Resequencing to Fine Map Known Idiopathic Pulmonary Fibrosis Risk Genes. Homing in on Causal Variants.

Genetic determinants of familial interstitial pneumonia, the familial form of idiopathic pulmonary fibrosis (IPF), were initially recognized as far back as 2001 as rare variants in the genes encoding SFTPC (surfactant protein C) and TERT (telomerase reverse transcriptase) (1, 2). Subsequent targeted candidate gene and whole-exome sequencing studies found that rare variation in candidate genes for familial IPF in the telomere complex (TERT, RTEL1 [regulator of telomere elongation helicase 1], and PARN [poly(A)-specific ribonuclease]) and surfactant protein pathways (SFTPA2 [surfactant protein A2] and SFTPC) also determined risk for sporadic IPF in individuals without a family history of IPF (Figure 1) (3–6). A linkage study in 2011 was the first to identify a SNP on chromosome 11p15.5 within the promoter of the mucin 5 gene (rs35705950 in MUC5B), with large effects on both familial and sporadic IPF risk replicated in subsequent genome-wide association studies (GWASs) (7).

Figure 1.

Allele frequencies and effect sizes of known risk loci for familial and sporadic idiopathic pulmonary fibrosis. Plot shows the distribution of allele frequencies for risk variants identified within the genes shown with estimated effect sizes based on prior studies.

For nearly a decade, GWASs have discovered 17 common genetic variants (allele frequency >5%) scattered across the genome associated with risk for IPF (Figure 1) (8–12). Among the most successfully replicated of these IPF risk loci was rs35705950 on MUC5B, confirming previous linkage studies and the large effect of this locus on familial and sporadic IPF pathogenesis (7). This led to the hypothesis that IPF results from increased MUC5B expression, causing excess production of mucus in the airways and thus impairing lung defense (13). The risk allele frequency of rs35705950 mirrors the observed prevalence of IPF, with a risk allele frequency of 10% in European individuals, the group with the highest IPF prevalence, whereas the risk allele is nearly nonexistent in African descent populations, in which IPF is much less prevalent. Although the MUC5B promoter polymorphism explains approximately 30% of the observed genetic risk (14), IPF is determined by environmental interactions with variation in multiple genes related to different pathogenic pathways identified by GWAS discoveries, including genes related to host defense (TOLLIP [Toll-interacting protein]), telomere maintenance (TERT, TERC [telomerase RNA component]), signaling (AKAP13 [A-kinase anchor protein 13]), and cell–cell adhesion (DSP [desmoplakin]) (8–12).

GWASs are based on chip genotyping data from subsets of SNPs to tag whole genomes and, more recently, additional imputed genotypes. Hence, GWASs are not usually sufficient to identify the causal variant and have the potential to miss uncovered or previously unknown rare variants. GWASs are also underpowered to detect rare variant associations owing to low frequency and the large number of rare variants throughout the genome. Because of these limitations, it can be difficult to conclusively determine the total number of causal variants within a genomic region. For example, there are contrasting reports of additional association signals within the chromosome 11p15.5 region independent of the MUC5B promoter polymorphism (9, 10). Hence, deep resequencing of known candidate genes for familial and sporadic IPF is required to fully characterize the genetic architecture for risk loci and map independent causative common and rare variants, both known and novel.

In this issue of the Journal (pp. 199–208), Moore and colleagues (15) describe targeted DNA resequencing of 16 genomic regions surrounding loci previously associated with risk for familial or sporadic IPF. Common variants (allele frequency ≥3%) were investigated individually in 3,624 individuals with IPF and 4,442 control subjects, and rare variants were investigated using gene-level and region-based tests in a subgroup of 7,116 subjects with confirmed European ancestry. This is the largest resequencing study of known loci for this uncommon respiratory disease, which has been evaluated mostly in smaller IPF genetic studies. The size of this large IPF cohort in combination with the use of deep, targeted gene resequencing analyzed with conditional analyses and gene-level rare variant tests allowed for the most detailed estimates of the contribution of multiple common and rare variants to IPF risk currently possible.

This resequencing study demonstrated several important aspects of the genetic architecture of known IPF risk loci. First, this study confirmed reported associations for rare and common variations initially described for the genomic regions investigated. However, the top associations were not always at the initially reported sentinel variant (i.e., near ZKSCAN1 [Zinc finger protein with KRAB and SCAN domains 1] and IVD [isovaleryl-coenzyme A dehydrogenase]), demonstrating the power of resequencing to detect causal variation. Second, known common variant associations were determined by a single, independent signal, including 11p15.5, demonstrating that the multiple signals previously reported for MUC5B (including TOLLIP) were likely due to linkage disequilibrium with rs35705950 (9). Third, GWAS loci with common IPF risk variants have the potential to harbor rare variations independently associated with IPF. Specifically, rare variants in the 11p15.5 region downstream of RP13-870H17.3 (near MUC2), TERT, and FAM13A (family with sequence similarity 13 member A) were independently associated with IPF, of which rare variant associations in 11p15.5 and FAM13A were novel. Fourth, this study confirms that loci with rare variants associated with familial IPF potentially contain rare variants associated with sporadic IPF but do not harbor common genetic determinants. Finally, these analyses confidently confirmed prior causal variants, including the MUC5B promotor variant, which had an effect size similar to those in multiple independent studies (16).

Despite the novel and confirmatory aspects of this large resequencing study, there remain aspects of the genetics of IPF not fully addressed. First, the known loci evaluated in this study account for only a small proportion of the observed risk for sporadic IPF. Hence, additional IPF risk loci could be identified in larger cohorts with more detailed genotype data inclusive of whole-genome sequencing for future GWASs. Second, many genetic associations are regulatory loci in noncoding regions identified with mRNA expression data (expression quantitative trait loci [eQTL] analyses) from lung tissue (10, 12). The challenge of evaluating IPF risk loci outside of coding regions relates in part to the lack of available lung tissue from patients with IPF for eQTL and epigenetic studies to identify regulatory variation for a causal gene within an associated noncoding region. Third, most genetic studies of IPF have been performed in European white descent cohorts because of the high prevalence of IPF in white individuals and to minimize confounding by varying ancestral backgrounds. IPF does not appear exclusively in white individuals, and studies limited to a single race could miss IPF risk loci unique to different ancestries. Finally, this study compared variation between subjects with IPF and control subjects, which is not informative regarding effects on severity, progression, or treatment response, which require genetic studies limited to IPF cases.

This resequencing study provides the most detailed characterization of the genetic architecture of known IPF risk loci, including the cumulative and independent genetic factors, both common and rare, constituting IPF risk in familial and sporadic cases with varying effect sizes (Figure 1). In this instance, deep sequencing of known loci rigorously confirmed causal variation from prior studies while identifying novel risk variation. As the cost of next-generation sequencing improves and the number of patients with IPF with genetic data and tissue samples for epigenetic and eQTL studies increases, future studies will improve understanding of this fatal disease without a known cure.