Mendelian randomization analysis of lipids traits and lipid-lowering drug-targets in relation to cognitive status

Cognitive impairment is increasingly prevalent among older adults, posing a major public health challenge. While lipid-lowering medications are widely used to manage dyslipidemia in this population, the causal relationship between lipid levels and cognitive function remains unclear. This study investigates genetically- predicted lipid traits, lipid-lowering drug-targets, and their associations with cognitive status.

Genetic variants associated with lipid features or lipid-lowering drug-targets were identified using data from the GLGC. Cognitive summary statistics were obtained from six independent GWAS datasets. Significant drug-targets were further assessed using SMR and co-localization to account for potential confounding effects.

No significant associations were found between lipid traits or eight of the nine lipid-reducing drug-targets and cognitive outcomes. However, genetic inhibition of NPC1L1 was associated with a higher risk of cognitive impairment across three distinct datasets (β1 = -0.191 [95 % CI: -0.307 to -0.075], p = 0.023; β2 = -0.288 [95 % CI: -0.394 to -0.182], p = 5.616 × 10^-6; β3 = -0.488 [95 % CI: -0.760 to -0.217], p = 0.011). SMR analysis revealed significant associations between NPC1L1 expression in subcutaneous adipose tissue and cognitive outcomes (β1 = 0.020 [95 % CI: 0.003 to 0.037], p = 0.019; β2 = 0.026 [95 % CI: 0.009 to 0.043], p = 0.003; β3 = 0.054 [95 % CI: 0.009 to 0.099], p = 0.019).

These findings do not support dyslipidemia as a causal factor for cognitive impairment. However, NPC1L1 inhibition, as targeted by ezetimibe, may increase the risk of cognitive decline, independent of its lipid-lowering effects.