Acrylamide ERP Resubmission Review

EXPERT REVIEW PANEL FOR ACRYLAMIDE METHODS Wednesday, May 3, 2023 Meeting Time: 10:00AM – 12:30PM (Eastern US) MEETING MINUTES

In Attendance:

Expert Review Panel Members:

Xi Feng 1 , San Jose State University Jason Beasley, Reading Scientific Services LTD (Mondelez) Abhijit Bhar, IADFAC Aurélien Desmarchelier, Société des Produits Nestlé Ashley Griffin, Nestle Ajai Prakash Gupta, FSSAI Lukas Vaclavik, Eurofins Food Chemistry Testing

1 ERP Chair

Method Authors:

Aurélien Desmarchelier Aude Bebius Erik Konings

AOAC Staff:

Deborah McKenzie Allison Baker

Tien Milor Delia Boyd Saliha Argubie

Other Attendees:

Karen Mandy

Simone Carron Markus Lacorn

Giampaolo Perinello

I.

Welcome, Introductions, and ERP Overview

Allison Baker (AOAC INTERNATIONAL) welcomed attendees and presented housekeeping information to the participants of the ERP Meeting. The quorum was confirmed after a rollcall of ERP panel members. Xi Feng (ERP Chair) called the meeting to order and welcomed attendees. He then gave a brief overview of the agenda and the method to be reviewed for First Action - ACRYL-01: Acrylamide in coffee, cereals, baby food, cocoa, petfood, potato products, vegetable crisps, biscuit, tea, nuts and spices by LC-MS/MS. Deborah McKenzie (AOAC INTERNATIONAL) provided an overview of the meeting, ERP process, and expectations of the Expert Review Panel.

II. Submission of Method for First Action: ACRYL-01: Acrylamide in coffee, cereals, baby food, cocoa, petfood, potato products, vegetable crisps, biscuit, tea, nuts and spices by LC-MS/MS. 1. Abhijit Bhar, Lukas Vaclavik, Ashley Griffin, and Xi Feng gave their reviews of the method, ACRYL-01: Acrylamide in coffee, cereals, baby food, cocoa, petfood, potato products, vegetable crisps, biscuit, tea, nuts and spices by LC-MS/MS in response to the AOAC Call for Methods for SMPR 2022.006. They each gave a brief summary of the candidate method, which used quantitative LC-MS/MS to determine acrylamide in multiple food matrices (i.e., potato-based products, baby food, cereal and bakery products, cocoa products, coffee, tea, herbs and spices, petfood, and nuts). The proposed method is derived from EN 16618:2015 with modifications to the sample preparation, scope of application, and LC conditions. Each reviewer gave a summary of the method to the other members of the Expert Review Panel and provided their reviews of the method, providing information in support of the method and general application package. Each reviewer then stated their recommendation for the method to the group. Reviewers Bhar, Griffin, Vaclavik, and Feng recommended that the method be adopted as a First Action. 2. Motion : a. To move candidate method, ACRYL-01: Acrylamide in coffee, cereals, baby food, cocoa, petfood, potato products, vegetable crisps, biscuit, tea, nuts and spices by LC- MS/MS , to OMA First Action Status as is. i. (Vaclavik moved and Bhar seconded; Consensus: 5 in favor, 2 abstain. Motion Passes.

b. To accept these requirements (below) for Final Action Status.

i. (Griffin moved, Vaclavik seconded; Consensus: 6 in favor; 1 abstain. Motion Passes. ii. Recommendations to the Method Author for Final Action Status: 1. Provide tables to support linearity 2. Include data to verify LOQ and recovery estimation 3. Revise typo in chromatogram (pet food) 4. Include unspiked chromatograms closer to LOQ 5. Include more detail in RPD values and spike recoveries (i.e., were averages used?)

6. Generate reproducibility data on in-house materials 7. Account for dry pet food vs high moisture pet food 8. Determine shelf life of solution 9. Include equilibration time in operating procedure

10. Explain reasoning for the use of internal standards 11. Include method user feedback

III.

Adjournment (Pham)

Andrew Pham adjourned the meeting at 11:16am ET.

AOAC Official Me thods of Analys is (OMA) ACRYL-01: Acrylamide in coffee, cerea ls , baby food, cocoa , petfood, pota to products , vegetable cris ps , bis cuit, tea , nuts and s pices by LC-MS/MS.Firs t Action 2022.xx.

Request from Expert Review Panel on Acrylamide (03.05.2023)

1. Provide Tables to support linearity We provide 2 types of documents explaining how the linearity was evaluated:

(1) Extract from our in-house validation file (see Document 1 attached below) including the part dedicated to linearity. Linearity was assessed by 7-point calibration curves covering a 1-100 ng/mL range, each datapoint being analyzed in triplicate. Plots of the regression lines shown in Document 1 demonstrated linearity over the 1-100 ng/mL range whatever the internal standard or the LC-MS/MS instrument applied. Document 1: Extract of in-house validation file – Part related to linearity:

Linearity_ValidationFi le.docx

(2) Two result sheets (n=2) collected during the validation for green tea and black pepper. In our process, raw data are exported from the supplier software, and then reformatted by means of an excel macro which also enable automated calculations. In each batch of analysis, a calibration curve is systematically included for concentration calculation and linearity assessment (see Document 2 ). Documents 2: Result files from validation (Recovery and Repeatability) in Tea and Pepper

Pepper.xlsx

Tea.xlsx

Note: The submitted manuscript contains a part describing how linearity assays were conducted ( part H) and which criteria were required to demonstrate linearity in ( part G.d ). No changes were applied in the manuscript.

2.

Data to verify LOQ and recovery estimation (a) LOQ

LOQ values were fully re-evaluated as defined in SMPR 2022.06 i.e., by signal-to-noise (S/N) extrapolation in non-fortified samples. Six chromatograms were considered for each sample, and the S/N values were retrieved from the supplier software (Sciex OS). Using the sample content or the assigned value in µg/kg, the concentration corresponding to S/N = 10 was determined. To establish the LOQ with a 99.7% confidence interval, the estimated LOQ was calculated as the concentration corresponding to S/N=10 + 3 STDEV. Results are now displayed as such in the (modified) Table 4.

