Allergen Analysis – key considerations (including gluten & food intolerance compounds)

Analysis may be required for a variety of reasons many of which are based on the needs of some form of risk assessment:

• Protect consumers at risk
• Establish allergen status of raw materials/ingredients
• Identify equipment that is difficult to clean
• Validate the effectiveness of cleaning, swabbing/allergen changeovers
• Verify the risk of allergen carry-over in the finished product
• Confirm free-from claims
• Monitor the effect of critical changes
• Inform the risk assessment in incidents and complaints
• Enforcement and forensic: seized exhibits, replica meals^[1]

Having identified the purpose, we need to think about sampling and how the sample is handled, what type of analysis we will carry out and what method we will apply. We need to consider how to gain assurance that the method is fit for purpose, and how the results are reported. The results must be interpreted in the context of the risk assessment. Figure 1 summarises these steps.

Figure 1: Analytical route map (Practical Guidance on the Application of Allergen Quantitative Risk Assessment - Courtesy of ILSI-Europe, 2022)

Analytical results are only useful if the sample submitted to the laboratory was meaningfully taken. See IFST Information Statement: 'Sampling for Food Analysis'. However, there is relatively little practical guidance on sampling for analysis related to food hypersensitivity. To fill this gap, guidance has been produced by the European section of the International Life Sciences Institute (ILSI-Europe). A brief overview is presented here. Sampling and analysis of raw materials to verify they are what they are claimed to be is relatively straightforward, although not trivial and not dealt with here. The more complex question is often “is there any unintended allergen present and if so, at what concentration?”. Figure 2 illustrates a strategy to determine the number of samples required.

Figure 2: Strategy to determine the appropriate number of samples

The ILSI guidance envisages three levels of concern, low medium and high. The sorts of common and different factors bearing on these are shown in figure 2. At a low level of concern, if present at all, the unintended allergen is known to turn up regularly and homogenously in the food product of interest. A small number of samples will usually suffice to give useful quantitative analytical data. As concern increases, less can be assumed about the frequency and homogeneity of the unintended allergen presence. If it is likely to be particulate in nature, irregularly and unpredictably distributed in the food product greater numbers of samples are required to characterise both the amount present and the dispersion of the data. Here, rules of thumb provide much-needed consistency to the sampling plan. One such is to take the cubed root of the number of units in the consignment as the number of samples to be taken and individually analysed. In particularly challenging circumstances the sampling plan should involve a much larger number of samples, e.g. the square root of the number of units plus one. This becomes akin to the sort of sampling plan necessary for mycotoxin analysis, and for the same reason, ‘hot-spots’ in the consignment. There are other things to think about such as increasing sample and test portion sizes to reduce dispersion. Please see the ILSI-Europe guidance and references therein.   

For more information see Allergen Bureau sampling guidance.

Figure 4 gives a summary of the main analytical techniques applied to food hypersensitivity analyses:

Figure 4: Summary of the main analytical techniques applied to food hypersensitivity analyses.

Cross-reactivity is defined as a positive response to a sample that does not contain any of the target analyte (Abbott et al., 2010). Those who develop allergen ELISAs routinely check their assay for cross-reactivity against a wide battery of food commodities. Particular care is needed with foods that are genetically similar to the target because of the higher probability of expression of proteins with amino acid sequence homologies. Examples include:

• Cross-reactivity between members of the Prunus family including apricot, cherry, peach, plum and almond, as well as mahaleb (Prunus mahaleb L.)
• Cross-reactivity between peanut and soya, soya is closely related to other members of the Leguminosae family as shown by Lacorn et al., 2018.
• The most well-known example of ELISA cross-reactivity is within the Apiaceae family, also known as the Umbelliferae, or, more familiarly, as the celery, carrot, or parsley family. Owing to extensive cross-reactivity, a specific ELISA is not possible for celery, which necessitates a PCR approach.
• A relatively recent emerging cross-reactivity is with regard to several mustard ELISAs which also react to canola protein. This can be solved by PCR provided the primers/probes have been validated as negative to the relevant Brassica species. A further problem has arisen with cross-reactivity to Sinapis arvensis, the wild mustard or charlock, for which PCR is also cross-reactive. An LC-MS method remains to be validated and this issue is a current problem.
• Clinical experience of patients cross-reacting to different foods that possess similar proteins is often a guide to laboratory cross-reactivity. For example, Lipid Transfer Proteins (LTPs) are found in plants and foods that contain plants. Lipid Transfer Protein Syndrome is an allergy affecting people who have become sensitised to LTPs. LTPs are small proteins found in a wide variety of plant foods; common foods include tomatoes, apples, raspberries, stoned fruit, grapes, walnuts, almonds, corn, and barley. Patients may react to one or multiple foods.

