{"paper_id":"a038645b-4b24-4fca-8a6c-c74bf0b21ca4","body_text":"We analyzed genome-wide data from three previously\ndescribed independent IPF case–control collections (named here as the\nChicago [ 5 ], Colorado [ 6 ], and UK [ 8 ] studies; please refer to the online supplement for\nsummaries of these collections). Two more independent case–control\ncollections (named here as the UUS [United States, United Kingdom, and Spain]\nand Genentech studies) were included as replication datasets. The new UUS study\nrecruited cases from the United States, United Kingdom, and Spain and selected\ncontrols from UK Biobank ( 19 ) (full\ndetails on the recruitment, genotyping, and quality control of UUS cases and\ncontrols can be found in the online supplement). The previously described ( 20 ) Genentech study consisted of cases\nfrom three IPF clinical trials and controls from four non-IPF clinical trials\n( see  the online supplement). All studies were restricted to\nunrelated individuals of European ancestry, and we applied stringent quality\ncontrol measures (full details of the quality control measures of each study can\nbe found in the online supplement and Figure E1 in the online supplement). All\nstudies diagnosed cases using American Thoracic Society and European Respiratory\nSociety guidelines ( 21 – 23 ) and had appropriate institutional\nreview board or ethics approval.\nGenotype data for the Colorado, Chicago, UK, and UUS studies were imputed\nseparately using the Haplotype Reference Consortium r1.1 panel ( 24 ) ( see  the online\nsupplement). For individuals in the Genentech study, genotypes were derived from\nwhole-genome sequencing data. Duplicated individuals between studies were\nremoved ( see  the online supplement).\nIn each of the Chicago, Colorado, and UK studies\nseparately, a genome-wide analysis of IPF susceptibility, using SNPTEST ( 25 ) v2.5.2, was conducted adjusting for\nthe first 10 principal components to account for fine-scale population\nstructure. Only biallelic autosomal variants that had a minor allele count\n≥10 were in the Hardy–Weinberg Equilibrium\n( P  > 1 × 10 −6 )\nand were well-imputed (imputation quality\n R 2  > 0.5) in at least two studies\nwere included. A genome-wide meta-analysis of the association summary statistics\nwas performed across the Chicago, Colorado, and UK studies using R v3.5.1\n(discovery stage). Conditional analyses were performed to identify independent\nassociation signals in each locus ( see  the online\nsupplement).\nSentinel variants (defined as the variant in an association signal where no other\nvariants within 1 Mb showed a stronger association) of the novel signals\nreaching genome-wide significance in the meta-analysis\n( P  < 5 × 10 −8 ),\nand nominally significant ( P  < 0.05) with\nconsistent direction of effect in each study, were further tested in the\nreplication samples. We considered novel signals to be associated with IPF\nsusceptibility if they reached a Bonferroni-corrected threshold\n( P  < 0.05/number of signals followed up)\nin a meta-analysis of the UUS and Genentech studies (replication stage;\n see  the online supplement). Previously reported signals\nwith\n P  < 5 × 10 −8 \nin the discovery meta-analysis were deemed a confirmed association.\nTo further refine our association signals to include\nonly variants with the highest probabilities of being causal, Bayesian\nfine-mapping was undertaken. This approach takes all variants within the\nassociated locus and, using the GWAS association results, calculates the\nprobability of each variant being the true causal variant (under the assumptions\nthat there is one causal variant and that the causal variant has been measured).\nThe probabilities are then combined across variants to define the smallest set\nof variants that is 95% likely to contain the causal variant (i.e., the 95%\ncredible set) for each IPF susceptibility signal ( see  the\nonline supplement).