Level / Assigned value (µg/kg)

Benchmark levels (µg/kg)

Estimated LoQ (µg/kg)

Matrix category

Sample

Potato Crisps FAPAS 30109 French Fries FAPAS TYG085RM Vegetable Crisps FAPAS 30101

718 71 1105

750 500 - 150 40 40 350

3 1 2 1 1

Potato based products

Infant Food FAPAS 3092 (infant biscuits) Infant Food FAPAS 30102

164 196 23 205

Baby food

Infant cereal (in-house reference) Biscuit (cookie) FAPAS 30104

0.4 0.5

Other cereal & bakery products

Cocoa products

Cocoa powder

88

-

3

Roasted coffee (in-house reference) Coffee (Instant) (in-house reference)

338 738 613 561

400 850 850 850

1 3 4 -

Coffee

Coffee (Instant) FAPAS 3087 Coffee (Instant) FAPAS 30107

Dry pet food

Pet food croquettes

13

-

0.2

Tea

Green tea

33

-

1

Herbs and spices

Black pepper

671

-

2

Nuts

Roasted almonds

413

-

1

LOQ determined according to SMPR 2022.06 were extremely optimistic from our point of view. This was obviously due to the absence of blank samples and the impossibility to spike at low levels. To discuss objectively the reported LOQ, we have added a paragraph in the manuscript as follows: In the absence of blank or very low contaminated materials, LOQ values were assessed by extrapolating the concentration in matrix samples (non-fortified) corresponding to a S/N = 10. With this procedure, LOQs fall within a 0.2- 4 µg/kg range, in agreement with the requirements set at ≤ 20 µg/kg for baby food and ≤ 50 µg/kg for all other matrix categories (Table 4). Such extrapolated LOQs were considered overly optimistic. However, approximating LOQ values at the lowest possible level can be of poor significance for the food categories containing substantial amounts of AA levels, typically > 100 µg/kg. More importantly, we felt imperative to demonstrate accurate and precise measurements in non-fortified samples, ideally with levels of AA at or below current EU benchmark levels (see Table 4).

(b) Recovery estimation

 For clarity purpose, the part H.c Performance characteristics\Validation was revised. Samples were divided in three categories (Non-blank samples, In-house reference materials and Quality control and reference materials). For each category, we reformulated the validation scheme and added the calculations applied for recovery determination: (Corrected) Recovery for non-blank samples spiked (%) = − � Where is the mass fraction (in µg/kg) for non-blank sample spiked at the level of interest; � is the average mass fraction (in µg/kg) for non-blank samples measured ‘as is’ (n=2); is the fortification level (in µg/kg). and Recovery for reference sample (% ) = Where and represent the experimental mass fraction and the reference mass fraction expressed in µg/kg, respectively.  Typical examples of recovery calculation in green tea and black pepper are provided ( cf Document 2 attached in this document). In each experiment, excel calculations are organized similarly, including linearity assessment, setting of identification criteria and mass fraction calculations. Calculated amounts were ultimately compared against the fortification levels or assigned values for recovery determination. For exhaustivity purpose, all data collected during validation have been gathered in Document 3 : Documents 3: Summary of all validation data Same formula than above is applied for Quality Control and Reference Material.

Validation_All data.xlsx

3. Typo in chromatogram (pet food) Done. Title of Figure 3 has been modified accordingly:

Figure 3. Chromatographic profile of the transition at m/z 72→ 55 in (1) FAPAS TYG085RM (French fries, 71 µg/kg), (2) FAPAS 30104 (Cookie, 205 µg/kg), (3) FAPAS 30101 (Vegetable crisps, 1105 µg/kg), (4) FAPAS 3092 (Infant food, 164 µg/kg), (5) FAPAS 30102 (Infant food, 196 µg/kg), (6) FAPAS 3087 (Instant coffee, 613 µg/kg), (7) In-house reference material (instant coffee, 736 µg/kg), (8) In-house reference material (roast coffee, 336 µg/kg), (9) In-house reference material (cereal, 23 µg/kg), (10) Cocoa (88 µg/kg), (11) FAPAS 30109 (Potato crisps, 718 µg/kg), (12) Green tea (34 µg/kg), (13) Black pepper (671 µg/kg) and (14) Roasted almonds (413 µg/kg) and (15) Petfood (dry) (113 µg/kg). Unspiked chromatogram closer to LOQ The extracted ion chromatograms (XIC) displayed in the Figure 3 of the manuscript were those obtained for 15 out of 16 matrices included in the presented validation. • The XIC of 1 sample (FAPAS 30107) was omitted in Figure 3 since two other instant coffee samples were already displayed (in-house sample and FAPAS 3087). • 14 out of 15 XIC displayed in Figure 3 were from unspiked samples. These samples analyzed ‘as such’ represent thus the lowest possible MS signals achieved during validation. • Only 1 sample (Petfood dry) was represented at a 113 µg/kg concentration level (i.e., 13 µg/kg basal content + 100 µg/kg fortification level) for the following reason: o After contacting Nestlé Purina (requestor of adding dry petfood in the SMPR), levels of interest for acrylamide in dry petfood are within a 100- 500 µg/kg range. o Displaying the chromatogram as low as 13 µg/kg in a petfood sample would be of rather poor interest from their perspective. o For transparency purpose, XIC in the dry petfood sample ‘as is’ (13 µg/kg) and fortified at 100 µg/kg for AA are displayed below. Unspiked sample Basal: 13 µg/kg RSDr/RSDiR: 4% - 7% Sample spiked at 100 µg/kg Level: 100+13 = 113 µg/kg RSDr/RSDiR: 3% - NA 4.

We are open to modify Figure 3 with the unspiked dry petfood sample at 13 µg/kg. Very consistent results were achieved at this level (Precision < 7 %). Nevertheless, we think it could bring unnecessary debates on additional interferences at trace level (To be discussed in a 2 nd ERP meeting?

5.

Detail in RPD values, spike recoveries (averages?)

