Genetics
Mendelian randomisation finds genetically higher insulin causally lowers Lp(a), helping explain why low Lp(a) tracks with diabetes risk (Cardiovasc Diabetol 2024)
Original title: Causal associations between insulin and Lp(a) levels in Caucasian population: a Mendelian randomization study
Prior observational studies have linked low Lp(a) to increased type 2 diabetes (T2D) risk, but Mendelian randomisation (MR) evidence for this link has been inconsistent, and in vitro work suggests high insulin can suppress apolipoprotein(a) synthesis. Using genetic instruments for fasting insulin from a European GWAS meta-analysis (n = 151,013) and Lp(a) summary data from a UK Biobank GWAS (n = 361,194), this two-sample MR study found genetically predicted fasting insulin negatively associated with Lp(a) (beta -0.15, SE 0.05, P = 0.003). The weighted median method supported a causal relationship (beta -0.26, SE 0.07, P = 0.0002), though MR-Egger did not reach significance (beta -0.22, SE 0.13, P = 0.11), with additional sensitivity analyses (MR-PRESSO, leave-one-out, MR Steiger) supporting robustness. The findings suggest hyperinsulinaemia, common in T2D, causally lowers Lp(a), offering a mechanistic explanation for why low Lp(a) is observationally associated with higher diabetes risk.
Original abstract
Background: Numerous observational studies have demonstrated that circulating lipoprotein(a) [Lp(a)] might be inversely related to the risk of type 2 diabetes (T2D). However, recent Mendelian randomization (MR) studies do not consistently support this association. The results of in vitro research suggest that high insulin concentrations can suppress Lp(a) levels by affecting apolipoprotein(a) [apo(a)] synthesis. This study aimed to identify the relationship between genetically predicted insulin concentrations and Lp(a) levels, which may partly explain the associations between low Lp(a) levels and increased risk of T2D.
Methods: Independent genetic variants strongly associated with fasting insulin levels were identified from meta-analyses of genome-wide association studies in European populations (GWASs) (N = 151,013). Summary level data for Lp(a) in the population of European ancestry were acquired from a GWAS in the UK Biobank (N = 361,194). The inverse-variance weighted (IVW) method approach was applied to perform two-sample summary-level MR. Robust methods for sensitivity analysis were utilized, such as MR‒Egger, the weighted median (WME) method, MR pleiotropy residual sum and outlier (MR-PRESSO), leave-one-out analysis, and MR Steiger.
Results: Genetically predicted fasting insulin levels were negatively associated with Lp(a) levels (β = - 0.15, SE = 0.05, P = 0.003). The sensitivity analysis revealed that WME (β = - 0.26, SE = 0.07, P = 0.0002), but not MR‒Egger (β = - 0.22, SE = 0.13, P = 0.11), supported a causal relationship between genetically predisposed insulin levels and Lp(a).
Conclusion: Our MR study provides robust evidence supporting the association between genetically predicted increased insulin concentrations and decreased concentrations of Lp(a). These findings suggest that hyperinsulinaemia, which typically accompanies T2D, can partially explain the inverse relationship between low Lp(a) concentrations and an increased risk of T2D.
Summary written by lp-a.org from the published abstract; figures as published. Page updated 18 August 2026. Methods.