Skip to main content
Download PDF

Investigating associations between blood metabolites, later life brain imaging measures, and genetic risk for Alzheimer’s disease

Alzheimer's Research & Therapy volume 15, Article number: 38 (2023) Cite this article

Abstract

Background

Identifying blood-based signatures of brain health and preclinical pathology may offer insights into early disease mechanisms and highlight avenues for intervention. Here, we systematically profiled associations between blood metabolites and whole-brain volume, hippocampal volume, and amyloid-β status among participants of Insight 46—the neuroscience sub-study of the National Survey of Health and Development (NSHD). We additionally explored whether key metabolites were associated with polygenic risk for Alzheimer’s disease (AD).

Methods

Following quality control, levels of 1019 metabolites—detected with liquid chromatography-mass spectrometry—were available for 1740 participants at age 60–64. Metabolite data were subsequently clustered into modules of co-expressed metabolites using weighted coexpression network analysis. Accompanying MRI and amyloid-PET imaging data were present for 437 participants (age 69–71). Regression analyses tested relationships between metabolite measures—modules and hub metabolites—and imaging outcomes. Hub metabolites were defined as metabolites that were highly connected within significant (pFDR < 0.05) modules or were identified as a hub in a previous analysis on cognitive function in the same cohort. Regression models included adjustments for age, sex, APOE genotype, lipid medication use, childhood cognitive ability, and social factors. Finally, associations were tested between AD polygenic risk scores (PRS), including and excluding the APOE region, and metabolites and modules that significantly associated (pFDR < 0.05) with an imaging outcome (N = 1638).

Results

In the fully adjusted model, three lipid modules were associated with a brain volume measure (pFDR < 0.05): one enriched in sphingolipids (hippocampal volume: ß = 0.14, 95% CI = [0.055,0.23]), one in several fatty acid pathways (whole-brain volume: ß =  − 0.072, 95%CI = [− 0.12, − 0.026]), and another in diacylglycerols and phosphatidylethanolamines (whole-brain volume: ß =  − 0.066, 95% CI = [− 0.11, − 0.020]). Twenty-two hub metabolites were associated (pFDR < 0.05) with an imaging outcome (whole-brain volume: 22; hippocampal volume: 4). Some nominal associations were reported for amyloid-β, and with an AD PRS in our genetic analysis, but none survived multiple testing correction.

Conclusions

Our findings highlight key metabolites, with functions in membrane integrity and cell signalling, that associated with structural brain measures in later life. Future research should focus on replicating this work and interrogating causality.

Background

Brain changes accompanying ageing are varied and can include pathologies that lead to cognitive impairment, the commonest of which is Alzheimer’s disease (AD). Identifying non-invasive and scalable markers of brain health and pathology in later life, including but not limited to those associated with AD, would be valuable for research and therapeutic trials. This has led to large efforts in detecting blood-based markers, with candidates such as neurofilament light and phosphorylated-tau showing particular promise [1]. Blood metabolites—the products of chemical reactions occurring in the body—may also present as potential candidates for this goal. Due to their proximity to core biological processes, they are uniquely placed to capture physiological changes and may allow insights into the processes associated with emergence of disease [2]. Additionally, since they are potentially modifiable [3, 4], they could represent possible targets for intervention.

Existing research has identified associations between several metabolite classes and imaging markers related to neurodegeneration, including particular lipids and amino acids [5,6,7,8,9]. However, these studies have been directed towards clinical cohorts, where pathology is already advanced. Additionally, little is known about the involvement of groups of interrelated metabolites; using systems-level approaches could offer an improved understanding of these complex relationships and facilitate the identification of candidate markers. We previously employed a systems-level approach to explore the metabolic correlates of late midlife cognitive function in the Medical Research Council National Survey of Health and Development (NSHD; the British 1946 birth cohort) [10]. We identified groups of highly coexpressed metabolites that associated with cognitive outcomes and key metabolites within these to explore further, including acylcarnitines, modified nucleotides and amino acids, vitamins, and sphingolipids, although many associations with late midlife cognitive outcomes were explained by social factors and childhood cognitive ability [10].

Here, we aimed to investigate the early metabolic correlates of later life brain imaging measures relevant to AD and neurodegeneration—Aβ pathology, whole-brain volume, and hippocampal volume—using metabolite data collected at age 60–64 and imaging data measured 5–11 years later. To provide a deeper understanding on the nature of potential relationships and how they may contribute to AD risk, we explored whether any key metabolites were additionally associated with polygenic risk for AD.

Methods

Participants

The NSHD is a broadly representative birth cohort study, originally following 5362 individuals since their birth in mainland Britain during one week in March 1946 [11]. At age 69–71, 502 participants enrolled in Insight 46, the neuroscience sub-study. At a University College London clinic, they underwent comprehensive clinical and cognitive tests, MRI, and 18F-florbetapir positron emission tomography (PET) imaging [12, 13]. Compared to the full NSHD cohort, participants of Insight 46 were of slightly higher cognitive ability, more socially advantaged, and of better overall health [13]. Further details on participant eligibility and recruitment can be found elsewhere [12].

Insight 46 participants with full metabolite data, and who completed the scanning procedure and were dementia-free, were included for module and hub metabolite analyses (N = 437; 47.6% female, 18.9% Aβ-positive). For PRS analyses, NSHD participants with metabolomics and genetic data were included (N = 1638; 50.4% female).

Ethical approval was obtained from the National Research Ethics Service Committee London (14/LO/1173). All participants provided written informed consent.

Materials

Metabolite quality control

At age 60–64, blood samples were collected in ethylenediaminetetraacetic acid (EDTA) tubes by trained research nurses (96% fasted). Samples were stored at – 80 °C.

Using Ultrahigh Performance Liquid Chromatography-Tandem Mass Spectrometry (UPLC-MS/MS), levels of 1401 metabolites were detected and measured by Metabolon Inc. (Durham, NC, USA) among 1814 NSHD participants. All samples were received by Metabolon at the same time point. Metabolites were assigned to nine families (lipids, amino acids, xenobiotics, peptides, nucleotides, cofactors and vitamins, carbohydrates, energy and partially characterised molecules) and further organised into pathways based on their proposed biological function informed by the Kyoto Encyclopaedia of Genes and Genomes (KEGG) database (Supplementary Table 1). Unknown metabolites were assigned to an “Unknown” family and pathway and denoted by a number prefixed by an “X”; these were included in all analyses.

Metabolite data underwent strict quality control (QC), as detailed in [10], resulting in 1019 metabolites (Supplementary Notes).

Genetic quality control

Initial QC and imputation were performed centrally by the NSHD study team (Supplementary Notes). For this analysis, we removed variants that were rare (MAF < 5%), with a low call rate (< 98%), or that deviated from Hardy–Weinberg equilibrium (p < 1 × 10−5). We additionally removed participants with a low call rate (< 98%), mismatching biological and reported sex, or that were related (PIHAT < 0.1). All QC was performed using PLINK v1.9 (https://www.cog-genomics.org/plink2) [14]. Following QC, genetic and metabolomic data were available for 1638 participants.

