Plasma Urate and Parkinson's Disease in Women

Abstract

Plasma urate has been consistently associated with a lower risk of Parkinson's disease in men, but it is less clear if this relation exists in women. Between 1990 and 2004, the authors conducted a nested case-control study among participants of the female-only Nurses’ Health Study. In controls (n = 504), plasma urate was positively associated with age, body mass index, alcohol consumption, hypertension, and use of diuretics and was inversely associated with physical activity and postmenopausal hormone use, as expected. Mean urate levels were 5.04 mg/dL for cases (n = 101) and 4.86 mg/dL for controls (P = 0.17). The age-, smoking-, and caffeine-adjusted rate ratio comparing women in the highest (≥5.8 mg/dL) with those in the lowest (<4.0 mg/dL) quartile was 1.33 (95% confidence interval: 0.69, 2.57; P_trend = 0.4). Further adjustment for body mass index, physical activity, history of hypertension, and postmenopausal hormone use did not change the results. Unlike in men, these findings do not support the hypothesis that urate is strongly associated with lower rates of Parkinson's disease among women.

The cause of neurodegeneration in Parkinson's disease remains unclear, but oxidative stress has been implicated (1). Urate, a potent antioxidant in humans that scavenges peroxynitrites and hydroxyl radicals, may be protective. Two prospective cohorts of men found that those with higher plasma urate concentrations had lower rates of Parkinson's disease (2, 3), while a third study, including 68 men and women, found a similar association but did not report gender-specific results (4). The latest prospective community-based study found similar results in men (cases: n = 57) and in women (cases: n = 38), but the association among women was not statistically significant (5). A history of gout, a condition characterized by hyperuricemia, was associated with Parkinson's disease in both men and women in one study (6) but in men only in another (7). We therefore sought to examine the relation between plasma urate and Parkinson's disease in the female-only Nurses’ Health Study.

MATERIALS AND METHODS

In 1976, the Nurses’ Health Study enrolled 121,700 female nurses (30–55 years) who returned a mailed questionnaire on lifestyle and medical history (8). Participants return follow-up questionnaires biennially to record newly diagnosed illnesses and to update lifestyle information. During 1989–1990, blood samples were collected from 32,826 participants.

Lifetime diagnosis of Parkinson's disease was self-reported in 1994 and biennially thereafter. Confirmation of cases was described elsewhere (9). Briefly, cases were confirmed when certainty of diagnosis as reported by treating neurologists or internists via questionnaire was definite or probable or when medical record reviews by our movement disorder specialist (M. A. S.) indicated either a final diagnosis of Parkinson's disease by a neurologist or evidence of at least 2 cardinal signs of Parkinson's disease (resting tremor, rigidity, bradykinesia) without features suggesting other diagnoses. Treating neurologists confirmed 88% of cases, treating internists confirmed 7.4%, and the remainder was confirmed by medical record review. Participants were followed for incident Parkinson's disease from blood draw until they returned the 2004 questionnaire or died.

A colorimetric enzyme assay (Hitachi 911; Roche Diagnostics, Indianapolis, Indiana) determined urate concentrations (coefficient of variation = 3.95%).

From 4 to 6 randomly selected controls who were alive and had not reported Parkinson's disease at the time of diagnosis were matched to cases on birth year (±1 year), race (white vs. other), fasting status (>8 hours vs. less/unknown), and year, month, and time of blood draw (2-hour intervals).

Mean plasma urate levels across groups were compared by using random-effects models that account for correlation within matched sets. Conditional logistic regression estimated odds ratios and 95% confidence intervals; with this matched design study, odds ratios estimate incidence rate ratios. Quartiles were based on the distribution among controls, and tests for trend modeled urate as a continuous variable. Given the sample size, the power to detect an odds ratio of 0.5 comparing the top with the bottom quartile was estimated to be >80%. Confounders were extracted from the questionnaire preceding blood draw. Effect modification was explored through separate analyses across levels of potential modifiers, and formal tests were conducted by modeling the product of urate and the modifier of interest. Analyses were conducted with SAS, version 9, software (SAS Institute, Inc., Cary, North Carolina).

RESULTS

During follow-up, 105 confirmed cases were identified and matched with 518 controls. Four (3.8%) cases and 14 (2.9%) controls were excluded because of gross hemolysis of the sample. The average age at onset was 68 years (range: 49–80 years), which on average was 7.1 years after blood draw (range: <1–14 years). The mean urate level was 5.04 mg/dL for cases and 4.86 mg/dL for controls (P = 0.17). As expected, urate in controls was positively associated with age, body mass index, alcohol consumption, hypertension, and use of diuretics and inversely associated with physical activity and postmenopausal hormone use (Table 1). The age-, smoking-, and caffeine-adjusted rate ratio of Parkinson's disease comparing the highest (≥5.8 mg/dL) with the lowest (<4 mg/dL) quartile of urate was 1.33 (95% confidence interval (CI): 0.69, 2.57) (Table 2). Adjustment for body mass index, alcohol, physical activity, hypertension, and postmenopausal hormone use did not materially change the results (Table 2). In an analysis restricted to cases with onset at least 7.1 years (median follow-up) after blood draw, the rate ratio was 0.73 (95% CI: 0.52, 1.03; P = 0.07) for a 1-mg/dL increase in urate.

