Track I, Session I: Chemical

Medicines and Excipients-Evolution

of Validation Practices
Wednesday, April 17, 2013 (9:00 a.m. to 11:00 a.m.)
IPC–USP Science & Standards Symposium
Partnering Globally for 21st Century Medicines
Moderator: Milind Joshi, Ph.D.
Chair, USP South Asia Stakeholder Forum
Acceptable Analytical Method Variation
Setting System Suitability
Requirements
Todd L. Cecil, Ph.D.
Vice President, Chemical Medicines
USP
Method Variation

 Sources
– Instrument Characteristics
– Sample Characteristics
– Method Parameters
– Environmental Affects
– Analyst
– Instrument Settings
– Many others
 Measuring Variation

 Random Error  Systematic Error

– Indeterminate Error – Determinate Error
– Experimental error – Discoverable source
(in theory)

Estimated using Precision Estimated using Accuracy

Validation to Measure Variability

 Developed in late 1980’s for the Pharmaceutical

Industry
– PhRMA -> USP <1225> -> ICH Q2A -> USP <1225>
 Defined “Analytical Performance Characteristics”
– Accuracy Intermediate precision
Repeatability
– Precision
Reproducibility
– Specificity Ruggedness
– Detection Limit Robustness

– Quantification Limit Trueness

– Linearity Bias

– Range
Acceptable Variability

Depends upon two factors

Application Expectation

 Test • Analyst Experience

 Procedure • Instrument knowledge

 Acceptance criteria • Matrix complexity

Acceptable Validation

 ICH and USP do not describe acceptable limits

 Therefore, Acceptable Validation/ Variation is
open to interpretation by:
– Bench Chemist
– Supervisory Chemist
– Regulatory Affairs Professional
– The Regulator
– The Pharmacopeial Professional
– And fights ensue…
There is Another Way!

 Recent publications
– Pharma’s Analytical Target Profile (ATP)
– USP’s Performance-Based Procedures
 Upcoming publications
– USP Validation and Verification Expert Panel
– USP Statistics Expert Committee
– USP “Requirements for Compendial Validation
<1200>” (working title)
Defining Another Way Forward

 Critical Validation Parameters

– What are the critical features (parameters) of
an acceptable procedure?

 Procedure Performance Measures

– How do we measure the critical parameters?

 Procedure Performance Acceptance Criteria

– What defines “good enough” for each
performance measure?
 Measuring the Parameters

 Precision – % RSD with sufficient degrees of freedom

 Accuracy – Spike Recovery or Comparison to Primary
Standard
 Specificity – Resolution or Spike Recovery
 Linearity – Slope, Intercept, R2
 Range – Precision and accuracy
 Limit of Detection – Precision
 Limit of Quantification – Precision
 Collapsing the Parameters

 Precision – Measure of Random Error

 Accuracy – Measure of Systematic Error
 Specificity – Measure of Systematic Error
 Linearity – Measure of Systematic Error
 Range –?
 Limit of Detection – Measure of Random Error
 Limit of Quantification – Measure of Random Error

 Why are we measuring things so many different ways?

 Does agreement mean quality? or are we hiding behind tradition?
 IF we combine critical components can we gain efficiency?
 Extract from <1200>

Category I* Category II* Category II* (Semi-

(Quantitative) quantitative)
Analytical
Performance <1225> <1200> <1225> <1200> <1225> <1200>
Characteristics
Accuracy Y 1 Y 4 ? N
Precision Y 1 Y 5 N 5
Specificity Y 2 Y 2 Y 2
Detection Limit
N N N N Y 6

Quantification
N N Y 4 N N
Limit
Linearity Y 3 Y 4 N N
Range Y 3 Y 4 ? N
1 Covered in the Precision-Accuracy Study
2 Covered in the Specificity Study
3 Covered in the Range Study
4 Covered in Accuracy Study
5 Covered in Precision Study
6 Covered in the Detectability Study
 Precision and Accuracy Study

 When properly combined Precision and Accuracy yield a

probability of passing.
Bias-%CV Tradefoff, 98%-102% limits, True Value = 100, Prob'y Passing 0.95

1.2

0.8
%CV

0.6

0.4

0.2

0
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60 1.80 2.00
Bias
 Study Detail

 Precision
– % RSD of 6 independent samples at 100%
 Accuracy
– Δ from RS label at 100%
– The data obtained for Precision can be used for
Accuracy
Combine with Acceptance Criteria to calculate
probability
Result = NORMDIST(Upper, Mean, SD, TRUE)
-NORMDIST (Lower, Mean, SD, TRUE)
 Limit: NLT 0.95
 Specificity Study

 Specificity is a special case of Accuracy.

