Article Text

Download PDFPDF
Original research
Respiratory traits and coal workers’ pneumoconiosis: Mendelian randomisation and association analysis
  1. Ting Wang1,
  2. Wenqing Sun2,
  3. Hongyan Wu1,
  4. Yuxin Cheng3,
  5. Yan Li2,
  6. Fanqing Meng1,
  7. Chunhui Ni2
  1. 1 Pathology, Nanjing Drum Tower Hospital, The Affiliated Hospital of Nanjing University Medical School, Nanjing, Jiangsu, China
  2. 2 Department of Occupational Medicine and Environmental Health and Key Laboratory of Modern Toxicology of Ministry of Education, Nanjing Medical University, Nanjing, Jiangsu, China
  3. 3 Comprehensive Cancer Centre, Nanjing Drum Tower Hospital, The Affiliated Hospital of Nanjing University Medical School, Nanjing, Jiangsu, China
  1. Correspondence to Professor Chunhui Ni, Department of Occupational Medicine and Environmental Health and Key Laboratory of Modern Toxicology of Ministry of Education, Nanjing Medical University, Nanjing 210008, Jiangsu, China; chninjmu{at}126.com; Dr Ting Wang; wangti08{at}163.com

Abstract

Objectives Susceptibility loci of idiopathic pulmonary fibrosis and chronic obstructive pulmonary disease were also significantly associated with the predisposition of coal worker’s pneumoconiosis (CWP) in recent studies. However, only a few genes and loci were targeted in previous studies.

Methods To systematically evaluate the genetic associations between CWP and other respiratory traits, we reviewed the reported genome-wide association study loci of five respiratory traits and then conducted a Mendelian randomisation study and a two-stage genetic association study.

Results Interestingly, we found that for each SD unit, higher lung function was associated with a 66% lower risk of CWP (OR=0.34, 95% CI: 0.15 to 0.77, p=0.010) using conventional Mendelian randomisation analysis (inverse variance weighted method). Moreover, we found susceptibility loci of interstitial lung disease (rs2609255, OR=1.29, p=1.61×10−4) and lung function (rs4651005, OR=1.39, p=1.62×10−3; rs985256, OR=0.73, p=8.24×10−4 and rs6539952, OR=1.28, p=4.32×10−4) were also significantly associated with the risk of CWP. Functional annotation showed these variants were significantly associated with the expression of FAM13A (rs2609255, p=7.4 ×10−4), ANGPTL1 (rs4651005, p=5.4 ×10−7), SPATS2L (rs985256, p=1.1 ×10−5) and RP11-463O9.9 (rs6539952, p=7.1 ×10−6) in normal lung tissues, which were related to autophagy pathway simultaneously according to enrichment analysis.

Conclusions These results provided a deeper understanding of the genetic predisposition basis of CWP.

  • coal dust
  • respiratory
  • genetic susceptibility

Statistics from Altmetric.com

Key messages

What is already known about this subject?

  • With the strategy of genome-wide association study (GWAS), several susceptibility loci of CWP had been identified in Chinese Han population.

  • Recently, several studies have shown that susceptibility loci associated with idiopathic pulmonary fibrosis and chronic obstructive pulmonary disease were also significantly associated with the predisposition of pneumoconiosis.

  • This brings new insights to the research approach of CWP, because it is usually difficult to recruit enough cases of CWP compared with the GWAS of other complex diseases.

What are the new findings?

  • We found each SD higher lung function was associated with a 66% lower risk of CWP (OR=0.34, 95% CI: 0.15 to 0.77, p=0.010) using conventional Mendelian randomisation analysis (inverse variance weighted method).

  • Moreover, we found susceptibility loci of interstitial lung disease (rs2609255, OR=1.29, p=1.61×10−4) and lung function (rs4651005, OR=1.39, p=1.62×10−3; rs985256, OR=0.73, p=8.24×10−4 and rs6539952, OR=1.28, p=4.32×10−4) were also significantly associated with the risk of CWP.

  • Functional annotation showed these variants were significantly associated with the expression of FAM13A (rs2609255, p=7.4 ×10−4), ANGPTL1 (rs4651005, p=5.4 ×10−7), SPATS2L (rs985256, p=1.1 ×10−5) and RP11-463O9.9 (rs6539952, p=7.1 ×10−6) in normal lung tissues, which were related to autophagy pathway simultaneously according to enrichment analysis.

How might this impact on policy or clinical practice in the foreseeable future?

