Dietary Intakes and Plasma Polyunsaturated Fatty Acid Levels in Parkinson’s Disease


 The study aims to determine the correlation between dietary intakes and plasma concentrations of PUFA and their associations with clinical severities in PD. We designed a case-control study to assess dietary fat intakes and disease severity in 38 PD patients and 33 healthy volunteers. Plasma levels of five PUFAs including α-linolenic acid (ALA), eicosapentaenoic acid, docosahexaenoic acid, linoleic acid (LA), and arachidonic acid (AA) were measured using gas chromatography. No differences were observed in dietary total energy and lipid intakes including PUFAs between two groups. PD patients had lower plasma levels of ALA, LA, and AA than controls. The associations between dietary intake and plasma concentrations of PUFAs were insignificant, whereas the association between dietary intake and plasma concentration of AA was exceptionally significant in controls. Plasma levels of ALA and LA were inversely associated with motor rating scores in PD patients, while that of AA was positively associated with non-motor symptoms.Plasma levels of ALA, LA, and AA were lower in PD patients than those in healthy controls, suggesting absorption problems in PD or expedited utilization of PUFAs. Contradictory relations on motor and non-motor symptoms imply conflicting clinical effects of PUFAs, raising questions about the usefulness of PUFA supplementation.


Introduction
Parkinson's disease (PD) is the second most common neurodegenerative disease affecting approximately 1% of people aged 65 years or more. Environmental factors are important in the pathogenesis of PD, speci cally in patients older than 50 years 1 . Daily food intake is consistently studied as an important environmental factor 2 .
Regarding fatty acid, the role of polyunsaturated fatty acid (PUFA) has been widely studied. In animal studies, omega-3 (ω-3) PUFA has the following bene cial effects: it protects nigral dopamine neurons, mitigates motor severity, or delays the development of levodopa-induced dyskinesia 3,4 . On the contrary, abundant PUFA in neural plasma membranes could be sources of oxygen radicals through lipid peroxidation 5 and could promote α-synuclein (αSyn) oligomerization with further aggregation according to in vitro and in vivo mouse models 6,7 . Clinical studies assessing the associations between PUFA intake and PD risk also remained controversial 8-11 , while direct analyses of lipid metabolites showed decreased PUFA level in PD 12,13 . Although a few clinical trials using ω-3 PUFA supplementations reported clinical bene ts of PUFA, the number of patients was small or vitamin E was coadministered in these trials 14,15 .
Previously, either food questionnaire or direct measurement of PUFA was adapted to investigate the role of PUFA in PD, which was insu cient to reveal the association between dietary habit and plasma PUFA level. In this study, we simultaneously examined the dietary fat intake and plasma concentration of PUFA, allowing us to determine the factors associated with plasma PUFA.

Results
Demographic factors, co-morbidity, and other lipid levels were previously described 16 , which were not different between PD patients and controls, while WCS and K-NMSS score were signi cantly higher in PD patients than those in controls (Table 1).
There were no signi cant differences in dietary intakes including total energy, total fat, cholesterol, and total fatty acids, saturated fatty acids, monounsaturated fatty acids, and PUFA between PD patients and controls (Table 2). The ratio of ω-6 to ω-3 PUFA (ω-6/ ω-3 ratio) based on dietary intakes was not different between the two groups.
The plasma levels of α-linolenic acid (ALA, C18:3n3) (p=0.017), linoleic acid (LA, C18:2n6) (p=0.003), and arachidonic acid (AA, C20:4n6) (p=0.024) were lower in PD patients than those in controls (Table 3 and Figure 1). The levels of eicosapentaenoic acid (EPA, C20:5n3) and docosahexaenoic acid (DHA, C22:6n3) were lower in PD patients compared to controls, but the difference between the two groups was insigni cant. The ω-6/ω-3 ratio in the plasma was insigni cantly different between PD patients and controls. The plasma level of LA was lower in patients taking catechol-O-methyltransferase inhibitor than that in patients not taking There were no signi cant associations between dietary intakes and plasma levels of PUFA in PD patients (Supplementary Figure   S1). In the control group, AA intake was associated with plasma concentrations (ρ=0.424, p=0.016).
We set outliers as any measures that are less than 1.5 interquartile range (IQR) below the lower quartile or more than 1.5 IQR above the upper quartile. When outliers of both dietary intakes and plasma levels of PUFA were removed and analyzed, the comparison results and statistical signi cance did not change between PD patients and healthy controls (data not shown). However, the strength of association between diet-plasma PUFA measures was weakened in the control group, when outliers were excluded for analysis (ρ=0.346, p=0.071). The correlation between plasma PUFA concentration and clinical parameters also did not change, but the positive correlation between plasma AA level and UPDRS I was no longer statistically signi cant (ρ=0.155, p=0.382).