 Interferences considered
 Separation Sciences
– Resolution of NLT 1.5
Specificity Study

 Non-Chromatographic Procedures are harder

 Spiked samples with interferences
 Measure the error caused by the addition of an
interference
 Limit is linked to the acceptance criteria of the
analyte
– The error caused by all interferences cannot
exceed the allowable bias from the Precision-
Accuracy Study
Range Study

 Retasked Range
 Precision-Accuracy evaluation at 80%, 90%,
100%, 110%, and 120%
 Instead of Mean in the calculation, use recovery
value
 Recovery Value = [Mean]/[Known]*100%
 Limit: Each concentration is NLT 0.95
Linearity

 Response vs Concentration
 Calibration curve
 Technique dependent application
 Calculated vs Known Concentration
 Slope =1
 Intercept =0
 Accuracy evaluation
 How do you measure linearity?
 Slope: not correlated to error
 Intercept: not correlated to error
 R2: limited correlation to error
Linearity

 Slope and Intercept

– Overwhelmed by random noise
– Not correlatable to systematic noise
– Adds no additional information
 Basis for Linearity is not supported
 Range adequate
Limit of Detection

 Only applies to “Limit” procedures

 S/N of 3
– independent samples at LOD
– “adequate precision and accuracy”
 What is adequate?
 What is the purpose of the test?
Limit Test

 Measure a Standard solution of the impurity at the limit

 Measure a Sample solution
– Is the response of the impurity in the Sample < Standard

– Pass
– Is the response of the impurity in the Sample ≥ Standard

– Fail
 Is the Δ between pass/fail adequate?
 If the limit is 0.1%, then acceptable values are
– 0.14% to 0.05%
– LOD does not assure the measurement
–Detectability does.
Detectability

 A new term included in <1200>

 Replaces LOD
3 steps
– 1: Standard of impurity at limit
– 2: Sample spiked with impurity at limit
– 3: Standard spiked with impurity at 100%-%RSD* for the impurity
– *can be estimated with Horwitz

 If 1=2 and 3<2, then the difference is detectable

 Otherwise, procedure is not adequate
What is Horwitz?
Limit of Quantitation

 Why do we make this measurement?

–10x S/N . . .
 A meaningful quantity in development?
–Yes
 Necessary to validate?
–No
 Validation presumes
–A known procedure
–A typical value for the analyte
–Known acceptance criteria
 You already know the typical range of the analyte…
 Use the Range Study
–80%-120% for Assay; 50%-150% for impurities
Summary

 USP is challenging validation concepts

 Including DOE and QbD through the ATP
 Include measurable parameters and clear criteria
 Focus on Precision and Accuracy results
 Use Specificity to aid understanding of Accuracy
 Retask Range
 Introduce detectabiltiy
 Eliminate LOD, LOQ and Linearity
 But Wait, There’s More…

 Setting System Suitability Requirements

 Validation
is measured only once
 System suitability is measured on a daily basis
– Traditionally uses instrument dependent measurements

– Resolution
– Tailing
– %RSD
 System suitability rarely linked to variance
 UseValidation protocol to evaluate Precision and Accuracy
across the days run.
 System suitability can then be linked to validation
 Specificity
should represent necessary minimums, but should
exceed criteria of validation
Precision, Accuracy & Linearity
Harry Yang, Ph.D.
Member, USP Statistics Expert Committee
Method Validation

 Validation is a snapshot (at any given time) of

the assay’s performance.
 It is confirmation that the assay is fit for its
intended use
 Required by regulatory guidelines
History of ICH Guidelines on Method Validation
Other Related Guidelines
Common Validation Characteristics

 Accuracy
 Precision
 Repeatability
 Intermediate precision
 Specificity
 Limit of detection
 Limit of quantitation
 Linearity
 Range
Common Validation Characteristics

 Accuracy
 Precision
 Repeatability
 Intermediate precision
 Specificity
 Limit of detection
 Limit of quantitation
 Linearity
 Range
Precision

 Closeness between a series of measurements

obtained from multiple sampling of the same
homogeneous sample
Precision

 Repeatability: intra-assay precision. Usually

same day, operator, equipment
 Intermediate Precision: same laboratory but
different operators, equipment, etc.
 Reproducibility: precision between laboratories
 Expressed as standard deviation (SD) or relative
standard deviation (RSD)

n n

 Xn (X i  X )2
SD
X i 1
, SD  i 1
, RSD 
n n 1 X
Accuracy

 The closeness of agreement between the value

which is accepted either as a conventional true
value or an accepted reference value and the
value found (ICH Q2(R1))
Bias Precision

E[( X  T ) 2 ]  (  X  T ) 2  var[ X ]

