<<

Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018 1

FOOD BIOLOGICAL CONTAMINANTS Performance Validation of the Microbiologique MicrofilmTM Test System for AOAC Research Institute Performance Tested Method SMCertification

AOAC Performance Tested Method SM 051702

Abstract anna ShapovaLova and hariSh K. Janagama Molecular Epidemiology, Inc., 15300 Bothell Way NE, The Microfilm™ Test System is intended for quantitative Lake Forest Park, WA 98155 and consists of three types of Microfilms for aerobic Long vuong and aLex Friedrich plate count (Microfilm APC), total coliform and Escherichia coli IEH and Consulting Group, Inc., 15300 Bothell count (Microfilm TCEc), and yeast and mold count (Microfilm Way NE, Lake Forest Park, WA 98155 YMC). This study evaluated the performance of the Microfilm dyLan JohnSon Test System against International Organization for Standardization Molecular Epidemiology, Inc., 15300 Bothell Way NE, (ISO) methods on 20 food matrixes and 2 environmental surfaces. Lake Forest Park, WA 98155 Ruggedness, robustness, and stability were also determined, LourdeS m. nadaLa while inclusivity and exclusivity studies were performed on Molecular Epidemiology, Inc., 15300 Bothell Way N.E., Microfilm TCEc and YMC. An independent evaluated Lake Forest Park, WA 98155; Microbiologique, the performance on four food matrixes and one environmental Inc., 8215 Lake City Way NE, Seattle, WA 98115 surface. No significant differences and high correlation coefficients van nguyen were observed between the Microfilm Test System and the Microbiologique, Inc., 8215 Lake City Way NE, Seattle, corresponding ISO methods (ISO 4833-1:2013 for APC, ISO WA 98115 4832:2006 for total coliform count, ISO 16649-2: 2001 for E. coli, eLpidio ceSar nadaLa, Jr and ISO 21527 Part 1 and Part 2 for YMC) in spiked food matrixes Molecular Epidemiology, Inc., 15300 Bothell Way NE, and environmental samples. These results were corroborated by Lake Forest Park, WA 98155 the independent laboratory. Inclusivity and exclusivity studies for manSour Samadpour Microfilm TCEc showed expected results for all theE. coli strains Molecular Epidemiology, Inc. and IEH Laboratories and tested (blue-violet or violet color), while the related coliforms Consulting Group, Inc., 15300 Bothell Way N.E., showed the expected blue-green colonies on the Microfilm. Lake Forest Park, WA 98155 Similarly, all 100 fungal strains tested showed typical growth on Microfilm YMC. Exclusivity testing on Microfilm TCEc and subMitting CoMpany YMC showed no growth of nontarget organisms. Robustness and Microbiologique Inc., 8315 Lake City Way NE, Seattle, ruggedness studies showed no significant differences in mean WA 98115 difference counts at varying incubation temperatures and times. Stability studies on three lots of the Microfilm Test System showed independent laboratory that it is stable at 2–25°C for 12 months and at 45°C for 6 weeks. University of Washington, School of Public Health, 4225 Roosevelt Way NE, Suite 100, Seattle, WA 98195 Participants reViewers yi chen Method authors U.S. Food and Drug Administration, Center for Food Safety Tam L. mai and Applied Nutrition, 5100 Paint Branch Pkwy, College Park, Molecular Epidemiology, Inc., 15300 Bothell Way NE, MD 20740 Lake Forest Park, WA 98155 Jim agin Shao-Lei Sung Q Laboratories, 1400 Harrison Ave, Cincinnati, OH 45214 Microbiologique, Inc., 8215 Lake City Way NE, Seattle, michaeL BrodSKy WA 98115 Brodsky Consultants, 73 Donnamore Crescent, Thornhill, ON L3T 4K6, Canada Submitted February 8, 2018. The method was independently tested, evaluated, and certified by Scope of Method the AOAC Research Institute as a Performance Tested Method SM. See http://www.aoac.org/testkits/steps.html for information on certification. Supplemental information is available online at: http://aoac. (a) Target organisms.—Total aerobic bacteria, total publisher.ingentaconnect.com/content/aoac/jaoac coliform/Escherichia coli, yeasts, and molds. Corresponding author’s e-mail: [email protected] (b) Matrixes.—(1) Food.—Meat products [raw beef, 2% DOI: https://doi.org/10.5740/jaoacint.18-0036 fat; raw pork, 20% fat; ready-to eat (RTE) lunch meat], poultry

Reproduced from Journal of AOAC International 101 (2018). DOI: https://doi.org/10.5740/jaoacint.18-0036 Lourdes Nadala: Participant of the 15th UM, 1987-1988.

168 169 2 Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018

products (raw chicken, 2% fat; raw turkey, 1% fat), fish and General Information seafood products (raw shrimp, surimi), fruits and vegetable- based products (spinach, lettuce, mung bean sprouts), nut-based Total aerobic count, coliforms, generic E. coli, yeasts, and products (peanut butter, almonds), dairy products (pasteurized molds are routinely enumerated in foods as part of product milk, 2% fat; ice cream), chocolate/bakery products (wheat release criteria or, in the case of swabs, for environmental flour, frost and topping mix), pasta (uncooked noodles), spices monitoring. Tests typically include plating on specific (black pepper, cumin), and animal feed (dry pet food). microbiological media, such as described in the U.S. Food (2) Environmental swabs.—Stainless steel and plastic. and Drug Administration Bacteriological Analytical Manual (c) Summary of validated performance claims.— methods, ISO standards, U.S. Department of Agriculture Performance equivalent to that of the International Organization methods, etc., and ready-to-use films. The Microfilm Test for Standardization (ISO) standards for total aerobic count System provides rapid and easy-to-use enumeration methods (ISO 4833-1:2013, Microbiology of the food chain–Horizontal for aerobic plate count (APC), total coliform count (TCC), method for the enumeration of microorganisms–Part 1: Colony E. coli, yeast, and molds. count at 30°C by the pour plate technique; 1), total coliforms (ISO 4832:2006, Microbiology of food and animal feeding Materials and Methods stuffs–Horizontal method for the enumeration of coliforms– Colony count technique; 2), E. coli (ISO 16649-2:2001, Test Kit Information Microbiology of food and animal feeding stuffs–Horizontal method for the enumeration of ß-glucuronidase-positive (a) Kit name.—Microbiologique Microfilm Test System. E. coli–Part 2: Colony count technique at 44°C using 5-bromo- (b) Cat No.—Microfilm Test System (M-2007): Microfilm 4-chloro-3-indolyl ß–D-glucuronide; 3), and yeasts and molds APC (M-1022), TCEc (M-1021), and Microfilm YM (M-1020). (ISO 21527 Part 1 and Part 2:2008, Microbiology of food and (c) Ordering information.—In the United States and Canada.— animal feeding stuffs–Horizontal method for the enumeration of Tel: 1-888-998-4115, E-mail: [email protected]. yeasts and molds; 4, 5) as reference culture methods. Other.—Tel: +1-206-525-0412.

Principle Test Kit Components

The Microbiologique Microfilm System is applicable for (a) Microfilm APC.—50 sterile sheets contained in sterile, enumeration of total aerobic bacteria (APC), total coliform/E. coli resealable pouches. (TCEc), or yeast and mold (YMC) in food and environmental (b) Microfilm TCEc.—50 sterile sheets contained in sterile, samples. For solid foods, the sample is first mixed with buffered resealable pouches. peptone water (BPW) or appropriate diluent, then mixed well by (c) Microfilm YMC.—50 sterile sheets contained in sterile, hand-massage or stomacher, and then 1 mL is inoculated onto resealable pouches. each of the three Microfilms that have been pre-impregnated with different media. For food samples with high bacterial load, serial Additional Supplies and Reagents Required But Not dilution may be performed prior to plating on the Microfilms. Supplied With the Kit The inoculated Microfilms are then incubated at 36 ± 1°C for 24 h (APC and TCEc) or 26 ± 1°C for 48 h (YMC) to allow for (a) Rinse/Diluent Solution.—One pouch sufficient for the formation of visible colonies. The colonies are counted, and making 2 L of Rinse solution (may be ordered separately from the count value is used to calculate the original concentration of Microbiologique, Inc.): BPW for Microfilm APC and TCEc, CFUs expressed as CFU/g or CFU/surface swab sample. and Peptone Water Broth, 0.1%, for Microfilm YMC. The System consists of three distinct indicator/selective (b) Swabs.—Sterile, individually wrapped (required for media impregnated into separate support matrixes. Each of environmental sample collection only; Microbiologique- the indicator/selective media contains growth nutrients, recommended swab in packs of 50 may be ordered separately). indicator dyes, and selective reagents needed to support rapid (c) Whirl Pak.—18 ounces, for sample preparation/rinsing. and specific growth for each class of microorganism being (d) Sterile measuring cylinder. enumerated. To calculate the number of CFUs, colonies are (e) Laboratory balance.—2000 g with 0.1 g accuracy. counted as colored spots on the Microfilm. For Microfilm (f) Sterile water. APC, colonies are colored, making them easily visible, (g) Pipet.—200 μL and 1000 μL and corresponding tips. thus facilitating the counting of colonies (Figure 1A). For (h) Fine-tipped marking pen. Microfilm TCEc, the medium contains growth nutrients (i) Gloves.—Nitrile, powder-free. supportive for coliforms and selective agents that inhibit (j) Alcohol lamp (>90% alcohol) or Bunsen burner and the growth of Gram-positive bacteria. Colonies of coliforms stainless steel forceps (round-tipped ). appear bluish-green to green, while E. coli colonies are dark blue–violet to violet (Figure 1B). Microfilm YMC contains Apparatus growth nutrients and an antibacterial agent. Yeasts form small, raised blue, light blue, green, or light green colonies (a) Vortex machine. with a defined edge and a uniform color (Figure 1C), while (b) Blender, stomacher, grinder, or similar device for mold colonies are normally larger with a diffused edge homogenizing sample. and show various colors depending on the genus/species (c) Incubators.—36 ± 1°C and 26 ± 1°C. (Figure 1D).

