Dietary fats, cholesterol and iron as risk factors for Parkinson's disease

1. Introduction 

Although the etiology of Parkinson's disease (PD) is poorly understood, epidemiologic evidence suggests that dietary factors may contribute to the development of PD [1]. Dietary influences on PD risk may be mediated by oxidative stress, which can lead to dopaminergic cell loss. The cells in the substantia nigra are normally subject to a high degree of oxidative stress, due in part to the metabolism of dopamine and subsequent creation of hydroxyl radicals via the Fenton reaction [2]. Dietary intake of iron, a catalyst in the Fenton reaction, has been shown to increase PD risk either in association with animal fat [3], or alone [4], [5]. Dietary fats, including cholesterol, have been associated with PD in some studies, and are thought to contribute to oxidative stress, but results have been conflicting [3], [4], [5], [6], [7], [8], [9], [10]. Findings regarding PD risk related specifically to dietary cholesterol have indicated either no association [6], [8], [9], or an increased risk [4]. Elevated serum levels of cholesterol have recently been related to a decreased PD risk [11], [12]. In a previously reported analysis from this study, saturated fat and total fat showed no association with PD risk, but individual fatty acids and cholesterol were not analyzed at that time [5]. Here, we examine the levels of dietary fat and cholesterol as well as iron intake, and their associations with PD risk.

3. Results 

The relations between PD and fats and cholesterol are shown in Table 2. Total fat, saturated fat, and the other fatty acids listed (monounsaturated, polyunsaturated, oleic acid, linoleic acid, linolenic acid, eicosapentaenoic acid and docosahexaenoic acid) showed no consistent associations with PD. Cholesterol intake in the highest quartile compared with the lowest quartile was associated with a decreased risk of PD in men (OR=0.53, 95%CI: 0.33, 0.86), trend p=0.007 but not in women (OR=1.13, 95%CI: 0.59, 2.19), trend p=0.819.

Table 2. Associations of Parkinson's disease with lipids (fat and cholesterol)

