Bronchial allergen challenges in children – safety and predictors

BACKGROUND: In allergic asthma, the diagnosis of house dust mite (HDM) allergy is mainly based on the patient's history, allergy testing by the skin prick test (SPT) or the levels of allergen-specific IgE. We retrospectively analysed data from 350 bronchial provocations with HDM and related it to the following parameters: specific IgE, bronchial hyperresponsiveness (BHR) to methacholine testing (MCT) and exhaled NO (eNO). METHODS: Approximately 350 patients (5-18 yr of age) with allergic asthma and a positive SPT to HDMs were included. To define the sensitivity and specificity for the detection method of an early asthmatic response (EAR), a receiver-operating characteristic (ROC) curve was plotted. The accuracy was measured by the area under the ROC curve (AUC). A logistic regression model was used to predict the individual probability of a positive challenge. The results of the regression model were validated in a prospective group of n = 75 patients. RESULTS: The following cut-off values showed the best combination of sensitivity and specificity: specific IgE Dermatophagoides farinae 19.6 kU/l (AUC, 0.88), PD(20) FEV(1) 0.13 mg methacholine (AUC, 0.73) and eNO 20.1 ppb (AUC, 0.71). The following equation predicted the individual probability of a positive challenge in the retrospective and prospective group: p = 1(.) [1 + exp[-(-1.78 + 2.46.(10) log D. far - 1.25(.10) logPD(20) metha)]](-1) , (AUC = 0.88). CONCLUSIONS: The value of using the specific IgE and MCT as predictors was confirmed in a large number of patients. We also showed, for the first time, that the eNO predicted the EAR. The logistic regression model is repeatable with a good accuracy.