Skip to main content
Download PDF

Interplay between polygenic propensity for ageing-related traits and the consumption of fruits and vegetables on future dementia diagnosis

BMC Psychiatry volume 22, Article number: 75 (2022) Cite this article

Abstract

Background

Understanding how polygenic scores for ageing-related traits interact with diet in determining a future dementia including Alzheimer’s diagnosis (AD) would increase our understanding of mechanisms underlying dementia onset.

Methods

Using 6784 population representative adults aged ≥50 years from the English Longitudinal Study of Ageing, we employed accelerated failure time survival model to investigate interactions between polygenic scores for AD (AD-PGS), schizophrenia (SZ-PGS) and general cognition (GC-PGS) and the baseline daily fruit and vegetable intake in association with dementia diagnosis during a 10-year follow-up. The baseline sample was obtained from waves 3–4 (2006–2009); follow-up data came from wave 5 (2010–2011) to wave 8 (2016–2017).

Results

Consuming < 5 portions of fruit and vegetables a day was associated with 33–37% greater risk for dementia in the following 10 years depending on an individual polygenic propensity. One standard deviation (1-SD) increase in AD-PGS was associated with 24% higher risk of dementia and 47% higher risk for AD diagnosis. 1-SD increase in SZ-PGS was associated with an increased risk of AD diagnosis by 66%(95%CI = 1.05–2.64) in participants who consumed < 5 portions of fruit or vegetables. There was a significant additive interaction between GC-PGS and < 5 portions of the baseline daily intake of fruit and vegetables in association with AD diagnosis during the 10-year follow-up (RERI = 0.70, 95%CI = 0.09–4.82; AP = 0.36, 95%CI = 0.17–0.66).

Conclusion

A diet rich in fruit and vegetables is an important factor influencing the subsequent risk of dementia in the 10 years follow-up, especially in the context of polygenetic predisposition to AD, schizophrenia, and general cognition.

Peer Review reports

Introduction

Dementia, of which approximately two-thirds constitute Alzheimer’s Disease (AD) cases [1], is associated with a progressive decline of brain functioning leading to a significant loss of autonomy, reduced quality of life and a shortened life expectancy [2]. Confronting the growing burden of dementia requires an identification of the mechanisms by which its risk is exacerbated or attenuated, especially if they entail modifiable risk factors. In turn, this can highlight ways to reduce disease occurrence in the general population.

Considering there is a substantial genetic contribution to dementia [3] and AD [4], one approach to measuring the genetic susceptibility to this disease has been polygenic score (PGSs) analyses [5], which demonstrated that individual differences in risk for dementia and AD diagnoses are driven by thousands of common genetic markers associated with AD [3, 6, 7]. PGSs can also be used to assess propensity to a condition that may never be expressed phenotypically, highlighting shared genetic risk between traits and health conditions [8, 9]. Indeed, recent evidence showed that the polygenic underpinning of AD overlaps with general cognitive ability [10]. While steeper cognitive decline frequently heralds the onset of dementia spectrum [11], similar to dementia, individual differences in cognitive function are also driven by thousands of common genetic markers scattered across the whole genome [12, 13]. More recently, it was highlighted that polygenetic susceptibility to schizophrenia is associated with developmental cognitive deficit in the general population of adults [14]. The link between dementia and schizophrenia is further reiterated by findings demonstrating that schizophrenia is associated with a more than two-fold higher risk of all-cause dementia [15]. Thus, applying PGSs for these traits to estimating an individual risk for dementia diagnosis may provide insights into the genetic make-up of this complex disease informing the search for its biological mechanisms.

Nonetheless, lifestyle factors also contribute to individual-level risk of dementia. Accumulating evidence showed that dietary intake of fruit and vegetables, which serve as primary sources of antioxidants [16], is associated with dementia risk [16, 17]. However, this relationship of fruit and vegetables consumption with dementia risk is inconclusive [17] and susceptible to misinterpretation given previous studies have not accounted for genetic factors, which may also influence dietary habits [18, 19]. Indeed, the results from the twin and family studies alluded to a genetic effect on the number of calories consumed and on the preference for specific nutrients and food items [18].

Because genetic variants are determined randomly at conception and segregated independently of environmental influences, it is often assumed that a genetic susceptibility to dementia is deterministic. However, the genetic risk might be attenuated by a favourable dietary intake of fruit and vegetables [20]. Similarly, inherited DNA variation and dietary intake may contribute independently to a susceptibility to dementia. It is equally feasible that higher genetic propensity may exacerbate the effect of lower fruit and vegetable intake in moderating the risk of future dementia diagnosis. A clearer understanding of this gene-by-environment (G × E) interaction will help understand how genetic predisposition to dementia interacts with the daily intake of fruit and vegetables in determining risk of dementia diagnosis in the general population.

Therefore, we used a large population-representative cohort of older adults to investigate whether higher PGSs for AD, schizophrenia, and general cognition, chosen a priori to index genetic variants associated with ageing-related conditions [3, 10, 14], were associated with dementia diagnosis 10 years later. We further investigated the interactions between these PGSs with daily intake of fruit and vegetables in relation to the risk of dementia diagnosis. For prevention and intervention purposes, it is important to show if daily intake of fruit and vegetables precedes the risk for dementia onset. If it does, then interventions that successfully increase daily intake of fruit and vegetables may also translate into a reduced risk for dementia in the following years. Therefore, the fruit and vegetable daily intake was measured at baseline only. We hypothesised that there would be significant interaction effects between PGSs and the baseline daily intake of fruit and vegetables in association with the dementia diagnosis during the 10-year follow-up period.