	Nutrient quartiles	g/day/ 2000kcal	Men				Women
			Cases (n=266)	Controls (n=351)	OR*	95%CI*	Cases (n=154)	Controls (n=209)	OR*	95%CI*
	Total fat
	1	0–60.8	85	80	1.0	–	48	60	1.0	–
	2	60.8–69.2	62	84	0.76	0.48, 1.22	36	56	0.77	0.42, 1.39
	3	69.2–77.9	54	95	0.62	0.39, 1.00	33	45	0.91	0.50, 1.66
	4	>77.9	65	92	0.81	0.51, 1.29	37	48	1.03	0.57, 1.87
	p for trend					0.269				0.855
	Saturated fat
	1	0–19.9	75	87	1.0	–	46	53	1.0	–
	2	19.9–23.7	69	95	0.88	0.56, 1.39	36	45	0.87	0.47, 1.62
	3	23.7–27.5	61	77	0.99	0.61, 1.59	37	63	0.66	0.37, 1.18
	4	>27.5	61	92	0.82	0.51, 1.31	35	48	0.81	0.44, 1.50
	p for trend					0.511				0.313
	Monounsaturated fat
	1	0–22.2	87	77	1.0	–	44	63	1.0	–
	2	22.2–26.0	59	87	0.66	0.41, 1.05	44	53	1.21	0.67, 2.15
	3	26.0–30.0	55	95	0.63	0.39, 1.01	29	45	0.94	0.50, 1.74
	4	>30.0	65	92	0.74	0.46, 1.17	37	48	1.23	0.67, 2.25
	p for trend					0.187				0.694
	Polyunsaturated fat
	1	0–9.8	77	75	1.0	–	35	65	1.0	–
	2	9.8–12.3	70	92	0.94	0.59, 1.49	52	48	2.14	1.19, 3.85
	3	12.3–15.1	54	94	0.70	0.43, 1.13	23	46	1.03	0.52, 2.01
	4	>15.1	65	90	0.91	0.56, 1.46	44	50	1.92	1.04, 3.52
	p for trend					0.462				0.179
	Oleic acid
	1	0–20.2	80	79	1.0	–	44	61	1.0	–
	2	20.2–24.2	74	89	0.92	0.58, 1.45	43	51	1.14	0.64, 2.04
	3	24.2–27.8	47	90	0.61	0.37, 1.00	27	50	0.77	0.41, 1.45
	4	>27.8	65	93	0.85	0.53, 1.35	40	47	1.31	0.72, 2.37
	p for trend					0.252				0.659
	Linoleic acid
	1	0–8.3	77	76	1.0	–	37	64	1.0	–
	2	8.3–10.7	73	93	0.94	0.59, 1.49	50	47	1.85	1.03, 3.32
	3	10.7–13.7	55	91	0.72	0.44, 1.17	25	49	0.96	0.50, 1.85
	4	>13.7	61	91	0.84	0.52, 1.35	42	49	1.67	0.91, 3.06
	p for trend					0.310				0.318
	Linolenic acid
	1	0–9.1	90	85	1.0	–	38	55	1.0	–
	2	9.1–1.0	67	95	0.72	0.46, 1.14	39	45	1.52	0.81, 2.85
	3	1.0–1.1	54	84	0.71	0.44, 1.14	36	56	1.04	0.56, 1.95
	4	>1.1	55	87	0.75	0.47, 1.21	41	53	1.23	0.66, 2.28
	p for trend					0.234				0.796
	Eicosapentaenoic acid (EPA)
	1	0–0.05	47	87	1.0	–	37	53	1.0	–
	2	0.05–0.08	77	89	1.59	0.97, 2.59	37	51	1.02	0.54, 1.90
	3	0.08–0.11	66	84	1.39	0.84, 2.30	38	56	0.84	0.46, 1.55
	4	>0.11	76	91	1.34	0.81, 2.20	42	49	1.30	0.70, 2.43
	p for trend					0.440				0.563
	Docosahexaenoic acid (DHA)
	1	0–0.08	54	89	1.0	–	36	51	1.0	–
	2	0.08–0.12	75	81	1.46	0.90, 2.35	37	59	0.88	0.47, 1.64
	3	0.12–0.17	67	89	1.16	0.71, 1.88	42	51	1.07	0.58, 1.98
	4	>0.17	70	92	1.05	0.65, 1.72	39	48	1.25	0.66, 2.37
	p for trend					0.872				0.407
	Cholesterol (mg/day/2000kcal)
	1	0–192	80	76	1.0	–	52	64	1.0	–
	2	192–243	63	75	0.84	0.52, 1.35	43	65	0.88	0.50, 1.53
	3	243–312	66	93	0.70	0.44, 1.10	31	47	0.90	0.49, 1.66
	4	>312	57	107	0.53	0.33, 0.86	28	33	1.13	0.59, 2.19
	p for trend					0.007				0.819

*Odds ratio (OR) adjusted for age, education, ethnicity, caffeine, alcohol, kcal and smoking. Variables are adjusted for total energy intake (weight/day/kcal).

An increased risk from dietary iron intake, exhibiting a dose–response gradient was found in men (OR highest quartile=1.82, 95%CI: 1.11, 2.99), trend p=0.013, but not in women (OR highest quartile=1.12, 95%CI: 0.59, 2.12), trend p=0.314 (Table 3). We also performed an analysis of dietary iron intake stratified on intake of multivitamins or iron supplements (<1/day versus ≥1/day). The reference group was defined as persons with iron intake below the median of controls and <1/day multivitamin or iron supplement intake. Compared with this reference group, we found PD risk increased (OR=1.46, 95%CI: 0.99, 2.15), for the above-median dietary iron intake in subjects with <1/day multivitamin or iron supplement. There was only a weak association for ≥1/day multivitamin or iron supplement in persons with below-median iron intake (OR=1.26, 95%CI: 0.85, 1.87), which was similar to those with ≥1/day multivitamin or iron supplement and the above-median dietary iron intake (OR=1.17, 95%CI: 0.82, 1.67).

