Gout and the risk of parkinson's disease: A cohort study
Several studies have suggested that higher serum uric acid levels lead to a lower risk of Parkinson's disease (PD) because uric acid exerts antioxidant effects on neurons. Our objective was to examine the relationship between gout and the risk of PD in persons age ≥65 years.
We conducted a population-based cohort study using the British Columbia Linked Health Database and PharmaCare data (i.e., prescription drug data for those age ≥65 years). We compared incidence rates of PD between 11,258 gout patients and 56,199 controls matched on age, sex, date of gout diagnosis, and length of medical record. Cox proportional hazards models were used to estimate the relative risk (RR) of PD, adjusting for age, sex, prior comorbid conditions, and use of diuretics and nonsteroidal antiinflammatory drugs.
Over an 8-year median followup, we identified 1,182 new cases of PD. Compared with individuals without gout, the multivariate RR of PD among those with gout was 0.70 (95% confidence interval [95% CI] 0.59–0.83). In subgroup analyses, the inverse association was similarly present in both sexes and was evident among those who did not use diuretics (RR 0.66, 95% CI 0.54–0.81), but not among diuretic users (RR 0.80, 95% CI 0.58–1.10, P for interaction 0.35).
Our population-based data provide evidence for a protective effect of gout on the risk of PD and support the purported protective role of uric acid.
Hyperuricemia develops in humans due to the absence of uricase, which is the result of mutations that occurred during early hominoid evolution (1). Hyperuricemia can be detrimental to humans, as demonstrated by its proven pathogenetic roles in gout, a common inflammatory arthritis, and by its putative roles in hypertension, cardiovascular disorders, and certain renal conditions, including nephrolithiasis (2). The evolutionary advantage of relative hyperuricemia in humans remains largely unknown, although the antioxidant property of uric acid may account for the greater longevity of humans compared with most other primates (3). Uric acid can scavenge superoxide, hydroxyl radical, and singlet oxygen, as well as chelate transition metals. In particular, in vitro and in vivo studies have shown that uric acid exerts antioxidant effects on neurons (4, 5).
With these antioxidant properties, uric acid has been hypothesized to protect against oxidative stress, a prominent contributor to dopaminergic neuron degeneration in Parkinson's disease (PD). Indeed, 3 prospective cohorts (6–8) and 2 retrospective case–control studies (9, 10) have evaluated a potential association between serum uric acid and the risk of developing PD, finding in all cases a lower risk of PD among individuals with higher levels of serum uric acid. Correspondingly, a recent case–control study based on the UK General Practice Research Database (GPRD) showed that a history of gout was associated with a 31% lower risk of developing PD and that the risk reduction was limited to men (11). If confirmed true, these data would indicate that there may be beneficial health effects associated with hyperuricemia, which may have therapeutic implications for treating hyperuricemia and gout. As more potent urate-lowering agents are in development, this potential adverse neurologic issue associated with lower uric acid levels may become important in setting appropriate target serum uric acid levels in patients with gout (12–14).
Our objective was to examine the link between gout and PD in persons age ≥65 years in British Columbia (BC), Canada. We investigated this age group because BC PharmaCare data provided prescription and dispensing records limited to this group, and the recent GPRD study found that the inverse link between gout and PD was only seen in individuals age ≥60 years.
MATERIALS AND METHODS
The BC Linked Health Database (BCLHD) is an extensive data resource for applied heath services and population health research covering the entire population of the province of BC (4.3 million residents in 2005) (15). The database contains an integrated longitudinal file of all persons registered with the health care system, including visits to health care professionals and laboratory tests ordered, hospital visits and interventions, as well as all prescriptions (through PharmaCare for residents age ≥65 years). From the BCLHD, we have created a musculoskeletal cohort that represents ∼3 million BC residents with any musculoskeletal symptoms or diagnoses at least once in their computerized BCLHD records between 1991 and 2004. The current cohort study population consisted of 11,258 individuals age ≥65 years with gout and 56,199 control individuals matched by age, sex, date of gout diagnosis (index date), and length of available medical record in the database from the BCLHD musculoskeletal cohort. Prevalent cases of gout recorded in the first year were excluded from the analysis (n = 6,100).
