Recent development on Validation Method of
Chemical Analysis for (Phyto)Pharmaceutical Preparations
Gunawan Indrayanto
Plant Biotechnology Research Group,
Faculty of Pharmacy, Airlangga University
Surabaya 60286, Indonesi
a
[email protected]
This presentation is for Education or Training only;
this presentation is not for commercial purposes
Why chemical/Pharmaceutical analysis
is very important?
Chemical and Pharmaceutical Analysis:
•
Pharmaceutical/ Chemical/Food/Cosmetic Industries
•
Diagnostic Laboratories
•
Forensic Laboratories
•
Government Laboratories (BPOM, FDA etc)
•
Government Laboratories (BPOM, FDA etc)
•
Research
How can we have a reliable results of the
chemical analysis?
Areas
of Chemical Analysis
•
Quantification:
–
How much
of substance X, Y, Z etc is in the
sample?
•
Detection:
–
Does the sample
contain
substance X, Y, Z etc?
–
Does the sample
contain
substance X, Y, Z etc?
•
Identification:
–
What is the
identity
of the substance(S) in the
sample?
•
Separation:
–
How can the species of interest
be separated
from
the sample matrix for better isolation,
“
Valid
”
Methods
Calibrated
Software
?
Reference
Validation
method
compounds
Target compound
must be
stable
in the
Routine Analytical Method
IQ,
OQ, PQ
are valid
Calibrated
Instruments
Qualified
Person
must be
stable
in the
selected solvent
Reliable
and
Validated
In order to determine
the “Quality”
of any (Herbal) Drugs,
all chemicals
(API, metabolites, impurities, degradation
products, heavy metals, pesticides, preservatives, sweetener,
toxin) must be evaluated (
qualitatively and quantitatively
).
analysis’s
methods are
Validation method
for specific compound(s):
drugs, metabolites,
drugs, metabolites,
Farmakope Indonesia V
Acceptance criteria of accuracy and
precision are depended
(For compound/marker approach)
Massart’s Golden rules of Validation method
(Cited by Gonzales & Herrador,
Data Elements Required for Analytical Method Validation
Type of Analytical Procedure
Assay Impurity testing Quantitative Limit tests
(For Drugs/Cosmetic/Food analysis)
Accuracy Yes Yes Yes Yes * No * n.a No No
Precision Yes n.a Yes n.a No n.a Yes n.a No n.a
Repeatability n.a Yes n.a Yes n.a No n.a n.a n.a No
Interm. Prec. n.a Yes n.a Yes n.a No n.a n.a n.a No
Specificity Yes Yes Yes Yes Yes Yes * n.a Yes Yes
DL No No No No Yes No * n.a No No
QL No No Yes Yes No No * n.a No No
Linearity Yes Yes Yes Yes No No * n.a No No
Range Yes Yes Yes Yes * No * n.a No No
Step taken for validation Process
Analytical goal
Method selection (Published / new)
Optimization
Stability testing (SST and pre-validation)
Stability testing (SST and pre-validation)
Testing validation parameters
Validation report
Transfer of Method (R&D to QC)
For analysis using (Chromatography)
SST
must be determined
( for standard/ QC samples):
According to USP /39/40 FI V the parameters are:
•Rs (usually > 1.75)
•N
•Tailing Factor ( nicer if ca, 1)
Requirements for doing verification of an official method
•Tailing Factor ( nicer if ca, 1)
•S/N ratio (>5)
•Replicate Injections: see each of monographs or
% RSD (maximal)=
Definition of Markers (2008)
:Not always = maybe “yes”
maybe “no”
Concentrations of
Selected (important) markers could
determine the activities
Pre-Validation: Stability testing
•
Scope:
Usage of autosampler for overnight runs with samples
in solution for hours in laboratory environment raises
concern about the stability of samples.
