Very low Lp(a) raises new-onset diabetes and fatty liver risk in UK Biobank, but Mendelian randomisation finds no causal link (Atheroscler Plus 2024)

Background And Aims: We aimed to study the association of very low serum Lipoprotein(a) [Lp(a)] concentrations with new-onset type 2 diabetes (T2D) and non-alcoholic liver disease (NAFLD) in the context of statin usage in the UK Biobank, a large prospective population cohort.

Methods: Using an extended biomarker dataset, we identified 47,362 participants with very low Lp(a) concentrations (<3.8 nmol/L) from a total of 451,479 participants. With a median follow-up of 12.3 years, we assessed the risk of new-onset cardiometabolic diseases in participants stratified by statin usage with Cox proportional hazards models. We performed two-sample Mendelian randomization MR analyses to test causal relationship between genetically predicted Lp(a) and T2D and NAFLD.

Results: Taking the participants with Lp(a) within reportable range as the reference group, the hazard ratios (HR) for T2D were 1.07 (95 % confidence interval, CI 1.01-1.13) and for NAFLD 1.30 (95 % CI 1.20-1.41) respectively for participants with very low Lp(a) (<3.8 nmol/L). The risk for new-onset T2D was higher in participants using statins (adjusted HR 1.15; 95 % CI 1.05-1.27). The risk estimates for new-onset NAFLD were comparable in the analysis stratified by statin use. There was no evidence for causal links between genetically predicted Lp(a) and T2D nor NAFLD in two-sample MR analyses.

Conclusions: Very low Lp(a) was associated with higher risks of T2D and NAFLD in a prospective analysis of the UK Biobank. The association with T2D was influenced by lipid lowering medication usage. MR analyses did not support causality for these inverse associations.