Methods and materials

Sample

Data were drawn from the English Longitudinal Study of Ageing (ELSA) study, which is a nationally representative sample of the English population aged ≥50 years [21]. The ELSA study started in 2002–2003 (wave 1) with participants recruited from the Health Survey for England, who were then followed-up every 2 years. To ensure that individuals in their early fifties continued to be represented in the ELSA cohort, the sample is periodically refreshed with younger participants to ensure that the full age spectrum is maintained [21]. Comparisons of ELSA with the national census showed that the baseline sample was representative of the non-institutionalised general population aged 50 and above in the United Kingdom [21]. As participants were asked about their fruit and vegetable intake beginning at wave 3 (2006–2007) for the core members, or wave 4 (2008–2009) for those who joined the study through a refreshment sample [22], these waves formed our baseline. Follow-up data were taken from wave 5 (2010–2011) to wave 8 (2016–2017). We excluded participants with diagnosed dementia, stroke, and schizophrenia at baseline. The full process of sample selection is depicted in Fig. 1. The rates of attrition and mortality in our sample during a 10-year follow-up period are presented in Additional file 1: Table 1. Ethical approval for each of the ELSA waves was granted by the National Research Ethics Service (London Multicentre Research Ethics Committee). All participants gave informed consent. All methods and the study are GDPR compliant.

Fig. 1
figure 1

Flowchart depicting the selection process of the analytical sample

Full size image

Study variables

Ascertainment of dementia cases

Because there are gold standard indicators of dementia [23], dementia was ascertained at each wave using a physician made diagnosis of dementia or Alzheimer’s disease (AD). The diagnoses were reported at each wave by the participants who were capable of participating personally in the study, or by their carer. At each subsequent wave of follow-up, the participants who reported to have been given a diagnosis of dementia were asked if they still had the diagnosis, and reasons for disputing having the diagnosis of dementia. Only those participants who confirmed having dementia or AD diagnosis were classified as having dementia or AD. In the event that ELSA participants were unable to respond to the main interview themselves, the 16 items IQCODE was administered to an informant (family member or long-term caregiver), who knew the respondent very well. The purpose of the IQCODE was to compare the present functional and cognitive performance with the prior performance during the past 2 years employing a 5-point scale, each item is scored from 1 = much improved to 5 = much worse. A threshold of ≥3.38 or more on the IQCODE was used to define dementia [24, 25] with high-sensitivity (0.82) and specificity (0.84) [26]. In the articles that we have previously published using ELSA and cited here, we found that 3.38 performed well [3, 23, 27,28,29,30]. Overall, 83.5% of dementia cases were identified from reports of physician diagnosed dementia or AD, and 16.5% were identified based on the IQCODE score. This approach to identifying dementia incidence, including AD has been widely used in population-based cohorts reinforcing its validity [3, 28,29,30].

Survival age

Survival age was calculated from baseline age when all participants were dementia-free to the date when an ELSA participant received the first self-report physician diagnosis of dementia, or the diagnoses of dementia ascertained through the IQCODE assessment during follow-up. For those without dementia, the survival age was calculated using the period spanning from study entry until either the point of their death, the last wave before dropout, or wave 8. Mortality data were obtained from the National Health Service central register; all individuals included in the analyses provided written consent for the linkage.

Fruit and vegetable intake

Fruit intake was measured by asking participants how many small glasses of fruit juice and how many tablespoons of various types of fruits they consumed the previous day. Vegetable intake was assessed by asking participants how much salad was eaten using a cereal bowl as the standard and how many tablespoons of either vegetables or pulses were consumed the previous day. Consistent with recommendations from the World Health Organization [31] and the U.K. National Health Service which recommend consuming five or more servings of fruits and vegetables per day [32], the total number of fruit and vegetable eaten during 1 day was categorised into a binary variable measuring whether participants consumed the recommended a minimum of 5 daily servings of fruit and vegetables a day (< 5 vs. ≥5 portions) to lower risk of serious long-term conditions [22]. As such, our findings would allow for comparison with previous studies that have used this cut-off point when analysing fruit and vegetable consumption [22].

Covariates

The set of covariates included sex (male) and genetic ancestry (as was measured with principal components (see below)), was included among the covariates to account for any ancestry differences in genetic structures that could bias our results. Specifically, because the ELSA participants had a homogeneous European background, we did not expect population stratification to have a strong influence on our results; therefore, we included four principal components as covariates. Because the apolipoprotein E gene (APOE-ε4) is one of the most significant determinant of dementia risk [33, 34], the APOE-locus should be treated as an independent factor in analyses of dementia sample when calculating polygenic scores [34]. Therefore, we included the ε4 allele of the apolipoprotein E gene (APOE-ε4) status as a confounder in our analyses. Consistent with previous research [35], APOE-ε4 status was defined according to absence (APOE ε2/2, ε2/3 and ε3/3) or presence (APOE ε2/4, ε3/4 and ε4/4) of APOE-ε4 alleles.