Table 3. Associations of Parkinson's disease with iron from food by gender

	Iron quartiles	mg/day/ 2000kcal	Men				Women
			Cases (n=266)	Controls (n=351)	OR*	95%CI*	Cases (n=154)	Controls (n=209)	OR*	95%CI*
	1	0–11.9	49	86	1.0	–	35	54	1.0	–
	2	11.9–13.9	60	83	1.23	0.74, 2.04	34	57	0.81	0.44, 1.52
	3	13.9–15.7	72	96	1.47	0.90, 2.42	48	44	1.56	0.84, 2.91
	4	>15.7	85	86	1.82	1.11, 2.99	37	54	1.12	0.59, 2.12
	P for trend					0.013				0.314

*Odds ratio (OR) adjusted for age, education, ethnicity, caffeine, alcohol, kcal and smoking. Iron is adjusted for total energy intake (weight/day/kcal).

Joint effects of dietary saturated fat and iron, as well as cholesterol and iron are shown in Table 4. Because saturated fat intake tended to decrease PD risk and iron intake increased PD risk, the reference group used was high saturated fat (above the median) and low iron (below the median). Using this reference combination, a modest increase in PD risk was seen for the combination of low saturated fat and high iron intake (OR=1.50, 95%CI: 1.07, 2.11) for all subjects combined. A non-significant increased PD risk was observed for low saturated fat and high iron when men and women were analyzed separately. Joint effects of cholesterol and iron were analyzed using a reference group of high cholesterol (above the median) and low iron (below the median). Increase in the risk for PD risk was observed in the low cholesterol, and high iron combination, but the association was stronger in men (OR=1.96, 95%CI: 1.25, 3.06) than in women (OR=1.35, 95%CI: 0.75, 2.44).

Table 4. Joint effects of saturated fat and cholesterol with iron from food by gender

	Nutrient split at median		All subjects				Men				Women
			Cases (n=420)	Controls (n=560)	OR*	95%CI*	Cases (n=266)	Controls (n=351)	OR*	95%CI*	Cases (n=154)	Controls (n=209)	OR*	95%CI*
	High SatFat, low Fe		109	169	1.0	–	64	96	1.0	–	45	73	1.0	–
	Low SatFat, low Fe		69	111	0.92	0.62, 1.37	45	73	0.86	0.52, 1.43	24	38	1.00	0.52, 1.93
	High SatFat, high Fe		85	111	1.28	0.87, 1.88	58	73	1.30	0.80, 2.12	27	38	1.22	0.64, 2.33
	Low SatFat, high Fe		157	169	1.50	1.07, 2.11	99	109	1.43	0.91, 2.24	58	60	1.66	0.96, 2.85
	High cholest, low Fe		96	159	1.0	–	64	108	1.0	–	32	51	1.0	–
	Low cholest, low Fe		82	121	1.10	0.74, 1.64	45	61	1.26	0.75, 2.12	37	60	0.95	0.51, 1.79
	High cholest, high Fe		86	121	1.33	0.90, 1.96	59	92	1.25	0.78, 1.99	27	29	1.71	0.83, 3.51
	Low cholest, high Fe		156	159	1.65	1.16, 2.35	98	90	1.96	1.25, 3.06	58	69	1.35	0.75, 2.44
	Quartiles: comparison of the lowest and highest quartiles
	SatFat	Fe
	High	Low	36	62	1.0	–	23	45	1.0	–	13	17	1.0	–
	Middle	Middle	326	426	1.34	0.86, 2.11	206	260	1.60	0.92, 2.80	120	166	0.90	0.40, 1.99
	Low	High	58	72	1.41	0.81, 2.47	37	46	1.60	0.79, 3.22	21	26	1.10	0.42, 2.93
	Cholest	Fe
	High	Low	33	56	1.0	–	19	41	1.0	–	14	15	1.0	–
	Middle	Middle	331	452	1.18	0.74, 1.89	207	281	1.53	0.85, 2.77	124	171	0.72	0.32, 1.60
	Low	High	56	52	1.63	0.90, 2.95	40	29	2.70	1.26, 5.76	16	23	0.70	0.25, 1.91

