Uric Acid and High-Density Lipoprotein Cholesterol Are Differently Associated with Alzheimer’s Disease and Vascular Dementia

INTRODUCTION

There is substantial evidence for the involvement of circulating uric acid (UA) in neurodegenerative processes. As a product of purine metabolism, UA accounts for a large proportion of naturally-occurring antioxidant activity [1] and thereby is thought to have a neuroprotective action both in neurodegenerative disorders and their experimental models [2]. This neuroprotective action appears to extend to a relationship with relatively improved outcome in stroke patients [3], which we have recently confirmed in patients in a general hospital neurology department [4].

However, these findings contrast strongly with the observation that UA is a risk factor for cerebrovascular disease [5] as it is for other vascular diseases. There are reported associations with atherosclerosis [6] as UA can increase endothelial cell activation and platelet adhesion [7].

The association of UA with dementia is unclear. A recent study of the emergence of dementia in a large elderly cohort showed an association with higher baseline UA [8], also significant in the small vascular dementia (VaD) subgroup, and a non-significant higher UA associated with the Alzheimer’s disease (AD) subgroup. A large consortium study did not identify genetic factors contributing to UA concentrations as a risk factor for AD however [9]. This is in contrast with the observation in the Taiwanese population that a diagnosis of gout, a consequence of elevated UA, is associated with lower incidences of both vascular and non-vascular dementia [10]. A meta-analysis of 21 studies with AD is consistent with this in showing an overall reduction in UA compared to healthy controls, although there were substantial inconsistencies between the results of the individual studies [11].

Contrasting with results from the incidence study of Latourte et al. [8], VaD is reported to exhibit reduced serum UA in a Chinese cohort [12]. A meta-analysis has indicated UA to be lower in AD, but not consistently associated with VaD [13]. Thus, a direct comparison between AD and VaD cohorts from the same population might throw further light on the relationship between different dementias and circulating UA.

There is also increasing evidence implicating disturbances of lipid metabolism in dementing disorders. Elevated cholesterol is a well-established risk factor for cerebrovascular disease including stroke and VaD as it is for cardiovascular disease [14]; the same is reportedly found for AD [15]. Cholesterol has also been implicated in the neurodegenerative mechanisms underlying AD [16]; apolipoprotein E, coded for by the AD risk gene APOE, is a cholesterol transporter [16], the aberrant metabolism of amyloid protein occurs in a lipid environment influenced by cholesterol [17, 18], and dietary fat appears to influence risk for AD [19]. Thus, cholesterol as a further risk factor may modify the association between UA and dementia.

We have undertaken an investigation in a general hospital setting to assess the relationship of UA with dementia, to compare subjects with VaD and AD, and to determine whether there are also associations or inter-relationships with other metabolic measures including blood lipids.

METHODS

The study cohort was drawn from 225 subjects admitted to the Puer People’s Hospital, Department of Neurology for the first time for investigation associated with symptoms of dementia. All patients included underwent neuroimaging, the great majority magnetic resonance imaging (MRI), following which diagnosis was made according to imaging findings, symptoms, and clinical history. Diagnosis of AD was performed according to the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) criteria [20], while diagnosis of VaD was performed according to the VASCOG (Vascular Behavioral and Cognitive Disorders) criteria [21].

Exclusion criteria were 1) treatment for other physical disorders, including gout and elevated homocysteine, 2) epilepsy, 3) kidney disease, liver disease, or multiple organ failure, 4) cancer, or 5) substance abuse, prior severe mental illness or psychoactive drug treatment. Subjects with dementia associated with non-vascular causes other than AD, such as Parkinson’s or Lewy body diseases, were also excluded from the study. This yielded a dementia series of 187 subjects. Current pharmacotherapy was recorded; this included statins, non-steroidal anti-inflammatory drugs (NSAIDs; mainly aspirin), donepezil, and antidepressants. A cohort of 79 healthy controls within the age range of the dementia series was identified from subjects who had undergone routine medical assessment in the hospital for employment purposes. Following the dementia-control analysis, further analysis was undertaken with the VaD and AD subgroups; subjects with mixed dementia were excluded.