  • Our findings may shed a light on the aetiology and underlying biological mechanisms of CWP and might have crucial relevance to the identification of high-risk individuals with occupational or environment dust exposure for intervention.

Introduction

Pneumoconiosis is the most common occupational disease in China. According to the latest official data on the country’s occupational health status from the National Health Commission, the total number of reported occupational cases up until 2018 was 97 500, and 90% of reported occupational diseases were pneumoconiosis. Furthermore, >200 million workers are exposing to multiple occupational health risks, including dust, chemicals and poison.1 Occupational prevention and control of pneumoconiosis has become an important challenge for China currently.

Coal worker’s pneumoconiosis (CWP) is the main subtype of pneumoconiosis in China, accounting for nearly 50% of pneumoconiosis cases.2 Although occupational inhalation of silica or free crystalline silicon dioxide is considered as the primary aetiological factor, only a small proportion of workers develop CWP at a similar cumulative exposure, indicating that genetic predisposition of CWP may exist. With the strategy of genome-wide association study (GWAS), several susceptibility loci of CWP had been identified in Chinese Han population.3 4 Recently, several studies have shown that susceptibility loci associated with idiopathic pulmonary fibrosis (IPF) and chronic obstructive pulmonary disease (COPD) were also significantly associated with the predisposition of pneumoconiosis.5 6 This brings new insights to the research approach of CWP, because it is usually difficult to recruit enough cases of CWP compared with the GWAS of other complex diseases. Systematic evaluation the associations of susceptibility loci from other respiratory traits may help us better understand the genetic susceptibility basis of CWP.

In the current study, we first systematically reviewed the reported associations of IPF, interstitial lung disease (ILD), COPD, lung function (LF) and asthma. Then, we conducted a two sample Mendelian randomisation (MR) study based on these associations and our previous CWP GWAS study. Besides, we also evaluated the associations of these loci with the GWAS data and validated the promising associations in an independent cohort study.

Methods

Genetic variants associated with five respiratory traits

We systematically collated reported non-tumour-related complex traits in the respiratory system from GWAS Catalogue by 30 September 2019. Then five respiratory traits which included >1000 cases in the screening stage were kept for further analysis, including IPF, ILD, COPD, LF and asthma. For each trait, the associations were mainly derived from the GWAS (or GWAS meta-analysis) with the largest sample size. A new GWAS study on ILD was published when we integrated the results, which was also reintegrated into this study.7 The effect sizes and corresponding p values were extracted for variants satisfying GWAS significance (p<5×10–8) from the original publications. As a result, we derived 129 single nucleotide polymorphisms (SNPs) for asthma from Zhu et al,8 81 SNPs for COPD from Sakornsakolpat et al,9 16 SNPs for IPF from Allen et al,10 17 SNPs for ILD from Fingerlin et al 11 and Hobbs et al 7 and 267 SNPs for LF from Shrine et al.12

Genetic associations of SNPs with CWP

Summary statistics on CWP, including OR estimates and SEs for the above instrumental SNPs, were derived from our previous GWAS scan in 202 CWP cases and 198 exposed controls, which were recruited from 4 coal mines from 2006 to 2012 in Jiangsu. Imputation was performed with SHAPEIT V213 and IMPUTE2 (http://mathgen.stats.ox.ac.uk/impute/impute_v2.html) taking the 1000 Genome Project Phase III as the reference set.14 The association analysis was conducted with PLINK V.1.9 (https://www.cog-genomics.org/plink2/) assuming an additive model.15 Significant eigenvectors, accumulating time of exposure to dust, job type as well as smoking status were adjusted in the analysis.

Mendelian randomisation

We performed MR analyses for asthma, COPD, IPF, ILD and LF on CWP risk separately to estimate whether people with these complex respiratory system traits are more susceptible to CWP. For each trait, only the most significant variant was kept in each linkage disequilibrium (LD) region (r2 ≥0.1). For instrumental SNPs which were absent in the CWP dataset, proxy SNPs (r2 >0.8) from the 1000 Genomes Project were used as a substitute. However, several instrumental SNPs were monomorphic in the CWP GWAS data, which were probably due to the heterogeneity among populations. As a result, we kept 89 instrumental SNPs for asthma, 65 instrumental SNPs for COPD, 12 instrumental SNPs for IPF, 17 instrumental SNPs for ILD and 220 instrumental SNPs for LF. We performed the MR analysis with random-effects inverse variance weighted (IVW) meta-analysis of the Wald ratio, weighted median method as well as the MR-Egger regression simultaneously.16 Besides, we also evaluated the directional pleiotropy based on the intercept obtained from the MR-Egger analysis, and whether the MR estimate was driven or biassed by a single SNP with a leave-one-out analysis.

