BackgroundResults from the POISED trial suggest that discontinuation of peanut oral immunotherapy can increase the risk of regaining clinical reactivity to peanut. ObjectiveWe sought to determine whether those who achieved sustained unresponsiveness (SU) or sustained high threshold (SHT) have different baseline sequential epitope-specific (es-) IgE profiles than those who achieved transient desensitization (TD). MethodsSubjects in the POISED trial (NCT02103270) were randomized to peanut (n=95) or placebo (n=25) for 24 months. OIT-desensitized subjects were then assigned to no peanut (PN-0, n=51) or 300mg (PN-300, n=30) for 12 months. SU and SHT were determined by those in PN-0 and PN-300, respectively, passing 4000mg peanut oral challenge. Specific IgE and IgG4 levels to peanut, Ara h 1-3 proteins and 64 allergenic epitopes were measured. We developed machine learning glmnet models with bootstrap simulations using baseline data to predict SU/SHT. ResultsEighty (84%) subjects were desensitized to peanut. Of those, 13% (n=8) and 37% (n=13) achieved SU/SHT in PN-0 and PN-300. Decreases in epitope-and protein-specific IgE levels and increases in IgG4 levels were observed during 2 years of OIT. At baseline, patients with SU in Peanut-0 but not Peanut-300 had lower es-IgE and protein-sIgE levels compared to the TD group. A machine-learning model with 12 baseline es-IgEs and age could predict SU/SHT with an accuracy of 94%, AUC 0.97, Sensitivity 1.00, Specificity 0.91. ConclusionsPatients who achieved SU/SHT have different baseline protein-and epitope-specific IgE profiles than those with TD. These profiles may help identify patients with an increased likelihood of achieving SU/SHT.
Read full abstract