AOAC GFA Working Group Documents for Review (November 2022)

AOAC Gluten & Food Allergens Stakeholder Program

Gluten Working Group Documents (Validation Guidance)

November 2022

AOAC INTERNATIONAL 2275 Research Blvd., Suite 300 Rockville, MD, 20850

UNITED STATES dboyd@aoac.org 301.924.7077 x126

Guidelines for Validation of Qualitative Gluten Methods, with Specific Examples for Lateral-Flow and Dipstick Type Devices

Contents 1 Scope ...............................................................................2 2 Applicability .....................................................................2 3 Terms and Definitions ......................................................2 4 Method Validation ...........................................................5 4.1 Method Developer Validation Study.........................5 4.1.1 Scope..................................................................5 4.1.2 Selectivity study .................................................5 4.1.3 Matrix / POD study ............................................6 4.1.4 Data analysis and reporting for selectivity and matrix / POD studies ...................................................6 4.1.5 Intended use statement.....................................7 4.1.6 Robustness study ...............................................7 4.1.7 High-Dose Hook Effect Study .............................7 4.1.8 Method Instructions ..........................................7 4.1.9 Product consistency and stability study.............8 4.2 Independent Lab Validation Study............................8 4.2.1 Scope..................................................................8 4.2.2 Matrix Study.......................................................8 4.2.3 Data Analysis and Reporting ..............................8 4.3 Collaborative Study...................................................8 4.3.1 Scope..................................................................8 4.3.2 Number of Collaborators ...................................9 4.3.3 Matrix / POD study ............................................9 4.3.4 Data analysis and reporting ...............................9 4.3.5 Collaborator comments .....................................9 4.4 Matrix Extension .......................................................9 4.4.1 Matrix Extension for single lab validation studies ....................................................................................9 4.4.2 Matrix Extension for multi-site collaborative studies.........................................................................9

Q UALITATIVE G LUTEN M ETHOD V ALIDATION G UIDELINES PG . 2

Metrological traceability of values assigned to calibrators and control materials ISO 5725-1:1994, Accuracy (trueness and precision) of measurement methods and results—Part I: General principles and definitions USP 31:2008, U.S. Pharmacopeia General Information/ Validation of Alternative Microbiological Methods AOAC Appendix N: ISPAM Guidelines for Validation of Qualitative Binary Chemistry Methods AOAC Appendix F: Guidelines for Standard Method Performance Requirements, Official Methods of Analysis (2016) FDA ORA-LAB 5.4.5 (2020) Volume II —Methods, Method Verification and Validation, Document No IV-02, Version 2, Section 2—Microbiology. 3.1 Analyte Chemical / molecular entity or entities, which may be surrogate(s), measured by the measurement system (see also Measurand ). 3.2 Biological Replicate One sample is extracted with a given number of replicates. Example for three biological replicates: Three separate test portions should be weighed, and each test portion should be tested following the testing method as specified by the method developer. 3.3 Candidate Method The method submitted for validation 3.4 Candidate Method Result The final result of the qualitative analysis for the candidate method. 3.5 Claimed Detection Capability (CDC) An analyte concentration, expressed in mg/kg of gluten (or µ g per surface area), that demonstrates a Probability of Detection (POD) of at least 0.95. This may be claimed as one level that achieves a POD of 0.95 across all validated matrices, or individually per matrix. The claimed detection capability must be verified empirically during method validation. 3.6 Collaborator An intended user who participates in the collaborative study. 3.7 Cross-reactivity A positive response to a sample that does not contain any of the analyte.

1 Scope

The purpose of this document is to provide comprehensive technical guidelines for conducting validation studies for qualitative gluten methods, for example methods submitted for AOAC INTERNATIONAL (AOAC) Performance Tested Methods SM (PTM) status and/or for AOAC Official Methods of Analysis SM (OMA) status. The requirements for method developer (single laboratory) validation studies, independent validation studies, and collaborative validation studies are described. Specific examples are provided for lateral-flow- and dipstick-style devices. These guidelines are intended to be applicable to the validation of candidate qualitative gluten methods, whether proprietary or non-proprietary, including those that may be submitted to AOAC for OMA status or PTM certification. Unforeseen circumstances may necessitate divergence from these guidelines in certain cases, and these must be reviewed by AOAC or another appropriate agency (other than the method developer). The AOAC PTM Program requires a method developer (single-laboratory) validation (SLV), and an independent laboratory study. The AOAC OMA Program requires an SLV (also known as the pre-collaborative study) and a collaborative study to achieve Final Action status. A harmonized PTM-OMA Program can be followed in which PTM certification is sought and, if successful, serves as the SLV phase of the OMA Program. 2 Applicability Where appropriate, definitions have been taken from international standards and the source is cited. Sources of definitions include the following: CODEX STAN 118-1979: Standard for foods for special dietary use for persons intolerant to gluten ISO/IEC Guide 99:2007, International vocabulary of metrology—Basic and general concepts and associated terms (VIM) ISO 3534-2:2006 , Statistics—Vocabulary and symbols—Part 2: Applied statistics ISO 17511:2020, In vitro diagnostic medical devices— Measurement of quantities in biological samples— 3 Terms and Definitions