Genetic data

Quality control

The genome-wide genotyping was performed at University College London Genomics in 2013–2014 using the Illumina HumanOmni2.5 BeadChips (HumanOmni2.5-4v1, HumanOmni2.5-8v1.3). Quality control followed specifications outlined in Additional file 1. To improve genome coverage, we imputed untyped quality-controlled genotypes to the Haplotype Reference Consortium [36, 37] using the University of Michigan Imputation Server [36]. Post-imputation, we kept variants that were genotyped or imputed at INFO> 0.95, in low linkage disequilibrium (R2 < 0.1) and with Hardy-Weinberg Equilibrium p-value> 10− 5. After the sample quality control 7,179,780 variants were retained for further analyses. An overview of the summary of these quality control steps are presented in Additional file 1: Table 2. To account for any ancestry differences in genetic structures that could bias results, principal components analysis was conducted retaining top principal components [38].

Polygenic scores

Polygenic score for AD (AD-PGS), schizophrenia (SZ-PGS) and general cognition (GC-PGS) were calculated using summary statistics from genome-wide association studies [39,40,41]. As creating polygenic scores based on pruning and p-value threshold has been criticised for discarding potentially important information and limiting prediction accuracy [42, 43], we calculated PGSs using polygenic risk score approach with continuous shrinkage (PRS-CS) [42]. PRS-CS utilises a Bayesian regression framework and places continuous shrinkage priors to address linkage disequilibrium without losing potentially important data. Simulated and real data analyses showed that PRS-CS improves the predictive performance of PGS over existing methods, such as LDpred [43], across a wide range of genetic architectures [42]. There were weak negative correlations between GC-PGS vs AD-PGS (r2 = − 0.04, p = 0.001), and GC-PGS vs SZ-PGS (r2 = − 0.16, p < 0.001) (Additional file 1: Table 5). To aid interpretability of the results, all PGSs were standardised (mean = 0, SD = 1). In this context, a 1 standard deviation increase in the explanatory variable is equivalent to a unit increase in the standardized version of the variable.

Statistical analysis

Imputing missing values

In the present study, some variables had missing values (Additional file 1: Table 3). Given analyses of complete cases (i.e., subset with no missing data in any of variables included for analysis) can result in reduced power and precision of estimates [44, 45], we imputed missing values employing missForest in RStudio version 3.6.2 [46]. MissForest is a nonparametric imputation method based on random forest, which handles continuous and categorical variables equally well and accommodates non-linear relation structures [46, 47]. MissForest has been shown to outperform the well-known imputation methods, such as k-nearest neighbours and parametric multivariate imputation by chained equations (MICE) in the presence of large proportion of missingness, non-linearity and variable interactions. To improve the quality of the imputed missing values we additionally included auxiliary variables [48], such as baseline age, marital status, smoking and presence of any lifelong health conditions. Distributions of the variables before and after imputation are presented in Additional file 1: Table 4, which demonstrate that imputed values are closely aligned with the observed values for all imputed variables.

Association analyses

To investigate the impact of AD-PGS, SZ-PGS and GC-PGS and the baseline daily intake of fruit and vegetables on the risk of dementia diagnosis during the 10-year follow-up period, we utilised the accelerated failure time (AFT) survival model for right-censored data separately for each PGS. To identify the best-fitting parametric model (i.e., exponential, Weibull, lognormal and generalized gamma), we employed the Akaike information criterion [49, 50], which showed that the Weibull model was the most appropriate for our analyses. For each PGS, we constructed the hazard function based on the survival age, as this approach is not restricted to the length of the follow-up period. To investigate whether our findings were applicable to all dementia or were specific to AD cases, we repeated the analyses limiting them to either AD cases only or removing individuals with a diagnosis of AD from the sample (non-AD cases).

Interactions

Interactions between PGSs and the baseline daily fruit and vegetable intake were investigated using multiplicative and additive models. The multiplicative model tests interaction as the departure from multiplicativity according to which the combined effect of two risk factors differs from the product of their individual effects; whereas, the additive interaction tests whether the combined effect of two risk factors differs from the sum of their individual effects [51]. To present the results from the additive interactions, we derived the relative excess risk due to interaction (RERI) and attributable proportion due to interaction (AP) [51, 52], with corresponding 95% confidence intervals. Here, RERI = 0; AP = 0 shows no interaction or exactly equal to additivity of the individual effects of the two risk factors; RERI> 0; AP > 0 highlight positive interaction or more than additivity of the individual effects of the two risk factors; RERI< 0; AP < 0 show negative interaction or less than additivity of the individual effects of the two risk factors.

Sensitivity analyses

We re-ran AFT survival models using the total number of fruit and vegetable eaten as a continuous variable (Additional file 1: Tables 6–7). As the frequency distribution of the total number of fruit and vegetable eaten was severely skewed, this variable was normalised by taking the logarithm to base 10 (log10fruits & vegetables) to allow the use of parametric models (Additional file 1: Figure 1). The p < 0.05 was considered statistically significant. Association analyses were conducted in STATA release 14 (STATA CorpLP, USA).

Results

Sample characteristics

The sample comprised 6784 individuals for whom the quality-controlled genome-wide genotyping and dementia status during follow-up were available (Table 1). Of these, 271 (4.0%) were classified as having dementia (i.e., cases) over the 10-year follow-up, and 6513 (96.0%) remained dementia-free (i.e., controls). Of all dementia cases, 69 (25.5%) had the diagnosis of AD and 109 (40.2%) were APOE-ε4 carriers. The baseline mean age for the entire sample was 64.5 years (standard deviation (SD) = 9.3, median = 63, range = 50–101). 64.6% (N = 175) of future dementia cases compared to 53.0% (N = 3452) controls consumed < 5 portions of fruit and vegetables a day at baseline (x2 = 14.0, p < 0.001).