‘RPD’ Targeted performances in terms of precision were derived from the (modified) Horwitz equation as stipulated in SMPR 2022.06 and in Commission Regulation 2017/2158. To make it clearer, Table 5a was more detailed for the precision parameter. Numeri cal values for RSD R and RSD r are directly indicated in the Table, taking examples at 20- 50 -200- 300 and 500 µg/kg concentration levels. Table 5a is now: (a) Target performance characteristics according to SMPR 2022.06 Parameter Criterion (%) LOQ ≤ 20 µg/kg (baby foods and bread) and ≤ 50 µg/kg for all other matrix category Recovery 75 -110 % Precision: For concentrations C < 120 µg/kg, RSD R = 22 % (Modified Horwitz equation) e.g., 20 µg/kg: RSD R / RSR r = 22 / 15 % 50 µg/kg: RSD R / RSR r = 22 / 15 % a Precision criteria were calculated according to (modified) Horwitz equation stipulated in [4,13]. Additionally, the text in Result & Discussion was slightly modified to discuss the precision criteria based on the Horwitz equation. Text is now: “While for concentrations C < 120 µg/kg, RSD R is set at 22 % (modified Horwitz equation), for concentrations 120 µg/kg < C < 138 mg/g, RSD R is predicted at 2C – 0.15 (Horwitz equation). Corresponding repeatability is expressed as 0.66 x RSD R. As shown in Table 5b and 5c , all precision values were aligned with the modified Horwitz equation.” ‘Recovery’ For better clarity, the part H.c Performance characteristics\Validation was deeply revised. Samples were defined in three categories i.e., ‘Non-blank samples’, ‘In-house reference materials’ and ‘Quality control and reference materials’. For each category, we explained calculations for recovery determination. Specifically, recovery calculation in non-blank samples fortified at 100 and 300 µg/kg, was modified as follows: ‘ Non- blank samples’ (n=5) of cocoa, pet food, green tea, roasted almonds and black pepper were first analyzed ‘as is’, with duplicate analyses performed under 6 different conditions over 3 days, resulting in a total of 12 measurements. The objective was to assess the method precision at the basal level i.e., both repeatability (RSDr, in %) and intermediate reproducibility (RSDiR, in %) parameters. In distinct experiments, recovery (REC, in %) was determined by spiking six replicates of non-blank samples at 100 and 300 μg/kg levels. To control for the native content, duplicate measurements of the sample ‘as is’ were conducted in these assays. The average of these measurements was then subtracted from the measured mass fraction using the following formula: (Corrected) Recovery for non-blank samples spiked (%) = − � For concentrations 120 µg/kg < C < 138 mg/g RSD R = 2C - 0.15 (Horwitz equation) RSD r - 0.66 times RSD R e.g., 200 µg/kg: RSD R / RSR r = 20 / 13 % 300 µg/kg: RSD R / RSR r = 19 / 13 % 500 µg/kg: RSD R / RSR r = 18 / 12 %

6. Generate reproducibility data on in-house reference materials Three samples were analyzed by different Nestlé laboratories with the aim to define consensus values. For each sample, 4 to 5 laboratories were involved in the determination of the consensus value, each performing duplicate analysis over 6 days. It should be noted that this study was conducted in January 2020, and the analytical conditions used were slightly different from those mentioned in the proposed manuscript. Since then, modifications have been made to the mobile phase composition, column length, and gradient run. Therefore, generating reproducibility data could be questioned by the ERP. For transparency purpose, we display the results of the internal study in the tables below. For the reasons mentioned above, no changes were inserted in the manuscript (To be discussed in a 2 nd ERPmeeting?) Pure soluble coffee – Median value: 738 µg/kg

Lab 1

Lab 2

Lab 3

Lab 4

682 679 689 722 687 730

660 682 693 742 711 734

724 766 690 776 673 719

735 757 686 800 659 681

801 759 820 768 835 644

850 752 794 752 798 628

705 772 775 754 759 769

706 769 758 748 728 749

Roast and ground coffee – Median value: 338 µg/kg Lab 1 Lab 2

Lab 3

Lab 4

317 322 327 351 328 362

326 317 325 343 337 343

326 337 355 376 339 332

328 349 348 351 359 338

350 347 338 339 307 322

350 326 315 337 316 340

368 324 291 361 353 326

387 331 285 365 354 326

Infant cereals – Median value: 23 µg/kg

Lab 1

Lab 2

Lab 3

Lab 4

Lab 5

22 29 22 26 26 26

22 28 22 28 28 26

29 23 22 22 23 23

23 22 22 22 24 23

19 23 24 24 19 22

19 19 23 21 18 22

23 23 23 24 24 22

23 23 24 23 24 23

20 23 23 22 20 22

20 21 23 22 20 21

7. Dry petfood vs high moisture pet food Done. As the SMPR stipulated dry petfood in the scope of SMPR 2022.06, we only reported validation data for a dry petfood sample. Dry petfood is now systematically employed for consistency purpose across the manuscript. Therefore, storage at room temperature is appropriate. 8. Determine shelf life of solution Stability was assessed for the acrylamide stock standard solution at 1 mg/mL in water (part E.a) and for the calibration solutions (part E.g). • Stability of the 1 mg/mL stock standard solution in water was tested for 6-month at – 20 °C. Two solutions were independently prepared at a 6-month interval. The same day, both solutions were diluted appropriately and analyzed in the same sequence of analysis. Average of peak area measured in stored solutions (n=5) shall not deviate by more than 15 % compared to average of peak area obtained from freshly prepared solution (n=5) .  Average of peak area in stored solutions matched at 99% compared to freshly prepared ones. • The stability of calibration solutions was also tested for a 6-months duration at -20°C. Two calibrations curves were prepared at a 6-month interval. In the stored calibration curve, concentration of acrylamide at each data point shall not deviate by more than 15 % compared to the freshly prepared one.  In our experiments, accuracy at each fortified data point was within a 90-99% range . Stability information were completed accordingly in parts Parts E.a. and E.g .: • The acrylamide stock standard solution was found stable at - 20 °C for at least 6 months. • Calibration solutions were found stable at - 20 °C for at least 6 months. • Other mentions referring to the stability of standard solutions were withdrawn. With the same wording than above, part H as a paragraph dedicated to Stability .— further develop in the Results and discussion part: According to current reference standards [11,15] , the stability of AA at 1 mg/mL in water is ensured for at least 3 months at 4-6 °C. In our study, the shelf life could be extended to ≥ 6 months at – 20 °C. Indeed, two separate solutions prepared according to E.a. at a 6-month interval, matched at 99%. Similarly, calibration solutions were found stable for ≥ 6 months at – 20 °C. The accuracy of each data point in stored calibration solutions was within a 90-99% range when compared to freshly prepared solutions. Practically, calibration solutions prepared as described in E.g., can be conveniently produced in high amount (e.g., 50 -mL), aliquoted and thawed just prior to analysis. 9. Equilibration time in operating procedure Taking equilibration time for 10-20 column volumes, the recommended equilibration time for the column mentioned in our method (4.6 mm x 150 mm – flow rate at 0.6 mL/min) would correspond to 29- 58 min, respectively. We modified the text accordingly as: ‘(a) Instrument check test.—Allow both LC and MS systems to equilibrate at the initial LC and MS conditions for 30-60 min (corresponds to 10-20 column volumes at 0.6 mL/min) before starting a batch of injections to ensure that the system pressure is stable and there is no leak’. 10. Explain reason for internal standard

Done. We have added one paragraph in the Result and Discussion part. During the validation process, for ruggedness purpose, we voluntarily alternated the isotopically labeled standards (IS) and the LC-MS/MS instruments between each experiment. Acrylamide (AA) being susceptible to interference(s), an interference free analysis is equally important for either d3-AA or 13C3-d3-AA. Results consistency across batches of analysis and identification criteria (i.e., ion ratio, retention time) for AA and its IS were demonstrated. 11. Method user feedback Feedback from users was collected from Nestlé Quality Assurance Center (NQAC) laboratory based in Dublin and R&D laboratory based in Marysville. Document 4: User feedback.