\nTo identify which genes might be implicated by the IPF susceptibility signals, we\nidentified whether any variants in the credible sets were genic coding variants\nand defined as deleterious (using Variant Effect Predictor [VEP] [ 26 ]). In addition, we tested to see if\nany of the credible set variants were associated with gene expression using\nthree expression quantitative trait loci (eQTL) resources (the Lung eQTL study\n[ n  = 1,111] [ 27 – 29 ],\nthe NESDA-NTR [Netherlands Study of Depression and Anxiety-Netherlands Twin\nRegister] blood eQTL database [ n  = 4,896]\n[ 30 ], and 48 tissues in GTEx [ 31 ] [ n  between 80 and\n491];  see  the online supplement). Where IPF susceptibility\nvariants were found to be associated with expression levels of a gene, we tested\nwhether the same variant was likely to be causal both for differences in gene\nexpression and IPF susceptibility. We only report associations with gene\nexpression where the probability of the same variant driving both the IPF\nsusceptibility signal and gene expression signal exceeded 80%\n( see  the online supplement).\nTo investigate whether the IPF susceptibility variants that were in noncoding\nregions of the genome might be in regions with regulatory functions (for\nexample, in regions of open chromatin), we investigated the likely functional\nimpact of those variants using DeepSEA (deep learning-based sequence analyzer)\n( 32 ). Taking all of the IPF\nsusceptibility variants together, we tested for overall enrichment in regulatory\nregions specific to particular cell and tissue types using FORGE (functional\nelement overlap analysis of the results of GWAS experiments) ( 33 ) and GARFIELD (GWAS analysis of\nregulatory or functional information enrichment with LD correction) ( 34 ). Finally, we investigated whether the\ngenes that were near to the IPF susceptibility variants were more likely to be\ndifferentially expressed between IPF cases and controls in four lung epithelial\ncell types, using SNPsea ( 35 ). More\ndetails are provided in the online supplement.\nAs previous studies have reported shared genetic\nsusceptibility for IPF and other lung traits ( 12 ,  13 ,  15 ), we investigated whether the new and\npreviously reported IPF susceptibility signals were associated with quantitative\nlung function measures in a GWAS of 400,102 individuals ( 36 ) or with ILAs in a GWAS comparing 1,699 individuals\nwith an ILA and 10,247 controls ( 37 ).\nLung function measures investigated were FEV 1 , FVC, the ratio\nFEV 1 /FVC (used in the diagnosis of COPD), and peak expiratory\nflow. We applied a Bonferroni corrected  P  value threshold to\ndefine variants also associated with ILAs or lung function.\nThe contribution of as yet unreported variants to IPF\nsusceptibility was assessed using polygenic risk scores. For each individual in\nthe UUS study, the weighted score was calculated as the number of risk alleles,\nmultiplied by the effect size of the variant (as a weighting), summed across all\nvariants included in the score. Effect sizes were taken from the discovery GWAS\nand independent variants selected using a linkage disequilibrium\n r 2  ≤ 0.1. As we wanted to\nexplore the contribution from as yet unreported variants, we excluded variants\nwithin 1 Mb of each IPF susceptibility locus from the risk score calculation\n( see  the online supplement).\nThe score was tested to identify whether it was associated with IPF\nsusceptibility, adjusting for 10 principal components to account for fine-scale\npopulation structure, using PRSice v1.25 ( 38 ). We altered the number of variants included in the risk score\ncalculation using a sliding  P  threshold\n( P T ) such that the variant had to have a\n P  value < P T  in the\ngenome-wide meta-analysis to be included in the score. This allows us to explore\nwhether variants that do not reach statistical significance in GWAS of current\nsize contribute to disease susceptibility. We used the recommended significance\nthreshold of  P  < 0.001 for determining\nsignificantly associated risk scores ( 38 ).\n\nFollowing quality control, 541 cases and 542 controls from\nthe Chicago study, 1,515 cases and 4,683 controls from the Colorado study, and 612\ncases and 3,366 controls from the UK study were available ( Table 1  and Figure E1) to contribute to the discovery stage of\nthe genome-wide susceptibility analysis ( Figure\n1 ). For the replication stage of the GWAS, after quality control, there\nwere 792 cases and 10,000 controls available in the UUS study and 664 cases and\n1,874 controls available in the Genentech study ( see  the online\nsupplement).\nDemographics of Study Cohorts\nDefinition of abbreviations :\nHRC = Haplotype Reference Consortium;\nUUS = United States, United Kingdom, and Spain.\nAge only available for 103 Chicago controls.\nAge available for 602 UK cases.\nSex only available for 500 Chicago cases.\nSex only available for 510 Chicago controls.