Chemistry-PTM-OMA_Modules_1-4
Module C1: Selectivity and Calibration Experiments
Calibration Experiment (Quantitative) • The calibration of the candidate method relates the response of an instrument to the concentration of an analyte. The relationship between concentration and response can be linear or non‐linear (curvilinear). • Determine the dose response curve (instrument response versus concentration) using at least six concentrations (including a blank) over the concentration range of interest. Test at least four replicates at each concentration. • Fit calibration curve with an appropriate regression model. Plot residuals vs. concentration and examine for non‐random patterns. Residuals should be at <15% for each concentration but can be at 20% for the lowest nonzero concentration. If a non‐random pattern is observed, use a different calibration model.
Calibration Examples
RU
Biotin Response Curve
Claimed Method Range 30‐150
1430
180
y = 1.0082x ‐ 2.8026 R² = 0.9979
160
1230
140
120
1030
100
830
80
Response
60
630
Method Result (ppm)
40
430
20
0
230
0
20
40
60
80
100
120
140
160
0 10 20 30 40 50 60 70 80 90 100 110
Sodium Sulfite Concentration (ppm)
ng/ml
Concentration
Calibration Residuals
Claimed Method Range 30‐150 ppm Residual Plot
2
1.5
1
0.5
0
Residual
0
20
40
60
80
100
120
140
160
‐0.5
‐1
‐1.5
‐2
Sodium Sulfite Concentration (ppm)
Residuals are all <2%.
Selectivity (Quantitative and Qualitative)
Selectivity is a study designed to demonstrate: • The method’s ability to detect or determine the full range of target analytes • The method’s ability to detect or determine the target analyte(s) without interference from matrix or components with similar properties • The method’s ability to not detect or determine molecules similar in chemical structure or activity to the analyte that are likely to be present in matrices of interest.
Selectivity (Quantitative and Qualitative)
• Develop a list of potential interferents. • Test potential interferents in the presence and absence of the full range of analytes. • Report any observed positive or negative interference. • Positive interference is the enhancement of the analytical response in the presence or absence of the analyte. • Negative interference is the reduction of the analytical response in the presence of the analyte.
Selectivity Example
Histamine Concentration, ppm
Compound, 1000 ppm
Raw Sardine
Raw Mackerel
Raw Anchovy
Cooked Tuna
Canned Tuna in Water
Canned Tuna in Oil
Canned Mackerel in Tomato
Canned Pickled Sardine
Canned Salted Anchovy
Raw Tuna
Water
3 0 0 0 0 1 0 1 1 1 1
4 9 1 6 1 9 0
5 0 0 0 0 0 0
3 4 0 8 4 2 0
0 0 0 0 0 0 1 7 6 0 6
2 0 2 2 4 3 2 4 6 0 8
5 0 0 0 3 5 0 1 1 1 3
7 4 0 4 5 5 0
1 0 1 0 1 0 0
19 11 10 16 13 19 15
3‐Methylhistamine
Tyramine
L‐Phenylalanine L‐Histidine L‐Tyrosine Tryptamine L‐Tryptophan Cadaverine
0
ND
ND
ND
ND
ND
ND
ND 11 17
ND
ND 18 10 13 19 33 48 63 64 65 63 62 65 ND
ND a
7
0
5
3
Putrescine Anserine Carnosine Agmatine
26
15
26
17
4 4
0 1
3 2
2 5
0 2
250
42 46 60 56 57 46 58 50
49 50 49 47 47 46 47 45
42 45 56 52 53 48 52 37
42 48 53 55 54 52 56 48
39 50 51 56 51 55 58 49
50 50 53 48 48 54 59 36
24 45 45 51 54 52 48 40
190
Water
51 50 47 40 49 56 36
50 53 52 54 53 56 51
3‐Methylhistamine
Tyramine
L‐Phenylalanine L‐Histidine L‐Tyrosine Tryptamine L‐Tryptophan Cadaverine
50
ND 49 46 50 53
ND 47 80 49 52
ND 49 54 48 46
ND 55 80 49 52
ND 54 54 52 55
ND 54 55 51 58
ND 50 51 50 57
ND 59 67 48 54
ND 56 66 54 54
65 69 65 65 71
Putrescine Anserine Carnosine Agmatine
>200
>200
145
>200
>200
>200
>200
>200
>200
a ND = Not determined due to insolubility
Module C2: PTM and OMA Matrix Studies
Overview
• The Matrix Study determines the method performance in the claimed matrices. • Matrix Studies can include the use of reference materials, reference standards and/or reference methods. • Reference material (RM) – Material, sufficiently homogeneous and stable with respect to one or more specified properties, which has been established to be fit for its intended use in a measurement process (NIST). • Certified Reference Material (CRM) – RM characterized by a metrologically valid procedure for one or more specified properties, accompanied by a certificate that provide the value of the specified property, its associated uncertainty, and a statement of metrological traceability.