True value Mean measurement

Bias

 Closeness of agreement between the average

value obtained from a large series of test results
and an accepted reference value

Bias   X  T

µT µX
Accuracy = bias + precision
 Validation of Accuracy and Precision

 Model

or

or
An Example

Test result Bias Intermediate Repeatability

X  T  X  T precision

Burdick, LeBlond, Sandell, Yang, 2013

An Example

Test result Bias Intermediate Repeatability

X  T  X  T precision

Burdick, LeBlond, Sandell, Yang, 2013

 Assessment of Bias: Traditional Approach
X
 t n 1, 0.025
s/n

 Test the hypothesis that bias = 0

H0: µ = 0 vs. H1: µ ≠ 0

Y
 t n 1,0.025
Reject H0 if s2 / n (which is the same as p-value < 0.05)

n n

Yn  (Y  Y )
i
2

where Y  i 1
, s2  i 1
, t n 1,0.025 - cutpoint of t-distribution
n n 1
Assessment of Bias: Traditional Approach

 P-value < 0.05 is equivalent to that the 95%

confidence interval contains zero, i.e.
 
0   Y  t n 1, 0.025s / n , Y  t n 1, 0.025s / n 
 

p-value < 0.05

p-value ≥ 0.05
Issue with the Traditional Approach

 Penalize more precise assay

 Award small sample size

 
0   Y  t n 1, 0.025s / n , Y  t n 1, 0.025s / n 
 

With of the 90% CI is

proportional to assay
precision (s) and
reciprocal
of the squared root of
sample size n.

Huberta et al, 2004

Equivalence Method

 Bias is deemed acceptable if the 90%

confidence interval of bias is bounded by pre-
specified acceptance limits (e.g., ±15%)

 
 Y  t n 1, 0.025s / n , Y  t n 1, 0.025s / n 
Huberta et al, 2004  
Comparison Between Significance and Equivalence

Is bias acceptable?
Significance Equivalence
Yes No

Yes No

Yes Yes

No Yes
Equivalence Method

 Bias is deemed acceptable if the 90%

confidence interval at each concentration level is
contained with in pre-specified range (LAL, UAL)

Plot of Bias vs. True Value

UAL

LAL

True value
a b c
 Accessing Conformance to Acceptance Criteria: Precision

 Intermediate precision is considered acceptable

if the 95% confidence interval is bounded by a
pre-selected number UAL

< UAL

Burdick, LeBlond, Sandell, Yang, 2013

 Total Error Approach

 Bias cannot be assessed independent of

precision

Huberta et al, 2004; Hoffman & Kringle, 2007

Total Error Approach

 Measured value = True value + Method Bias + Method error

Y = Test result - True value

Total Error Approach

 Accuracy of a method is acceptable if it is very

likely that the difference between every
measurement of a sample and the true value is
inside pre-chosen acceptance limits

Huberta et al, 2004

Total Error Approach

 Risk = 1 - Probability of meeting acceptance criterion

Huberta et al, 2004

Methods for Testing H0:

 Beta-expectation tolerance interval (Huberta et al, 2004)

 With 100β% confidence that bias of a future measurement is bounded
by λ
 Average (expected) probability for bias of a future observation is no
smaller than 100β%

 Beta-content tolerance interval (Hoffman & Kringle,

2007)
 With 100γ% confidence that bias of 100β% future measurements is
bounded by λ

 Bayesian analysis (Burdick, LeBlond, Sandell, Yang,

2013)
 Conditional on validation data, probability for bias of a future
observation is no smaller than 100β% P(  Y  X     | data)   .
T
Accuracy Profile

