1. Background: Root Causes and Sources of Hazards from Food Allergens The control of allergens is a significant concern for food manufacturers. However, the absence of universally agreed acceptable allergen levels has led to the overuse of fail-safe warning or pre-cautionary labelling. This lack of guidance is causing a great deal of confusion on the best approach to control allergen risks. It is estimated that allergen associated incidents cause 30,000 individuals to seek emergency treatment and 150 200 deaths annually in the USA. There are 160 potentially allergenic foods but only 8 (5%) give rise to 90% of allergic reactions. According to the FDA Food Allergen Coordinator, Dr Steven M. Gendel, and his team 1 the FDA Recall Food Registry (RFR) shows that 34% of all recalls in USA were due to undeclared allergens (figure 1). The proportion of all recalls due to undeclared allergens increased from 25% in 1999 to 40% in 2012. Figure 1. Causes for food product recalls 1 FDA Recall Enterprise System (RES) analyze recall data to produce trend data and identify root causes that can be summarized as follows; Most product recalls involve bakery and snack type products Milk, wheat and soy were the most common adulterants Errors in packaging were the most common root causes Table 1. Allergen recalls 1 Data analysis showed that Five food types, bakery products, snack foods, candy, dairy products and dressings, were most often involved in food allergen recalls during this period. Bakery products accounted for almost as many food allergen recalls as all of the other top five foods combined (table 1). Table 2. Root causes for allergen recalls 1 The allergens most often involved in recalls were milk, wheat and soy. Peanuts and tree nuts combined caused fewer recalls than any one of these top three allergens. 1 Gendel et al.: Learning from the FDA Food Allergen Recalls and Reportable Foods; (04/2014) Food Safety Magazine 1
Just over 20 percent of the recalls involved mislabeling for multiple allergens; many involved a combination of milk, wheat, soy and egg. This may reflect the many different ways that these foods are used and the variety of different ingredients that are derived from each of them. 88% of root causes were related to labelling issue (table 2), and only 12% of root causes were due to some form of cross contact in manufacturing of which there are many possible causes, only one of which is cleaning. Gendel states that 13 distinct root causes of recalls were identified, all of which were related to packaging, primarily o packaging confusion in manufacturing; poor differentiation between stock items o wrong terminology used to describe the contents giving misleading information o incorrect information due to failed carry-over of information relating to the ingredients The RES group state that the main lessons to be learned included 1. The causes of undeclared allergens were simple and preventable and require a. regular review of formulation and ingredients b. double checks on labels immediately prior to packing to ensure that the product and packaging match 2. Packaging and labelling control are as important for allergen control as cleaning and GMP. While GMPs and preventive controls are critical in averting the unintended presence of allergens through cross-contact, it is equally important to be sure that all the allergens that are used or that are components of ingredients are declared. 3. Allergen-related problems occur more frequently in some types of foods than in others. a. Difficult nature of some products requires the shared use of equipment b. Production of dry mixes creates challenges in the production environment (aerosols, dust). 4. Imported ingredients and consumer-ready food with unclear and/or incorrect declarations are emerging issues and close attention to supplier quality assurance is required. Zoning within manufacturing facilities and Operational Pre-requisite Programs are now viewed as best practice and highly desirable for allergen management. These measures control the movement of people and equipment as well as the manufacturing environment, and minimise and contain adventitious crosscontact of potential hazards that are not immediately associated with product contact surfaces. 2. Cleaning and monitoring Cleaning is a GMP requirement and pre-requisite for minimizing the risk of cross-contamination in food manufacturing with all foreign matter including allergens. Cleaning is a multi-stage process designed to remove all food residues that has been successfully deployed and improved for decades. The composition of allergenic foodstuff such as wheat flour and milk contains only 1-3 % allergenic protein, and most ready-to-eat foods will contain much smaller amounts of allergen ingredients. It is expected that traditional 2
