Life course plasma metabolomic signatures of genetic liability to Alzheimer's disease.

Mechanisms through which most known Alzheimer's disease (AD) loci operate to increase AD risk remain unclear. Although Apolipoprotein E (APOE) is known to regulate lipid homeostasis, the effects of broader AD genetic liability on non-lipid metabolites remain unknown, and the earliest ages at which metabolic perturbations occur and how these change over time are yet to be elucidated. We examined the effects of AD genetic liability on the plasma metabolome across the life course. Using a reverse Mendelian randomization framework in two population-based cohorts [Avon Longitudinal Study of Parents and Children (ALSPAC, n = 5648) and UK Biobank (n ≤ 118,466)], we estimated the effects of genetic liability to AD on 229 plasma metabolites, at seven different life stages, spanning 8 to 73 years. We also compared the specific effects of APOE ε4 and APOE ε2 carriage on metabolites. In ALSPAC, AD genetic liability demonstrated the strongest positive associations with cholesterol-related traits, with similar magnitudes of association observed across all age groups including in childhood. In UK Biobank, the effect of AD liability on several lipid traits decreased with age. Fatty acid metabolites demonstrated positive associations with AD liability in both cohorts, though with smaller magnitudes than lipid traits. Sensitivity analyses indicated that observed effects are largely driven by the strongest AD instrument, APOE, with many contrasting effects observed on lipids and fatty acids for both ε4 and ε2 carriage. Our findings indicate pronounced effects of the ε4 and ε2 genetic variants on both pro- and anti-atherogenic lipid traits and sphingomyelins, which begin in childhood and either persist into later life or appear to change dynamically.

Hypertensive disorders of pregnancy and midlife maternal cognition in a prospective cohort study.

Hypertensive disorders of pregnancy (HDP) are associated with an increased risk of cardiovascular disorders, with recent evidence linking pre-eclampsia with vascular dementia. We examined associations of HDP with cognitive performance measured in midlife, in a prospective cohort study, the Avon Longitudinal Study of Parents and Children. Six cognitive function domains were measured 20 years after pregnancy at a mean age of 51 years. The cognition tests were repeated at clinics in the following two years. Cognitive function domains measured were immediate and delayed verbal episodic memory, working memory, processing speed, verbal intelligence, and verbal fluency. Exposures were pre-eclampsia, gestational hypertension (GH), and a combined category of any HDP, all compared to normotensive pregnancy. Of 3393 pregnancies included in the analysis, GH was experienced by 417 (12.3%) and pre-eclampsia by 57 (1.7%). GH was associated with lower verbal episodic memory, in the delayed logic memory test (-0.16 SDs; 95% CI -0.30, -0.03; p = .015) and there was weak evidence of an association with the immediate logic memory test (-0.13 SDs; -0.27, 0.001; p = .058). However, we did not see steeper declines by age for women with GH and there was no evidence of associations with other cognitive domains or for pre-eclampsia with any domains. Results were not substantially changed after controlling for midlife blood pressure. Our findings suggest that a history of GH is associated with slightly reduced episodic memory 20 years after pregnancy, but we found no evidence of a quicker age-related decline compared to women with normotensive pregnancies.

A brief comparison of polygenic risk scores and Mendelian randomisation.

Mendelian randomisation and polygenic risk score analysis have become increasingly popular in the last decade due to the advent of large-scale genome-wide association studies. Each approach has valuable applications, some of which are overlapping, yet there are important differences which we describe here.

Investigation of genetic determinants of cognitive change in later life.

