Abstract

Background: An unbiased ‘genome-first’ approach has the potential to expand the understanding of common coding and predicted loss-of-function (pLOF) variants associated with non-alcoholic fatty liver disease (NAFLD). Utilizing this approach, we aimed to uncover the functionality of pathogenic variants in TM6SF2.

Methods: We leveraged exome sequencing data from 44,297 patients in the Penn Medicine Biobank and interrogated 121 non-synonymous missense variants and 12 pLOF variants for associations with ICD-10 coded NAFLD, non-alcoholic steatohepatitis (NASH), and hepatocellular carcinoma (HCC). Significant variants were further analysed with serum parameters, liver fat scores, and imaging adjusted for age, sex, body mass index, principal components of ancestry, and carriage of PNPLA3 rs738409:G. We replicated findings in the UK Biobank (UKB) and differences were statistically significant when p < 0.05. In silico prediction of the structure of TM6SF2 was generated using ColabFold Alphfold2 notebook with MMseqs2.

Results: E167K homozygotes are at an increased risk of ICD-diagnosed NAFLD (p<0.0001, OR:4.8), NASH (p<0.0001, OR:5.1), and HCC (p=0.001, OR:1.77). Similarly, E167K heterozygotes exhibited a similar trend with NAFLD and NASH only. Additionally, L156P heterozygotes were at an increased risk of physician-diagnosed NAFLD (p<0.001, OR:2.16) and NASH (p<0.0001, OR:2.10). Only L156P homozygotes were at an increased risk for HCC (N=2, p<0.0001, OR:532.0). We successfully replicated associations with ICD-diagnosed NAFLD, NASH, and HCC in the UKB for both variants. Additional diagnoses on imaging, coupled with evidence of elevated CT-derived hepatic fat scores, in the PMBB strengthened these associations in E167K and L156P carriers (p<0.01). Metabolomic analyses reveals lower circulating total cholesterol, triglycerides, fatty acids, and total choline in E167K and L156P carriers (Bonferroni-corrected, p<0.0001). In silico predictions validate both mutations to cause structural disruptions on the EXPERA domain leading to loss of TM6SF2 protein function. This hypothesis was strengthened when we interrogated pLOF variants and discovered a stop-gain truncation codon (W35X) associated with an increased risk of physician-diagnosed NAFLD (p=0.03, OR:15.9), NASH (p=0.01, OR:23.4) and elevated liver fat scores (p<0.001) in the same directionality as E167K and L156P.

Conclusion: We confirm that E167K and L156P induces a loss-of-function effect on the protein structure, thus leading to liver disease, and identified a stop-gain codon of potential therapeutic utility towards NAFLD.