.5.6 Justification of Specifications
3.2.P.6 Reference Standards
3.2.P.7 Container Closure systems
3.2.P.8 Stability
3.2.P.8.1 Stability Summary and Conclusions
3.2.P.8.2 Post Approval Stability Protocol and Stability commitment
3.2.P.8.3 Stability Data

In September 2002, the ICH issued guideline M4, Organization of the Common
Technical Document (CTD) for the Registration of Pharmaceuticals for Human Use.
Each CTD contains 5 modules:
282 N. Subbarao and K. Huynh-Ba

Module 1 – Region Specific Information

Module 2 – Summaries
Module 3 – Quality (CMC)
Module 4 – Non-clinical Study Reports
Module 5 – Clinical Study Reports

Stability data is included in Module 3. Subsections of Module 3, which include

portions of the stability report described in 13.4.3 of this chapter, are highlighted in
bold in Table 13.2 above. Stress studies to support the capability of the analytical
method are also described in CMC Module 3.2 of the CTD. However, information
such as stability data of intermediates designed to support holding times are usually
included in the Manufacturing sections.

13.4.4.1 Drug Substance Stability Sections in CMC

Stability data for primary submission studies and supporting studies for drug sub-
stance are included in CMC section 3.2.S.7. The amount of stability data required at
submission depends on the stability storage condition and the proposed shelf-life of
the drug substance. A minimum of 12-months’ long-term condition storage data and
6 months’ accelerated or intermediate condition data will usually be needed on three
primary batches for products intended for storage at room temperature. Cross-over
data must be provided for any analytical method changes.

13.4.4.2 Drug Product Stability Sections in CMC

CMC section 3.2.P.8 includes stability data for drug products. The amount of sta-
bility data required at submission depends on the intended storage condition and
the proposed shelf-life of the drug product. A minimum of 12 months’ long-term
storage condition data and 6 months of accelerated study data is usually required
for products intended to be stored at room temperature.

13.5 Conclusions
Stability raw data and results must be reviewed and evaluated promptly after the
analysis. The analyst must also review the stability profile of the batch, as well as
stability data of the product after each data point generated. Many companies have
implemented LIMS to help making reporting and evaluating stability data more effi-
cient. The stability report is an important segment in the CMC document package.
Every company must have an OOS and OOT SOP. If a laboratory error cannot be
shown to be the root cause of an OOS or OOT incident, then a cross-functional
investigation must be initiated. OOS and OOT investigations are important, as they
continue to be one of the leading causes of warning letters.
13 Evaluation of Stability Data 283

References
1. International Conference on Harmonization (2003) Q1A(R2): Stability testing of new drug
substances and products (second revision).
2. International Conference on Harmonization (2003) Q1E: Evaluation of stability data.
3. PhRMA CMC Statistics and Stability Expert Teams (April 2003) Identification of out-of-trend
stability results. Pharm Technol pp 38–52.
4. FDA (1998) Guide to inspection of quality control laboratories.
5. FDA/CDER (October 2006) Guidance for industry: Investigating out-of-specification (OOS)
test results for pharmaceutical production.
6. International Conference on Harmonization (1999) Q6A: Specifications: test procedures and
acceptance criteria for new drug substances and new drug products: chemical substances.
7. International Conference on Harmonization (1999) Q6B: Specifications: test procedures and
acceptance criteria for biotechnological/biological products.