Parkinson's disease (PD), which is characterised by bradykinesia, rigidity with cogwheeling, rest tremor and postural instability, is one of the most common neurodegenerative disorders. Although genetic factors are thought to play an important role in some cases of PD, environmental and lifestyle factors are probably responsible for the majority of cases(Reference de Lau and Breteler1). Although symptomatic treatment has improved in recent years, a cure for PD is not yet available. Because the cause of onset is largely unknown, epidemiology is important in defining the cause of PD and in evaluating preventive measures.
The importance of research on the possible role of dietary factors in PD is emphasised by the fact that diet is modifiable. Given that increased homocysteine levels might accelerate dopaminergic cell death in PD through neurotoxic effects(Reference Duan, Ladenheim and Cutler2), increasing intake of B vitamins involved in the regulation of homocysteine metabolism (i.e. folate, vitamin B6, vitamin B12 and riboflavin) might decrease the risk of PD through decreasing plasma homocysteine(Reference Obeid, McCaddon and Herrmann3, Reference Gaenslen, Gasser and Berg4). However, epidemiological evidence for the association of intake of these B vitamins with PD is sparse(Reference Hellenbrand, Boeing and Robra5–Reference Chen, Zhang and Schwarzschild8), and no studies have been conducted in non-Western populations.
Here, we examined the association between dietary intake of folate, vitamin B6, vitamin B12 and riboflavin and the risk of PD using data from a multicentre hospital-based case–control study in Japan(Reference Miyake, Sasaki and Tanaka9, Reference Murakami, Miyake and Sasaki10). Population-specific genetic characteristics (e.g. single-nucleotide polymorphisms influencing homocysteine metabolism)(Reference Wilcken, Bamforth and Li11) especially explain the need to investigate this topic in a Japanese population.
Subjects and methods
Study population
The present study was based on a multicentre hospital-based case–control study conducted at eleven collaborating hospitals in Japan (three university and one national hospitals in Fukuoka, and three university, three national and one municipal hospitals in Kinki). Recruitment of patients with PD was done at all the collaborating hospitals. Eligible cases were patients who were within 6 years of the onset of PD, and had received treatment at one of the collaborating hospitals during the period 1 April 2006 to 31 March 2008. PD was diagnosed by the collaborating neurologists using the UK PD Society Brain Bank criteria(Reference Hughes, Daniel and Kilford12). The neurologists in charge invited 298 eligible PD patients to take part; of these, 250 patients were cooperative in answering the questionnaires, whereas 48 declined (response rate: 84 %).
Recruitment of control subjects was done at three of the eleven collaborating hospitals (one university hospital in Fukuoka, and one university and one national hospitals in Kinki). Eligible control subjects were inpatients and outpatients without neurodegenerative diseases (i.e. orthopaedic surgery, ophthalmology, otorhinolaryngology, plastic surgery and oral surgery) during the period 1 April 2006 to 31 March 2008. Controls were not matched to cases either individually or by group. A total of 528 patients were approached by the attending doctors or our research nurses for recruitment as controls; of whom, 372 agreed, while 156 declined (response rate: 70 %).
For analysis, one case and four controls were excluded due to missing data on the factors under study, leaving 249 cases and 368 controls for the final analysis. The present study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects/patients were approved by the ethics committees of the eleven collaborating hospitals (Faculty of Medicine, Fukuoka University; Utano National Hospital; Osaka City University Graduate School of Medicine; Graduate School of Medical Sciences, Kyushu University; Wakayama Medical University; Kyoto University Graduate School of Medicine; Kurume University School of Medicine; Minami-Kyoto National Hospital; Toneyama National Hospital; Kyoto City Hospital and National Omuta Hospital). Written informed consent was obtained from all subjects/patients.
Measurements
Case and control subjects filled out a set of two self-administered questionnaires, and mailed them to the data management centre or handed them to research nurses. Research technicians completed missing or illogical data by telephone or direct interview.
