Updated population minimal eliciting dose distributions for use in risk assessment of 14 priority food allergens

https://doi.org/10.1016/j.fct.2020.111259Get rights and content

Highlights

  • The current study provides new and updated population minimal eliciting dose (EDp) distributions for 14 allergenic foods.

  • Updated dataset increased from the previously reported 1750 allergic individuals to currently over 3400 allergic individuals.

  • Utilizes novel Stacked Model Averaging statistical method for interval-censored data.

  • The results presented can be used in the risk assessment and subsequent risk management of allergenic foods.

Abstract

Food allergy and allergen management are important global public health issues. In 2011, the first iteration of our allergen threshold database (ATDB) was established based on individual NOAELs and LOAELs from oral food challenge in roughly 1750 allergic individuals. Population minimal eliciting dose (EDp) distributions based on this dataset were published for 11 allergenic foods in 2014. Systematic data collection has continued (2011–2018) and the dataset now contains over 3400 data points. The current study provides new and updated EDp values for 14 allergenic foods and incorporates a newly developed Stacked Model Averaging statistical method for interval-censored data. ED01 and ED05 values, the doses at which 1%, and respectively 5%, of the respective allergic population would be predicted to experience any objective allergic reaction were determined. The 14 allergenic foods were cashew, celery, egg, fish, hazelnut, lupine, milk, mustard, peanut, sesame, shrimp (for crustacean shellfish), soy, walnut, and wheat. Updated ED01 estimates ranged between 0.03 mg for walnut protein and 26.2 mg for shrimp protein. ED05 estimates ranged between 0.4 mg for mustard protein and 280 mg for shrimp protein. The ED01 and ED05 values presented here are valuable in the risk assessment and subsequent risk management of allergenic foods.

Introduction

Food allergy and allergen management are significant global public health issues. Food-allergic individuals must adhere to specific avoidance diets to prevent the occurrence of allergic reactions (Muraro et al., 2014; Boyce et al., 2010) and as such, a number of countries and regulatory bodies who recognized this importance have enacted critical laws, regulations or standards for labeling of “priority allergens” to enable avoidance (Gendel, 2012; Yeung and Robert, 2018). However, allergen control in the processing facility and throughout the ingredient supply chain is a challenging venture for manufacturers in a globalized economy (Yeung and Robert, 2018). When ingredients and technologies are sourced worldwide from multiple business partners, complexity rises, which can increase the chance for the unintended presence of allergens in food, leading to potential harm as demonstrated by the number of large scale recalls that have affected a large number of food companies (Garber et al., 2016; Gendel, 2018; Sayers et al., 2016; Walker et al., 2016). Fear of causing unintended harm to an allergic consumer, coupled with the lack of regulation regarding the procedures or warnings on labels surrounding unintentional allergen presence in foods has led to the proliferation of voluntary precautionary allergen labels on packaged foods (Allen and Taylor, 2018).

Depending on the level of sensitivity of an allergic individual (minimal eliciting dose of an individual), exposures from the unintended presence of residues of allergenic foods that are not declared on product labels (undeclared allergens) can pose a risk to food-allergic individuals (Blom et al., 2018; Remington et al., 2015; Zurzolo et al., 2019). In general, precautionary (advisory) allergen labeling (i.e. may contain) is used to inform consumers of products with a potential unintended allergen presence (UAP). However, it is now well-recognized that excessive use of precautionary labeling results in unintended consequences, in particular a decline in trust of the label and ignoring of the warnings which negates the original intent of the labeling (Allen and Taylor, 2018). The implementation of risk assessment approaches as a basis for the establishment of action levels for precautionary (advisory) allergen labeling is needed to quantify the level of risk. By quantifying the risk of UAP, it is possible to allow for risk management strategies that would sufficiently protect food-allergic consumers. PAL would be the final risk management option and using quantitative risk assessment as a support for the risk management decision making process would prevent PAL from being overly burdensome which has been shown to contribute to a reduction in the quality of life of food-allergic individuals.

In 2007, the Allergen Bureau of Australia & New Zealand (ABA) released their VITAL (Voluntary Incidental Trace Allergen Labeling) allergen management program with the goal of limiting precautionary allergen labeling related to the presence of an unintended allergen. As part of the VITAL 2.0 updated guidance in 2011, the VITAL Scientific Expert Panel identified doses of allergens where 1% and/or 5% of the respective allergic-population would be predicted to experience any objective allergic reaction (the ED01 and the ED05) and recommended Reference Doses (based on the ED01 and/or the 95% lower confidence interval of the ED05) to guide the risk management and the application of precautionary labels on food products (Allen et al., 2014; Taylor et al., 2014).

