Introduction
It is often a challenge to diagnose Parkinson's disease (PD) and atypical Parkinsonian disorders solely on clinical criteria. Image analysis using statistical mapping can improve early differential diagnosis of PD from atypical syndromes such as progressive supranuclear palsy (PSP) and multiple system atrophy (MSA). Using brain network analysis and [18F]fluorodeoxyglucose (FDG) PET data, we have previously generated a PD-related covariance pattern (PDRP) to describe its unique abnormal metabolism. The expression of this network in individual patients is a sensitive imaging marker to assess the disease severity and progression, as well as to predict treatment outcome following medical and neurosurgical interventions. However we have found that PDRP is not expressed in patients with PSP, suggesting the existence of a specific metabolic topography of its own. In this study we sought to determine whether specific abnormal metabolic brain topography is associated with patients with PSP.
Methods
We used FDG scans from 20 PSP patients whose diagnosis was clinically confirmed at follow-up (age = 69 ± 9 years; disease duration = 2.9 ± 1.2 years) and 22 age-matched normal controls (age = 56 ± 11 years). Imaging studies were done on a GE Advance PET camera in 3D mode. All patients were scanned at least 12 hours off medications. Images of cerebral glucose metabolic rates were generated and transformed into a standard brain space. By examining spatial covariance based on principal component analysis, we performed calculations in the combined brain scans to identify principal components whose expressions can significantly separate PSP and control subjects. The specificity of this PSP network was then evaluated by computing prospectively its expression in PD and MSA patients with the same mean age and disease course.
Results
Network analysis disclosed a significant PSP-related metabolic covariance pattern (PSP-RP) discriminating PSP patients from normal controls (p < 0.0001). This characteristic pattern showed hypermetabolism in the thalamus, parietal, occipital, temporal, lateral frontal and motor cortices. Concurrently, hypometabolism was seen in the caudate and brainstem, as well as in the midline frontal cortex and insula. PSP-RP expressions in the PD and MSA cohorts were significantly lower than that in the PSP patients and did not differ from the normal group.
Conclusion
This study reveals unique metabolic brain topography that is present in PSP but absent in PD and MSA patients. It can serve as a disease marker to help accurate diagnosis of PSP by clinicians. The quantification in single patient will be useful in differentiating PSP from early stage PD and MSA. It will be of interest to investigate whether PSP_RP correlate with clinical symptoms and experimental therapies.