Scanning procedure

The scanning procedure and data processing were undertaken by the Insight 46 team. Aβ-PET and MRI were acquired contemporaneously using a single Biograph mMR 3 Tesla PET/MRI scanner (Siemens Healthcare) [12]. Aβ burden was quantified over 10 min, approximately 50 min after intravenous injection with 18F-florbetapir (370 mBq. Standardised uptake value ratios (SUVRs) were derived using a grey matter cortical composite and an eroded subcortical white matter reference region. A cut-off of > 0.6104 was used to define Aβ positivity being the 99th percentile of the lower (Aβ-negative) Gaussian distribution [15]. Participants below this threshold were defined as Aβ-negative. Data were processed using an in-house pipeline, including attenuation correction using pseudo-CT [12]. For volumetric T1-weighted MRI images, visual QC was performed as detailed in [12] and processed using the following: MAPS [16] for whole-brain volume (with manual editing if needed), STEPS [17] for left and right hippocampal volumes (with manual editing if needed), and SPM12 (fil.ion.ucl.ac.uk/spm) [18] for total intracranial volume (TIV). Hippocampal volume was calculated as the mean volume of the left and right hippocampi.

Covariables

In line with our previous analysis in the full NSHD [10, 19], covariables were as follows: sex, blood collection information (clinic location, age, fasting status), age at scan, APOE genotype (ε4 carrier/non carrier; blood samples at age 53 or 69 years), BMI (60–64 years nurse visit), lipid medication (yes/no; self-reported use in 24 h preceding blood collection at 60–64 years), childhood cognitive ability (15 years), highest level of educational attainment (no qualifications/ ‘O level’/ ‘A level’ or higher; 26 years), childhood socioeconomic position (SEP) (father’s current or last known occupation categorised according to the UK Registrar General; 11 years), and midlife SEP (own occupation categorised as for childhood SEP; 53 years).

Statistical analysis

We previously imputed missing covariable data using multiple imputation chained Eqs. (100 iterations and 50 imputations) [20] in the full NSHD metabolomics dataset. Further details of missing data can be found in Table 1. Unless otherwise specified, we carried out all analyses in R version 3.6.0 (details of all software and packages used can be found in Supplementary Notes). A visual summary of our analytical pipeline can be found in Fig. 1.

Table 1 Characteristics of participants included in this analysis
Full size table
Fig. 1
figure 1

Analysis pipeline

Full size image

Coexpression network analysis

To explore associations of clusters (termed “modules”) of co-expressed metabolites, we applied weighted gene coexpression network analysis (WGCNA) [21,22,23] to metabolite data, as detailed previously [10]. First, metabolite data were adjusted for sex and blood clinic information, and the standardised residuals were used for WGCNA. Fourteen modules of highly connected metabolites were then identified, and the first principal component of each module (termed “module eigenvalue”) was derived to allow for relationships between modules and outcomes to be examined. Overrepresentation analyses were conducted, using the hypergeometric test, to identify enriched pathways within the module and provide insight into potential biological function [10]. Modules were allocated an arbitrary colour name using the WGCNA package for ease of discussion.

Hub metabolites

Metabolites that are highly connected to their module (termed “hub metabolites”) are likely to be functionally important and thus present as valuable marker candidates [24]. We previously identified associations between 35 hubs, defined using correlations between metabolites and module eigenvalues exceeding r = 0.65 (termed “module membership”; kME), and cognitive outcomes in the NSHD [10]. As these metabolites were selected on the premise of showing associations with cognitive function, we additionally looked for hubs that may be important in brain imaging outcomes. To do this, we extracted any additional metabolites exceeding the 0.65 threshold [10] in modules showing significant (pFDR < 0.05) associations in the present analysis.

Regression analysis

To allow for direct comparisons, we standardised all continuous predictors and outcomes to a mean of 0 and standard deviation of 1. We then tested relationships between (a) modules and (b) hub metabolites using linear regression (for whole-brain volume and hippocampal volume) and logistic regression (for Aβ status). Model 1 adjusted for basic covariables: sex, blood collection information, age at scan, APOE genotype, and total intracranial volume (for whole-brain volume and hippocampal volume only). As modules were already adjusted for sex and blood clinic information, these covariables were not additionally included for module analyses. Model 2 additionally adjusted for lipid-related factors: BMI and lipid medication use. Finally, model 3 further adjusted for childhood cognitive ability, educational attainment, and SEP (parental and midlife).

Analyses were conducted on each imputed dataset and pooled using Rubin’s rules [25]. Regression assumptions were checked by examination of the residuals. We applied false discovery rate (FDR) correction using the Benjamini–Hochberg procedure [26] with an alpha = 0.05. FDR correction was applied separately to each analysis (module and hub) and each outcome.

Polygenic risk scores

To explore whether levels of key metabolites were influenced by genetic risk for AD, we investigated associations between AD polygenic risk scores (PRS)—a weighted sum of genetic variants associated with a trait or disease—and hub metabolites and modules that significantly associated with an imaging outcome (pFDR < 0.05). We obtained genome-wide association study summary statistics from Kunkle et al. [27] (N = 63,926; 21,982 AD clinically ascertained cases, 41,944 controls), which were used as the base data for PRS analyses. Using PRSice-2 [28], we computed PRS in the NSHD, both including and excluding SNPs in the APOE region (chr 19, GRCh37 coordinates 44,912,079 to 45,912,079) [29]. Two p-value thresholds (PT)—previously identified to be optimal for PRS including and excluding the APOE region—were used for SNP selection: 5 × 10−8 (suggested for APOE region included) and 0.1 (suggested for APOE region excluded) [30], resulting in four PRS. SNPs in linkage disequilibrium (r2 > 0.001 within a 250-kb window) were clumped, and the SNP with the lowest p-value was retained.

We first standardised predictors and outcomes to a mean of 0 and standard deviation of 1. Then, we regressed PRS on key metabolites and modules, adjusting for sex, age, blood collection details, and seven genetic principal components (to control for population stratification). We applied FDR correction to each analysis—module and hub metabolite—using the Benjamini–Hochberg procedure [26] with an alpha = 0.05.

Additional analysis

We conducted several additional analyses to test the robustness of our findings (see Supplementary Notes for full details). In brief, we first investigated whether WGCNA modules, which were curated in the full NSHD, were preserved in the Insight 46 subset. Then, for our main analyses, we applied a more conservative Bonferroni correction to our findings (module: p < 3.57 × 10−3; hub metabolite: p < 1.06 × 10−3). Finally, we explored whether lifestyle and related factors (lifetime smoking, diet, exercise, blood pressure, and alcohol intake) explained any of our results.

