Potential correlation between DMSs of aging-related genes and clinical index of asthma
To further assess whether the differential methylation of the 9 aging-related genes is related to the occurrence and severity of asthma, we detected the correlation between the differential 68 DMSs in aging-related genes and the lung function indicators of asthma patients. The results demonstrated that there were 10 DMSs significantly were associated  with lung function. The maximum correlation coefficient for each DMSs was presented in Figure 3. The remaining correlation analysis data was showed in Figure S1. For these 10 DMSs, three DMSs (Chr4:75310649, Chr6:108883024, Chr17:7591672) were closely related to at least three clinical indicators. In addition, other three DMSs (Chr4:75310649, Chr20:32274088, Chr6:108882977) were related to two clinical indicators. It has also been shown that the correlation coefficients of the 10 DMSs were all greater than 0.38 withp -value < 0.05. It was also particularly noteworthy that Chr17:7591672 was closely related to four lung function indicators (FVC, FEV1, PEF, FEF25), with a correlation coefficient of 0.671 and a p -value equal to 0.0001. These data strongly suggested that theses differential DNAm of the specific aging-related DMSs may influence the occurrence and severity of asthma. The complete data for the 10 DMSs and clinical indicators were showed in Table 3.
Feasibility ofcandidate DMSs as biomarkers of asthma
Since the differential 10 DMSs have been confirmed to be closely associated to the clinical lung function of asthma patients, we further evaluated their potential as biomarkers for asthma patients. First, ROC analysis of the methylation levels of each candidate DMS was performed. The areas under the curve (AUC) of 9 DMSs (p -value< 5%) were between 65.3% and 76.3%, and the AUC of 6 DMSs was greater than 70% (Figure 4A and Table 4). Besides, logistic regression was conducted and the ROC of 9 candidate DMSs showed that the AUC of the predicted probability of the 9 candidate DMSs was as high as 95.4%, and the result was statistically significantly (p -value < 0.1%, Figure 4B). These results indicated that the 9 candidate DMSs had the potential value to be the biomarkers for asthma. Meanwhile, to verify the above results, PCA analysis consisting of 9 candidate DMSs was executed. The result revealed that the methylation levels of the total 9 DMSs could effectively distinguish asthma patients from HCs (Figure 4C).
To better understand the possible value of the 9 DMSs, we further calculate the methylation change rate of the 9 DMSs in HCs and asthma patients, which is a description of the possibility of methylation status alteration. Then, the status of the changed methylation or unchanged methylation was determined using the optimal cutoff value. The optimal cutoffs of the 9 DMSs were calculated according to the Youden index which was presented in Table 3. The methylation change rate of HCs and asthmatic patients were included in Figure 5. Specially, the methylation change rate of the total 9 DMSs in HCs showed a significant decreasing trend, whereas significantly increased methylation change rate was observed in asthma patients (Figure 5A). The methylation change rate of the total 9 DMSs in asthma was 33.3% ~ 100%, and the rate in HCs was only 0 ~ 55.6%. Notably, the change rate of a single DMS in asthma patients was between 47.27% and 89.09%, while it was 1.96% ~ 41.17% in HCs (Figure 5B). Similarly, asthma patients had a higher rate of methylation change. Statistical results showed that the methylation change rate of the total 9 DMSs was significantly increased in asthma patients (p -value < 0.1%, Figure 6A). In addition, ROC analysis was implemented according to the methylation change rate of the 9 DMSs in all samples (Figure 6B) and there was a higher AUC compared to previous method (AUC=0.98). Moreover, the PCA analysis results also indicated that the methylation change rate of 9 DMSs could better distinguish asthma patients from HCs (Figure 6C).