Cognitive decline is a major health concern and identification of genes that may serve as drug targets to slow decline is important to adequately support an aging population. Whilst genetic studies of cross-sectional cognition have been carried out, cognitive change is less well-understood. Here, using data from the TOMMORROW trial, we investigate genetic associations with cognitive change in a cognitively normal older cohort. We conducted a genome-wide association study of trajectories of repeated cognitive measures (using generalised estimating equation (GEE) modelling) and tested associations with polygenic risk scores (PRS) of potential risk factors. We identified two genetic variants associated with change in attention domain scores, rs534221751 (p = 1 × 10-8 with slope 1) and rs34743896 (p = 5 × 10-10 with slope 2), implicating NCAM2 and CRIPT/ATP6V1E2 genes, respectively. We also found evidence for the association between an education PRS and baseline cognition (at >65 years of age), particularly in the language domain. We demonstrate the feasibility of conducting GWAS of cognitive change using GEE modelling and our results suggest that there may be novel genetic associations for cognitive change that have not previously been associated with cross-sectional cognition. We also show the importance of the education PRS on cognition much later in life. These findings warrant further investigation and demonstrate the potential value of using trial data and trajectory modelling to identify genetic variants associated with cognitive change.

The bidirectional effects between cognitive ability and brain morphology: A life course Mendelian randomization analysis.

Introduction: Little is understood about the dynamic interplay between brain morphology and cognitive ability across the life course. Additionally, most existing research has focused on global morphology measures such as estimated total intracranial volume, mean thickness, and total surface area. Methods: Mendelian randomization was used to estimate the bidirectional effects between cognitive ability, global and regional measures of cortical thickness and surface area, estimated total intracranial volume, total white matter, and the volume of subcortical structures (N=37,864). Analyses were stratified for developmental periods (childhood, early adulthood, mid-to-late adulthood; age range: 8-81 years). Results: The earliest effects were observed in childhood and early adulthood in the frontoparietal lobes. A bidirectional relationship was identified between higher cognitive ability, larger estimated total intracranial volume (childhood, mid-to-late adulthood) and total surface area (all life stages). A thicker posterior cingulate cortex and a larger surface area in the caudal middle frontal cortex and temporal pole were associated with greater cognitive ability. Contrary, a thicker temporal pole was associated with lower cognitive ability. Discussion: Stable effects of cognitive ability on brain morphology across the life course suggests that childhood is potentially an important window for intervention.

Efficacy of metformin targets on cardiometabolic health in the general population and non-diabetic individuals: a Mendelian randomization study.

BACKGROUND: Metformin shows beneficial effects on cardiometabolic health in diabetic individuals. However, the beneficial effects in the general population, especially in non-diabetic individuals are unclear. We aim to estimate the effects of perturbation of seven metformin targets on cardiometabolic health using Mendelian randomization (MR). METHODS: Genetic variants close to metformin-targeted genes associated with expression of the corresponding genes and glycated haemoglobin (HbA1c) level were used to proxy therapeutic effects of seven metformin-related drug targets. Eight cardiometabolic phenotypes under metformin trials were selected as outcomes (average N = 466,947). MR estimates representing the weighted average effects of the seven effects of metformin targets on the eight outcomes were generated. One-sample MR was applied to estimate the averaged and target-specific effects in 338,425 non-diabetic individuals in UK Biobank. FINDINGS: Genetically proxied averaged effects of five metformin targets, equivalent to a 0.62% reduction of HbA1c level, was associated with 37.8% lower risk of coronary artery disease (CAD) (odds ratio [OR] = 0.62, 95% confidence interval [CI] = 0.46-0.84), lower levels of body mass index (BMI) (ß = -0.22, 95% CI = -0.35 to -0.09), systolic blood pressure (SBP) (ß = -0.19, 95% CI = -0.28 to -0.09) and diastolic blood pressure (DBP) levels (ß = -0.29, 95% CI = -0.39 to -0.19). One-sample MR suggested that the seven metformin targets showed averaged and target-specific beneficial effects on BMI, SBP and DBP in non-diabetic individuals. INTERPRETATION: This study showed that perturbation of seven metformin targets has beneficial effects on BMI and blood pressure in non-diabetic individuals. Clinical trials are needed to investigate whether similar effects can be achieved with metformin medications. FUNDING: Funding information is provided in the Acknowledgements.