Overview • Standard Reference Material (SRM) – A CRM issued by NIST that also meets additional NIST‐specific certification criteria and is issued with a certificate of analysis that reports the results of its characterizations and provides information regarding the appropriate use(s) of the material. Uses include method development, calibration, or ensuring the integrity of measurement quality assurance programs. • Reference Standard – A traceable standard preparation of analyte with documented identity and purity. • Reference Method – A pre‐existing official regulatory or other recognized analytical method against which the candidate method will be compared.
Overview
• RMs, CRMs, and SRMs that are relevant matrixes can be used as test materials in a validation study. • Reference standards can be used to prepare solutions to fortify matrixes to specified analyte concentrations for use as test materials in a validation study. • Reference methods applicable to the relevant matrixes can be used in a method comparison validation study. • In rare cases, such as new or complex analytes, reference methods, RMs, CRMs, SRMs, and reference standards may not be available. In these cases, alternate strategies should be explored.
Sources of Reference Materials and Reference Standards • National Institute of Standards and Technology (NIST) https://www.nist.gov/srm • Environmental Protection Agency (EPA) https://cfpub.epa.gov/si/si_public_record_Report.cfm?Lab=ORD&dirEntryId=38019 • Canadian National Research Council (NRC) https://nrc.canada.ca/en/certifications‐ evaluations‐standards/certified‐reference‐materials/certified‐reference‐materials • US Pharmacopeia (USP) http://www.usp.org/reference‐standards/reference‐standards‐ catalog • Institute of Reference Materials and Measurements (IRMM) https://ec.europa.eu/jrc/en/science‐update/welcome‐irmm‐reference‐materials‐ catalogue • Chemical Suppliers
Sources of Reference Methods
• AOAC Official Methods of Analysis (OMA) http://www.eoma.aoac.org/subscriber/main.asp • FDA Laboratory Methods https://www.fda.gov/Food/FoodScienceResearch/LaboratoryMethods/default.htm • USDA Food Safety Inspection Service (FSIS) Chemistry Laboratory Guidebook (CLG) https://www.fsis.usda.gov/wps/portal/fsis/topics/science/laboratories‐and‐procedures/guidebooks‐and‐ methods/chemistry‐laboratory‐guidebook/chemistry‐laboratory‐guidebook • USDA Federal Grain Inspection Service (FGIS) https://www.ams.usda.gov/services/fgis/standardization/reference‐methods • CEN/ISO Methods https://www.iso.org/store.html • Health Canada Health Products and Food Branch (HPFB) Compendium of Methods https://www.canada.ca/en/health‐ canada/services/food‐nutrition/research‐programs‐analytical‐methods/analytical‐methods/compendium‐methods.html • American Public Health Association (APHA) Standard Methods for the Examination of Dairy Products https://ajph.aphapublications.org/doi/book/10.2105/9780875530024 • APHA Standard Methods for the Examination of Water and Wastewater https://www.standardmethods.org • US Pharmacopeia (USP)/National Formulary https://www.uspnf.com
Terminology
PTM and OMA SLV Matrix Study Designs
Method Type Study Design
Statistics
Acceptance Criteria Determined by SMEs; suggested recovery and precision criteria in OMA Appendix K 95% CI on dPOD must include 0 if method comparison If no reference method, determine lowest concentration that yields POD = 1.0.
