The project was approved by the medical ethics committee of the Chinese Medicine Hospital of Xinjiang Uygur Autonomous Region., All participants signed an informed consent form before the experiment. Data on environment, occupation, smoking history, tumors and general situation, according to the unified epidemiological questionnaire, was obtained from all patients. In this study, according to the GOLD (Global Chronic Obstructive Pulmonary Disease Action) report,21 315 clinical patients diagnosed with COPD were included in the case group, of which 147 were smokers (smoking more than two cigarettes per day, or had quit smoking before enrollment) and, 166 people were non-smokers (never smoked). The diagnosis of COPD by hospital pulmonologists is based on clinical examination and spirometry. The severity of COPD is classified according to fixed forced expiratory volume in 1 s (FEV1) / forced vital capacity (FVC <0.7) and the predicted value of FEV1. The healthy controls were 314 healthy subjects from the same geographic area with normal lung volume according to the GOLD standard, including 52 smokers and 118 non-smokers. Subjects with other lung diseases, infections, inflammation, endocrine, gastrointestinal, liver, kidney and tumor diseases, and excessive alcohol consumption (≥ 40 g /day) were excluded. Peripheral venous blood (5 ml) was collected in a test tube containing EDTA, centrifuged, and stored at - 80 ℃ for subsequent use.

To verify the effectiveness of the genotyping of the selected population, the SNPs on chromosomes 11 and 7 were based on the smallest level according to the CHB (Beijing Han nationality) data in the Thousand Genome Project (http://www.internationalgenome.org/) SNPs with locus frequency less than 5% were initially screened. Subsequently, the HaploReg v4.1 (https://pubs.broadinstitute.org/mammals/haploreg/haploreg.php) was used to predict the potential functions of selected SNPs, which are related to promoter histone marks, enhancer histone marks, DNAse, proteins bound and selected eQTL hits, indicating the mechanisms by which these SNPs may affect the disease occurrence and development. Finally, we used a few previously reported SNPs located adjacent to LOC105375199 (rs10245353 and rs2290263), Loc105375199 (rs4719841 and rs7780562) and MIR4300HG (rs7934083 and rs78750958) as candidate loci for this study. Genomic DNA was extracted according to the GoldMag® nanoparticles method and its concentration was measured using the NanoDrop 2000. We conducted Multiplexed SNP MassEXTEND experiment, SNP genotyping and data management and analysis using the Sequenom MassARRAY Assay Design 3.0 software, Sequenom MassARRAY RS1000, and Sequenom Typer 4.0 software, respectively, as reported previously.22,23

To analyze the mass spectrometry results, the undetected, repeated data and the success rate of typing 90% of the sites or samples were deleted. This study used Welch's t-test to analyze the average age between the cases and control groups. Hardy Weinberg equilibrium (HWE) analysis is a method of quality control of the genetic data of the control group. Theoretically, the control group should follow the Hardy Weinberg equilibrium, p > 0.05, indicating that the control group reaches genetic balance, ensuring that the data of the control group is credible. Fisher precision test and Wald test were used to compare the allele/genotype frequency of each locus between the cases and the controls, and genetic model analysis was performed using SNPStats software (http://bioinfo.iconcologia.net/snpstats/start.htm) to assess the association between the SNPs and COPD risk. Multi-factor dimensionality reduction (MDR) analysis is a nonparametric, nongenetic approach suitable for case-control studies that only require individual genetic data (such as SNPs) and no other special conditions to analyze gene-gene interactions are needed.24,25 One-way analysis and ANOVA were used to analyze the differences in clinical characteristics between the different genetic models. Multiple comparisons were Bonferroni-corrected.26

The basic information of the participants in this study is shown in Table 1. A total of 629 subjects, 315 and 314 patients in the COPD and control groups, respectively, were included in this study. There were 239 men and 76 women in the COPD group and 237 men and 77 women in the control group. There were no significant differences in age, gender, smoking, complication, wheezing, chest tightness, recent respiratory infection status and BMI between the two groups. Detailed information such as the average value and standard deviation value of patients’ respiratory rate (per minute), pulse rate (pe minute), forced vital capacity (L), quantity per second (L), and rate per second (s) are shown in Table 1.