Feedbacks NQACs.docx

Specific Request from Expert Review Panel on Acrylamide (03.05.2023) Reviewer 1: All comments were adressed in the previous pages.

Reviewer 2: I would also like a little more clarification on the data in table 5c- are these average recoveries and basal levels since the analysis was carried out on 6 different days? this is spelled out more clearly in table 5d and I would just want the same clarity in 5c The part H.c Performance characteristics\Validation has been deeply modified. We defined different categories of samples as ‘Non-blank samples’, ‘In-house reference samples’ and ‘Reference materials and quality control materials’. For each category, the validation scheme is included along with the calculations applied for recovery determination. Considering the clarification conducted in H.c , we found clearer to limit the titles of Tables 5 b-d according to the categories just mentioned as: (b)‘In-house reference samples’. Recovery (REC) and precision (RSD r and RSD iR ), in %. (c) ‘Non-blank samples’. Recovery (REC) and precision (RSD r and RSD iR ), in %. (d) ‘ Quality control and reference materials’. Accuracy, in %. Reviewer 3: Consider mentioning use of enzymatic solution for cereal samples in Principle. It also may be worth explaining the reason for using the enzymatic solution for cereal samples in the manuscript. Done. In the part Principle we have added a specific comment to this point as follows: ‘For cereal-based samples only, the efficiency of the initial extraction step is assisted by adding a commercial solution of α -amylase’. Consider adding sample particle size requirements for cereals, potato crisps, tea and dry petfood (F,b,2) Only one grinder applied to roasted coffee bean allowed us to control particle size i.e., ZM 200 with 0.5 mm sieve size. For other matrices particle size was not measured but sample was checked for homogeneity and visual aspects. Consider using " method blank" instead of "reagent blank" throughout the method and manuscript". Should requirements for blank (negative control) be included in the method (e.g. response in blank NMT 30% of respone in LCL?). Done, we have modified accordingly: ‘(a) Method blank.—For each series of analysis, a method blank should be included. This is carried out by performing successively extraction and clean-up procedures (starting from F.d) in the absence of any test portion. Ensure that the peak area of AA in the method blank is absent or less than 1/3 compared to peak area of AA measured in the lowest calibrant.’ Identification requirement in SANTE/11312/2021 for ion ratios is as follows: "Ion ratio from sample extracts should be within ±30% (relative) of average of calibration standards from same sequence". Can authors comment on using ion ratio requirement of ±30%? At the time of the validation, we factually used identification criteria as established in Commission Decision 2002/657/EC. Nevertheless, in the meantime, more recent documents (e.g., SANTE/11312/2021 or EU 2021/808/CR) have been released.

For accuracy purpose, we modified the part G.e mentioning that we used the 2002/657/EC document to establish the identification criteria. Still, we added a note about the two recent documents that can also be taken as reference. To point out that identification criteria have been recently revised in EU Commission Decision 2021/808 [15] or in Document N° SANTE/11312/2021 [16] which can also be taken as reference for setting dentification criteria. For spike recovery calculations, was marginal recovery calculation used on all occasions? If total recovery calculation was used, this should be indicated in relevant tables. Part H.c Performance characteristics\Validation has been deeply modified. Calculations are displayed to provide more clarity. Reviewer 4: All comments were addressed in the previous pages.

AOAC Official Method 2022.xx 1 Acrylamide in coffee, cereals, baby food, cocoa, dry petfood, potato products, vegetable 2 crisps, biscuit, tea, nuts and spices by LC-MS/MS. 3 First Action 2022.xx. 4 Aude Bebius 1 , Frederique Reding 1 , Viviane Theurillat 1 , Valérie Leloup 1 , Erik Konings 2 , 5 Thierry Delatour 2 , Aurélien Desmarchelier 2 . 6 1 Nestlé Research, Orbe, Route de Chavornay, 1350 Orbe, Switzerland. 7 2 Société des Produits Nestlé S.A., Nestlé Research, Vers-chez-les-Blancs, 1000 Lausanne 26, 8 Switzerland. 9 Abstract 10 Background: Acrylamide (AA) is a process contaminant naturally formed during roasting, 11 frying or baking of starchy food at temperatures > 120 °C. Fried potato products, bread, coffee, 12 biscuits and crackers are important contributors of dietary exposure to AA, a molecule 13 classified as probable human carcinogen. While maximum levels for AA in food have not been 14 set so far, legislation is established through benchmark levels (EU), recommended 15 specifications (Republic of Korea) or safe harbor levels (California Office of Environmental 16 Health Hazard Assessment). Considering existing risks of misquantification inherent to the 17 analysis of very small and polar molecules such as AA, there is a need for a consensus standard 18 dedicated to the determination of AA in a broad variety of food. 19 Objective: The study describes the performance characteristics achieved by a quantitative LC- 20 MS/MS method dedicated to the determination of AA in food. Representative samples from 21 multiple food categories (n=9) were validated including potato-based products; baby food; 22 cereal and bakery products; cocoa products; coffee; tea; herbs and spices; dry petfood and nuts. 23 Targeted performance requirements in terms of limit of quantification, recovery and precision 24 were as defined per SMPR 2022.06. 25 Method: The proposed method derives from EN 16618:2015 standard pending modifications 26 brought to the (1) sample preparation (simplified, potentially automated); (2) scope of 27 application (significantly extended) and (3) LC conditions (improved selectivity). In more 28 detail, the protocol involves an initial extraction with water while isooctane is simultaneously 29 added for defatting purpose. After shaking and centrifugation, the supernatant is collected and 30 diluted with water before being purified by two successive solid phase extraction (SPE) 31