Recovery, bias, LOD, LOQ, s r Regression analysis per matrix
Quantitative All claimed matrixes in MD lab; 1/5 matrixes in IL; co‐fortification if multianalyte method 4‐6 concentrations (incl 0) 5 replicates per concentration Qualitative All claimed matrixes in MD lab; 1/5 matrixes in IL; co‐fortification if multianalyte method ≥6 concentrations (incl 0) – 2 concs yielding fractional positive results* 20‐30 replicates per concentration *Fractional positives are defined as POD = 0.25‐0.75
Intermediate reproducibility Probability of Detection (POD)
OMA Collaborative (MLV) Study Designs
Method Type # Valid Data Sets Study Design per Collaborator
Statistics
Recovery, bias, LOD, LOQ, s r , s R
Quantitative ≥8
≥5 Materials 2 Replicates per material
Qualitative
≥10
1 Matrix 12 Replicates High 12 Replicates Low (fractional pos.*) 12 Replicates Non‐spiked
LPOD s r
with 95% CI with 95% CI
s R
s L If method comparison: dLPOD + CI (CI must include “0”)
*Fractional positives defined as LPOD = 0.25‐0.75
Choosing Matrices
Choose matrices: • Regulatory requirements • Likely to contain the target analyte • Recent public interest (e.g., outbreaks, recalls, known health risks) • Associated with trade disputes • Marketing considerations
Preparation of Test Materials
• Obtain sufficient amount of single lot of matrix for candidate method analyses and reference method analyses (if applicable) • Screen food item for endogenous analyte • Recommend using both the candidate and reference methods • Naturally contaminated or incurred matrices are preferred • If analyte not found, or the contamination level is too low, then fortify the matrix with the target analyte. • Incorporate into matrix using an appropriate technique with appropriate holding conditions • Details are analyte/matrix dependent and will be specified in protocols • Analyte concentrations should cover the intended range of the candidate method
Contamination of Surfaces
• Contaminate surface areas in a random blinded manner with a volume of the appropriate material that allows even distribution across the surface without excessive pooling of liquid • 2.25 mL for 12” x 12” area • 250 µL for 4” x 4” area • 100 µL for 1” x 1” area • Dry overnight (≥16 h) at room temperature – surfaces must be visibly dry • Swipe surface area with sponge or swab according to the method.
Performing the Study – Test Portions
• From each material, select the replicate test portions needed for the Candidate and Reference Methods • Randomize and blind code the test portions for each method • Analyze test portions for each method • In some cases, intermediate reproducibility may be desired • Second analyst, different day, separate equipment
Analyzing Data – Quantitative Methods
• Unblind the data • Plot candidate method results vs. known concentration (or reference method results) • Perform regression analysis • Perform outlier tests (Cochran and Grubbs) to identify significantly outlying data points. Data points should rarely be removed, and only for justifiable cause.
Analyzing Data – Quantitative Method SLV
Calculate for each concentration of each matrix: • Mean – the average result from replicate test portions • Repeatability (s r • Relative Standard Deviation of Repeatability (RSD r ) ‐ s r
) ‐ the standard deviation from replicate test portions
divided by the
mean candidate method result • Bias ‐ the difference between the mean candidate method result and the true value (known fortification level or reference method result) • Recovery – the ratio of the mean candidate method result and the true value (known fortification level or reference method result)
Limit of Detection (LOD) and Limit of Quantitation (LOQ) Limit of Detection (LOD)
• The measured quantity value, obtained by a given measurement procedure, for which the probability of falsely claiming the absence of a component in a material is β, given a probability α of falsely claiming its presence • IUPAC recommends default values for αand β equal to 0.05 (or 95% confidence) Limit of Quantitation (LOQ), also called Limit of Determination • The lowest amount of analyte in a laboratory sample which can be quantitatively determined with a defined confidence (predetermined goals for bias and imprecision are met)