Dietary habits during the preceding month were assessed using a self-administered, semi-quantitative, comprehensive, diet history questionnaire (DHQ). Details of the DHQ's structure and calculation of dietary intake, and validity for commonly studied nutritional factors have been published elsewhere(Reference Sasaki, Yanagibori and Amano13–Reference Murakami, Sasaki and Takahashi16). Briefly, the DHQ is a structured sixteen-page questionnaire which asks about the consumption frequency and portion size of selected foods commonly consumed in Japan as well as about general dietary behaviour and usual cooking methods(Reference Sasaki, Yanagibori and Amano13, Reference Murakami, Sasaki and Takahashi16). Estimates of daily intake for foods (150 items in total), energy and selected nutrients were calculated using an ad hoc computer algorithm for the DHQ(Reference Sasaki, Yanagibori and Amano13, Reference Murakami, Sasaki and Takahashi16) based on the Standard Tables of Food Composition in Japan(17, 18). Because only a small number of subjects used dietary supplements weekly or more often (e.g. vitamin B complex, 4 % and multivitamins, 7 %), the use of dietary supplements was not incorporated into the analysis. Dietary glycaemic index, a measure of carbohydrate quality, was calculated according to a procedure described elsewhere(Reference Murakami, Miyake and Sasaki10, Reference Murakami, Sasaki and Takahashi16, Reference Murakami, Sasaki and Takahashi19). To minimise the influence of dietary misreporting, an ongoing controversy in studies that collect dietary information using self-reported instruments(Reference Murakami, Sasaki and Takahashi20), dietary variables were energy adjusted using the density method (except for glycaemic index)(Reference Murakami, Miyake and Sasaki10, Reference Murakami, Sasaki and Takahashi16, Reference Murakami, Sasaki and Takahashi19). In a previous study of forty-seven Japanese women, Pearson's correlation coefficients between the DHQ and 3 d estimated dietary records were 0·48 for energy, 0·37–0·75 for energy-providing nutrients and 0·38–0·68 for other nutrients(Reference Sasaki, Yanagibori and Amano13); in another study of ninety-two Japanese women and ninety-two Japanese men(Reference Murakami, Sasaki and Takahashi16), Pearson's correlation coefficients between the DHQ and 16 d weighed dietary records were 0·58 and 0·39 for folate, 0·63 and 0·58 for vitamin B6, 0·51 and 0·41 for vitamin B12, 0·51 and 0·39 for riboflavin, 0·42 and 0·49 for cholesterol, 0·50 and 0·58 for dietary glycaemic index, 0·42 and 0·48 for vitamin E, 0·45 and 0·47 for vitamin C, 0·63 and 0·40 for β-carotene, 0·74 and 0·82 for alcohol, 0·43 and 0·38 for caffeine, and 0·68 and 0·52 for Fe, respectively (Sasaki S, unpublished results, 2006), suggesting satisfactory validity of the DHQ in terms of B vitamins and other dietary variables.
Body weight and height were self-reported as part of the DHQ. BMI was calculated as weight (kg) divided by the square of height (m2). The second questionnaire on non-dietary factors potentially associated with PD elicited information on sex, age, education and smoking habits. This questionnaire was developed for this survey based on comprehensive literature review of epidemiological studies on risk factors for PD, although its validity has not been investigated.
Statistical analysis
All statistical analyses were performed using SAS statistical software version 9.1 (SAS Institute, Inc., Cary, NC, USA). Dietary intake (amount per 4184 kJ of energy) of B vitamins (i.e. folate, vitamin B6, vitamin B12 and riboflavin) was categorised at quartile points on the distribution of control subjects. Using logistic regression analysis, crude and multivariate-adjusted OR and 95 % CI for PD for each quartile category of dietary intake were calculated (i.e. models 1 and 2). The lowest quartile category of dietary intake was used as a reference category. Multivariate-adjusted OR (model 2) were calculated by adjusting for potential non-dietary confounding factors(Reference Miyake, Sasaki and Tanaka9, Reference Murakami, Miyake and Sasaki10), including sex (men or women), age (years, continuous), region (Fukuoka or Kinki), pack-years of smoking (none, 0·1–29·9 or ≥ 30·0), education ( < 10, 10–12 or ≥ 12 years) and BMI (kg/m2, continuous). Further adjustment (i.e. model 3) was done for potential dietary confounding factors (continuous), i.e. variables associated with PD in this population(Reference Miyake, Sasaki and Tanaka9, Reference Murakami, Miyake and Sasaki10), namely cholesterol (mg/4184 kJ) and glycaemic index, and variables possibly associated with PD according to the literature(Reference de Lau and Breteler1, Reference Gaenslen, Gasser and Berg4), namely vitamin E (mg α-tocopherol/4184 kJ), vitamin C (mg/4184 kJ), β-carotene (μg/4184 kJ), alcohol (% of energy), caffeine (mg/4184 kJ) and Fe (mg/4184 kJ). Finally, additional adjustment for intake of other B vitamins (continuous) was done (i.e. model 4). We also conducted four sensitivity analyses (using model 4): that confined to cases diagnosed less than 3 years from onset; that excluding control subjects having diseases of the digestive tract; that confined to subjects recruited from three of the eleven collaborating hospitals at which cases were recruited; and that excluding dietary supplement users. Trends of association were assessed by a logistic regression model which assigned consecutive integers to the levels of the independent variable. All reported P values are two-tailed, and P values < 0·05 were considered statistically significant.
