Alpha-synuclein measured in cerebrospinal fluid from patients with Alzheimer’s disease, mild cognitive impairment, or healthy controls: a two year follow-up study

The present study included a total of 74 patients and 25 controls enrolled in a two-year follow-up study from the summer of 2009. All subjects were ethnic Norwegians between 53 and 78 years of age and recruited through the Department of Neurology, University Hospital of Trondheim, by two neurologists (SBS or GRG). Neurological examination was performed on both patient and control groups, including the Mini Mental State Examination (MMSE) [29] and the Ten-Word Test with delayed recall as taken from ADAS Cog [30], as well as cerebral MRI at 3 T at baseline and after two years. APOE genotyping was performed on blood samples according to the method described elsewhere [31]. General exclusion criteria were insufficient sight and hearing to complete the cognitive testing, a present psychiatric or malignant disease, use of anti-coagulating medication, or high alcohol consumption.

At inclusion 26 patients were diagnosed with early AD (National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (NINCDS-ADRDA) criteria [32]), and 48 with aMCI (The International Working Group on Mild Cognitive Impairment Criteria [33]). Patients returned to the hospital one and two years after inclusion, with lumbar puncture and neuropsychological tests performed on each occasion. A few patients missed certain time-points of lumbar puncture. During follow-up, 23 of the aMCI patients kept their initial diagnosis (designated MCI-MCI), while 25 developed AD (designated MCI-AD). Patients diagnosed with AD from baseline comprised the AD-AD group. These three patient groups were compared in the statistical analyses. A detailed overview of the clinical work-up is described elsewhere [34].

Twenty-five healthy elderly volunteers were recruited as controls from societies for retired people in central Norway, or caregivers not genetically related to the patients. Lumbar puncture was performed at inclusion, whereas neuropsychological examination was performed at both inclusion and after two years. Control individuals were assessed as being healthy for their age without signs of neurological disorders, and without first-degree relatives with dementia.

Patients and healthy controls underwent lumbar puncture, usually at the L4/L5 intervertebral space, and CSF samples were obtained as described previously [34]. Briefly, 1 mL aliquots of CSF were collected directly into Corning 2 mL polypropylene cryovials immersed in ice-water, and frozen within 30 min of lumbar puncture and stored at −80 °C [35, 36]. Only three CSF samples from patients were centrifuged prior to freezing due to contamination from blood. Erythrocyte counts in the remaining samples were (mean ± SD) 1.6 ± 2.8/μL (overall range 0–17 erythrocytes/μL). The albumin quotient (Q_A = albumin in CSF/albumin in plasma) [37] was calculated for all samples. All samples were transferred to −20 °C approximately one week prior to analysis, and were slowly thawed in ice-water on the morning of analysis. The CSF samples were analyzed by commercially-available enzyme-linked immunosorbent assay (ELISA) monoplex kits; α-synuclein (Invitrogen: cat.no KHB0061), Aβ1-42 (Innogenetics: Aβ42, cat. no. 80324), total tau protein (t-tau, cat. no. 80323), 181-phosphorylated tau (p-tau, cat. no. 80317), and Aβ-40 (IBL International: cat. no. 81015). All assays were run according to the manufacturers’ instructions. Patient and healthy control samples were run on the same plates, in duplicate and averaged. An internal control was run on five out of seven α-synuclein ELISA plates, giving an inter-assay CV of 10.2 %, while the intra-assay CV was 2.9 % (n = 7). CV values for the core biomarkers are given elsewhere [34], but biomarker results were unknown at the time of clinical diagnosis. None of the individuals were excluded from the study due to increased hemoglobin levels in the samples, which is known to be a confounding factor in α-synuclein quantification in CSF, resulting in artificially increased values [18].

The four study groups (AD-AD, MCI-AD, MCI-MCI, healthy control individuals) were compared with respect to categorical demographic and clinical characteristics at the time of inclusion by means of the chi-square tests (overall and pairwaise comparisons). One-way ANOVA was conducted to compare characteristics measured on a continuous scale, and also mean baseline values of α-synuclein, Aβ42, Aβ40, t-tau and p-tau. In pairwise comparison between groups, the least significant difference (LSD) procedure was applied, adjusting the nominal significance level to 1 %. A linear mixed model with subject as random factor, and with interaction terms added, was applied to compare mean changes in α-synuclein levels during follow-up (measurements at inclusion, and after 1 and 2 years, respectively) between the three patients groups. Both non-linear (categorical) and linear (per year) time patterns were considered. Moreover, linear associations between α-synuclein and other biomarkers (Aβ42, Aβ40, t-tau, p-tau) were examined with Pearson’s correlation coefficient and linear regression analysis. The determination coefficient (R²) is reported as the measure of percent explained variance in the linear regression model.

