Search for a command to run...
Spatial covariance analysis has been used with (18)F-fluorodeoxyglucose (FDG) PET to detect and quantify specific metabolic patterns associated with Parkinson's disease (PD). However, PD-related patterns cannot necessarily serve as biomarkers of the processes that underlie the atypical parkinsonian syndromes. In this FDG PET study, we used strictly defined statistical criteria to identify disease-related metabolic patterns in the imaging data from patients with multiple system atrophy (MSA) and progressive supranuclear palsy (PSP), the two most common of these atypical conditions. We found that MSA and PSP were each associated with a specific, highly stable metabolic brain network (P < 0.0001, bootstrap estimation). The MSA-related pattern was characterized by decreased metabolism in the putamen and cerebellum. The PSP-related pattern was characterized by metabolic decreases in the brainstem and medial frontal cortex. For both conditions, pattern expression was significantly elevated in patients relative to age-matched healthy control subjects (P < 0.001). For each condition, we validated the associated disease-related metabolic pattern by computing its expression on an individual scan basis in two independent patient cohorts, and in one subsequent healthy volunteer cohort. We found that for both MSA and PSP, prospective assessments of pattern expression accurately discriminated patients from controls (P < 0.001). These findings suggest that the major atypical parkinsonian syndromes are associated with distinct patterns of abnormal regional metabolic activity. These disease-related networks can potentially be used in conjunction with functional brain imaging as quantifiable biomarkers for the assessment of these pathological conditions.