Estimating Dementia Risk Using Multifactorial Prediction Models.

Importance: The clinical value of current multifactorial algorithms for individualized assessment of dementia risk remains unclear. Objective: To evaluate the clinical value associated with 4 widely used dementia risk scores in estimating 10-year dementia risk. Design, Setting, and Participants: This prospective population-based UK Biobank cohort study assessed 4 dementia risk scores at baseline (2006-2010) and ascertained incident dementia during the following 10 years. Replication with a 20-year follow-up was based on the British Whitehall II study. For both analyses, participants who had no dementia at baseline, had complete data on at least 1 dementia risk score, and were linked to electronic health records from hospitalizations or mortality were included. Data analysis was conducted from July 5, 2022, to April 20, 2023. Exposures: Four existing dementia risk scores: the Cardiovascular Risk Factors, Aging and Dementia (CAIDE)-Clinical score, the CAIDE-APOE-supplemented score, the Brief Dementia Screening Indicator (BDSI), and the Australian National University Alzheimer Disease Risk Index (ANU-ADRI). Main Outcomes and Measures: Dementia was ascertained from linked electronic health records. To evaluate how well each score predicted the 10-year risk of dementia, concordance (C) statistics, detection rate, false-positive rate, and the ratio of true to false positives were calculated for each risk score and for a model including age alone. Results: Of 465 929 UK Biobank participants without dementia at baseline (mean [SD] age, 56.5 [8.1] years; range, 38-73 years; 252 778 [54.3%] female participants), 3421 were diagnosed with dementia at follow-up (7.5 per 10 000 person-years). If the threshold for a positive test result was calibrated to achieve a 5% false-positive rate, all 4 risk scores detected 9% to 16% of incident dementia and therefore missed 84% to 91% (failure rate). The corresponding failure rate was 84% for a model that included age only. For a positive test result calibrated to detect at least half of future incident dementia, the ratio of true to false positives ranged between 1 to 66 (for CAIDE-APOE-supplemented) and 1 to 116 (for ANU-ADRI). For age alone, the ratio was 1 to 43. The C statistic was 0.66 (95% CI, 0.65-0.67) for the CAIDE clinical version, 0.73 (95% CI, 0.72-0.73) for the CAIDE-APOE-supplemented, 0.68 (95% CI, 0.67-0.69) for BDSI, 0.59 (95% CI, 0.58-0.60) for ANU-ADRI, and 0.79 (95% CI, 0.79-0.80) for age alone. Similar C statistics were seen for 20-year dementia risk in the Whitehall II study cohort, which included 4865 participants (mean [SD] age, 54.9 [5.9] years; 1342 [27.6%] female participants). In a subgroup analysis of same-aged participants aged 65 (±1) years, discriminatory capacity of risk scores was low (C statistics between 0.52 and 0.60). Conclusions and Relevance: In these cohort studies, individualized assessments of dementia risk using existing risk prediction scores had high error rates. These findings suggest that the scores were of limited value in targeting people for dementia prevention. Further research is needed to develop more accurate algorithms for estimation of dementia risk.

Distinct metabolic features of genetic liability to type 2 diabetes and coronary artery disease: a reverse Mendelian randomization study.