1

cartridges. Confirmatory detection of AA is conducted by LC-MS/MS in the multiple reaction 32 monitoring mode (MRM) and isotopic dilution was applied for quantification approach using 33 either 2,3,3- d 3 -acrylamide ( d 3 -AA), or 13 C 3 -2,3,3- d 3 -acrylamide ( 13 C 3 -d 3 -AA) as labelled 34 internal standard. 35 Results: A total of 16 samples from 9 matrix categories were included in the validation process. 36 First, full validation was conducted in coffee (instant, roast), infant cereal, cocoa powder, pet 37 food (croquettes), tea (green tea), spices (black pepper) and nuts (roasted almonds) with 38 satisfactory performances both in terms of recovery (97-108 %) and precision (RSD r and 39 RSD iR < 12 %). Then, the method applicability was demonstrated through the analysis of 40 quality control materials and reference materials including French fries, potato crisps, 41 vegetable crisps, instant coffee, infant food and biscuit (cookie), with accuracy values 42 determined within a 94-107 % range. 43

44

2

Introduction

45

The 20 th anniversary of the discovery of AA in food has just been marked. Indeed in 2002, 46 scientists in Sweden identified significant amounts of AA in carbohydrate-rich foods (e.g., fried 47 potato products, beetroot, bread) heated at temperature > 120 °C [1] . Since then, the dietary 48 intake of AA has been a subject of high concern as the International Agency for Research on 49 Cancer (IARC) classified AA as ‘probable human carcinogen’ in 1994 [2] . While the link 50 between dietary exposure to AA and human cancer is still an open question, reducing the 51 exposure to AA to the highest possible extent remains a priority [3] . 52 Maximum levels intended to lower the exposure to AA in food have not been established so 53 far, but the regulatory landscape is getting stricter. The European legislation is contained in 54 Commission Regulation (EU) 2017/2158 through the establishment benchmark levels and 55 mitigation measures in certain foods [4] , further complemented by Commission 56 Recommendation (EU) 2019/1888 [5] to increase the monitoring of AA in a non-exhaustive 57 list of food. Current benchmark levels will be revised in 2023, while the setting of maximum 58 levels for AA in certain foods is under consideration [6] . Another regulatory update will be 59 effective from April 2023 in the U.S. state of California. The Office of Environmental Health 60 Hazard Assessment (OEHHA) amended Title 27, California Code of Regulation, addressing 61 the exposure to AA in thermally processed food [7] . Maximum concentrations are set as lowest 62 achievable levels and products having AA below such levels will not be subjected to warning 63 labels. In Asia, the Republic of Korea in 2021 established ‘Recommended Specifications’ for 64 AA in food [8] . Interestingly, in addition to classical food matrices such as coffee, French fries, 65 cereals, baby food etc…, a limit in tea was introduced for the first time. From a regulation 66 standpoint, an effective monitoring AA in a very broad variety of food is nowadays eagerly 67 awaited. 68 As a small molecule (molecular weight at 71.08 g/mol) with polar properties (Log P = -0.7), 69 AA is prone to interference, thus to misquantification. Recently, we investigated how 70 uncontrolled sample preparation or LC-MS/MS parameters could impair quantitative 71 measurements of AA [9,10] : In a first method, sample purification combining liquid-liquid 72 extraction and solid phase extraction (SPE) was necessary to get rid of an unknown interference 73 that caused up to a 20-fold overestimation factor in cocoa [9] . In a second example, a method 74 based on the EN 16618:2015 standard [11] could lead to a 40 % overestimation in coffee in 75

3

case of coelution between AA and an interference identified as N-acetyl-ß-alanine. Twenty 76 years after its discovery, reliable measurements of AA in food remain a challenge. 77 In April 2021, to address the current analytical gaps and the regulatory framework, an AOAC 78 initiative was launched to develop a voluntary consensus standard for AA in food [12] . The 79 Standard Method Performance Requirement (SMPR 2022.06) released by the AOAC working 80 group was further adopted in 2023 [13] for the surveillance of AA in potato products, baby 81 food, bread, cereal and bakery products, cocoa products, coffee, tea, herbs, spices, pet food and 82 nuts. Targeted l imits of quantification were set at ≤ 20 µg/kg in bread and baby food or 83 ≤ 50 µg/kg for any other food matrix. Other criteria such as recovery (75-110 %) and precision 84 (based on Horwitz equation) were equivalent to those described in Commission Regulation 85 (EU) 2017/2158 [4 ]. 86 The proposed study describes a single laboratory validation (SLV) for the LC-MS/MS 87 determination of AA in a broad set of matrices. Full validation was conducted in coffee (instant, 88 roast), infant cereal, cocoa powder, pet food (croquettes), tea (green tea), spices (black pepper) 89 and nuts (roasted almonds). The method applicability was further demonstrated by the analysis 90 of quality control samples and reference materials made of French fries, potato, vegetable 91 crisps, instant coffee, infant food and biscuit (cookie). Chromatographic conditions were 92 optimized to ensure baseline separation of AA from other potential matrix interferences 93 mentioned in SMPR 2022.06, namely N-acetyl-ß-alanine, 3-aminopropanamide, lactamide 94 [13] . 95 Material Safety Data Sheets (MSDS) should be available for all chemicals, inherent risks and 97 corresponding safety precautions shall be identified. Follow general safety precautions and 98 environmental aspects as described in the local Safety, Health & Environment rules in place. 99 Acrylamide (AA) is a mutagen and teratogen and is classified by IARC as Probably 100 Carcinogenic to Humans (Group 2A). It is toxic in powder and solution if swallowed, inhaled, 101 or absorbed through the skin. If possible, use pre-made AA in solution rather than creating 102 dilutions from pure powder in the laboratory. 103 Solvents, standard and/or standard solutions of AA should always be handled under a fume 104 hood. Avoid any contact with the skin. Wear a laboratory coat, disposable vinyl gloves, and 105 safety glasses during sample preparation. 106 Safety considerations 96