Table 1 Baseline sample characteristics of ELSA participants
Full size table

PGS, fruit and vegetable intake

Independently from PGSs, consuming < 5 portions of fruit and vegetables a day was associated with a higher risk for dementia diagnosis ranging from 33 to 37% (Table 2). One standard deviation (1-SD) increase in AD-PGS was associated with a 24% greater risk of dementia diagnosis (HR = 1.24, 95%CI = 1.01–1.51) and a 47% greater risk of AD (HR = 1.47, 95%CI = 1.00–2.18) during the 10-year follow-up. There was a significant multiplicative interaction effect between SZ-PGS and the baseline daily intake of fruit and vegetables in relation to AD diagnosis risk during follow-up (HR = 1.66, 95%CI = 1.05–2.64). Accordingly, 1-SD increase in SZ-PGS was associated with an increased risk of AD diagnosis by an average 66% in participants who consumed < 5 portions of fruit or vegetables a day at baseline. Having a higher PGS for general cognition was associated with a reduced risk for non-AD diagnosis (HR = 0.80, 95%CI = 0.65–0.98). There was a significant additive interaction between GC-PGS and < 5 portions of the baseline daily intake of fruit and vegetables in association with AD diagnosis during the 10-year follow-up (RERI = 0.70, 95%CI = 0.09–4.82; AP = 0.36, 95%CI = 0.17–0.66) (Additional file 1: Table 6).

Table 2 Multivariate AFT model estimating associations of PGSs for ageing-related traits and risk for dementia diagnosis
Full size table

Sensitivity analyses

When using a continuous variable of the baseline daily number of fruit and vegetable consumed, the results related to AD-PGS remained unchanged (Additional file 1: Table 6). 1-SD increase in SZ-PGS was associated with a greater risk of dementia diagnosis by an average 24% during the 10-year follow-up (HR = 1.24, 95%CI = 1.04–1.49). There was a significant multiplicative interaction effect between SZ-PGS and the baseline daily intake of fruit and vegetables in relation to the risk of dementia (HR = 0.84, 95%CI = 0.71–0.99) and AD diagnosis (HR = 0.64, 95%CI = 0.47–0.88) during follow-up. Accordingly, 1-SD increase in SZ-PGS was associated with a decreased risk of dementia and AD diagnosis by an average 16 and 36%, respectively, per every portion of fruit or vegetables. All previous findings related to GC-PGS and participants consuming > 5 servings of fruit and vegetables did not remain when we used the baseline fruit and vegetable consumption as a continuous variable (Additional file 1: Tables 6–7).

Discussion

To our knowledge, this is the first study to have investigated interactions between polygenic propensity for AD, schizophrenia and general cognition, and the baseline daily intake of fruit and vegetables in estimating risk for dementia and AD diagnoses in the following 10 years using a large population-representative sample of older adults. Consistent with the genetic liability threshold model according to which the combined effect of many genetic risk variants with other factors causes an individual to cross the threshold leading to the development of a condition [53], we identified significant interactions between the polygenic propensity for schizophrenia and the baseline daily fruit and vegetable intake in association with an increased risk for dementia and AD onset in the following 10 years. This, combined with an assertion that a long prodromal phase of AD allows for the modification of a lifestyle-related factor toward reducing AD risk [54], may suggest that the earlier a healthy lifestyle approach is adopted the more protective against AD it would be.

Consistent with the evidence-based support for fruit and vegetable intake as an important factor influencing risk for dementia [16, 17], we found that consuming less than 5 servings of fruit and vegetable a day was associated with an increased risk of dementia in the following 10 years independently from the genetic propensity for this disease and the APOE-ε4 status. As pathologic processes involved in dementia include oxidative stress [55] and inflammation [56], consuming less fruit and vegetable than recommended to lower risk of serious long-term conditions [22] may lead to diminished numbers of antioxidant and anti-inflammatory compounds present in the brain to decrease risk for oxidative damage, down-regulate inflammation and strengthen the neurons’ antioxidant defence [57]. Consequently, the brain’s ability to withstand more pathology before clinical symptoms of dementia become detectable may be significantly decreased. While these findings reiterate that increasing intake of fruits and vegetables to at least to 5 portions a day may lower the risk of dementia; the non-significant relationship of fruit and vegetable intake with AD risk may reflect the relatively small sample of AD cases available in the ELSA cohort.

Our results further indicate that a higher aggregate of loci for AD may exert its effect by accelerating the clinical presentation of dementia, and to a greater extent of AD. This finding reiterates the results from previous studies [3, 6, 7] and highlights the existence of specific population subgroups that may be at higher risk of dementia based on their polygenetic loading. Some of the mechanisms put forward to explain this relationship assert that common genetic variants associated with AD may affect the immune response, regulation of endocytosis, cholesterol transport and protein ubiquitination [58]. Moreover, we found that higher GC-PGS, which services as an indicator for polygenic predisposition for higher general cognition [12], was associated with a reduced risk for non-AD diagnosis during the 10-year follow-up period in participants. This is consistent with the notion that a higher polygenic propensity to general cognition serves as a protective factor against dementia-related outcomes. However, the interaction effect showed that the protective effect of higher polygenic propensity for a better cognition may be insufficient to counteract the impact of consuming less fruits and vegetables a day than recommended by the World Health Organization [31] and the U.K. National Health Service against developing the non-AD in the following 10 years. Nonetheless, these results did not remain when the analyses were repeated using the baseline fruit and vegetable consumption as a continuous variable. This may imply that categorising the total consumption of fruit and vegetable based on the recommended a minimum of 5 daily servings of fruit and vegetables a day (< 5 vs. ≥5 portions) to lower risk of serious long-term conditions [22] captures important risks that the continuous variable does not.