Table 1.

Age-adjusted Characteristics of Controls According to Quartiles of Urate Among US Women in 1990^a

	Urate Concentrationb
	Quartile 1 (<4.0 mg/dL)	Quartile 2 (4.0–<4.8 mg/dL)	Quartile 3 (4.8–<5.8 mg/dL)	Quartile 4 (5.8–10.3 mg/dL)
No.	134	118	128	124
Age at blood collection in 1988–1990, years	59.9	60.2	61.2	62.0
Body mass index, kg/m2	23.5	24.2	26.4	28.3
Pack-years of smoking	10.9	10.2	15.7	12.5
Caffeine intake, mg/day	258	286	314	268
Total alcohol intake, g/day	8.8	8.3	10.2	12.0
Dairy, servings/day	2.0	1.8	2.1	2.1
Meat as main dish, servings/day	1.00	0.76	0.82	0.73
Fish, servings/day	0.27	0.20	0.22	0.22
Physical activity, METs/week	16.2	16.6	15.3	14.2
Postmenopausal, %	92.2	91.8	90.2	95.4
Postmenopausal hormone users, %	42.0	44.2	34.8	27.0
Thiazide users, %	11.9	10.1	16.0	27.6
Other diuretics, %	10.8	11.7	11.3	22.8
Aspirin users (>1 time/week), %	27.5	23.6	27.3	31.5
Other NSAIDs (>1 time/week), %	21.4	19.8	20.5	16.3
History of high blood pressure, %	25.4	21.4	34.3	51.1
History of self-reported gout, %	0.8	0	2.6	2.2
Ferritin, ng/mL	75	92	100	134
Total cholesterol, mg/dL	221	232	232	242
HDL cholesterol, mg/dL	67	64	61	53
LDL cholesterol, mg/dL	129	139	137	145

Abbreviations: HDL, high density lipoprotein; LDL, low density lipoprotein; MET, metabolic equivalent; NSAID, nonsteroidal antiinflammatory drug; SI, Système International [d'Unités] (International System [of Units]).

^aAll variables except age are age adjusted by direct standardization to all controls.

^bSI (metric) equivalents by quartile: <232, 232–<280, 280–<340, and 340–613 μmol/L.

Table 2.

Association Between Urate in Quartiles and Rates of Parkinson's Disease in US Women, 1990–2004

Urate	No. of Controls	No. of Cases	Adjusted for Age, Smoking, and Caffeinea			Multivariable Adjustedb
			Rate Ratio	95% CI	P Value	Rate Ratio	95% CI	P Value
Quartile 1	134	21	Referent			Referent
Quartile 2	118	35	1.90	1.02, 3.52		1.69	0.89, 3.20
Quartile 3	128	19	1.01	0.51, 2.00		0.91	0.44, 1.89
Quartile 4	124	26	1.33	0.69, 2.57		1.14	0.55, 2.37
Trend			1.08	0.91, 1.27	0.4	1.07	0.88, 1.30	0.5

Abbreviations: CI, confidence interval; HDL, high density lipoprotein; LDL, low density lipoprotein.

^aResults were similar when adjusted for only age and matching factors.

^bAdjusted for age (years); pack-years of smoking; caffeine intake (by quartiles, mg/day), alcohol intake (0, 1–9, 10–19, ≥20 g/day), body mass index (<23, 23–<25, 25–<27, ≥27 kg/m²); physical activity (metabolic equivalents/week); postmenopausal hormone use (premenopausal, postmenopausal, nonuser; postmenopausal, user; unknown status); and history of hypertension in 1988 (yes/no). Further adjustment for dairy intake (by quartiles, servings/day); regular use of aspirin (>1 per week vs. less); other nonsteroidal antiinflammatory drug use (>1 per week vs. less); use of thiazide or other diuretics (yes/no); and plasma levels of ferritin, total cholesterol, and LDL and HDL cholesterol (by quartiles) did not materially change the results.

Over 90% of cases and 91% of controls were postmenopausal at blood draw. The mean urate level was 0.42 mg/dL higher among controls who had never used postmenopausal hormones (40%) compared with ever-users (P = 0.001). In the multivariate-adjusted model, the rate ratio of Parkinson's disease for a 1-mg/dL increase in urate was 0.61 (95% CI: 0.31, 1.17; P = 0.13) among never users of postmenopausal hormones and was 1.00 (95% CI: 0.70, 1.41; P = 0.9) among ever users. Age, smoking, and caffeine did not modify the relation between urate and Parkinson's disease.