cleaning methods conducted correctly will be sufficient to remove the allergen component. Most cleaning methods result in 3-4 log reduction in product residues (data obtained from ATP testing). After cleaning, allergens are expected to be present at <1 ppm i.e. the limit of detection of most commercial test kits. The contribution of allergen cross contamination from a cleaned surface into subsequent finished product itself is therefore likely to add a very small non-detectable risk in the finished product. Gross failure of cleaning or lack of cleaning would be required to make a gluten-free product noncompliant (see example Appendix 3). Specific detection methods alone give partial information about overall safety and risk, and should be used as a balanced analytical approach. There are several methods for specific allergens of which immunological methods e.g. quantitative plate ELISA tests and qualitative lateral flow tests (LFT) in dipstick formats are the most commonly used. However, the relatively high cost is often an impediment to their widespread adoption. Plate ELISA tests are more sensitive (typically <0.1 ppm) but require a skilled analyst. LFTs are more convenient and have a limit of detection of 1 10 ppm but their performance can be variable. All methods for surface measurement are dependent on manual swabbing. Samples are recovered subsequently from the swab via a resuspension step which means that even ELISA tests are semiquantitative for this application. By contrast, simple rapid hygiene tests such as ATP bioluminescence and non-specific protein tests are widely used by industry and are well established proven methods of cleaning validation and verification. The benefit of such methods is that they are simple, rapid, sensitive and cost effective. In addition, trend analysis of results from regular monitoring yields more valuable information than infrequent testing. Accordingly, methods with the greatest sensitivity and broadest spectrum will give the greatest assurance of surface cleanliness and hence demonstrate a low cross-contamination hazard and risk from food residues and allergens. Jackson et al. (2008) conducted an in-depth survey 2 and overview of cleaning and other controls to prevent allergen cross contact in food processing operations. This multi-disciplinary team of experts from FDA, academia and industrial blue chip companies states that: There was no agreement on minimum level of allergen that causes a reaction in a sensitive consumer 2 Jackson et al.: Cleaning and Other Control and Validation Strategies to Prevent Allergen Cross-Contact in Food-Processing Operations. (2008) J Food Protect 71(2):445-58 3
There were many different causes/opportunities for cross contact in food processing both direct and indirect. Specifically, o no agreement regarding the best cleaning methods to remove food allergens through either wet or dry cleaning and there was o no agreement on safe residue levels Several test methods are used in industry to measure cleanliness during allergen control. However, each has its own limitations and there is no single method that satisfies all requirements. The paper concludes that Comparisons of immunochemical allergen specific methods and nonspecific methods (ATP and total protein) for determining cleaning efficiency are needed. Therefore, Hygiena conducted such a study at Campden BRI, the largest membership based food and drink research centre in the world. The results are described here. 4
3. Comparison of methods to assess cleanliness and residual allergens A thorough comparative study was conducted by Campden BRI in a pilot plant to simulate a factory cleaning procedure for the removal of food residues including four known allergens. The objective was to measure residues of ATP, total protein and the four specific allergens at appropriate stages during a simulated cleaning cycle. The foodstuff soil on a stainless steel surface was tested for gluten, casein, egg and peanut throughout the cycle. A slurry was prepared from a common commercial ready meal (beef with noodles) consisting of several main food groups. Allergens stated on the packaging were egg, wheat (gluten), soya and unsuitable for peanut allergy sufferers. For the purpose of the study the slurry was supplemented with 0.3g of freezedried peanut powder to ensure all allergens were at a detectable level at the start of the experiment. The homogenized product was diluted 1:1:1; food solids : semi-skimmed milk (1.8% fat) : tap water. The main constituents of the freshly prepared slurry were egg noodles (34%), yellow bean and chilli sauce, marinated beef (16%), beansprouts, red pepper, and spring onion. Ten stainless steel sheets (50 x 50 cm) were divided into 10 x 10 cm squares using black permanent marker creating a 5 x 5 grid. The surfaces were thoroughly cleaned with detergent and rinsed before the trial was initiated. The slurry (10g) was applied evenly to each 10 x 10 cm square on each stainless sheet and allowed to dry at 57 C for 