BACKGROUND: Type 2 diabetes (T2D) and coronary artery disease (CAD) both have known genetic determinants, but the mechanisms through which their associated genetic variants lead to disease onset remain poorly understood. METHODS: We used large-scale metabolomics data in a two-sample reverse Mendelian randomization (MR) framework to estimate effects of genetic liability to T2D and CAD on 249 circulating metabolites in the UK Biobank (N = 118,466). We examined the potential for medication use to distort effect estimates by conducting age-stratified metabolite analyses. FINDINGS: Using inverse variance weighted (IVW) models, higher genetic liability to T2D was estimated to decrease high-density lipoprotein cholesterol (HDL-C) and low-density lipoprotein cholesterol (LDL-C) (e.g. , HDL-C:  -0.05 SD; 95% CI -0.07 to -0.03, per doubling of liability), whilst increasing all triglyceride groups and branched chain amino acids (BCAAs). IVW estimates for CAD liability suggested an effect on reducing HDL-C as well as raising very-low density lipoprotein cholesterol (VLDL-C) and LDL-C. In pleiotropy-robust models, T2D liability was still estimated to increase BCAAs, but several estimates for higher CAD liability reversed and supported decreased LDL-C and apolipoprotein-B. Estimated effects of CAD liability differed substantially by age for non-HDL-C traits, with higher CAD liability lowering LDL-C only at older ages when statin use was common. INTERPRETATION: Overall, our results support largely distinct metabolic features of genetic liability to T2D and CAD, illustrating both challenges and opportunities for preventing these commonly co-occurring diseases. FUNDING: Wellcome Trust [218495/Z/19/Z], UK MRC [MC_UU_00011/1; MC_UU_00011/4], the University of Bristol, Diabetes UK [17/0005587], World Cancer Research Fund [IIG_2019_2009].

Exploring the causal effects of genetic liability to ADHD and Autism on Alzheimer's disease.

Few studies suggest possible links between attention deficit hyperactivity disorder (ADHD), autism spectrum disorder (ASD) and Alzheimer's disease but they have been limited by small sample sizes, diagnostic and recall bias. We used two-sample Mendelian randomization (MR) to estimate the bidirectional causal association between genetic liability to ADHD and ASD on Alzheimer's disease. In addition, we estimated the causal effects independently of educational attainment and IQ, through multivariable Mendelian randomization (MVMR). We employed genetic variants associated with ADHD (20,183 cases/35,191 controls), ASD (18,381 cases/27,969 controls), Alzheimer's disease (71,880 cases/383,378 controls), educational attainment (n = 766,345) and IQ (n = 269,867) using the largest GWAS of European ancestry. There was limited evidence to suggest a causal effect of genetic liability to ADHD (odds ratio [OR] = 1.00, 95% CI: 0.98-1.02, P = 0.39) or ASD (OR = 0.99, 95% CI: 0.97-1.01, P = 0.70) on Alzheimer's disease. Similar causal effect estimates were identified as direct effects, independent of educational attainment (ADHD: OR = 1.00, 95% CI: 0.99-1.01, P = 0.76; ASD: OR = 0.99, 95% CI: 0.98-1.00, P = 0.28) and IQ (ADHD: OR = 1.00, 95% CI: 0.99-1.02. P = 0.29; ASD: OR = 0.99, 95% CI: 0.98-1.01, P = 0.99). Genetic liability to Alzheimer's disease was not found to have a causal effect on risk of ADHD or ASD (ADHD: OR = 1.12, 95% CI: 0.86-1.44, P = 0.37; ASD: OR = 1.19, 95% CI: 0.94-1.51, P = 0.14). We found limited evidence to suggest a causal effect of genetic liability to ADHD or ASD on Alzheimer's disease; and vice versa.

The causes and consequences of Alzheimer's disease: phenome-wide evidence from Mendelian randomization.

Alzheimer's disease (AD) has no proven causal and modifiable risk factors, or effective interventions. We report a phenome-wide association study (PheWAS) of genetic liability for AD in 334,968 participants of the UK Biobank study, stratified by age. We also examined the effects of AD genetic liability on previously implicated risk factors. We replicated these analyses in the HUNT study. PheWAS hits and previously implicated risk factors were followed up in a Mendelian randomization (MR) framework to identify the causal effect of each risk factor on AD risk. A higher genetic liability for AD was associated with medical history and cognitive, lifestyle, physical and blood-based measures as early as 39 years of age. These effects were largely driven by the APOE gene. The follow-up MR analyses were primarily null, implying that most of these associations are likely to be a consequence of prodromal disease or selection bias, rather than the risk factor causing the disease.