All subjects had undergone routine blood testing; fasting plasma samples were collected before 8am by venipuncture in heparin-lithium anticoagulant tubes. After centrifugation, plasma samples were aliquoted in polypropylene tubes for biochemical analyses using routine assays. Data for UA, total cholesterol, high- and low-density lipoprotein cholesterol (HDL and LDL), and triglycerides were collected; for the majority of subjects in the dementia series measures of HbA1c and BMI were obtained. Abnormal or elevated results were defined as follows: elevated UA > 420umol/L, elevated HbA1c>7%. The results were compared between groups and relationships between the various measures were investigated. Statistical analysis was performed using SPSS v21. Results for each measure were tested for normal distribution (Kolmogaroff-Smirnoff test); data deviating significantly from normality was log-transformed and the test repeated. T-tests, univariate analysis of variance followed by Dunnett’s test, chi-squared test and Pearson correlation were used to compare samples and determine correlations. The study was approved by the Hospital Research Ethics Committee; as an anonymized retrospective study, requirement for informed consent was waived.

RESULTS

Of the series of 187 selected subjects with dementia, 111 received a diagnosis of VaD, 60 of AD, and 16 mixed dementia. A significant difference in age between the dementia and control groups was apparent (Table 1); age was included as a covariate in further comparisons between these groups. Not all subjects had all measures recorded; sample sizes for each measure are recorded in Tables 1 and 2. The three cholesterol measures were normally distributed; UA and TG deviated significantly from a normal distribution. After log-transformation, these two variables were found to be normally distributed and these transformed data were used in subsequent analyses.

DISCUSSION

These findings indicate that in a naturalistic cohort of subjects attending a Chinese neurology department, a substantial and highly significant reduction in UA is seen in AD, while no abnormality of UA was found in the VaD group; a strong difference between the two dementia groups is apparent.

One interpretation of the UA findings is that the antioxidant properties of UA provides protection against the development of AD and thus a reduction in UA is a risk factor for AD. Certainly there is evidence for this from both experimental findings and from results of both prospective and cross-sectional studies although, as briefly reviewed in the introduction, there is some inconsistency between reports.

On the other hand, there appear to be opposing and conflicting effects of UA in cerebrovascular disease, with elevated UA proposed as a risk factor [5], although the functional outcome of ischemic stroke is improved in the absence of reduced UA [3, 4]. The finding that the effect of UA on stroke outcome is confounded in hyperglycemic patients [22] indicated that there might be an interaction between hyperglycemia and UA in the associations with dementia due to cerebrovascular causes, i.e., VaD. However, there was no influence of elevated HbA1c on the main findings of a reduction below control values of UA in AD but not in VaD. These findings indicate that, independent of hyperglycemia, reduced UA is not associated with VaD.

Substantial reductions in dementia were seen in the three cholesterol measures. However, a significant reduction only in total cholesterol was observed in the AD group, while a highly significant reduction in HDL was specific to patients with VaD.

That relative deficits were observed in all three cholesterol measures is notable—it appears that the reduction in potentially atherosclerogenic LDL may be offset by the reduction in protective HDL. These changes were independent of statin use; statins had no apparent effect on lipid measures. This may be related to the fact that subjects presenting initially with elevated cholesterol are more likely to receive statins, the use of which is greater in the VaD group, which would then tend to reduce total cholesterol and LDL, while they have little effect on HDL [23].

While total cholesterol was reduced in both AD and VaD groups, an effect independent of statin treatment, a substantial and significant reduction below control values in HDL was only seen for the VaD group. A simple interpretation of this suggests that the VaD subjects have a lower capacity for the transport and removal of cholesterol and thus are at greater risk of atherosclerotic damage. However, HDL has a variety of subtypes with different functions that are not fully understood; further investigation of such lipid fractions is needed to provide a better understanding of the role of HDL in dementing disorders. Analysis of a large series of elderly subjects has shown how HDL, rather than LDL or total cholesterol, is associated with general cognitive ability, and how reductions in four subfractions of HDL contribute the strongest metabolic associations with dementia [24]. This association of reduced HDL with dementia has also been observed in a Chinese AD cohort [25], along with independent effects of increased LDL and total cholesterol. Our sample does not demonstrate significant differences in AD other than a reduction in total cholesterol, and no significant difference in lipids between AD and VaD.