Discussion
In this study, we examined ve essential PUFAs based on the subjects' history of dietary intakes and plasma levels in PD.
Previous epidemiological studies investigated the role of PUFAs on the risk of PD, showing controversial results [8][9][10][11]17 . A recent review summarized that the risk of PD was negatively associated with the intakes of PUFA, ω-3 PUFA, or ALA and ω-3/ω-6 ratio, while it was positively associated with the intakes of AA and ω-6 PUFA 18 . In our study, diet history of PUFA intake and calculated ω-6/ω-3 ratio were insigni cantly different between PD patients and controls. Although we did not intend to assess the risk of PD associated with PUFAs, our ndings were inconsistent with those of previous diet studies.
In this study, we recruited patients with established diagnosis of PD. Considering that the mean disease duration of PD patients was 4 years, food consumption behavior could be modi ed by seeking "healthier" foods after the diagnosis, blunting unhealthy pattern during preclinical or early stage of the disease. Food behavior change could be indirectly conjectured by a study showing that the larger portion of PD patients (76%) in Korea took complementary or alternative medicine than that in Western countries 19,20 .
Previously, changes in the pattern of food intakes before and after the diagnosis of PD are not fully studied, except the increased total amount of energy intakes after PD diagnosis 21 . Further studies are required to assess the in uence of PD diagnosis on the change of the pattern of food intake.
Only a couple of studies directly measured plasma levels of PUFA 12,13 . Plasma levels of PUFAs also decreased in our PD patients, while the levels of other lipids were not different between PD patients and controls (Tables 1 and 3) 12,13 . Previous studies reported decreased PUFAs as a part of pro ling serum metabolites and did not pay attention to diet history. The discrepancy between dietary intake and plasma concentrations of PUFA was unexpected and rst reported in PD.
Absorption and utilization of PUFAs should be considered when assessing the association between plasma PUFAs and PD. First, PUFAs are absorbed in the small intestines; thus, small intestinal bacterial overgrowth (SIBO) can hinder PUFA absorption 22,23 . SIBO was frequently reported in PD patients and had negative effects on clinical features 24 . In ammation or deranged osmolarity by the bacteria may hamper normal absorption 22,24 . In this study, a difference in dietary intakes of PUFAs between PD patients and controls was not observed, and the positive association between dietary intakes and plasma levels of AA in the control group suggested that malabsorption was the cause of decreased PUFAs in PD. Second, PUFAs are utilized by neurons, and expedited utilization could be hypothesized in association with pathologic change of local milieu such as oxidative stress in PD. Fatty acidbinding protein (FABP) is a protein translocating PUFA from the extracellular space inside the neurons. In PD, it was reported that FABP expression was elevated in the substantia nigra (SN) 25 . This upregulation is understandable because the SN is highly energydemanding and resultantly vulnerable to oxidative stress, and PUFAs have protective mechanisms against oxidative stress. However, previous studies reported different levels of PUFAs in the brain. In the SN, PUFA level decreased along with lipid peroxidation, engendering harmful effects on the neurons 5 . In other studies, levels of PUFAs increased in the frontal cortices of patients with Lewy body disease such as PD, and PUFA was also associated with αSyn oligomerization 6 . Therefore, the effect of PUFAs on the neurons is double-edged, ful lling the demand of endangered neurons under oxidative stress but threatening neurons by lipid peroxidation and αSyn oligomerization.
In clinical analyses, the plasma levels of PUFAs were negatively associated with the severity of motor symptoms, implying a bene cial effect of PUFAs. Previous studies using ω-3 PUFAs produced inconsistent results, whereas clinical studies using ω-6 PUFAs were not conducted because of the unfavorable results obtained regarding the association between ω-6 PUFAs and PD risk 10, 11,18 . In our study, both ω-3 (ALA) and ω-6 (LA) PUFAs had negative associations with parkinsonian motor symptoms, while another ω-6 (AA) had a positive association with the severity of nonmotor symptoms. Therefore, the clinical bene t of PUFAs on PD is not straightforward.
There are several limitations to be considered. First, since the present study is a cross-sectional case-control study on PD patients, it is di cult to draw the causal interferences of PUFA on PD. Second, obtaining diet history can be affected by non-standardized method. In this study, nutritionists interviewed all the participants using standard FFQ of which the reliability and validity was established 26,27 . Finally, there is an issue related to the reliability of a single cross-sectional measurement of PUFA levels. In this study, to prevent possible dietary effect, we asked the participants to fast more than 12 hours before blood sampling. And we simultaneously measured the levels of non-PUFA lipid to ensure the reliability of PUFA measurement, which were not different between PD patients and controls. Although we did not repeat the measurement of PUFA levels, repeated measurement of PUFA levels over 2-3 years showed consistent results in women from Nurses' Health Studies 28 .
In conclusion, the simultaneous assessment of food intake history and plasma PUFA is rstly done to show possible absorption problem in PD. Overconsumption of PUFAs by vulnerable neurons with both bene cial and detrimental effect in PD is also conceivable. Finally, con icting associations between PUFAs and clinical symptoms raise questions about the bene t of PUFA supplementation in PD.

