Serum short-chain fatty acids and its correlation with motor and non-motor symptoms in Parkinson’s disease patients

PD patients were recruited from Taizhou Hospital of Zhejiang Province from July, 2020 to January, 2021. The inclusion criteria were: (1) agreement to participate in the research; (2) aged between 60 and 75 years; (3) diagnosis of PD according to Diagnostic criteria of Parkinson’s disease in China (2016 edition) [15]; (4) no use of antibiotics for three months; (5) no use of omega-3, probiotics for two weeks; (6) no use of lipid-lowering medicine for one month. The healthy controls were recruited from Health Management Center at Taizhou Hospital of Zhejiang Province from July, 2020 to January, 2021 and met the above inclusion criteria except diagnosis of PD.

Exclusion criteria were: (1) secondary parkinsonism, atypical parkinsonism, Alzheimer’s disease, cerebrovascular disease or other central nervous system diseases; (2) celiac disease; (3) chronic pancreatitis; (4) history of gastrointestinal surgery; (5) inflammatory bowel disease; and (6) history of cancer within three years.

All participants provided demographics of age, sex, smoke, alcohol consumption, hypertension, diabetes mellitus, liver function index, lipid profiles, and medical history. For all patients, the following data were recorded: disease duration from onset to study, Hoehn-Yahr stage, motor symptom related-Unified Parkinson’s Disease Rating Scale (UPDRS) part III score, motor complications (end-of-dose phenomenon, dyskinesia, and freezing), non-motor symptoms (cognitive impairment, anxiety, depression, paresthesia, dysautonomia, sleep disorders and rapid-eye-movement sleep behavior disorder (RBD)), medication usage and Levodopa-equivalent daily dose (LEDD). The UPDRS scores were rated during the on state. Motor complications and paresthesia were evaluated by clinical face-to-face interviews with patients. Sleep disorders was evaluated according to Parkinson disease sleep scale. Autonomic function was evaluated using the Scale for Outcomes in Parkinson’s Disease-Autonomic Dysfunction. Patients that often had symptoms related to postural changes, urinary dysfunction caused by non-primary or secondary causes of the urinary system, or constipation, were considered to have dysautonomia. The cognitive state was evaluated using Mini-mental State Examination (MMSE) scale. The anxiety and depression states were evaluated using Hamilton Anxiety Scale (HAMA) and Hamilton Depression Scale (HAMD), respectively. RBD was diagnosed with REM Sleep Behavior Disorder Questionnaire Hong Kong (RBDQ-HK). All the assessments were performed by two experienced physicians specialized in movement disorders (Suzhi Liu and Yajing Wang, with 21 and 10 years of work experience on Parkinson’s Disease, respectively). LEDD was calculated according to the reference [16].

Blood was sampled into serum-separating tubes by professional nurses in the morning after overnight fasting. After 30 min at the room temperature, the blood samples were centrifuged at 4000 rpm for 5 min at 4 °C. The serum was collected into a 1.5 mL Eppendorf tube and stored at −80 °C until assay.

Concentrations of serum SCFAs (including heptanoic acid) were analyzed using a gas chromatography-mass spectrometer (GC-MS). Serum samples were deproteinized by phosphoric acid and extracted with ether. After being centrifuged at 4000 rpm for 10 min, the supernatant was collected and injected to GC-MS. An Agilent 7890B-7000D GC-MS was fitted with a capillary column HP-INNOWAX 25 m × 0.20 mm × 0.4 μm. The injector temperature was 240 °C, and the carrier gas flow rate was set to 1 mL/min. The ion source and transmission line temperatures were 200 and 250 °C, respectively. The electron bombarding voltage was 70 eV, and single ion monitoring was applied.

Continuous variables were expressed as mean ± Std.Error and compared using student’s t-test if data were normally distributed or otherwise, expressed as median (IQR) and analyzed using Mann-Whitney U test. The Kolmogorov-Smirnov test was used to evaluate the normality of the distribution of the variables. Categorical variables were expressed as number (%) and compared by χ2 test or Fisher’s exact test. A correlation was investigated using Spearman nonparametric correlation analysis method. Statistical significance values were set at α ≤ 0.05 (two-sided), and correlation magnitude levels were defined as weak (<0.3), moderate (0.3–0.59), and strong (≥0.6). Data analysis was conducted using SPSS software, version 16.0 (SPSS Inc.). Multiple testing of the nine SCFAs between PD and controls was then corrected by Bonferroni method. The potential confounding variables predictive of SCFAs was determined by multiple linear regression using stepwise method.

