FOOD ALLERGEN TESTING

1. FOOD ALLERGEN TESTING Peter Cressey ESR Allergen Seminar Waipuna Hotel and Conference Centre 21 March 2007

2. Presentation Content • What are we testing for? • What techniques are available? • What are the strengths and weaknesses of various methods? • Common pitfalls and problems in allergen testing • When to use testing

3. What are we testing for? • True food allergies are an immunological response to specific food proteins • Test methods can target proteins or the genetic material that produces the proteins (DNA) • The test method needs to be as specific as possible for the food material of concern • The test method need to be able to detect the material of concern in as wide as possible a range of circumstances (e.g. processed informs of the food including heat treated)

4. What Techniques are Available? • Protein-based methods (usually antibody based) that determine proteins characteristic of the allergenic source material (but not necessarily the allergenic proteins) • DNA-based methods that determine a characteristic sequence of base pairs (not necessarily related to the allergenic proteins) Most allergen testing utilises protein-based methods (Enzyme-Linked ImmunoSorbent Assay – ELISA) detecting a specific sequence of amino acids (epitope)

5. Rapid Antibody-Based Methods • So called ‘dipstick’ or ‘lateral flow’ methods • Extracted proteins are captured on coloured particles with antibodies attached • Particles are allowed to migrate through a support medium, containing a zone of antibodies • A positive result is visualised as a coloured line on a dipstick or test strip

7. Rapid Allergen Methods • Fast (<10 minutes) • Sensitive (detection limits ~5 ppm) • Qualitative • Suitable for screening of foods and environmental swabs • Currently available for peanut, gliadin (gluten), milk

8. Sandwich ELISA

9. Sandwich ELISA • Extracts added to reaction well coated with bound antibodies • Second set of antibodies are then added with enzyme linked to the antibody • Coloured chemical added, the enzyme converts the chemical to a different form with a different colour

11. Sandwich ELISA • Test methods are exacting and usually take 1-2 hours to complete • Sensitive (sub-ppm to ~5 ppm, depending on kit) • Quantitative* • Available for most major allergenic source materials, except fish

12. Pitfalls with ELISA • Detection level too high (some kits designed for adulteration detection – percentage level, not ppm) • Kit not validated for substrate or state of allergen (heat-treated, non-heat-treated) • Not following kit insert instructions • Hydrolysed or fermented proteins • Non-homogeneity in test material • Too low detection limit – what do you do with the result?

13. Limit of Detection too High • Until recently one of the few soy kits available had a detection limit of 5000 ppm (0.5%) soy protein • Soy allergy sufferers have been reproted to react to as little as 88 mg soy protein or 18 g of a food containing 5000 ppm of soy protein • This kit was designed for grain adulteration not allergen detection

14. Kit not Validated for Substrate or State of Allergen • With some kits cooked egg protein only showed 5-20% of the response of raw egg protein. However, most egg allergic individuals are allergic to both raw and cooked egg • Almond protein was recovered from a biscuit matrix quantitatively, however recoveries from a fruit pulp were less than 20% of expected values • Tropomyosin, the target protein for crustacean kits varies significantly in content between prawn species and between prawns an dotehr crustacea

15. Not Following Kit Insert • Elisa kit methods are highly empirical and changing times, temperatures and concentration may not have the expected results • A US regulatory agency caused an unjustified Class I recall through a decision to multiply and analytical result by a factor of 25 • We have no local examples of this, BECAUSE WE DON’T DO IT

16. Hydrolysed or Fermented Proteins • The sequence of amino acids utilised by the test may be different to the sequence causing the allergic reaction • Hydrolysis and fermentation break up protein structure and may destroy either the test or allergic sequence • Consequently, the food may test positive while it is not allergenic (probably okay) or it may test negative when it is still allergenic (not okay)

17. Non-Homogeneity of Test Sample • The test is only as good as the sample tested! • Allergen contamination of a food has the potential to be distributed in a non-uniform manner (‘hot spots’) • Sampling needs to take this into account

18. Limit of Detection Too Low • Some kits are extremely sensitive (e.g. limit of detection 0.5 ppm) and will elicit positive test results at even lower concentration (e.g. 0.1 ppm) • The most sensitive individiuals reported in the literature react to about 0.1 mg of protein from the allergenic source • At a detection level of 0.1 ppm, the most sensitive allergy sufferers would need to consume 1 kg of the offending food to experience an adverse reaction • However, positive test results cannot be ignored, no matter how low

19. Qualitative or Quantiative? • Many kits can adequately detect the qualitative presence of allergenic source material • Quantitative measurement requires a target protein that is: • Always present at about the same concentration • Unaffected by food processing • Able to be extracted quantitatively from a range of foods • Many kits appear to be quantitative when, in fact, they are only qualitative

20. Test kit validation – what is ESR doing? • Sensitivity – what is the actual limit of detection? • Specificity – does the kit cross react with other than the target proteins? • Precision – how repeatable are results? • Accuracy – does the kit result represent the actual amount of material present? • Confounders – does processing affect the performance of the kit (e.g. heating of ingredients)?

21. Test kit validation – typical results (Tepnel peanut kit) • Sensitivity – LOD approximately 1.5 ppm of peanut • Specificity – no detectable cross reaction with range of foods (soy, almond, cashew, lentils, sesame, cornmeal, linseed) • Precision – CoV of approximately 20% • Accuracy – average spike recovery 106% • Heat affect – 90% recovery after heating for 2 hours at 100 C • Validated for biscuit matrix, other matrices would need to be validated on a case by case basis

22. Performance of Currently Available Test Kits • While kits all appear similar, there seems to be considerable variation in the degree of development and validation carried out • Peanut kits from major manufacturers are generally the best and have been through AOAC method assessment process • Some kits are perfectly satisfactory for screening, but not really suitable for quantitative testing • Some kits (e.g. crustacea) would only be of limited value for screening, unless the test situation was well defined (i.e. known allergen source in known food)

23. Conclusions • A range of antibody-based assays are available for most common food allergens • Assays require no or minimal laboratory resources (unless quantitation is required) • Limitations of assay need to be acknowledged • Assays should be validated for particular situations

24. When Should Allergen Testing be used? • Investigation of consumer complaints (critical incidents) • Validation of sanitation methods • Incoming raw product validation • (Finished product testing?)