Individuals with gout were identified as those with 2 visits at least 1 day apart with the International Classification of Diseases, Ninth Revision (ICD-9) code of 274 between March 31, 1992 and April 1, 1997. The positive predictive values (PPVs) of ICD-9 codes for gout have been reported to range from 61% in a US managed care setting (16) to 100% in a US Veterans Affairs rheumatology clinic setting (17). Similar to the GPRD study (11), we further divided gout patients according to treatment status, where treated patients consisted of those with a gout diagnosis and at least 1 prescription for an anti-gout medication (allopurinol, probenecid, colchicine, or sulfinpyrazone), and untreated patients consisted of those who did not receive any prescription for anti-gout medication during the study period. Seventy-two percent of gout patients received at least 1 prescription for an anti-gout medication, with allopurinol accounting for 87% of these prescriptions. The validity of gout diagnoses recorded in the GPRD was 90% when they were combined with the use of anti-gout medication (11, 18). The incidence rates of gout per 1,000 person-years in our cohort according to age categories 65–84 and ≥85 years were 5.69 and 7.53 cases among men, respectively, and 2.41 and 2.84 cases among women, respectively. These rates closely corresponded with incidence estimates reported in the GPRD (19).
Covariate exposure data included history of comorbid medical conditions (hypertension, diabetes, hyperlipidemia, and chronic obstructive pulmonary disease [COPD]), Charlson comorbidity score (20, 21), and medication use (diuretics including thiazides and furosemide, and nonsteroidal antiinflammatory drugs [NSAIDs]) at study baseline. These covariates were chosen based on their known associations with gout or with PD risk or known determinants (1, 8, 11, 22).
Our primary outcome was the first recorded event of diagnosis for PD using the ICD-9 code for PD, 332. This code was shown to have a PPV of 86% (95% confidence interval [95% CI] 83–88%) in a validation study using a population of nursing home residents in 5 US states (23). We added at least 2 prescriptions of anti-Parkinsonian medications to improve the specificity of our definition of PD. In the GPRD study, the PPV associated with this definition for PD was 90% (11). The incidence rates of PD per 1,000 person-years in our cohort according to age categories 65–69, 70–74, 75–79, and ≥80 years were 1.7, 2.8, 3.8, and 4.3 cases among men, respectively, and 1.3, 1.7, 2.1, and 2.5 cases among women, respectively. These rates were within the range of published incidence estimates (24). For example, these incidence rates tended to be lower than or similar to those reported in the Rotterdam (25), Italian (26), and Rochester (27) studies, but tended to be higher than those reported in the Taiwanese (28), Hawaiian (29), or Kaiser Permanente (30) cohorts.
We computed person-time of followup for each individual from the study baseline (first occurrence of gout visit or index date in control cohort) to the date of PD diagnosis, deregistration from the medical service plan, death from any cause, or end of the study period (March 31, 2004), whichever came first. We used Cox proportional hazards modeling (PROC PHREG; SAS Institute, Cary, NC) to estimate the relative risk (RR) for PD in all multivariate analyses. These models were adjusted for age, sex, history of comorbid medical conditions (hypertension, diabetes, hyperlipidemia, and COPD), Charlson comorbidity score, and prescription medication use (diuretics and NSAIDs). We conducted analyses stratified by sex (men versus women), age group (<75 versus ≥75 years), and diuretic use (user versus nonuser) to assess possible effect modification. We tested the significance of interactions with a likelihood ratio test by comparing a model with the main effects of history of gout and the stratifying variable and the interaction term with a reduced model with only the main effects. We conducted lag analyses by advancing the PD diagnosis date by 2 years to account for the preclinical phase of the disease (11). We also repeated the analyses to include prevalent cases of gout recorded in the first year. For all RRs, we calculated 95% CIs. All P values were 2-sided.