10 minutes. For cleaning (see figure 2) a stationary power hose located 90cm away from the stainless steel sheet was used exclusively. The water/detergent/disinfectant sprays were applied at a pressure of 25 bar or 1 PSI (Pounds per Square Inch). A bucket of fresh water was present for purging the connecting hose after each chemical spray. Pre-trials were carried out to determine a) the right distance and spray times to sufficiently cover and then rinse the stainless steel sheets, and b) that a gradual reduction of food particles on the surface was achieved during the simulated cleaning cycle. Mechanical scrubbing was excluded from the study because this manual activity is subjective and difficult to replicate. The detergent Somplex Fatsolve (Diversey) was used at a concentration of 1-2% with a contact time of 10-15 minutes. The disinfectant Suma D10 J-Flex (Diversey), a QUAT based detergent Figure 2. Simulated industrial cleaning process to assess several detection methods for the measurement cleanliness and removal of food residue and allergens. 5
disinfectant, was used for cleaning and disinfection applied at a 1% solution with a contact time of 30 seconds. Surface swabs were collected using a randomised sampling plan. Ten replicate samples were collected and tested for each method at each stage of cleaning i.e. Stage 1: before drying, Stage 2: after pre-rinse, Stage 3: after detergent and rinse, and Stage 4: after disinfectant and rinse The applied test methods included 2 quantitative ELISA tests for gluten and peanut as benchmarks and a range of rapid specific and non-specific tests for measuring product residues. 1. High sensitivity ATP: EnSURE Luminometer with SuperSnap swab device sensitive (limit of detection) to 0.1 fmols ATP and giving quantitative results in Relative Light Units (RLU). 2. High sensitivity Total Protein: AllerSnap (Hygiena) incubated for 30 mins at 37 C yields a semiquantitative result based on colour change from green to purple with a limit of detection of 1 3 microgram ( μg) total protein. 3. Lateral Flow Device Casein: Reveal 3-D Casein (Neogen Corporation); horizontal lateral flow immunochromatographic strips for the qualitative detection of casein residues in 5 minutes at room temperature. The limit of detection is stated to be low ppm (presumed to be <10 ppm). 4. Lateral Flow Device Gluten: RIDA QUICK Gliadin (r-biopharm); vertical lateral flow immunochromatographic strips for the qualitative detection of gliadin (gluten) residues in 5 minutes at room temperature with a limit of detection of approx. 0.5 μg gliadin/100 cm 2 (approx.1 μg gluten/100 cm 2 ) dried on to a surface. 5. Lateral Flow Device Egg: (r-biopharm) vertical lateral flow immuno-chromatographic detection strips for the qualitative detection of egg residues in 10 minutes at room temperature with a limit of detection of 1-3 µg (also cross reacts with chicken meat and skin). 6. Lateral Flow Device Peanut: (r-biopharm) vertical lateral flow immuno-chromatographic strips for the qualitative detection of peanut in 10 minutes at room temperature with a limit of detection of 10 μg dried on to a surface. 7. ELISA Gliadin immunoassay: RIDA SCREEN (r-biopharm); plate ELISA for detection of gliadin in 30 mins with detection limit of 2 mg/kg (ppm) gliadin (or 4 mg/kg) gluten and a limit of quantification of 5 mg/kg (ppm) gliadin (or 10 mg/kg) gluten. 8. ELISA Peanut immunoassay: RIDA SCREEN FAST Peanut (r-biopharm); plate ELISA for detection of peanut in 30 mins with detection limit of 1.5 ppm peanut (mg/kg or 0.00015 % peanut) and a limit of quantification of 2.5 ppm peanut (mg/kg or 0.00025 % peanut) 6
4. Results Table 3 summarizes the results at each stage of cleaning and for each test as the number of positive samples where the product residue was detected out of the total number of replicates (X/Y). The ELISA tests were performed in triplicate (X/3) whereas all other test used 10 replicates (X/10). The most sensitive tests were ELISA and the high sensitivity ATP test The high sensitivity total protein test (AllerSnap) had similar or better sensitivity than the lateral flow tests (LFT). The Gluten LFT had better sensitivity than all the other LFT. The LFT for egg did not give any meaningful results. Table 3. Product residues detected by 8 methods during the 4-stage cleaning process Pass/Fail Stage 1 Stage 2 Stage 3 Stage 4 dry slurry pre clean detergent & rinse disinfectant & rinse ELISA gluten 1/1 3/3 3/3 3/3 ELISA peanut 1/1 3/3 1/3 0/1 EnSURE / SuperSnap 10/10 10/10 10/10 10/10 AllerSnap 10/10 10/10 5/10 0/10 LFT gluten 9/10 10/10 10/10 0/8 a LFT peanut 9/10 b 7/10 c 0/10 0/10 LFT casein 10/10 d 10/10 e 3/10 e 0/6 a LFT egg failed in this study a invalid results removed, b 2 faint results, c 5 faint results, d 3 faint tests, e 1 faint result ELISA Gluten tests showed a gradual reduction in the amount of allergen detected during the cleaning cycle. 