Genome-wide Enrichment of TERT Rare Variants in Idiopathic Pulmonary Fibrosis Patients of Latino Ancestry

To the Editor:

Idiopathic pulmonary fibrosis (IPF) is a rare lung disease characterized by progressive fibrosis and irreversible decline in lung function. Genome-wide rare variant studies have either included only patients of European ancestry (1, 2) or have been disproportionally enriched with patients of European ancestry so that rare variant genetics of patients of non-European ancestry with IPF have been understudied. In this report, we evaluate the enrichment of rare genetic variants in patients of non-European ancestry with IPF.

Patients

This study was approved by the Institutional Review Board at Columbia University, and written consent was obtained from all participants. Detailed information regarding patient and control cohorts, including enrollment and sequencing sites, was previously published (3). Briefly, all case subjects carried a diagnosis of IPF or familial pulmonary fibrosis. The full cohort included multiethnic, unrelated cases (n = 2,966) and control subjects (n = 29,817) (3).

Methods

Sequence data for cases and control subjects from either Columbia or TOPMed (Trans-Omics for Precision Medicine) were processed using the program-specific bioinformatic pipelines for variant calling and variant annotation (3). Qualifying variants were defined as rare (mean allele frequency < 0.0005 in global gnomADv2 Genomes and all ExAC or gnomADv2 exome ancestry-specific subgroups) and predicted to be damaging (stop gained, start lost, frameshift, splice acceptor/donor site, or missense with consensus in silico predicted damaging effect by PolyPhen, REVEL [Rare Exome Variant Ensemble Learner], and PrimateAI). We used unsupervised Louvain clustering using principal components of ancestry, obtained using a set of predefined variants (4), to group cases and control subjects into 11 ethnically similar clusters for the Columbia cohorts and 7 ethnically similar clusters for the TOPMed cohorts, as previously described. Self-identified ancestry was available for the Columbia cohorts, against which predicted ancestry of unsupervised clusters was compared. Peddy (5) was used to infer ancestry of the TOPMed cohorts with a probability cutoff of 75% used for ancestry prediction. Individuals not meeting this cutoff for any ancestry were categorized as “Other Admixed”. Individuals classified as “Latino” in the Columbia cohort were grouped with those classified as “American (AMR)” in the TOPMed cohort. Enrichment of rare variants in each gene was tested for an association with IPF through an exact two-sided Cochran-Mantel-Haenszel test after grouping of the case and control clusters by ancestry. Genomic inflation was assessed for the non-European group using a permutation approach for estimating expected P values, as previously described (3).

Results

A total of 241 unrelated patients, which represents 8% of the total, were grouped into non-European cases. Slightly less than half (n = 120) of the non-European cases represent individuals of Latino ancestry; the remainder represents those of African, South Asian, East Asian, and Other Admixed ancestry. The remainder of the cases were grouped into European ancestry clusters (n = 2,725, 92% of the total cohort).

Gene burden analysis of deleterious rare (protein-truncating and missense) variants for 241 cases and 12,509 non-European control subjects showed no evidence of genomic inflation (λ = 0.996) (Figure 1A). We found an excess of TERT rare damaging variants that exceeded genome-wide significance (odds ratio [OR], 67.1; 95% confidence interval [CI], 23.1–195.0; P = 9.4 × 10⁻¹⁴). Analysis by ancestry demonstrated an excess of rare, damaging TERT variants in the Latino subgroup (OR, 80.9; 95% CI, 17.3–383.8; P = 2.6 × 10⁻⁸) (Figures 1B and 2). African (OR, 28.7; 95% CI, 0.6–300.6; P = 4.3 × 10⁻²) and South Asian (OR, 32.4; 95% CI, 4.5–362.0; P = 1.7 × 10⁻⁴) ancestry groups showed a trend toward enrichment that did not exceed study-wide significance (P < 2.63 × 10⁻⁶ accounting for ∼19,000 genes). There were too few rare variant carriers of East Asian or Other Admixed ancestry to calculate meaningful ORs. All qualifying TERT rare variants discovered in non-European cases were submitted to ClinVar (SCV002520369).

Figure 1.

Analysis of non-European idiopathic pulmonary fibrosis and familial pulmonary fibrosis cases and control subjects. Quantile–quantile plot of observed versus expected P values comparing the burden of rare deleterious variants in protein-coding genes in cases and control subjects. (A) Collapsing analysis of 241 cases and 12,509 ancestry-matched non-European control subjects identify significant enrichment of rare deleterious variants in TERT using a model of autosomal dominant inheritance. (B) Similarly, collapsing analysis of 120 Latino cases and 4,959 ancestry-matched control subjects showed significant enrichment of TERT rare variants. The genetic model includes heterozygous variants (with an ExAC/gnomAD allele frequency < 0.0005 predicted to be a protein-truncating (stop, frameshift, splice site), insertion/deletion, or deleterious missense variant (PolyPhen damaging, REVEL ⩾ 0.5, and PrimateAI ⩾ 0.8).

Figure 2.

Forest plot of top significant genes from ancestry-stratified gene burden analysis. Odds ratio derived from Cochran-Mantel-Haenszel test of cases and control subjects in European and non-European clusters. Subsets of non-European cases and control subjects are indicated only for TERT. P values marked by the asterisk are less than a study-wide Bonferroni multiplicity-adjusted significance of P < 2.63 × 10⁻⁶.

Although the non-European group did not show enrichment of PARN, RTEL1, and KIF15 rare deleterious variants, these groups all showed a trend in the same direction as the European ancestry group. No other genes exceeded study-wide significance. For TERT and KIF15, the inclusion of patients of non-European ancestry with IPF led to higher ORs and increased evidence in favor of rare deleterious variant contributions, compared with the analysis of the European ancestry group alone.

Discussion

Despite over 75% of the world’s population being of non-European ancestry (6), relatively few genetic studies have focused on these populations. The lack of diversity in genetic studies is problematic because genetic variant effect sizes and polygenic risk prediction scores cannot be readily extrapolated to populations of different ancestries (7). This knowledge gap exacerbates health inequities in the modern era of personalized medical care.

We report the first genome-wide assessment of rare, deleterious genetic variants in patients of non-European ancestry with IPF. Enrichment of TERT rare deleterious variants was found to exceed genome-wide significance for patients of non-European ancestry, specifically, the Latino subgroup. We find that for at least two genes (TERT and KIF15), the inclusion of patients of non-European ancestry led to a greater OR in the meta-analysis. Like prior studies (8, 9), this approach demonstrated the increased power of multiethnic studies over single-ethnicity studies for improving genetic discoveries.

Here, we adjusted for population substructure in cases and ancestry-matched control subjects and used an exact test to assess the contributions of ultrarare variants in cases and control subjects. Although rare and ultrarare variants are less influenced by population structure than common variants, decreased representation of non-European subjects in reference cohorts may lead to inflated rates of rare qualifying variants. We found no evidence of genomic inflation using our methodology. We acknowledge the uncertainty in comparing the frequency of qualifying variants for subjects of different non-European ancestries, given the smaller sample sizes and wide but overlapping CIs. However, this study justifies this approach in identifying rare variant genetic contributions for individuals of all ancestries.

To our knowledge, this is the first study that confirms the involvement of rare deleterious TERT variants surpassing genome-wide significance for patients of Latino and non-European ancestry with IPF. Indeed, short telomere lengths have been described for a subset of patients from Spain and Mexico with IPF, and short telomeres are associated with earlier onset and more rapidly progressive disease (10). To better understand the genetic underpinnings of patients of all ancestries with IPF, additional work will be needed to broaden patient recruitment to normalize imbalances.