168 169 Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018 3

(a) (b)

E. coli

Coliform

(c) (d)

C

Figure 1. Photographs of the Microfilm Test System: (a) Microfilm APC inoculated withStaphylococccus aureus ATCC 51740; (b) Microfilm TCEc inoculated with E. coli ATCC 25922 (Dark blue–violet colonies) and Klebsiella oxytoca MEI 70309 (green colonies); (c) Microfilm YMC inoculated with Candida glabrata MEI 283 (green colonies with defined edge); and (d)Aspergillus niger ATCC 16404 (green colonies with diffused edge). ATCC: American Type Culture Collection, Bethesda, MD; MEI: Molecular Epidemiology, Inc., Seattle, WA. Color images are available online at http://aoac.publisher.ingentaconnect.com/content/aoac/jaoac

Safety Precautions each dilution tube well on a (approximately 10 s) before proceeding to the next dilution transfer. Follow routine good laboratory and aseptic techniques. (b) Environmental swabs.—Aseptically remove swab Dispose of used Microfilms following local regulations on handle, and stomach the swab with the initial liquid for 1 min biohazard waste. or hand-massage for 2 min. Serial dilutions are performed as above. General Preparation Inoculation onto Microfilm Remove each Microfilm from packaging following aseptic techniques. Reseal the bag and store at 2–25°C. (1) Aseptically remove from the pouch and set aside the number of Microfilms needed. Label sample identification (ID) Sample Preparation and Microfilm Test on the top corner of the respective Microfilm. (2) Lift the top plastic sheet and apply 1.0 mL of the sample (a) Food samples.—Aseptically weigh 50 g of well-mixed to each of the three Microfilms (APC, YMC, and TCEc) by sample into a Whirl-Pak or stomacher bag. Add 450 mL of dispersing the liquid over the entire surface of the film, then sterile Diluent Solution, and stomach for 1 min or hand- carefully release the plastic sheet to cover the film. massage for 2 min to obtain a 10-1 dilution. Serial 10-fold (3) Place the inoculated Microfilms in a 36 ± 1°C dilutions can be prepared using the same Diluent Solution. Mix for APC and TCEc, and a 26 ± 1°C incubator for YMC.

170 171 4 Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018

Interpretation and Test Result Report Validation Study

Enumeration of colonies.—Interpretation is based on the This validation study was conducted under the AOAC suggested extraction/dilution protocol. Take into consideration Research Institute (RI) Performance Tested Method SM program the dilution used for inoculation onto the film. Therefore, if and according to the AOAC International Methods Committee additional dilutions have been incurred to get samples within Guidelines for Validation of Microbiological Methods for Food range of upper enumeration limits, these dilution factors should and Environmental Surfaces (6). Method developer studies, be considered in the final CFU calculations. Refer to counting such as inclusivity/exclusivity [for Microfilm TCC/E. coli count rules below on correct method of enumeration. (ECC) and Microfilm YMC only], matrix performance studies, Microfilms are manually counted at 24 ± 2 h (APC, TCEc) product consistency and stability studies, and robustness and at least 48 h (YMC): testing, were conducted by Microbiologique Inc. at the Institute For APC.—After the incubation period (24 ± 2 h), count all for Environmental Health Laboratories (Seattle, WA). The visible colored colonies, regardless of size or intensity. Multiply independent laboratory study was conducted by the School of count by the appropriate dilution factor and report result. Public Health, University of Washington, and included a matrix For E. coli.—After the incubation period (24 ± 2 h), count study on five matrixes: four food and one surface. all dark blue–violet to violet colonies, regardless of size or intensity. Multiply count by the appropriate dilution factor and report result. Inclusivity and Exclusivity Studies For TCC.—After the incubation period (24 ± 2 h), count all blue-green and violet colonies, regardless of size or intensity. Inclusivity and exclusivity studies were performed to test the Multiply count by the appropriate dilution factor and report result. ability of the Microfilm TCEc and Microfilm YMC to support For yeast and mold.—After incubation period (at least 48 h), the growth of target organisms (inclusivity) and exclude the count all colonies with defined edges as yeast and all other nontarget organisms (exclusivity). colonies with diffused edges as mold. Multiply each count by Microfilm TCEc.— Fifty coliform bacteria (except E. coli) and the appropriate dilution factor and report result. 52 well-characterized strains of E. coli from the American Type Culture Collection (ATCC) and the Molecular Epidemiology, Additional Notes Inc. (MEI; Seattle, WA) culture collection were tested on one lot of Microfilm TCEc for specificity to coliforms and E. coli, respectively. Thirty nontarget strains from ATCC and MEI were (a) For APC and TCEc.—(1) Statistically reliable range tested for their ability to grow on Microfilm TCEc. (10–250 CFU).—Count all CFU on the Microfilm. Record All test strains were individually inoculated into Tryptic Soy dilution(s) used and total number of colonies counted (e.g., 40 -1 Broth (TSB) medium and incubated at 36 ± 1°C for 24 h and used colonies on 10 dilution film = 400 CFU). to prepare cell suspensions with an approximate cell density of (2) High range (>250 CFU).—When number of CFU per 10–100 CFU/mL on the test Microfilm. One mL of cell suspension Microfilm exceeds 250 for all dilutions tested, record the count was applied on individual Microfilm TCEc and incubated, and as too numerous to count and retest the sample. the growth/colony color of each strain was examined. (3) Low range (<10 CFU).—If Microfilms from all dilutions Microfilm YMC.—One hundred fungal strains (50 each of yield fewer than 10 colonies, record the actual number of -1 molds and yeasts) were tested on Microfilm YMC. Strains were colonies on the lowest dilution (e.g., 5 colonies on 10 dilution each inoculated onto Sabouraud Dextrose Agar (SDA) plates film = 50 CFU). and incubated at 26 ± 1°C for 48–72 h (or until adequate growth (4) Consecutive dilutions with count values in the countable was observed). A cell suspension in BPW was prepared for each range.—When Microfilms from consecutive dilutions are strain and then stabilized at 2–8°C prior to inoculation. All fungal within range, compute the counts from both Microfilms and suspensions were enumerated and adjusted to an estimated cell report the arithmetic average as the final result, unless one density of approximately 150 CFU/mL on the test Microfilm count is more than twice the other, then report the lower count YMC. One mL was applied to the Microfilm and incubated at value. 26 ± 1°C for 48 h, and growth/colony of each strain was examined. (5) Consecutive dilutions with no observable CFU.—When For exclusivity testing, 30 bacterial strains were prepared as Microfilms from all dilutions have no colonies, report count as individual inoculum, and each was tested on Microfilm YMC less than 1 times the corresponding lowest dilution used (e.g., -1 and incubated at 26 ± 1°C for 48 h. no CFU counted on 10 dilution film = <10 CFU). (b) For yeast and mold.—(1) Statistically reliable range (1–150 CFU).—Count all CFU on the Microfilms. Record the Matrix Study dilution(s) used and total number of colonies counted (e.g., 2 colonies on 10-1 dilution film = 20 CFU). Twenty food matrixes were tested for method comparison (2) High range (>150 CFU).—When number of CFU per between the Microfilm Test System and the corresponding Microfilm exceeds 150 for all dilutions, record the count as too ISO reference method: raw ground beef (7% fat), raw pork numerous to count and retest the sample using an appropriate (20% fat), RTE lunch meat, raw chicken (2% fat), raw turkey dilution of the sample. (1% fat), raw shrimp, surimi, spinach, lettuce, mung bean (3) No CFUs.—When Microfilms from all dilutions have no sprouts, peanut butter, almonds, pasteurized milk (2% fat), ice colonies, report count as less than 1 times the corresponding cream, wheat flour, frost and topping mix, uncooked noodles, lowest dilution used (e.g., no CFU counted on 10-1 dilution black pepper, cumin, and dry pet food. Two environmental film = <10 CFU). surfaces (stainless steel and plastic) were spiked with cocktails