50 PD patients and 50 normal controls were recruited in this study. The main clinical characteristics and laboratory results are described in Table 1. No difference was observed in age and gender between PD patients and controls (P = 0.450 and 0.675 respectively). No difference existed in the smoke and alcohol consumption rate between PD patients and controls (P = 0.269 and 0.359 respectively). PD and controls exhibited the same rate of hypertension and diabetes mellitus comorbidity (both P = 1.000). The two groups exhibited no difference in laboratory characteristics, including bilirubin, creatinine, total cholesterol, triglycerides, or high-density lipoprotein cholesterol. However, low-density lipoprotein cholesterol in PD group was significantly lower than in the control group (mean ± Std.Error, PD vs. control, 2.91 ± 0.09 vs 2.47 ± 0.12, P = 0.005) (Table 1).

PD patients had a disease duration of 51.5(48) (median (IQR)) months and the Hoehn-Yahr stage was 2.5(0.5). The UPDRS part III score was 42.82 ± 2.20 (mean ± Std.Error) and MMSE score was 20.24 ± 0.73 (mean ± Std.Error). In addition, patients with longer disease duration appeared to have higher UPDRS part III score (R = 0.494, P = 0.000) (Supplementary Fig. 1). The prevalence of motor complications such as end-of-dose phenomenon, dyskinesia, and freezing was described. The prevalence of non-motor symptoms, including anxiety, depression, paresthesia, dysautonomia, sleep disorders and RBD in PD patients was calculated. The usage of anti-Parkinson’s agent was also calculated and LEDD is 555.5 ± 38.9 mg (Table 1).

The levels of serum propionic, butyric and caproic acids were significantly lower in PD group than in control group after correction by Bonferroni method (P < 0.0056 (0.05/9)). Heptanoic acids were significantly higher in PD patients than controls (P < 0.0056). No differences were observed between the two groups in pentanoic acid (P = 0.008), isobutyric acid (P = 0.019), acetic acid (P = 0.072), isovaleric acid (P = 0.450), or isocaproic acid (P = 0.937) (Fig. 1).

Although demographic variables of PD and control group were similar, multiple linear regression using stepwise method was applied to determine the explanatory variables predictive of SCFAs. The model showed that: 1) PD (P < 0.000), smoke (P = 0.022), triglycerides (P = 0.019) and creatinine (P = 0.047) were significant predictors for the decrease of propionic acid, accounting for 39.6% (P = 0.000) of the variance indicated by the adjusted R-squared; 2) PD (P < 0.000), alcohol consumption (P < 0.001) and low-density lipoprotein cholesterol (P = 0.044) were significant predictors for the decrease of butyric acid, accounting for 42.3% (P = 0.000) of the variance; 3) PD (P = 0.005), alcohol consumption (P = 0.005) and high-density lipoprotein cholesterol (P = 0.038) were significant predictors for the decrease of caproic acid, accounting for 15.2% (P = 0.000) of the variance; 4) PD (P = 0.021) was also a significant predictor for the increase of heptanoic acid, accounting for 4.3% (P = 0.021) of the variance.

In order to evaluate the relationship between SCFAs and PD progression, the contents of SCFAs in different Hoehn-Yahr stages were analyzed. As there were only 2 and 1 patient in stage 1 and 4 respectively, patients at stage 1 and stage 2, and patients at stage 3 and stage 4 were pooled in the analysis. Between these two pooled groups of patients (stage 1–2 vs. 3–4), the levels of serum SCFAs were not significantly different (p > 0.05). The motor symptom in PD was evaluated by UPDRS part III. We observed that only the serum level of propionic acid was correlated with UPDRS part III scores (Supplementary Table 1). The correlation coefficient of propionic acid with UPDRS part III scores was −0.365, and P-value was 0.009, indicating moderate correlation magnitude levels (Fig. 2A).

After we observed a correlation between serum SCFA and motor symptoms in PD patients, we analyzed the potential relation between SCFAs and non-motor symptoms. We observed that only the level of serum propionic acid was correlated with cognitive impairment and depression. As shown in Fig. 2, B and C, serum propionic acid was negatively correlated with MMSE score (R = -0.416, P = 0.003), but positively correlated with HAMD score (R = 0.306, P = 0.03). Notably, HAMD score was negatively correlated with MMSE score (R = -0.375, P = 0.007) (Supplementary Fig. 2), implying that depressed patients exhibit cognitive impairment or vice versa.

No correlation was observed between SCFAs and motor complications. The serum SCFAs concentration was comparable in patients with and without motor complications (data not shown).

All PD patients were administered with anti-PD drug. To investigate the influence of anti-Parkinson medication on peripheral blood SCFAs, the serum SCFAs concentration was compared between people who received a specific anti-Parkinson medication and those who did not receive it. The results showed that the serum SCFAs appeared not to be influenced by levodopa administration as correlation between SCFAs and LEDD was not observed. However, serum propionic acid concentration was significantly higher in PD patients taking trihexyphenidyl (n = 14) (P = 0.028) (Fig. 3A) or tizanidine (n = 16) (P = 0.032) (Fig. 3B). Other anti-Parkinson medication’s influence on serum SCFAs concentration was not observed.