Study participants and genotyping

To identify additional susceptibility loci of CWP, we also systematically evaluate the GWAS reported associations of other respiratory traits in our CWP cohort. For variants significantly associated with CWP in the GWAS scan (p<0.05), we replicated the associations in an indecent sample set including 703 CWP cases and 705 exposed controls. The participants were unrelated ethnic Han Chinese, which were recruited from the coal mines of Xuzhou Mining Business Group in Jiangsu. All of the subjects were underground coal miners with occupational exposure to silica. Demographic information of the subjects was summarised in online supplemental table 1. A written informed consent as well as a sample of 5 mL whole blood were obtained from each subject before taking part in the research.

Supplemental material

Genotyping was performed using the TaqMan allelic discrimination assay on the ABI 7900 system (Applied Biosystems, Foster City, California, USA), and the genotypes were called using the SDS 2.3 Allelic Discrimination Software (Applied Biosystems) by technicians who were blinded to the case-control status. The primers and probes are available on request if necessary. We took several measures for quality control, including: (i) mixing case and control samples on each plate in a blinded fashion; (ii) including two water controls in each plate as blank control; (iii) selecting 5% of the samples for replicated genotyping.

Public database

To explore the potential mechanisms of the significant loci, HaploReg V.4.1 (https://pubs.broadinstitute.org/mammals/haploreg/haploreg.php) was used to annotate the identified variants as well as their high LD variants (pairwise with r2 ≥0.6). Furthermore, the GTEx Portal V8 (http://www.gtexportal.org/home/) was used for gene co-expression analysis and the expression quantitative trait loci (eQTL) analysis.17 The normalised RNA-seq expression data of 578 normal lung tissues were available from GTEx V8, of which 515 samples had released whole-genome sequencing genotypes. The Kyoto encyclopaedia of genes and genomes database (KEGG, http://www.genome.jp/kegg/pathway.html/) were used to define human biological pathways.18

Statistical analysis

Logistic regression was used to calculate the ORs and their 95% CIs between SNPs and risk of CWP. In the replication stage, job type, accumulating time of exposure to dust and smoking status were adjusted as covariates. To combine results from GWAS and replication stage, we used meta-analysis with a fixed-effect model if there was no heterogeneity (p value for heterogeneity >0.05); otherwise, we used random-effects model. The heterogeneity between groups was calculated with χ2-based Cochran’s Q statistics. A linear regression model was used to perform a genome-wide co-expression analysis. Then KEGG enrichment analysis was used to explore the potential mechanisms based on the genes significantly co-expressed with the identified susceptibility genes (defined as pFDR <0.05 at a false discovery rate). The enrichment analysis was performed using the ‘clusterProfiler’ R package. All the above analyses were performed using R (V.3.6.1).

Results

After evaluating the associations between genetically predicted respiratory traits and CWP risk, we found higher LF was associated with significantly lower odds of CWP (table 1). Using conventional MR analysis (IVW method), we found that with every SD, higher LF was associated with a 66% lower risk of CWP (OR=0.34, 95% CI: 0.15 to 0.77, p=0.010). Consistent associations were observed using MR-Egger method (OR=0.11, 95% CI: 0.02 to 0.61, p=0.013), and weighted median method (OR=0.21, 95% CI: 0.05 to 0.92, p=0.039) in terms of direction and magnitude (figure 1). No significant heterogeneity and pleiotropy were observed for the association between LF and CWP. There was no single SNP driving the overall effect of LF on CWP according to the funnel plot and the leave-one-out sensitivity analysis (online supplemental figure 1). However, we did not observe the associations between other respiratory traits and CWP risk.

Table 1

Mendelian randomisation estimates of the associations between five respiratory traits and risk of coal worker’s pneumoconiosis

Figure 1

Scatter plot of each single nucleotide polymorphism (SNP) associated with lung function and the risk of coal worker’s pneumoconiosis (CWP). Scatter plots show the per-allele association with CWP risk plotted against the per-allele association with lung function (with vertical and horizontal black lines showing 95% CI for each SNP). The scatter plot is overlaid with the Mendelian randomisation estimate of the effect of lung function on worker’s pneumoconiosis risk.