Results

Participant characteristics

Participant characteristics can be found in Table 1 (see Supplementary Notes for characteristics split by Aβ status).

Metabolite coexpression network modules

Overall, we identified three modules that showed associations with brain volume outcomes (pFDR < 0.05) and none (p > 0.05) with Aβ status. Full results can be found in Supplementary Table 2 and are visualised in Fig. 2. Results of the fully adjusted model are discussed hereafter.

Fig. 2
figure 2

Source data are available in [7]

Module results. A Heatmap showing relationships between modules and brain imaging outcomes in the basic model (model 1) and final model (model 3). The basic model was adjusted for age, sex, blood clinic location, and APOE genotype, and the final model additionally for BMI, lipid medication use, childhood cognitive ability, educational attainment, and SEP (parental and midlife). Tiles are coloured by effect direction (blue = associated with better outcomes, red = associated with worse outcomes). Effect sizes are presented inside the tiles. Associations significant after multiple testing correction are represented by a solid fill and nominal (p < 0.05) by a fainter fill. Module names in bold were additionally associated with a cognitive outcome in our previous analysis. Data for all models (1-3) are present in Supplementary Table 2. B Table presenting enriched pathways in each module, with the most highly enriched pathways presented first. No pathways were enriched for the tan module.

Full size image

We report associations between higher expression of two lipid modules and smaller whole-brain volumes: the brown module, enriched in diacylglycerol (DAG) and phosphatidylethanolamine (PE) pathways (ß =  − 0.066, 95%CI = [− 0.11, − 0.019], p = 0.006, pFDR = 0.044) and the blue module, enriched in various fatty acids pathways (ß =  − 0.072, 95% CI = [− 0.12, − 0.026], p = 0.0021, pFDR = 0.035). Higher expression of the yellow module, enriched in sphingolipid metabolism and related pathways, was associated with a larger hippocampal volume (ß = 0.14, 95% CI = [0.055, 0.23], p = 0.0017, pFDR = 0.035).

Hub metabolites

We explored associations between 81 metabolites that were highly connected (kME > 0.65) in significant modules identified in 3.2 and 35 that were identified to be hubs in our previous study on cognitive function. Across all models, we report 30 key metabolites after FDR correction, of which 13 were previously associated with cognitive outcomes (Fig. 3). No associations were detected for any metabolite and Aβ status after FDR correction, although some nominal associations were observed (3 cyan and 1 yellow metabolite; Supplementary Table 1 and Fig. 3).

Fig. 3
figure 3

Hub metabolite results. Heatmap showing relationships between key hub metabolites (pFDR < 0.05) and brain imaging outcomes in the basic model (model 1) and final model (model 3), as well as relationships between these metabolites and Alzheimer’s disease polygenic risk scores (best threshold shown). The basic model was adjusted for age, sex, blood clinic location, and APOE genotype, and the final model additionally for BMI, lipid medication use, childhood cognitive ability, educational attainment, and SEP (parental and midlife). Tiles are coloured by effect direction (blue = associated with better outcomes, red = associated with worse outcomes). Effect sizes are presented inside the tiles. Associations significant after multiple testing correction are represented by a solid fill and nominal (p < 0.05) by a fainter fill. Metabolites are organised by both module (indicated via the colour panel on the right) and pathway (specified next to the module colour panel). Metabolite names in bold were additionally associated with a cognitive outcome in our previous analysis. Data for all models (1-3) are present in Supplementary Table 1. AD, Alzheimer’s disease; DMTPA, 2,3 dihydroxy-5-methylthio-4-pentenoic acid; NSHD, National Survey of Health and Development; PRS, polygenic risk score

Full size image

In the fully adjusted model, 22 metabolites were associated (pFDR < 0.05) with an imaging outcome (whole-brain volume: 22; hippocampal volume: 4). Of the 22, 10 metabolites belonged to the yellow module (pathways: sphingolipid metabolism) and were positively associated with larger brain volumes. Twelve metabolites were negatively associated with brain volumes: six belonging to the blue module (pathways: fatty acid (monohydroxy; dicarboxylate), long chain PUFA (n3 and n6), glycerolipid metabolism), four to the brown module (pathways: PE, phosphatidylcholine (PC), DAG), and two to additional modules identified in our previous analysis of cognitive function (pathways: fatty acid metabolism (acyl carnitine); methionine, cysteine, SAM and taurine metabolism).

Polygenic risk scores

We investigated whether modules and hub metabolites that associated with an imaging measure (pFDR < 0.05) were also associated with polygenic risk for AD (APOE region included and excluded). We observed no significant relationships following FDR correction, and no relationships were observed for metabolite modules at either the nominal or adjusted level of significance.

At the nominal threshold, we report relationships between higher AD PRS and decreased levels of five hub metabolites (Fig. 3 and Supplementary Table 3). Three DAG and phosphoethanolamine hub metabolites were associated with the APOE AD PRS only (PT = 5 × 10−8; ß range =  − 0.061 to − 0.050, p range = 0.012 to 0.04, pFDR > 0.05). Two PUFA were associated with the non-APOE AD PRS (PT = 0.1; DPA: ß =  − 0.077, 95% CI = [0.13, − 0.029], p = 0.0018; linolenate: ß =  − 0.054, 95%CI = [− 0.10, − 0.0056], p = 0.029, pFDR > 0.05); these associations weakened with the APOE region additionally included (PT = 0.1; DPA: ß =  − 0.058, 95% CI = [− 0.11, − 0.0097], p = 0.019, linolenate: ß =  − 0.040, 95% CI = [− 0.088, 0.0088], p = 0.11).

Additional analysis

Full results from our additional analyses are discussed in Supplementary Notes. Briefly, modules showed moderate to large preservation in the Insight 46 subset of the NSHD (Supplementary Fig. 1). Following Bonferroni correction (module: p < 3.57 × 10−3; hub metabolite: p < 1.06 × 10−3), all modules and nine metabolites remained associated with an outcome, with 21 no longer reaching the adjusted level of significance. Further adjustment for life course factors resulted in minimal changes (Supplementary Tables 4 and 5).

Discussion

In a population-based cohort, we identified three modules of coexpressed lipids (phospholipids and DAGs, fatty acids, and sphingolipids) that were associated with whole-brain or hippocampal volume and 22 highly connected metabolites within these that present as potential markers for additional study. We report no significant metabolic associations for Aβ status following multiple testing correction, nor for AD polygenic risk in our genetic analyses, although some relationships were seen at the nominal level. Taken together, these findings highlight associations between lipids and later life brain structure, with no strong evidence to suggest relationships are specific to AD-related pathology (as measured through Aβ-PET and PRS).

Fatty acids in whole-brain volume

First, we found that higher expression of the blue module—enriched in several fatty acid pathways—was associated with smaller whole-brain volumes. We identified eight hub metabolites in this module, belonging to fatty acid pathways (long chain PUFA (n3 and n6); monohydroxy; dicarboxylate; medium chain), as well as glycerol from the glycerolipid metabolism pathway. These pathways are tightly linked: fatty acids and glycerol constitute phospholipids and triglycerides, and dicarboxylate and monohydroxy fatty acids are oxidative products of PUFA and other fatty acids.