170 171 Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018 5 of inoculum consisting of one gram-positive strain, one E. coli food matrix, five replicates per inoculum level were prepared strain, and one non-E. coli coliform, or yeast and mold at four for testing. Ten different inoculum cocktails were used for levels (Low: 10–100 CFU/g; Medium: 103–104 CFU/g; High: inoculation to ensure a variety of species were tested (Table 1). >104 CFU/g; and uninoculated). Depending on the food type, unstressed, heat-stressed, or Inoculum preparation, sample inoculation, and lyophilized organisms were used for inoculation: stabilization.—Two sets of test matrixes were inoculated as (a) Raw and cold-processed foods.—Unstressed organisms follows: one set with a composite of three microorganisms (one and stabilized at 2–8°C for 2 days. Gram-positive, one E. coli strain, one non–E. coli coliform) and (b) Heat-processed foods.—Heat-stressed organisms were the second set with one yeast and one mold). Each set of food prepared as follows and stabilized at room temperature for matrixes was divided into four portions (50 g per test portion): 2 weeks. Each of three portions was inoculated with 1–2 log CFU/g or mL Bacteria.—One colony of each organism was grown at (Low), 2–4 log CFU/g or mL (Medium), and >4 log CFU/g or 35°C to the exponential phase (approximately 108 CFU/mL; mL (High), respectively. One portion was uninoculated. For each the growth curves of all tested organisms are predetermined).

Table 1. List of inoculum mix composition, levels of inoculum and food/environmental matrixes for method comparison study

Inoculum cocktail Strain Level for each organism, CFU/g or mL Food matrix

A A1: Uninoculated Almond, surimi Enterococcus raffinosus MEI 28542 Low: 10–100 (1–2 log) Citrobacter freundii MEI 10187 Medium: 100–10000 (2–4 log) E. coli ATCC BAA-178 High: >10000 (>4 log) A2: Candida krusei ATCC 6258 Aspergillus clavatus MEI 78201 B B1: Uninoculated Pasteurized milk, 2% Staphylococcus aureus ATCC 29247 Low: 10–100 (1–2 log) fat RTE lunch meat Klebsiella pneumoniae MEI 10625 Medium: 100–10000 (2–4 log) E. coli ATCC 43887 High: >10000 (>4 log) B2: Pichia fermentans MEI 27049-2 A. parasiticus MEI 27261-24 C C1: Uninoculated Stainless steel, HDPE E. faecalis ATCC 19433 Medium: 100–10000 (2–4 log) plastic K. oxytoca MEI 70309 High: >10000 (>4 log) E. coli ATCC 25922 C2: C. glabrata MEI 283 A. niger ATCC 16404 D D1: Uninoculated Ground beef, 7% fat Listeria innocua ATCC 51742 Low: 10–100 (1–2 log) ground pork, 20% fat C. freundii ATCC 8909 Medium: 100–10000 (2–4 log) E. coli ATCC 11775 High: >10000 (>4 log) D2: Trichosporon asahii MEI 17179 A. aculeatus MEI 12923 E E1: Uninoculated Ground chicken, 2% fat S. aureus ATCC 51740 Low: 10–100 (1–2 log) ground turkey, 1% fat E. coli ATCC 11840 Medium: 100–10000 (2–4 log) K. pneumoniae MEI 8326 High: >10000 (>4 log) E2: Saccharomyces kudriavzevii MEI 28746 A. awamori MEI 23114 F F1: Uninoculated Raw shrimp, spinach E. gallinarum ATCC 49608 Low: 10–100 (1–2 log) E. coli MEI 90672-04b Medium: 100–10000 (2–4 log) Enterobacter aerogenes ATCC 13048 High: >10000 (>4 log) F2: S. exiguus MEI 20323 A. phoenicis MEI 14299

172 173 6 Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018

Table 1. (continued)

Inoculum cocktail Strain Level for each organism, CFU/g or mL Food matrix

G G1: Uninoculated Lettuce, mung bean S. aureus ATCC 27659 Low: 10–100 (1–2 log) sprout K. pneumoniae MEI 8927 Medium: 100–10000 (2–4 log) E. coli MEI 19505 High: >10000 (>4 log) G2: S. cerevisiae MEI 347 A. tamarii MEI 27259 H H1: Uninoculated Peanut butter, ice E. faecalis ATCC 29212 Low: 10–100 (1–2 log) cream, frost and topping mix E. coli MEI 27178 Medium: 100–10000 (2–4 log) C. koseri MEI 2640 High: >10000 (>4 log) H2: Cryptococcus humicola MEI 294 Fusarium incarnatum MEI 9663 I I1: Uninoculated Dry pet food, uncooked S. sciuri ATCC 29062 Low: 10–100 (1–2 log) noodles E. coli MEI 71259 Medium: 100–10000 (2–4 log) E. cloacae ATCC 13047 High: >10000 (>4 log) I2: C. membranaefaciens MEI 21462 Penicillium flavigenum MEI 31334 J J1: Uninoculated Wheat flour, black Streptococcus agalactiae ATCC 12386 Low: 10–100 (1–2 log) pepper, cumin E. coli MEI 79168 Medium: 100–10000 (2–4 log) E. sakazakii ATCC 51329 High: >10000 (>4 log) J2: C. lactis-condensi MEI 85392 P. glabrum MEI 27713

Three milliliters of bacterial suspension was aliquoted into enumerated, and aliquoted into sterile screw-capped tubes, and screw-cap sterile tubes and then incubated at 50 ± 0.5°C then incubated at 50 ± 0.5°C in a water bath– for 40 min. in a water bath–shaker for 10 to 25 min. After heat shock, The cell suspension was enumerated before and after heat stress cell concentration was estimated by plating in triplicate on to determine the degree of injury as above (Table 3). selective (Violet Red Bile Agar for coliforms, Baird Parker (c) Dry foods with lyophilized culture.—The lyophilization Agar for Staphylococcus aureus, and Slanetz and Bartley’s M medium was prepared by dissolving 10 g of skim milk and 5 g Enterococcus Agar for Enterococcus faecalis; 7) and nonselective of sodium glutamate in 80 mL distilled water and then made up agar and the degree of injury calculated (Table 2) according to to 100 mL, aliquoted in 5 mL portions into glass screw-capped Appendix J, Microbiology Guidelines, AOAC Official Methods tubes, and autoclaved at 115°C for 13 min. of AnalysisSM (6). Bacteria.—One colony from a 24-hr Trypticase Soy Agar Yeast and mold.—Yeasts and molds were revived from plate culture of each organism was grown in 30 mL of TSB frozen stocks and inoculated on SDA. Plates were incubated at at 35°C overnight (approximately 108 CFU/mL) and harvested 26 ± 1°C for 48–72 h. A cell suspension was then prepared by by centrifugation (× 3000 rpm, Beckman GS-6R, 30 min, 4°C), scraping colonies from the and resuspending in BPW, washed twice with sterile saline, and resuspended in 10 mL of

Table 2. Percent injury of heat-stressed bacteria used for inoculation of processed food matrixes

Strain Heat treatment, 50 ± 0.5°C Pre-heat stress, CFU/mLa Post-heat stress, CFU/mLb Injury, %

Klebsiella pneumoniae MEI 10625 12 min 3.63 × 108 1.67 × 108 53.99 Citrobacter koseri MEI 2640 10 min 3.90 × 107 1.47 × 107 62.31 E. coli MEI 7178 20 min 6.00 × 108 1.97 × 108 67.20 E. coli ATCC 43887 12 min 3.63E × 108 8.33 × 108 77.06 Staphylococcus aureus ATCC 27659 25 min 1.74 × 107 3.77 ×106 78.33 Enterococcus faecalis ATCC 29212 25 min 4.50 × 108 2.23 × 108 50.37

a Nonselective media = Tryptic Soy Agar (TSA). b Selective media = Violet Red Bile Agar (VRBA) for E. coli and other coliforms, Baird Parker Agar for S. aureus, M Enterococcus Agar for E. faecalis (7).

