Neural correlates of early cognitive dysfunction in Parkinson's disease

Abstract

Objective: Dementia is a common and feared aspect of Parkinson's disease but there are no robust predictors of cognitive outcome. Visuoperceptual deficits are linked to risk of dementia in Parkinson's disease but whether they predict cognitive change is not known, and the neural substrates of visuoperceptual dysfunction in Parkinson's have not yet been identified.

Methods: We compared patients with Parkinson's disease and unaffected controls who underwent BOLD fMRI while performing our previously validated visuoperceptual task and tested how functional connectivity between task-specific regions and the rest of the brain differed between patients who performed well and poorly in the task.

Results: We show that task performance at baseline predicts change in cognition in Parkinson's disease after 1 year. Our task-based fMRI study showed that the performance in this task is associated with activity in the posterior cingulate cortex/precuneus. We found that functional connectivity between this region and dorsomedial prefrontal cortex was reduced in poor performers compared with good performers of this task.

Interpretation: Our findings suggest that functional connectivity is reduced between posterior and anterior hubs of the default mode network in Parkinson's patients who are likely to progress to worsening cognitive dysfunction. Our work implicates posterior default mode nodes and their connections as key brain regions in early stages of dementia in Parkinson's disease.

Figure 1

Experimental task. (A) Each trial started with a plain gray screen with a central fixation cross for 400 msec. On each trial, a different image of a cat or dog was shown that was skewed by a variable amount (four levels of skew between 0 a.u. and 2.8 a.u., defined in pilot sessions), order pseudorandomized. The image was shown for 280 msec. This was followed up by a screen with the letters C and D, indicating cat and dog. Participants indicated whether they had seen a cat or a dog by pressing a key on a response pad whilst inside the scanner. (B) Control trials with noisy images were identical in structure, but increasing amounts of visual noise were added (four levels of noise: between 0 and 1.2 a.u., defined in pilot sessions).

Figure 2

Relationship between baseline performance in the skew task and change in MoCA after 12‐month follow‐up.

Figure 3

Neural correlates of skew performance. (A) SPM showing main effect of increasing difficulty across tasks in unaffected individuals, overlaid on the mean T1 image of all participants. Threshold for display < 0.001 uncorrected. (B) SPM showing the interaction of skew task and increasing difficulty in unaffected individuals, overlaid on the mean T1 image of all participants. Threshold for display < 0.001 uncorrected. Scale bars represent T value of coordinates.

Figure 4

Neural correlates of skew performance in high‐ versus low‐performing Parkinson's patients. SPM showing greater BOLD activity in high‐ versus low‐performing Parkinson's patients in the left parietal (A) and prefrontal regions (B). Threshold for display < 0.001 uncorrected, with cluster level correction applied. Scale bar represents T value of coordinates.

Figure 5

Psychophysiological interactions. SPM showing difference in functional connectivity to seed voxel identified in Figure 3B between high‐ and low‐performing patients with Parkinson's disease. Thresholded at < 0.001 uncorrected for display, with cluster level correction applied. Scale bar represents T value of coordinates.