Causal effects of circulating cytokine concentrations on risk of Alzheimer's disease and cognitive function.

BACKGROUND: There is considerable evidence suggesting a role of neuroinflammation in the pathogenesis of Alzheimer's disease. Establishing causality is challenging due to bias from reverse causation and residual confounding. METHODS: We used two-sample MR to explore causal effects of circulating cytokine concentrations on Alzheimer's disease risk and cognitive function. We employed genetic variants from the largest publicly available genome-wide association studies (GWASs) of cytokine concentrations (N = 8,293), Alzheimer's disease (71,880 cases/383,378 controls), prospective memory (N = 152,605 to 462,302), reaction time (N = 454,157 to 459,523) and fluid intelligence (N = 149,051). RESULTS: Evidence suggest that 1 standard deviation (SD) increase in levels of CTACK (CCL27) (OR = 1.09 95%CI: 1.01 to 1.19, p = 0.03) increased risk of Alzheimer's disease. There was weak evidence of a causal effect of MIP-1b (CCL4) (OR = 1.04 95% CI: 0.99 to 1.09, p = 0.08), Eotaxin (OR = 1.08 95% CI: 0.99 to 1.17, p = 0.10), GROa (CXCL1) (OR = 1.04 95% CI: 0.99 to 1.10, p = 0.15), MIG (CXCL9) (OR = 1.17 95% CI: 0.97 to 1.41, p = 0.10), IL-8 (Wald ratio: OR = 1.21 95% CI: 0.97 to 1.51, p = 0.09) and IL-2 (Wald Ratio: OR = 1.21 95% CI: 0.94 to 1.56, p = 0.14) on Alzheimer's disease risk. A 1 SD increase in concentration of Eotaxin (IVW: OR = 1.05 95% CI: 0.98 to 1.13, p = 0.14), IL-8 (OR = 1.21 95% CI: 1.07 to 1.37, p = 0.003) and MCP1 (OR = 1.07 95% CI: 1.03 to 1.13, p = 0.003) were associated with lower fluid intelligence, and IL-4 (OR = 0.86 95%CI: 0.79 to 0.98, p = 0.02) with higher. CONCLUSIONS: Our findings suggest a causal role of cytokines in the pathogenesis of Alzheimer's disease and fluid intelligence.

Little genomic support for Cyclophilin A-matrix metalloproteinase-9 pathway as a therapeutic target for cognitive impairment in APOE4 carriers.

Therapeutic targets for halting the progression of Alzheimer's disease pathology are lacking. Recent evidence suggests that APOE4, but not APOE3, activates the Cyclophilin-A matrix metalloproteinase-9 (CypA-MMP9) pathway, leading to an accelerated breakdown of the blood-brain barrier (BBB) and thereby causing neuronal and synaptic dysfunction. Furthermore, blockade of the CypA-MMP9 pathway in APOE4 knock-in mice restores BBB integrity and subsequently normalizes neuronal and synaptic function. Thus, CypA has been suggested as a potential target for treating APOE4 mediated neurovascular injury and the resulting neuronal dysfunction and degeneration. The odds of drug targets passing through clinical trials are greatly increased if they are supported by genomic evidence. We found little evidence to suggest that CypA or MMP9 affects the risk of Alzheimer's disease or cognitive impairment using two-sample Mendelian randomization and polygenic risk score analysis in humans. This casts doubt on whether they are likely to represent effective drug targets for cognitive impairment in human APOE4 carriers.

Associations between Adverse Childhood Experiences and the novel inflammatory marker glycoprotein acetyls in two generations of the Avon Longitudinal Study of Parents and Children birth cohort.