\nSmoking status only recorded for 236 UK cases.\nSmoking status only recorded for 753 idiopathic pulmonary fibrosis cases\nin UUS.\nSmoking status only recorded for 481 of the Genentech controls.\nManhattan plot of discovery analysis results. The  x  axis\nshows chromosomal position, and the  y  axis shows the\n−log( P  value) for each variant in the discovery\ngenome-wide analysis. The red line shows genome-wide significance\n( P  < 5 × 10 −8 ),\nand variants in green met the criteria for further study in the replication\nanalysis (i.e., reached genome-wide significance in the discovery\nmeta-analysis and had  P  < 0.05 and\nconsistent direction of effects in each study). Genes in gray are previously\nreported signals that reach significance in the discovery genome-wide\nmeta-analysis. Genes in black are the novel signals identified in the\ndiscovery analysis that reach genome-wide significance when meta-analyzing\ndiscovery and replication samples. The signals that did not replicate are\nshown in red. For ease of visualization the  y  axis has been\ntruncated at 25.\nTo identify new signals of association, we meta-analyzed the genome-wide association\nresults for IPF susceptibility for the Chicago, Colorado, and UK discovery studies.\nThis gave a maximum sample size of up to 2,668 cases and 8,591 controls for\n10,790,934 well-imputed ( R 2  > 0.5)\nvariants with minor allele count ≥10 in each study and which were available\nin two or more of the studies (Figure E2).\nThree novel signals (in 3p21.31 [near  KIF15 ,  Figure 2A ], 7p22.3 [near  MAD1L1 ,  Figure 2B ], and 8q24.12 [near\n DEPTOR ,  Figure 2C ])\nshowed a genome-wide significant\n( P  < 5 × 10 −8 )\nassociation with IPF susceptibility in the discovery meta-analysis and were also\nsignificant after adjusting for multiple testing\n( P  < 0.01) in the replication stage comprising\n1,467 IPF cases and 11,874 controls ( Tables\n2  and E1). Two additional loci were genome-wide significant in the\ngenome-wide discovery analysis but did not reach significance in the replication\nstudies. The sentinel variants of these two signals were a low-frequency intronic\nvariant in  RTEL1  (MAF = 2.1%, replication\n P  = 0.012) and a rare intronic variant in\n HECTD2  (MAF = 0.3%, replication\n P  = 0.155). Conditional analyses did not\nidentify any additional independent association signals at the new or previously\nreported IPF susceptibility loci (Figure E5).\nRegion plots of three novel idiopathic pulmonary fibrosis susceptibility loci\nfrom discovery genome-wide meta-analysis. Each point represents a variant\nwith chromosomal position on the  x  axis and the\n−log( P  value) on the  y  axis.\nVariants are colored in by linkage disequilibrium with the sentinel variant.\nBlue lines show the recombination rate, and gene locations are shown at the\nbottom of the plot. Region plots are shown for the three replicated novel\nidiopathic pulmonary fibrosis susceptibility loci, i.e.,\n( A ) the susceptibility signal on chromosome 3 near\n KIF15 , ( B ) the susceptibility signal\non chromosome 7 near  MAD1L1 , and ( C ) the\nsusceptibility signal on chromosome 8 near  DEPTOR .\nDiscovery and Replication Association Analysis Results for the Five Signals\nReaching Significance in the Discovery Genome-Wide Association Studies that\nHave Not Previously Been Reported as Associated with Idiopathic Pulmonary\nFibrosis\nDefinition of abbreviations :\nChr = chromosome; CI = confidence\ninterval; MAF = minor allele frequency;\nOR = odds ratio; Pos = position;\nrsid = reference SNP cluster ID.\nThe minor allele is the effect allele, and the MAF is taken from across\nthe studies used in the discovery meta-analysis.\nTo identify the likely causal genes for each new signal, we investigated whether any\nof the variants were also associated with changes in gene expression ( Table 3 ). The sentinel variant (rs78238620)\nof the novel signal on chromosome 3 was a low-frequency variant\n(MAF = 5%) in an intron of  KIF15  with the minor\nallele being associated with increased susceptibility to IPF and decreased\nexpression of  KIF15  in brain tissue and the nearby gene\n TMEM42  in thyroid ( 31 )\n(Figure E7 and Tables E2 and E3i). The IPF risk allele for the novel chromosome 7\nsignal (rs12699415, MAF = 42%) was associated with decreased\nexpression of  MAD1L1  in heart tissue ( 31 ) (Figure E8 and Tables E2 and E3ii). For the signal on\nchromosome 8, the sentinel variant (rs28513081) was located in an intron of\n DEPTOR , and the IPF risk allele was associated with decreased\nexpression of  DEPTOR  (in colon, lung, and skin [ 27 – 29 ,  31 ]) and RP11-760H22.2 (in\ncolon and lung [ 31 ]). The risk allele was\nalso associated with increased expression of  DEPTOR  (in whole blood\n[ 30 ]),  TAF2  (in colon\n[ 31 ]), RP11-760H22.2 (in adipose [ 31 ]), and KB-1471A8.1 (in adipose and skin\n[ 31 ], Figure E9 and Tables E2 and\nE3iii). There were no variants predicted to be highly deleterious within the\nfine-mapped signals for any of the loci.