*Odds ratio (OR) adjusted for age, education, ethnicity, caffeine, alcohol, kcal and smoking (and gender for all subjects combined). Variables are adjusted for total energy intake (weight/day/kcal). Median cut-points based on the nutrient weight/day/kcal intake for each person need to be described based on a fixed number of kcal in order to obtain a mg quantity for each nutrient. For an individual with a 2000kcal/day diet the median for iron is 13.9mg, the median for saturated fat is 23.7g, and the median for cholesterol is 243mg. Quartile cut-points are listed in Table 2, Table 3.

In order to compare quartile distributions, a reference group combining the highest quartile of saturated fat and the lowest quartile of iron was defined. Compared to this reference, the lowest saturated fat and the highest iron showed a modest increase in PD risk for both genders combined (OR=1.41, 95%CI: 0.81, 2.47), with the risk slightly higher in men (OR=1.60, 95%CI: 0.79, 3.22) than in women (OR=1.10, 95%CI: 0.42, 2.93). Using this quartile stratification with cholesterol, the risk of PD from the lowest cholesterol in combination with the highest iron increased in men (OR=2.70, 95%CI: 1.26, 5.76) but not in women (OR=0.70, 95%CI: 0.25, 1.91).

4. Discussion 

We observed a moderately reduced PD risk associated with the highest levels of dietary cholesterol intake, although this association was limited to men. The highest levels of iron intake were found to increase PD risk in men, but not in women. When combining low dietary cholesterol and high dietary iron, the risk was even greater than each factor alone, although this association was also limited to men. In addition, the combination of low saturated fats and high iron conferred moderate PD risk elevations in men and women combined, although the effects were considerably stronger in men.

The nutritional values obtained from this analysis are within normal limits. The upper quartile of iron for this population (>15.7mg for a 2000kcal/day diet) is within a normal range, as the recommended intake for iron is 8mg/day for adults over 50 years of age and the Tolerable Upper Intake Level (UL) is 45mg/day for age 14 and over [15]. The median for total fat (69g, or 31% of total kcal/day for a 2000kcal/day diet) is also within the Acceptable Macronutrient Distribution Range (AMDR) for adults of 20–35% of total kcal total fat [16]. Since the body can synthesize cholesterol, no recommended intake is determined [16].

In contrast to our findings, PD and diet have demonstrated an increased risk of PD from cholesterol [4] or no association [6], [8], [9]. Of interest though is that eggs, which are high in cholesterol, were observed to decrease PD risk in one study from Sweden [17]. The lack of association between PD and saturated fat in our study is in agreement with others [6], [9], although an increase in PD risk has also been found from saturated fat intake [4], [8]. The increased risk for PD from dietary iron found here agrees with our previous findings [5] as well as with findings from studies in New York [3] and Michigan [4]. The study in New York also found that the iron effect disappeared when animal fat was considered [3], indicating that PD risk from iron was reduced with higher levels of animal fat. This also agrees with our observation of a lower risk of PD from high iron with high levels of saturated fat. The different associations that we observed in men and women are difficult to explain, but a previous study by de Lau [11] has also shown gender-specific associations; a reduced PD risk from serum cholesterol was found only in women.

Iron catalyzes the Fenton reaction converting hydrogen peroxide (H2O2) to the highly reactive hydroxyl radical. Oxidative stress is thought to occur in idiopathic PD as part of a cascade of events involved in dopaminergic cell death [2]. An increase in oxidative stress from the Fenton reaction is coherent mechanistically with the increased risk seen from elevated iron intake.