Over an 8-year median followup period, we identified a total of 1,182 new cases of PD. The baseline characteristics of the study population according to gout history are shown in Table 1. The mean age was 74.1 years and men comprised 66% of the study population. Individuals with gout showed a higher frequency of comorbid medical conditions as well as NSAID use.
Table 1. Baseline characteristics according to history of gout diagnosis*
|Age, mean ± SD years||74.1 ± 6.5||74.1 ± 6.5||1.000|
|Men||37,330 (66.4)||7,482 (66.5)||0.943|
|Comorbid medical conditions|| || || |
| History of hypertension||15,879 (28.3)||4,711 (41.9)||0.001|
| History of diabetes||4,413 (7.9)||1,173 (10.4)||0.001|
| History of hyperlipidemia||4,345 (7.7)||1,143 (10.2)||0.001|
| History of COPD||10,295 (18.3)||2,584 (23.0)||0.001|
| Charlson comorbidity score, mean ± SD||1.2 ± 1.8||1.4 ± 1.8||0.001|
|Prescription medication use|| || || |
| Diuretic use||9,152 (16.3)||3,521 (31.3)||0.001|
| NSAID use||37,676 (67.0)||10,666 (94.7)||0.001|
Compared with individuals without a history of gout, the crude RR of PD among those with a gout diagnosis was 0.75 (95% CI 0.63–0.89). After adjusting for age, sex, comorbid medical conditions, and use of diuretics and NSAIDs, the multivariate RR was 0.70 (95% CI 0.59–0.83). Individuals with gout who received anti-gout treatment had a lower multivariate RR (0.66, 95% CI 0.54–0.81) than those who did not receive any anti-gout medication (0.79, 95% CI 0.59–1.05) (Table 2).
Table 2. RRs of Parkinson's disease according to history of gout*
|No. of Parkinson's disease cases||1,026||156||108||48|
|Incidence per 1,000 person-years||2.47||1.85||1.76||2.09|
|Crude RR (95% CI)||1.0 (referent)||0.75 (0.63–0.89)||0.71 (0.59–0.87)||0.85 (0.63–1.13)|
|Multivariate RR (95% CI)†||1.0 (referent)||0.70 (0.59–0.83)||0.66 (0.54–0.81)||0.79 (0.59–1.05)|
In subgroup analyses, the association between a history of gout and a lower risk of PD similarly existed in both sexes (Table 3). Similar levels of inverse association were also observed in the 2 age groups, <75 and ≥75 years. Finally, the protective effect of gout on PD was significant among individuals who did not use diuretics (RR 0.66, 95% CI 0.54–0.81), but not among diuretic users (RR 0.80, 95% CI 0.58–1.10).
Table 3. Multivariate relative risks and 95% confidence intervals of Parkinson's disease (PD) stratified by sex, age, and diuretic use*
|No. of PD cases||310||872|| ||586||596|| ||207||975|| |
|No gout||1.0 (referent)||1.0 (referent)|| ||1.0 (referent)||1.0 (referent)|| ||1.0 (referent)||1.0 (referent)|| |
|All gout||0.67 (0.48–0.95)||0.71 (0.58–0.86)||0.773||0.66 (0.52–0.85)||0.74 (0.58–0.94)||0.651||0.80 (0.58–1.10)||0.66 (0.54–0.81)||0.349|
|Treated gout||0.71 (0.48–1.04)||0.65 (0.51–0.82)||0.751||0.61 (0.46–0.82)||0.72 (0.55–0.95)||0.500||0.75 (0.52–1.08)||0.63 (0.49–0.81)||0.470|
|Untreated gout||0.58 (0.30–1.14)||0.85 (0.62–1.18)||0.304||0.80 (0.53–1.21)||0.77 (0.51–1.17)||0.838||0.98 (0.56–1.70)||0.73 (0.52–1.03)||0.405|
When we repeated the analyses after advancing the PD diagnosis date by 2 years to examine the potential impact of insidious onset of PD cases, the multivariate RRs for PD were 0.71 (95% CI 0.59–0.87) for gout and 0.66 (95% CI 0.52–0.83) for treated gout. In analyses that included prevalent cases of gout recorded in the first year, we found similar negative associations with risk of PD for gout (RR 0.68, 95% CI 0.59–0.77) and treated gout (RR 0.64, 95% CI 0.55–0.74).
