The Use of Genetic Information to Define Idiopathic Pulmonary Fibrosis in UK Biobank

To the Editor: Idiopathic pulmonary fibrosis (IPF) is a rare disease with prevalence of 50 in 100,000 cases in the UK.1 Genome-wide association studies have identified 20 independent single nucleotide polymorphisms (SNPs) that are associated with IPF risk to date.2, 3, 4 A single common SNP in the MUC5B gene promoter region (rs35705950) has a large effect on IPF risk with each copy of the T allele that is associated with a 4- to 5-fold increased risk of IPF.4,5 Most datasets for genetic studies of IPF were derived from dedicated IPF cohort studies, registries, and clinical trials, which are usually modest in size. Large general population cohorts, such as UK Biobank, represent a valuable resource for increasing IPF case sample sizes for molecular epidemiologic studies. However, observed effect size estimates for rs35705950 on IPF risk in general population cohorts, for which cases are defined with the use of the International Classification of Diseases, revision 10 (ICD-10)6 J84.1 code, are smaller than those that are estimated in clinically-derived datasets.7 Although this attenuation could be explained by misclassification of IPF cases, the misclassification may be mitigated by the substantial gain in statistical power that can be leveraged from very large biobanks. However, more accurate classification of cases and control subjects in biobanks could provide more accurate effect estimates for use in further analyses. Given this, we proposed that the IPF risk effect size of rs35705950 could be used to evaluate and refine the choice of codes to define IPF cases. We applied this approach in UK Biobank.