0.03mg/L Gliadin/gluten residues were detected after the disinfection and rinsing stage (figure 3). ELISA Peanut showed a gradual reduction in the amount of allergen detected as the cleaning cycle but it was less sensitive than the ELISA Gluten test. After washing with detergent and rinsing, only 1 of the 3 replicates detected the presence of peanut residue above its calculated limit of detection of 0.13mg/L. After the disinfectant and rinse stage only one sample was collected although no residue was detected above the calculated limit of detection (<0.13mg/L). The high sensitivity ATP test (EnSURE & SuperSnap) successfully detected the removal of the food residue at all levels. It was as sensitive as the ELISA Gluten test and more sensitive than the ELISA peanut test. Closer inspection of the ATP swab data revealed that the median measurement after the disinfectant step was 14 RLU (figure 3). Any result above 2 RLU (0.1 fmols ATP) would be considered a positive. Accordingly, the ATP 7
Figure 3. Comparison of high sensitivity ATP and Protein tests. Numbers on the x-axis denote cleaning stages. AllerSnap positive detections out of 10 and median ATP RLU values are also shown. test had sufficient sensitivity to be able to detect up to ten-fold lower residue levels of this slurry, i.e. a 4 log reduction of food residue. Verification for other foodstuffs and processing conditions would be required. The high sensitivity total protein test (AllerSnap) detected product residues at stages 1, 2 and 3 but not after the disinfectant and rinse step. The limit of detection of AllerSnap is 1-3 µg protein per swab and gave a similar performance to the ELISA peanut test. However, the ELISA Gluten test detected lower levels of products residue after the disinfectant stage. Figure 4 compares the performance of the high sensitivity total protein test (AllerSnap) to the high sensitivity ATP test (SuperSnap) and demonstrates the greater sensitivity of the ATP test that also gives faster results (15 seconds). This shows that ATP (a common component of all foodstuffs) can be detected in residues sooner and in smaller quantities than total protein or specific allergens. AllerSnap provided consistent and reliable results and detected residues at all stages of cleaning except the disinfectant stage. AllerSnap gave results equivalent to or better than LFTs (figure 4). LFT allergen tests gave variable results and did not detect specific allergen residues at all stages of cleaning. In addition, a notable proportion of devices from all lateral flow tests were affected by variations in the result output. Lateral flow tests are designed to give a definitive presence/absence answer although they are known to have limitations. For example, too much food residue is known to cause interference and can give false negative results for allergen tests (the so called hook effect or poisoning ). This was Figure 4. Comparison of AllerSnap with LFT specific allergen tests. Columns refer to percentage positive detections. Numbers on the x-axis denote cleaning stages as shown in figure 2. 8
observed as faint lines, which complicated result interpretation. Therefore, a faint line may suggest a positive result or may indicate a faulty test. Only the gluten LFT detected residues in all 10 replicates after the detergent cleaning step at which stage only 2 out of 10 samples gave a positive detection with the milk casein LFT. The peanut LFT detected 2 out of 10 clear positive samples after the pre-rinse stage although 5 other samples were weakly positive. Peanut residues were detected by the ELISA peanut test but not the LFT after the application of detergent. Furthermore, LFT peanut yielded two invalid results while LFT egg did not provide any reliable results. This test gave false positives with all samples including a negative control. 5. Discussion The results from the study described above demonstrate that good cleaning can remove all food residues including its allergenic components to levels below the limit of detection of the test. Several methods can be applied to monitor and verify the cleaning process including specific and non-specific methods that can be equally effective. A combination of methods can provide a greater assurance of cleanliness. A case study described in Appendix 1 confirms this as well as giving a cost benefit. So what is the relative importance of cleaning in an allergen control program? Analysis of recall data by FDA shows that 88% of non-compliance cases are due to labelling issues and 12% are from other causes. Data from a snapshot Food Standards Agency UK survey show that <50% of those foods with precautionary labelling as may contain... actually contained allergens 3. Jackson s excellent survey and overview (2008) states that there are many different causes/opportunity for cross contact in food processing both direct and indirect and suggested several