The estimates of each of the instrumental SNPs were shown in online supplemental table 2, and we observed 11 SNPs significantly associated with the risk of CWP in the GWAS scan. After evaluating the associations with an independent cohort with 703 CWP cases and 705 exposed controls (table 2 and online supplemental table 3), we found four SNPs were consistently associated with the risk of CWP (OR=1.27, p=1.52×10−3 for rs2609255; OR=1.32, p=0.016 for rs4651005; OR=0.75, p=7.76×10−3 for rs985256 and OR=1.24, p=6.64×10−3 for rs6539952). After combining results of the two stages, we found that rs2609255 (4q22.1, previously reported in ILD, OR=1.29, p=1.61×10−4), rs4651005 (1q25.2, previously reported in LF, OR=1.39, p=1.62×10−3), rs985256 (2q33.1, previously reported in LF, OR=0.73, p=8.24×10−4) and rs6539952 (16q24.1, previously reported in LF, OR=1.28, p=4.32×10−4) were significantly associated with the risk of CWP (table 2). In the subgroup analysis, we did not observe obvious heterogeneity among different subgroups for these variants (online supplemental table 4).

Table 2

Association details of the identified variants associated with coal worker’s pneumoconiosis risk

To evaluate the joint effect of the four newly identified CWP variants, we performed a joint effect analysis according to the numbers of risk alleles. As expected, we observed that the risk of CWP increased significantly along with the number of risk alleles in an allele-dosage manner (ptrend=2.65×10−10) (table 3). Compared with individuals with 0–2 risk alleles, those carrying 6–8 risk alleles showed a 3.72 times higher risk of CWP (OR=3.72, 95% CI: 2.41 to 5.82, p=4.86×10−9).

Table 3

Joint effect of risk alleles for rs2609255, rs4651005, rs985256 and rs6539952 on pneumoconiosis risk

An in silico validation was then performed to explore the potential mechanisms of the identified four variants. None of the neighbouring SNPs within the same LD block (pairwise r2 >0.6) of the four variants was located at exon regions, while several variants were located at pivotal regulatory elements (including promoter, enhancer, transcription factor binding sites and DNases) (online supplemental table 5), which indicated that regulating gene expression might be the main mechanism for these SNPs.

For each SNP, we performed an eQTL analysis for genes within 1 Mb according to GTEx V8. We found that rs2609255 was significantly associated with the expression of FAM13A (p=7.4 ×10−4) in normal lung tissues; significant associations were also observed for rs4651005-ANGPTL1 (p=5.4 ×10−7), rs985256-SPATS2L (p=1.1 ×10−5) and rs6539952-RP11-463O9.9 (p=7.1 ×10−6) (figure 2). KEGG enrichment analysis was then performed according to the genes significantly co-expressed with the four genes based on the expression data from GTEx V8. Interestingly, we found that genes co-expressed with FAM13A, ANGPTL1, SPATS2L and RP11-463O9.9 were enriched in autophagy pathway simultaneously (online supplemental figure 2).

Figure 2

Expression quantitative trait loci analysis of the identified four variants. (A) rs2609255-FAM13A, (B) rs4651005-ANGPTL1, (C) rs985256-SPATS2L and (D) rs6539952-RP11-463O9.9. The x-axis represents different alleles of the identified single nucleotide polymorphism, and the y-axis shows the normalised expression of corresponding gene.

Discussion

Besides exposure to free crystalline silica, genetic susceptibility also played an important role in the development of pneumoconiosis. Even though studies have shown susceptibility loci of some respiratory traits are also associated with the risk of CWP, previous studies usually directed at a few sites. In this study, we systematically evaluated the associations between other respiratory traits and the risk of CWP through an MR and association analysis. We found that LF was inversely related to the risk of CWP overall; besides, susceptibility loci of ILD (rs2609255-FAM13A) and LF (rs4651005-ANGPTL1, rs985256-SPATS2L and rs6539952-RP11-463O9.9) were also associated with the susceptibility of CWP.

Although ILD and IPF are mainly characterised by pulmonary fibrosis, which is similar to CWP, we did not observe associations between these traits in the MR study, which indicating the different aetiologies and pathogenesis. Instead, we found individuals with lower LF were more likely to get CWP under the similar coal dust exposure. Bodies of evidences have demonstrated the dose-response relationships of cumulative coal mine dust exposure and LF impairment.19 With an MR design,20 our research suggested low basic LF might be one of the risk factors of CWP after exposure to dust. This was probably due to higher LF usually associated with better mucociliary clearance.21 The result indicated that LF evaluation is necessary in dust-exposed workers.