How peripheral cholesterol fractions reflect its status in the brain remains unclear. There seems to be little correlation between cholesterol measures in plasma and brain, although elevated plasma levels can increase oxidative metabolites of cholesterol which may then cross the blood-brain barrier to perturb cholesterol function in the brain [26]. Increases in these oxysterols in cerebrospinal fluid have been observed in AD [27].

The negative correlation between UA and HDL seen in the dementia group has been observed previously in hyperlipidemic subjects [28]. This is a relatively weak effect but indicates that, to a small extent, the negative effects of a reduction of UA on oxidative stress may be offset by the vascular protective influence of a relative increase in HDL. The role of inflammation in dementia, particularly AD [29], may also be of relevance and indicates an opposing relationship between the two measures. While HDL [30] has anti-inflammatory effects, there is evidence that under some circumstances UA can be pro-inflammatory [7]. The importance of this is unclear in the context of dementing disorders; the antioxidant effect of UA may also be protective against the increase in oxidative free radicals that are a consequence of inflammatory processes [31]. Furthermore, HDL may be reduced by inflammation, although the underlying mechanisms remain unclear [32]. It is suggested that the evidence for inflammation is stronger than that for lipid disturbances in the pathobiology of AD [33]. Our data indicates that lipid disturbances, while present in AD, are supplemented by UA deficits, while in VaD there are more substantial lipid abnormalities.

While we were unable to report a systematic assessment of cognitive function and stage of disease in the patients studied here, they were recently diagnosed and, for most subjects, are likely to be at an early stage of their clinical disease. Establishing APOE genotype would have provided potentially valuable information, given the importance of apolipoprotein E in cholesterol transport, although this was not available to the current study. While the patients in the studied subgroups met diagnostic criteria for either AD or VaD, it is clear that such strict diagnoses are often approximations and some patients may have elements of both disorders; postmortem examination or more elaborate imaging techniques not available to the present study would be needed to definitively establish accurate diagnosis. It should be noted that the proportions of subjects in the dementia groups studied reflect the relatively higher incidence of VaD in a general hospital neurology department and not the prevalence of VaD and AD in the population. The control group provided a useful comparator, although data additional to UA and lipids was not available. However, it remains possible that some of the apparent differences between dementia and control subjects may be non-specific, and a further medical control group might be more appropriate. The study is limited in its scope as few clinical variables were available for collection; it is likely there are other factors, such as smoking and alcohol use, that may influence disease risk or modulate the effect of UA. Nevertheless, it is has been reported in a large cohort that smoking has no influence on UA or its correlations with lipid measures [25]. Sample sizes were not large, particularly of the AD group, although they compare favorably with those in several other reports. A further limitation is the retrospective and essentially cross-sectional nature of the study; the relationships observed between dementia subtypes and biochemical measures are informative associations but it cannot be concluded without further longitudinal study that they are risk factors. However, the findings do add to previous studies suggesting reduced UA to be a risk factor in AD [11]. A clearer understanding of the role of cholesterol measures in dementia subgroups would have been obtained from a cohort of subjects not receiving statins; however, given the ubiquitous use of these drugs in controlling elevated cholesterol in elderly populations, results in drug-free subjects may have little value in application to clinical practice.

In conclusion, these results indicate a clear distinction between AD and VaD in their association with plasma UA, a major contributor to total antioxidant activity, in which only AD has significant reduced concentrations. They also demonstrate deficits in measures of cholesterol in dementia. It is notable that HDL is reduced below control values in VaD with no significant deficit in AD, although in this sample we could not demonstrate a significant difference between the two dementia groups, a limitation that a larger AD sample might address. These findings, obtained in the naturalistic setting of a general hospital neurology department, may have value in contributing to clinical assessment as well as further indicating the potential of improving antioxidant status in AD and those at risk of this disease [34, 35].

DISCLOSURE STATEMENT

Authors’ disclosures available online (https://www.j-alz.com/manuscript-disclosures/19-1111r1).