preventative measures (Appendix 2). The cleaning process itself is but one factor and although dry cleaning has more potential to create allergen problems it needs to be balanced against the requirements for pathogen control. Where cleaning has been cited as a probable cause of allergen cross contact in recalls involving egg and peanut, the likely reasons were gross lapses in cleaning practice or failure to schedule processing. Jackson et al. (2008) also stated that there are several test methods used in industry to measure cleanliness during allergen control but each have their own limitations and there is no single method that satisfies all requirements. ELISA methods are often held up as the gold standard but even their performance is influenced by food manufacturing processes. 3 Food Standards Agency UK: Survey of allergen labelling and allergen content of processed foods; Project Code FS241038; web access: http://www.food.gov.uk/science/research/allergy-research/fs241038 last accessed: 15/05/2015 9
The prime purpose of cleaning is to remove soil (product residue) from processing equipment. Allergens are generally a lesser component of the food residues, so the ideal test for cleaning verification and validation is an instant sensitive test for common components of product residue e.g. ATP and total protein tests which are well established and accepted. Cleaning is optimized by correct use of appropriate chemical and physical actions and there is no evidence to suggest that allergen removal requires additional or different treatments. However the reuse/recycling of CIP chemicals may require investigation of the potential to carry over unwanted protein residues. For cleaning validation studies, the use of specific allergen tests is also recommended by several food safety organizations. There are no international standards for any method to measure the efficacy of cleaning because each processing facility is unique and one size does not fit all. Manufacturers are expected to do the best that they can and are encouraged to frequently monitor performance and gather data for trend analysis. Jackson et al. (2008) commented that although the presence of an allergenic food in swab samples or rinse water indicates that the allergen cleaning protocol or its execution requires revision, it does not necessarily indicate the presence of the allergenic protein in the finished product. The transfer of allergenic protein from equipment surfaces to foods is a complex process that depends on many factors, including the adhesion properties of the protein to the surface, the abrasiveness of the subsequently processed food, the composition of the food contact surface, the temperature of processing, the concentration of the allergenic protein, and the properties of the allergenic protein (e.g. physical form and solubility in the subsequent food being processed). Residual level of allergen would also have to be extremely high to create a non-compliant at risk scenario (see Appendix 3 regarding gluten), yet most post-cleaning verification test show a negative result or not detected at the limit of detection typically 1 10 ppm. 10
6. Summary & Conclusion In a simulated industrial cleaning process, 8 test methods were used to measure non-specific food residues and specific allergens residues at each of 4 stages of cleaning. The cleaning process successfully removed all residues to levels near or below the limit of detection of the test methods. The most sensitive tests were ELISA allergen tests and high sensitivity ATP tests The high sensitivity total protein test (AllerSnap) had a similar or better sensitivity compared to the LFT. The Gluten LFT had a better sensitivity than all the other LFT. A combination of methods can provide a greater assurance of cleanliness and a cost effective monitoring program. Whereas cleaning is an important CCP in the control of allergen, the risk of cross contamination due to inadequate cleaning is relatively low. Packaging is the main cause of allergen non-compliance and recalls. The management of procedures to prevent cross contact within the entire manufacturing process together with the control of labelling is more important to minimize risk, non-compliance and expensive recalls. Prepared by Dr Martin Easter and Dr Andreas Rossbach ( Hygiena International Ltd) Unit E, 3 Regal Way, Watford, Herts., WD24 4YJ, United Kingdom Tel +44 1923 818821 ; enquiries@hygiena.net 11