Since publication of the updated VITAL 2.0 Reference Doses, a number of stakeholders and national agencies have begun to adopt the use of Reference Doses (FAVV SciCom, 2017; NVWA BuRO, 2016; Sjögren Bolin, 2015; Waiblinger and Schulze, 2018). However, wide variations exist regarding approaches to the implementation of the data and models underpinning Reference Doses. Additional scientific improvements, in both the amount of data for individual eliciting doses from oral food challenges and the development of more advanced statistical modeling approaches, have also become available and their use could encourage the harmonization of risk assessment or risk management approaches and applications.

Previously, three parametric models (log normal, log logistic, Weibull) were used with Interval-Censoring Survival Analysis (ICSA) to fit results from individual oral food challenge data and to estimate the dose of an allergen (EDp) at which a proportion (p) of the allergic population would be likely to react (Allen et al., 2014; Ballmer-Weber et al., 2015; Taylor et al., 2014, 2010, 2009). However, limitations to this approach do exist. First and foremost, each of the parametric models provides a different EDp estimate. No biological basis exists for selecting between the different models. This limitation encourages multiple possible interpretations of the same data. Additionally, these previously available ICSA models were not able to incorporate random effects into their calculations. Random effects enter the models through study-to-study heterogeneity (i.e. differing protocols, participant recruitment, dosing schemes, possible regional genetic or environmental differences, etc). As a consequence, estimation of a reference dose using any single model may result in bias and estimates whose parameters do not fully reflect the true uncertainties. In other fields of risk assessment, especially regarding benchmark dose (BMD) approaches for chemical risk assessment, single model selection and rejection has been acknowledged as a suboptimal approach and a more advanced, preferred method for BMD calculation is model averaging (EFSA Scientific Committee, 2017). Bayesian model averaging has also been incorporated into the EPA's Benchmark Dose Software v3.1.1 (BMDS 3.1.1) application in an effort to incorporate available state-of-the-art methods in the publicly available BMD software (US EPA et al., 2019). However, model averaging techniques for interval-censored data were not available.

To fill this need, the National Institute for Occupational Safety and Health (NIOSH), the Netherlands Organisation for Applied Scientific Research (TNO) and the University of Nebraska-Lincoln, developed a Bayesian “Stacked Model Averaging” approach for interval-censored data (Wheeler et al., 2019). This approach combines parametric survival estimates from multiple models into a single EDp estimation based upon a weighted average of survival estimates designed to estimate the true survival curve and the EDp estimations from food allergen minimal eliciting dose distribution data. Furthermore, the approach allows for the inclusion of additional models not previously available; these distributions might better fit the food allergen dose distribution data. The Stacked Model Averaging approach also allows inclusion of random effects for representation of the study-to-study heterogeneity of the allergen dose distribution curve (Wheeler et al., 2019). This state-of-the-art, flexible method allows for more robust EDp estimations than previously available ICSA methods when using interval-censored data from oral food challenges.

In 2011, the first iteration of our allergen threshold database (ATDB), containing roughly 1750 individual data points, was established. On the basis of this dataset, the first population eliciting dose (EDp) results were published in 2014 (Allen et al., 2014; Taylor et al., 2014). The systematic collection of oral food challenge data has continued from 2011 until the fall of 2018 in a similar fashion and the dataset now contains over 3400 data points. The current study provides updated EDp values using stacked model averaging for the previously reported allergenic foods, as well as several new foods based on the expanded database where data have recently become available.

Section snippets

Methods

Publications were selected based upon the criteria outlined previously (Taylor et al., 2009), in particular focusing on results from low-dose oral challenges. Additionally, unpublished data were obtained from clinical records where possible. Data from double-blind, placebo-controlled food challenges (DBPCFCs) were preferred, except in the case of data from infants and very young children where blinding was not considered necessary. In the selected available studies, subjects were challenged

Results

From 2011 until August 2018, over 2516 titles and abstracts were identified from PubMed and Scopus and were screened for further review, 570 peer-reviewed articles were kept for full PDF review and 47 were identified as containing quantitative data in a useable format for the current study (Supplementary Table 1). Further data were added from unpublished clinical datasets if available (~25% of total data available). Already in 2011, sufficient data were available for the derivation of the ED01,

Discussion

In this current study, data from over 3400 low-dose oral clinical challenges were used to determine the ED01 (estimated protection levels of 99%) and the ED05 (estimated protection levels of 95%) with the respective 95% confidence intervals through the Stacked Model Averaging for 14 different allergens.