In this population-based study of Canadians age ≥65 years, we found a 30% reduction in the risk of PD among individuals with a history of gout. The inverse association between gout and the risk of PD was independent of age, sex, prior comorbid conditions, and use of diuretics and NSAIDs. The inverse association was evident among both men and women and individuals that did not use diuretics. These findings lend further support to the purported protective role of uric acid against PD. As potent urate-lowering agents are being developed (12–14), this potential adverse neurologic issue associated with lower urate levels may become clinically relevant. Furthermore, these findings are consistent with the notion that hyperuricemia and gout may have certain beneficial effects, which may agree with the evolutionary advantage of relative hyperuricemia in humans (1, 3).
Our findings expand on previous epidemiologic findings of the inverse association between serum urate levels and risk of PD, and confirm the recent GPRD report of an inverse link between history of gout and the risk of PD (11). Two retrospective case–control studies (9, 10) and 3 prospective studies (6–8) have reported an inverse association between serum uric acid levels and the risk of PD among men. The Honolulu Heart Study based on 7,968 Japanese American men found that subjects with baseline serum uric acid measurements over the median value had a 40% lower risk of incident PD (6). Similarly, the Rotterdam Study showed that a 1 SD increase in serum uric acid level was associated with a 30% lower risk of PD in men (7). Recently, the Health Professionals Follow-up Study reported that men in the highest quartile of uricemia had a 57% lower risk of PD compared with men in the lowest quartile (8). Whereas these studies evaluated the impact of serum uric acid, the GPRD case–control study expanded the investigation to the impact of gout (11). Based on 1,052 PD cases and 6,634 controls, Alonso and colleagues found that individuals with a history of gout had a 31% lower risk of developing PD, a finding that closely agrees with our results. Because these data were observational, we cannot exclude the possibility that unmeasured factors in these studies might have contributed to the observed associations, or that low uric acid levels might have served as an epiphenomenon of PD pathogenesis. However, we are not aware of any concrete mechanism that would lead to both prior low serum uric acid levels and a future risk of PD. Overall, these data provide consistent, prospective evidence that elevated uric acid levels are associated with a lower future risk of PD.
Furthermore, the potential impact of serum uric acid in the progression of already existing PD has also been examined in 2 clinical trials: the Parkinson Research Examination of CEP-1347 (PRECEPT) (31) and Deprenyl and Tocopherol Antioxidative Therapy of Parkinsonism (DATATOP) trials (32). In the PRECEPT study of 806 subjects with PD, those with higher serum urate levels had slower progression of PD symptoms and signs (31). A better outcome was also associated with high levels of uric acid in cerebrospinal fluid in the DATATOP trial (32). Taken together, all of these data suggest that elevated uric acid levels have a neuroprotective effect on PD.
The potential mechanisms behind the observed protective association of gout in PD include the antioxidant properties of uric acid. Uric acid can reduce oxidative stress through its actions as an effective scavenger of peroxynitrite (ONOO−) and hydroxyl radicals (OH•) (33). In the 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) model of PD, uric acid completely prevented the death of the dopaminergic cells induced by homocysteine plus rotenone or iron (34). Possible mechanisms for a neuroprotective action of uric acid include the suppression of oxyradical accumulation and preservation of mitochondrial function (35), inhibition of the cytotoxic activity of lactoperoxidase (36), repair of DNA damaged by free radicals (37), and protection from dopamine-induced apoptosis (38). The beneficial effects of uric acid have also been shown in animal models of neurologic disorders, including multiple sclerosis and spinal cord injury (39).