Subjects
Thirty-eight PD patients and 33 controls aged between 50 and 75 years were included. Patients were diagnosed with PD according to the United Kingdom Parkinson's Disease Society Brain Bank Clinical Diagnosis Criteria 29 . The inclusion criterion was as follows: patients not receiving antibiotics, immune-related drugs, and vaccines (for 3 months, respectively), lipid-lowering drugs (for 1 month), and vitamin supplements, ω-3 fatty acids, prebiotics, and probiotics (for 2 weeks, respectively) 16 . The exclusion criterion was as follows: patients with secondary parkinsonism, other signi cant central nervous system diseases, malignancy within the last 3 years, and signi cant gastrointestinal disorders 16 . The study has been approved by local ethics committee (Institutional Review Boards of Kyung Hee University Hospital, # KHUH 2017-08-035) and all subjects provided written informed consents before inclusion in the study in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki).

Clinical and Nutritional Evaluation
All subjects were interviewed by a trained interviewer, using a structured questionnaire to obtain information on demographics, height, weight, and medical history (disease duration and medications). We used a validated 138-item semiquantitative food frequency questionnaire (FFQ) to assess usual dietary intake over the past 12 months, which were selected based on commonly were used for the subjects' clinical evaluation. Levodopa-equivalent daily dose was calculated based on the previous reference 30 .

Sample Collection and Plasma Lipid Analysis
Blood samples were collected in the morning after the subjects were required to fast over 12 hours. We measured the levels of cholesterol, triglycerides, high-and low-density lipoprotein, and PUFA including ALA, EPA, DHA, LA, and AA.
All blood samples were prepared in a nonadditive blood collecting tube by adding 1 mg/mL of ethylenediaminetetraacetic acid and 1 mg/mL of reduced glutathione. Plasma was separated from the whole blood by centrifugation at 3000 rpm for 10 minutes immediately. The plasma was treated with 10 µl/ml of butylated hydroxytoluene-methanol and stored at − 80℃.
For PUFA analysis, the samples were dissolved in methanol containing methyl heptadecanoate (3 g/L) as an internal standard. Acetyl chloride was added to the dissolved samples for acid-catalyzed esteri cation and transesteri cation, and incubated for 45 minutes. For neutralization, potassium carbonate solution (6% in water) was mixed gently to the samples with hexane and subsequently placed on ice for 30 minutes. After cooling, tubes were centrifuged for 10 minutes at 3000 rpm. The extracted lipids of the supernatant were dried under nitrogen gas and washed with hexane by solvent.
The solution was injected by pulsed split mode into an Agilent 7890B gas chromatograph with a ame ionization detector using Agilent DB-WAX capillary column (30 m × 0.25 µm × 0.25 µm). The inlet and detector temperatures were 250℃ and 270℃, respectively, and nitrogen was used as a carrier gas at a rate of 1 mL/min. The concentration of long chain fatty acid was calculated using the peak height ratio method 31 .

Data and Statistical Analyses
We applied Student's t-test and Mann-Whitney U test to compare the means of two independent continuous variables according to assumption of normal distribution. Categorical data were compared by Pearson's chi-squared test analysis. Spearman's rank correlation coe cient analysis was applied to determine bivariate correlation. The Statistical Package for the Social Sciences (SPSS) software version 25.0 (SPSS Inc., Chicago, IL, USA) was used for statistical analysis with p < 0.05 considered as statistically signi cant (two-tailed).

Data availability
The datasets collected and analyzed for the current study are available from the corresponding author on reasonable request.