The basic information of all the SNPs is shown in Table 2, including SNP-ID, gene, chromosomes, position, alleles, minor allele frequency (MAF), p-HWE, OR (95%CI) and p-value. After Bonferroni correction, the results showed that individuals with the A allele of rs4719841 had a lower risk of COPD than those with the G allele (p < 0.0001). Individuals with the G allele of rs78750958 had a lower risk of COPD than those with the A allele (p < 0.0001). HaploReg software function prediction showed that the selected SNPs (rs10245353 rs2290263, rs4719841, rs7780562, rs7934083 and rs78750958) are related to siPhy cons, DNAse, motif changes, selected eQTL hits, promoter histone marks, enhancer histone marks, and proteins, suggesting that these SNPs may affect the occurrence and development of the disease. A multi-genetic model was used to analyze the correlation between the selected SNPs and the risk of COPD, and the results showed that rs4719841 and rs7934083 were related to the risk of COPD (p < 0.0017) (Table 3). Specifically, the A/A genotype of rs4719841 was observed to reduce the risk of COPD. In the co-dominant, dominant and additive models, rs7934083 reduced the risk of COPD.

To further clarify the correlation between the six selected SNPs and the risk of COPD, we conducted 10 stratified analyses, including stratification by age, gender, complications, BMI, asthma, recent respiratory infection, smoking, wheezing, chest tightness, and GOLD spirometric grade (Supplementary Table 1). The results showed that COPD risk due to some of the selected SNPs was associated with age, gender, and non-smoking categories, whereas in case of other selected SNPs, COPD risk was not associated with stratified analyses.

In subjects aged > 70 years, we found that rs7934083 was associated with reducing COPD risk (p < 0.0017), while in subjects aged ≤ 70 years, rs4719841 was associated with a reduction in the risk of COPD (p < 0.0017) (Table 4). The results of the male stratification analysis showed that rs4719841 reduced the risk of COPD in the allele, collinearity model, additive model and recessive model (p < 0.0017). The results of the male stratification analysis also showed that individuals with the T allele rs7934083 had a lower risk of COPD than individuals with the C allele (p = 0.0007). In the female stratification analysis, we only observed that rs7934083 was associated with a reduced risk of COPD (p < 0.0017). In the stratified analysis of non-smokers, we observed that the other five selected SNPs were not related to the risk of COPD except for rs4719841, which is related to a reduced risk of COPD (Table 5).

To better understand the impact of selected SNP interactions on COPD risk, we conducted MDR analysis (Supplementary Table 2 and Fig. 1). The dendrogram and frauctman-reiringold represent the interactions between these SNPs. In Fig. 1(A), the connection between nodes is shorter, indicating that the redundant interaction is stronger. A negative double-track entropy in Fig. 1(B) indicates that it is an antagonistic effect. A positive value indicates a synergistic effect. The results showed that there was a specific antagonism between the selected genes. Supplementary Table 2 shows the model combination of candidate SNP interactions for the COPD risk assessment. Among the many locus models, the best combination is a three-locus model consisting of rs4719841, rs7934083 and rs78750958, which increases the risk of COPD (test accuracy=0.618, CVC=10/10, OR=2.93, 95% CI: 2.11 – 4.05, p < 0.001).3.4 3.5 Genetic models and clinical characteristics.

Figure 1.

The dendogram (A) and fruchterman Rheingold (B) of SNP-SNP interaction for COPD risk. (A) Short connections among nodes represent stronger redundant interactions. (B) Negative percent entropy indicates redundancy.

(0.26MB).

In addition, we explored the relationship between different SNP models and the clinical indices of patients, including respiratory rate (times per minute), pulse rate (times per minute), forced vital capacity (L), one-second volume (L), and one- second rate (S). There was no significant correlation between the genetic models of rs10245353, rs2290263, and rs78750958 and the above clinical parameters (p > 0.05). However, the genetic models of rs4719841, rs7934083, and rs7780562 are related to forced vital capacity, rate per second, and respiratory rate/forced vital capacity of COPD patients (p < 0.05), as shown in Table 6.