e.g., degradation by hydrolysis, photolysis, adhesion to
glassware, etc.,
•
Recommendations:
•
Recommendations:
Generate data to support standard and sample solution
stability under normal lab conditions for duration of
test procedure, e.g., 24 hours, 3-5 hours for TLC Profile
Acceptance Criteria
Maximal peak area difference is ± 2 %, compare to the
fresh solution
Reference standards:
originality (USP, BP, SIGMA, MERCK, etc);Purity of the standards
-Using mass Balance Method
Mass balance for primary standard;
(It need (a) calibrated micro/analytical balances)
Smallest minimum weight
Smallest
net weight (w)
according USP: 38 /2015; 39/2016
For analytical balance d = 0.1 or 0.01mg ; micro-balance 0.001 mg,
w will be = 82 mg , 8.2 mg
,
0.82 mg
(if SD = 0 or , < 0.41 d)
Or w = 2000 SD
USP 35/FI V : Tolerance not exceed 0. 1 % of the amount weight;
Selectivity/Specificity
Specificity/Selectivity
USP, ICH, FDA: Specificity
AOAC, IUPAC : Selectivity
Specificity is
the ability to determine
the analyte unambiguously
in the presence of other components, this includes degradation
products, metabolites, matrix component etc.
products, metabolites, matrix component etc.
Selectivity evaluation for pharmaceutical preparations
:
Identity of API should be tested, and its selectivity should be
proved.
Please be careful if the identification method that described
in Pharmacopeia/ Official method is not identical with its
in Pharmacopeia/ Official method is not identical with its
assay method (e.g. Identification: IR, color reactions; assay :
HPLC
)
A B
C R & D/QC Lab. at Pharmaceutical Industry
Nice if
Densitograms (λλλλ = 260 nm) obtained from: (1) solution of standard mometasone furoate, (2) extract from excipients of laboratory-made cream, (3) extract of laboratory-made cream, (4) solution of nipagin, (5) solution of nipasol, (6) extract of commercial ointment-1, (7) extract of commercial lotion, (8) extract of commercial cream-1, (9) extract of commercial ointment-2 and (10) extract of commercial cream-2. Peak identities: (A) mometasone furoate, (B) nipagin and nipasol, (C) unknown.
Wulandari, L, Tan, KS., Indrayanto,G. (2003), J. Liq. Chromatogr. R & T, 26, 109-117
The target peak must be proved regarding its identity and purity
Peaks should be well separated
(F.Melianita, J. Witha, S.Arifin, WF.Karina, G. Indrayanto (2009).
Rs should be not less 1.75
IF Impurities, related compounds, possible Degradation products
were unavailable
Cited from: D. Widiretani, I. Luailia, G. Indrayanto,
If we have no
forced experiments to prove the selectivity.
Identity and purity check
of peaks A and B must be performed
Cited from: D. Widiretani, I. Luailia, G. Indrayanto,
Determination of Xanthorrhizol in the rhizomes of Curcuma xanthorrhiza by GC FID
Xanthorrhizol ?
(F. Melinata & G. Indrayanto (2006), Unpublished work)
Should tried
Purity Check using PDA detector (HPLC) or Densitometer (TLC-Scanner)
M.W. Dong, Modern HPLC
for Practicing Scientist, Wiley, 2006
The purity of the peak can be calculated by
Acetylsalicylic acid
200 nm
N=12804, Tf=1.20
Evaluation of the shape of acetylsalicylic acid
Evaluation of the shape of acetylsalicylic acid
chromatogram using 5 wavelengths
chromatogram using 5 wavelengths
Analysis using LC MS/MS triple Quad
(Mw: 868.04)
M. Distl et al. (2009) Potato Research 52, 39-56
The power of MRM
Why we need do the “
Why we need do the “identity test
identity test” and “
” and “
purity test
purity test” of the peaks for
” of the peaks for each
each samples for
samples for
quantitative analysis ?
quantitative analysis ?
•
If the peak is not the target compound,
the conclusion will be not correct.
•
If “impurity components” are included
in the target peak, the quantitative-results
in the target peak, the quantitative-results
will be wrong
•
By Using LC-MS/MS or GC-MS/MS
the target peak doesn’t need to be pure, it can
contain > 1 compounds which have same target
Linearity
•
The linearity of a method is its ability to provide measurement results that are directly proportional to the concentration of the analyte, or are directly proportional after mathematical transformation.• The linearity is usually documented as the ordinary least squares (OLS) curve, or simply as linear regression curve, of the measured responses (peak area or height) as a function of increasing analyte concentrations
•
Peak area
shall be preferred compared to peak heights for making the calibration curve• The linearity of the detector can be obtained by diluting of the analyte stock solution, whilst the linearity of the analytical method can be determined by making a series of concentration of the analyte from independent
preparations (weighing, spiking)
•
Analytical Methods Committee, Analyst (1988), 113: 1469-1471.