Huberta et al, 2004

References

1 Graybill FA, Wang CM. Confidence intervals on nonnegative linear combinations of variances. J Am Stat Assoc. 1980;75:869–
873.
2. Nijhuis MB, Van den Heuvel ER. Closed-form confidence intervals on measures of precision for an interlaboratory study. J
Biopharmaceutical Stat. 2007;17:123–142.
3. Satterthwaite FE. An approximate distribution of estimates of variance components. Biometric Bull. 1946;2:110–114.
4. Huberta P, Nguyen-Huub JJ, Boulangerc B, et al. Harmonization of strategies for the validation of quantitative analytical
procedures: a SFSTP proposal—part I. J Pharm Biomed Anal. 2004;36:579–586.
5. Huberta P, Nguyen-Huub JJ, Boulangerc B, et al. Harmonization of strategies for the validation of quantitative analytical
procedures: a SFSTP proposal—part II. J Pharm Biomed Anal. 2007;45:70–81.
6. Huberta P, Nguyen-Huub JJ, Boulangerc B, et al. Harmonization of strategies for the validation of quantitative analytical
procedures: a SFSTP proposal—part III. J Pharm Biomed Anal. 2007;45:82–96.
7. Mee RW. b-expectation and b-content tolerance limits for balanced one-way ANOVA random model. Technometrics.
1984;26:251–254.
8. Hahn GJ, Meeker WQ. Statistical Intervals: A Guide for Practitioners. New York:Wiley; 1991:204.
9. Hoffman D, Kringle R. Two-sided tolerance intervals for balanced and unbalanced random effects models. J Biopharm Stat.
2005;15:283–293.
10. Montgomery D. Introduction to Statistical Quality Control. 3rd ed. New York: Wiley; 1996:441.
11. Kushler RH, Hurley P. Confidence bounds for capability indices. J Quality Technol. 1992:24(4):188–195.
12. Wolfinger RD. Tolerance intervals for variance component models using Bayesian simulation. J Quality Technol.
1998;30:18–32.
13. Ntzoufras I. Bayesian Modeling in WinBUGS. New York: Wiley; 2009:308–312.
14. Spiegelhalter D, Thomas A, Best A, and Gilks, W (1996) BUGS 0.5 Examples Volume 1(version i), Example 7, Dyes, pp 24-
26. Available from http://www.mrc-bsu.cam.ac.uk/bugs/documentation/Download/eg05vol1.pdf (accessed November 20,
2012).
15. Burdick R, LeBlond D, Sandell D, Yang H. Statistical methods for validation of method accuracy and precision.
Pharmacopeia Forum, May –June Issue, 39 (3)
.16. USP. USP 36–NF 31, Validation of Compendial Procedures <1225>. Rockville, MD: USP; 2013:983–988.
17. ICH. Validation of analytical procedures: text and methodology Q2(R1). 2005.
http://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q2_R1/Step4/Q2_R1__Guideline.pdf.
Accessed 27 November 2012.
Linearity
Two Types of Linearity

 Response vs concentration linear curve

 This is a calibration curve
 It provides a means to convert a signal to a desired measured
value

 Predicted concentration vs known concentration

 This is a surrogate for Accuracy
 Slope should be 1 and intercept should be 0

Todd L. Cecil, Personal communication, 2013

Calibration Curve

 We wish to measure the concentration of an

analyte in a test sample.
 Standards = known concentrations of an analyte
 To estimate the concentration, we create a
standard curve
Standard Curve

Novick and Yang, 2013

Standard Curve

Sample

Novick and Yang, 2013

Test of Linearity - ICH Q2(R1) guideline

 Evaluate linearity by visual inspection

Novick and Yang, 2013

Test of Linearity – Pearson Correlation

r=1 r = -1

r=0
Test of Linearity – Lack of Fit (LOF)

 Determine how close the predicted values to the mean

values at each concentration level

Evidence of lack of fit

The EP6-A Guidelines

 Clinical and Laboratory Standards Institute

 http://www.clsi.org/source/orders/free/ep6-a.pdf
 Compare straight-line to higher-order polynomial
curve fits
 Recommendation: Test higher-order coefficients.

Novick and Yang, 2013

The EP6-A Guidelines

Novick and Yang, 2013

Literature

Novick and Yang, 2013

Drawbacks of Significance Test

 Conduct hypothesis testing with linearity claim

as the null hypothesis
 Rely on failing to reject the null hypothesis to
conclude linearity
 Penalize precise assay
 Award small sample size
More Literature

Novick and Yang, 2013

Two Practical Approaches

 Two one-sided tests (TOST) for calibration error

 Estimate bias in concentration due to approximating either
quadratic curve or proportional model using linear line
 Bias is expressed as a function of a ratio of two model
parameters. Thus the Fieller’s Theorem can be applied to
obtained 90% confidence interval of the bias
 Akaike information criterion (AIC)
 Based on the principle of parsimony – the smallest possible
number of parameters for adequate representation of the data

where N – total number of data points, and K – the total number

of estimated regression model parameters
LeBlond, Tan and Yang, (2013a, 2013b)
Estimating Calibration Bias: Linear vs Quadratic

Models:

Bias:

Assumption: Concentration levels used in the experiment are

symmetrically spaced.

LeBlond, Tan and Yang, (2013a, 2013b)

90% Confidence Interval (CI) of Bias

Fieller’s exact 90% confidence Interval

Linearity is accepted if the above 90% CI is contained

Within pre-specified limits.

LeBlond, Tan and Yang, (2013a, 2013b)

Linear Model vs Proportional Model

Models:

Bias:

LeBlond, Tan and Yang, (2013a, 2013b)

 90% Confidence Interval of Bias

90% CI of ratio of two model parameters:

90% CI of bias in concentration:

Linearity is accepted if the above

90% CI is contained
Within pre-specified limits.