Appendix 1: Case study - Detecting allergens in a Ready Meal Factory. A production facility manufacturing ready meals and vegetable dishes for major supermarket retailers but also makes a nut product on a less frequent basis. The site needs to ensure that its cleaning has been effective to remove nut allergens after the manufacturing of nut products and before releasing the production area back to general manufacturing. The products contained 3 different tree nuts but for the sake of completeness nine nut allergens were tested in the cleaning validation exercise and all nine nut allergens need to be shown to be absent before release of the lines and equipment. An off-site contract laboratory was used to conduct specific ELISA-based allergen tests with a turnaround time of 10 working days during which the production facility could not be used thus losing valuable production time. A minimum of ten different samples were taken at various points of the production facility and each sample was tested for 9 tree nut allergens at considerable cost. Previous cleaning validation exercises using only the specific allergen tests had not always passed first time thus requiring repeat testing and the production line out of use for further 10 days. This exercise incurred considerable cost, and the facility needed a faster, more reliable, and cost effective way to validate the cleaning. The EnSURE luminometer with SuperSnap gives a high sensitivity ATP test to a sensitivity level of 0.1 fmol ATP and results were obtained in 15 seconds to give immediate feedback and corrective action. Surfaces that failed at greater than 10 RLU were re-cleaned and re-tested. When all surfaces passed with SuperSnap the surfaces were then swabbed with a high sensitivity total protein detection swabs (AllerSnap). If the protein test gave negative results showing absence at the 1 μg level, the specific allergen tests were employed. Subsequently, all the specific allergen tests were shown to be negative and the line was released back to production. The staff felt extremely confident that the outcome of the specific allergen tests would be negative following the initial pre-validation using the SuperSnap and AllerSnap tests. Pre-validation screening enabled the site to make significant savings by avoiding repeat testing and further lost production. The Hygiene Manager commented that the combined method approach was very beneficial in releasing the nut production back into general production and that This process gave me confidence that we would get it right first time with the allergen swabs. This not only saved on cost but more importantly guaranteed food safety. All our allergen swabs came back clear and the area was released back to general production on plan. I would definitely employ this process again. The regular use of high sensitivity ATP and high sensitivity protein tests enable high standards of cleaning to be maintained that can be supplemented with specific allergen tests as required. 12
Cleaning is one the CCPs for allergen control and a variety of detection methods are available to validate the cleaning processes. Specific allergen tests have limitations and are expensive whereas other methods have sensitivity but lack specificity. A combination of three high sensitivity detection methods (ATP, protein and specific allergen tests) provide a more comprehensive, sensitive and rapid result and deliver a timely, cost effective solution. 13
Appendix 2: Prevention of cross-contact during processing (from Jackson et al., 2008) 1) Scheduling of processing runs. a) Schedule long runs of products containing allergenic ingredients to minimize changeovers. b) Segregate allergenic and non-allergenic product production areas, or if this is not possible c) Process non-allergenic foods before allergenic products. d) Schedule sanitation immediately after production of foods containing allergenic ingredients. e) When product design permits, add allergenic ingredients as late in the process as possible. 2) Use of dedicated systems. a) Dedicate processing equipment and lines, if possible, to prevent allergen cross-contact. b) Dedicate tools, containers, and utensils and color code or clearly mark them. c) Minimize reuse of processing and/or cooking media (water or oil). d) Restrict personnel working on processing lines containing allergenic ingredients from working on non-allergenic production lines. 3) Control of rework and work in progress. a) Use color-coded tags to identify and record when reworked products with allergenic ingredients are produced, where they are stored, the products to which they are reworked into, and when these products are added back into the line. b) Use rework containing unique allergenic foods and/or ingredients only in the same formulation (e.g., like into like practice) 4) Maintain equipment to ensure that the systems are operating as designed. Jackson et al.: Cleaning and Other Control and Validation Strategies to Prevent Allergen Cross-Contact in Food-Processing Operations. (2008) J Food Protect 71(2):445-58 14
Appendix 3: Worked example of gluten cross contamination Gross cross contamination which would be visible with the naked eye would have to occur to make a non-compliant product. Theoretical example of contamination of glutenfree bread with gluten from the preceding production run due to poor surface cleaning. 15