Circulating asymmetric dimethylarginine and cognitive decline: A 4‐year follow‐up study of the 1936 Aberdeen Birth Cohort

The underlying mechanisms leading to dementia and Alzheimer's disease (AD) are unclear. Asymmetric dimethylarginine (ADMA), an endogenous inhibitor of nitric oxide synthase, may be associated with cognitive decline, but population‐based evidence is lacking.


| INTRODUCTION
Alzheimer's disease (AD) is a neurodegenerative disorder characterised by a rapid decline in cognition in old age, which is distinct from the normal ageing process. The causes of late onset AD are largely unknown and despite extensive research, there is still no clear consensus on blood-based biomarkers or effective treatments. 1,2 The vast majority of research has focused on a set of histopathological hallmarks which are often present in diseased brains. However, evidence from observational studies suggests a poor correlation between plaque density and degree of dementia in AD, 3,4 and multiple pharmacological interventions targeting these histopathological hallmarks have proven ineffective. 5 There is growing observational evidence for an overlap of risk factors in AD and cardiovascular disease. For example, diabetes is associated with a 2-to 3-fold increased risk of vascular events 6,7 but also a 50% increased risk of dementia 8 and a 20% increase in the rate of cognitive decline. [9][10][11] Furthermore, large genomewide association studies have confirmed associations between 'vascular risk' genes and the development of AD, 12 with the strongest known genetic risk factor being a mutation in the apolipoprotein E4 gene, which encodes a cholesterol transporter protein. 13 Vascular endothelial dysfunction, an abnormality that is associated with the presence of virtually all known cardiovascular risk factors, is frequently seen in AD. Despite this evidence, there is a lack of population-based research investigating the roles of vascular changes, oxidative stress, lipid metabolism and inflammation in AD pathogenesis. 14,15 Nitric oxide (NO) has an established role as a mediator in the pathogenesis of cardiovascular disease 16 but also potentially in AD. 17,18 Impaired NO release from dysfunctional endothelial cells precedes reduced blood flow and regional metabolic deficiency. This may result in hypoxic events, leading to activation of the immune response and resultant cell death. 17 In support of this hypothesis, recent interventions to increase the concentrations of endogenously produced NO have halted AD pathogenesis in mice. 19 Furthermore, abnormal NO signalling has been reported in the brains of both human sufferers and animal models of AD. 20,21 Asymmetric dimethylarginine (ADMA) is an endogenous inhibitor of nitric oxide synthase (NOS), an essential enzyme for the synthesis of NO. 17 Therefore, ADMA may play an important pathophysiological role in the abnormal NO signalling observed in both CVD and AD and could be a potential target for novel therapeutic interventions.
A number of small-scale studies have investigated the association between ADMA concentrations and cognitive decline in older adults. [22][23][24][25] However, no study has adjusted for childhood intelligence, which explains up to 50% of cognitive decline in old age. [26][27][28] Investigating the association between ADMA and cognitive decline with full adjustment for potential confounders could have implications for presymptomatic diagnosis of AD and may reveal novel drug targets.
This study aims to investigate the relationship between plasma ADMA concentrations and the change in cognitive performance across four years in the 1936 Aberdeen Birth Cohort (ABC36) study.

| Study population
The ABC36 is a well-studied cohort population described in greater depth elsewhere. 29

| Study procedures
Assessment interviews were conducted in three stages, carried out by a trained research nurse. The first section comprised assessment of demographic and clinical data, which included smoking and alcohol history, family history of dementia, diabetes or stroke and a physical examination including measurement of body mass index (BMI). Written informed consent was obtained from all participants at the start of the interview. All study procedures were approved by the Grampian Research Ethics Committee (GREC).
Occupation, deprivation scores and years of formal education were used as a composite measure of socioeconomic status (SES).
Occupation was categorised in accordance with the Standard Occupational Classification issued by the Scottish Government. 31 Each participant was given a deprivation code corresponding to the postcode of their home address in accordance with the Scottish Index of Multiple Deprivation ranking system. 32 During the physical examination, blood pressure readings were recorded (sitting). Hypertension was identified when mean systolic pressure was above 144 mmHg or when mean diastolic pressure was above 94 mmHg. Subjects receiving medication for the treatment of hypertension were included in the hypertensive category. Subjects were asked to disclose smoking habits into the following classifications: never smoked, previous smoker and current

Key points
• This cohort study investigated the association between serum ADMA levels and change in cognitive performance in late-middle age (from 63 to 67 years) for the first time using a prospective cohort study de/sign.
• Circulating ADMA levels displayed strong, linear and inverse associations with cognitive function after 4 years.
Therefore, blood-based ADMA has potential to aid the early detection of cognitive decline in late-middle age and must be explored using larger studies.
• These findings suggest a potential role of nitric oxide impairment in the pathophysiology underlying cognitive impairment and possible progression to AD, with implications for presymptomatic diagnosis or novel therapy.
smoker. As an additional assessment, hippocampal volumes were analyzed through MRI imaging assessment at WIII. The MR acquisition procedure has been described in greater depth elsewhere. 33  respectively. The test is comprised of five sets of 12 questions, which become increasingly difficult, hence greater cognitive ability is associated with a higher score. 35 For the purposes of this study, scores from WI and WIII were used to measure cognitive trajectories, with the difference between the two scores being the primary outcome measure. For the primary analyses, subjects were grouped into three categories based on change in cognition from WI to WIII.