Results
Characteristics of the study population are shown in Table 1. About 40 % of both case and control subjects were males, and recruited from the Fukuoka region. Cases were more likely to be older, have lower BMI and be non-smokers. Cases also had a significantly lower intake of folate, alcohol and caffeine, and a significantly higher intake of cholesterol. There were no differences between cases and controls in education and other dietary variables, or in sex or region.
Table 1 Characteristics of the study population
(Mean values and standard deviations)
* For categorical variables, the χ2 test was used; for continuous variables, the independent-samples t test was used.
Table 2 shows sample characteristics of the lowest and highest quartiles of dietary intake of B vitamins. There were more women and non-smokers in higher quartiles of dietary folate, vitamin B6 and riboflavin. There was a positive association between intake of folate and riboflavin and age. Vitamin B6 intake was inversely associated with BMI. There were positive associations among intake of B vitamins examined; Pearson's correlation coefficients ranged from 0·39 (folate and vitamin B12) to 0·68 (folate and riboflavin). Intake of B vitamins was associated positively with intake of cholesterol, vitamins E and C, β-carotene and caffeine, and inversely with dietary glycaemic index and alcohol.
Table 2 Sample characteristics for the lowest (Q1) and highest (Q4) quartiles of dietary intake of B vitamins
(Mean values and percentages)
* For categorical variables, the χ2 test was used; for continuous variables, a linear trend test was used.
Table 3 shows crude and multivariate OR for PD according to the quartile of dietary intake of B vitamins. After adjustment for potential non-dietary confounding factors (model 2), dietary intake of folate was inversely associated with the risk of PD. Multivariate OR for PD in the first, second, third and fourth quartiles of dietary folate were 1 (reference), 0·81 (95 % CI 0·51, 1·29), 0·65 (95 % CI 0·40, 1·06) and 0·50 (95 % CI 0·30, 0·85), respectively (P for trend = 0·007). However, the association between folate and PD disappeared after further adjustment for potential dietary confounding factors (model 3; P for trend = 0·68) and other B vitamins (model 4; P for trend = 0·87).
Table 3 OR for Parkinson's disease according to the quartile of dietary intake of B vitamins
(Odds ratios and 95 % confidence intervals)
* Values are median (range).
† Crude model.
‡ Adjusted for non-dietary factors, including sex (men or women), age (years, continuous), region (Fukuoka or Kinki), pack-years of smoking (none, 0·1–29·9 or ≥ 30·0), education ( < 10, 10–12 or ≥ 12 years) and BMI (kg/m2, continuous).
§ Adjusted for non-dietary factors (variables used in model 2) and dietary factors (continuous), including cholesterol (mg/4184 kJ), dietary glycaemic index, vitamin E (mg α-tocopherol/4184 kJ), vitamin C (mg/4184 kJ), β-carotene (μg/4184 kJ), alcohol (% of energy), caffeine (mg/4184 kJ) and Fe (mg/4184 kJ).
∥ Adjusted for non-dietary and dietary factors (variables used in model 3) and intake of other B vitamins (continuous), including folate (μg/4184 kJ), vitamin B6 (mg/4184 kJ), vitamin B12 (μg/4184 kJ) and riboflavin (mg/4184 kJ).
