Risk Factors Associated with Peanut Allergy in a High Risk Infant Cohort (CoFAR)
Monday, March 5, 2018
South Hall A2 (Convention Center)
Scott H. Sicherer, MD FAAAAI, Robert A. Wood, MD FAAAAI, Tamara T. Perry, MD, Stacie M. Jones, MD, FAAAAI, Donald Y. M. Leung, MD PhD FAAAAI, Alice Henning, MS, Peter Dawson, PhD, A. Wesley Burks, MD FAAAAI, Robert W. Lindblad, MD, Hugh A. Sampson, MD FAAAAI
RATIONALE: To determine baseline demographic, clinical and serum/skin test parameters associated with development of peanut allergy (PNA) in a cohort of 3-15 month olds with likely egg/milk allergy and/or moderate-severe atopic dermatitis and a positive egg/milk skin prick test (SPT), but no known PNA.

METHODS: The primary endpoint was PNA [confirmed/convincing diagnosis at any time or last classified as serologic PNA (<2 yrs, ≥5 kUA/L, otherwise ≥14 kUA/L)] among 511 participants followed a median of 7.3 years. Univariate logistic regression was used to explore associations; baseline factors with p<0.15 were analyzed by stepwise multiple logistic regression, stratified by PNA status and run on randomized development and validation datasets.

RESULTS: 205/511 (40.1%) were classified as PNA. Baseline factors associated with PNA with p<0.01 included: increased AD severity, higher egg and peanut SPT, higher egg, milk, peanut, Ara H1, Ara H2, and Ara H3 IgE, higher peanut IgG and peanut IgG4, and increased peanut consumption during pregnancy. P-values were between 0.01 and 0.05 for younger age, non-white race, lack of breastfeeding, and increased peanut consumption during lactation. The stepwise model identified younger age at enrollment, higher peanut and Ara H2 IgE, and lack of breastfeeding as risk factors. The final model predicted 79.4% in the development and 74.8% in the validation dataset (AUC=0.83 for both). Models using stricter PNA criteria found higher Ara H2 as predictive.

CONCLUSIONS: We identified key risk factors associated with PNA, lack of breastfeeding, higher Ara H2 and peanut-specific IgE, which can be used to predict outcomes.