Impaired synaptic function is linked to cognition in Parkinson's disease

Abstract

Objective: Cognitive impairment is frequent in Parkinson's disease, but the underlying mechanisms are insufficiently understood. Because cortical metabolism is reduced in Parkinson's disease and closely associated with cognitive impairment, and CSF amyloid-β species are reduced and correlate with neuropsychological performance in Parkinson's disease, and amyloid-β release to interstitial fluid may be related to synaptic activity; we hypothesize that synapse dysfunction links cortical hypometabolism, reduced CSF amyloid-β, and presynaptic deposits of α-synuclein. We expect a correlation between hypometabolism, CSF amyloid-β, and the synapse related-markers CSF neurogranin and α-synuclein.

Methods: Thirty patients with mild-to-moderate Parkinson's disease and 26 healthy controls underwent a clinical assessment, lumbar puncture, MRI, ¹⁸F-fludeoxyglucose-PET, and a neuropsychological test battery (repeated for the patients after 2 years).

Results: All subjects had CSF amyloid-β 1-42 within normal range. In Parkinson's disease, we found strong significant correlations between cortical glucose metabolism, CSF Aβ, α-synuclein, and neurogranin. All PET CSF biomarker-based cortical clusters correlated strongly with cognitive parameters. CSF neurogranin levels were significantly lower in mild-to-moderate Parkinson's disease compared to controls, correlated with amyloid-β and α-synuclein, and with motor stage. There was little change in cognition after 2 years, but the cognitive tests that were significantly different, were also significantly associated with cortical metabolism. No such correlations were found in the control group.

Interpretation: CSF Aβ, α-synuclein, and neurogranin concentrations are related to cortical metabolism and cognitive decline. Synaptic dysfunction due to Aβ and α-synuclein dysmetabolism may be central in the evolution of cognitive impairment in Parkinson's disease.

Figure 1

Median CSF neurogranin concentrations. The median concentration of CSF neurogranin in patients with mild‐to‐moderate Parkinson's disease and healthy controls, error bars represent 95% confidence intervals. (Mean CSF neurogranin in the Parkinson's disease subjects was 251 (range: 125–556), and in healthy controls, 397 (range: 119–988).)

Figure 2

Correlations between FDG‐PET and CSF Aβ42 in Parkinson's disease. The statistical map shows two clusters in each hemisphere, one frontal (yellow) and one posterior (turquoise) cluster, in which cortical metabolism in mild‐to‐moderate Parkinson's disease is correlated with CSF Aβ42 concentrations. Mean CSF Aβ42 was 943 ng/L (range 707–1280). Cluster‐wise P = 0.0001 for all clusters. The clusters comprised 36% of the total surface area (27% of the left and 45% of the right hemisphere surface area). R ² was 0.389 in the left and 0.433 in the right combined hemisphere clusters (excluding the effects of age and gender).

Figure 3

Correlations between FDG‐PET and CSF Aβ38 in Parkinson's disease. The statistical map shows the three clusters in the left hemisphere and the two clusters in the right hemisphere in which cortical metabolism in mild‐to‐moderate Parkinson's disease is positively correlated with CSF Aβ38 concentrations. For the left hemisphere, cluster‐wise P = 0.0001 for the two mainly lateral clusters, whereas for the mainly medial (precuneal; green) cluster, cluster‐wise P = 0.0074. For the right hemisphere, cluster‐wise P = 0.0001 for the larger, posterior cluster, whereas cluster‐wise P = 0.00170 for the frontal. The clusters comprised 36% of the total surface area (38% of the left hemisphere surface area, and 33% of the right). R ² was 0.319 in the left and 0.304 in the right combined hemisphere clusters (excluding the effects of age and gender). Mean CSF Aβ38 was 1877 ng/L (range: 819–4551).

Figure 4

Correlations between FDG‐PET and CSF Aβ40 in Parkinson's disease. The statistical map shows one cluster in the left hemisphere and two clusters in the right hemisphere in which cortical metabolism in mild‐to‐moderate Parkinson's disease is positively correlated with CSF Aβ40 concentrations. For the left hemisphere cluster and the right hemisphere occipital cluster, cluster‐wise P = 0.0001. For the right hemisphere parietal cluster (shown in blue), cluster‐wise P = 0.0061. The clusters comprised 6% of the total surface area (8% of the left hemisphere surface area, and 5% of the right). R ² was 0.277 in the left and 0.250 in the right combined hemisphere clusters (excluding the effects of age and gender). Mean CSF Aβ40 was 5125 ng/L (range: 2212–10512).

Figure 5

Correlations between FDG‐PET and CSF total α‐synuclein in Parkinson's disease. The statistical map shows the clusters in each hemisphere in which cortical metabolism in mild‐to‐moderate Parkinson's disease is correlated with CSF t‐α‐syn concentrations. Mean CSF t‐α‐syn was 278 (range: 160–476). Cluster‐wise P = 0.0002 for the more frontal left clusters and the lateral right clusters, and P = 0.021 for the remaining left, and P = 0.028 for the remaining right cluster. The clusters comprised 13% of the total surface area (8% of the left and 19% of the right surface area). R ² was 0.337 in the left and 0.339 in the right combined hemisphere clusters (excluding the effects of age and gender).

Figure 6

Correlations between FDG‐PET and CSF neurogranin in Parkinson's disease. The statistical map shows the clusters in each hemisphere in which cortical metabolism in mild‐to‐moderate Parkinson's disease is correlated with CSF neurogranin concentrations. Mean CSF neurogranin was 251 (range: 125–556). Cluster‐wise P = 0.008 for the lateral left and P = 0.002 for medial left cluster, whereas P = 0.0002 for the single, large right cluster. The clusters comprised 10.5% of the total surface area (4% of the left and 16% of the right surface area). R ² was 0.330 in the left and 0.346 in the right combined hemisphere clusters (excluding the effects of age and gender).