4

Method

107

A. Principle 108 The protocol involves an initial extraction with water while isooctane is simultaneously added 109 for defatting purpose. After shaking and centrifugation, the supernatant is collected and diluted 110 with water (1+1) before being purified by two successive solid phase extraction (SPE) 111 cartridges that are (1) Isolute® Multimode cartridge which combines non-polar (C18), strong 112 cation exchange (SO 3 - ) and strong anion exchange (-NR 3 + ) retention mechanisms. Used in the 113 “pass-through” mode, interference components (acidic, basic and non-polar) are retained on 114 the column, while AA is collected in a purified state; and (2) Isolute® ENV+® cartridge which 115 is suitable for the extraction of polar analytes such as AA not retained by C18 or C8 sorbents. 116 Eventually, the SPE eluate is partially evaporated before LC-MS/MS analysis in the multiple 117 reaction monitoring mode (MRM). 118 Quantification is performed by the isotopic dilution approach using either d 3 -AA, or 13 C 3 - 13 C 3 - 119 d 3 -AA as labelled internal standards. 120 Positive identification of AA in sample is conducted according to the confirmation criteria 121 defined in EU Commission Decision 2002/657/EC [14] . 122 (a) Conical tubes.— 50-mL, polypropylene (e.g., Falcon, art. 352070, Becton Dickinson 124 Labware). 125 (b) Disposable culture tube, rimless .—4-mL, Pyrex®, (e.g., art. 99445-10, VWR). 126 (c) Amber glass HPLC vials.— 1.5 mL (e.g., art. 5184-3556, Agilent Technologies). 127 (d) Vial caps .—PTFE, silicone septa (e.g., art. 5182-3556, Agilent Technologies). 128 (e) Round bottom tubes .—5 mL, polypropylene, 12 x 75 mm, (e.g., art. 115201, Greiner 129 Bio-One Ltd) for Automated SPE system. 130 (f) Syringe filter .— PVDF LC syringe filters, 25 mm, 0.2 µm (e.g., art. WAT200808, 131 Waters) or PVDF hydrophilic syringe filters, 25 mm, 0.45 µm (e.g., art. CH4525-PV, 132 ThermoFisher Scientific). 133 Equipment for manual SPE: 134 (g) SPE cartridges .—ISOLUTE® Multimode 500 mg/3 mL (e.g., art. 904-0050-B, 135 Biotage). 136 (h) SPE cartridges .—ISOLUTE® ENV+ 500 mg/6 mL (e.g., art. 915-0050-C, Biotage). 137 B. Materials 123

5

Equipment for automated SPE:

138

(i) SPE cartridges .—ISOLUTE® Multimode 500 mg/3 mL (e.g., art. 904-0050-B, 139 Biotage). 140 (j) SPE cartridges .—ISOLUTE® ENV+ 500 mg/3 mL (Tabless) (e.g., art. 915-0050-BG, 141 Biotage). 142 (k) Caps for SPE cartridges .—natural Polyethylene, 3 mL cartridge (e.g., art. 2954699, 143 Gilson). 144 (a) Ultra-centrifugal mill .— with sieve 0.5 mm sizes (e.g., Mill ZM 200, Retsch). 146 (b) Knife mill .—(e.g., Grindomix GM 200, Retsch). 147 (c) Cryogenic grinder .—(e.g., 6875D Freezer/Mill, SPEX SamplePrep). 148 (d) Homogenizer .— (e.g., Polytron PT 3100 D, Kinematica AG). 149 (e) Analytical balance .—with 0.01 g and 0.01mg readability (e.g., PM 2000 and 150 AX205DR, Mettler). 151 (f) Vortex mixer .—(e.g., Vortex genie 2, art. G560E, Scientific Industries Inc.). 152 (g) Universal hot plate magnetic stirrer .—(e.g., IKA RCT basic IKAMAG, art. Z645052, 153 Sigma-Aldrich). 154 (h) Orbital Shaker .—0-300 rpm speed range (e.g., art. KS 501, Fisher scientific). 155 (i) Centrifuge .—use at 5000 rpm (e.g., Sorval RC-6 PLUS with rotor F13S – 14x50c, 156 ThermoFisher Scientific). 157 (j) LC-MS/MS instrument .—(e.g., Agilent 1260 or Agilent 1290 LC systems (Agilent 158 Technologies) coupled to either Sciex 5500 or Sciex 6500+ triple quadrupole 159 spectrometers (Sciex, Foster City, CA, USA) equipped with a Turbo V™ Ion Source. 160 (k) HPLC column .—Shodex column RSpak DE-413, 150 mm x 4.6 mm, 4 µm (art. 161 F7001005, ShowaDenko) equipped with HPLC pre-column .—Shodex guard column 162 RSpak DE-G 4A, 10 mm x 4.6 mm, 10 µm (art. F6700150). 163 Note: Detailed information on LC and MS/MS instrumental conditions are given in Tables 1 164 and 2 respectively. 165 Equipment for manual SPE: 166 (l) Evaporator .—for manual SPE (e.g., TurboVap LV, art. 415000, Biotage). 167 (m) SPE vacuum manifold .—(e.g., Visiprep, 24-port model, art. 57265, Sigma Aldrich). 168 (n) SPE vacuum pump trap kit .— (e.g., art. 51120-U, Sigma Aldrich). 169 C. Equipment 145

6

(o)

Disposable liners for SPE manifold .—(e.g., art. 57059, Sigma Aldrich).

170

Equipment for automated SPE:

171

(p) Automated Solid Phase Extraction system .—(e.g., Gilson GX-274 ASPEC™ equipped 172 with 3-mL cartridges mobile racks). 173 Note: Detailed parameters about automated SPE are provided in Table 3 . 174

D.

Chemicals and reagents

175

(a) Acrylamide, analytical standard .—purity ≥ 99 % (e.g., art. 23701, Sigma Aldrich). 176 (b) Acrylamide, internal standard .—(e.g., 2,3,3-d 3 -Acrylamide, D enrichment ≥ 98 %, 177 purity ≥ 98 %, art. 636568) or 178 (c) Alternative acrylamide, internal standard .—(e.g., 13 C 3 -2,3,3-d 3 -Acrylamide, D 179 enrichment ≥ 98 %, 13 C enrichment ≥ 99 %, purity ≥ 98 %, art. 798924) both from 180 Sigma Aldrich. 181 (d) Isooctane .—for analysis EMSURE® ACS, Reag. Ph Eur (e.g., art 1.04727, Sigma 182 Aldrich). 183 (a) Methanol .—hypergrade for LC-MS LiChrosolv® (e.g., art. 106035, Sigma Aldrich). 184 (b) Methanol .—for analysis EMSURE® ACS, Reag. Ph Eur (e.g., art. 1.06009, Sigma 185 Aldrich). 186 (c) Water .—e.g., ultrapure water (e.g., Milli Q water) or LC-MS Grade (LiChrosolv®, art. 187 1.15333, Merck). 188 (d) Alpha-amylase (e.g., Fungamyl® 2500 SG, Novozymes). 189 E. Reagents and solution preparation 190 (a) Acrylamide stock standard solution 1000 µg/mL in water .—Weigh 10.00 mg ± 0.01 mg 191 of AA into a 10-mL volumetric flask. Dilute to volume with water. This solution is stable at 192 - 20 °C for at least 6 months. 193 (b) Acrylamide working standard solution 0.5 µg/mL in water .—Transfer 100 µL of the 194 stock solution (E.a) into a 200-mL volumetric flask, and complete to the mark with water. Store 195 at 4 °C for maximum 3 months. 196 (c) 2,3,3-d 3 -acrylamide (d 3 -AA) stock standard solution 200 µg/mL in water (Internal 197 Standard (IS)) .—Weigh 10.00 mg ± 0.01 mg of d 3 -AA into a 50-mL volumetric flask. Dilute 198 to volume with water. Store at - 20 °C for maximum 9 months. 199