172 173 Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018 7

Table 3. Percent injury of heat-stressed fungi used for inoculation of processed food matrixes

CFU/mLa

Strain Heat treatment, 50 ± 0.5°C Pre-heat stress Post-heat stress Injury, %

Pichia fermentans MEI 27049-2 10 min 3.80 × 108 1.73 × 108 54.47 Cryptococcus humicola MEI 294 20 min 3.05 × 106 4.33 × 105 78.00 Aspergillus parasiticus MEI 27261-24 10 min 2.25 × 107 7.30 × 106 68.00 Fusarium incarnatus MEI 19663 10 min 2.40 × 108 8.60 × 107 64.20 a Sabouraud Dextrose Agar counts. the lyophilization medium (“High” inoculum). The Medium Product Consistency and Low inoculum levels were prepared by diluting the High stock in the lyophilization medium. The cell suspensions were Lot-to-lot study.—To assess product consistency, three lots/ aliquoted in 0.2 mL volumes into sterile ampules, frozen at batches of each type of Microfilm were tested using a cocktail of –70°C for 30 min, and then lyophilized following routine microorganisms consisting of a Gram-positive bacterium, E. coli, procedure for lyophilizing microbial cultures. non–E. coli coliform, a yeast, and a mold (Table 5) at the following Yeast and mold.—Yeast inoculum was prepared as described inoculum levels: Uninoculated (0), Low (1–2 log CFU/mL), above except the medium used for preparing the initial yeast and High (>4 log CFU/mL). One milliliter of inoculum in BPW suspension was prepared in Sabouraud Dextrose Broth was inoculated into each appropriate Microfilm and incubated incubated at 25°C. Mold was grown on SDA plates at 25°C for following the conditions specified for each type of Microfilm. 72 h. A stock spore suspension was prepared using 10 mL of Stability study.—For product stability, a single lot of each 12% skim milk and was used to prepare subsequent Medium type of Microfilm was incubated at different temperatures and Low inoculum levels, after which 0.2 mL aliquots in and tested at the following time points: 2–8°C (month 0, 3, ampules were lyophilized as above. 6, 12, and 18), 25°C (month 0, 2, 3, 4, 5, 6, 12, and 18), and Testing procedure.—The analysts performed the validation study according to the instructions outlined in the product insert Table 5. Microbial composition and levels of inoculum for and the reference method(s). Furthermore, the analysts were ruggedness, lot-to-lot consistency, and stability studies blinded from the sample IDs to eliminate bias. Ruggedness study Inoculum level, CFU/mL Two sets (per food matrix), previously artificially contaminated at different levels of inoculum and stabilized as described above, R1 (APC and TCC/ECC): were serially diluted in the appropriate diluent (Supplemental Enterococcus raffinosus MEI 28542 Uninoculated (0) Table 1), then 1 mL was applied on Microfilms and incubated. Citrobacter freundii MEI 8225 Medium: 100–10000 Colonies were then counted following the Instructions for Use E. coli ATCC BAA-1428 (2–4 log) (IFU), and the values were calculated and recorded. Digital R2: (YMC) pictures of the Microfilms were taken and archived. The same Candida krusei ATCC 6258 samples were tested with the corresponding reference ISO Aspergillus clavatus MEI 78201 methods (Table 4). Preparation of samples (50 g) and dilutions T1 (APC and TCC/ECC): were performed following ISO 6887 Parts 1–5, depending on Staphylococcus aureus ATCC 29247 Uninoculated (0) the food matrix (8–12). Media composition and preparation and Klebsiella pneumoniae MEI 10625 Medium: 100–10000 procedures for each ISO method were strictly followed. E. coli ATCC 43887 (2–4 log) T2 (APC and TCC/ECC): Table 4. Candidate method (Microfilm) and corresponding Pichia fermentans MEI 27049-2 ISO reference standard/method A. parasiticus MEI 27261-24 Lot-to-lot reproducibility Candidate method: Microfilm ISO reference standard/method L1: E. faecalis ATCC 19433 Uninoculated (0) ISO 4833-1:2013, Microbiology of the food K. oxytoca MEI 70309 Low: 10–100 (1–2 log) chain–Horizontal method for the enumeration of APC microorganisms–Part 1: Colony count at 30°C by E. coli ATCC 25922 High: >10000 (> 4 log) the pour plate technique L2: C. glabrata MEI 283 For E. coli: ISO 16649-2: 2001, Microbiology of food and animal feeding stuffs–Horizontal method A. niger ATCC 16404 for the enumeration of ß-glucuronidase-positive E. Stability study coli–Part 2: Colony count technique at 44ºC using TCEc 5-bromo-4-chloro-3-indolyl ß–D-glucuronide S1: S. aureus ATCC 51740 Uninoculated (0) For coliforms: ISO 4832:2006, Microbiology of food and animal feeding stuffs–Horizontal method for the E. coli ATCC 11840 Medium: 100–10,000 enumeration of coliforms–Colony count technique K. pneumoniae MEI 8326 (2–4 log) S2: ISO 21527-1:2008 and 21527-2:2008, Microbiology YMC of food and animal feeding stuffs–Horizontal Saccharomyces kudriavzevii MEI 28746 method for the enumeration of yeasts and molds A. awamori MEI 23114

174 175 8 Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018

at 45°C (week 0, 1, 2, 3, 4, 5, and 6). Five replicates each of Results and Discussion uninoculated BPW and BPW inoculated with Medium level (2–4 log CFU/mL) of a cocktail of microorganisms (Table 5) Method Developer Validation Studies were prepared, blind coded, and randomized for testing at the above time points using the appropriate Microfilm. Inclusivity and exclusivity studies (Microfilm TCEc and Ruggedness study.—The two parameters that were tested Microfilm YMC only).—Microfilm TCEc.—Of the 50 non- for this study were incubation temperature and incubation time E. coli coliform inclusivity strains tested (Supplemental (Table 5). A cocktail inoculum following Table 1 was prepared. Table 2), all were detected as bluish-green colonies on the Each parameter was tested in replicates of five per inoculum Microfilm TCEc. With regard to the E. coli inclusivity testing, level and tested with the appropriate Microfilm. all 52 E. coli strains showed violet colonies on the Microfilm TCEc (Supplemental Table 3). Data Analysis In terms of exclusivity testing, all 30 strains tested on Microfilm TCEc failed to grow or did not show visible growth Analysis was performed by manual counting of the colonies on Microfilm TCEc (Supplemental Table 4). as directed in the IFU (candidate method) and corresponding Microfilm YMC.—All of the fungal inclusivity strains ISO test reference method and ISO 7218:2007/Amd 1:2013(E): showed growth on Microfilm YMC, showing various colors Microbiology of food and animal feeding stuffs–General (Supplemental Table 5a). Yeast colonies were generally smaller requirements and guidance for microbiological examinations, than mold colonies and showed a defined edge, while most of including Amendment 1(13). On certain occasions, a hand-held the mold strains exhibited a diffused edge. A few of the strains magnifying lens was used in examining plates with pinpoint showed delayed growth of pronounced, countable colonies at colonies in pour-plate methods or with agar overlay in ISO methods 48 h (e.g., Aspergillus ruber, Eurotium chevalieri, Fusarium for APC, TCC, and E. coli counts. These values were recorded verticillioides, Wallemia sebi, Zygosaccharomyces bailii, and manually, transformed into logarithmic values, and analyzed for Z. rouxii), although faint growth was observed in most of the following: them. Further incubation to 60–72 h resulted in pronounced (a) Repeatability (Sr).—Standard deviation of replicates countable colonies for all these strains. None of the exclusivity for each analyte at each concentration of each matrix for each strains tested showed visible colonies on the Microfilm YMC method according to the following formula: (Supplemental Table 5b).

n Matrix study.—All results from the method developer’s 2 ∑()XXi − validation studies were considered in data and statistical i=1 Sr = analyses. Analyses [i.e., mean difference (log10), reverse n −1 transformed mean difference (CFU/g or CFU/mL), 95% CI, (b) Mean difference between candidate and reference and the standard deviation of repeatability (Sr) and relative methods.—Mean difference between candidate and reference standard deviation (RSD)] from method comparison between method transformed results with 95% confidence interval (CI) the Microfilm System and the corresponding ISO reference for each method at each inoculum level of each food matrix or methods on 20 food matrixes and environmental surfaces at three environmental surface. inoculum levels (per g) [Low (1–2 log), Medium (2–4 log), and (c) Mean difference of reverse transformed log10 values High (>4 log)] and two environmental surfaces at two inoculum between candidate and reference methods.—Mean difference of levels (Medium and High) were performed. A summary of the 2 reverse transformed log10 values between candidate and reference linearity R and P values from method comparison between the method with 95% CI for each method at each inoculum level for Microfilm Test System and the corresponding ISO standard each food matrix or environmental surface. methods are summarized in Table 6. Overall linear regression (d) Nested one-way analysis of variance (ANOVA).— graphs comparing the Microfilm and ISO standard methods are P value <0.05 indicates significance at the 95% confidence presented in Figure 2a–2e, while those for individual matrixes level. Statistical analysis was performed using a nested one-way are shown in Supplemental Figures S1–S5. ANOVA by F-test using Statistical Analysis System (SAS) v9.4 APC (Microfilm APC).—The mean difference (log10) for (SAS Institute, Cary, NC). the Microfilm APC across the 20 food matrixes ranged from (e) Linear regression analysis (Pearson’s) was also 0.0040 to 0.1183 at Low inoculum level (except for ground performed to determine the correlation coefficient (R2) of the pork 0.2669), from 0.0016 to 0.1663 at Medium inoculum Microfilms and the reference methods. level (except for ground pork –0.3987), and from 0.0010 to 0.1588 at High inoculum level (Supplemental Table 6). The Sr Independent Laboratory Method Validation Study for Microfilm APC within each inoculum level (including the uninoculated samples) ranged from 0.0000 to 0.6549, while the Four food matrixes (almonds; RTE lunch meat; surimi; reference method showed an Sr ranging from 0.0000 to 0.5730 and pasteurized milk, 2% fat) and one environmental surface (Supplemental Table 8). The RSD of both the Microfilm APC (stainless steel) were inoculated by the method developer and the ISO reference method were close to each other and in following the scheme described in the method comparison the majority of cases were well below 10%. The highest RSD study. The food and surface swab samples were coded, values for Microfilm/ISO were noted for uninoculated samples randomized, and delivered to the School of Public Health of ice cream (60.1/57.2), uncooked noodles (136.9/136.9), Laboratory, University of Washington, for testing. Results from peanut butter (12.7/12.56), top frosting (138.5/136.9), and the independent laboratory study were submitted to the AOAC almond (15.5/9.01) at Low inoculum level (Supplemental upon completion of the study and decoding of samples. Table 8). The Microfilm APC RSD and ISO 4833-1 RSD were

