To analyze the predictive value of the fractional of exhaled nitric oxide (FeNO) and blood eosinophil (B-Eos) counts on the severity of airway hyperresponsiveness in asthma patients, then explore a prediction model for the severity of AHR.

This study retrospectively collected 1347 patients diagnosed with asthma in our hospital from January 2014 to December 2019, and identified a cohort of 520 patients who had simultaneous completed datasets of FeNO and B-Eos. According to the methacholine challenge test (MCT) results, the population was divided into severe AHR group (MCT is moderate or severely positive, n=183) and mild AHR group (MCT is very mild or slightly positive, n=337). The differences in demographics, lung function, FeNO and B-Eos are analyzed between these two groups. Logistic regression is used to construct a multi-factor regression model, then the risk of severe AHR is displayed by nomogram and forest chart.

FeNO and B-Eos in the severe AHR group were significantly higher than those in the mild AHR group (73 vs. 36 ppb, 394 vs. 243 cells/μl, P<0.001). Logistic regression showed that age, gender, FEV₁/FVC ratio, B-Eos, and FeNO were independent risk factors for severe AHR. The model incorporating these risk factors has a sensitivity of 49.7% and a specificity of 87.8%. The receiver operating characteristic (ROC) curve analysis shows that the AUC of the regression model is significantly higher than that of FeNO or B-Eos alone (0.797 vs. 0.715 or 0.644, P<0.001). When comparing the risk of having severe AHR in different subgroups, the adjusted odds ratio (aOR) of having severe AHR elevated progressively with the gradual increase in FeNO or B-Eos (P<0.001). While, the multivariable aOR of having severe AHR was 1.57 for females (P=0.041), 3.38 for patients with lower FEV₁/FVC ratio (<70%, P<0.001).

FeNO or B-Eos alone has moderate diagnostic accuracy for predicting severe AHR. The nomogram constructed by the multi-factor regression model can be used to predict the probability of severe AHR.