Conversely, low intake of vitamin B6 was associated with an increased risk of PD, independent of non-dietary confounding factors (model 2). Multivariate OR for PD in the first, second, third and fourth quartiles of dietary vitamin B6 were 1 (reference), 0·64 (95 % CI 0·40, 1·03), 0·75 (95 % CI 0·47, 1·19) and 0·58 (95 % CI 0·36, 0·94), respectively (P for trend = 0·052). This association between vitamin B6 and PD was generally retained after further adjustment for dietary confounders (model 3; P for trend = 0·13) and other B vitamins (model 4). Multivariate OR for PD in the first, second, third and fourth quartiles of dietary vitamin B6 were 1 (reference), 0·56 (95 % CI 0·33, 0·94), 0·69 (95 % CI 0·38, 1·25) and 0·48 (95 % CI 0·23, 0·99), respectively (P for trend = 0·10). Intake of vitamin B12 and riboflavin was not associated with the risk of PD.
Discussion
In the present case–control study in Japan, we found that after adjustment for potential dietary and non-dietary confounding factors, low intake of dietary vitamin B6 was associated with an increased risk of PD. In contrast, no independent association was seen between intake of folate, vitamin B12 and riboflavin and the risk of PD. To our knowledge, this is the first study to examine the relationship of dietary B vitamins with PD in non-Western populations.
A limited number of epidemiological studies have examined this topic. A case–control study in Germany found that a higher intake of folate, vitamin B6 and vitamin B12, but not of riboflavin, was associated with a lower risk of PD(Reference Hellenbrand, Boeing and Robra5). A prospective cohort study in the Netherlands showed that a higher intake of vitamin B6, but not of folate or vitamin B12, was associated with a decreasing risk of PD(Reference de Lau, Koudstaal and Witteman6). These are partially consistent with the present findings. Conversely, a case–control study in the US found no association for intake of folate, vitamin B6 or vitamin B12(Reference Johnson, Gorell and Rybicki7), while a large US-based prospective study in health professionals found no association between intake of folate, vitamin B6 or vitamin B12 and PD risk(Reference Chen, Zhang and Schwarzschild8). Considering the presence of food fortification with folic acid, however, the USA may not be the ideal setting to assess this association(Reference de Lau and Breteler1). These discrepancies among studies may be explained, at least in part, by differences in study population and design, dietary assessment methods used, definitions of PD applied and potential confounding factors considered.
The association between the intake of nutrients that influence homocysteine concentration and the risk of PD can be explained by homocysteine metabolism because of the potential neurotoxic effects of homocysteine(Reference de Lau and Breteler1). Previous studies have generally shown that dietary intake of folate and related B vitamins such as vitamin B6 and riboflavin is moderately associated with plasma homocysteine concentration, whereas no such association is expected for vitamin B12, at least when folate intake is within a normal range(Reference Ganji and Kafai21–Reference Selhub, Jacques and Wilson25). The findings are generally inconsistent with the present result that low intake of vitamin B6, but not of folate, vitamin B12 or riboflavin, is associated with an increased risk of PD. The association between vitamin B6 and PD might be explained by neuroprotective properties of vitamin B6 through antioxidant capacities (unrelated to homocysteine metabolism)(Reference Mahfouz and Kummerow26, Reference Chumnantana, Yokochi and Yagi27), and through its role in dopamine synthesis(Reference Tan, Cheah and Fook-Chong28). Alternatively, vitamin B6 intake might be a marker of dietary patterns that are related to decreased PD risk (e.g. high intake of a wide range of antioxidants). Additionally, a potential effect of blood levels of B vitamins involved in the regulation of homocysteine metabolism is not ruled out by the present study based on dietary intake. Thus, future research which directly examines the association of PD with blood concentrations of homocysteine and related B vitamins would be of interest.
Several limitations of the study warrant mention. First, diagnosis of PD is based on clinical symptoms, and it is assumed that some proportion of cases may be misdiagnosed, which could cause bias towards the null. Nevertheless, cases were identified according to strict diagnostic criteria, and thus the possibility of misclassification of PD is small.
One may speculate that data obtained from PD patients based on the two self-administered questionnaires are not reliable on the basis that PD affects several aspects of symptoms including movement, mood, behaviour, thinking and sensation. Nevertheless, responses to the questionnaires were carefully checked for completeness by research technicians trained for this survey, who clarified any missing values or errors of logic by telephone or direct interview with subjects. The possible influence of the symptoms of PD on the quality of data derived from the questionnaires should thus have been minimised.