174 175 Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018 9

Table 6. Summary of linearity correlation coefficient (R2) and P valuesa from method comparison between the Microfilm System and the applicable ISO reference methods performed by the method developer and the independent laboratory

Method developer

Microfilm TCEc Microfilm TCEc Microfilm YMC Microfilm YMC Microfilm APC TCC ECC yeast counts mold counts

Matrix/ surface R2 P R2 P R2 P R2 P R2 P

Almond 0.9929 0.9919 0.9949 0.9735 0.9955 0.9683 0.9711 0.9627 0.9700 0.9884 Black pepper 0.9926 0.9972 0.9919 0.8704 0.9835 0.8704 0.9980 0.9569 0.9963 0.9802 Cumin 0.9988 0.9968 0.9492 0.8454 0.9145 0.7248 0.9980 0.9711 0.9932 0.9711 Dry pet food 0.9970 0.9998 0.9963 0.9996 0.9925 0.9649 0.9978 0.9939 0.9900 0.9604 Ground beef, 7% fat 0.9970 0.9844 0.9972 0.9699 0.9942 0.9839 0.9985 0.9917 0.9987 0.9945 Ground chicken, 2% fat 0.9888 0.7990 0.9981 0.9821 0.9949 0.9563 0.9825 0.9203 0.9959 0.9799 Ground pork, 20% fat 0.9167 0.4680 0.9895 0.8915 0.9979 0.9847 0.9957 0.9615 0.9976 0.9694 Ground turkey, 1% fat 0.9826 0.9957 0.9933 0.9607 0.9930 0.8913 0.9721 0.9799 0.9992 0.9879 Ice cream 0.9949 0.9521 0.9979 0.9933 0.9980 0.9815 0.9947 0.9721 0.9943 0.9858 Milk 0.9997 0.9960 0.9960 0.9951 0.9946 0.9242 0.9966 0.9690 0.9964 0.9653 Mung bean sprouts 0.9160 0.9385 0.9331 0.7261 0.9989 0.9649 0.9592 0.9124 0.9946 0.9769 Noodles, uncooked 0.9988 0.9804 0.9974 0.9678 0.9988 0.9873 0.9955 0.9632 0.9947 0.9960 Peanut butter 0.9973 0.9720 0.9983 0.9672 0.9981 0.9470 0.9970 0.9924 0.9967 0.9722 Romaine lettuce 0.9304 0.8021 0.9881 0.9901 0.9950 0.9665 0.9729 0.8507 0.9701 0.9318 RTE lunch meat 0.9997 0.9916 0.9994 0.9936 0.9954 0.9493 0.9976 0.9600 0.9927 0.9695 Shrimp 0.9992 0.9979 0.9982 0.9924 0.9993 0.9501 0.9946 0.9848 0.9855 0.9910 Spinach 0.9352 0.7777 0.9937 0.9554 0.9983 0.9651 0.9879 0.9149 0.9976 0.9540 Surimi 0.9987 0.9947 0.9978 0.9797 0.9941 0.8796 0.9982 0.9744 0.9986 0.9809 Top frosting 0.9989 0.9733 0.9967 0.9854 0.9975 0.9304 0.9949 0.9720 0.9969 0.9715 Wheat flour 0.9668 0.9888 0.9885 0.9799 0.9915 0.9073 0.9794 0.9743 0.9914 0.9552 Plastic 0.9996 0.9837 0.9991 0.9544 0.9985 0.9544 0.9985 0.9715 0.9959 0.9561 Stainless steel 0.9991 0.5715 0.9985 0.9966 0.9986 0.9296 0.9977 0.9501 0.9979 0.9789

Independent laboratory

R2 P R2 P R2 P R2 P R2 P

Almond 0.9845 0.9589 0.9075 0.9528 0.9593 0.9151 0.8962 0.9307 0.8932 0.8448 Milk, 2% fat 0.9968 0.9975 0.9971 0.983 0.9919 0.9749 0.9954 0.9777 0.9947 0.9544 RTE lunch meat 0.9868 0.9968 0.9988 0.9917 0.989 0.9409 0.9924 0.9007 0.9969 0.9687 Surimi 0.999 0.996 0.9986 0.9817 0.9963 0.8656 0.9487 0.8898 0.9972 0.9902 Steel 0.9794 0.8728 0.9651 0.9682 0.9979 0.9413 0.9952 0.9467 0.9986 0.9716 a P value <0.05 by nested one-way ANOVA is significant at 95% confidence.

still remarkably similar, and the higher values were attributed to samples with Low, Medium, or High inoculum levels. The mean a statistical artifact in calculating RSDs in uninoculated samples log10 counts of aerobic bacteria (APC) showed no significant where the background flora was near the limit of detection and differences between the candidate method (Microfilm APC) Poisson distribution effects lead to larger variations among and the reference method (ISO 4833-1) in all 20 matrixes tested replicates. (Table 6) and high overall correlation coefficient (R2 = 0.996; The Microfilm APC showed excellent correlation with the Figure 2a). reference ISO method with R2 values ≥0.98, except for ground TCC (Microfilm TCEc).—With respect to TCC, the mean 2 2 pork (R = 0.9167), mung bean sprouts (R = 0.9160), spinach difference in log10 counts between Microfilm TCEc and (R2 = 0.9352), and romaine lettuce (R2 = 0.9304; Supplemental ISO 4832 ranged from 0.0001 to 0.1247 and from 0.0073 to Figure S1), which are still considered as indicative of a strong 0.1453 for Medium and High inoculum levels, respectively correlation between the two methods (14). The relatively lower (Supplemental Table 10). With the Low inoculum level, 2 correlation (R <0.98) between the Microfilm APC and ISO however, the differences in mean log10 counts had outliers: method in these food samples may be explained by the high ground pork (–0.3508) and cumin (1.1932). The Sr for background microbial flora (approximately 6 logs), hence no Microfilm TCEc coliform counts ranged from 0.000 to 0.5372 distinctions could be made between uninoculated or inoculated (Supplemental Table 12) and from 0.0000 to 0.5069 for the

176 177 10 Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018

(a) (b) (c)

(d) (e)

Figure 2. Overall linear regression graphs comparing the Microfilm and ISO standard methods are shown for (a) APC, (b) TCC, (c) ECC, (d) yeast, and (e) mold.