7

(d) 2,3,3-d 3 -acrylamide (d 3 -AA) working standard solution 4 µg/mL in water (IS) .—Add 2 200 mL of stock standard solution (E.c) into a 100-mL volumetric flask, and complete to the mark 201 with water. Store at - 20 °C for maximum 9 months. 202 (e) 13 C 3 -2,3,3-d 3 -acrylamide ( 13 C 3 -d 3 -AA) ca 20 µg/mL in water (IS) .— 13 C 3 -d 3 -AA is 203 supplied in small amount (ca. 1 mg). To record the weighed mass as precisely as possible, 204 weigh the original commercial bottle (m1). Transfer the content into a 50-mL volumetric flask. 205 Rinse the bottle with water and complete the flask to the mark with water. Weight again empty 206 dried commercial bottle (m2) to record weighed amount of 13 C 3- d 3 -AA by difference (m IS = 207 m1- m2). Store at - 20°C for maximum 9 months. 208 (f) 13 C 3 -2,3,3- d 3 -acrylamide ( 13 C 3 -d 3 -AA) working standard standard solution 2 µg/mL in 209 water (IS) .—Into a 100-mL volumetric flask, dilute an adequate volume of stock solution (E.e) 210 to obtain a concentration of 2 µg/mL (± 0.3 µg/mL) in water. Store at - 20 °C for maximum 9 211 months. 212 (g) Calibration solutions .—Into separate 50-mL volumetric flasks (n=7), transfer 0 / 0.15 213 / 0.5 / 1.5 / 2.5 / 5 and 10 mL of the AA 0.5 µg/mL working standard solution (E.b), 214 corresponding to 0 / 1.5 / 5 / 15 / 25 / 50 and 100 ng/mL concentration. Add in each flask 0.1 215 mL of the IS working standard solution. Complete to the mark with water. Calibration solutions 216 are stable at - 20°C for at least 6 months. 217 (h) Rinsing solution during SPE cleanu-up Methanol-water (60+40) .— Into a 200-mL 218 glass bottle, transfer 120 mL of methanol and 80 mL of water (measured separately in 200 and 219 100-mL volumetric cylinders) and mix. Store at ambient temperature for maximum one month. 220 (i) Enzymatic solution for cereals samples preparation.— Prepare the enzymatic solution 221 by weighing Fungamyl® 2500 SG, applying a 1/400 (w/v) dilution factor in water. As an 222 example, for 24 samples preparation, dilute 1.20 g ± 0.05 g of enzyme into 480 mL of water. 223 Homogenize using magnetic stirrer plate and let it rest for 10 min before using. This solution 224 can’t be stored, it must be freshly prepared. 225

F.

Laboratory test sample preparation

226

(a) Laboratory sample.— A representative sample should be sent to the laboratory, ideally 227 with a minimum of 100 g. Sample must not be damaged or changed during transport and 228 storage. It should be transported in air-tightness container, protected from light, and stored at 229

8

- 20 °C ± 3 °C if analyses are not scheduled within 2 weeks. For short storage periods up to 230 two weeks: 231 (1) Instant coffee, roasted coffee, spices, tea.— storage is recommended at 4 °C. 232 (2) Nuts.— storage is recommended at - 20 °C until analysis. 233 (3) Cereals, dry petfood, cocoa, crips .— storage at room temperature 234

(b)

Test sample preparation.— recommended following grinding conditions:

235

(1) Roasted coffee beans .— grind 100 g ± 5 g of the laboratory sample using an ultra - centrifugal Mill (e.g., ZM 200 with 0.5 mm sieve size). Be careful to not overheat the

236 237 238 239 240

sample during milling.

(2) Cereals, potato crips, tea, dry pet food.—G rind 100 g ± 5 g of the laboratory sample

with a knife mill (e.g., Grindomix GM 200, Retsch) at 8000 rpm for 10 s.

(3) Cocoa, spices.— no specific preparation. 241 The day of the analysis, bring the whole laboratory sample to room temperature and 242 homogenize it by mixing. 243 (4) Nuts.— Grind 100 g ± 5 g of the laboratory sample with a cryomill (e.g, 6875D 244 Freezer/Mill, SPEX SamplePrep). Store rapidly at - 20 °C after grinding (F.a) and 245 weigh immediately prior to test portion preparation (see F.c). 246 (c) Test portion preparation.— Into a 50-mL polypropylene tube, weigh appropriate 247 amounts of sample as follows: 248 (1) Pure soluble coffee, potato crisps, cocoa: 1.000 g ± 0.100 g 249 (2) Roast and ground coffee, breakfast cereals, dry petfood, tea, nuts, spices: 250 2.000 g ± 0.100 g 251 (3) Infant cereals: 4.000 g ± 0.100 g 252 In each tube, add 100 μL of selected IS working standard solution 253

(d)

Extraction.—

254

Add 5 mL of isooctane.

255 256 257 258 259

Depending on the selected matrix, add:

(1) 20 mL of water

(2) 20 mL of the enzymatic solution for cereal-based products.

Shake during 30 s with a vortex mixer.

