The Influence of Agricultural Work and Plasma Uric Acid on Hospital Admission for Alzheimer’s Disease

INTRODUCTION

The neurodegenerative disorders including Parkinson’s disease (PD) and Alzheimer’s disease (AD) contribute a major medical, social, and economic burden for health care of the elderly worldwide. Treatment for these disorders is essentially symptomatic; effective preventative strategies remain elusive despite decades of research seeking to identify pathogenic mechanisms. Etiological factors are varied but for both PD and AD include some relatively infrequent single genetic mutations, as well as other genetic and environmental risks.

One risk factor for neurodegenerative disease with strong supportive epidemiological evidence is exposure to environmental neurotoxins such as pesticides. Both known and presumed pesticide exposure have been associated with increased occurrence of PD [1]. AD too has, in some studies, been related to prior exposure to pesticide chemicals [2]. Such exposure is a hazard that may be associated with agricultural work, particularly in economically- and educationally-deprived communities; its neurotoxic effects are relatively frequent in farmers in Southern China [3].

A further factor associated with risk for neurodegenerative disease is uric acid (UA). UA is a product of purine metabolism present in the blood and it has well-established toxic effects in gout as well as contributing risk to, e.g., cardiovascular disease. However, it also contributes substantially to the antioxidant capacity of the body, for example by scavenging reactive oxygen species (ROS). As such it appears to have neuroprotective properties that may be important in neurological disease; reductions in circulating UA are found neurodegenerative disorders including PD [4] and AD [5], and it may also influence the outcome of ischemic stroke [6].

We have previously demonstrated that both undertaking agricultural work and having a relative reduction in blood UA concentrations are factors associated with hospital admission for PD in a semi-rural region of Southwest China [7]. We have also identified that lower UA is associated with a diagnosis of AD, but not of vascular dementia (VaD), in subjects attending the same hospital neurology department [8]. In the current study we have collected data from a larger series of subjects with dementia to determine whether agricultural workers are over-represented, as they are in PD, in those receiving a diagnosis of AD as opposed to VaD, in this population and how the result might relate to circulating concentrations of UA.

METHODS

The subjects were from a cohort of 452 patients first admitted to the Department of Neurology of Puer People’s Hospital with a diagnosis of dementia and included a series of 187 subjects previously studied [8]. The study was approved by the Hospital Research Ethics Committee; as an anonymized retrospective study, requirement for informed consent was waived.

All patients included underwent neuroimaging, primarily structural magnetic resonance imaging (MRI), following which diagnosis was made according to imaging findings, symptoms, and clinical history; for AD this met the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) criteria [9], while diagnosis of VaD met the Vascular Behavioral and Cognitive Disorders (VASCOG) criteria [10]. This included cognitive assessment using the Mini-Mental State Examination; all neuroimaging was reviewed by one of us (HXL) to confirm diagnosis.

Exclusion criteria were: treatment for other physical disorders, including gout and elevated homocysteine; epilepsy; kidney disease, liver disease, or multiple organ failure; cancer; substance abuse or prior severe mental illness. Subjects with dementia associated with non-vascular causes other than AD, such as PD or Lewy body disease, were also excluded from the study. Information on occupation was collected in order to identify those with a background in farming or related agricultural work. Current pharmacotherapy was recorded in almost all subjects; this included statins, non-steroidal anti-inflammatory drugs (NSAIDs; mainly aspirin), antidepressants and donepezil, and data on treatment with each of these four drug classes was collected.

All subjects had undergone routine blood testing on admission; fasting plasma samples were collected before 8 am by venipuncture in heparin-lithium anticoagulant tubes. After centrifugation, biochemical analysis of plasma samples was undertaken from which UA data were collected.

Statistical analysis was performed using SPSS v21. Results for UA were tested for normal distribution (Kolmogorov-Smirnoff test); as they deviated significantly from normality all statistical testing of UA was undertaken with log-transformed data.

RESULTS

From the series of 452 subjects with a dementia diagnosis, two subgroups of patients with clinically confirmed AD or VaD were studied, excluding those with a mixed VaD and AD diagnosis (Table 1). We observed no significant association between diagnosis and agricultural work, indicating that agricultural workers in this sample did not demonstrate a relatively increased prevalence of AD (χ² = 0.186, p = 0.9). Hence there was no significant difference between the proportions of agricultural workers in the AD and the VaD groups (40% and 42% respectively; Table 1). This contrasts with our previous finding in PD [7]; direct comparison with that previously reported PD group identifies a significantly greater proportion of subjects admitted with PD than with AD had a history of agricultural work (67% and 40% respectively; χ² = 22.42, p < 0.0001).

There was a notable difference in age between agricultural and non-agricultural workers, with the former in each diagnostic group presenting at a younger age. However, the lack of a significant difference in the proportion of agricultural workers in the two dementia diagnostic groups was also apparent for the subjects in the younger (<73 years) subgroup (χ² = 3.023, p = 0.22); the same was true for the older subgroup (χ² = 1.100, p = 0.58).

Of the sample studied above, 406 subjects had data from routine UA analysis, of which 121 and 166 had confirmed diagnoses of AD and VaD respectively. UA was significantly associated with sex (t = 5.99; p < 0.001), but not age (r = – 0.085; p = 0.165); sex was included as a cofactor in further analyses. We could confirm, in this larger series, the association of reduced UA with AD, compared with VaD (Table 1), that we previously observed in this population [8]. In the healthy control sample of 79 subjects (age 62.5±10.3 years) from that earlier study, the UA concentrations (348±86μmol/L) were very similar to those of the current VaD group and hence are substantially greater than the AD sample.