The variant rs2609255 located at 4q22 and was first reported as a susceptibility locus of IPF in non-Hispanic white subjects, which was also replicated in the Japanese population.22 Recently, Wang et al found the variant was also associated with the susceptibility of silicosis with a case control study including 177 cases and 204 controls in a Chinese population.5 Of interest, our results indicated the variant was also linked to CWP. The variant rs2609255 is an eQTL SNP for FAM13A in lung tissue, and previous studies have shown that FAM13A was differentially expressed in respiratory epithelial cells during the differentiation of human pulmonary type II cells in vitro as well as in patients with cystic fibrosis.

The variant rs4651005 was reported to be associated with the forced expired volume in 1 s (FEV1) and forced vital capacity (FVC) in a large-scale GWAS of LF recently.12 The variant located at an enhancer marked by histone H3K4me1 in lung fibroblast primary cells according to Encode database and associated with the expression of ANGPTL1 in lung tissues according to GTEx-V8. According to previous findings, ectopic expression of ANGPTL1 suppressed the epithelial-to-mesenchymal transition (EMT) by reducing the expression of the zinc-finger protein SLUG in pulmonary epithelial cells.23 Consistently, EMT of alveolar epithelial cells was considered as an important process of pulmonary fibrosis, which could increase the deposition of extracellular matrix and promoted myofibroblasts.24 25

Another two susceptibility loci of FEV1/FVC in previous study, rs985256 at 2q33.1 and rs6539952 at 16q24.1, were also identified as susceptibility loci of CWP in current study. Even though rs985256 did not locate at an expression regulatory region, a high LD variant rs7575286 (r2=0.96) located at an enhancer marked by H3K4me1 and H3K27ac in lung fibroblast primary cells and associated with the expression of SPATS2L (p=1.9 ×10-5). SPATS2L has been reported as an important bronchodilator response gene in subjects with asthma and associated with therapy response in children with asthma exacerbations.26 27 The variant rs6539952 was predicted to be associated with the expression of RP11-463O9.9, whose function was still unclear by now.

Through co-expressed pathway analysis, the autophagy pathway was shown to be an important pathway associated with the identified genes and involved in the development of CWP. As summarised by Zhao et al, several studies have demonstrated that autophagy mitigated the apoptosis of fibroblasts and the senescence of alveolar epithelial cells. In addition, silica-induced pulmonary fibrosis can be alleviated by decreasing apoptosis of alveolar epithelial cells in silicosis.28 However, evidences have also shown that excessive macrophage autophagy aggravated the pathogenesis of silicosis fibrosis by promoting the proliferation and migration of lung fibroblasts in silicosis.29

Even though new risk factor and susceptibility genes of CWP were identified, providing new insights into understanding the genetic basis of pneumoconiosis, this study still has some limitations. First, the sample size of GWAS study is limited, and the statistical power is insufficient. Second, the newly identified gene did not reach the level of GWAS significance, which probably due to the limited sample size. At last, the biological mechanisms behind these associations need further functional experimental verification.

In conclusion, with an MR and association analysis, we found LF was inversely related to the risk of CWP and identified four novel CWP susceptibility loci of rs2609255-FAM13A, rs4651005-ANGPTL1, rs985256-SPATS2L and rs6539952-RP11-463O9.9 in Han Chinese population. These findings may advance our understanding of the pathogenesis and susceptibility of CWP.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • TW, WS, HW and YC contributed equally.

  • Correction notice This article has been corrected since it was published Online First. The author name Yuxin Cheng was previously spelt incorrectly as Yuxing Cheng.

  • Contributors TW, WS, HW and YC are joint first authors. Study design was led by TW, FM and CN. Data collection was carried out by WS, HW and YC. Statistical analyses were conducted by TW and YL. TW, WS, HW and YC drafted the manuscript. FM and CN revised the manuscript and approved the final version.

  • Funding This work was supported by the National Natural Science Foundation of China Nos. (81602813; 81874258).

  • Competing interests None declared.

  • Patient consent for publication Not required.

  • Ethics approval This study was approved by the institutional review board of Nanjing Medical University.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data availability statement Data are available on reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information. The detailed information of SNPs derived from published GWAS would be available on request from the corresponding authors.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.