For commodities such as crustaceans and fish, where multi-species allergies are common, cross-reactivity may be welcomed as a screening approach.

The Polymerase Chain Reaction deployed in real-time PCR (qPCR) is readily available to detect and quantify the amount of allergen species-specific DNA in a sample.

For some foods, for example, celery, no ELISA is available owing to cross-reactivity in the botanical family to which celery belongs. Celery is thus routinely assayed by qPCR.

The advantages of qPCR include that DNA is often a more robust molecule than many proteins. qPCR gives the required sensitivity, multiplexing, and DNA specificity in a relatively rapid, uncomplicated setup and workflow. There are many real-time PCR instruments available and the application of qPCR gives complementarity when confirmation of the species is required.

The main disadvantage leveled against qPCR is that the analytical target is DNA rather than protein, which is the hazard. This is valid, and for egg and milk, DNA methods are not applicable since it is not possible to distinguish DNA from hens’ eggs from DNA from chicken meat. Similarly, cows’ milk and beef both contain bovine DNA. However, in most other cases validated qPCR findings should not be dismissed without detailed evidence that no protein is present.

Quantification of the concentration of unintended allergen species present in a sample is possible, initially on a DNA copy number to copy number basis, unless a suitable food-based reference material (RM) is used to construct the calibration curve. More meaningful quantification requires conversion factors, that remain to be fully harmonised. Digital PCR (dPCR) is capable of absolute quantification without a calibrator.

For more information, refer to the AMC Technical Brief on PCR.

Biosensors are proposed as potential alternatives to traditional methods, such as ELISA, for direct, reproducible, real-time, and relatively inexpensive detection of food allergens. They are analytical devices incorporating a biological, or biologically derived, sensing element (such as antibodies), either intimately associated with, or integrated within a physicochemical transducer. The transducer converts the recognition event into a measurable chemical or physical signal, to produce a digital electronic signal proportional to the concentration of a specific analyte, or group of analytes. These can be generic proteins, from the food causing the allergenic effect, or specific biomarker proteins identified for different food allergens.

The combination of new selective bio- and biomimetic receptors, with the unique optical, electrical, or electrochemical properties of nanomaterials, has allowed the development of new nanosensors with the potential for improvements in food safety, quality control, and traceability. They offer excellent prospects for the development of portable, miniaturised, real-time, easy-to-use, rapid, and cost-effective sensors capable of screening multiple allergens when required. Some of the advantages of nanosensor applications, for allergen detection, include low detection limits (single molecule detection), reduction of reagent volumes, shorter analysis times, or multianalyte detection. Their use has also led to improved sensitivity, selectivity, robustness, and accuracy of conventional analytical methods, higher efficiency and sample throughput, as well as better performance for the analysis of complex food samples. Relatively few sensors based on nanoparticles have been described for the analysis of food allergens.

From an analytical perspective, new methods are required for the detection of trace concentrations of allergens in complex food matrices, as well as for hidden food allergens, or new allergenic foods prior to them being commercially available. Other challenges in this field include the exploration of new ways to enhance the signal-to-noise ratio, signal amplification, enhancement of the transduction efficiency, sample preparation, and biosensor validation. In any case, it is vital to connect laboratory research with real-world applications. Biosensors are of interest to the food industry and regulatory bodies, responsible for the protection of the health of food-allergic consumers. The food industry could play an essential role by encouraging research in this field. At the time of writing biosensor applications for food allergen analysis largely remain in the research and clinical fields.

It is axiomatic that laboratories carrying out food hypersensitivity related analysis have quality assurance in place - planned and systematic activities within the quality system. These include familiarisation with and validation of the method. Validation includes assessment of the method performance characteristics, typically accuracy (trueness, or ‘bias’), applicability (for the matrix and concentration range), the limit of detection, limit of quantification, precision (repeatability including intermediate precision and reproducibility), selectivity, sensitivity, linearity and uncertainty. These terms are simply explained in the IFST Information Statement: ‘How to Choose and Instruct a Laboratory for Chemical Food Analysis’ which references the authoritative international texts on the subject. Quality control activities to monitor the performance of the method, e.g. control charts, and accreditation to ISO/IEC 17025 ‘General requirements for the competence of testing and calibration laboratories’, provide assurance that the method remains fit for purpose.

