(222) Discovery of Novel Genetic Variants in Severe CF Liver Disease as Determined by Whole Genome Sequencing

Background: Liver disease is the third leading cause of death in people with (pw) cystic fibrosis (CF). Many pw CF have liver dysfunction, but severe CF liver disease with portal hypertension due to cirrhosis (CFLD) occurs in only 3-5%. One genetic variant (SERPINA1; Z allele) is known as a risk allele for CFLD [1]. The Cystic Fibrosis Genome Project (CFGP) performed whole genome sequencing (WGS) [2] in 516 pw CFLD, and 3,566 pw CF validated to not have CFLD (CFnoLD), to identify additional modifier genes for CFLD.

Methods: CFLD and CFnoLD subjects were recruited worldwide or from North America, respectively. Validation of severe CFLD (defined as portal hypertension due to cirrhosis) was by review of medical records and source documents. To account for age of onset, while enabling corrections for cryptic relatedness, we transformed CFLD time-to-event to a quantitative phenotype using martingale residuals. A single nucleotide polymorphism (SNP) association test of 206 validated liver disease SNPs, based on literature review of relevant liver pathophysiologies, was performed; then a genome wide association study (GWAS) of ~8 million bi-allelic SNPs which passed QC, missing rate < 0.02, and MAF ≥0.01.

Results: The mean age of severe CFLD diagnosis was 10.3 years, whereas 22.8 years was the censored latest age of not having liver disease (CFnoLD). CFnoLD subjects had a higher prevalence of F508del homozygotes (66% vs 55% for CFLD subjects), which reflected study design and enrollment. Males had a higher CFLD frequency compared to females (ratio 1.8:1). Only rs28929474 (Z-allele) in SERPINA1 was significant in the candidate SNP analysis of 206 SNPs following correction for multiple testing (p=1.1x10^-4; FDR=0.02; MAF=0.018; beta=0.11). GWAS analysis, after adjusting for sex, PCs, and site, showed SNP associations in the CFTR gene, due to imbalance of F508del in disease and control populations. After conditioning on the top SNP in CFTR, SNPs at CFTR were no longer significant. Two SNPs were genome-wide significant (< 5x10^-8): near PKD1 (rs147350387; p=8.1x10^-10; MAF=0.019; beta=0.18) and near FNBP1 (rs10988538; p=4.7x10^-9; MAF=0.017; beta=0.18). There were also two “suggestive” (< 5x10^-7) SNPs, in DUSP6 (rs769700; p=1.5x10^-7; MAF=0.203; beta=0.05) and in ANKUB1 (rs62268801; p=4.7x10^-7, MAF=0.047, beta= -0.09). Transcriptome-wide association study (TWAS) evidence for expression versus severe CFLD showed significant associations (FDR qvalue < 0.15) in CXCR1 and AAMP (increased expression associated with increased CFLD risk) and TRBV23-1 (increased expression associated with decreased CFLD risk).

Conclusions: We confirm that severe CFLD is associated with the SERPINA gene Z allele and identify novel genetic associations with biological plausibility for modifying CFLD risk. Associated SNPs include genes involved in bile duct development and congenital hepatic fibrosis (PKD1); cytoskeleton, intracellular signaling and endocytosis (FNBP1); cell differentiation, proliferation and survival (DUSP6); cilium assembly and movement (ANKUB1); immune function (CXCR1 and TRBV23-1); and angiogenesis (AAMP).

Acknowledgements: In memory of Dr. Peter Durie, whose vision and collaboration helped enable the success of this project, and who is sadly missed by the CF research community. This work is supported by CFF grants CUTTIN18XX1, BAMSHA18XX0, and KNOWLE18XX0 and is submitted on behalf of the CF Genome Project.

References: [1] Bartlett JR et al JAMA 2009
[2] Raraigh KS et al J Cyst Fibros 2021