One patient from the MCI-AD group had a considerably elevated CSF α-synuclein concentration at baseline, and another at two years (though both were in the range of the kit). The albumin quotient for these two patients (and patients in general), did not indicate a disturbance of the blood–brain barrier [37] (data not shown). The analysis of changes in α-synuclein was first carried out with the two extreme values excluded. Mean baseline concentrations and standard deviations of α-synuclein in all study groups, and mean values after 1 and 2 years in the three patient groups are shown in Table 2a. The highest baseline mean level of α-synuclein measured in CSF was found in the MCI-AD group, with the other groups having approximately similar mean levels, lowest for the AD-AD group (p = 0.26, test for overall difference between groups). The individual variation was large in all study groups, but the standard deviation was larger in the two MCI groups than in the AD-AD and the control groups (Table 2a). The difference in mean α-synuclein baseline level became more pronounced when the two extreme values were included in the analysis (Additional file 1: Table S1a). The overall test for difference between the four study groups was then closer to statistical significance (p = 0.08), and the test for difference between the MCI-AD and the AD-AD groups was significant at the 5 % level, but not at the 1 % level (p = 0.019).

Individual time patterns of α-synuclein levels in the three patient groups are shown in Fig. 1a-c. Most patients had levels between 200–750 pg/mL. Except for one patient in the AD-AD group, values above 1000 pg/mL occurred only in the MCI-MCI and MCI-AD groups. A decrease in α-synuclein during the first year of follow-up was seen in all groups (extreme values excluded), but was more pronounced in the MCI-AD group than in the AD-AD and MCI-MCI groups (Table 2b, mean reduction of 16.7 vs. 11.4 and 9.0 pg/mL, respectively). In the MCI-AD group, α-synuclein values decreased also in the second year, in contrast to the pattern seen in the AD-AD and MCI-MCI groups (Table 2b). However, no significant changes in α-synuclein over time were seen within any of the groups, and the mean reduction over time did not differ between the groups (p = 0.93 and p = 0.67, test for categorical and linear interaction, respectively). Moreover, the mean α-synuclein levels did not differ significantly between the groups either after one year (p = 0.23) or after two years (p = 0.33). The mean α-synuclein values predicted on the basis of the linear mixed model fitted observed data rather well (Fig. 1d). In the analyses with the two extreme values included, the mean reduction in α-synuclein levels over time in the MCI-AD group was considerably more pronounced (Additional file 1: Table S1b). Nevertheless, no significant differences were found, either within or between the three patient groups.

Patients were included in the present study between one and seven years after onset of the first symptom related to aMCI or AD. In view of the large variation in individual α-synuclein levels at baseline, the mean α-synuclein level at time of inclusion in the study was compared between groups defined by duration of symptoms prior to inclusion (1–2 years vs. ≥ 3 years). In the MCI-AD group, those with the shortest duration of symptoms had significantly higher mean α-synuclein CSF concentration (802.2 pg/mL, confidence interval: 553.9 - 1051.5 pg/mL) than patients in the group with longer duration (442.8 pg/mL, confidence interval: 320.3 - 565.4 pg/mL), two sample t-test, p = 0.01, extreme value at baseline excluded. No significant differences were found according to duration of symptoms in the MCI-MCI or AD-AD groups.

In CSF samples from healthy control individuals, Aβ40, t-tau and p-tau all showed strong linear associations with CSF α-synuclein (determination coefficient R² in range 57.3-61.8 %, p < 0.0005) whereas the association with Aβ42 was weaker (R² of 24.7 %). In a combined analysis of all patient groups, the association between baseline CSF α-synuclein and Aβ40 was similar to that observed in healthy individuals (Fig. 2b), whereas the strength of linear association with t-tau and p-tau was slightly weaker (Fig. 2c and d, R² = 41.6 % and 42.3 %, respectively, p < 0.0005). No significant association was found between Aβ42 and α-synuclein (p = 0.69), Fig. 2a. In a model with mutual adjustment for all these biomarkers, all factors together explained 67.3 % of the variation in α-synuclein, but only Aβ40 and p-tau contributed significantly. The direction of linear association between each biomarker and α-synuclein was consistently observed within all four study groups, except that Aβ42 was negatively correlated in the MCI-AD group (Pearson’s correlation coefficient −0.325) and positively correlated in the other study groups (Pearson’s correlation coefficient in range 0.075 – 0.499). The opposite direction of association with Aβ42 in the patient groups probably explains the lack of association in combined analysis of all patients.

The mean concentration of α-synuclein did not differ significantly between APOE ɛ4 carriers and non-carriers in any group (Additional file 2 Table S2), though a more pronounced difference was detected in the MCI-AD group. In this patient group, carriers of APOE ɛ4 had a mean CSF α-synuclein level of 887.1 ± 692.9 pg/mL whereas non-carriers had a mean level of 442.3 ± 181.1 pg/mL. However, the difference was not statistically significant.