Statistics analysis
Continuous variables were expressed as the mean ± standard deviation or median (quartiles [Q]1, [Q]3) depending on their distribution. Categorical variables are summarized as numbers and percentages. For continuous variables, the differences among different inflammatory asthma phenotypes and healthy were analyzed using A student’s t-test, or ANOVA if they were normally distributed, otherwise Wilcoxon rank-sum test were used to assess. Categorical data were compared using the chi-squared test or Fisher’s test. Correlations between targeted metabolites and clinical indexes in asthma patients in the validation set were expressed using Spearman’s correlation coefficients and adjusted by false discovery rate (FDR) correction (Benjamini–Hochberg method). Between two groups, targeted metabolites were examined by receiver operating characteristic (ROC) curves and the area under the curve (AUC) of each ROC was calculated to show their capability of discriminating different groups in validation set. Logistic and negative binomial regression models were established to assess the association between metabolites and severe asthma exacerbation during follow-up in the validation set. These analyses were conducted using SPSS version 25.0 (IBM Corp, Armonk, NY, USA). In all statistical analyses, a double tailed P -value of < 0.05 was considered statistically significant.