•
W. Horwitz, Referee (1995), December, 2.
•
J. van Loco; M. Elskens; C. Croux; H. Beernaert. Accred. Qual.
Assur (2002), 7: 281-285
‘Correlation coefficient (r) can not be used alone
for proving the linearity (
r close to 1 is not
a reliable indicator of linearity
)
Assur (2002), 7: 281-285
•
G. Indrayanto; M. Yuwono, in:J. Cazes (ed), Encyclopedia of
Chromatography (Supp.), Marcel Dekker, Inc, New York, NY
10016, 2003.
•
P.Araujo, J.Chromatogr.B. (2009) 877, 2224-2234
Evaluation of linearity/calibration curve
•
Relative process standard deviation (Vxo)- Funk’set al. 1992 (< 5 %); “sdv” (CATS from CAMAG) < 5
•Mandel’ test (should be OK)
•Residual test (should be OK)
•ANOVA-linearity testing (should be OK, p < 0.01)
•RSD of the Plot of response factor Vs. concentration (RSD < 2.5 – 5 %)
et al. (Xp must be less than the lowest concentration of
•Xp value- Funk’s et al. (Xp must be less than the lowest concentration of the calibration curve)
• % Y intercept < 2
• “r” value (can not be used alone), even “r” is close to one
•Fisher Ratio (according to Araujo 2009)
•G value, Quality Coefficient (According to D Beer et al., 2007)
Some mathematical equations
for evaluation of linearity (Funk’s et al., 1992)
The importance of the Xp’s Value for testing the linearity
Xp values must be < X1
Lowest Highest
Proving Linearity according to Araujo (2009)
– The calibration spots should be independent of each other,
therefore spots should be applied from different, independently
weighed solutions – not by diluting from one and the same
stock solution.
Journal of Planar Chromatography 23 (2010) 3, 173–179 DOI: 10.1556/JPC.23.2010.3.1 173
0933-4173/$ 20.00 © Akadémiai Kiadó, Budapest
stock solution.
– As a consequence, calibration spots should be applied using
uniform volume of solutions, but different concentrations, to
Working range (WR) of a calibration curve can be defined as:
“the ratio of upper concentration (X
u)
and
lower concentration (X
l)”
In QC of Pharmaceutical Industry, WR = 1.5 – 2
Our recommendation: ± 20 % from target concentration ( with n = 4/5)
If WR > 10, It is recommended to use
“a weighting regression calibration curve”
Requirement of assay using Single point Calibration
• Linear response function OK
• Negligible constant systemic error (CL is not significantly different to zero, or
• Negligible constant systemic error (CL is not significantly different to zero, or Intercept < 2 % from target compound )
• Homogeneity of variance
(J. Ermer & P. Nethercote, Validation in PharmaceuticaAnalysis, Wiley-VCH, 2015).
Confidence range of intercept:
Detection limit (DL), Quantification limit (QL):
•
Impurities
•
Degradation products
•
Cleaning validation
•
Heavy metals
•
Heavy metals
•
Toxic metabolites
•
Pesticides
•
Waste products
•
Bio-analytical methods
Detection Limit & Quantitation Limit
Detection limit (DL) defined as the lowest
concentration of an analyte that can be detected under the analytical condition to be used, but cannot be measured quantitatively.
Quantitation limit (QL) is the lowest concentration that can be determined with acceptable accuracy and precision under the analytical conditions. Generally QL can be estimated as 3 times of DL.
DL and QL for instrumental (chromatographic)
analytical methods can be defined in the term of the signal-noise ratio (2-3/1 for DL and 10/1 for QL)
By constructing a linear regression of relative low concentrations of analyte, accurate value of DL can be calculated, in this case DL = Xp. We
recommend using of 5-10 relatively low
Determination of DL using method of Funk et al (DL = Xa/Xp): QL = 3DL; X2 and X1 (highest and lowest concentrations of standards) should be as low as possible!
Evaluation of Accuracy
Accuracy or trueness of an analytical method is given by the
extend by which the value obtained deviates from the true value.
Firstly accuracy can be determined by analyzing a sample with known
concentration and comparing between the measured and true value.
The second approach is by comparing test results obtained from new method with results from the existing method that known to be accurate.
The third and fourth approaches are based on the percent recovery of known analyte spiked into blank matrices or products. The last
technique is known as standard addition method.
For spiked samples into blank matrices, it is recommended to prepare sample in 5 different concentrations in the level of 80-120% of the target concentration.