The dietary assessment questionnaire used (i.e. DHQ) has not been validated in the present population, although it has been validated in other Japanese groups(Reference Sasaki, Yanagibori and Amano13–Reference Murakami, Sasaki and Takahashi16). The possibility of misclassification might have biased the magnitude of the observed associations towards the null. More importantly, current rather than past dietary habits were assessed, not only because the critical period of life during which PD develops is the subject of much controversy(Reference Vingerhoets, Snow and Lee29, Reference Morrish, Sawle and Brooks30) but also because retrospective estimation of past diet is generally difficult(Reference Byers, Rosenthal and Marshall31), and the utility of our DHQ for past diet is unknown. Thus, the present study is inevitably based on the assumption that the relative food intake patterns of cases and controls have remained fairly stable over time. Nevertheless, this may be reasonable given that there seem to be no special diets for PD followed in Japan, albeit that dopamine shortage can affect food preferences(Reference Wang, Volkow and Fowler32). In this regard, the results of a sensitivity analysis confined to cases diagnosed less than 3 years from onset (n 109) were similar to those of the overall analysis, with the exception of a positive association for vitamin B12. The multivariate OR in the highest (compared with the lowest) quartile were 0·51 (95 % CI 0·16, 1·66) for folate (P for trend = 0·20), 0·30 (95 % CI 0·11, 0·82) for vitamin B6 (P for trend = 0·0496), 2·55 (95 % CI 1·10, 5·92) for vitamin B12 (P for trend = 0·09) and 0·49 (95 % CI 0·17, 1·37) for riboflavin (P for trend = 0·19).
Because our controls were not normal healthy persons, but rather inpatients and outpatients without neurodegenerative diseases, they may be unlikely to provide an estimate of the exposure prevalence in the population from which the cases arose, although the response rate was relatively high (70 %). In Japan at least, the use of normal healthy people in the community as controls is not always sufficient, given the relatively low response rate observed in a case–control study in Japan (50 %)(Reference Miyake33). Some control subjects had diseases of the digestive tract such as stomach cancer (n 2), or diseases associated with long-term modification of the diet such as diabetes (n 42; no controls with ulcerative colitis or Crohn's disease). Nevertheless, the results of a sensitivity analysis excluding these subjects were similar to those of the overall analysis, with the exception of no association for vitamin B6. The multivariate OR in the highest (compared with the lowest) quartile were 1·03 (95 % CI 0·41, 2·58) for folate (P for trend = 0·94), 0·53 (95 % CI 0·25, 1·11) for vitamin B6 (P for trend = 0·15), 1·15 (95 % CI 0·60, 2·22) for vitamin B12 (P for trend = 0·97) and 0·59 (95 % CI 0·26, 1·33) for riboflavin (P for trend = 0·12). Furthermore, although our control subjects were selected from three of the eleven collaborating hospitals at which cases were recruited, the results of a sensitivity analysis confined to patients recruited from these three hospitals (n 153) were similar to those of the overall analysis, with the exception of no association for vitamin B6. The multivariate OR in the highest (compared with the lowest) quartile were 0·65 (95 % CI 0·23, 1·81) for folate (P for trend = 0·53), 0·68 (95 % CI 0·30, 1·55) for vitamin B6 (P for trend = 0·50), 1·94 (95 % CI 0·93, 4·06) for vitamin B12 (P for trend = 0·12) and 0·57 (95 % CI 0·23, 1·41) for riboflavin (P for trend = 0·17).
As mentioned above, the percentage of subjects using dietary supplements such as vitamin B complex and multivitamins in the present study was not large (12 % (n 29) for cases and 10 % (n 38) for controls). Moreover, the exclusion of these dietary supplement users from the analysis did not materially alter the results. The multivariate OR in the highest (compared with the lowest) quartile were 1·06 (95 % CI 0·41, 2·77) for folate (P for trend = 0·81), 0·45 (95 % CI 0·21, 0·98) for vitamin B6 (P for trend = 0·12), 1·09 (95 % CI 0·56, 2·15) for vitamin B12 (P for trend = 0·93) and 0·58 (95 % CI 0·25, 1·38) for riboflavin (P for trend = 0·13). Thus, it is unlikely that dietary supplements had a major effect on the present findings.