LeBlond, Tan and Yang, (2013a, 2013b)

Test Linearity for More General Experiment Design Conditions

 An equally-spaced experimental design is not a

necessary condition
 Linearity can be tested under general conditions

Yang, Novick and LeBlond, 2013; Novick and Yang, 2013

References
1. USP. USP 36–NF 31, Validation of Compendial Procedures <1225>. Rockville, MD: USP; 2013:983–988.
2. ICH. Validation of analytical procedures: text and methodology Q2(R1). 2005.
http://www.ich.org/fileadmin/Public_Web_Site/ICH_Products/Guidelines/Quality/Q2_R1/Step4/Q2_R1__Guideline.pdf.
Accessed 27 November 2012.
3. Clinical and Laboratory Standards Institute. EP06-A02 Evaluation of the linearity of quantitative measurement procedures: a
statistical approach. 2003. http://www.techstreet.com/standards/clsi/ep06a?product_id=1277866. Accessed 27
November 2012.
4. Anscombe FJ. Graphs in statistical analysis. Am Stat. 1973;27(1):17–21.
5. Van Loco J, Elskens M, Croux C, Beernaert H. Linearity of calibration curves: use and misuse of the correlation coefficient.
Accred Qual Assur. 2002;7:281–285.
6. Bruggemann L, Quapp W, Wennrich R. Test for nonlinearity concerning linear calibrated chemical measurements. Accred
Qual Assur. 2006;11:625–631.
7. Mandel J. (1964) The Statistical Analysis of Experimental Data. New York: Wiley; 1964.
8. Mark H, Workman J., Chemometrics in Spectroscopy, Linearity in Calibration How to Test for Non-linearity, Spectroscopy
2005;20(9):26–35
9. Liu J, Hsieh E. Evaluation of linearity in assay validation. In: Encyclopedia of Biopharmaceutical Statistics. 2nd ed. London:
Informa Healthcare; 2010:467–474
10. Finney DJ. Statistical Method in Biological Assay. 2nd ed. London: Charles Griffin; 1964:27–29.
11. Berger RL, Hsu JC. Bioequivalence trials, intersection-union tests and equivalence confidence sets. Stat Sci.
1996:11(4):283–319.
12. Burnham KP, Anderson DR. Model Selection and Multimodel Inference: A Practical Information–Theoretic Approach. 2nd
ed. New York: Springer; 1998:31.
13. Burnham KP, Anderson DR. Multimodel inference: understanding AIC and BIC in model selection. Sociol Meth Res.
2004;33(2):261–304.
14. David LeBlond, Charles Y Tan, Harry Yang (2013), Confirmation of Analytical Method Calibration Linearity, Pharmacopeial
Forum 39(5), pp XX – XX.
15. David LeBlond, Charles Y Tan, Harry Yang (2013), Confirmation of Analytical Method Calibration Linearity: Practical
Application, Pharmacopeial Forum ??(??), pp ?? – ??.
16. Steve Novick and Harry Yang (2013), Directly Testing the Linearity Assumption for Assay Validation, Accepted for
publication in Journal of Chemometrics.
17. Steve Novick and Harry Yang (2013), Directly Testing the Linearity Assumption for Assay Validation, Accepted for
publication in Journal of Chemometrics, The 36th Mid-west Biopharmaceutical Statistics Workshop, Muncie, Indiana,
May, 2013i
18. Harry Yang, Steve Novick and David LeBlond (2013). Testing linearity under general experimental conditions. In
preparation.
Lifecycle Management of Analytical
Procedures
Joachim Ermer, Ph.D.
Member, USP Validation and Verification Expert Panel
Objectives of Expert Panel Validation & Verification

 Adaptation of the lifecycle concept [ ICH Q8] and of

modern concepts for process validation
to analytical procedures
 to holistically align analytical procedure variability with the
requirements of the product to be tested
 to demonstrate that the analytical procedure meets the
predefined criteria over the whole lifecycle
 to facilitate continual improvement
 Proposal to revision and compile USP General Chapters <1225>,
<1226> and <1224> into a single General Information Chapter on
Lifecycle Management of Analytical Procedures
 Stimuli article to be published in PF 39(5), Sep - Oct 2013
Quality by Design – Also Relevant for Analytics

 “systematic approach that begins with

predefined objectives and emphasizes product
and process understanding and process control,
based on sound science and quality risk
management” [ICH Q8]
 systematic approach that begins with
predefined objectives and emphasizes analytical
procedure understanding and analytical control,
based on sound science and quality risk
management”
 Alignement of Process and Analytical Procedure

