Mendelian randomization identifies blood metabolites previously linked to midlife cognition as causal candidates in Alzheimer’s disease

There are currently no disease-modifying treatments for Alzheimer’s disease (AD), and an understanding of preclinical causal biomarkers to help target disease pathogenesis in the earliest phases remains elusive. Here, we investigated whether 19 metabolites previously associated with midlife cognition—a preclinical predictor of AD—translate to later clinical risk, using Mendelian randomization (MR) to tease out AD-specific causal relationships. Summary statistics from the largest genome-wide association studies (GWASs) for AD and metabolites were used to perform bidirectional univariable MR. Bayesian model averaging (BMA) was additionally performed to address high correlation between metabolites and identify metabolite combinations that may be on the AD causal pathway. Univariable MR indicated four extra-large high-density lipoproteins (XL.HDL) on the causal pathway to AD: free cholesterol (XL.HDL.FC: 95% CI = 0.78 to 0.94), total lipids (XL.HDL.L: 95% CI = 0.80 to 0.97), phospholipids (XL.HDL.PL: 95% CI = 0.81 to 0.97), and concentration of XL.HDL particles (95% CI = 0.79 to 0.96), significant at an adjusted P < 0.009. MR–BMA corroborated XL.HDL.FC to be among the top three causal metabolites, in addition to total cholesterol in XL.HDL (XL.HDL.C) and glycoprotein acetyls (GP). Both XL.HDL.C and GP demonstrated suggestive univariable evidence of causality (P < 0.05), and GP successfully replicated within an independent dataset. This study offers insight into the causal relationship between metabolites demonstrating association with midlife cognition and AD. It highlights GP in addition to several XL.HDLs—particularly XL.HDL.FC—as causal candidates warranting further investigation. As AD pathology is thought to develop decades prior to symptom onset, expanding on these findings could inform risk reduction strategies.

More than 50 million people worldwide currently live with dementia, and with an aging world population, this figure is expected to increase to more than 152 million by 2050 (World Alzheimer Report 2018). The most common dementia type is Alzheimer’s disease (AD), characterized by impaired everyday function, severe cognitive decline—particularly working, episodic, and declarative memory (1)—and a range of neuropsychiatric symptoms (2). It represents a major source of global morbidity and mortality and poses significant human and economic costs (3).Disappointingly, AD drug development has proven difficult, with a 99.6% failure rate in the decade of 2002 to 2012, and this rate continues at the same low level today (4). Numerous reasons have been proposed as to why such clinical trials have failed, including incomplete understanding of true causal mechanisms and a failure to intervene early enough in the pathological cascade. It is therefore necessary to discover biomarkers that can identify individuals at high risk of developing AD and at the earliest possible stages of pathology onset. Moreover, it is important for these to be potentially modifiable so as to offer targets for preventative or therapeutic strategies.Metabolomics represents one avenue that may give a deeper insight into AD etiology. Metabolites are small molecules (<1,500 atomic mass units) with a role in metabolism (5). As the products of many biological processes, they sit at the end of the systems biology pathway and therefore represent effective intermediate phenotypes to a given disease because of their proximity to the clinical endpoint (6, 7). Due to 1) their noninvasive nature of measurement, 2) the fact that they are potentially modifiable through diet and lifestyle, and 3) the ability of many to cross the blood brain barrier, blood metabolites are both practical and valuable markers of biological processes and disease states in dementia (8).Markers of lipid metabolism have received particular attention in this context, as the impairment of lipid metabolism has been associated with AD (5, 8–11) and beta-amyloid (Aβ) burden (12, 13). Relevant to early intervention, they have also been associated with cognitive performance and brain function during normal aging (14, 15). Recently, using a large British population-based birth cohort, we investigated associations between 233 blood metabolites and both memory and processing speed at 60 to 64 y of age as well as changes in these cognitive domains from 60 to 64 to 69 y old. Associations with several metabolite classes were observed, including fatty acids (FAs), various compositions of high-density lipoproteins (HDLs), and glycoprotein acetyls (GP) (16).However, it is not yet established whether these metabolites are causally associated with dementia and AD. Using knowledge from these preclinical associations to investigate translatability to later AD risk could hold special utility in informing early treatment intervention, particularly if a causal relationship can be shown. This study therefore aims to expand our observational findings and assess whether 19 blood metabolites previously associated with late midlife cognition causally associate with later clinical AD status. Both univariable and Bayesian multivariable Mendelian randomization (MR) approaches are harnessed to interrogate independent as well as group associations, and a range of sensitivity analyses are performed to further scrutinize results. Identifying candidate blood metabolites, which are detectable preclinically and on the causal pathway to later AD diagnosis, will aid in facilitating further research into early intervention strategies and more targeted therapeutics.     

The many weak instruments problem and Mendelian randomization

Instrumental variable estimates of causal effects can be biased when using many instruments that are only weakly associated with the exposure. We describe several techniques to reduce this bias and estimate corrected standard errors. We present our findings using a simulation study and an empirical application. For the latter, we estimate the effect of height on lung function, using genetic variants as instruments for height. Our simulation study demonstrates that, using many weak individual variants, two‐stage least squares (2SLS) is biased, whereas the limited information maximum likelihood (LIML) and the continuously updating estimator (CUE) are unbiased and have accurate rejection frequencies when standard errors are corrected for the presence of many weak instruments. Our illustrative empirical example uses data on 3631 children from England. We used 180 genetic variants as instruments and compared conventional ordinary least squares estimates with results for the 2SLS, LIML, and CUE instrumental variable estimators using the individual height variants. We further compare these with instrumental variable estimates using an unweighted or weighted allele score as single instruments. In conclusion, the allele scores and CUE gave consistent estimates of the causal effect. In our empirical example, estimates using the allele score were more efficient. CUE with corrected standard errors, however, provides a useful additional statistical tool in applications with many weak instruments. The CUE may be preferred over an allele score if the population weights for the allele score are unknown or when the causal effects of multiple risk factors are estimated jointly. © 2014 The Authors. Statistics in Medicine published by John Wiley & Sons Ltd.     