For standard addition method, the spiking concentrations are 30-60 % of the label claimed
ll1.workcast.net/10311/.../MelissaHannaBrown_QbD_SepSci.pdf (2014)
Accuracy is combination of “trueness” and “precision”
PRECISION
For general rule, the standard deviation of the method should be lower
than 1/6 of the specification range or RSD value was not more than 2 %.
For method validation purposes, precision is determined by
multiple application of the complete analytical procedure on
one homogenous real sample.
According to ICH, both repeatability and intermediate precision should be
According to ICH, both repeatability and intermediate precision should be
tested
Repeatability defined as precision under the same conditions i.e. same
analyst, equipment, reagents time and column/TLC-Plate.
Intermediate precision is performed by repeatability testing on the
different combinations of analyst, equipment, reagents and time within one
laboratory. It is recommended to do 6-10 measurements on each of
repeatability study
RSD maximum for System Precision
SST replicates injections
% RSD
=
% RSD
=
Ermer. J & Nethercote, P (Eds), Method validation In Pharmaceutical Analysis, Wiely-VCH, 2015
Evaluation of the Precision
by one-way ANOVA (2 levels of precisions: Sr and SR)
If within-condition variance (repeatability) =
S
r2, and
between-condition variance is expressed as
(S
2)(S
B2)Intermediate precision variance is
S
R2= S
r2+ S
B2USP: Accuracy: (1) % recovery, and (2) evaluating linearity of estimated and actual concentrations (= recovery curve)
M. Yuwono & G. Indrayanto (2005), Validation method of Chromatography Methods of Analysis, Profiles of Drugs Substances, Excipients and Related Methodology,Vol. 32, Elsevier Academic Press, San Diego, New York, Boston, London, Sydney, Tokyo, Toronto, pp. 243-258
OK
Recovery CURVE
Line 2: constant systematic error
Line 3: constant and proportional systematic error
Evaluation of Accuracy:
Recovery CURVE (Funk’s
et al., 1992
):
To prove whether systematic errors did not occur, a linear regression of recovery curve of Xf (concentration of the analyte measured by the propose method) against Xc
(nominal concentration of the analyte) should be constructed
Xf = af + bf. Xc
and the confidence range (Cr) of the intercept {VB(af)} and slope {VB(bf)} from the recovery curves were calculated for p = 0.05
recovery curves were calculated for p = 0.05
Acceptance criteria
According to AOAC 2013
Trueness/Accuracy
Range:
97- 103 %
98 ± 2 %
E. Rozet, Ph. Hubert, Presentation University de Liege, Erasme, January 2012, reproduced with permission
USP Medicines Compendium, 2013)
Acceptable value
100 ± 1
https://mc.usp.org/.../10AssessValidation.pdf (discontinue March 30, 2015
For Mean± RSD
Calculation based on Cp and Cpk (“process/method capability”)
according to Kromidas
p =
p=
Cp =
(Kromidas (1999, 2011), Validerung in der Analytik, Wiley-VCH)
Cpk1 =
, Cpk2 =
Select the smaller values between Cpk 1,2
Cp / Cpk > 1.33: Very nice; 1<Cp / Cpk<1.33: Nice; Cp / Cpk <1: bad data
Probability method according USP Medicines Compendium, 2013)
According to Ermer, J & Nethercote, Eds. (2015) in his book Method Validation in Pharmaceutical Analysis:
Ti (tolerance interval = Distribution range) = ± Kf x SD
(Kf = Student t factor, or see Table:
(http://www.bessegato.com.br/UFJF/resources/tolerance_table.pdf)
Result of accuracy (R) = Mean ± Ti
(
tolerance interval
)
R can be calculated using : http://statpages.org/tolintvl.htmlAcceptance criteria for accuracy & precision according to Ermer & Nethercote:
R can be calculated using : http://statpages.org/tolintvl.html
http://statpages.org/tolintvl.html
Example: Specification range : 105-95
%Using Tolerance Table WSU for p (0.95, 0.90) k = 3.712: range 96.402-101.598% Using t table (p=95%), k = 2.571, range : 97.201- 100.79 %
Or
using new method: Accuracy Profile:
(combination of trueness, precision, (robustness) and uncertainty)
Upper Limit
G. Indrayanto In Profile of Drugs Substances,
Prediction interval
R = Mean ± ( ) x SD
Tolerance Interval (TI)
……….. : Upper specification limit (USL) %
………Lower specification Limit(LTL) %
R = Result of Accuracy: Mean ± CI/PI/TI
R (mean ± TI) must be included in the range of (USL-LTL),
if process variation is 2 %, in this case TI max should be (
λ
-2) = 3 %;
Our experiences showed that TI can be replaced by capability method
(Cp/Cpk); more works need to be performed to prove this
OK NOT OK
95 %
-x---x---x---x---x..