Upregulation of this module and hub metabolites may thus represent changes in lipid metabolism, including enhanced lipid breakdown, accumulation of free fatty acids and glycerol in the blood, and alterations in fatty acid oxidation—all of which have been linked to neurodegeneration and AD [31,32,33]. Worth noting, however, is that while n6 PUFA have been typically linked to risk of neurodegenerative disease, n3 have been linked to decreased risk [34], although both are an area of contention [35]. Here, we identified one hub (docosapentaenoate; DPA, 22:5n3) to be an n3 fatty acid, contrasting with the general consensus among the literature. We additionally observed some suggestive evidence of an association between AD genetic risk and PUFA, particularly with the APOE region removed; a higher AD PRS was associated with decreased levels of two long chain PUFA hubs (linolenate (18:3n3 or 3n6) and DPA), albeit only at a nominal significance level. These effect directions align with those reported in the literature but were in the unexpected direction based on findings in our imaging analysis. While it is not possible to draw definitive conclusions, we believe these results highlight the need for further work in larger, independent samples.

We further identified two module hubs (16-hydroxypalmitate and hexadecanedioate) which are products of microsomal omega-oxidation—a minor oxidation pathway for fatty acids. Accumulation thus points to defects in mitochondrial β-oxidation pathways, perhaps induced by an overload of free fatty acids or vice versa [36, 37]. Notably, defective β-oxidation and compensatory omega-oxidation pathways are thought to induce oxidative stress [38, 39]—a key mechanism linked to neurodegeneration, which is marked among other things by reduction in brain volume [40]. Nevertheless, to our knowledge, these metabolites have not been linked to brain health and neurodegeneration previously, although they have been found to play a role in other phenotypes, such as mortality and blood pressure [36].

Phospholipids and DAGs in whole-brain volume

Similar to the module of fatty acids, higher expression of the brown module—enriched in PEs and DAGs—associated with smaller whole-brain volumes. These pathways have key roles in membrane structure and cell signalling: PEs are a class of phospholipid which form important components of cellular membranes [41], and are a precursor to DAGs, which are components of cellular membranes and secondary messengers [42]. We reported associations between higher expression of this module and worse short-term memory in our previous analysis of cognitive function in the NSHD, although these were mostly explained by BMI and lipid medication [10]. Here, our results were independent of these factors with minimal attenuations overall.

Within the module, we identified four hub metabolites—two phospholipids and two DAGs. Both subclasses have been previously linked to AD and neurodegeneration [43, 44], and higher blood levels have been hypothesised to represent alterations in membrane integrity and subsequent degradation [45, 46], although some associations in the opposite direction have also been reported [32, 47]. Three module hubs were additionally associated with an AD PRS at the nominal level; this relationship appeared to be driven predominantly by APOE and may therefore reflect pleiotropic pathways. Again, these were in the opposite direction to our findings for whole-brain volume and did not survive multiple testing correction, and so should be interpreted with caution.

Sphingolipids in hippocampal volume and whole-brain volume

We highlighted associations between higher levels of sphingolipids, a lipid class that contain important constituents of cellular membranes [48], and larger hippocampal and whole-brain volumes. Our findings were observed for sphingomyelins in particular, a subclass that are especially abundant in the CNS, where they form pivotal components of neuronal membranes and play key roles in signal transduction [49]. Given their biological role, it is unsurprising that sphingolipids have been previously linked to brain health and pathology [50, 51]. Here, we report associations between a module enriched in sphingolipids and hippocampal volume, as well as several sphingomyelin hub metabolites and whole-brain volume.

We previously reported similar findings for sphingolipids and several cognitive outcomes; however, after we adjusted for childhood cognitive ability and education in particular, these were entirely attenuated [10]. Interestingly, we did not observe the same pattern here, with relationships showing negligible attenuations overall, although it is worth noting that childhood cognitive ability and education show much weaker associations with structural brain measures and thus less likely to confound associations. We hypothesised that attenuations could represent earlier relationships we are unable to capture without longitudinal metabolite data, shared genetic or environmental underpinnings, or confounding by reverse causation, i.e. increased sphingolipid levels may be consequential to higher cognitive function in early life, and education may be capturing shared components. With no earlier measure of brain volume, the latter could extend to our present findings, although previous research has linked sphingomyelins to longitudinal markers of pathology [50, 52, 53]. Another possibility is that sphingolipids may have different involvements in cognitive function and later life brain volume, or may be particularly important during sensitive age periods, for example in cognitive development as well as during vulnerable periods in later life with regard to structural integrity [48]. Expanding this to longitudinal data, alongside interrogating relationships using MR, will allow for a greater insight into our findings.

Limited relationships were seen for Aβ status and AD polygenic risk

Interestingly, we saw limited metabolic relationships for Aβ status; no module showed associations (p > 0.05), and a handful of metabolites were associated at the nominal threshold only. Possibly, there are no robust associations between these metabolites and Aβ, or independent of APOE, although other work has reported differently [5, 7]. Alternatively, this may reflect power, particularly given the relatively young age of Insight 46 participants and the smaller sample that were Aβ PET-positive at this stage. Our measure of amyloid load (PET) reflects the deposition of fibrillar amyloid plaque; it is possible that measuring upstream, soluble forms of Aβ may both increase the numbers who are amyloid positive and allow for an exploration of whether different metabolic pathways are associated with different stages of β-amyloid formation. Further follow-up studies, including CSF measures of soluble Aβ burden, are planned. In addition, five metabolites were associated with an AD PRS at the nominal level, but none survived multiple testing correction, and no other metabolic associations were seen at either threshold. As modules and metabolites were selected for PRS analyses based on showing significant associations in our imaging analyses, which were observed for whole-brain or hippocampal volume, and not Aβ, this suggests that these metabolites may not be specific to AD. Alternatively, it may also reflect power; expanding our work in larger samples is warranted.

Strengths and limitations

Strengths of the study include age-matched cohort with information on a large range of confounders across the life course, including rarely available measures of cognitive development, as well as data on both imaging and subdomains of cognitive function. Moreover, the metabolomics data in this study represent a far more comprehensive proportion of the metabolome than in past clinical metabolomic studies of neuroimaging parameters [7, 53]. Nevertheless, the results of this study should be interpreted in the context of the following limitations. First, our findings may not extend to the general population. Study participants are all white and, compared to the full NSHD cohort, participants enrolled in Insight 46 were on average of slightly better self-rated health, cognitive ability, and SEP. Next, our findings may be subject to residual confounding; further study is needed to disentangle causal relationships. It is additionally possible that some metabolite degradation may have occurred despite storage at – 80 °C [54, 55]. However, strict quality control and best practices were implemented, and factors related to storage time and potential technical differences were adjusted. Finally, as there are currently few cohorts with genetic, serum LC–MS, and brain imaging data, we are not yet able to replicate our analyses elsewhere.