We further detected a modifying effect of polygenic predisposition to schizophrenia on the association between consuming less than 5 portions of fruit and vegetable and AD diagnosis in the following 10 years. Specifically, a higher genetic load of schizophrenia alleles appears to increase risk of AD diagnosis, but not dementia, among older individuals who did not have a diet rich in fruits and vegetables. These findings, in combination with previous reports that AD onset is often associated with psychosis [59,60,61,62], may suggest that multiple differently regulated aetiological pathways may give rise to similar clinical presentation, but only in the context of environmental exposures, such as daily intake of fruit and vegetables. As neuroinflammation has been implicated in both of AD and schizophrenia [63, 64], it may be hypothesised that fruit and vegetables attenuate the role played by genetic markers for schizophrenia in association with AD diagnosis via antioxidant and anti-inflammatory properties [16].

Methodological considerations

In the present study, we analysed a large sample of adults who were nationally representative of older adults in England. Our study further included an almost equal proportion of women and men all of whom were from socio-economically diverse backgrounds. The present study has a unique value given the current sparseness of data on diet before dementia onset [19]. The further strengths of this study included its prospective collection of information on dementia diagnose and risk factors. Nonetheless, several limitations warrant a discussion. Using doctor diagnosis to identify most dementia cases may imply that the presented dementia and AD cases might have been underestimated [3]. Although dementia and AD were ascertained using a combined algorithm based on a physician made diagnosis and a higher score on the informant reports (IQCODE), it is still reliant on a self-reported diagnosis reported by either the participant themselves or their carers and render more severe cases. As it is estimated that dementia remains undetected in almost 30 to 50% of primary care patients in the UK [65], we cannot exclude a possibility that some participants within the “dementia-free’ group may have been the preclinical stages of dementia and who, if followed for long enough, might eventually develop dementia. Because all covariates that were included in the models were set at birth, we minimised chances of collider bias effecting our findings [66]; on the other hand, however, because of this nature of the analyses, we did not adjust the confounding effect on the dietary intake.

It is being argued that when investigating the G × E interaction to properly control for confounders, the covariate × environment and the covariate × gene interaction terms in the same model ought to be included [67]. However, considering that we have 13 covariates and small numbers in the outcomes, entering 36 interaction terms in each model is likely to lead to overfitting [68]. Therefore, to avoid overfitting, we were unable to adjust our models for interactions between the covariates as advised [67]. Because the construction of PGSs is largely dependent on the availability of the summary statistics from genome-wide association studies (GWASs), the vast majority of which is based on the participants of European descent [69], further work is necessary to develop the PGSs model in non-white populations. Further, the methodology employed to collect data on the baseline daily consumption of fruit and vegetables may not necessarily represent a long-term habit. The measures of the baseline fruit and vegetable consumption may have been restricted because intake during the previous day rather than a typical day was queried [22]. The reason for focusing on the previous day is that memory is stronger for recent experiences. Finally, the proportion of missing data in the present study was comparable to other population-representative longitudinal cohort of similar design [21, 70, 71] and within the range for missForest to handle it efficiently [72, 73].

Conclusion

Our findings reiterate that increasing intake of fruits and vegetables to at least 5 portions a day may lower the risk of dementia and AD independently from the genetic propensity for this disease and the APOE-ε4 status. Consuming at least 5 portions of fruit and vegetable a day may lower the risk of dementia and AD, especially among those who have higher polygenic predisposition to AD and schizophrenia. Future emphasis on promoting healthy behaviours especially in later life, although challenging, may prove valuable in reducing the public health burden of dementia.

Availability of data and materials

The English Longitudinal Study of Ageing (ELSA) was developed by a team of researchers based at University College London, the Institute for Fiscal Studies and the National Centre for Social Research. The data are linked to the UK Data Archive and freely available through the UK data services and can be accessed here: https://discover.ukdataservice.ac.uk.

Abbreviations

AD:

Alzheimer’s diagnosis

APOE:

Apolipoprotein E

ELSA:

English Longitudinal Study of Ageing

GC:

General cognition

GWAS:

Genome-wide association study

IQCODE:

Informant Questionnaire on Cognitive Decline in the Elderly

PGS:

Polygenic score

PRS-CS:

Polygenic risk score approach with continuous shrinkage

RERI:

Relative excess risk due to interaction

SZ:

Schizophrenia

References

  1. Hendrie HC. Epidemiology of dementia and Alzheimer's disease. Am J Geriatr Psychiatry. 1998;6(2 Suppl 1):S3–18.

    Article  PubMed  CAS  Google Scholar 

  2. Ljubenkov PA, Geschwind MD. Dementia. Semin Neurol. 2016;36(4):397–404.

    Article  PubMed  Google Scholar 

  3. Ajnakina O, Cadar D, Steptoe A. Interplay between socioeconomic markers and polygenic predisposition on timing of dementia diagnosis. J Am Geriatr Soc. 2020.

  4. Gatz M, et al. Role of genes and environments for explaining Alzheimer disease. Arch Gen Psychiatry. 2006;63(2):168–74.

    Article  PubMed  Google Scholar 

  5. Dudbridge F. Power and predictive accuracy of polygenic risk scores. PLoS Genet. 2013;9(3):e1003348.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  6. Lee SH, et al. Estimation and partitioning of polygenic variation captured by common SNPs for Alzheimer's disease, multiple sclerosis and endometriosis. Hum Mol Genet. 2013;22(4):832–41.