90.0 94.7 99.5 104.2 110.0 %
If µ1 = 99.5; 99.5 – µ2 ≤ 4.7 and 99.5 – µ2 ≥ - 4.7 (σ = 4.7); so the value of µ2 must be 94.7 < µ 2 <104.2,
Excel:
Matrix Effect
• Purpose
• To evaluate whether you have a
concentration dependent systematic
error due the matrix
• i.e. ion suppression
• How
• comparison of standard curve with
matrix matched standard curve
• Conclusion
ROBUSTNESS/RUGGEDNESS
Robustness can be defined as a measure
of its capability to remain unaffected by
small, but deliberate variations in
method parameters and provides an
indication of its reliability during normal
indication of its reliability during normal
usage.
Ruggedness of a method is the degree of
reproducibility of test results obtained by
the analysis of same samples under a
Performing an Ideal Robustness Study
•
The selected parameters should be studied
by using
an Experimental design
e.g
.
Plackett-Burmann Design or Fractional
Factorial Design
.
•
The effect(s) are calculated by using a
•
The effect(s) are calculated by using a
Mu
ltivariate analysis program
(e.g.
Analysis of effect)
•
The
most significant parameter (s)
can be
evaluated
•
For herbal drugs’s profile, select 3 marker
G. Indrayanto In Profile of Drugs Substances, Excipients and Related Methodology, Vol 37,
Data from: D. Widiretani, I. Luailia, G. Indrayanto, J. Planar Chromatogr.
Calculation of each effect in Robustness study
Reporting analysis results
for drug ‘assay:
: Mean/Result ± Kf x SD
,
instead of
Mean ± SD (2015, Ermer & Nethercote)
Or:
(1992; Funk et al.)
Mean/Result
± Kf x SD, or
β
<
λ
should be included in the range (upper limit – lower limit, or ±
λ
)
https://dl.dropboxusercontent.com/u/1926781/vma-1.4.3a.jar
Metabolite Profiling/Finger Printing
Market
Metabolomics : What can we do?
Modified from the Presentation of Dr.S.Park (Seoul National Univ.)
- Differentiate groups
- Find markers for the groups
- Predict the unknowns
Disease diagnosis
Plant/microorganism-metabolite profiling
Metabolomics
Principle of Metabolomics (Metabolite Profiling)
Substrate -> Product
Modified from: Schreiber, S., Nature Chemical Biology, 1, 64 (2005)
Fundamental Issue in MVS: Dimension Reduction
Bucketing
0 .0
1st step to PCA model: calculate „bucket density“
m/z
In order to enable a PCA calculation the three dimensional LC-MS dataset (RT, mass, intensity) has to be transform into a two dimensional
In
te
n
si
ty
Bucketing of „Raw Data“ = Rectangular Bucketing
Preparation for Statistics = Transformation of LC-MS Data into Table
LC-MS chromatograms of N samples
Time
Metabolomics needs Statistics
How to deal with many data sets?
In preparation for statistics, we need to convert the 3D LC-MS data into a
matrix:
LC peaks in QC sample remain
QC samples run interspersed over 150 injections of complex coffee metabolomics samples:
QC Injection No.1
QC Injection No.147
~150 Injections
QC Injection No.77
The
compact
does not need a coffee break
…LC peaks in QC sample remain
constant in intensity
QC Samples cluster closely in PCA scores plot
Data acquisition and feature extraction
worked reliably
PCA scores plot
-1.0 -0.5 0.0 0.5 1.0 PC 1 -0.75
-0.50 -0.25 0.00
Application of Metabolite Profiling
for QC Herbal Drugs
WAG: Wild American Ginseng
CAG: Cultivated American Ginseng ASG: Asian Ginseng
Classification work flow
Load model for
classifcation
1
4
Load Analyses to be classified classifcation
Analyses
are processed with bucketing parameters
from model
Classification result
2