\nGene Expression and Spirometric Results for the Three Novel IPF\nSusceptibility Loci\nDefinition of abbreviations :\nChr = chromosome; CI = confidence\ninterval; eQTL = expression quantitative trait loci;\nIPF = idiopathic pulmonary fibrosis;\nrsid = reference SNP cluster ID.\nAnnotation of the variant was taken from Variant Effect Predictor (VEP).\nA list of all variants included in the credible sets with their\nannotations and eQTL results can be found in Table E3. For\ncolocalization, only genes where there was a greater than 80%\nprobability of colocalization between the IPF risk signal and gene\nexpression of that gene are reported in this table. In the\ncolocalization column, ↑ denotes that the allele that increases\nIPF risk was associated with increased expression of the gene, ↓\ndenotes that the IPF risk allele was associated with decreased\nexpression of the gene, and ↕ denotes that the IPF risk allele\nwas associated with increased expression in some tissues and decreased\nexpression in others. Full results from the eQTL and colocalization\nanalyses can be found in Table E2. The spirometric results for the three\nnovel IPF risk loci are taken from Shrine and colleagues ( 36 ) using the allele associated\nwith increased IPF risk as the effect allele, with β being the\nchange in  z -score units. Results for all IPF risk\nvariants can be found in Table E6.\nWe confirmed genome-wide significant associations with IPF susceptibility for 11 of\nthe 17 previously reported signals (in or near  TERC, TERT, DSP, \n7q22.1,  MUC5B, ATP11A, IVD, AKAP13, KANSL1, FAM13A , and\n DPP9 ; Table E1 and Figure E4). The signal at\n FAM13A,  while genome-wide significant in the discovery\nmeta-analysis, was not significant in the Chicago study. This was the only signal\nreaching genome-wide significance in the discovery genome-wide meta-analysis that\ndid not reach at least nominal significance in each study in the discovery analysis.\nThree further previously reported signals at 11p15.5 (near  MUC5B )\nwere no longer genome-wide significant after conditioning on the\n MUC5B  promoter variant (Table E1), consistent with previous\nreports ( 6 ,  39 ).\nOf the 14 IPF susceptibility signals (i.e., the 11 previously reported signals we\nconfirmed and three novel signals), the only variant predicted to have a potential\nfunctional effect on gene regulation through disruption of chromatin structure or\ntranscription factor binding motifs (using DeepSEA) was rs2013701 (in an intron of\n FAM13A ), which was associated with a change in DNase I\nhypersensitivity in 18 cell types and FOXA1 in the T-47D cell line (a breast cancer\ncell line derived from a pleural effusion, Table E4). The 14 IPF susceptibility\nsignals were found to be enriched in DNase I hypersensitivity site regions in\nmultiple tissues including fetal lung tissue (Figures E10 and E11). No enrichment in\ndifferential expression in airway epithelial cells between IPF cases and healthy\ncontrols was observed for the 14 IPF susceptibility signals when using SNPsea (Table\nE5).\nPrevious studies have reported an overlap of genetic association loci between lung\nfunction and IPF ( 12 ). We undertook a\nlookup of the 14 IPF susceptibility loci in the largest GWAS of lung function in the\ngeneral population published to date ( 36 ).