PROCESS ANALYTICAL PROCEDURE

Quality Target Analytical Target

Product Profile Profile

Prospective summary of the quality Defines the objective of the test

characteristics of a drug product and quality requirements
to ensure quality, safety, efficacy for the reportable result
Analytical Target Profile (ATP)

 Developed starting 2008 by EFPIA / PhRMA Working

Group “Analytical Design Space”
 M. Schweitzer, M. Pohl et al.: QbD Analytics. Implications and
Opportunities of Applying QbD Principles to Analytical
Measurements, Pharmaceutical Technology, Feb. 2010, 2-8
http://pharmtech.findpharma.com/pharmtech/article/articleDetail.
jsp?id=654746
 Quality (data) attributes of the reportable result
 performance requirements for use
 accuracy and measurement uncertainty including precision
Analytical Target Profile (ATP)

 Based on the understanding of the target measurement

uncertainty
 Maximum allowed uncertainty to maintain acceptable levels of
confidence

 Reference point for assessing the fitness of an analytical

procedure
 towards predetermined performance requirements
 In development phase and during all changes within the lifecycle
 linked to the purpose, not to a specific analytical technique.
Analytical Target Profile (ATP)

 Any analytical procedure that conforms to the ATP is

acceptable
 USP Medicines Compendium, General Chapter <10>
 May be also established for existing procedures
 including compendial procedures
 based on (monograph) specifications, existing knowledge
ATP Example Assay

 The procedure must be able to quantify [Analyte]

 in presence of X, Y, Z
 over a range of A% to B% of the nominal
concentration
 with an accuracy and uncertainty such that the
reportable result falls
 within ±1.0% of the true value
 with at least a 90% probability
 determined with 95% confidence
Three Stage Approach to Analytical Validation

 Aligned with process validation terminology:

Stage 1

Analytical Control Strategy

Knowledge management
Procedure Design (Development and Understanding

Risk assessment
Stage 2
Procedure Performance Qualification (PPQ)

Stage 3
Continued Procedure Performance Verification

Changes
Stage 1 – Procedure Design

 According to ATP requirements

 Procedure selection, development and
understanding
 Identification and investigation of potential
analytical variables
 Risk assessment
 Robustness studies (Method Design Space)
 Analytical Control Strategy
 Knowledge gathering and preparation
Stage 2 - Procedure Performance Qualification (PPQ)

 Confirmation the analytical procedure, operated in the

routine environment is capable of delivering
reproducible data which consistently meet the ATP
 Includes analytical transfer
 Implementation of compendial procedures
 Precision study to finalize the Analytical Control Strategy
 e.g. format of the reportable result (number of determinations)
 May / should be built on results generated in Stage 1
 Iterative character of procedure development/optimisation
Stage 3 – Continued Procedure Performance Verification

 To provide ongoing assurance that the analytical

procedure remains in a state of control throughout its
lifecycle
 Routine Monitoring: Ongoing program to collect and
process data that relate to method performance, e.g.
 from analysis / replication of samples or standards during batch
analysis
 by trending system suitability data
 by assessing precision from stability studies
[J. Ermer et al.: J. Pharm. Biomed. Anal. 38/4 (2005) 653-663]
Continual Improvements (Changes)

 Risk assessment to evaluate

 Impact of the respective change
 Required actions to demonstrate (continued)
appropriate performance
 Accordingly, apply
 Stage 3 (if within Method Design Space)
 Stage 2 (e.g. transfer)
 Stage 1 (e.g. outside Method Design Space, new
procedure)
2010-2015 V&V Expert Panel

 Gregory P. Martin, (Chair) Complectors Consulting

 Kimber L. Barnett, Pfizer Inc.
 Christopher Burgess, Burgess Analytical Consultancy, Ltd.
 Paul D. Curry, Abbvie,
 Joachim Ermer, Sanofi-Aventis GmbH
 Gyongyi S. Gratzl, Ben Venue Laboratories, Inc.
 Elizabeth Kovacs, Apotex, Inc.
 David J. LeBlond, Statistical Consultant
 Rosario LoBrutto, Teva Pharmaceuticals USA
 Anne K. McCasland-Keller, Eli Lilly & Company
 Pauline L. McGregor, PMcG Consulting
 Phil Nethercote, GlaxoSmithKline
 David P. Thomas, Johnson & Johnson Pharmaceutical R&D
 M. L. Jane Weitzel, Quality Analysis Consultants
 Government Liaison(s): Lucinda F. Buhse, FDA
 USP Scientific Liaison(s):
Todd L Cecil, Kenneth Freebern, Walter Hauck, Horacio N. Pappa, Tsion Bililign
Analytical Method Validations
Current Practices and Industry
Perspective
Rajiv A. Desai, Ph.D.
Dishman Pharmaceuticals and Chemicals Ltd.
ATP Example Impurity