reference method. The RSDs across the matrixes were around (13.42/3.66), ground turkey (6.13/10.64), milk (13.31/6.64), 10%, with the highest RSD values for Microfilm/ISO noted romaine lettuce (16.53/3.85), spinach (14.68/19.29), top frosting for uninoculated samples of ground chicken (18.2/15.3), (10.47/5.81), and wheat flour (11.16/28.34; Supplemental Table ground turkey (23.8/12.2), spinach (23.8/16.2), and wheat 16). There was no notable trend on the RSD of the Microfilm or flour (34.6/22.5), and Low-level almond, cumin, and milk with the ISO 16649-2. In cumin, however, recovery of colonies (Microfilm/ISO RSD values of 12.7/15.9, 23.0/139.3, and was observed with Microfilm but not with the ISO method in 11.8/15.8, respectively). The higher RSDs for Microfilm and Low-level-inoculated samples. Consequently, because the counts the ISO method in uninoculated samples may be attributed to a were at the limit of detection, reproducibility for Microfilm was statistical artifact in calculating RSDs in uninoculated samples, artificially low, in contrast to the absence of growth (<10 CFU/g) where the background flora was near the limit of detection and in all replicates for the ISO method, thus giving a seemingly Poisson distribution effects lead to larger variations among good reproducibility. replicates. In the case of Low-level-inoculated samples, the One-way nested ANOVA gave P values >0.05 for all the Microfilm TCC RSDs were lower than that of the ISO RSDs, matrixes tested, indicating no significant differences between the indicating better reproducibility for the Microfilm TCC. Microfilm TCEcE. coli counts and ISO 16649-2 E. coli counts. Nested one-way ANOVA showed no significant differences Similarly, the counts between the two showed high correlation (P >0.05) between the Microfilm TCEc and ISO 4832 average coefficients, with R2 values ranging from 0.91 (cumin) to ≥0.98 mean difference log counts in all 20 food matrixes tested for the remaining matrixes (Table 6) and an aggregate R2 = (Table 6). In addition, the counts from both methods showed 0.988 (Figure 2c). The relatively low R2 value for cumin may excellent correlation with R2 values ranging from 0.93 for mung be attributed to the higher coliform counts (by approximately bean sprouts to ≥0.99 (Supplemental Figure S2) for most of the one log) obtained with Microfilm TCEc than the ISO method. remaining matrixes, with an overall R2 = 0.992 (Figure 2b). This suggests that the Microfilm TCEc offers a more favorable E. coli count (Microfilm TCEc).—With respect to E. coli medium for E. coli (and coliforms) than the ISO method, hence enumeration, the log10 mean difference on counts from the its higher recovery of coliforms in this particular matrix. Microfilm TCEc and ISO 16649-2 plates showed a wide range, Yeast count (Microfilm YMC).—With respect to yeast depending on the inoculum level, with the widest range at the enumeration, the mean log10 differences between the Microfilm Low inoculum level, i.e., from –0.0047 to 0.4072 with an outlier YMC and the reference method (ISO 21527) counts ranged from of 1.5416 (Supplemental Table 14). With Medium and High 0.0058 to 0.1366 for the Low inoculum level, from –0.0087 to inoculum levels, the mean differences ranged from 0.0096 to 0.2586 for the Medium inoculum level, and from –0.0018 to 0.3154 and from 0.0224 to 0.3757, respectively. The Sr across 0.1962 for the High inoculum level (Supplemental Table 18) inoculum levels ranged from 0.0000 to 0.4827 for Microfilm across all matrixes. The Sr for Microfilm YMC yeast counts TCEc and from 0.0000 to 0.5209 for ISO 16649-2 (Supplemental ranged from 0.0000 to 0.8669 and from 0.0000 to 0.6849 for the Table 16). The RSD values for both Microfilm and the ISO ISO reference method (Supplemental Table 20). The RSD values standard method for E. coli were around 10%. The highest values for both Microfilm YMC and ISO standard method were mostly of Microfilm/ISO 16649-2 RSD were noted for uninoculated around 10% or below. The highest RSDs for Microfilm YMC/ ground turkey (15.86/11.37) and Low-level-inoculated ISO 21527 were noted for uninoculated ground pork (23.78/20), samples of almond (6.99/15.10), black pepper (12.18/17.76), ground turkey (223.61/223.61), shrimp (15.08/19.07), and cumin (22.50/NA), dry pet food (25.35/17.91), ground beef wheat flour (97.60/61.85), and Low-level-inoculated samples

176 177 Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018 11 of almond (62.25/13.16), ground pork (16.32/16.37), uncooked between Microfilm TCEc and ISO 4832 at both the Medium noodles (15.26/17.59), romaine lettuce (4.12/10.21), and wheat and High inoculum levels ranged from –0.0023 to 0.0610 flour (13.86/9.03; Supplemental Table 20). (Supplemental Table 11). The Sr for Microfilm TCEc coliform One-way nested ANOVA showed no significant differences counts at each inoculum level ranged from 0.0420 to 0.1566, between the Microfilm YMC and ISO 21527 Part 1 or Part 2 while the reference method had Sr values ranging from 0.0467 average log yeast counts in all 20 food matrixes tested (Table 6). to 0.1626 (Supplemental Table 12). RSDs were below 3% for A high degree of correlation, i.e., an aggregate R2 = 0.991, was both Microfilm TCEc and ISO Standard 4832 (Supplemental observed for all food matrixes (Figure 2d). Table 12). Mold count (Microfilm YMC).—With respect to mold Nested one-way ANOVA showed no significant differences enumeration, the mean log10 differences between the Microfilm between the Microfilm TCEc and ISO 4832 average log counts YMC and the reference method (ISO 21527) counts were in both plastic and stainless steel swab samples tested (Table 6). within a tight range (≤0.2 log), regardless of inoculum level and In addition, TCCs from both methods showed excellent matrix, e.g., from 0.0068 to 0.2066 for the Low level (except for correlation with R2 values ≥0.99. an outlier for almond, –0.4855), from –0.0079 to 0.2154 for the E. coli count (Microfilm TCEc).—With respect to E. coli Medium level, and from –0.0009 to 0.2932 for the High level enumeration, the log10 mean difference on counts from the (Supplemental Tables 22 and 23). The Sr ranged from 0.0000 Microfilm TCEc and ISO 16649-2 plates ranged from 0.1709 to 0.9314 for Microfilm YMC mold counts and from 0.0000 to to 0.3154 (Supplemental Table 15). The Sr for the Microfilm 0.8761 for the ISO reference method (Supplemental Table 24). TCEc for E. coli counts ranged from 0.0457 to 0.1970, while the The RSDs for both Microfilm YMC and ISO 21527 were around Sr values for the ISO reference method ranged from 0.0436 to 10% or below. The highest RSDs for mold counts by Microfilm 0.2283 (Supplemental Table 16). The Microfilm TCEc E. coli YMC/ISO 21527 were noted for uninoculated dry pet food counts showed high correlation with the ISO reference method (138.53/136.93), ground pork (223.61/223.61), romaine lettuce with R2 value ≥0.99. RSD values for both methods were <5% (94.10/57.84), shrimp [not applicable (NA)/223.61], spinach (Supplemental Table 16). (24.31/25.10), and wheat flour (97.60/61.85), and Low-level- Nested one-way ANOVA showed no significant differences inoculated samples of almond (56.60/13.10), dry pet food between the Microfilm TCEc E. coli counts and ISO 16649-2 (13.72/7.92), ground pork (11.59/8.34), mung bean sprouts counts on the plastic and stainless steel swab samples inoculated (14.10/15.74), uncooked noodles (11.46/8.67), romaine lettuce with Medium and High inoculum levels (Table 6). (6.85/10.30), RTE lunch meat (13.05/5.83), and Medium-level Yeast count (Microfilm YMC).—With respect to yeast wheat flour (15.95/18.97; Supplemental Table 24). It should enumeration, the mean log10 differences between the Microfilm be noted that the high RSDs for both the Microfilm and ISO YMC and the reference method (ISO 21527) counts ranged 21527 were observed in uninoculated samples that had low from 0.0770 to 0.1876 for both surfaces inoculated with levels of mold contamination. Because of Poisson distribution Medium and High inoculum levels (Supplemental Table 19). effects, some replicates may not contain the target organism; The Sr for Microfilm YMC yeast counts ranged from 0.0684 to consequently, low reproducibility of replicates is expected. 0.1596, while the ISO reference method had Sr values ranging One-way nested ANOVA showed no significant differences from 0.0776 to 0.1286 on plastic and stainless steel surfaces between the Microfilm YMC and ISO 21527 Part 1 or Part 2 (Supplemental Table 20). RSD values for both methods were average log counts for molds in all 20 food matrixes and around 6% (Supplemental Table 20). environmental surfaces tested (Table 6), and a high degree of One-way nested ANOVA showed no significant differences correlation, i.e., R2 ≥ 0.98, was observed in all food matrixes between the Microfilm YMC and ISO 21527 Part 1 or Part 2 except for romaine lettuce (R2 = 0.97; Figure 2e), which is still average log yeast counts in both environmental surfaces indicative of a strong correlation between the Microfilm and the (Table 6). A high degree of correlation, i.e., R2 ≥ 0.99, was ISO reference method (14). observed in yeast counts from Microfilm YMC and the reference Environmental swab testing study.—APC (Microfilm method (ISO 21527). APC).—The mean difference (log10), reverse transformed Mold count (Microfilm YMC).—With respect to mold mean difference (CFU/mL), and 95% CI for the Microfilm APC enumeration on the two environmental surfaces tested, the on the two environmental swab samples (plastic and stainless mean log10 differences between the Microfilm YMC and the steel) were below 0.2 log and ranged from –0.0227 to 0.1154 (at reference method (ISO 21527) counts ranged from 0.0200 both inoculum levels tested, Medium and High; Supplemental to 0.2148 for stainless steel and from –1.3187 to –1.6360 for Table 7). The Sr values ranged from 0.0000 to 0.1571 for the plastic (Supplemental Table 23). In terms of reproducibility, the Microfilm APC and from 0.0000 to 0.2172 for the reference Sr ranged from 0.0694 to 0.1173 and from 0.0849 to 0.0955 for method (Supplemental Table 8). The RSDs for environmental Microfilm YMC and ISO 21527, respectively. The RSD values samples were below 1% (Supplemental Table 7). In terms of for both methods were around 5% or below for both plastic and the correlation coefficient between the Microfilm APC and stainless steel swabs (Supplemental Table 24). reference ISO method, the aggregate R2 was ≥0.99. One-way nested ANOVA showed no significant differences Statistical analysis by nested one-way ANOVA of mean between the Microfilm YMC and ISO 21527 Part 1 average log10 counts of aerobic bacteria (APC) showed no significant log10 counts for molds in both surfaces tested. Furthermore, a differences between the candidate method (Microfilm APC) and high degree of correlation between the Microfilm YMC and the reference method (ISO 4833-1) on the two environmental ISO 21527 mold counts, i.e., R2 ≥ 0.99, was observed for both surfaces tested (Table 6). surfaces (Table 6). TCC (Microfilm TCEc).—With respect to TCC on the same Product consistency.—Lot-to-lot study.—To assess product environmental samples, the mean difference in log10 counts consistency, three lots/batches of each type of Microfilm were