\nThe sentinel variants of 12 of the 14 IPF susceptibility loci were at least\nnominally associated ( P  < 0.05) with one or more\nlung function trait in general population studies ( Tables 3  and E6). After adjustments for multiple testing\n( P  < 5.2 × 10 −4 ),\nthe previously reported variants at  FAM13A ,  DSP ,\nand  IVD  were associated with decreased FVC, and variants at\n FAM13A ,  DSP , 7q22.1\n( ZKSCAN1 ), and  ATP11A  were associated with\nincreased FEV 1 /FVC. Similarly, for the three novel susceptibility\nvariants, all showed at least a nominal association with decreased FVC and increased\nFEV 1 /FVC. We observed a nominally significant association of the\n MUC5B  IPF risk allele with decreased FVC and increased\nFEV 1 /FVC. The IPF risk alleles at  MAPT  were\nsignificantly associated with both increased FEV 1  and FVC. To determine\nhow the variants identified for IPF susceptibility are related to differences in\nlung function between cases and controls, we investigated whether variants known to\nbe associated with lung function show an association in our IPF GWAS. Of the 279\nvariants reported ( 36 ) as associated with\nlung function (Table E7), 8 showed an association with lung function after\ncorrections for multiple testing (located in or near  MCL1 ,\n DSP ,  ZKSCAN1 ,  OBFC1 ,\n IVD ,  MAPT , and two signals in\n FAM13A ).\nAs interstitial lung abnormalities may be a precursor to IPF in a subset of patients,\nand there have been previous reports of shared genetic etiology between IPF and ILAs\n( 37 ,  40 ,  41 ), we investigated\nwhether our three new signals and the 11 previously reported signals were associated\nwith ILAs in the largest ILA GWAS reported to date ( 37 ). Eight of the IPF susceptibility loci were at least\nnominally significantly associated with either ILAs or subpleural ILAs with\nconsistent direction of effects (i.e., the allele associated with increased IPF risk\nwas also associated with increased ILA risk). The new  KIF15 ,\n MAD1L1 , and  DEPTOR  signals were not associated\nwith ILAs (although the rare risk allele at  HECTD2  that did not\nreplicate in our study showed some association with an increased risk of subpleural\nILAs [ P  = 0.003] with a large effect size\nsimilar to that observed in the IPF discovery meta-analysis).\nTo quantify the impact of as yet unreported variants on IPF susceptibility, polygenic\nrisk scores were calculated excluding the 14 IPF susceptibility variants (as well as\nall variants within 1 Mb). The polygenic risk score was significantly associated\nwith increased IPF susceptibility despite exclusion of the known genetic association\nsignals (including  MUC5B ). As the  P T \nfor inclusion of variants in the score was increased, the risk score became more\nsignificant reaching a plateau at around\n P T  = 0.2 with risk score\n P  < 3.08 × 10 −23 \nand explaining around 2% of the phenotypic variation (Figure E12), suggesting that\nthere is a modest but statistically significant contribution of additional as yet\nundetected variants to IPF susceptibility. Further increasing\n P T  beyond 0.2 did not improve the predictive\naccuracy of the risk score.\n\nWe undertook the largest GWAS of IPF susceptibility to date\nand identified three novel signals of association that implicated genes not\npreviously known to be important in IPF.\nThe strongest evidence for the new signal on chromosome 8 implicates\n DEPTOR , which encodes the dishevelled, Egl-10 and Pleckstrin\ndomain–containing mTOR-interacting protein.  DEPTOR  inhibits\nmTOR (mammalian target of rapamycin) kinase activity as part of both the mTORC1 and\nmTORC2 protein complexes. The IPF risk allele at this locus was associated with\ndecreased gene expression of  DEPTOR  in lung tissue (Table E2).\nTGFβ-induced DEPTOR suppression can stimulate collagen synthesis ( 42 ), and the importance of mTORC1 signaling\nvia 4E-BP1 for TGFβ-induced collagen synthesis has recently been demonstrated\nin fibrogenesis ( 43 ).\n MAD1L1 , implicated by a new signal on chromosome 7 and eQTL\nanalyses of nonlung tissue, is a mitotic checkpoint gene, mutations in which have\nbeen associated with multiple cancers including lung cancer ( 44 ,  45 ). Studies\nhave shown that  MAD1 , a homolog of  MAD1L1 , can\ninhibit  TERT  activity (or possibly enforce expression of\n TERT  when the promoter E-box is mutated) ( 45 ,  46 ). This could\nsuggest that  MAD1L1  may increase IPF susceptibility through reduced\ntelomerase activity. Another spindle-assembly–related gene ( 47 ),  KIF15,  was implicated\nby the new signal on chromosome 3 (along with  TMEM42 ).\nThe genome-wide study also identified two signals that were not replicated after\nmultiple testing adjustments.  