 Impurity: The procedure must be able to quantify

[impurity] relative to [drug]
 in the presence of components likely to present in the
sample
 over the range from reporting threshold to the
specification limit.
 The accuracy and precision of the procedure must be
such that the reportable result falls
 within ± X% of the true value for impurity levels from 0.05% to
0.15% with 80% probability with 95% confidence,
 and within ± Y% of the true value for impurity levels >0.15%,
with 90% probability determined with 95% confidence.
 Purpose of the Analytical measurement is
to get consistent, reliable and accurate data
 Source of Impurities in the Drug Substance and products

Origin of Impurities

From Equipment and

Impurities in Drug Substances Packaging material

Earlier stage
material
Residual solvents

Side reactions Degradents Extractables Leachables

Genotoxic Impurities
General Process for the Synthesis of Drug Substance

Stage 1
Solvent W
A + B C + ( traces of A and B )

Stage 2
Solvent X
C + D E+ ( traces of C and D ) + M ( reaction between A and C )
Reagent R

Stage 3
Solvent Y
E + F Crude API + ( traces of E and F ) + traces of D + degradent of E
Metal catalyst
Stage 4
Solvent Z
Crude API Final API + Traces of earlier stage material
Side reactions
Degradents
Solvents
Reagents
 Analytical Method Validation Criteria ….

- Suitability of Instrument

 - Status of Instrument Qualification and calibration

 - Suitability of reference standard , reagent, placebo, etc

 - Suitability of documentation, written analytical procedure

 approved protocol with pre-established acceptance criteria

USP General Chapter <1224>

Transfer of Analytical Procedures

1. Comparative testing of same lot or standards

2. Co-validation between laboratories

3. Complete or partial validation of Analytical procedures

by receiving laboratory and hence a transfer waiver
USP General Chapter <1225>

Validation of Compendial Procedures

As per cGMP regulations 211.194(a), the test methods

with established specifications, must meet standards of
accuracy and reliability

As per 211.194(a)(2) users are not required to validate

the accuracy and reliability of these methods. But
verify their suitability under actual conditions of use.
 Data Elements Required to be Validated

Analytical Category II
Performance Category I Category III Category IV
Charecteristics Quantitative Limit tests
Accuracy Yes Yes * * No

Precision Yes Yes No Yes No

Specificity Yes Yes Yes * Yes

Detection limit No No Yes * No

Quantitation limit No Yes No * No

Linearity Yes Yes No * No

Range Yes Yes * * No

• May be required, depending on the nature of the specific test

Category I : Procedures for Quantitation of major component or Active substance

Category II : Procedures for determining Impurities

Category III : Procedures for determining performance characteristics ( eg., dissolution, drug release, etc )

Category IV : Identification Tests

 USP General Chapter <1225>

Rationale for revisiting the compendial method

An appropriate justification for a testing procedure

Elaborating the capability of the proposed method over other

types of determinations.

For revisions, a comparison should be provided for the

limitation of the current method and advantage offered by the
new method.
USP General Chapter <1226>

Verification of Compendial Procedures

Verification for a compendial test procedure is an

assessment of whether the procedure can be used for its
intended purpose, under actual conditions of use for a
specific drug substance or drug product.

User should have the appropriate experience, knowledge

and training to understand and be able to perform the
compendial procedure.
USP General Chapter <1226>

Verification of Compendial Procedures

If the verification of the compendial procedure is not

successful and the USP staff is unable to resolve the
problem, it may be concluded that the procedure may not
be suitable for use

It may be necessary to develop and validate an alternate

procedure. This alternate method can be submitted to
USP , along with appropriate data to support the inclusion
or replacement of the current compendial procedure.
 US General Chapter <1226>

Verification of Compendial Procedures

Method verification should be based on an assessment of

the complexity of both the procedure and the material to
which the procedure is applied

Verification should assess whether the compendial method is

suitable for the drug substance and the drug product matrix.
Taking into account the drug substance synthetic route, the
method of manufacture for the drug product or both.
 US General Chapter <1226>

Verification of Compendial Procedures

Drug substance from different suppliers may have different

impurity profile that may not necessarily be addressed by the
compendial method

Excepients in the drug products can vary widely among

manufacturers and may interfere directly or cause formation
of impurities that are not considered by the compendial
procedure.
 US General Chapter <621>

Chromatography

System Suitability is an integral part of chromatography

methods

These are based on the concept that equipment, electronics,

analytical operations and samples analysed constitute an
integral system that can be evaluated as such.