Conclusions

Our findings highlight relationships between groups of lipids and structural brain measures, as well as key metabolites within these that are likely to be driving associations. Future work should be directed towards understanding if these metabolites associate with longitudinal changes in brain volumes and whether relationships are causal; this could advance our understanding of brain health and neurodegeneration and reveal possible targets of intervention.

Availability of data and materials

Anonymised data are available upon request to bona fide researchers (https://skylark.ucl.ac.uk/NSHD/doku.php). Analytical code can be accessed at https://github.com/beckigreen/metabolites-brain-ad-prs.

Abbreviations

Aβ:

Amyloid-beta

AD:

Alzheimer’s disease

BMI:

Body mass index

DAG:

Diacylglycerol

EDTA:

Ethylenediaminetetraacetic acid

FDR:

False discovery rate

FLAIR:

Weighted fluid-attenuated inversion recovery

KEGG:

Kyoto Encyclopaedia of Genes and Genomes

kME:

Module membership

LCMS:

Liquid chromatography-mass spectrometry

MR:

Mendelian randomisation

n3 and n6:

Omega 3 and 6

NSHD:

National Survey of Health and Development

PC:

Phosphatidylcholine

PE:

Phosphatidylethanolamine

PRS:

Polygenic risk score

PUFA:

Polyunsaturated fatty acid

QC:

Quality control

SEP:

Socioeconomic position

SUVR:

Standardised uptake volume ratio

WGCNA:

Weighted gene coexpression network analysis

References

  1. Alawode DOT, Heslegrave AJ, Ashton NJ, Karikari TK, Simrén J, Montoliu-Gaya L, et al. Transitioning from cerebrospinal fluid to blood tests to facilitate diagnosis and disease monitoring in Alzheimer’s disease. J Intern Med. 2021;290(3):583–601.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  2. Beebe K, Kennedy AD. Sharpening precision medicine by a thorough interrogation of metabolic individuality. Comput Struct Biotechnol J. 2016;14:97–105.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  3. Wishart DS. Emerging applications of metabolomics in drug discovery and precision medicine. Nat Rev Drug Discov. 2016;15(7):473–84.

    Article  CAS  PubMed  Google Scholar 

  4. Voyle N, Kim M, Proitsi P, Ashton NJ, Baird AL, Bazenet C, et al. Blood metabolite markers of neocortical amyloid-β burden: discovery and enrichment using candidate proteins. Transl Psychiatry. 20166(1):e719–e719.

  5. Kim M, Snowden S, Suvitaival T, Ali A, Merkler DJ, Ahmad T, et al. Primary fatty amides in plasma associated with brain amyloid burden, hippocampal volume, and memory in the European Medical Information Framework for Alzheimer’s Disease biomarker discovery cohort. Alzheimers Dementia. 2019;15(6):817–27.

    Article  Google Scholar 

  6. de Leeuw FA, Karamujić-Čomić H, Tijms BM, Peeters CFW, Kester MI, Scheltens P, et al. Circulating metabolites are associated with brain atrophy and white matter hyperintensities. Alzheimers Dementia. 2021;17(2):205–14.

    Article  Google Scholar 

  7. Nho K, Kueider-Paisley A, Arnold M, MahmoudianDehkordi S, Risacher SL, Louie G, Serum metabolites associated with brain amyloid beta deposition, cognition and dementia progression. Brain Communications. 2021;3(3):fcab139. https://0-doi-org.brum.beds.ac.uk/10.1093/braincomms/fcab139.

  8. Proitsi P, Kim M, Whiley L, Simmons A, Sattlecker M, Velayudhan L, et al. Association of blood lipids with Alzheimer’s disease: a comprehensive lipidomics analysis. Alzheimers Dementia. 2017;13(2):140–51.

    Article  Google Scholar 

  9. Xu J, Bankov G, Kim M, Wretlind A, Lord J, Green R, et al. Integrated lipidomics and proteomics network analysis highlights lipid and immunity pathways associated with Alzheimer’s disease. Transl Neurodegener. 2020;9(1):36.

    Article  PubMed  PubMed Central  Google Scholar 

  10. Green R, Lord J, Xu J, Maddock J, Kim M, Dobson R, et al. Metabolic underpinnings of late midlife cognitive outcomes: findings from the 1946 British Birth Cohort. Brain Communications. 2021. 4:fcab291.

  11. Stafford M, Black S, Shah I, Hardy R, Pierce M, Richards M, et al. Using a birth cohort to study ageing: representativeness and response rates in the National Survey of Health and Development. Eur J Ageing. 2013;10(2):145–57.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  12. Lane CA, Parker TD, Cash DM, Macpherson K, Donnachie E, Murray-Smith H, et al. Study protocol: Insight 46 – a neuroscience sub-study of the MRC National Survey of Health and Development. BMC Neurol. 2017;17(1):75.

    Article  PubMed  PubMed Central  Google Scholar 

  13. James S-N, Lane CA, Parker TD, Lu K, Collins JD, Murray-Smith H, et al. Using a birth cohort to study brain health and preclinical dementia: recruitment and participation rates in Insight 46. BMC Res Notes. 2018;11(1):885.

    Article  PubMed  PubMed Central  Google Scholar 

  14. Purcell S, Neale B, Todd-Brown K, Thomas L, Ferreira MAR, Bender D, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet. 2007;81(3):559–75.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  15. Lu K, Nicholas JM, Collins JD, James S-N, Parker TD, Lane CA, et al. Cognition at age 70: Life course predictors and associations with brain pathologies. Neurology. 2019;93(23):e2144–56.

    Article  PubMed  PubMed Central  Google Scholar 

  16. Leung KK, Barnes J, Modat M, Ridgway GR, Bartlett JW, Fox NC, et al. Brain MAPS: an automated, accurate and robust brain extraction technique using a template library. Neuroimage. 2011;55(3):1091–108.

    Article  PubMed  Google Scholar 

  17. Jorge Cardoso M, Leung K, Modat M, Keihaninejad S, Cash D, Barnes J, et al. STEPS: Similarity and truth estimation for propagated segmentations and its application to hippocampal segmentation and brain parcelation. Med Image Anal. 2013;17(6):671–84.

    Article  CAS  PubMed  Google Scholar 

  18. Malone IB, Leung KK, Clegg S, Barnes J, Whitwell JL, Ashburner J, et al. Accurate automatic estimation of total intracranial volume: a nuisance variable with less nuisance. Neuroimage. 2015;104:366–72.

    Article  PubMed  Google Scholar 

  19. Proitsi P, Kuh D, Wong A, Maddock J, Bendayan R, Wulaningsih W, et al. Lifetime cognition and late midlife blood metabolites: findings from a British birth cohort. Transl Psychiatry. 2018;8(1):1–11.