178 179 12 Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018

tested. The mean differences in log counts among the three lots statistically significant difference was observed at 33°C for of each type of Microfilm ranged from 0.0196 to 0.0098 (Low Microfilm APC, but the mean difference between 10 thelog level) and from –0.0009 to 0.0018 (High Level) for Microfilm count and the overall mean log difference in count (–0.0266) APC, from –0.0051 to 0.0306 (Low level) and from –0.0095 was below the limit for practical difference of 0.20 (15). Thus, to 0.0073 (High level) for Microfilm TCEc (TCCs), from results from the ruggedness studies indicate that the Microfilm –0.0303 to 0.0719 (Low level) and from –0.0071 to 0.0543 Test System provides valid results at the temperatures specified (High level) for Microfilm TCEc E.( coli counts), from –0.0718 for each type of Microfilm. to 0.0434 (Low level) and from –0.0113 to 0.0084 (High level) for Microfilm YMC yeast counts, and from –0.0995 to 0.0838 (Low level) and from –0.0647 to 0.0395 (High level) Independent Laboratory Method Validation Study for Microfilm YMC mold counts (Supplemental Table 26a and 26b). Significant differences were observed between Lot 1 The independent laboratory method validation study was on the Low-level inoculum mean log yeast counts and Lot 3 performed by Scott Meschke’s laboratory, School of Public Health, for Microfilm YMC mold counts; however, the mean log University of Washington, according to a study design provided differences in counts between these lots and the overall mean by the AOAC RI. Four food matrixes and an environmental were below 0.20, the limit for practical difference (15). The Sr swab sample spiked with a panel of microorganisms (one Gram- within each lot and between lots ranged from 0.02 to 0.07 for positive, a non–E. coli coliform bacterium, and E. coli in one Microfilm APC, from 0.02 to 0.12 for Microfilm TCEc TCCs, cocktail, and a yeast and a mold strain in a second cocktail) from 0.05 to 0.20 for Microfilm TCEc E. coli counts, from following the procedure outlined in the Independent Laboratory 0.01 to 0.07 for Microfilm YMC yeast counts, and from 0.04 Method Validations Study in the Methods section of this report to 0.07 for Microfilm YMC mold counts. These results also were tested in five replicates per inoculum level. The matrixes indicate good reproducibility among the three lots of each type tested included almond, surimi, pasteurized milk (2% fat), of Microfilm. and RTE lunch meat at three inoculum levels [Low (1–2 log), Stability study.—Results from real-time stability testing at Medium (2–4 log), and High (>4 log)] and stainless steel surface 2–8°C and at 25°C showed that all three types of Microfilms swab samples inoculated at Medium and High levels. All test were stable for 1 year and at 45°C for 6 weeks (Supplemental materials (uninoculated and inoculated) were blind-coded and Table 27a–c), i.e., mean log counts between time points were provided by Microbiologique to the independent laboratory. The not statistically significant for Microfilm APC or Microfilm operators were trained on the candidate and reference methods as TCEc for TCCs and E. coli counts. However, statistical described in the Method section above. significance was observed for Microfilm yeast counts, but closer APC (MicrofilmAPC).—The mean difference (log10), for examination of the mean log10 difference (overall mean minus the Microfilm APC and the ISO 4833 APC counts in the food the individual means for each time point) showed that the values matrixes tested ranged from 0.0000 to –0.2179 (Supplemental were mostly below 0.2, the limit of practical difference (15), Table 6) and from –0.4609 to –0.7834 for stainless steel samples except at Week 2 and Week 4 (–0.25. and 0.29, respectively; (Supplemental Table 7). The Sr for Microfilm APC across the Supplemental Table 27b). These two values may simply reflect matrixes/surface tested ranged from 0.0000 to 0.4564, with the the natural variability associated with the use of a range of fresh higher variation observed in almonds. Similarly, a wide range inoculum (2–4 logs) prepared at each time point that could lead of standard deviation was also observed in almonds with the to inherent variation. reference method, i.e., from 0.0000 to 0.5292 (Supplemental The interim 12-month data from real-time stability testing at Table 9). The RSD values for Microfilm APC and ISO 4833 2–8°C and 25°C indicate that all average counts at each time were around 5% except in almonds, which had RSD values point across all types of Microfilms were within the inoculum of 25.10/10.65, 18.43/65.47, 13.94/7.9984 and 8.26/9.63 in range (2–4 logs) as determined by the appropriate ISO method. uninoculated, and Low, Medium, and High inoculum levels, In addition, the reproducibility at each time point was high as respectively (Supplemental Table 9). When the mean log indicated by the low Sr values at each time point. Based on these differences were examined, the values for the inoculated samples data, the Microfilm Test System is assigned a 1 year shelf life were not practically different, i.e., below 0.2. In addition, the R2 at 2–25°C. values between the Microfilm APC and reference ISO method Ruggedness study.—With respect to incubation time, on all five matrixes were ≥0.98 (Supplemental Figure S6), the counts remained the same at all incubation times tested which indicate strong correlation between both methods (14). regardless of the type of Microfilm, i.e., 22 ± 0.5, 23 ± 0.5, Statistical analysis by nested one-way ANOVA of mean 25 ± 0.5, 26 ± 0.5, and 27 ± 0.5 for Microfilm APC and log10 counts of aerobic bacteria (APC) showed no significant Microfilm TCEc, and 46 ± 0.5, 47 ± 0.5, 46 ± 0.5, 47 ± 0.5, differences between the Microfilm APC and the reference 46 ± 0.5, 47 ± 0.5, and 46 ± 0.5, 47 ± 0.5 for Microfilm YMC method (ISO 4833-1) in all four matrixes tested and stainless (Supplemental Table 28a and 28b). steel (Table 6). With varying incubation temperature, the Microfilm TCEc TCC (Microfilm TCEc).—With respect to TCC on the same (both total coliform and E. coli counts) and Microfilm YMC set of food matrixes and environmental sample, the mean (both yeast and mold counts) did not show statistically difference in log10 counts between Microfilm TCEc and ISO significant differences in counts at all the temperatures tested, 4832 ranged from 0.0001 to –0.3226 (Supplemental Tables 10 i.e., 33 ± 1.0, 34 ± 1.0, 35 ± 1.0, 36 ± 1.0, and 37 ± 1.0 for and 11). The Sr for Microfilm TCEc coliform counts ranged from Microfilm TCEc (Supplemental Table 28c) and 24 ± 1.0, 0.0000 to 0.8383 for all the matrixes tested and stainless steel. 27 ± 1.0, and 29 ± 1.0 for Microfilm YMC (Supplemental In comparison, the Sr of the reference method showed a similar Table 28d). With regard to Microfilm APC, however, a range, i.e., from 0.0000 to 0.7387 (Supplemental Table 13). The