9

Shake at 300 rpm for 30 min at room temperature with an orbital shaker. 260 Centrifuge at 5000 rpm for 10 min at 20°C. 261 Discard the upper phase (isooctane). 262 Pipette 2 mL of aqueous extract and dilute with 2 mL water. 263 Filter through e.g., PVDF syringe filters into culture tubes for SPE steps. This step is critical 264 for automated SPE. 265 267 Condition with 3 mL methanol and 10 mL water. Prevent the sorbent from drying. 268 Load 0.5 mL of the 2-mL diluted sample extract (to discard). 269 Load and collect in a 15-mL tube the remaining diluted sample extract. 270 Finish eluting with 1 mL water. 271 (2) Isolute ENV+ cartridge: 272 Condition with 5 mL methanol and 5 mL water. Prevent the sorbent from drying. 273 Load the fraction obtained from the Isolute Multimode cartridge. 274 Let the extract soak for 2 min onto the cartridge phase. 275 Rinse twice with 2 mL water. 276 Load 0.5 mL of MeOH-water (60+40) solution (E.h) (to discard). 277 Elute and collect with 1.5 mL MeOH-water (60+40) solution (E.h). 278 Dry carefully the ENV+ cartridge under vacuum to collect the test sample solution remaining 279 as AA might be stocked on the phase of the cartridge. 280 (f) Solid Phase Extraction with automated SPE.— The main parameters of the automatic 281 SPE manifold are presented in Table 3 . 282 283 Depending on the MS detector sensitivity, evaporate the extract to 0.5-1 mL under a gentle 284 stream of nitrogen at ambient temperature. Transfer test sample into a LC amber glass micro- 285 vials for further LC-MS/MS analysis. 286 287 (a) Instrument check test.— Allow both LC and MS systems to equilibrate at the initial LC and 288 MS conditions for 30-60 min (corresponds to 10-20 column volumes at 0.6 mL/min) before 289 starting a batch of injections to ensure that the system pressure is stable and there is no leak. 290 (e) Solid Phase Extraction with Manual SPE.— 266 (1) Isolute Multimode cartridge (g) Preparation of test solutions G. Operating procedure

10

(b) Sample analysis sequence setup.— After system equilibration, start the sequence by 291 injecting one solvent blank (water) and one method blank. Inject the calibration curve, ensuring 292 that AA and AA-IS are detected at the expected retention time. Inject a solvent blank to check 293 for possible carry-over. Depending on the number of samples, inject alternatively six test 294 samples and one solvent blank to ensure the absence of carry-over. End the sequence by 295 reinjecting at least 1 calibration point of the calibration curve and two solvent blanks to rinse 296 the column. 297 (c) Data treatment.— Process the data using the appropriate instrument-dedicated software. 298 Peak areas are used for subsequent calculations. 299 (d) Calibration.— AA is quantified by the isotope dilution approach using an external 300 calibration curve. Draw the calibration curve ‘Area Ratio = Conc Ratio x Slope + Intercept’ by 301 plotting peak area ratio between AA and IS (= y-axis ) versus the injected amount ratio for AA 302 and IS (= X-axis). Calculate the slope and the intercept by linear regression and check that the 303 calibration curve is linear: Regression coefficient R 2 should be ≥ 0.99 ; the relative standard 304 deviation of the average of response factors (= y/x) should be < 20 % and deviation of the back- 305 calculated concentrations from true concentration should be ≤ ± 20 %. 306 (e) Identification and confirmation.— AA and IS were considered as positively identified when 307 all confirmation criteria as defined in EU Commission Decision 2002/657/EC [14] were 308 fulfilled: (1) the peak area ratios from the different transition reactions recorded for both 309 unlabeled and labeled species were within the tolerances set in [14]. For the transition reactions 310 mentioned in Table 2b , tolerance was set at 50 %. (2) the ratio of the chromatographic retention 311 time of AA to that of the IS, i.e., the relative retention time of AA, corresponds to that of the 312 averaged relative retention time in the calibration solutions within a 2.5 % tolerance. 313 To point out that identification criteria have been recently revised in EU Commission Decision 314 2021/808 [15] or in Document N° SANTE/11312/2021 [16] which can also be taken as 315 reference for setting dentification criteria. 316 (f) Expression of result.— Calculate mass fraction of AA, w, in µg/kg using the following 317 equation: 318 w = − × 319

11

Where A AA is the peak area of AA; A IS is the peak area of the IS; I is the intercept of the 320 regression line; S is the slope of the regression line; m IS is the mass of IS added to the test 321 portion, in ng; and m a is the mass of the test portion, in g. 322 323 (a) Scope.— The manuscript presents the results of a single laboratory validation (SLV) 324 fulfilling the analytical requirements outlined in SMPR 2022.006 [13] . Validation data were 325 acquired in instant coffee, roasted coffee, cereal-based products (infant- and breakfast-cereals), 326 petfood (dry), cocoa, crisps (potato and vegetable), spices (black pepper), tea (green tea) and 327 nuts (roasted almonds). 328 (b) Samples.— As shown in Table 4 , a total of 16 individual samples from 9 matrix categories 329 as defined in SMPR 2022.006 [13] were analyzed during the validation process. 330 ‘Non-blank samples’ (n=5) of cocoa powder, dry pet food (croquettes), green tea, black pepper 331 and roasted almonds were bought from Swiss retailers. 332 ‘In-house reference samples’ (n=3) were also included for the validation with one pure soluble 333 coffee (reference value = 736 ± 96 μg/kg), one roast and ground coffee (reference value = 336 334 ± 45 μg/kg) and one infant cereal (reference value = 23 ± 6 μg/kg). To establish reference 335 values, each sample was sent to six different laboratories, each conducting duplicate analyses 336 under six different conditions (varying LC-MS/MS, analysts, calibration solutions, SPE lots or 337 days of analysis). After a statistical treatment that involved the exclusion of outlier data, the 338 median of remaining values was established as the reference value. 339 ‘Quality control and Reference materials’ (n=8) were a selection of samples from FAPAS 340 (Sand Hutton, York, United Kingdom) that included French fries (pre-cooked) (TYG085RM, 341 reference value = 71 ± 4 μg/kg); biscuit (cookie) (T30104QC, 205 μg/kg, range for |z| ≤ 2 = 342 122– 288 μg/kg); vegetable crisps (T30101QC, 1105 μg/kg, range for |z| ≤ 2 = 757 –1453 343 μg/kg); infant foodstuffs (biscuit for infants) (T3092QC, 164 μg/kg, range for |z| ≤ 2 = 95 –233 344 μg/kg); infant food (T30102QC, 196 μg/kg, range for |z| ≤ 2 = 116 – 276 μg/kg) and instant 345 coffees (T3087QC, 613 ± 12 μg/kg and T30107QC, 561 ± 16 μg/kg) and potato crisps (T30109, 346 718 μg/kg, range for |z| ≤ 2 = 476 – 959 μg/kg−1). 347 (c) Validation.— Performance characteristics in terms of recovery, accuracy and precision were 348 determined by the analysis of reference materials and through spiking experiments . To 349 demonstrate the method ruggedness, experiments were conducted by alternating the following 350 parameters: (a) two operators, (b) two distinct internal standards ( d 3 -AA or 13 C 3 -d 3 - AA), (c) 351 H. Performance Characteristics

12