Limit of Detection (LOD) and Limit of Quantitation (LOQ) Estimate LOD and LOQ for each matrix one of several ways: 1. Plot s r vs. mean result from matrix study and perform regression a. Estimate LOD as , where is the mean value of blank replicates, is the intercept, and is the slope b. Estimate LOQ as LOD x 3. 2. Analyze 10 blank test portions and 10 test portions containing a low 0 3.3 1 1.65 b X s m b s m 0 X
concentration of analyte (near the expected LOQ) a. Estimate LOD as the mean blank result + 3 SD low conc b. Estimate LOQ as the mean blank result + 10 SD low conc 3. Analyze 10 blank test portions a. Estimate LOD as the mean blank result + 3 SD b. Estimate LOQ as the mean blank result + 10 SD
Limit of Detection (LOD) and Limit of Quantitation (LOQ) Next, validate the estimated LOQ (LOQ est ): • Fortify matrix at the LOQ est concentration • Analyze 10 replicates • Demonstrate RSD r <20%
Data Table – Quantitative Method SLV Candidate Method Results for Histamine in Frozen Tuna Compared to a Known Spike Value
Test Portion
Spike ppm
Result ppm
Spike ppm
Result ppm
Spike ppm
Result ppm
Spike ppm
Result ppm
Spike ppm
Result ppm
Spike ppm
Result ppm
1 2 3 4 5 6 7 8 9
0 0 0 0 0 0 0 0 0 0
0.0
2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5 2.5
2.1 2.3 2.1 2.4 2.1 2.3 2.2 2.3 2.1 2.3
5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
5.4 5.1 4.8 5.3 5.2 4.9 4.9 5.6 5.0 5.2
10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0
10.5 10.2 11.0 11.5 10.1 12.1 10.8 10.5 11.1 10.8 0.69 9.8
20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0 20.0
24.4 19.7 24.0 22.4 20.2 21.0 22.9 23.6 21.4 23.0 22.3 1.62
50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0 50.0
49.5 57.3 51.2 54.9 58.9 59.4 53.6 55.0 56.3 51.7 54.8 3.31
‐0.1 0.0 0.3 0.1 0.2 0.3 0.2 0.3 ‐0.1
10
Mean 0.12
Mean
2.22 0.11
Mean
5.14 0.25
Mean
Mean
Mean
s r
0.16
s r
s r
s r
s r
s r
RSD r 6.04% Recovery 88.8% Recovery 102.8% Recovery 108% Recovery 111.3% Recovery 109.6% Bias ‐0.28 Bias 0.14 Bias 0.8 Bias 2.3 Bias 4.8 4.95% RSD r 4.86% RSD r 6.39% RSD r 7.26% RSD r
Data Graph – Quantitative Method SLV
Candidate Method Results for Histamine in Frozen Tuna Compared to a Known Spike Value
70
60
50
40
30
y = 1.1016x ‐ 0.1851 R² = 0.9939
20
10
0
Candidate Method Result, ppm
0
10
20
30
40
50
60
-10
Histamine Concentration, ppm
Data Graph – LOD and LOQ est
Determination of LOD and LOQ est
4
y = 0.0604x + 0.0646 R² = 0.99
3
2
1
Standard Deviation
0
0
10
20
30
40
50
60
Mean Concentration, ppm
Calculation of LOD and LOQ est From the graph of s r vs. mean cand :
y = 0.0604x + 0.0646
0.12 3.3 0.0646 1 1.65(0.0604)
s m
3.3 1 1.65
X
LOD = = = 0.37 ppm 0 b
LOQ est
= 3 x LOD = 3 x 0.37 = 1.1 ppm
Analyzing Data – Quantitative Method MLV
Calculate for each material: • Grand Mean – the mean result across all laboratories • Repeatability (s r portions among laboratories • Relative Standard Deviation of Repeatability (RSD r ) ‐ s r
) ‐ the mean standard deviation from duplicate test
divided by the
mean candidate method result • Reproducibility (s R
) ‐ the standard deviation across all laboratories
• Relative Standard Deviation of Reproducibility (RSD R ) – s R the mean candidate method result
divided by
Data Table – Quantitative Method MLV Raw Data
From Feng et al. (2018) J AOAC Int . 101, 1566‐1577
Data Table – Quantitative Method MLV Statistical Summary
From Feng et al. (2018) J AOAC Int . 101, 1566‐1577
Analyzing Data – Qualitative Methods
• Use Probability of Detection (POD) statistics • POD is the proportion of positive analytical outcomes for a qualitative method for a given matrix at a given bacterial level or concentration • POD = x/N • Unblind the data • Analyze each level of each matrix separately • Calculate POD and 95% confidence interval (CI) for each method • POD C for the candidate method results • POD R for the reference method results
Analyzing Data – Qualitative Method SLV
• Method comparison achieved by estimating bias of candidate method • Bias estimated as dPOD, the difference between two POD values • dPOD C = POD C – POD R = bias between candidate and reference method results • Calculate 95% confidence intervals on dPOD values • If the confidence interval on dPOD does not contain zero, then the two POD values are statistically different (bias is significant). • If no reference method, determine lowest level that achieves POD = 1.0. • For antibiotics in milk (FDA CVM program), use probit analysis to determine concentration that yields 90% POD with 95% confidence.