BACKGROUND: Adverse childhood experiences (ACEs) are associated with increased risk of non-communicable diseases in adulthood, potentially mediated by chronic low-grade inflammation. Glycoprotein acetyls (GlycA) is a marker of chronic and cumulative inflammation. We investigated associations between ACEs and GlycA at different ages, in two generations of the population-based Avon Longitudinal Study of Parents and Children (ALSPAC) birth cohort. METHODS: ALSPAC offspring's total ACE scores were generated for two age periods using prospectively collected data: 0-7y and 0-17y. GlycA was measured using high-resolution proton nuclear magnetic resonance at mean ages 8y, 18y, and 24y. Sample sizes ranged from: n = 5116 (8y) to n = 3085 (24y). ALSPAC mothers (n = 4634) retrospectively reported ACEs experienced before age 18y and GlycA was assessed at mean age 49y. We used multivariable linear regression to estimate associations between ACEs (total ACE score and individual ACEs) and subsequent GlycA in both samples, adjusting for key confounders. RESULTS: Mean GlycA levels were similar in offspring and mothers and over time. In offspring, there was no evidence that ACEs (total score or individual ACE) were associated with GlycA at age 8y or 18y, or 24y after adjustment for maternal age at birth and parity, maternal marital status, household occupational social class, maternal education, maternal smoking, own ethnicity, sex, and age in months. In mothers, there was evidence of a positive association between the total ACE score and GlycA at age 49y (adjusted mean difference 0.007 mmol/L; 95%CI: 0.003, 0.01). Emotional neglect was the only individual ACE associated with higher GlycA after adjusting for confounders and other ACEs. CONCLUSION: Results suggest the association between ACEs and GlycA may emerge in middle age. Future research should explore the extent to which inflammation in adulthood mediates well-documented associations between ACEs and adverse health outcomes in later life.

Exposure to multiple childhood social risk factors and adult body mass index trajectories from ages 20 to 64 years.

BACKGROUND: While childhood social risk factors appear to be associated with adult obesity, it is unclear whether exposure to multiple childhood social risk factors is associated with accelerated weight gain during adulthood. METHODS: We used the Medical Research Council National Survey of Health and Development, a British population-based birth cohort study of participants born in 1946, height and weight were measured by nurses at ages 36, 43, 53 and 60-64 and self-reported at 20 and 26 years. The 9 childhood socioeconomic risk factors and 8 binary childhood psychosocial risk factors were measured, with 13 prospectively measured at age 4 years (or at 7 or 11 years if missing) and 3 were recalled when participants were age 43. Multilevel modelling was used to examine the association between the number of childhood social risk factors and changes in body mass index (BMI) with age. RESULTS: Increasing exposure to a higher number of childhood socioeconomic risk factors was associated with higher mean BMI across adulthood for both sexes and with a faster increase in BMI from 20 to 64 years, among women but not men. Associations remained after adjustment for adult social class. There was no evidence of an association between exposure to childhood psychosocial risk factors and mean BMI in either sex at any age. CONCLUSIONS: Strategies for the prevention and management of weight gain across adulthood may need to tailor interventions in consideration of past exposure to multiple socioeconomic disadvantages experienced during childhood.

Genomic analysis of diet composition finds novel loci and associations with health and lifestyle.

We conducted genome-wide association studies (GWAS) of relative intake from the macronutrients fat, protein, carbohydrates, and sugar in over 235,000 individuals of European ancestries. We identified 21 unique, approximately independent lead SNPs. Fourteen lead SNPs are uniquely associated with one macronutrient at genome-wide significance (P < 5 × 10-8), while five of the 21 lead SNPs reach suggestive significance (P < 1 × 10-5) for at least one other macronutrient. While the phenotypes are genetically correlated, each phenotype carries a partially unique genetic architecture. Relative protein intake exhibits the strongest relationships with poor health, including positive genetic associations with obesity, type 2 diabetes, and heart disease (rg ≈ 0.15-0.5). In contrast, relative carbohydrate and sugar intake have negative genetic correlations with waist circumference, waist-hip ratio, and neighborhood deprivation (|rg| ≈ 0.1-0.3) and positive genetic correlations with physical activity (rg ≈ 0.1 and 0.2). Relative fat intake has no consistent pattern of genetic correlations with poor health but has a negative genetic correlation with educational attainment (rg ≈-0.1). Although our analyses do not allow us to draw causal conclusions, we find no evidence of negative health consequences associated with relative carbohydrate, sugar, or fat intake. However, our results are consistent with the hypothesis that relative protein intake plays a role in the etiology of metabolic dysfunction.