Q UALITATIVE G LUTEN M ETHOD V ALIDATION G UIDELINES PG . 3

3.18 Matrix Totality of components of a material system except the analyte (ISO 17511). For example, the food, beverage, or environmental surface material to be included in the validation as per the intended use of the method. 3.19 Measurand The quantity intended to be measured (the specification of the measurand should be sufficiently detailed to avoid any ambiguity). See also Analyte . 3.20 Method A procedure that includes sample processing, assay and data interpretation. 3.21 Probability of Detection (POD) The proportion of positive analytical outcomes for a qualitative method for a given matrix at a given gluten level or concentration. POD is concentration dependent. 3.22 Qualitative Method A method of analysis with two possible outcomes. (AOAC Appendix N) 3.23 Repeatability Variation arising when all efforts are made to keep conditions constant by using the same instrument and operator (in the same laboratory) and repeating during a short time period. Expressed as the repeatability standard deviation (SDr); or % repeatability relative standard deviation (%RSDr). 3.24 Reproducibility Variation arising when identical test materials are analyzed in different laboratories by different operators on different instruments. The standard deviation or relative standard deviation calculated from among-laboratory data. Expressed as the reproducibility standard deviation (SDR); or % reproducibility relative standard deviation (%RSDR). (ISO 5725-1:1994) 3.25 Robustness Study A study which tests the capacity of a method to remain unaffected by small but deliberate variations in method parameters and which provides an indication of its reliability during normal usage (USP 31). This includes an analysis of the stability and variation of production lots (for methods sold as kits). 3.26 Sample The batch of matrix from which replicate test portions are removed for analysis. The sample contains analyte, if present, at one homogeneous concentration.

3.8 Cross-reactivity Study The examination of matrices that do not contain the analyte, which are potentially cross-reactive, to determine that they are not detected by the method. 3.9 Data Set Results from a combination of an analytical instrument, device, or equipment and an operator, technician or analyst. Results from a single test site may serve as multiple data sets dependent on nonredundant operators and acceptable separation of effort. 3.10 Enzyme-Linked Immunosorbent Assay (ELISA) An analytical procedure characterized by the recognition and binding of specific antigens by antibodies and signal generation by an enzyme-substrate reaction. 3.11 Fractional Range of POD Validation criterion that is satisfied when a sample yields both positive and negative responses within a set of replicate analyses. 3.12 Gluten A protein fraction fromwheat, rye, barley, oats or their crossbred varieties and derivatives thereof, to which some persons are intolerant and that is insoluble in water and 0.5M NaCl. Throughout this document, the word ‘wheat’ refers to all Triticum species, such as durum wheat, spelt and khorasan wheat. (CODEX STAN 118-1979). 3.13 High-Dose Hook Effect Saturation of binding sites, such that detection is compromised or such that there is no longer a unique relation between the signal and the analyte concentration. Also known as Prozone Effect or Overload Effect. 3.14 Incurred Test Material Prepared from a food matrix into which a gluten source (e.g. flour) has been incorporated prior to subjecting the matrix to a given food processing operation. 3.15 Independent Testing Site A testing site not owned, operated or controlled by the same entity as the method developer. 3.16 Interference Study The examination of matrices expected to be tested with the method, to demonstrate that they do not interfere with detection of the analyte. 3.17 Lateral Flow Device (LFD) An analytical method characterized by use of an immunochromatography platform for the detection of specific analyte.

Q UALITATIVE G LUTEN M ETHOD V ALIDATION G UIDELINES PG . 4

3.27 Selectivity The degree to which the method can detect the analyte in the presence of other substances, matrices, or potentially interfering materials 3.28 Spiked Test Material A food matrix into which a gluten source (e.g. flour) has been incorporated after all relevant food processing operations have been completed. 3.29 Technical Replicate One extracted sample is analyzed more than one time. Example for two technical replicates: A sample is extracted once and the resulting extract is tested on two separate lateral flow devices. 3.30 Test Material A material used for method validation that either contains gluten at a given concentration in the context of a food or environmental matrix or is a blank matrix free of gluten. 3.31 Test Portion The quantity of material taken from the test sample (or, if both are the same, from the laboratory sample) and on which the test or observation is actually carried out. [ISO 6206:1979, 3.2.14] 3.32 Test Sites Sites that simulate where the method is intended to be used, such as a traditional laboratory (e.g. public health or commercial laboratory) or non-laboratory indoor facility (e.g. manufacturing site, restaurant).

Q UALITATIVE G LUTEN M ETHOD V ALIDATION G UIDELINES PG . 5

criteria cannot be claimed, and must be reported in the method instructions. A method claiming wheat must give a positive response to all listed Triticum species, otherwise a claim for the specific Triticum species must be made. For methods claiming any Triticum species, only common wheat ( Triticum aestivum ) should be used in all other studies described in this guidance. Cross Reactivity The matrices identified in Annex A, Table 2, at full, undiluted concentration (with some exceptions as noted), will be prepared and analyzed with the candidate method as it is designed for testing food products. Only if the method’s intended use is limited to environmental surface swabbing should the cross-reactivity study be performed on environmental surfaces. Test one replicate per sample. The method should produce negative results for all samples. In the event that an unclaimed matrix tests positive, it or another example of the same matrix may be retested in six independent test portions (biological replicates), with no positive results, to rule out cross-reactivity. Any cross reactive matrix must be reported to end users as part of the method limitations. All data must be reported, and retests must be explained. Interference The matrices identified in Annex A, Table 2 will be spiked with gluten from each claimed gluten source at 20 mg/kg (or less, depending on the regulatory threshold of the target market/country), or two times the environmental surface detection limit for methods limited to environmental surface testing. Sample preparation is described in Annex B. The prepared samples/surfaces will be analyzed with the candidate method as it is designed. Test one replicate per sample. For each sample, the method should produce a positive result. In the event that the single sample replicate tests negative, it may be retested in 20 biological replicates, with no failures allowed, to rule out interference. Findings that certain matrices interfere with gluten detection should be investigated further, using additional similar matrices, to determine the full scope of interference. Unless the interference can be shown to be limited to very few and specific matrices, and this claimed limitation would be of no concern to the method’s end