    Article  Google Scholar 

  20. van Buuren S, Groothuis-Oudshoorn K. mice: Multivariate imputation by chained equations in R. J Stat Softw. 2011;45(1):1–67.

    Google Scholar 

  21. Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9(1):559.

    Article  PubMed  PubMed Central  Google Scholar 

  22. Langfelder P, Horvath S. Fast R functions for robust correlations and hierarchical clustering. J Stat Softw. 2012;46(11): i11.

    Article  PubMed  PubMed Central  Google Scholar 

  23. Zhang B, Horvath S. A general framework for weighted gene co-expression network analysis. Stat Appl Genet Mol Biol. 2005;4:Article17.

  24. Horvath S, Dong J. Geometric interpretation of gene coexpression network analysis. PLoS Comput Biol. 2008;4(8): e1000117.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Rubin DB. Multiple imputation for nonresponse in surveys. New York: Wiley; 2004. 326 p.

  26. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Stat Soc: Ser B (Methodol). 1995;57(1):289–300.

    Google Scholar 

  27. Kunkle BW, Grenier-Boley B, Sims R, Bis JC, Damotte V, Naj AC, et al. Genetic meta-analysis of diagnosed Alzheimer’s disease identifies new risk loci and implicates Aβ, tau, immunity and lipid processing. Nat Genet. 2019;51(3):414–30.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  28. Euesden J, Lewis CM, O’Reilly PF. PRSice: Polygenic Risk Score software. Bioinformatics. 2015;31(9):1466–8.

    Article  CAS  PubMed  Google Scholar 

  29. Zettergren A, Lord J, Ashton NJ, Benedet AL, Karikari TK, Lantero Rodriguez J, et al. Association between polygenic risk score of Alzheimer’s disease and plasma phosphorylated tau in individuals from the Alzheimer’s Disease Neuroimaging Initiative. Alzheimers Res Therapy. 2021;13(1):17.

    Article  CAS  Google Scholar 

  30. Leonenko G, Baker E, Stevenson-Hoare J, Sierksma A, Fiers M, Williams J, et al. Identifying individuals with high risk of Alzheimer’s disease using polygenic risk scores. Nat Commun. 2021;12(1):4506.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  31. Chew H, Solomon VA, Fonteh AN. Involvement of lipids in Alzheimer’s disease pathology and potential therapies. Front Physiol. 2020;11:598.

    Article  PubMed  PubMed Central  Google Scholar 

  32. Kao Y-C, Ho P-C, Tu Y-K, Jou I-M, Tsai K-J. Lipids and Alzheimer’s disease. Int J Mol Sci. 2020;21(4):1505.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  33. Fonteh AN, Cipolla M, Chiang AJ, Edminster SP, Arakaki X, Harrington MG. Polyunsaturated fatty acid composition of cerebrospinal fluid fractions shows their contribution to cognitive resilience of a pre-symptomatic Alzheimer’s disease cohort. Front Physiol. 2020;11:83.

    Article  PubMed  PubMed Central  Google Scholar 

  34. Dyall SC. Long-chain omega-3 fatty acids and the brain: a review of the independent and shared effects of EPA. DPA and DHA Front Aging Neurosci. 2015;7:52.

    PubMed  Google Scholar 

  35. González-Domínguez R, García-Barrera T, Gómez-Ariza JL. Using direct infusion mass spectrometry for serum metabolomics in Alzheimer’s disease. Anal Bioanal Chem. 2014;406(28):7137–48.

    Article  PubMed  Google Scholar 

  36. Menni C, Graham D, Kastenmüller G, Alharbi NHJ, Alsanosi SM, McBride M, et al. Metabolomic identification of a novel pathway of blood pressure regulation involving hexadecanedioate. Hypertension. 2015;66(2):422–9.

    Article  CAS  PubMed  Google Scholar 

  37. Wanders RJA, Komen J, Kemp S. Fatty acid omega-oxidation as a rescue pathway for fatty acid oxidation disorders in humans. FEBS J. 2011;278(2):182–94.

    Article  CAS  PubMed  Google Scholar 

  38. Natarajan SK, Ibdah JA. Role of 3-hydroxy fatty acid-induced hepatic lipotoxicity in acute fatty liver of pregnancy. Int J Mol Sci. 2018;19(1):322.

    Article  PubMed  PubMed Central  Google Scholar 

  39. Gusdon AM, Song K, Qu S. Nonalcoholic fatty liver disease: pathogenesis and therapeutics from a mitochondria-centric perspective. Oxid Med Cell Longev. 2014;2014: e637027.

    Article  Google Scholar 

  40. Klein JA, Ackerman SL. Oxidative stress, cell cycle, and neurodegeneration. J Clin Invest. 2003;111(6):785–93.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. Fullerton MD, Hakimuddin F, Bonen A, Bakovic M. The development of a metabolic disease phenotype in CTP: phosphoethanolamine cytidylyltransferase-deficient mice. J Biol Chem. 2009;284(38):25704–13.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  42. Carrasco S, Mérida I. Diacylglycerol, when simplicity becomes complex. Trends Biochem Sci. 2007;32(1):27–36.

    Article  CAS  PubMed  Google Scholar 

  43. Toledo JB, Arnold M, Kastenmüller G, Chang R, Baillie RA, Han X, et al. Metabolic network failures in Alzheimer’s disease: a biochemical road map. Alzheimers Dementia. 2017;13(9):965–84.

    Article  Google Scholar 

  44. Wood PL, Medicherla S, Sheikh N, Terry B, Phillipps A, Kaye JA, et al. Targeted lipidomics of fontal cortex and plasma diacylglycerols (DAG) in mild cognitive impairment and Alzheimer’s disease: validation of DAG & accumulation early in the pathophysiology of Alzheimer’s disease. J Alzheimers Dis. 2015;48(2):537–46.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  45. González-Domínguez R, García-Barrera T, Gómez-Ariza JL. Application of a novel metabolomic approach based on atmospheric pressure photoionization mass spectrometry using flow injection analysis for the study of Alzheimer’s disease. Talanta. 2015;131:480–9.

    Article  PubMed  Google Scholar 

  46. Calzada E, Onguka O, Claypool SM. Phosphatidylethanolamine metabolism in health and disease. Int Rev Cell Mol Biol. 2016;321:29–88.

    Article  CAS  PubMed  Google Scholar 

  47. Huo Z, Rana BK, Elman JA, Dong R, Engelman CD, Johnson SC, et al. Metabolic profiling of cognitive aging in midlife. Frontiers in Aging Neuroscience. 2020;12:397.

    Article  Google Scholar 

  48. Olsen ASB, Færgeman NJ. Sphingolipids: membrane microdomains in brain development, function and neurological diseases. Open Biol. 7(5):170069.