RTEL1 , a gene involved in telomere\nelongation regulation, has not previously been identified in an IPF GWAS; however,\nthe collective effect of rare variants in  RTEL1  has been reported\nas associated with IPF susceptibility ( 48 – 54 ). The ubiquitin E3\nligase encoded by  HECTD2  has been shown to have a proinflammatory\nrole in the lung, and other  HECTD2  variants may be protective\nagainst acute respiratory distress syndrome ( 55 ). However, the lack of replication for these signals in our data\nsuggests that further exploration of their relationship to interstitial lung\ndiseases is warranted.\nBy combining the largest available GWAS datasets for IPF, we were able to confirm 11\nof 17 previously reported signals. Conditional analysis at the 11p15.5 region\nindicated that previously reported signals at  MUC2  and\n TOLLIP  were not independent of the association with the\n MUC5B  promoter variant. Previously reported signals at\n EHMT2 ,  OBFC1 , and  MDGA2  were\nonly found to be associated in one of the discovery studies and showed no evidence\nof an association with IPF susceptibility in the other two discovery studies. Only\nthe 11 signals that we confirmed in our data were included in subsequent\nanalyses.\nThe IPF susceptibility signals at  DSP, FAM13A , 7q22.1\n( ZKSCAN1 ), and 17q21.31 ( MAPT ) have also been\nreported as associated with COPD, although with opposite effects (i.e., the allele\nassociated with increased risk of IPF being associated with decreased risk of COPD).\nSpirometric diagnosis of COPD was based on a reduced FEV 1 /FVC ratio. In\nan independent dataset of 400,102 individuals, eight of the IPF signals were\nassociated with decreased FVC and with a comparatively weaker effect on\nFEV 1 . This is consistent with the lung function abnormalities\nassociated with IPF, as well as the decreased risk of COPD. Of note, only around 3%\nof previously reported lung function signals ( 36 ) also showed association with IPF susceptibility in our study. This\nsuggests that while some IPF susceptibility variants might represent genes and\npathways that are important in general lung health, others are likely to represent\nmore disease-specific processes.\nUsing polygenic risk scores, we demonstrated that, despite the relatively large\nproportion of disease susceptibility explained by the known genetic signals of\nassociation reported here, IPF is highly polygenic with potentially hundreds (or\nthousands) of as yet unidentified variants associated with disease\nsusceptibility.\nA strength of our study was the large sample size compared with previous GWAS and the\navailability of an independent replication dataset. A limitation of our study was\nthat the controls used were generally younger in all studies included, and there\nwere differences in sex and smoking distributions in some of the studies. As age,\nsex, and smoking status were not available for all individuals in four of our\ndatasets, we were unable to adjust for these variables without substantially\nreducing our sample size. However, cases and controls in the UUS and UK datasets\nwere matched for age, sex, and smoking. The three novel signals replicated in all of\nthe discovery and replication datasets, providing reassurance that the signals we\nreport are robust despite differences between the datasets. As we had limited\ninformation beyond IPF diagnosis status for a large proportion of the individuals\nincluded in the studies, we cannot rule out some association with other age-related\nconditions that are comorbid with IPF. However, other age-related conditions were\nnot excluded from either the cases or controls. For the signals near\n KIF15  and  MAD1L1 , there was substantial\nevidence for an association with gene expression in nonlung tissues but not in\neither of the two (nonfibrotic) lung tissue eQTL datasets. This could reflect cell\ntype-specific effects that are missed when studying whole tissue or effects that are\ndisease-dependent. Finally, our study was not designed to identify rare functional\nvariant associations. As both common and rare variants are known to be important in\nIPF susceptibility ( 39 ), this is a\nlimitation of our study.\nIn summary, we report new biological insights into IPF susceptibility and demonstrate\nthat further studies to identify the genetic determinants of IPF susceptibility are\nneeded. Our new signals of association with IPF susceptibility provide increased\nsupport for the importance of mTOR signaling in pulmonary fibrosis as well as the\npossible implication of mitotic spindle-assembly genes.","source_license":"CC-BY-4.0","license_restricted":false}