4 Method Validation Qualitative binary methods are those that produce one of two possible responses (as defined in AOAC Appendix N). This guidance has been developed for use with candidate methods that are designed to detect gluten. If a candidate method’s intended use is not covered by this document or existing standard method performance requirements (SMPRs), the standing AOAC expert review panel (ERP) for gluten, or other qualified agency, may determine the appropriate cross-reactivity/interference panels, and performance requirements. 4.1 Method Developer Validation Study 4.1.1 Scope The method developer validation study is intended to determine the performance of a method under the controlled conditions of a laboratory. The study is designed to evaluate performance parameters including cross reactivity, interference, POD at regulatory level(s), robustness, between-device variation, lot-to-lot variability and product stability. From a regulatory perspective, gluten has four potential sources – wheat, rye, barley and oat grains – and multiple regulatory levels. Developers must determine which of these sources and levels their method is intended to detect, and perform interference and POD studies for each claimed gluten source. For all studies, the sample test portions must be blind coded and randomized to the analyst, as well as to the reader (the person determining the result) if the method relies on visual determination of the test result. 4.1.2 Selectivity study Breadth The gluten source materials identified in Annex A, Table 1, should be tested at two times the lowest CDC (as long as that is equal to or below 20 mg/kg, otherwise test at 20 mg/kg) in a rice flour matrix. Only if the method’s intended use is limited to environmental surface swabbing should this study be performed on environmental surfaces. Test one replicate per sample. The claimed gluten sources (wheat (all Triticum species), rye, barley and/or oats) should all produce positive results. In the event that the single sample replicate tests negative, it may be retested in 20 biological replicates, with no failures allowed, to rule out an insufficient response to that gluten source. All data must be reported, and retests must be explained. Any gluten sources that do not meet these

Q UALITATIVE G LUTEN M ETHOD V ALIDATION G UIDELINES PG . 6

Pressure/Extruded type of processing, but was able to make incurred samples for all other types of processing, they could not claim the “Cereals (Not Fermented, Hydrolyzed or Fractionated)” category. However, they could make a limited claim for “Raw, Processed, Baked, Fried and Dehydrated Cereals”. Method developers with the ability to produce fermented, hydrolyzed or fractionated matrix samples that were incurred with gluten prior to these processes may make individual claims based on the fermentation organism, hydrolyzing agent or fractionation process. Example claims would be “Soy Tempeh fermented with Rhizopus oligosporus ”, “Modified corn starch hydrolyzed with sodium hydroxide”, or “Wheat starch fractioned with water”. For each gluten source-matrix combination, the developer must analyze a blank (no gluten) sample, one sample per each CDC, one sample at half of the lowest CDC, and one sample at least 1.5 times the highest CDC, prepared as described in Annex B. If the sample at half the lowest CDC produces a POD of 0%, an additional sample must be prepared and analyzed with a concentration midway between the half-lowest CDC sample and lowest CDC sample. If the sample at half the lowest CDC produces a POD of 100%, an additional sample must be prepared and analyzed with a concentration midway between the blank (no gluten) sample and the half-lowest-CDC sample. For environmental swabbing assays and CIP/rinse water validations, the developer will treat each surface and/or CIP/rinse water solution as a separate matrix, and analyze the prepared gluten source-matrix materials as described above. Perform (20) biological replicate tests of each gluten source-matrix-spike level sample. For environmental swabbing, perform (20) replicates of each gluten source matrix-spike level sample on each surface claimed. The samples at and above the CDC must produce a POD of 0.95 or higher. The blank samples must produce a POD of 0.05 or lower. 4.1.4 Data analysis and reporting for selectivity and matrix / POD studies A table or tables should be prepared for the results of the cross-reactivity and interference studies, listing the cross reactivity test results, as well as the test results for each matrix, gluten-source and spike level used in the interference study. For the matrix / POD study, analyze the data for positive or negative responses, and prepare a table listing the matrix, gluten source, gluten level, number of samples tested ( N ), number of positive responses ( x ), POD and 95% lower

users, this type of interference may not meet the requirements for validation. Any interfering matrices must be reported in the method instructions. 4.1.3 Matrix / POD study The purpose of the laboratory matrix study is to measure the probability of detection at the claimed CDC(s), and below and above this level(s), in a controlled laboratory setting for all gluten sources, matrices and surfaces claimed in the method’s intended use statement. A matrix/POD study must be performed in each claimed matrix. In order to ensure that each claimed gluten source is represented, the gluten sources must be rotated across the claimed matrices as shown in Tables 1 and 2. Matrices Claimed Matrix 1 2 3 4

Wheat Rye Barley

Wheat Rye Barley

1

Wheat

Wheat

2 3 4

Rye

Rye

Barley

Barley

Wheat Table 1. Rotation of gluten sources across claimed matrices for methods claiming to detect wheat, rye and barley. This pattern would continue for five matrices and greater. Matrices Claimed Matrix 1 2 3 4 5