For ELISA, special focus should also be on checking for cross reactivity, matrix interference, and practical issues such as pipetting, plate washing, calibration curves and the use of in-house and external RMs and proficiency testing rounds. It is also important to remember that the reactions in an ELISA may not have time to come to equilibrium if the assay is to be completed within the time allocated hence the kit manufacturer’s procedure must be followed exactly, especially as to timing and temperatures.

In qPCR the same considerations of quality assurance and quality control apply and in addition special attention is required to the quantity and quality of the extracted DNA. Measurement of DNA concentration and purity can be by optical density (absorbance), agarose gel electrophoresis or fluorescent DNA-intercalating dyes. Absorbance at 260 nm can be used to calculate the concentration of nucleic acids in solution and for reliable spectrophotometric DNA quantification, a series of dilutions is recommended.  Strong absorbance close to 230 nm indicates organic compounds or chaotropic salts in the extracted DNA. Substances including urea, EDTA, guanidine hydrochloride, and phenolate ions all exhibit absorbance maxima at or close to 230 nm. The absorbance ratio at 260 nm over 230 nm should be greater than 1.5, and preferably close to 1.8. Values that differ significantly from these target values are usually indicative of extraction reagent carryover. See also the specific advice on quantification by qPCR by Burns (2019).

Again, for LC-MS/MS the same considerations of quality assurance and quality control apply but there are particular challenges associated with allergen mass spectrometry. The experimental design aimed at developing a new LC-MS/MS method is outlined showing the selection of protein and peptide targets, sample preparation, preliminary MS work-up, follow-up work preferably using an incurred material to assess peptide recovery before moving on to ‘real’ foods and validation.  

In more detail, the protein molecular weight should be known, along with the full amino acid sequence and the protein should be unique to the foodstuff to be detected. At a minimum, the sequence should not be present in commonly analysed matrixes and other food ingredients. The protein expression should be characterised and preferably expression variability should be minimal. In other words, not altered to an appreciable extent by the tissue, species/cultivar, year/season, response to environmental/disease/stress factors, or post-harvest treatment. The protein should ideally not be subject to modification (either post-translational or processing-induced). If present, modifications must be considered when developing peptide targets and MS methods. The protein should not be subject to hydrolysis during storage or food processing and should be released from the matrix into the solution for detection. The protein should be reproducibly digested by the endoprotease enzyme used in the chosen sample preparation technique.

Data analysis is crucial in all analytical situations and although this aspect has similarities between ELISA and LC-MS/MS, in the latter, as a newer technique applied to allergen analysis, it is particularly challenging. The three essential components are: assurance around the detection of positive signals, calibration, and signal integration. Appropriate best practice is available (Johnson and Downs, 2019), along with transparent reporting criteria. These include full descriptions of:

• calibration material
• mathematical operations for calibration
• strategy used to select signals for use in quantification and those to be used qualitatively. 

Adoption of these criteria will improve the evaluation, comparability, and transferability of LC-MS/MS methods.

An excellent example is the reference method for the determination of total allergenic protein content in processed food (milk in cookies) at clinically relevant concentrations, traceable to the SI and with well characterised uncertainty.

Apart from FODMAPs, the compounds that give rise to food intolerance are mainly small molecules for which the difficulties of analysis exhibited by proteins are not encountered. Thus, HPLC and Ultra-Performance Liquid Chromatography (UPLC) methods, with suitable standards are readily available. For FODMAPs, relatively larger carbohydrate molecules High-Performance Anion Exchange chromatography with Pulsed Amperometric Detection is the most suitable approach. For sulphites a variety of methods are available including enzymatic analysis, ion-exclusion chromatography with amperometric detection, and variations on the Monier-Williams wet chemical reference method. It should be borne in mind that the Monier-Williams method has long been known to be interfered with positively by foods such as dried garlic and soya proteins. These foods typically assay around 200 mg per kg and 30 mg per kg as sulphur dioxide (SO₂) respectively in the Monier-Williams method. Other foods such as onions and cabbage also typically give responses in the range of 3 – 20 mg per kg as SO₂. These interferences are caused by the presence of volatile sulphur-containing compounds. In case of doubt, an LC-MS/MS should be applied.