  49. Schneider N, Hauser J, Oliveira M, Cazaubon E, Mottaz SC, O’Neill BV, et al. Sphingomyelin in brain and cognitive development: preliminary data. eNeuro. 2019 6(4):ENEURO.0421–18.2019.

  50. Mielke MM, Haughey NJ. Could plasma sphingolipids be diagnostic or prognostic biomarkers for Alzheimer’s disease? Clin Lipidol. 2012;7(5):525–36.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  51. Sliz E, Shin J, Ahmad S, Williams DM, Frenzel S, Gauß F, et al.; NeuroCHARGE Working Group. Circulating Metabolome and White Matter Hyperintensities in Women and Men. Circulation. 2022;145(14):1040–52.

  52. Mielke MM, Haughey NJ, Bandaru VVR, Weinberg DD, Darby E, Zaidi N, et al. Plasma sphingomyelins are associated with cognitive progression in Alzheimer’s Disease. J Alzheimers Dis. 2011;27(2):259–69.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  53. Varma VR, Oommen AM, Varma S, Casanova R, An Y, Andrews RM, et al. Brain and blood metabolite signatures of pathology and progression in Alzheimer disease: A targeted metabolomics study. PLoS Med. 2018;15(1): e1002482.

    Article  PubMed  PubMed Central  Google Scholar 

  54. Kirwan JA, Brennan L, Broadhurst D, Fiehn O, Cascante M, Dunn WB, Schmidt MA, Velagapudi V. Preanalytical processing and biobanking procedures of biological samples for metabolomics research: a white paper, community perspective (for “Precision Medicine and Pharmacometabolomics Task Group”—The Metabolomics Society Initiative). Clin Chem. 2018;64(8):1158–82.

    Article  CAS  PubMed  Google Scholar 

  55. Haid M, Muschet C, Wahl S, Römisch-Margl W, Prehn C, Möller G, Adamski J. Long-term stability of human plasma metabolites during storage at -80 °C. J Proteome Res. 2018;17(1):203–11.

    Article  CAS  PubMed  Google Scholar 

Download references

Acknowledgements

We are very grateful to those study members who helped in the design of the study through focus groups and to the participants both for their contributions to Insight 46 and for their commitments to research over the last seven decades. We are grateful to the radiographers and nuclear medicine physicians at the UCL Institute of Nuclear Medicine and to the staff at the Leonard Wolfson Experimental Neurology Centre at UCL. We are particularly indebted to the support of the late Dr Chris Clark of Avid Radiopharmaceuticals who championed this study from its outset.

We thank the International Genomics of Alzheimer’s Project (IGAP) for providing summary results data for these analyses. The investigators within IGAP contributed to the design and implementation of IGAP and/or provided data but did not participate in analysis or writing of this report.

We acknowledge use of the research computing facility at King’s College London, Rosalind (https://rosalind.kcl.ac.uk), which is delivered in partnership with the National Institute for Health Research (NIHR) Biomedical Research Centres at South London & Maudsley and Guy’s & St. Thomas’ NHS Foundation Trusts and part-funded by capital equipment grants from the Maudsley Charity (award 980) and Guy’s & St. Thomas’ Charity (TR130505). We further acknowledge the use of BioRender.com in the creation of Fig. 1.

Funding

This study is principally funded by grants from Alzheimer’s Research UK (ARUK-PG2014-1946, ARUK-PG2017-1946), the Medical Research Council Dementias Platform UK (CSUB19166), the Wolfson Foundation (PR/ylr/18575), and the Alzheimer’s Association (SG-666374-UK). The genetic analyses are funded by the Brain Research Trust (UCC14191). Florbetapir amyloid tracer was kindly provided by Avid Radiopharmaceuticals (a wholly owned subsidiary of Eli Lilly. The NSHD is funded by the Medical Research Council (MC_UU_00019/1, MC_UU_00019/3).

R.G. is supported by the National Institute for Health Research (NIHR) Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London. M.A.S. was supported by the EPSRC-funded UCL Centre for Doctoral Training in Medical Imaging (EP/L016478/1). T.D.P. was supported by a Wellcome Trust Clinical Research Fellowship (200,109/Z/15/Z) and is supported by a NIHR Clinical Lectureship. D.M.C is supported by the UK Dementia Research Institute which receives its funding from DRI Ltd, funded by the UK Medical Research Council, Alzheimer’s Society and Alzheimer’s Research UK, as well as Alzheimer's Research UK (ARUK-PG2017-1946), the UCL/UCLH NIHR Biomedical Research Centre, and the UKRI Innovation Scholars: Data Science Training in Health and Bioscience (MR/V03863X/1). C.H.S. is supported by an Alzheimer’s Society Junior Fellowship (AS-JF-17–011). R.D. is supported by the following: (1) NIHR Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London, London, UK; (2) Health Data Research UK, which is funded by the UK Medical Research Council, Engineering and Physical Sciences Research Council, Economic and Social Research Council, Department of Health and Social Care (England), Chief Scientist Office of the Scottish Government Health and Social Care Directorates, Health and Social Care Research and Development Division (Welsh Government), Public Health Agency (Northern Ireland), British Heart Foundation and Wellcome Trust; (3) The BigData@Heart Consortium, funded by the Innovative Medicines Initiative-2 Joint Undertaking under grant agreement No. 116074. This Joint Undertaking receives support from the European Union’s Horizon 2020 research and innovation programme and EFPIA; it is chaired by DE Grobbee and SD Anker, partnering with 20 academic and industry partners and ESC; (4) the National Institute for Health Research University College London Hospitals Biomedical Research Centre; (5) the National Institute for Health Research (NIHR) Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London; (6) the UK Research and Innovation London Medical Imaging & Artificial Intelligence Centre for Value Based Healthcare; (7) the National Institute for Health Research (NIHR) Applied Research Collaboration South London (NIHR ARC South London) at King’s College Hospital NHS Foundation Trust. J.M.S. is supported by the Alzheimer's Association (SG-666374-UK) and acknowledges the support of the National Institute for Health Research University College London Hospitals Biomedical Research Centre. M.R. is supported by the MRC (MC_UU_00019/3). P.P. is supported by Alzheimer’s Research UK.

This paper represents independent research part-funded by the National Institute for Health Research (NIHR) Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London. The views expressed are those of the author(s) and not necessarily those of the NHS, the NIHR, or the Department of Health and Social Care.