Wheat Rye Barley Oats

Wheat Rye Barley

Wheat Rye

1

Wheat Wheat

2 3 4 5

Oats

Barley

Rye

Rye

Oats

Barley

Barley

Oats

Oats

Wheat Table 2. Rotation of gluten sources across claimed matrices for methods claiming to detect wheat, rye, barley and oats. This pattern would continue for six matrices and greater. Alternatively, a matrix/POD study for a matrix category may be performed by testing each claimed gluten source, per the rotation shown in Table 1 or 2, in at least 5 examples from the category, equally distributed across each available type of processing (Annex C). Samples under each type of processing must be incurred. As an example, a method wishing to make a claim for the “Cereals (Not Fermented, Hydrolyzed or Fractionated)” category would need to test one matrix from each of the five provided processing categories, and in each instance, gluten would need to be added to the matrix prior to the described processing step. If a method developer was unable to access suitable equipment for preparing incurred samples in the

Q UALITATIVE G LUTEN M ETHOD V ALIDATION G UIDELINES PG . 7

that have a ‘hook line’ indicating an excessive amount of analyte). In the event of any negative result, additional samples may be tested. Negative results at any level may indicate a hook effect, and any gluten concentration where a hook effect is seen must be reported in the kit instructions. 4.1.8 Method Instructions Following the cross-reactivity, interference, matrix / POD, intended use, and robustness studies, the method developer should finalize the method instructions, taking into account any information learned from the validation. The method instructions should describe the method in detail. For antibody-based/binding methods this would include 1) information about the antibodies (or other equivalent binding agents) used, 2) whether the method targets one or multiple proteins 3) whether the antibodies used are monoclonal or polyclonal, and 4) the protein(s) used to generate the antibodies, and whether those were modified, fractionated or synthesized. Instructions must include specific qualifications or training required to perform the method, as well as links or references to this training. The method instructions should also describe, or provide diagrams/pictures describing, the interaction of the sample and each reagent used, and the method by which they produce a result. Step-by-step instructions for sample preparation and performance of the method are required. Pictorial examples are encouraged. Details should be provided on result interpretation. For example, in kits where color change is a factor, end users should be directed on when to call a result positive based on color intensity. In this area as well, pictorial examples are encouraged. Include details on the reporting of results (positive, negative, < > the CDC, detected/not detected at the CDC, etc.), name of method, the units (wheat protein, gluten, gliadin, etc.), and the method for converting all results to gluten values. The instructions must clearly outline any known limitations of the method, including any sources of gluten that are not detected by the method, any known under- or over sensitivity to specific gluten sources, cross-reactivity or interference from any matrices, and information on specific fractions/proteins targeted. Appropriate storage conditions must be provided. The kit lot number and expiration date should be clearly indicated on product packaging.

confidence limit (LCL) and upper confidence limit (UCL) of the POD, as well as POD curves (see AOAC Appendix H). Gluten concentrations at and above the CDC must produce a POD of 0.95 or greater. Blanks samples must produce a POD of 0.05 or lower. 4.1.5 Intended use statement The method developer must provide a statement of the expected context(s) of use, expected matrices and expected analytical goals of the method. 4.1.6 Robustness study The method developer, in conjunction with the AOAC or other independent validation manager, is expected to make a good faith effort to determine which, and to what magnitude, parameters are most likely to vary in the hands of an end user. Each parameter should be varied both up and down by at least 20%. These parameters should be tested in a factorial or Plackett-Burman design, as described in Annex D. Robustness should be analyzed with a blank and a sample at the lowest CDC for one claimed gluten source, in a single incurred matrix. The matrix chosen should be one of the more challenging matrices (most highly processed) from the Matrix/POD study. Test ten biological replicates for each factorial pattern. POD should be calculated for each factorial pattern, and analyzed as described in Annex D. Data should be recorded in a table that describes each individual parameter variation, the individual results of each of the ten replicates for each factorial pattern, and the POD and POD 95% confidence interval for each factorial pattern. The results of the statistical analysis should be reported in a separate table. Any parameter change that has a significant effect on the POD (p < .05) should be reported in the method instructions. 4.1.7 High-Dose Hook Effect Study To determine whether the method is adversely affected by high concentrations of the analyte, a “hook effect” study must be performed. For each gluten source claimed, perform an extraction of a whole flour (or higher concentrate if commercially available, such as vital wheat gluten) per the kit instructions. Dilute this extract 1:10, 1:100, 1:1000 and 1:10,000 in the kit extraction buffer. Follow the kit instructions to prepare these extracts for testing, and then test each sample. Repeat this process four more times, resulting in five separate biological replicates for the whole gluten source and each dilution. All results are expected to be positive (or potentially invalid, for methods

Q UALITATIVE G LUTEN M ETHOD V ALIDATION G UIDELINES PG . 8

specific test method, which will be provided by the method developer. 4.2.2 Matrix Study An independent laboratory, approved by AOAC or the appropriate agency, will perform a matrix / POD study for each gluten source in at least one incurred matrix per claimed matrix category, or in at least one matrix per every five individual matrices claimed, as shown below:

4.1.9 Product consistency and stability study If the test method is sold as a kit or device prepared in lots or batches, a product consistency and stability study must be performed to ensure that the performance of the product is consistent from lot-to-lot and over time under normal storage conditions for the shelf life of the product. Lot-to-lot consistency and product stability can be measured in the same set of experiments. A minimum of three separate product lots must be evaluated. The product lots should span the shelf life of the kit. For example, if the kit shelf life is 12 months, an approximately 12-month-old kit, six-month-old kit and recently produced kit should be evaluated. For an initial single lab validation, accelerated aging may be used if kits at the end of their shelf life are not available - if this is done, then lot-to-lot stability should still be performed across 3 recent lots. Kits should be aged using increased temperature storage as described in ASTM F1980-16 or CLSI EP25-A. Real time data is needed for validations such as AOAC Official Method applications, and prior to the first AOAC Performance Tested Method renewal. Samples used in the evaluation should be made in any one matrix claimed for the method, or using a stable control material. Samples should consist of a blank, as well as a sample spiked at the lowest CDC of the kit for one claimed gluten source. 20 technical replicates of a single sample (one extract or one pooled extract) should be performed on each kit lot, following the method instructions. Data should be reported in a table listing the test kit lot, age of the lot (real time and accelerated), the individual results for each sample, as well as the POD and 95% POD confidence interval for each set of 20 replicates. The CDC sample must produce a POD of 0.95 or greater, and the blank sample must produce a POD of 0.05 or less, in each kit. Alternatively, method developers may provide internal lot to-lot and stability data for review, provided they meet the criteria of the combined study described above. 4.2 Independent Lab Validation Study 4.2.1 Scope The independent validation study, conducted under PTM guidance, should verify the analytical results obtained in the method developer study in a controlled laboratory setting. Independent labs must receive two types of training prior to performing the study. They shall receive training on the processes for independent studies (e.g. AOAC Method Validation Training Course), followed by training on the

Claimed Matrices Matrices tested by Ind. Lab

1-5

6-10

11-15

16-20

1

2

3

4

Table 3 . Number of matrices to be tested by the independent laboratory, as related to the number of claimed matrices. The matrix/POD study will be performed as described in 4.1.3, including the rotation of claimed gluten sources shown in Tables 1 or 2, depending on the method claims. The collaborating laboratories must analyze at least one environmental surface for every 5 environmental surfaces claimed. The selection of which matrices/surfaces are analyzed should be reflective of the range of difficulty associated with the claimed matrices. 4.2.3 Data Analysis and Reporting For the matrix / POD study, analyze the data for positive or negative responses, and prepare a table listing the matrix, gluten source, gluten level, number of samples tested ( N ), number of positive responses ( x ), POD and 95% lower confidence limit (LCL) and upper confidence limit (UCL) of the POD, as well as POD curves (LaBudde, R.A. (2009) Coverage Accuracy for Binomial Proportion 95% Confidence Intervals for 12 to 100 Replicates, TR297, Least Cost Formulations, Ltd, Virginia Beach,VA, http://www.lcfltd.com/Documents/tr297%20coverage%20 accuracy%20binomial%20proportions.pdf). Gluten concentrations at and above the CDC must produce a POD of 0.95 or greater. Blanks samples must produce a POD of 0.05 or lower.

4.3 Collaborative Study 4.3.1 Scope

The collaborative study evaluates the performance of the method across multiple collaborators. The collaborative study is a requirement of the AOAC OMA program, and is not required for the AOAC PTM program. Collaborators should receive training on the specific test method, which will be provided by the method developer.

Q UALITATIVE G LUTEN M ETHOD V ALIDATION G UIDELINES PG . 9

portions of each sample. All samples must be blinded when sent to the collaborator sites. 4.3.4 Data analysis and reporting For each gluten source-matrix-contamination level sample, calculate the POD for each collaborator. Then for each gluten source-matrix-contamination level, calculate the LPOD (mean POD across all collaborators) and standard deviation of collaborator PODs (sPOD). (Wehling et al 2011) Report the LPOD estimates with 95% confidence intervals, and the POD curves. Calculate repeatability and reproducibility as described in Wehling et al. 2011. Gluten concentrations at and above the CDC must produce a POD of 0.95 or greater. Blanks samples must produce a POD of 0.05 or lower. 4.3.5 Collaborator comments Comments on the candidate method should be encouraged from all collaborators, and any comments should be reported in the collaborative study report. 4.4 Matrix Extension 4.4.1 Matrix Extension for single lab validation studies A single lab matrix/POD study must be performed as described in 4.1.3 . A matrix/POD study must also be completed by an independent laboratory as described under 4.2.2 and reported as described under 4.2.3. 4.4.2 Matrix Extension for multi-site collaborative studies A single lab matrix/POD study must be performed as described in 4.1.3 . A minimum of ten collaborator sites will perform the matrix / POD studies as described under 4.3.3 and reported as described under 4.3.4 .

The method developer shall provide a data submission form for the collaborator sites. It is recommended that a trial run be conducted prior to the initiation of the validation study. The purpose of the trial run is to ensure that logistics, sample handling and data reporting processes are worked out and understood by all of the collaborators. A small sample set (2-4 samples) should be analyzed. The trial run should be conducted under the same conditions as the validation study. A period of time should be allotted for troubleshooting after completion of the trial run, including a discussion with each collaborator to address issues and answer questions. The data from this trial run should not be analyzed or included in the validation report. 4.3.2 Number of Collaborators The minimum number of collaborators is based on AOAC Appendix N guidelines. Presently these call for usable data from a minimum of ten independent testing sites, so it is recommended to use more than ten to ensure at least ten usable data sets. 4.3.3 Matrix / POD study The collaborator sites will perform the matrix / POD studies for each claimed gluten source in at least one of the incurred matrices for each matrix category claimed in the method developer study, following the rotation of claimed gluten sources shown in Tables 1 or 2, depending on the method claims. If the method developer study consisted of only individual matrices, rather than matrix categories, then the collaborator study will test at least one incurred matrix for every 5 matrices tested in the method developer study, as shown below: Table 4 . Number of matrices to be tested by each collaborator site, as related to the number of claimed matrices. The selection of the specific matrices used in the collaborative studies should be reflective of the range of difficulty and matrix category associated with the claimed matrices. Samples to be tested will include, for each matrix, a blank sample, a sample at half the lowest CDC (or the fractional range of the POD, if found), a sample at each CDC, and a sample at least 1.5 times the highest CDC. The collaborator sites will be provided with, and test, a minimum of eight test Claimed Matrices 1-5 6-10 11-15 16-20 Matrices tested by Collaborators 1 2 3 4