    Article  PubMed  CAS  Google Scholar 

  7. Ridge PG, et al. Alzheimer's disease: analyzing the missing heritability. PLoS One. 2013;8(11):e79771.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  8. Bellou E, Stevenson-Hoare J, Escott-Price V. Polygenic risk and pleiotropy in neurodegenerative diseases. Neurobiol Dis. 2020:104953.

  9. Gandal MJ, et al. Shared molecular neuropathology across major psychiatric disorders parallels polygenic overlap. Science. 2018;359(6376):693–7.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  10. Anttila V, et al. Analysis of shared heritability in common disorders of the brain. Science. 2018;360(6395).

  11. Rajan KB, et al. Cognitive impairment 18 years before clinical diagnosis of Alzheimer disease dementia. Neurology. 2015;85(10):898–904.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  12. Davies G, et al. Study of 300,486 individuals identifies 148 independent genetic loci influencing general cognitive function. Nat Commun. 2018;9(1):2098.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  13. Savage JE, et al. Genome-wide association meta-analysis in 269,867 individuals identifies new genetic and functional links to intelligence. Nat Genet. 2018;50(7):912–9.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  14. Kępińska AP, et al. Schizophrenia polygenic risk predicts general cognitive deficit but not cognitive decline in healthy older adults. Transl Psychiatry. 2020;10(1):422.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  15. Ribe AR, et al. Long-term risk of dementia in persons with schizophrenia: a Danish population-based cohort study. JAMA Psychiatry. 2015;72(11):1095–101.

    Article  PubMed  Google Scholar 

  16. Morris MC, et al. Dietary intake of antioxidant nutrients and the risk of incident Alzheimer disease in a biracial community study. Jama. 2002;287(24):3230–7.

    Article  PubMed  CAS  Google Scholar 

  17. Jiang X, et al. Increased consumption of fruit and vegetables is related to a reduced risk of cognitive impairment and dementia: Meta-analysis. Front Aging Neurosci. 2017;9:18.

    PubMed  PubMed Central  Google Scholar 

  18. Rankinen T, Bouchard C. Genetics of food intake and eating behavior phenotypes in humans. Annu Rev Nutr. 2006;26:413–34.

    Article  PubMed  CAS  Google Scholar 

  19. Hughes TF, et al. Midlife fruit and vegetable consumption and risk of dementia in later life in Swedish twins. Am J Geriatr Psychiatry. 2010;18(5):413–20.

    Article  PubMed  PubMed Central  Google Scholar 

  20. Lourida I, et al. Association of Lifestyle and Genetic Risk with Incidence of dementia. Jama. 2019;322(5):430–7.

    Article  PubMed  PubMed Central  Google Scholar 

  21. Steptoe A, et al. Cohort profile: the English longitudinal study of ageing. Int J Epidemiol. 2013;42(6):1640–8.

    Article  PubMed  Google Scholar 

  22. Boehm JK, et al. Longitudinal associations between psychological well-being and the consumption of fruits and vegetables. Health Psychol. 2018;37(10):959–67.

    Article  PubMed  PubMed Central  Google Scholar 

  23. Cadar D, et al. Cohort profile update: the harmonised cognitive assessment protocol sub-study of the English longitudinal study of ageing (ELSA-HCAP). Int J Epidemiol. 2020.

  24. Cadar D, et al. Individual and area-based socioeconomic factors associated with dementia incidence in England: evidence from a 12-year follow-up in the English longitudinal study of ageing. JAMA. Psychiatry. 2018.

  25. Jorm AF. A short form of the informant questionnaire on cognitive decline in the elderly (IQCODE): development and cross-validation. Psychol Med. 1994;24(1):145–53.

    Article  PubMed  CAS  Google Scholar 

  26. Quinn TJ, et al. Informant questionnaire on cognitive decline in the elderly (IQCODE) for the diagnosis of dementia within community dwelling populations. Cochrane Database Syst Rev. 2014;4:Cd010079.

    Google Scholar 

  27. Ma Y, et al. Higher risk of dementia in English older individuals who are overweight or obese. Int J Epidemiol. 2020;49(4):1353–65.

    Article  PubMed  PubMed Central  Google Scholar 

  28. Deckers K, et al. Modifiable risk factors explain socioeconomic inequalities in dementia risk: evidence from a population-based prospective cohort study. J Alzheimers Dis. 2019.

  29. Hackett RA, et al. Social engagement before and after dementia diagnosis in the English longitudinal study of ageing. PLoS One. 2019;14(8):e0220195.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  30. Rafnsson SB, et al. Loneliness, social integration, and incident dementia over 6 years: prospective findings from the English longitudinal study of ageing. J Gerontol B Psychol Sci Soc Sci. 2017.

  31. World Health Organization, F.a.A.O.o.t.U.N., Diet, nutrition and the prevention of chronic diseases: Report of a joint WHO/FAO expert consultation WHO Technical Report Series, No. 916 Geneva: World Health Organization., 2003.

  32. Naska A, et al. Fruit and vegetable availability among ten European countries: how does it compare with the 'five-a-day' recommendation? DAFNE I and II projects of the European Commission. Br J Nutr. 2000;84(4):549–56.

    Article  PubMed  CAS  Google Scholar 

  33. Escott-Price V, et al. Common polygenic variation enhances risk prediction for Alzheimer's disease. Brain. 2015;138(Pt 12):3673–84.