Is disrupted sleep a risk factor for Alzheimer's disease? Evidence from a two-sample Mendelian randomization analysis.

BACKGROUND: It is established that Alzheimer's disease (AD) patients experience sleep disruption. However, it remains unknown whether disruption in the quantity, quality or timing of sleep is a risk factor for the onset of AD. METHODS: We used the largest published genome-wide association studies of self-reported and accelerometer-measured sleep traits (chronotype, duration, fragmentation, insomnia, daytime napping and daytime sleepiness), and AD. Mendelian randomization (MR) was used to estimate the causal effect of self-reported and accelerometer-measured sleep parameters on AD risk. RESULTS: Overall, there was little evidence to support a causal effect of sleep traits on AD risk. There was some suggestive evidence that self-reported daytime napping was associated with lower AD risk [odds ratio (OR): 0.70, 95% confidence interval (CI): 0.50-0.99). Some other sleep traits (accelerometer-measured 'eveningness' and sleep duration, and self-reported daytime sleepiness) had ORs of a similar magnitude to daytime napping, but were less precisely estimated. CONCLUSIONS: Overall, we found very limited evidence to support a causal effect of sleep traits on AD risk. Our findings provide tentative evidence that daytime napping may reduce AD risk. Given that this is the first MR study of multiple self-report and objective sleep traits on AD risk, findings should be replicated using independent samples when such data become available.

Education, intelligence and Alzheimer's disease: evidence from a multivariable two-sample Mendelian randomization study.

OBJECTIVES: To examine whether educational attainment and intelligence have causal effects on risk of Alzheimer's disease (AD), independently of each other. DESIGN: Two-sample univariable and multivariable Mendelian randomization (MR) to estimate the causal effects of education on intelligence and vice versa, and the total and independent causal effects of both education and intelligence on AD risk. PARTICIPANTS: 17 008 AD cases and 37 154 controls from the International Genomics of Alzheimer's Project (IGAP) consortium. MAIN OUTCOME MEASURE: Odds ratio (OR) of AD per standardized deviation increase in years of schooling (SD = 3.6 years) and intelligence (SD = 15 points on intelligence test). RESULTS: There was strong evidence of a causal, bidirectional relationship between intelligence and educational attainment, with the magnitude of effect being similar in both directions [OR for intelligence on education = 0.51 SD units, 95% confidence interval (CI): 0.49, 0.54; OR for education on intelligence = 0.57 SD units, 95% CI: 0.48, 0.66]. Similar overall effects were observed for both educational attainment and intelligence on AD risk in the univariable MR analysis; with each SD increase in years of schooling and intelligence, odds of AD were, on average, 37% (95% CI: 23-49%) and 35% (95% CI: 25-43%) lower, respectively. There was little evidence from the multivariable MR analysis that educational attainment affected AD risk once intelligence was taken into account (OR = 1.15, 95% CI: 0.68-1.93), but intelligence affected AD risk independently of educational attainment to a similar magnitude observed in the univariate analysis (OR = 0.69, 95% CI: 0.44-0.88). CONCLUSIONS: There is robust evidence for an independent, causal effect of intelligence in lowering AD risk. The causal effect of educational attainment on AD risk is likely to be mediated by intelligence.