Aberrant functional connectivity within the basal ganglia of patients with Parkinson's disease

Abstract

Resting state functional MRI (rs-fMRI) has been previously shown to be a promising tool for the assessment of early Parkinson's disease (PD). In order to assess whether changes within the basal ganglia network (BGN) are disease specific or relate to neurodegeneration generally, BGN connectivity was assessed in 32 patients with early PD, 19 healthy controls and 31 patients with Alzheimer's disease (AD). Voxel-wise comparisons demonstrated decreased connectivity within the basal ganglia of patients with PD, when compared to patients with AD and healthy controls. No significant changes within the BGN were seen in AD, when compared to healthy controls. Moreover, measures of functional connectivity extracted from regions within the basal ganglia were significantly lower in the PD group. Consistent with previous radiotracer studies, the greatest change when compared to the healthy control group was seen in the posterior putamen of PD subjects. When combined into a single component score, this method differentiated PD from AD and healthy control subjects, with a diagnostic accuracy of 81%. Rs-fMRI can be used to demonstrate the aberrant functional connectivity within the basal ganglia of patients with early PD. These changes are likely to be representative of patho-physiological basal ganglia dysfunction and are not associated with generalised neurodegeneration seen in AD. Further studies are necessary to ascertain whether this method is sensitive enough to detect basal ganglia dysfunction in prodromal PD, and its utility as a potential diagnostic biomarker for premotor and early motoric disease.

Keywords: Alzheimer's disease; Basal ganglia; Functional connectivity; Parkinson's disease; Parkinsonism; fMRI.

Fig. 1

An unbiased template of the resting state basal ganglia network (BGN) generated from 80 healthy controls (Szewczyk-Krolikowski et al., 2014a). The spatial map is thresholded at z > 2.6.

Fig. 2

Voxel-wise analysis confined to the basal ganglia showing decreased connectivity in PD patients compared to a) healthy controls; b) patients with Alzheimer's disease; and c) patients with Alzheimer's disease and healthy controls, combined. Results show significantly reduced connectivity in PD in all regions of the basal ganglia. In image c, voxels showing reduced connectivity make up 68.9%, 99.0% 98.3% and 99.7% of the total volume of the caudate, pallidum, anterior putamen and the posterior putamen, respectively.Images are displayed in radiological convention (right is left). Clusters are thresholded at p > 0.05 after FWE correction.

Fig. 3

Regions of interest analysis. Average of the mean P.E. extracted bilaterally from the regions of interest, expressed as a proportion of the mean P.E. for that region in healthy controls (mean ± standard error of the mean). Results demonstrate significantly reduced P.E. in all regions in PD, compared to healthy controls and patients with Alzheimer's disease. HC, healthy controls; AD, Alzheimer's disease; PD, Parkinson's disease; P.E., parameter estimates; ant. Puta, anterior putamen; post. Puta, posterior putamen.

Fig. 4

The diagnostic utility of resting state connectivity. a) Mean component scores were calculated using a data-driven principal component analysis-based dimensionality reduction method by combining the mean parameter estimates from each of the regions of interest. Results demonstrate a significantly lower score in patients with PD, when compared to patients with Alzheimer's disease and healthy controls. ***p < 0.0001. b) Individual component scores were used to generate receiver operating characteristic (ROC) curves. Area under the curve was 0.80 (95% C.I. 0.68–0.93), 0.81 (95% C.I. 0.71–0.92) and 0.81 (95% C.I. 0.71–0.90) for the PD vs HC, PD vs AD and PD vs All comparisons, respectively. HC, healthy controls; AD, Alzheimer's disease; PD, Parkinson's disease; All, group comprised of healthy controls and patients with Alzheimer's disease.

Supplementary Fig. 1

Group-level map of the BGN for each study group. Images are displayed in radiological convention (right is left). Clusters are thresholded at p > 0.05 after FWE correction.