Author information

Author notes

  1. Jonathan M. Schott, Marcus Richards and Petroula Proitsi contributed equally to this work.

Authors and Affiliations

  1. Institute of Psychiatry, Psychology & Neuroscience, King’s College London, 16 De Crespigny Park, London, SE5 8AB, UK

    Rebecca E. Green, Jodie Lord, Jin Xu, Lachlan Gilchrist, Richard Dobson & Petroula Proitsi

  2. UK National Institute for Health Research (NIHR) Maudsley Biomedical Research Centre, South London and Maudsley Trust, London, UK

    Rebecca E. Green & Richard Dobson

  3. Department of Medical Physics and Biomedical Engineering, Centre for Medical Image Computing (CMIC), University College London (UCL), London, UK

    Marzia A. Scelsi & Carole H. Sudre

  4. Institute of Pharmaceutical Science, King’s College London, London, UK

    Jin Xu & Cristina Legido-Quigley

  5. MRC Unit for Lifelong Health & Ageing at UCL, University College London, Floor 5, MRC LHA at UCL, 1 – 19 Torrington Place, London, WC1E 7HB, UK

    Andrew Wong, Sarah Naomi-James, Dylan M. Williams, Carole H. Sudre & Marcus Richards

  6. Dementia Research Centre, UCL Queen Square Institute of Neurology, 8-11 Queen Square, London, WC1N 3BG, UK

    Sarah Naomi-James, Thomas D. Parker, Christopher A. Lane, Ian B. Malone, David M. Cash, Carole H. Sudre, William Coath, David L. Thomas, Sarah Keuss, Nick C. Fox & Jonathan M. Schott

  7. University College London, Institute of Health Informatics, London, UK

    Alex Handy & Richard Dobson

  8. Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden

    Dylan M. Williams

  9. Department of Brain Sciences, Imperial College London, London, W12 0NN, UK

    Thomas D. Parker

  10. UK DRI Centre for Care Research and Technology, Imperial College London, London, W12 0BZ, UK

    Thomas D. Parker

  11. UK Dementia Research Institute at University College London, London, UK

    David M. Cash & Nick C. Fox

  12. School of Biomedical Engineering and Imaging Sciences, King’s College London, London, UK

    Carole H. Sudre

  13. Neuroradiological Academic Unit, Department of Brain Repair and Rehabilitation, UCL Queen Square Institute of Neurology, University College London, London, UK

    David L. Thomas

  14. Health Data Research UK London, University College London, London, UK

    Richard Dobson

  15. NIHR Biomedical Research Centre at University College London Hospitals NHS Foundation Trust, London, UK

    Richard Dobson

  16. Steno Diabetes Center Copenhagen, Gentofte, Denmark

    Cristina Legido-Quigley

Authors
  1. Rebecca E. Green
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Jodie Lord
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Marzia A. Scelsi
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Jin Xu
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Andrew Wong
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Sarah Naomi-James
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Alex Handy
    View author publications

    You can also search for this author in PubMed Google Scholar

  8. Lachlan Gilchrist
    View author publications

    You can also search for this author in PubMed Google Scholar

  9. Dylan M. Williams
    View author publications

    You can also search for this author in PubMed Google Scholar

  10. Thomas D. Parker
    View author publications

    You can also search for this author in PubMed Google Scholar

  11. Christopher A. Lane
    View author publications

    You can also search for this author in PubMed Google Scholar

  12. Ian B. Malone
    View author publications

    You can also search for this author in PubMed Google Scholar

  13. David M. Cash
    View author publications

    You can also search for this author in PubMed Google Scholar

  14. Carole H. Sudre
    View author publications

    You can also search for this author in PubMed Google Scholar

  15. William Coath
    View author publications

    You can also search for this author in PubMed Google Scholar

  16. David L. Thomas
    View author publications

    You can also search for this author in PubMed Google Scholar

  17. Sarah Keuss
    View author publications

    You can also search for this author in PubMed Google Scholar

  18. Richard Dobson
    View author publications

    You can also search for this author in PubMed Google Scholar

  19. Cristina Legido-Quigley
    View author publications

    You can also search for this author in PubMed Google Scholar

  20. Nick C. Fox
    View author publications

    You can also search for this author in PubMed Google Scholar

  21. Jonathan M. Schott
    View author publications

    You can also search for this author in PubMed Google Scholar

  22. Marcus Richards
    View author publications

    You can also search for this author in PubMed Google Scholar

  23. Petroula Proitsi
    View author publications

    You can also search for this author in PubMed Google Scholar

Consortia

The Insight 46 study team

Contributions

RG, JMS, MR, and PP designed and set up the research and were responsible for the manuscript. RG performed the primary data analysis and wrote the first draft of the manuscript. MAS, AW, SNJ, DMS, TDP, CAL, IBM, DMC, CHS, WC, DLT, SK, NCF, JMS, MR, and PP were involved in data acquisition and generation. RG, JL, JX, AH, LG, MR, and PP provided input on data analysis and interpretation. All authors read and approved the final manuscript.

Corresponding authors

Correspondence to Jonathan M. Schott, Marcus Richards or Petroula Proitsi.

Ethics declarations

Ethics approval and consent to participate

Ethical approval was obtained from the National Research Ethics Service Committee London (14/LO/1173). The study was conducted in accordance with the Declaration of Helsinki, and all participants provided written informed consent.

Consent for publication

Not applicable.

Competing interests

MAS is currently a full-time employee of F. Hoffman-La Roche AG. However, her contributions to this research precede in time and are unrelated to her present employment. J.M.S. has received research funding from Avid Radiopharmaceuticals (a wholly owned subsidiary of Eli Lilly), has consulted for Roche Pharmaceuticals, Biogen, Merck, and Eli Lilly, given educational lectures sponsored by GE Healthcare, Eli Lilly, and Biogen. He is Clinical Advisor to UK DRI and Chief Medical Officer for ARUK.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1:

Supplementary file 1: Table 1. Full hub metabolite results (models 1-3).

Additional file 2:

Supplementary file 2: Table 2. Full module analysis results (models 1-3).

Additional file 3:

Supplementary file 3: Table 3. PRS results.

Additional file 4:

Supplementary file 4: Table 4. Associations between metabolites, modules, outcomes and lifestyle factors.

Additional file 5:

Supplementary file 5: Table 5. Associations between metabolites and modules and whole-brain volume following adjustment for smoking and alcohol intake.

Additional file 6:

Supplementary file 6: Figure 1. A. Plot showing the Zsummary preservation statistics for each module against the module size. Thresholds for moderate and strong evidence of module preservation are indicated. B. Plot showing the median rank for module preservation. Modules are ranked from most preserved (lowest number) to least preserved (highest number).

Additional file 7:

Supplementary file 7: Supplementary Notes. Supplementary Notes (metabolomics acquisition, metabolomic and genomic quality control, polygenic risk scores, software and packages used, additional analyses, participant characteristics split by amyloid status).

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Green, R.E., Lord, J., Scelsi, M.A. et al. Investigating associations between blood metabolites, later life brain imaging measures, and genetic risk for Alzheimer’s disease. Alz Res Therapy 15, 38 (2023). https://0-doi-org.brum.beds.ac.uk/10.1186/s13195-023-01184-y

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://0-doi-org.brum.beds.ac.uk/10.1186/s13195-023-01184-y

Keywords

Alzheimer's Research & Therapy

ISSN: 1758-9193