    Article  PubMed  PubMed Central  Google Scholar 

  34. Ware EB, et al. Considering the APOE locus in Alzheimer's disease polygenic scores in the health and retirement study: a longitudinal panel study. BMC Med Genet. 2020;13(1):164.

    CAS  Google Scholar 

  35. Zhang A, et al. Brain APOE expression quantitative trait loci-based association study identified one susceptibility locus for Alzheimer's disease by interacting with APOE epsilon4. Sci Rep. 2018;8(1):8068.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  36. Das S, et al. Next-generation genotype imputation service and methods. Nat Genet. 2016;48(10):1284–7.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  37. McCarthy S, et al. A reference panel of 64,976 haplotypes for genotype imputation. Nat Genet. 2016;48(10):1279–83.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  38. Price AL, et al. Principal components analysis corrects for stratification in genome-wide association studies. Nat Genet. 2006;38(8):904–9.

    Article  CAS  PubMed  Google Scholar 

  39. Schulze TG, et al. Molecular genetic overlap in bipolar disorder, schizophrenia, and major depressive disorder. World J Biol Psychiatry. 2014;15(3):200–8.

    Article  PubMed  Google Scholar 

  40. Jansen IE, et al. Genome-wide meta-analysis identifies new loci and functional pathways influencing Alzheimer's disease risk. Nat Genet. 2019;51(3):404–13.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  41. Davies G, et al. Genetic contributions to variation in general cognitive function: a meta-analysis of genome-wide association studies in the CHARGE consortium (N=53949). Mol Psychiatry. 2015;20(2):183–92.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  42. Ge T, et al. Polygenic prediction via Bayesian regression and continuous shrinkage priors. Nat Commun. 2019;10(1):1776.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  43. Vilhjalmsson BJ, et al. Modeling linkage disequilibrium increases accuracy of polygenic risk scores. Am J Hum Genet. 2015;97(4):576–92.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  44. Moons KG, et al. Using the outcome for imputation of missing predictor values was preferred. J Clin Epidemiol. 2006;59(10):1092–101.

    Article  PubMed  Google Scholar 

  45. Zhao Y, Long Q. Multiple imputation in the presence of high-dimensional data. Stat Methods Med Res. 2016;25(5):2021–35.

    Article  PubMed  Google Scholar 

  46. Stekhoven DJ, Peter B. MissForest—non-parametric missing value imputation for mixed-type data. Bioinformatics. 2012;28(1):112–8.

    Article  PubMed  CAS  Google Scholar 

  47. Oba S, et al. A Bayesian missing value estimation method for gene expression profile data. Bioinformatics. 2003;19(16):2088–96.

    Article  PubMed  CAS  Google Scholar 

  48. Collins LM, Schafer JL, Kam CM. A comparison of inclusive and restrictive strategies in modern missing data procedures. Psychol Methods. 2001;6(4):330–51.

    Article  PubMed  CAS  Google Scholar 

  49. Royston P, Parmar MK. Flexible parametric proportional-hazards and proportional-odds models for censored survival data, with application to prognostic modelling and estimation of treatment effects. Stat Med. 2002;21(15):2175–97.

    Article  PubMed  Google Scholar 

  50. George B, Seals S, Aban I. Survival analysis and regression models. J Nucl Cardiol. 2014;21(4):686–94.

    Article  PubMed  PubMed Central  Google Scholar 

  51. Knol MJ, et al. Estimating measures of interaction on an additive scale for preventive exposures. Eur J Epidemiol. 2011;26(6):433–8.

    Article  PubMed  PubMed Central  Google Scholar 

  52. Knol MJ, et al. Estimating interaction on an additive scale between continuous determinants in a logistic regression model. Int J Epidemiol. 2007;36(5):1111–8.

    Article  PubMed  Google Scholar 

  53. Falconer DS. The inheritance of liability to certain diseases, estimated from the incidence among relatives. Ann Hum Genet. 1965;29:51–71.

    Article  Google Scholar 

  54. Livingston G, et al. Dementia prevention, intervention, and care. Lancet. 2017;390(10113):2673–734.

    Article  PubMed  Google Scholar 

  55. Lovell MA, Markesbery WR. Oxidative damage in mild cognitive impairment and early Alzheimer's disease. J Neurosci Res. 2007;85(14):3036–40.

    Article  PubMed  CAS  Google Scholar 

  56. Heneka MT, O'Banion MK. Inflammatory processes in Alzheimer's disease. J Neuroimmunol. 2007;184(1–2):69–91.

    Article  PubMed  CAS  Google Scholar 

  57. Steele M, Stuchbury G, Münch G. The molecular basis of the prevention of Alzheimer's disease through healthy nutrition. Exp Gerontol. 2007;42(1–2):28–36.

    Article  PubMed  CAS  Google Scholar 

  58. IGAP. I.G.o.a.s.D.C., convergent genetic and expression data implicate immunity in Alzheimer's disease. Alzheimers Dement. 2015;11(6):658–71.

    Article  Google Scholar 

  59. Lyketsos CG, et al. Mental and behavioral disturbances in dementia: findings from the Cache County study on memory in aging. Am J Psychiatry. 2000;157(5):708–14.

