A total of 300 asthma patients and 418 healthy controls were enrolled. The average ages of asthma patients and controls were 43.6±13.48 and 44.09±13.75 years, respectively. No significant differences in sex, body mass index (BMI) and smoking history were observed between case and control groups (Table 1). Late-onset asthma (age of asthma onset ≥18 years) accounted for 74.3% in the case group. Most asthma individuals were outpatients (88.67%), and we could only get half of the patients' reports of eosinophil count, total serum immunoglobulin E (IgE), pulmonary function test and provocation or relaxation test. The other half of the patients' relevant tests were done in other hospitals, but we couldn’t acquire. 58.33% of the patients adopted the step 4 treatment plan according to Global Strategy for Asthma Management and Prevention (2018 update), 12.67% adopted step 5, 3.33% used step 3 and the other patients’ treatment information was lost.
Association analyses between CDHR3, EMSY SNPs and asthma susceptibility
The characteristics of the selected SNPs are listed in Table S1 and S2. Rs10899234 in EMSY and rs6967330 in CDHR3 were excluded due to their deviation from HWE in the control subjects (P<0.05). The genotyping assays failed for rs12155008, rs41270 and rs448024 in CDHR3.
After adjusting for confounding factors including age, sex, BMI and smoking history, four SNPs were found to be associated with asthma susceptibility (Table 2 and Figure S1). The A allele of rs3847076 in CDHR3 was associated with increased susceptibility to asthma under the additive model (P = 0.032, OR = 1.407, 95% CI: 1.030-1.923). For EMSY, both the TC/TT genotype and T allele of rs2508746 were associated with decreased risk of asthma (dominant model: P = 0.019, OR = 0.660, 95% CI: 0.465-0.935; additive model: P = 0.026, OR = 0.718, 95% CI: 0.536-0.961). The TG/TT genotype and T allele of rs12278256 were associated with reduced asthma risk (dominant model: P = 0.033, OR = 0.563, 95% CI: 0.332-0.953; additive model: P = 0.027, OR = 0.558, 95% CI: 0.332-0.937). Finally, the GG genotype of rs1892953 showed an association with increased asthma risk under the recessive model (P = 0.015, OR = 1.667, 95% CI: 1.104-2.518). After excluding people who were lack of smoking or BMI information, we used the online software SNPStats (https://snpstats.net/) for statistical analysis again and the results (shown in the Table S3) were similar to Table 1. However, it should be reminded that some significant associations maybe were expected just by chance.
Stratified analysis results by gender, smoking status, BMI status and onset age of asthma were shown in Table 3. The cut-off point of adult BMI in China is different from other countries, as 18.5≤BMI < 24kg/m2 meaning normal weight range and BMI≥24kg/m2 meaning overweight or obese. Allele A of rs3847076 was associated with increased susceptibility to asthma in male subgroup, smoking subgroup, BMI < 24kg/m2 subgroup and late onset asthma subgroup (P=0.023, OR=1.869; P=0.009, OR=2.168; P=0.005, OR=1.835 and P=0.023, OR=1.457, respectively). Similarly, rs2508746 TC+TT was related with decreased asthma susceptibility in the non-smoking subgroup, non-overweight subgroup, and late-onset asthma subgroup in dominant model (P=0.014, OR=0.618; P=0.027, OR=0.612 and P=0.016, OR=0.637, respectively). Meanwhile, rs1892953 GG shown increased risk of asthma in the female subgroup, non-smoking subgroup, non-overweight subgroup, and late onset asthma subgroup in recessive model (P=0.038, OR=1.738; P=0.04, OR=1.615; P=0.017, OR=1.910 and P=0.017, OR=1.680, respectively). Rs12278256 T was still associated with decreased asthma susceptibility in female subgroups, non-smoking subgroups, and non-overweight subgroups in additive model (P=0.032, OR=0.465; P=0.02, OR=0.508 and P=0.028, OR=0.481, respectively). The interaction between these four SNPs and smoking, sex and BMI were shown in Table S4. We got significant interaction between rs3847076 and rs1892953 and smoking, sex and BMI, while no interaction was found between rs12278256 and these clinical phenotypes. Meanwhile, significant interaction could also be observed between rs2508746 and either gender or BMI.
We further explored the relationship between eosinophil count, total serum IgE, pulmonary function test of asthma patients and gene variants. Eosinophil count was higher in asthma patients with genotype CC of rs3847076 comparing to individuals with genotype CA (Table S5). Total IgE was related with four variants of CDHR3 and one variant of EMSY (Table S6). Both FEV1% predicted and FEV1/FVC% were significant different in nine SNP genotypes, including rs2508746 and rs1892953. Higher FEV1/FVC% was also seen in genotype GG of rs12278256 (Table S7). Due to the small number of samples, further verification research is needed.
Haplotype and LD analysis
The LD between SNPs of CDHR3 and EMSY was low and those SNPs were divided into eight haplotype blocks with Haploview software (Figures 1 and 2). Only the haplotype consisting of GATCTGAGT in block 1 of EMSY was associated with decreased risk of asthma (P = 0.037, OR = 0.615, 95% CI: 0.388-0.975) (Table 4).
Functional prediction results
Four statistically significant SNPs were predicted using the software RegulomeDB and Haploreg v4 (Table S8). Rs144934374 is strongly linked to rs12278256 and its RegulomeDB scores is lower than that of rs12278256, suggesting that it may be the functional site represented by rs12278256. Acting as promoter histone marks or enhancer histone marks, or affecting DNAse is suggested to be associated with chromatin status, and binding proteins or altering regulatory motifs in ChIP-Seq suggest that transcription levels may be affected. It seems that these four SNPs may have certain effects on chromatin status and transcription level. Rs1892953 appears as an expression quantitative trait loci (eQTL) SNP in thyroid tissue .