Causal relationships between blood lipids and major psychiatric disorders: Univariable and multivariable mendelian randomization analysis

Abstract

Background: Whether the positive associations of blood lipids with psychiatric disorders are causal is uncertain. We conducted this two-sample Mendelian randomization (MR) analysis to comprehensively investigate associations of blood lipids with psychiatric disorders.

Methods: Univariable and multivariable models were established for MR analyses. Inverse variance-weighted (IVW) MR was employed as the main approach; weighted median and MR-Egger were used as sensitivity analysis methods. The possibility of violating MR assumptions was evaluated utilizing several sensitivity analyses, including heterogeneity statistics, horizontal pleiotropy statistics, single SNP analysis, leave-one-out analysis and MR-PRESSO analysis. As instrumental variables, we screened 362 independent single-nucleotide polymorphisms (SNP) related to blood lipids from a recent genome-wide association study involving 76,627 individuals of European ancestry, with a genome-wide significance level of p < 5 × 10^{- 8}. Summary-level information for the six psychiatric disorders was extracted from Psychiatric Genomics Consortium and Alzheimer Disease Genetics Consortium.

Results: We observed eight significant associations in univariable MR analysis, four of which were corroborated by multivariable MR (MVMR) analysis modified for the other three lipid traits: high-density lipoprotein cholesterol (HDL-C) level with the risk of PTSD (OR = 0.91, 95% CI = 0.85-0.97, p = 0.002) and AD (OR = 0.79, 95% CI = 0.71-0.88, p < 0.001) and triglycerides (TG) level with the risk of MDD (OR = 1.02, 95% CI = 1.003-1.03, p = 0.01) and panic disorder (OR = 0.83, 95% CI = 0.74-0.92, p < 0.001). In addition, four associations were not significant in MVMR analysis after adjustment for three lipid traits: total cholesterol (TC) level with the risk of PTSD, low-density lipoprotein cholesterol (LDL-C) level with the risk of MDD and AD and TG level with the risk of AD.

Conclusions: Our results show that blood lipids and psychiatric disorders may be related in a causal manner. This shows that abnormal blood lipid levels may act as reliable biomarker of psychiatric disorders and as suitable targets for their prevention and treatment.

Keywords: Alzheimer’s disease; Anxiety; Blood lipids; Depression; Mendelian randomization analysis; Panic; Post-traumatic stress disorder; Schizophrenia.

Fig. 1

Bidirectional Mendelian randomization study workflow. Abbreviations: LDL-C, Low density lipoprotein cholesterol; HDL-C, High density lipoprotein cholesterol; TG, Triglycerides; TC, Total cholesterol; MR, Mendelian randomization; IVW, inverse-variance weighted MR; SNP, single nucleotide polymorphism

Fig. 2

Forest Plot of Associations of blood lipids on Risk of psychiatric disorders. Abbreviations: LDL-C, Low density lipoprotein cholesterol; HDL-C, High density lipoprotein cholesterol; TC, Total cholesterol; TG, Triglycerides; MR, Mendelian randomization; IVW, inverse-variance weighted MR; CI, confidence interval; OR, odds ratio; MDD, Major Depressive Disorder; PTSD, Post-Traumatic Stress Disorder; AD, Alzheimer’s Disease. Results plotted are after pruning for instrument heterogeneity. The forest plot illustrates the OR of the impact of a 1-standard–deviation rise in genetically determined exposure (TC, HDL-C, LDL-C, TG) on outcome (psychiatric disorders) as a box, with error bars representing the 95% CI. The vertical dotted line delineates an OR of 1, i.e., no impact of the exposure on the result

Fig. 3

Scatter Plot of Associations of blood lipids on Risk of psychiatric disorders. Abbreviations: LDL-C, Low density lipoprotein cholesterol; HDL-C, High density lipoprotein cholesterol; TC, Total cholesterol; TG, Triglycerides; MR, Mendelian randomization; IVW, inverse-variance weighted MR; CI, confidence interval; OR, odds ratio; MDD, Major Depressive Disorder; PTSD, Post-Traumatic Stress Disorder; AD, Alzheimer’s Disease. The Scatterplot of independent instrument single-nucleotide polymorphism (SNP) exposure impacts vs. result effects from 2 independent samples augmented by the standard error of these effects on the vertical and horizontal sides (for presentation, alleles are coded so that all SNP exposure effects are positive). The solid lines denote the regression slopes fitted by the primary inverse variance–weighted (IVW) and complementary mendelian randomization (MR) approaches: slopes fitted by IVW MR method were very comparable in direction and magnitude to slopes fitted by MR-Egger and weighted median approaches for risk of the result