ANNEX A SELECTIVITY STUDY

*These may be omitted if they are being used as a gluten source in the validation matrix studies. + Oats that are not comingled with wheat, rye or barley may be difficult to source. Whole oat groats may need to be ground to generate a pure oat flour sample. Table 1. Gluten Sources (materials should be tested at two times the CDC, as long as that is equal to or below 20 ppm, in rice flour) Wheat Flour* ( Triticum aestivum ) Wheat Flour ( Triticum compactum ) Khorasan Wheat Flour ( Triticum turanicum ) Oat Flour* + ( Avena sativa ) Durum Wheat Flour ( Triticum durum ) Spelt Wheat Flour ( Triticum spelta ) RyeFlour* ( Secale cereale ) Einkorn Wheat Flour ( Triticum monococcum ) Triticale Flour (x Triticosecale ) Barley Flour* ( Hordeum vulgare ) Emmer Wheat Flour ( Triticum dicoccon )

Table 2. Commodities for Cross-Reactivity and Interference Studies (materials should be tested as normally purchased/used - any processing should be described (roasting, irradiation, etc.)

Flax seed flour/ meal (

Linum

Almond flour ( Prunus dulcis )

Salmon (

Oncorhynchus spp.)

usitatissimum )

Amaranth flour ( Amaranthus spp.)

Green pea flour ( Pisum sativum )

Sesame flour (

Sesamum indicum )

Guar gum, dilute 1:10 in rice flour ( Cyamopsis tetragonoloba )

Arrowroot ( Maranta arundinacea )

Sorghum flour ( Sorghum bicolor )

Black bean flour ( Phaseolus vulgaris )

Hazelnut flour ( Corylus avellana )

Soya flour (

Glycine max )

Sweet rice flour (

Oryza sativa

Brown rice flour ( Oryza sativa )

Milk powder, cow ( Bos taurus )

glutinosa )

Buckwheat flour ( Fagopyrum esculentum )

Millet flour ( Panicum miliaceum )

Tapioca flour/starch ( Manihot esculenta )

Petroselinum

Chestnut flour ( Castanea sativa ) Parsley flakes (

Tea, Ground (

Camellia sinensis )

crispum )

Coconut flour ( Cocos nucifera ) Ground coffee ( Coffea arabica or Coffea canephora )

Peanuts ( Arachis hypogaea )

Walnuts ( Juglans spp.) White bean flour ( vulgaris var. humilis )

Phaseolus

Pork sausage (

Sus domesticus )

Potato flour/starch (

Solanum

Corn meal ( Zea mays )

White rice flour (

Oryza sativa )

tuberosum )

Dried fruits or raisins ( Vitis vinifera )

Quinoa flour ( Chenopodium quinoa )

Yellow pea flour ( Lathyrus aphaca )

Xanthan gum, dilute 1:10 in rice flour (from Xanthomonas campestris )

Egg powder, chicken ( Gallus gallus domesticus )

Pea protein (

Pisum sativum )

Faba bean flour ( Vicia faba )

**subject to further research, this may be of interest as a gluten-like source Buy from reputable sources and ensure that you are getting the actual material, and that it’s gluten free. This can be done by testing using an appropriate validated method. If you have information on the specific varietal tested, include that information in the validation report, as well as including the part(s) of the material that is tested (skin, flesh, stone, pit, etc.). For a multi-component matrix like pork sausage, provide all ingredients. Table 3. Possible Additional Commodities (materials should be tested as normally purchased/used - any processing should be described (roasting, irradiation, etc.) Rye Grass ( Lolium perenne )** Hemp ( Cannabis sativa ) Romano Bean Flour ( Phaseolus coccineus ) Urad Dal Flour ( Vigna mungo ) Lima Bean Flour ( Phaseolus lunatus ) Spices Garfava Flour (mixture of Garbanzo Flour ( Cicer arietinum ) and Fava Bean Flour ( Vicia faba )) Protein Sources (e.g. Duckweed ( Lemna minor ), insect, algal, fungal), Garbanzo Bean/Gram Flour ( Cicer arietinum ), Chia ( Salvia hispanica ) Lupin Flour ( Lupinus spp. ), Kidney Bean Flour ( Phaseolus vulgaris ), Lentil Flour (Lens culinaris ), Cauliflower (Brassica oleracea var. botrytis ). Teff Flour ( Eragrostis tef) Carob ( Ceratonia siliqua ) Poppy Seeds ( Papaver spp .) Sunflower Kernels ( Helianthus annuus ) Carrageenan (dilute 1:10 in rice flour) Cocoa Beef meat Pork meat Poultry meat Cumin ( Cuminum cyminum ) Turmeric ( Curcuma longa ) Ginger Powder ( Zingiber officinale ) Marjoram ( Origanum majorana ) Cinnamon ( Cinnamomum verum ) Clove ( Syzygium aromaticum ) Paprika ( Capsicum annuum ) Sage ( Salvia officinalis ) Thyme ( Thymus vulgaris ) Coriander seed ( Coriandrum sativum )