Data Table – Qualitative Method SLV Matrix Pork (wt %) N a x b POD c 95% CI d Beef 0.00 % 20 0 0.00 0.00, 0.16 0.10 % 20 1 0.05 0.01, 0.24 0.50 % 20 19 0.95 0.76, 0.99 0.99 % 20 20 1.00 0.84, 1.00 2.44 % 20 20 1.00 0.84, 1.00 4.76 % 20 20 1.00 0.84, 1.00 Beef e 0.00 % 20 0 0.00 0.00, 0.16 0.10 % 20 0 0.00 0.00, 0.16 0.50 % 20 12 0.60 0.33, 0.78 1.00 % 20 20 1.00 0.84, 1.00 2.50 % 20 20 1.00 0.84, 1.00 5.00 % 20 20 1.00 0.84, 1.00 Bovine Liver 0.00 % 20 0 0.00 0.00, 0.16 0.10 % 20 0 0.00 0.00, 0.16 0.50 % 20 1 0.01 0.01, 0.24 1.00 % 20 19 0.95 0.76, 0.99 2.49 % 20 20 1.00 0.84, 1.00 4.99 % 20 20 1.00 0.84, 1.00
a N = Number of test portions b x = Number of positive test portions c POD = x/N d 95 % CI = 95 % confidence interval e Data from independent laboratory
POD Graphs
Data Table – Qualitative Method SLV Antibiotics in Milk Drug Concentration (µg/kg) N a x b POD c Conc. at 90 % POD with 95 % Conf. (µg/kg) d Amoxicillin 0 60 0 0.00 5.8 2 30 2 0.067 4 30 15 0.50 6 30 30 1.00 7 30 30 1.00 7.5 30 30 1.00 8 30 30 1.00 Penicillin G 0 60 0 0.00 2.9 1 30 1 0.033 2 30 11 0.40 2.5 30 24 0.80 2.75 30 27 0.90 3 30 29 0.967 4 30 30 1.00 5 30 30 1.00
a N = Number of test portions b x = Number of positive test portions c POD = x/N d By probit analysis
POD Graphs
Detection of Amoxicillin in Raw Milk
Detection of Penicillin G in Raw Milk
1.00
1.00
0.80
0.80
0.60
0.60
0.40
0.40
POD
POD
0.20
0.20
0.00
0.00
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
Amoxicillin Concentration, µg/kg
Penicillin G Concentration, µg/kg
Analyzing Data – Qualitative Method MLV
• First, analyze data from each laboratory and each material individually • Determine POD values and 95 % Confidence Intervals • If method comparison, calculate dPOD C statistically significant differences • Combine data across laboratories and calculate LPOD with 95% confidence interval for each material • If method comparison, calculate dLPOD C with 95 % confidence intervals and look for any statistically significant differences • Determine error estimates s r , s L , and s R and report with 95% confidence intervals. • Determine the p ‐value of the t‐test based on the X 2 distribution and compare to 0.10 to determine whether the interlaboratory effect, s L , is statistically significant. This is the homogeneity test of laboratory POD values. with 95 % confidence intervals and look for any
Data Table – Qualitative Method MLV Raw Data
From Lacorn et al. (2016) J AOAC Int., 99, 730‐737
Data Table – Qualitative Method MLV Statistical Summary
From Lacorn et al. (2016) J AOAC Int., 99, 730‐737
POD Graphs – Qualitative Method MLV
From Lacorn et al. (2016) J AOAC Int., 99, 730‐737
Module C3: Robustness Study
Overview • This study evaluates the ability of the method to remain unaffected by small variations in method parameters that might be expected to occur when the method is performed by an end user. • The method developer, in conjunction with the AOAC Project Manager, is expected to make a good faith effort to choose parameters that can be influenced by the end user and are most likely to affect the analytical performance and determine the range of variation that can occur without adversely affecting the analytical results
Overview Parameters are varied and deviations tested by an appropriate experimental design: • Plackett‐Burman design (e.g., 7 parameters, 8 treatments) • Full factorial design (k parameters, 2 k treatments, e.g., 3 parameters, 8 treatments) • Fractional factorial design of choice, 2 k‐1 or 2 k‐2 (k=4‐6 parameters)
Factorial Design
• Full factorial design • For selected parameters, vary conditions above and below nominal test conditions • Test every possible combination of all variables • Requires 2 n de terminations, where n is the number of parameters varied • Fractional factorial design • For selected parameters, vary conditions above and below nominal test conditions • Test a subset of a full factorial design depending upon the fractional design (e.g., ½, ¼) • Study materials prepared in the same manner as the matrix study materials
