Gender, prick test size and rAra h 2 sIgE level may predict the eliciting dose in patients with peanut allergy: Evidence from the Mirabel survey

Summary Background Peanut allergy management is based on active avoidance and access to emergency treatment including self‐injectable adrenaline. Knowing the dose at which a patient is likely to react is crucial for risk assessment and could significantly improve management by integrating a personalized approach. Objective To develop a threshold dose distribution curve model from routinely collected data. Methods The MIRABEL survey is an observational study of 785 patients with peanut allergy/sensitization conducted in France, Belgium and Luxemburg. The current analysis included the 238 participants for whom medical and oral food challenge data were available. Several statistical models (Kaplan‐Meier, Cox model, Weibull and Lognormal with predictive factors, basic Weibull and Lognormal) were compared to select the best model and predictive factor combination associated with the threshold doses. Inferences were made with a Bayesian approach. Results Patients were mainly children (mean age: 9 years [IQR: 6‐11]; 87% < 16 years) and males (62%). Median Ara h2 s IgE was of 8kUA/L [IQR: 1‐55] and median skin prick test size of 10 mm [IQR: 7‐13]. OFC was positive in 204 patients (86%). The median threshold dose was of 67 mg of peanut protein [IQR: 16‐244]. The dose at which 1% of the patients are likely to react with objective symptoms was 0.26 [0.03; 2.24] mg of peanut protein. Gender, size of the skin prick test (SPT) and Ara h 2 specific IgE level had a significant impact on the threshold dose distribution curve. The Cox model was the most effective to predict threshold doses with this combination of factors. Girls react to lower doses than boys with a beta coefficient associated to the risk and a 95% credible interval of 0.44 [0.04; 0.77]. The higher the size of the SPT and the Ara h 2 specific IgE level are, the higher the risk of reacting to a small amount of peanut, with beta coefficients associated to the risk and 95% credible intervals of 0.05 [0.02; 0.08] and 0.01 [0.01; 0.02], respectively. Conclusion and clinical relevance According to the model, routinely collected data could be used to estimate the threshold dose. The consequences could be the identification of high‐risk patients who are susceptible to react to small amounts of peanut and a personalized management of peanut allergy integrating the risk of allergic reaction. Limitations of this study are that assessors of OFC outcome were aware of SPT and Arah2 results, and a further validation study