No significant difference in pharmacotherapy was observed between agricultural workers and those who were not (p > 0.2 in each case), although there were differences associated with diagnosis in which statins and NSAIDs were prescribed more frequently in the VaD group and donepezil more frequently in AD (Table 2). After inclusion of these drug treatments as cofactors, the significant difference in UA between diagnostic groups remained. However, for NSAID treatment alone, a significant effect of drug on UA concentrations was apparent in the total sample independent of diagnosis (F = 4.672; p = 0.032) in which subjects receiving NSAIDs had higher plasma UA concentrations, although this did not reach statistical significance in the individual diagnostic subgroups.

DISCUSSION

These findings demonstrate the absence of a detectable relationship between a history of agricultural work and hospital admission for AD in the studied population, in strong contrast to our previous finding of a substantially higher frequency of such work in patients admitted with PD from the same semi-rural catchment area [7].

This result emphasizes how an association with agricultural work, as a likely proxy for exposure to pesticides in this relatively deprived and often poorly-educated population, is specific to PD in this population and does not apparently generalize to AD, another primarily sporadic, age-related, neurodegenerative proteinopathy. Our findings contrast with some other reports suggesting links between pesticide exposure and AD, although the evidence that this makes a substantial contribution to the incidence of AD is weak, compared to the strong association between indicators of pesticide exposure and the development of PD [1]. For example, a study looking at neurological disease in elderly Costa Rican subjects identified a substantially increased prevalence of PD, but not of AD, in subjects exposed to pesticides [11].

Why then are there differential effects of agricultural work, a likely indication of pesticide exposure in this population [3], on AD and PD? Our findings would be consistent with the suggestion that the neurotoxic effects of pesticide exposure are relatively specific in producing the pathology of PD, i.e., for a toxic effect on dopaminergic neurons not, in this population, generalizing to effects on other neuronal systems such as the cholinergic neurons that are particularly, but not solely, implicated in the degenerative process of AD [12]. Certainly, the evidence supports such a relative selectivity for dopaminergic neuronal degradation, rather than the neuronal systems most affected in AD, in the neurotoxic action of several pesticides [13].

It was notable that in both diagnostic groups the agricultural workers presented with dementia at a younger age without an effect on overall relative disease incidence. This could reflect rural work having an accelerating effect on cognitive deterioration without increasing lifetime dementia risk. Such a mechanism could involve pesticide exposure enhancing the consequences of other risk factors through, e.g., an increase in ROS, although other factors associated with farming and a rural environment could contribute, such as differences in diet, level of education or other developmental or lifestyle factors which may increase the risk of subsequent dementia.

It is hypothesized that exposure to pesticides associated with agricultural work could result in neuronal damage through oxidative stress mechanisms [14], although other alternative or additional mechanisms may be involved in their reported association with AD [15]. UA, as a systemic antioxidant, is thought to have protective effects against neurodegeneration by scavenging ROS, independent of their origins or sites of action. This possible protective mechanism appears to be reduced in the AD group: there is a reduction in circulating UA concentrations in AD compared with VaD. These findings confirm, in a larger cohort, our previous finding [8] and is also consistent with several, albeit not all, other reports indicating a relative reduction of blood UA to be associated with AD [16]. While agricultural work was associated with a 12% reduction in UA in the AD group, this did not reach statistical significance; the difference between agricultural and non-agricultural workers was minimal in the VaD group.

The relationship between NSAID treatment and UA observed here is an interesting one, although the effect was small, not being significant in the individual diagnostic subgroups. Others have observed that aspirin treatment of elderly subjects can result in increased circulating UA [17], thought to be brought about by a nephrotoxic effect resulting in reduced glomerular secretion of UA [18], and potentially a contraindication for prolonged aspirin use in the elderly. However, it is conceivable that an aspirin-induced elevation in UA might, by thereby increasing systemic antioxidant activity, provide a novel explanation contributing to the admittedly controversial observation, seen in cohort studies [19] but not in randomized controlled trials [20], that this treatment might have some protective effect against neurodegeneration leading to AD.

This study, as a retrospective cross-sectional investigation in a specific population, is limited in several respects, not least in sample selection relying on hospital admission. However, this is likely to include the great majority of disease cases in the general population served by the hospital. While there might be a difference in attendance rates between people from rural and urban areas, it is unlikely that any such difference would differentially affect AD and VaD patients. Data on other potentially confounding factors, such as alcohol use and tobacco smoking, have not been collected, although it has been previously reported in a large cohort that smoking has no influence on UA [21]. Long-term data on disease course would also be informative, but beyond the scope of the current retrospective study of hospital records. Similarly, this study design did not include direct measurement of exposure to potentially neurotoxic pesticides in the subjects; agricultural work is inevitably an inaccurate proxy for neurotoxic exposure. We are aware that such exposure, and its liability to have neuropathological consequences, may well vary between agricultural workers in different regions and of different ethnicity depending on environmental and genetic risk factors, so our results cannot necessarily be extrapolated to other populations. However, the difference in the proportion of admissions of PD and AD cases from agricultural workers in this population strongly indicates that the effect of a potentially important environmental risk factor for PD does not extend to AD, another neurodegenerative proteinopathy with a different, and more generalized, neuronal involvement.