DISCUSSION

In this nested case-control study based on a large cohort of female nurses, plasma urate levels were not significantly associated with Parkinson's disease risk, except for a modest and only marginally significant inverse association for Parkinson's disease risk 7 or more years after blood draw. These results are in contrast with previous observations in men where a high plasma urate level strongly and consistently predicted a lower risk of Parkinson's disease (2, 3, 5). The strengths of this study include its prospective design and large sample size. Although the diagnosis of Parkinson's disease was mostly confirmed by treating neurologists rather than by direct examination, a large clinico-pathologic study confirmed 90% of neurologist-diagnosed Parkinson's disease cases at autopsy (10); thus, the proportion of misdiagnosed cases is probably too small to explain the null findings. Further, the strong inverse association with smoking in this cohort is similar to that reported by others, suggesting accuracy of Parkinson's disease diagnoses (11). Measuring plasma urate once may not capture long-term exposure; although the strong inverse relation between urate and Parkinson's disease in men was also based on a single measurement, levels in women may vary more because of menopause and postmenopausal hormone use.

The findings reported here contrast with those of several prospective studies, particularly in men (2–5). In the Honolulu Heart Program, 7,968 men of Japanese or Okinawan ancestry were followed for 30 years. Those with baseline plasma urate levels above the median had a 40% (95% CI: 0, 60) lower rate of Parkinson's disease than those with levels below the median (2). In a nested case-control study of predominantly white US health professionals, men in the highest quartile of plasma urate had a 57% (95% CI: −2, 82; P_trend = 0.017) lower rate of Parkinson's disease than men in the lowest quartile (3). Similar to our findings, the inverse association was strengthened in analyses restricted to cases with onset after the median (4 years) interval between blood draw and onset. A prospective population-based cohort in Rotterdam found a 29% (95% CI: 2, 49) reduction in Parkinson's disease with a 1-standard deviation increase in plasma urate (4). The authors reported that the results did not differ by gender, but they had only 68 cases in total. Recently, a prospective, US community-based study found that men with higher levels of urate had lower odds of Parkinson's disease (top vs. bottom quintile: odds ratio = 0.3, 95% CI: 0.1, 0.7; P_trend = 0.02); an inverse association suggested in women (38 cases) was not significant (odds ratio = 0.5, 95% CI: 0.2, 1.7; P_trend = 0.4) (5).

Indirect evidence suggests that the urate–Parkinson's disease association may be gender specific. A case-control study nested in the General Practice Research Database comprising 3 million Britons (1,052 incident cases) found 40% (95% CI: 8, 60) lower rates of Parkinson's disease in men with a history of gout compared with men without gout, whereas no association was found in women (7). However, a study using the British Columbia Linked Health and PharmaCare databases found a 30% (95% CI: 17, 41) reduced risk of Parkinson's disease in both men and women with a history of gout (6). A review of dairy and milk consumptions, which both decrease urate, found an association with Parkinson's disease in men (pooled rate ratio = 1.8, 95% CI: 1.4, 2.4) but not in women (pooled rate ratio = 1.3, 95% CI: 0.8, 2.1), comparing study-specific highest and lowest categories (12). In 2 studies of Parkinson's disease progression, men with higher urate levels at baseline had lower rates of reaching disability sufficient to warrant dopaminergic therapy; in women, the association was weaker and not statistically significant (13, 14).

The inverse association in men may result from urate's antioxidant properties. Oxidative stress has been implicated in the pathogenesis of Parkinson's disease, and evidence of oxidative damage, such as DNA and protein damage and lipid peroxidation, has been found at autopsy of Parkinson's disease brains (1, 15). Dopamine metabolism can cause excess reactive oxygen species and hydrogen peroxide. Urate exhibits antioxidant properties such as scavenging reactive oxygen species and peroxynitrites, has ameliorated human dopaminergic cell damage in culture following exposure to pesticides or iron ions (16), and can reduce the death of dopaminergic neurons in primary midbrain culture (17). Alternatively, urate could be a marker for another protective molecule or simply a marker of oxidative stress and thus not causally related to Parkinson's disease.

On average, plasma urate levels are lower in women than men. The mean level among controls was 4.9 mg/dL in this study compared with 6.1 mg/dL in our study in men (3), a discrepancy not fully explained by age differences. Our findings that urate levels are higher in postmenopausal than premenopausal women and in never users of postmenopausal hormones than in ever users were very similar to those from a cross-sectional survey of US women (18). There was a suggestion that postmenopausal hormone use modified the association between urate and Parkinson's disease, where an inverse association was seen in women who were never users. However, it was not statistically significant and could be a chance finding. A gender difference has also been reported for the caffeine–Parkinson's disease association; men who regularly consume caffeinated drinks have lower rates of Parkinson's disease, whereas a similar relation was found only in women who never used postmenopausal hormones (9, 19, 20).

The question remains whether pharmacologic or dietary manipulation of plasma urate or its precursors will slow disease progression in Parkinson's disease patients. An association between raised plasma urate levels and cardiovascular disease morbidity and mortality should not preclude investigation of its potential therapeutic effect among patients, particularly men, diagnosed with Parkinson's disease, as the potential for slower disease progression may outweigh the risk of harm.

Glossary

Abbreviation

CI

confidence interval