For risk management of unintended allergen presence (UAP) in food, zero risk in food allergy management cannot be achieved similar to other risk factors encountered in everyday life in society (

CRediT authorship contribution statement

Benjamin C. Remington: Conceptualization, Writing - original draft, Methodology, Software, Validation, Formal analysis, Investigation, Visualization, Supervision, Data curation. Joost Westerhout: Methodology, Software, Data curation, Validation, Formal analysis, Investigation, Writing - review & editing, Visualization. Marie Y. Meima: Data curation, Investigation, Writing - review & editing. W. Marty Blom: Conceptualization, Investigation, Data curation, Writing - review & editing, Supervision,

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The authors would like to thank Jamie Kabourek, M.S., R.D. for her contributions to enable a smooth literature review process through her role as the Resource Manager for the Food Allergy Research and Resource Program at the University of Nebraska. The Food Allergy Research & Resource Program is a food industry-funded consortium with more than 100 member food companies. This research was party financially supported through Dutch Governmental TNO Research Cooperation Funds.

References (51)

  • B. Niggemann et al.

    Accurate oral food challenge requires a cumulative dose on a subsequent day

    J. Allergy Clin. Immunol.

    (2012)
  • H.A. Sampson et al.

    Standardizing double-blind, placebo-controlled oral food challenges: American academy of allergy, asthma & immunology-European academy of allergy and clinical immunology PRACTALL consensus report

    J. Allergy Clin. Immunol.

    (2012)
  • R.L. Sayers et al.

    How much of a problem is peanut in ground cumin for individuals with peanut allergy?

    J. Allergy Clin. Immunol.

    (2016)
  • M. Shaker et al.

    The health and economic outcomes of peanut allergy management practices

    J. Allergy Clin. Immunol. Pract.

    (2018)
  • S.L. Taylor et al.

    Establishment of reference doses for residues of allergenic foods: report of the VITAL expert panel

    Food Chem. Toxicol.

    (2014)
  • S.L. Taylor et al.

    Threshold dose for peanut: risk characterization based upon published results from challenges of peanut-allergic individuals

    Food Chem. Toxicol.

    (2009)
  • S.L. Taylor et al.

    Threshold dose for peanut: risk characterization based upon diagnostic oral challenge of a series of 286 peanut-allergic individuals

    Food Chem. Toxicol.

    (2010)
  • P.J. Turner et al.

    Fatal anaphylaxis: mortality rate and risk factors

    J. Allergy Clin. Immunol. Pract.

    (2017)
  • J. Westerhout et al.

    Deriving individual threshold doses from clinical food challenge data for population risk assessment of food allergens

    J. Allergy Clin. Immunol.

    (2019)
  • G.A. Zurzolo et al.

    Self-reported anaphylaxis to packaged foods in Australia

    J. Allergy Clin. Immunol. Pract.

    (2019)
  • K. Blumchen et al.

    Modified oral food challenge used with sensitization biomarkers provides more real-life clinical thresholds for peanut allergy

    J. Allergy Clin. Immunol.

    (2014)
  • J.A. Boyce et al.

    Guidelines for the diagnosis and management of food allergy in the United States: report of the NIAID-sponsored expert panel

    J. Allergy Clin. Immunol.

    (2010)
  • A.B. Do et al.

    Undeclared food allergens and gluten in commercial food products analyzed by ELISA

    J. AOAC Int.

    (2018)
  • S. Dua et al.

    Effect of sleep deprivation and exercise on reaction threshold in adults with peanut allergy: a randomized controlled study

    J. Allergy Clin. Immunol.

    (2019)
  • A.E.J. Dubois et al.

    How does dose impact on the severity of food-induced allergic reactions, and can this improve risk assessment for allergenic foods?

    Allergy

    (2018)
  • Cited by (127)

    • Utility of food allergy thresholds

      2024, Annals of Allergy, Asthma and Immunology
    View all citing articles on Scopus
    1

    Current address: Food Allergy Research and Resource Program, University of Nebraska, Lincoln, USA.

    View full text