Treated gout was associated with a lower risk of PD than untreated gout in our study (RR 0.66 versus 0.79), as well as in the GPRD study (odds ratio 0.61 versus 0.75) (11). Although these findings suggest an additional benefit from anti-gout medicine, particularly allopurinol, an anti-gout treatment history also reflects severe gout or high levels of serum uric acid at the time of treatment initiation (i.e., confounding by indication). Because the noncompliance rate (44%) (40) and nonpersistence rate (72%) (41) of allopurinol use are high, the long-term serum uric acid levels existing prior to the onset of PD could be higher among patients treated with any anti-gout medication than among untreated patients. Furthermore, this potential treatment effect cannot explain the previously reported link between serum uric acid levels and risk of PD in population studies, because the prevalence of anti-gout medication use in these studies was minimal (<1%) (6–8).
The inverse association between history of gout and risk of PD similarly existed in both sexes with gout and treated gout. The association among women was not observed in the GPRD, which may be explained by the definition for gout that was based on one recorded diagnosis (11). In our study, gout was defined using at least 2 recorded diagnoses. When we reanalyzed the data using 1 gout visit (n = 587 female PD cases), the association disappeared (RR for gout among women 0.87, 95% CI 0.69–1.10). The discrepancy could also stem from the difference in age ranges among women in the GPRD (≥40 years) (42) and in our study (≥65 years), because the inverse association was only seen in individuals age ≥60 years in the GPRD study (11). We also found a significant inverse association among gout patients who did not use diuretics. Although these data suggest that the potential benefit of uric acid may be less evident among individuals with higher levels of uric acid induced by diuretic use, the sample size was substantially smaller. Of note, diuretic use was not taken into account in the previous GPRD study (11). Because diuretic use is a well-established cause of hyperuricemia and gout, it would be important to incorporate this variable in future studies (43). Overall, these potential subgroup effects of sex and diuretic use call for confirmation in future studies.
Strengths and limitations of our study deserve comment. This study was performed using a population-based sample of Canadian women and men; therefore, findings are likely to be applicable to the general population. Because the definition of gout was based on administrative diagnosis codes and prescription records, a certain level of misclassification of exposure is inevitable, as described in Materials and Methods. A diagnosis of gout could have often been recorded based on the suggestive clinical presentation of gout, likely in combination with the presence of hyperuricemia, but without documentation of monosodium urate crystals. Nonetheless, an inverse link with these “soft cases” of gout with hyperuricemia would also support the hypothesized protective effect of hyperuricemia against PD. A potential detection bias (i.e., increased detection of PD due to clinical care associated with gout) would not explain the observed inverse association. Furthermore, any nondifferential misclassification of these diagnoses would have biased the study results toward the null and would not explain the significant associations observed in this study. Nonetheless, confirmation of these results using specific case definitions of gout and investigation of other similarly degenerative neurologic diseases would be valuable. Finally, use of an administrative data set precluded adjustment for unmeasured risk factors for PD such as coffee consumption and smoking status (44). However, coffee consumption and smoking did not confound the risk of PD in relation to serum uric acid levels (8) and history of gout (11), respectively. Therefore, further adjustment of these factors would not likely affect our results.
In conclusion, our population-based data provide evidence for a protective effect of gout on the risk of PD and support the purported protective role of uric acid. The protective effect may be evident among both men and women and those who do not use diuretics.
Dr. Choi had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. De Vera, Gao, Choi.
Acquisition of data. De Vera, Kopec, Choi.
Analysis and interpretation of data. De Vera, Rahman, Rankin, Gao, Choi.
Manuscript preparation. De Vera, Rahman, Rankin, Kopec, Gao, Choi.
Statistical analysis. De Vera, Rahman, Choi.
The authors thank Dr. Miguel Hernan for his critical review of the manuscript.