Although adjustments to compensate for a variety of potential confounding variables were attempted, residual confounding effects could not be ruled out. In particular, potential non-dietary confounding factors were assessed using a non-validated questionnaire. Finally, while information on cognitive impairment is unfortunately unavailable in the present study, there may be some subjects (particularly PD patients) with cognitive impairment, which can affect both dietary intake and the ability to recall diet, and hence the present results.
In conclusion, the present case–control study in Japan showed that low intake of vitamin B6, but not of folate, vitamin B12 or riboflavin, was independently associated with an increased risk of PD. Because epidemiological research on this topic is still sparse, further evidence from well-designed case–control and prospective cohort studies is required to accumulate evidence on potential beneficial role of dietary B vitamins in the development of PD.
Acknowledgements
None of the authors had any personal or financial conflict of interest. The present study was supported by Health and Labour Sciences Research Grants, Research on Intractable Diseases, Research Committee on Epidemiology of Intractable Diseases, from the Ministry of Health, Labour and Welfare, Japan. K. M. contributed to statistical analysis, data interpretation and manuscript writing. Y. M. contributed to study design, data collection, overall management and manuscript editing. S. S. contributed to design of the dietary study and manuscript editing. K. T., W. F. and C. K. contributed to study design, data collection and data management. Y. T., T. Y., T. O., N. K., N. S. and H. F. contributed to outcome definition and case recruitment. Y. H. and M. N. contributed to supervision of the design and execution of the study. All authors contributed to the preparation of the manuscript and approved the final version submitted for publication.
Appendix
Other members of the Fukuoka Kinki Parkinson's Disease Study Group are as follows: Yasuhiko Baba and Tomonori Kobayashi (Department of Neurology, Faculty of Medicine, Fukuoka University); Hideyuki Sawada, Eiji Mizuta and Nagako Murase (Clinical Research Institute and Department of Neurology, Utano National Hospital); Tsuyoshi Tsutada and Hiroyuki Shimada (Department of Geriatrics and Neurology, Osaka City University Graduate School of Medicine); Jun-ichi Kira (Department of Neurology, Neurological Institute, Graduate School of Medical Sciences, Kyushu University); Tameko Kihira and Tomoyoshi Kondo (Department of Neurology, Wakayama Medical University); Hidekazu Tomimoto (Department of Neurology, Kyoto University Graduate School of Medicine); Takayuki Taniwaki (Division of Respirology, Neurology and Rheumatology, Department of Medicine, Kurume University School of Medicine); Hiroshi Sugiyama and Sonoyo Yoshida (Department of Neurology, Minami-Kyoto National Hospital); Harutoshi Fujimura and Tomoko Saito (Department of Neurology, Toneyama National Hospital); Kyoko Saida and Junko Fujitake (Department of Neurology, Kyoto City Hospital); Naoki Fujii (Department of Neurology, Neuro-Muscular Center, National Omuta Hospital); Masatoshi Naito and Jun Arimizu (Department of Orthopaedic Surgery, Faculty of Medicine, Fukuoka University); Takashi Nakagawa, Hirofumi Harada and Takayuki Sueta (Department of Otorhinolaryngology, Faculty of Medicine, Fukuoka University); Toshihiro Kikuta and George Umemoto (Department of Oral and Maxillofacial Surgery, Faculty of Medicine, Fukuoka University); Eiichi Uchio and Hironori Migita (Department of Ophthalmology, Faculty of Medicine, Fukuoka University); Kenichi Kazuki, Yoichi Ito and Hiroyoshi Iwaki (Department of Orthopedic Surgery, Osaka City University Graduate School of Medicine); Kunihiko Siraki and Shinsuke Ataka (Department of Ophthalmology and Visual Sciences, Osaka City University Graduate School of Medicine); Hideo Yamane and Rie Tochino (Department of Otolaryngology and Head and Neck Surgery, Osaka City University Graduate School of Medicine); Teruichi Harada (Department of Plastic and Reconstructive Surgery, Osaka City University Graduate School of Medicine); and Yasushi Iwashita, Motoyuki Shimizu, Kenji Seki and Keiji Ando (Department of Orthopedic Surgery, Utano National Hospital).