178 179 Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018 13

RSD values of the Microfilm/ISO method were mostly below 5%, In addition, a high degree of correlation, i.e., R2 ≥ 0.99, was except for almond (Low, Medium, and High inoculum levels) observed in all food matrixes except for almond (R2 = 0.8962) with RSD values of 65.9492/56.0786, 12.6739/12.3518, and and surimi (R2 = 0.9487; Supplemental Figure S9). The R2 11.0707/11.2784, respectively (Supplemental Table 13). A higher values for almond and surimi are still considered indicative of RSD value for Microfilm (9.6906) was also noted in Low-level- good correlation between the two methods (14). inoculated milk compared to the ISO method with an RSD value Mold count (Microfilm YMC).—With respect to mold of 4.7059. However, when mean log differences between the enumeration, the mean log10 differences between the Microfilm two methods were compared, the values for inoculated almonds YMC and the reference method (ISO 21527) counts ranged were –0.0462, 0.0070, and –0.0221 for Low, Medium, and High from 0.0000 to –0.5351 (Supplemental Tables 22 and 23), while inoculum levels, respectively, and –0.1136 for milk, which are the Sr ranged from 0.0000 to 0.5243 (Supplemental Table 25). all below the limit of practical difference (15). Similarly, the Sr for the reference method ranged from 0.0000 No statistically significant differences were observed between to 0.7213 (Supplemental Table 25). The RSD values for both the Microfilm TCEc and ISO 4832 average log counts in all the methods were below 10%, except in almonds, where the RSD food matrixes tested and stainless steel. In addition, the counts values for the uninoculated and Low level for the ISO method from both methods showed strong correlation coefficients with were high at 223.61% and 55.9%, respectively (Supplemental 2 2 R values ≥0.96, except for almonds, which showed an R value Table 25). The high RSDs of the ISO method may be attributed to of 0.9075, which is still considered as indicative of a good the presence of the target organism at the limit of detection, and correlation (14) between the Microfilm and the ISO methods Poisson distribution effects could lead to low reproducibility. (Supplemental Figure S7, Table 6). Additionally, the ISO plates showed spreading mycelial growth, E. coli count (Microfilm TCEc).—With respect to E. coli hence accurate counting proved to be difficult. enumeration, the mean difference between the Microfilm TCEc No significant differences were found between the Microfilm and ISO 16649-2 E. coli counts ranged from 0.0000 to 0.3757 YMC and ISO 21527 Part 1 or Part 2 average log counts for (Supplemental Table 14), while the Sr ranged from 0.0000 to molds in the four food matrixes and environmental swab 0.8107 (Supplemental Table 17), with higher Sr values observed samples (stainless steel) tested (Table 6). A high degree of with almonds. In comparison, the reference method showed a correlation, i.e., R2 ≥ 0.99, was observed in all food matrixes 2 similar Sr range, i.e., from 0.0000 to 0.7014, with almonds showing except for almonds (R = 0.8932; Supplemental Figure S10), the highest Sr values. The RSD values for both methods were although this value is still considered as indicative of good below 5%, except in almonds with Microfilm/ISO RSD values correlation between the two methods (14). of 65.12/81.68, 12.68/13.20, and 10.57/11.11 for Low, Medium, and High, respectively, and milk (15.62/4.89; Supplemental Conclusions Table 17). However, the mean log differences for these samples were below 0.2, indicating no practical difference (15) between Results from both the method developer and independent results obtained with the Microfilm and ISO methods. validation studies indicate that the Microfilm Test System No significant differences were observed between the showed no significant differences in performance compared Microfilm TCEc E. coli counts and ISO 16649-2 counts on to the reference ISO standard methods in all the food matrixes all five samples tested (Table 6). In addition, a high degree of and environmental surfaces tested. In addition, results from correlation between the results from the Microfilm TCEc and the method developer’s validation studies showed that the ISO method, i.e., R2 ≥ 0.96, was observed in all matrixes tested Microfilm Test System is robust, reproducible from lot to (Supplemental Figure S8). lot across all types of Microfilms, and stable at the storage Yeast count (Microfilm YMC).—With respect to yeast temperatures tested. Real-time stability data indicate that enumeration by the independent laboratory, the mean log 10 the Microfilm Test System is stable for 12 months at 2–25°C differences between the Microfilm YMC and the reference storage and for 6 weeks when exposed to a higher temperature method (ISO 21527) counts ranged from 0.0000 to –0.5716 (45°C). (Supplemental Tables 18 and 19). The Sr for Microfilm YMC yeast counts ranged from 0.0000 to 0.7864 (Supplemental Acknowledgments Table 21). In comparison, the Sr values for the reference method ranged from 0.0000 to 0.2387 (Supplemental Table 21). The We would like to express our gratitude to Scott Meschke, RSD for both methods were below 10%, except for almonds Nicola Beck, and Alexandra Kossikat (University of Washington, (uninoculated and Low inoculum level with Microfilm/ISO School of Public Health, Seattle, WA) for their professional RSD values of 223.6068/NA and 55.9372/NA, respectively) work on the independent laboratory validation study. and uninoculated surimi for the Microfilm method (Microfilm/ ISO RSD values of 138.1050/NA; Supplemental Table 21). The higher RSDs obtained with Microfilm for these samples were References most likely due to recovery of yeasts in some of the replicates at the limit of detection. A similar observation was noted by the (1) ISO 4833-1:2013(E): Microbiology of the food chain–Horizontal method for the enumeration of microorganisms–Part 1: Colony method developer in yeast enumeration, where no counts were count at 30°C by the pour plate technique, International obtained from the ISO method, indicating that the Microfilm Organization for Standardization, Geneva, Switzerland provides a medium for better recovery of yeasts. (2) ISO 4832:2006(E): Microbiology of food and animal feeding No significant differences were observed between the stuffs–Horizontal method for the enumeration of coliforms– Microfilm YMC and ISO 21527 Part 1 or Part 2 average log yeast Colony count technique, International Organization for counts in all food matrixes tested and stainless steel (Table 6). Standardization, Geneva, Switzerland

180 181 14 Mai et al.: Journal of aoaC international Vol. 101, no. x, 2018

(3) ISO 16649-2:2001(E): Microbiology of food and animal (10) ISO 6887-3:2003: Microbiology of food and animal feeding feeding stuffs–Horizontal method for the enumeration of stuffs–Preparation of test samples, initial suspension and ß-glucuronidase -positive E. coli–Part 2: Colony count decimal dilutions for microbiological examination–Part 3: technique at 44°C using 5-bromo-4-chloro-3-indolyl ß-D- Specific rules for the preparation of fish and fishery products, glucuronide, International Organization for Standardization, International Organization for Standardization, Geneva, Geneva, Switzerland Switzerland (4) ISO 21527-1:2008(E): Microbiology of food and animal feeding (11) ISO 6887-4:2003: Microbiology of food and animal feeding stuffs–Horizontal method for the enumeration of yeasts and stuffs–Preparation of test samples, initial suspension and moulds–Part 1: Colony count technique in products with water decimal dilutions for microbiological examination–Part 4: activity greater than 0, 95, International Organization for Specific rules for the preparation of products other than milk Standardization, Geneva, Switzerland and milk products, meat and meat products, and fish and fishery (5) ISO 21527-2:2008(E): Microbiology of food and animal feeding products, International Organization for Standardization, stuffs–Horizontal method for the enumeration of yeasts and Geneva, Switzerland moulds–Part 2: Colony count technique in products with water (12) ISO 6887-5:2003: Microbiology of food and animal feeding activity less than or equal to 0, 95, International Organization stuffs–Preparation of test samples, initial suspension and for Standardization, Geneva, Switzerland decimal dilutions for microbiological examination–Part 5: (6) AOAC INTERNATIONAL Methods Committee Guidelines Specific rules for the preparation of milk and milk products, for Validation of Microbiological Methods for Food and International Organization for Standardization, Geneva, Environmental Surfaces, Appendix J. 2012, www.eoma.aoac. Switzerland org/app_j.pdf (13) ISO 7218:2007: Microbiology of food and animal (7) Slanetz, L.W., & Bartley, C.H. (1957) J. Bacteriol. 74, 591–595 feeding stuffs–General requirements and guidance for (8) ISO 6887-1:1999: Microbiology of food and animal feeding microbiological examinations, including Amendment 1:2013, stuffs–Preparation of test samples, initial suspension and International Organization for Standardization, Geneva, decimal dilutions for microbiological examination–Part 1: Switzerland General rules for the preparation of the initial suspension (14) Ott, R.L., & Longnecker, M. (2001) An Introduction to and decimal dilutions, International Organization for Statistical Methods and Data Analysis, 5th Edition. Duxbury Standardization, Geneva, Switzerland Press, Pacific Grove, CA (9) ISO 6887-2:2003: Microbiology of food and animal feeding (15) AOAC Research Institute (2014) Comparative Evaluation of the stuffs–Preparation of test samples, initial suspension and decimal 3M  Rapid Aerobic Count Plate for the Enumeration of dilutions for microbiological examination–Part 2 Specific rules Total Viable Count in a Variety of Foods (Part I) and Robustness for the preparation of meat and meat products, International Study (Part II), https://multimedia.3m.com/mws/media/1040389O/ Organization for Standardization, Geneva, Switzerland petrifilm-rapid-aerobic-count-plate-q-labs-study.pdf

180 181