A Streamlined Strategy for Basophil Activation Testing in a Multicenter Phase III Clinical Trial

Background: The basophil activation test (BAT) has been limited to research settings owing to technical issues. Novel approaches using dry, ready-to-use reagents and streamlined protocols offer greater flexibility and may open opportunities for easier implementation in clinical research.

Objective: Using a streamlined basophil activation test (sBAT) strategy and the settings of the baseline study of the Epicutaneous Immunotherapy in Toddlers with Peanut Allergy (EPITOPE) trial of EPicutaneous ImmunoTherapy, we aimed to assess the feasibility of implementing BAT in a multicenter trial and to evaluate its utility in predicting the outcomes of peanut double-blind placebo-controlled food challenge (DBPCFC).

Methods: Whole blood samples were collected from subjects aged 1 to 3 years (n = 241) undergoing baseline eligibility DBPCFC in the EPITOPE study across 15 clinical sites in North America. After preparation with sBAT reagents, processed samples were analyzed in a single central laboratory within 5 days of collection and preparation. The eliciting dose (ED) at DBPCFC was determined using, Practical Allergy (PRACTALL) criteria. Using a machine learning approach that incorporated BAT-derived features, clinical characteristics, and peanut-specific immunoglobulin E, the ability to predict outcomes of interest (ED ≤ 300 mg or > 300 mg] and use of epinephrine) was assessed using data randomly split into training (n = 182) and validation (n = 59) subsets.

Results: The expression of basophil activation markers CD203c and CD63 correlated with ED and severity outcomes of DBPCFC. Most informative concentrations of peanut extract in the sBAT assay for these associations were 1 ng/mL and 10 ng/mL. Using machine learning to assess the ability to predict the outcomes of DBPCFC, the best models using only the BAT-derived features provided relatively high sensitivities of 0.86 and 0.85 for predicting ED and epinephrine use, respectively, whereas specificities were lower, ranging from 0.60 to 0.80. Although including specific immunoglobulin E and skin prick test data in addition to those from sBAT did not improve the ability to identify individuals most at risk for severe reactions, it did improve the ability to identify patients with an ED greater than 300 mg.

Conclusions: In addition to facilitating implementation in multicenter trials, sBAT retains the potential of BAT to characterize allergic patients and confirms its potential to contribute to predicting the outcome of oral food challenges.