Cholesterol as a protective factor for PD has recently been suggested from a large study in Rotterdam, where it was found that higher levels of serum total cholesterol lowered the risk of PD, although this was restricted to women [11]. Another recent cohort study found a modest decreased PD risk with increasing serum cholesterol levels for men and women [18]. It has also been reported that the best predictor of the potent antioxidant coenzyme Q10 (CoQ10) in serum is serum cholesterol [19]. Thus, elevated cholesterol per se may not reduce risk, but instead may be a marker of CoQ10. Cholesterol and CoQ10 share a common pathway for synthesis [20]; and increasing cholesterol may have feedback on the production of CoQ10, either increasing or decreasing synthesis. Higher levels of CoQ10 might be beneficial by reducing oxidative stress due to H2O2 generated by dopamine metabolism [21]. CoQ10 as a therapeutic agent for PD is also under current investigation [22]. Cholesterol may have direct neuroprotective effects due to its requirement for cell membrane structure and function, and its role in myelin synthesis [23]. Cholesterol has also been found to induce the formation of synapses in cultured neurons by as much as seven times [24]. Although cholesterol in the brain is mostly from in situ synthesis and not transported from the circulation [23], it is still possible that plasma cholesterol may have some effect on brain cholesterol production or metabolism as cholesterol and lipid transport are complex. Cholesterol turnover in the brain is increased in neurodegenerative disorders, and large amounts of cholesterol turnover among both glial cells and neurons in the central nervous system during neuron repair and remodeling [25].

Possible relations between PD and oxidative stress and lipid peroxidation [4] led us to investigate various fatty acids, but our null results for fatty acids do not support these mechanistic hypotheses.

A limitation of this study is the reliance upon questionnaires, administered later in life, to elicit dietary habits throughout adult life, as there is potential for bias towards current dietary habits. This limitation is common to case–control studies of diet and has been addressed by validating the FFQ for periods as long as 20 years [26], and 24 years [27]. There is controversy regarding the period of life over which PD develops [28]. For this reason, we elicited data regarding typical dietary habits during most of adult life. The difference in time between PD onset and diagnosis should therefore be less problematic when dietary habits are considered over a wide time frame of adult life. An additional limitation of this study, as well as other retrospective studies of PD, is distinguishing the temporal relation between dietary habits and disease onset. Food choices may change due to the early manifestations of PD, possibly due to dopamine depletion [29]. Because of this, failure to take into account the temporal relations between changes in dietary habits and PD onset symptoms can be a source of bias. Our non-significant finding for iron intake in women may be due to a difference in iron status between men and women, which we could not investigate because serum iron levels were not available. In addition, BMI would also be most accurate if it had been measured rather than self-reported.

Our finding of no detectable difference in BMI between PD cases and controls is not in agreement with others who have found PD cases to lose weight of about 5lb on average in the 10 years before diagnosis [30], although our sample of subjects with BMI data is small.

An important strength of our study is the identification of a large number of incident rather than prevalent PD cases. Prevalent cases, used in most other epidemiologic studies, can introduce bias if the exposures of interest are related to the progression of the disease. In addition, our subjects were selected from a well-defined population, in an institution providing medical care to approximately 400,000 people in western Washington State.

Our finding of an increased risk of PD from a dietary intake of low cholesterol in the presence of high iron may be part of the chain of events leading to death of dopaminergic neurons. It is possible that increased cholesterol intake may increase either the level of cholesterol in the brain, or the level of the antioxidant CoQ10. Cholesterol and iron may have different mechanisms in relation to PD pathogenesis, and thus their effects would be expected to be independent. Other corroborative studies of PD risk related to dietary intakes of cholesterol and iron would add support to our observed associations. Public health recommendations arising from this and future research on dietary fats and PD need to be developed with careful consideration given to prevention of prevalent conditions, such as cardiovascular disease.