ANNEX B PREPARATION OF MATERIALS FOR GLUTEN METHOD VALIDATION

Until such time as a reference materials are available, the gluten source for all prepared samples should be commercial, unbleached whole wheat, whole rye, whole barley or whole oat flour. The chosen flour should be analyzed for Dumas or Kjeldahl nitrogen. Convert to percent crude protein by multiplying the nitrogen value by 5.83. Then convert to percent gluten by multiplying the crude protein value by the following factors, depending on the grain:

• Wheat 0.80 • Rye 0.48 • Barley 0.75 • Oats 0.15

These conversion factors are suggestions, and may vary across different grain samples. The factors come from two publications (1,2); the higher conversion factor for each grain has been chosen to err on the side of consumer safety. Alternately, method developers can use the wet chemical method in Wehling/Scherf (2) to arrive at the gluten content for wheat, rye and barley flours. (1) Schalk K, Lexhaller B, Koehler P, Scherf KA (2017) Isolation and characterization of gluten protein types from wheat, rye, barley and oats for use as reference materials. PLoS ONE 12(2): e0172819. doi:10.1371/journal.pone.0172819 (2) Wehling, P and Scherf, KA (2020) Preparation of Validation Materials for Estimating Gluten Recovery by ELISA According to SMPR 2017.21. Journal of AOAC International 103(1): 210-215. Finally, convert the percent gluten to mg/kg (ppm) gluten by multiplying the result by 10,000. As an example, a barley flour is tested and found to have a Dumas nitrogen level of 1.5%. This is multiplied by 5.83 to attain a crude protein level of 8.75%. Using the conversion factor for barley, the 8.75% crude protein is multiplied by 0.75 to obtain the gluten percent of 6.56%. This percent value is then multiplied by 10,000 to estimate the mg/kg (ppm) value at 65,600. This is equivalent to 65.6 mg of gluten per gram of flour. Making Spiked Materials Bulk spiked materials may be prepared for the selectivity, stability and lot-to-lot studies, and bulk spikes of raw materials are often made prior to the processing steps when making incurred samples. These methods can be used for any material that has a small particle size or uniform consistency, including flours, baking mixes, spices, meats, sauces, dressings, ice cream (melted), etc. They can also be used in other matrices that can be dried and ground to a flour-like consistency, such as nuts, seeds, and bread crumbs. Thorough blending is key to a successful trial. For dry materials like flours, or for liquid consistencies, blending can be done in a blender or tumbler-style mixer, or even by manual tumbling of material in a zippered plastic bag. Add the spike material uniformly within the matrix, rather than adding it all in one location prior to blending it in. Making spikes in very fine matrices with small particle size can be difficult, and re-milling of the matrix and spike may be necessary to achieve particle size homogeneity. While gluten is not water soluble, it can be uniformly dispersed in sauces, dressings and other liquids by either spiking

directly with flour, or making a suspension of gluten in the matrix, mixing it thoroughly to achieve uniformity, and using this to make the spikes. Make sure to mix the material again before any samples are taken from it. For paste-like items and meats, spread the matrix out on aluminum foil, parchment or other non-stick surface, sprinkle the spike material uniformly across the top, and then recombine the matrix and mix by kneading. Extremely high-speed or high-heat mixing can alter the gluten results, so mechanical blending should be done in short pulses, and only for the duration needed to achieve sufficient uniformity. Liquid suspensions made in the kit extraction buffer can be used to spike individual test portions for the interference portion of the selectivity study prior to extraction. Liquid spiking of test portions may not be used for the matrix or other studies. If this method is used, state in the validation report that the selectivity study only tests for analytical interference, not interference with the extraction. Options for adding gluten to the matrix, either as a spike or prior to processing of an incurred matrix, include (see Figure 1): 1. Creation of a mid or high-level stock followed by serial dilution. The gluten concentration in the stock should be chosen to allow the largest volume of stock material to be used in the preparation of each spike level. 2. Creation of mid or high-level stock used to then make each individual bulk preparation. 3. Creating bulk spike level samples directly from the source material (flour). 4. A combination of the above, in which spikes are made directly from the flour source for higher levels, then diluted to achieve lower levels) The method for creation of each sample must be described in the report. Any suitable validated quantitative method can be used to assess sample homogeneity. Assessing homogeneity of the high or mid level stock can be a good initial step before preparing lower level spikes. Homogeneity should be assessed for every bulk test material, or at least as many as needed to confirm that the mixing procedure is adequate to minimize distributional variance. Homogeneity should be assessed by testing 10 test portions, taken from throughout the material, individually extracted and run according to the method instructions of any validated quantitative assay (e.g., use 2 wells if the method calls for it). The preferred CV from the homogeneity data will depend on the method performance requirements, with the homogeneity SD below the required repeatability SD. Higher CVs may be expected at lower analyte concentrations. Use the stocks for testing on the same day if possible. Samples made in dry matrices, like flours, can be stored at room temperature for several days, remixing each stock thoroughly before use. Samples made in perishable matrices (dairy products, meats) should be refrigerated for no more than 2 days, remixing each stock thoroughly before use. Samples may also be stored frozen in working aliquot sized portions for an extended period. Making Incurred Materials The section above, Making Spiked Materials, describes the initial steps in making an incurred material. The spiking must occur prior to the major processing step in order for the end product to be considered an incurred matrix. Further considerations for common types of processing are provided below