    Article  PubMed  CAS  Google Scholar 

  60. Leroi I, et al. The epidemiology of psychosis in dementia. Am J Geriatr Psychiatry. 2003;11(1):83–91.

    Article  PubMed  Google Scholar 

  61. Merriam AE, et al. The psychiatric symptoms of Alzheimer's disease. J Am Geriatr Soc. 1988;36(1):7–12.

    Article  PubMed  CAS  Google Scholar 

  62. Berrios GE, Brook P. Visual hallucinations and sensory delusions in the elderly. Br J Psychiatry. 1984;144:662–4.

    Article  PubMed  CAS  Google Scholar 

  63. Heneka MT, et al. Neuroinflammation in Alzheimer's disease. Lancet Neurol. 2015;14(4):388–405.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  64. Steiner J, et al. Increased prevalence of diverse N-methyl-D-aspartate glutamate receptor antibodies in patients with an initial diagnosis of schizophrenia: specific relevance of IgG NR1a antibodies for distinction from N-methyl-D-aspartate glutamate receptor encephalitis. JAMA Psychiatry. 2013;70(3):271–8.

    Article  PubMed  Google Scholar 

  65. Connolly A, et al. Underdiagnosis of dementia in primary care: variations in the observed prevalence and comparisons to the expected prevalence. Aging Ment Health. 2011;15(8):978–84.

    Article  PubMed  Google Scholar 

  66. Arnold KF, et al. Reflection on modern methods: generalized linear models for prognosis and intervention-theory, practice and implications for machine learning. Int J Epidemiol. 2021;49(6):2074–82.

    Article  PubMed  Google Scholar 

  67. Keller MC. Gene x environment interaction studies have not properly controlled for potential confounders: the problem and the (simple) solution. Biol Psychiatry. 2014;75(1):18–24.

    Article  PubMed  Google Scholar 

  68. Austin PC, Steyerberg EW. Events per variable (EPV) and the relative performance of different strategies for estimating the out-of-sample validity of logistic regression models. Stat Methods Med Res. 2017;26(2):796–808.

    Article  PubMed  Google Scholar 

  69. Martin AR, et al. Clinical use of current polygenic risk scores may exacerbate health disparities. Nat Genet. 2019;51(4):584–91.

    Article  PubMed  PubMed Central  CAS  Google Scholar 

  70. Sonnega A, et al. Cohort profile: the health and retirement study (HRS). Int J Epidemiol. 2014;43(2):576–85.

    Article  PubMed  PubMed Central  Google Scholar 

  71. Power C, Elliott J. Cohort profile: 1958 British birth cohort (National Child Development Study). Int J Epidemiol. 2006;35(1):34–41.

    Article  PubMed  Google Scholar 

  72. Ajnakina O, et al. Development and validation of prediction model to estimate 10-year risk of all-cause mortality using modern statistical learning methods: a large population-based cohort study and external validation. BMC Med Res Methodol. 2021;21(1):8.

    Article  PubMed  PubMed Central  Google Scholar 

  73. Ajnakina O, et al. Predicting onset of early- and late-treatment resistance in first-episode schizophrenia patients using advanced shrinkage statistical methods in a small sample. Psychiatry Res. 2020;294:113527.

    Article  PubMed  Google Scholar 

Download references

Acknowledgements

The authors express gratitude to the ELSA research participants, and the professionals involved in conducting the study.

Funding

The English Longitudinal Study of Ageing is funded by the National Institute on Aging (grant RO1AG17644) and by a consortium of UK government departments coordinated by the Economic and Social Research Council (ESRC). O.A. is funded by the National Institute for Health Research (NIHR) (NIHR Post-Doctoral Fellowship - PDF-2018-11-ST2–020). E.F. is funded by the Medical Research Council (MRC) (MRC Doctoral Training Programme - MR/N013867/1). The views expressed in this publication are those of the authors and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health and Social Care.

Author information

Authors and Affiliations

  1. Department of Behavioural Science and Health, Institute of Epidemiology and Health Care, University College London, 1-19 Torrington Place, London, WC1E 7HB, UK

    Emma Ruby Francis, Dorina Cadar, Andrew Steptoe & Olesya Ajnakina

  2. Brighton and Sussex Medical School, Brighton, East Sussex, UK

    Dorina Cadar

  3. Department of Biostatistics & Health Informatics, Institute of Psychiatry, Psychology and Neuroscience, King’s College London, London, UK

    Olesya Ajnakina

Authors
  1. Emma Ruby Francis
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Dorina Cadar
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Andrew Steptoe
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Olesya Ajnakina
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

EF was responsible for data analysis and drafting the report. DC and AS were responsible for the study conception and design, and the critical revision of the report. OA was responsible for study conception and design, preparation of data, data analysis and finalising the final draft of the manuscript. All authors were responsible for interpretation of the results, critical revision and all approved the final version of the report.

Corresponding author

Correspondence to Olesya Ajnakina.

Ethics declarations

Ethics approval and consent to participate

Ethical approval for each of the ELSA waves was granted by the National Research Ethics Service (London Multicentre Research Ethics Committee). All participants gave informed consent. All methods and the study are GDPR compliant.

Consent for publication

Not applicable.

Competing interests

Not applicable.

Additional information

Publisher’s Note

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

Supplementary Information

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

Francis, E.R., Cadar, D., Steptoe, A. et al. Interplay between polygenic propensity for ageing-related traits and the consumption of fruits and vegetables on future dementia diagnosis. BMC Psychiatry 22, 75 (2022). https://doi.org/10.1186/s12888-022-03717-5

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12888-022-03717-5

Keywords

BMC Psychiatry

ISSN: 1471-244X

Contact us