Factors affecting chromatography

Mobile phase Composition, strength , temperature, pH, flow rate

Column Flow rate, dimention, Temperature, pressure, Stationary phase

 US General Chapter <621>

Adjustments allowed in HPLC Compendial methods

pH of Mobile phase : ± 0.2 units

Concentration of salts in buffer : within ± 10%

Ratio of components in mobile phase : ± 10%

Wavelength : ± 3 nm Column length : ± 70 %

Flow rate : ± 50% Column Temperature : ± 10 deg C

Injection volume : Can be reduced, but not increased

 US General Chapter <621>

Adjustments allowed in GC Compendial methods

Gas carrier flow rate : ± 50 %

Oven temperature : ± 10%

Temperature program : ± 20 %

Column length : ± 70 %

Injection volume and split volume : Can be adjusted, if detection

and repeatability are satisfactory
 Techniques Used for Analysis

Additional testing parameters are now considered along with the

conventional methods

Analytical Instruments moving from Research to Quality Control

NMR ICP XRD

LC-MS GC-MS NIR

Used mainly for low level detections of impurities

Method validation parameters to be selected appropriately along

with sampling and sample preparations
 QbD and PAT

Quality by Design (QbD ) is being encouraged by the Regulatory

guidelines, the analysis conducted at every step of the process
needs to be reliable.

Testing methods adopted under the Process Analytical technology

(PAT) should be able to provide real time analysis in the shortest
possible time.

Validation should be definitely done for analytical methods used

under the QbD and PAT environment. No matter what the stage
of the process and not just restricted to final product.

A validated method gives assurance of process control at each

stage, concept of QbD will be further reinforced.
Compendial Method and Non-compendial Method

Compendial Method  Verification / Validation

Non-compendial methods  Validation

Alternate to Compendial method  Validation + Equivalence

Potential Genotoxic Impurity (PGI)

Genotoxic Compounds have a potential to damage DNA at

any level of exposure. Its scientifically proved that there
are certain chemical structures which damage the DNA.

The accepted levels of such chemicals is required to be

maintained at a very low to avoid any cause of concern.
 Potential Genotoxic Impurity (PGI)

When can a specification of a drug substance exclude a limit of Potential

Genotoxic Impurity ?

1. Is just a theoretical impurity, but not found during manufacturing.

2. Is formed or introduced in intermediate steps and is controlled in the

intermediate stage and does not exceed 30% of the limit derived by TTC or
any defined acceptable limits

3. Is formed or introduced in final synthesis step, it should be included.

4. However, it is possible to apply skip test if the level does not exceed 30% of
the limit. Data of atleast 6 consecutive pilot scale batches or 3 consecutive
production batches would support the justification

Method validation becomes a very important aspect

which ever stage the analysis is performed

Guideline on the limits of genotoxic impurities' (EMEA/CHMP/QWP/251344/2006),

 Potential Genotoxic Impurity (PGI)

Threshold of toxicological concern (TTC) values for genotoxic impurities

above 1.5 μg /day will be treated on a case-by-case basis. For short-duration
treatments, the acceptability of higher levels will be in line with the principles
outlined below

Duration of Single ≤1 month ≤3 ≤6 ≤12

Exposure dose months months months
Allowable 120 μg 60 μg 20 μg 10 μg 5 μg
daily intake

For more than one PGI in a drug substance, the TTC limits will be individually
applied, if the impurities are structurally different.

For more than on PGI, but structurally similar, it is expected that the mode of
action would be same, hence a sum of the limits will be accepted.
Regulatory Audit Warning Letter

Your firm did not validate analytical methods used to test APIs.

The inspection revealed that your firm had not validated the HPLC
method for assay and related substances for finished API for human
use..

Your response states that XX of the APIs manufactured at your

facility, are compendial products. The remaining YY % are non-
compendial APIs had no method validation. You committed to complete
these method validations by (Date) . However, this does not address
product currently on the market, or product that will enter the market
tested with an unvalidated method. Your proposal to verify “key
parameters” for the first API batch produced does not provide the same
level of assurance as method validation.
 Regulatory Audit Warning Letter

Inadequate Instrument Qualification and Analytical Method Validation

Improvements to analytical techniques and transfer of methods to at-

or on-line applications emerged as important opportunities to reduce
risk and increase efficiency in today’s modern manufacturing facility.
A pharmaceutical company was cited for not adequately performing
the required steps to support the transition to a new testing approach.
There was no method comparison or equivalency study performed to
show that the “changes were superior to the original approved
method.
The data was used for OOS closure and lot release.
Four Level Control on Analysis and Results