Baked, Fried or Dehydrated Materials Baking, frying and dehydrating are processing methods that can be reasonably replicated at a small scale, in a laboratory. The same process applies for each. When possible, weigh the incurred material before and after processing. Any change in the analyte concentration above or below the expected value should be accounted for by the change in mass. When exact ppm values are needed, for example for a quantitative method, the moisture/weight change from processing must be accounted for in determining the amount of spike material to be added. If the entirety of the material cannot be weighed before and after processing, additional analyses can be performed to determine the potential analyte gain or loss, such as moisture content, protein, or zinc/other metals. If moisture/weight change results in a slightly higher ppm value than intended, higher level incurred samples can be mixed with blank, processed sample to achieve various concentrations. The lowest concentration achieved in this way should not be less than 10% of the concentration of the high-level incurred material. Larger discrepancies require a second incurred matrix to be made at a lower level. Pressure Treated/High Heat/Extruded These are processes that cannot normally be replicated outside of a manufacturing facility. If a manufacturer is particularly interested in the development of the assay, the kit developer may be able to partner with them to make gluten spikes on a pilot scale, using a similar method as described above for baked, fried and dehydrated products. In the absence of access to a manufacturing plant, some highly processed matrices can be incurred through “fortification”. An example would be a whole wheat puffed/extruded breakfast cereal. A pilot plant could create a mid-level spike (100 ppm, for example), which could be diluted down in a similarly processed blank matrix to create lower concentrations. Any validated method can be used to verify the absence of gluten in the non-gluten-containing matrix. Making Environmental Surface Samples Determine the surface area that’s expected to be swabbed. Typical area is 25 cm 2 – 100 cm 2 (approx. 4 in. 2 - 16 in. 2 ) Make suspensions from the flour in the kit extraction solution, or 60% ethanol solution. Create solutions at gluten concentrations ( µ g/ml) around the expected sensitivity level of the method, as described in the validation requirements. Pipette gluten suspension per outlined area, distributing the liquid as evenly as possible. Shake the suspension thoroughly before pipetting it into each square. Note the volume of solution added to each area, to allow the total µ g of gluten per swab area to be calculated. If the method is for swabbing of wet areas, the surfaces are ready for testing. If the method is meant to test dried-on material, allow the gluten suspension to dry completely (overnight if necessary). Cleaning solution studies for an environmental surface claim are voluntary. Cleaning solution studies are to be performed as described in the following section.

Making Rinse Water/CIP/Cleaning Solution Study Samples Make a high level suspension of gluten in kit extraction buffer or 60% ethanol, then dilute into water or water/cleaning solution to the desired gluten concentrations, around the expected sensitivity level of the method, as described in the validation requirements. The high level gluten solution in kit extraction buffer or ethanol should not make up more than 1% of the final CIP (cleaning solution)/rinse water preparation, to ensure that the sample is representative of a typical CIP (cleaning solution)/rinse water sample. If the method is designed for rinse water testing, and the cleaning solution will not be tested at its recommended working concentration, the dilution of the cleaning solution must be reported. Cleaning solution (CIP) validations must be performed separately for each cleaning agent. Method developers may choose to perform a validation in examples from each of the four main types of cleaning solutions: degreasers, detergents, abrasives, and acids. But the validation will only be reported for the specific cleaning agent that is used.

Figure 1. Options for generating bulk spike materials at various levels

Appendix C - Matrix Categories

NOT FERMENTED, HYDROLYZED OR FRACTIONATED

*

PROCESSED / BAKED / CURED / SMOKED / MARINATED

PRESSURE/HEAT-UHT/ PASTEURIZATION/ EXTRUSION

DEHYDRATED / DRIED / DRY CURED

RAW/ MINIMALLY PROCESSED RAW/ MINIMALLY PROCESSED

FOOD CATEGORY

FERMENTED

HYDROLYZED

FRACTIONATED

FRIED

FERMENTED

HYDROLYZED

FRACTIONATED

Binders, stabilizers, emulsifiers

Carrageenan, Xanthan gum, Guar gum

PROCESSED

Caramel, pralines, marzipan, nougats,

Pastilles, Lozenges, jelly beans, toffees, licorice, Chewing gum, Mints, Icing or Frosting (non chocolate), sauces used for toppings, non chocolate (butterscotch, marshmallow)

Candy

RAW/ MINIMALLY PROCESSED

PROCESSED, BAKED

FRIED

PRESSURE/EXTRUSION

DEHYDRATED

FERMENTED

HYDROLYZED

FRACTIONATED

Soy Milk, Oat milk, Rice milk, Teff milk (if hydrolyzed as part of processing); Modified food starch.

Bread, cakes, cookies, tortillas, fresh pasta, bakery products, confectionaries, crackers, bagels, muffins, grain-based protein bars

Sourdough, Malt, Malt extract, Sprouted flours, Soy, oat or rice milk yogurts, natto, tempeh, soy sauce, miso

Breaders / Batters for fish sticks and chicken nuggets, tortilla chips, donuts

Unmodified Wheat starch, soy protein isolate, tofu, Maltodextrin, Soy Lecithin

Whole or milled Sorghum, soybeans, corn, millet, teff, rice, fonio, oats; baking mixes

Breakfast cereals, Puffs /pellets

Bread crumbs, Dried Pasta

Cereal Grains

RAW/ MINIMALLY PROCESSED

Chemicals and Preservatives

Need to be validated per matrix

Pg. 1