| Childhood intelligence
Participants were categorised as rapid decliners if their RPM scores had decreased by 7 points or more from baseline assessment (WI).
Participants were allocated into the slow decline category if WIII RPM scores had decreased anywhere between 0 and 6 points below baseline scores. The maintainers and improvers had either maintained cognitive function or improved their RPM score between WI and WII assessments.

| Plasma ADMA levels
Blood samples were taken at baseline (WI) in 2000 and measurement of serum lipids undertaken. A random subset of blood samples were centrifuged at 4 C for 10 minutes at 2500 rpm to obtain plasma which was then frozen at −80 C. Samples were thawed at a later date and plasma ADMA concentrations were measured using hydrophilicinteraction liquid chromatography-electrospray tandem mass spectrometry as described previously. 36

| Statistical analysis
To investigate population sample characteristics, chi-square tests and tests for linearity were employed to compare the means between each subgroup of cognitive trajectory profile and selected characteristics at baseline. The corresponding significance value for each test was presented in the results.
The adjusted mean values (and corresponding 95% confidence intervals) of baseline ADMA were estimated for each category of change in RPM from baseline to WIII assessment using a direct standardization technique. The adjusted association between plasma Baseline characteristics of study population, by cognitive performance category

| RESULTS
A total of 93 participants with complete information at baseline and all subsequent assessments contributed to the main analyses ( Categories of cognitive performance correspond to the following differences in RPM score between baseline and follow-up assessments: rapid decliners = RPM score decrease by ≥7 points; slow decliners = RPM score decrease between 0 and 6 points; maintainers/improvers = RPM score remained unchanged or increased across follow-up. ADMA = asymmetric dimethylarginine; BMI = body-mass index; DBP = diastolic blood pressure; HDL = high density lipoprotein; LDL = low density lipoprotein; SBP = systolic blood pressure. *P < .05 for linear trend.

F I G U R E 1 Association between baseline ADMA across categories of cognitive function scores between baseline and 4-year repeat assessment.
Mean values (95% confidence intervals) were adjusted for gender, smoking, social deprivation, education and hippocampal volume. Areas of point estimates are inversely proportional to the variance of the coefficient. Participants were categorised as maintainers or improvers if they did not achieve a lower score at wave III (WIII) compared with wave I (WI). Similarly, participants were categorised as slow decliners if they achieved between 1 and 7 points below their baseline levels, and rapid decliners if performance at WIII was greater than 7 points below the baseline RPM score. The y-axis is standardised to represent ±0.5 SDs from the mean value of ADMA RPM performance categories before adjustment for potential confounders.
In confounder adjusted analysis, ADMA concentrations were negatively and linearly associated with an improvement in cognitive performance approximately four years after baseline (Figure 1). On average, one SD (roughly equivalent to 0.06 μmol/L) increase in ADMA at baseline was associated with a reduction in cognitive performance score of 1.26 (95% CI 0.14-2.26) points from baseline to WIII assessment.

| DISCUSSION
After full adjustment for potential confounders, higher plasma ADMA

| Underlying mechanisms
The exact cause of the association between plasma ADMA concentrations and cognitive function is unknown. However, a number of mechanisms have been suggested previously. 38  Various cohort studies have confirmed associations between higher ADMA concentrations and a number of chronic diseases, including type 2 diabetes, cardiovascular disease and stroke. [45][46][47] In addition, a recent review has also highlighted the potential involvement of ADMA in the pathogenesis of other chronic diseases beyond atherosclerotic diseases, including COPD and depression. 48 Together with a growing ageing population, the increasing prevalence of chronic disease risk factors is leading to rapid increases in multimorbidity and polypharmacy. Adverse drug reactions (ADRs) are themselves a major cause of morbidity, accounting for an estimated 6.5% of unplanned hospital admissions. 49

| Strengths and limitations
This study presents with a number of strengths. First, the prospective study design enabled us to study the association between ADMA concentrations and change in cognition for the first time, which has not been conducted to date. In addition, as discussed previously, childhood intelligence is one of the strongest predictors of cognition in old age, 26 Therefore, additional research is required to investigate whether peripheral ADMA concentrations adequately reflect the concentrations of this metabolite in the central nervous system.
A common source of bias in longitudinal studies of cognitive ageing is the retention of volunteer participants, as participants who drop out tend to score lower during initial cognitive tests, a group who are at greater risk of dementia. 50 With this knowledge, practitioners may have inflicted bias into the present study through inviting rapid decliners back for follow-up assessments earlier than nondecliners.
Moreover, refusal to participate was significantly associated with low childhood mental ability (MHT) scores (P < .05), reported as a source of bias elsewhere. 54 Furthermore, although we adjusted for multiple important factors, it is not possible to rule out the effects of residual confounding.
Importantly, no measure of renal function was included in the analyses. ADMA is removed from the body through a combination of intracellular metabolism and renal excretion, and hence renal function is an important confounding variable, which would explain some of the variance in plasma ADMA levels. However, this may have been less important in this present study since renal function is generally not altered in healthy individuals at age 63 years. Of these limitations, some were unavoidable consequences of observational research.

| Conclusion
In summary, the results from this study suggest that ADMA concentrations are associated with cognitive decline in older age. Therefore, plasma ADMA, or correlates thereof, may be involved in the pathogenesis underlying cognitive decline and progression to AD. However, larger studies investigating this association are warranted. Possible future implications include the discovery of novel drug targets or presymptomatic biomarkers for AD.