Plackett‐Burman Design
• W.J. and Steiner, E.H. (1975), Statistical Manual of the Association of Official Analytical Chemists, p 50‐55. • Wernimont, Grant T. (1985), Use or Statistics to Develop and Evaluate Analytical Methods, p 78‐82. • For each parameter, vary conditions above and below nominal test conditions • For example, test 8 combinations of 7 parameters and analyze data by parameter • Study materials prepared in the same manner as the matrix study materials
Examples of Robustness Parameters Parameters that should be varied will be method dependent. Examples of parameters that might be varied are as follows:
• temperature at any temperature critical step (e.g., storage, extraction, incubation, mixing) • times for any time critical step (e.g., storage, extraction, incubation, mixing) • order of reagent addition
• test portion mass/volume • laboratory sample holding time • test portion/solvent ratio • reagent volume(s) • reagent concentration
Plackett‐Burnam Design (8 combinations, 7 variables)
Determination #
Parameter
1 A B C D
2 A B
3 A b C d E
4 A b
5 a B C d e F g
6 a B
7 a b C D
8 a b
A or a B or b C or c D or d E or e F or f G of g
c
c
c
c
D
d e F
d E
D
E F
e
e
E F g
f
f
f
f
G
g
g
G
G
G
Observed Result
s
t
u
v
w
x
y
z
Fractional Factorial Design (4 variables, 2 4‐1 factorial design – 1/2 fraction resulting in 8 determination)
Determination #
Parameter
1 A B C D
2 a b C D
3 a B
4 A b
5 a B C d
6 A b C d
7 A B
8 a b
A or a B or b C or c D or d
c
c
c
c
D
D
d
d
Observed Result
s
t
u
v
w
x
y
z
Fractional Factorial Design (6 variables, 2 4‐2 factorial design ‐ 1/4 fraction resulting in 16 determination) Determination # 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 A a A a A a A a A a A a A a A a B B b b B B b b B B b b B B b b Parameter
A or a B or b C or c D or d E or e F or f
C C C C c c c
c
C C C C c c c c
D D D D D D D D d d d d d d d d
E e e E e E E e E
e e E e E E e
F F f
f
f
f
F F
f
f
F F F F f
f
Observed Result
k l
m n o p q r
s
t
u v w x y z
p
Method Parameter Robustness Examples The greater the reduction in combinations from a full factorial, the wider the variation in method parameters is needed to examine the effect of the parameter variation.
Robustness Test Parameter
Baseline Value (Package Insert Instructions)
Low Value
High Value
Full factorial
4.5 g 4.0 g
5.5 g 5.0 g
Test portion mass
5 g ± 0.5g
Plackett‐Burnam
Full factorial
110 min
130 min
Extraction time
120 min ± 10 min
Plackett‐Burnam
100 min
140 min
Full factorial
10 min
20 min
Time between extraction and instrument reading
15 min ± 5 min
Plackett‐Burnam
7 min
25 min
Robustness
Quantitative Methods • Use 1 test portion per treatment combination • Analyze according to candidate method with appropriate parameters altered • Analyze data by appropriate data analysis: ANOVA, multi‐factor regression, or generalized linear model software
Qualitative Methods • Use 1 test material at analyte concentration that yields fractional positives (2‐4 positives/5 replicates) • Test 5 replicates per treatment combination • Use generalized linear model for data analysis
Conclusions • Method developers should be encouraged that statistically significant findings in this experiment are not indicative of a faulty method, and the discovery of significance is not a roadblock to successful method validation. • Success of this experiment is not conditional on a conclusion of “no significant differences were observed.” • Any findings at this stage may be used to modify method parameters, emphasize areas of caution, or ease or tighten specific parameter ranges.