Exact optimum coin bias in Efron's randomization procedure

Efron's biased coin design is a restricted randomization procedure that has very favorable balancing properties, yet it is fully randomized, in that subjects are always randomized to one of two treatments with a probability less than 1. The parameter of interest is the bias p of the coin, which can range from 0.5 to 1. In this note, we propose a compound optimization strategy that selects p based on a subjected weighting of the relative importance of the two fundamental criteria of interest for restricted randomization mechanisms, namely balance between the treatment assignments and allocation randomness. We use exact and asymptotic distributional properties of Efron's coin to find the optimal p under compound criteria involving imbalance variability, expected imbalance, selection bias, and accidental bias, for both small/moderate trials and large samples. Copyright © 2015 John Wiley & Sons, Ltd.     

Start‐up designs for response‐adaptive randomization procedures with sequential estimation

Response‐adaptive randomization procedures are appropriate for clinical trials in which two or more treatments are to be compared, patients arrive sequentially and the response of each patient is recorded before the next patient arrives. However, for those procedures that involve sequential estimation of model parameters, start‐up designs are commonly required in order to provide initial estimates of the parameters. In this paper, a suite of such start‐up designs for two treatments and binary patient responses are considered and compared in terms of the numbers of patients required in order to give meaningful parameters estimates, the number of patients allocated to the better treatment, and the bias in the parameter estimates. It is shown that permuted block designs with blocks of size 4 are to be preferred over a wide range of parameter values. For the case of two treatments, normal responses and selected start‐up procedures, a design incorporating complete randomization followed appropriately by repeats of one of the treatments yields the minimum expected number of patients and is to be preferred. Copyright © 2015 John Wiley & Sons, Ltd.     

Association of mental health with the risk of coronary artery disease in patients with diabetes: A mendelian randomization study

Background and aimsObservational studies have shown an association between mental health and coronary artery disease (CAD) in patients with diabetes. Nevertheless, whether these associations are causal is still unknown. In this two-sample Mendelian randomization (MR) study, we aimed to assess the causality between mental health and CAD in patients with diabetes.Methods and resultsSingle-nucleotide polymorphisms (SNPs) associated with: depression (807,553 individuals), anxiety (83,556 individuals) and neuroticism (329,821 individuals) were identified from the largest genome-wide association studies (GWAS). Summary-level data for CAD were extracted from the recently published GWAS of 15,666 diabetic patients (3968 CAD cases and 11,696 controls). The inverse-variance weighted (IVW) method was used for the main analysis. Sensitivity analyses included weighted median, maximum likelihood, and the MR-Egger method. Genetic liability to depression was significantly associated with a higher risk of CAD in patients with diabetes (odds ratio [OR], 1.286; 95%CI,1.018–1.621;p = 0.035). For anxiety and neuroticism, no causal association with CAD in patients with diabetes was observed. Consistent results were obtained in most sensitivity analyses.ConclusionsThis MR study provides genetic evidence that depression is a potential risk factor for CAD in patients with diabetes. However, anxiety and neuroticism were not causally associated with CAD in patients with diabetes. Mental health treatments should be enhanced to prevent CAD in patients with diabetes.     

Lipoprotein(a): An independent,genetic, and causal factor for cardiovascular disease and acute myocardial infarction

Lipoprotein(a) [Lp(a)] is a circulating lipoprotein, and its level is largely determined by variation in the Lp(a) gene (LPA) locus encoding apo(a). Genetic variation in the LPA gene that increases Lp(a) level also increases coronary artery disease (CAD) risk, suggesting that Lp(a) is a causal factor for CAD risk. Lp(a) is the preferential lipoprotein carrier for oxidized phospholipids (OxPL), a proatherogenic and proinflammatory biomarker. Lp(a) adversely affects endothelial function, inflammation, oxidative stress, fibrinolysis, and plaque stability, leading to accelerated atherothrombosis and premature CAD. The INTER-HEART Study has established the usefulness of Lp(a) in assessing the risk of acute myocardial infarction in ethnically diverse populations with South Asians having the highest risk and population attributable risk. The 2018 Cholesterol Clinical Practice Guideline have recognized elevated Lp(a) as an atherosclerotic cardiovascular disease risk enhancer for initiating or intensifying statin therapy.     

Using next-generation sequencing as a genetic diagnostic tool in rare autosomal recessive neurologic Mendelian disorders

Next-generation sequencing was used to investigate 9 rare Chinese pedigrees with rare autosomal recessive neurologic Mendelian disorders. Five probands with ataxia-telangectasia and 1 proband with chorea-acanthocytosis were analyzed by targeted gene sequencing. Whole-exome sequencing was used to investigate 3 affected individuals with Joubert syndrome, nemaline myopathy, or spastic ataxia Charlevoix-Saguenay type. A list of known and novel candidate variants was identified for each causative gene. All variants were genetically verified by Sanger sequencing or quantitative polymerase chain reaction with the strategy of disease segregation in related pedigrees and healthy controls. The advantages of using next-generation sequencing to diagnose rare autosomal recessive neurologic Mendelian disorders characterized by genetic and phenotypic heterogeneity are demonstrated. A genetic diagnostic strategy combining the use of targeted gene sequencing and whole-exome sequencing with the aid of next-generation sequencing platforms has shown great promise for improving the diagnosis of neurologic Mendelian disorders.