Module C4: Product Consistency and Stability
Product Consistency (Lot‐to‐Lot Variability)
• Study to confirm that the manufacturing of the test kit and its components is consistent among lots • Testing is performed on 3 unique lots • Each lot must be a uniquely manufactured lot or consist of uniquely prepared reagents, supplies, and kit components • Can be combined with stability testing in a single study; if tested independent of stability testing, age of test kit is not a variable in this study, only variation among lots
Product Stability
• Study to support the shelf life statement and confirm that there is no observable change in performance of the method over the shelf life under normal storage conditions. Three possible study designs are: 1. Testing is performed on lots representing the full span of the shelf life of the kits (newly manufactured, middle of term, near expiration date). • This design allows the stability study to be performed concurrently with the product consistency study when unique lots are tested 2. Testing is performed on 1 lot by testing at least 5 time points over the shelf life of the kit 3. Preliminary stability testing is conducted using accelerated studies • Provides only a rough estimate of shelf‐life • Real time data supporting the entire shelf life of the kit under normal storage conditions must be submitted as soon as available
Product Consistency & Stability: Combined Real‐ Time Study Design for Quantitative Methods
• Three (3) lots of test kits
• One lot near the expiration date • One lot near the middle of the expiration period • One lot recently manufactured. • Conducted using a single material (e.g., matrix, calibration standards, reference standards, reference material) • Five (5) test portions evaluated for each lot. • 5 replicates of the target analyte at a high level • 5 replicates of the target analyte at a low level • 5 replicates of blanks • Test in a randomized blind coded fashion. • Decode results and analyze for effects on bias and repeatability. • Data demonstrating no statistical difference in detection between lots and no significant time slope are required.
Quantitative 3 lots of kits • 5 replicates of target analyte at low level • 5 replicates of target analyte at high level • 5 replicates of blanks Bias & repeatability
Product Consistency & Stability: Combined Real‐ Time Study Design for Qualitative Methods
• Three lots of test kits
• One lot near the expiration date • One lot near the middle of the expiration period • One lot recently manufactured. • Conducted using a single material (e.g., matrix, reference standards, reference material, calibration standards) • Twenty (20) total replicates evaluated for each lot. • 10 replicates of the target analyte at a fractional level (2‐8) • 10 replicates of blanks • Test in a randomized blind coded fashion. • Decode, calculate POD values and confidence intervals and analyze data for variable detection between lots/time points.
Qualitative 3 lots of kits • 10 replicates of target material at low level • 10 replicates of blanks
POD + CI
Independent Product Consistency Study
• For test methods that have multiple components with different lot identification numbers • Lots of the individual components will be interchanged prior to analysis • Testing will be conducted in same manner as combined consistency/stability study
Kit Component 1 Lot 1 Lot 2 Lot 3 Lot 1 Lot 2 Lot 3 Lot 1 Lot 2 Lot 3
Kit Component 2 Lot 1 Lot 1 Lot 1 Lot 2 Lot 2 Lot 2 Lot 3 Lot 3 Lot 3
Kit Component 3 Lot 3 Lot 2 Lot 1 Lot 1 Lot 2 Lot 3 Lot 2 Lot 1 Lot 3
Independent Stability Study
Quantitative
Qualitative
Real time or accelerated
Real time or accelerated
1 lot of kits
1 lot of kits
• 5 replicates of target material at low level • 5 replicates of target material at high level • 5 replicates of blanks 5 time points over the shelf life of the test kit
• 10 replicates of target material at low level • 10 replicates of blanks
5 time points over the shelf life of the test kit
Bias & repeatability
POD + CI
NT3
Accelerated Stability Study Design
Accelerated Stability Study
Claimed Shelf Life
Required Storage Temperature
Storage Temperature
Component
Shelf Life
Time Points
Test Solution 1
25°C
9 mos. (270 d)
55°C
0, 2, 5, 8, 11, 14 d
Test Solution 2
‐20°C
9 mos. (270 d)
25°C
0, 1, 2, 4, 6, 8 d
Test Solution 3
‐20°C
9 mos. (270 d)
25°C
0, 1, 2, 4, 6, 8 d
• Data generated is based on the Arrhenius model • Assuming Ea = 20 kcal ex. 1 year at 5°C ≈ 32 days at 25°C, and 1 year at 25°C ≈ 45 days at 45°C • Test kit is stored at an elevated temperature to age the product more quickly • Number of replicates/time points is consistent with standard stability study
Slide 8 NT3
Does this even apply to chemistry? Maybe it is only micro. I had never heard of it before. Nancy Thiex, 5/21/2019
Parting Thoughts
• Product consistency and stability testing can be conducted independently or combined • Fortified materials, extracts or diluents from the matrix studies may be used for this study • Work with your AOAC Technical Consultant to ensure that the testing performed meets the study design requirements for an AOAC validation
Made with FlippingBook Learn more on our blog