Adjustment for Age-related Metabolic Changes Improves the Differential Diagnosis of Parkinsonism

Objectives: Idiopathic Parkinson’s disease (IPD) and atypical parkinsonian syndromes may have similar symptoms at the early disease stage. Positron emission tomography (PET) with ¹⁸F-FDG was shown to be helpful in the differential diagnosis of Parkinsonism. Of note, metabolic changes are composed of disease-related abnormalities and age-associated changes. As the onset of Parkinsonism spans over a large range of age (30-80), the mixed age-related changes may affect accuracy of an early PET based diagnosis. Here, we test, whether an adjustment for the age-related changes may improve an FDG-PET based early differential diagnosis.

Methods: 206 patients with suspected parkinsonian clinical features were subjected to an ¹⁸F-FDG PET imaging. After the imaging, these patients were assessed by blinded movement disorders specialists for a mean of 2.1 years before a final clinical diagnosis of IPD (n=136), multiple system atrophy (MSA, n=40), and progressive supranuclear palsy (PSP, n=30) were made. PET images were spatially normalized and parcellated into 119 anatomical regions according to an AAL template. A longitudinal template of age-related changes [1] was applied to adjust for the age-related metabolic changes. Namely, an age-adjusted Z score of each anatomical region was calculated by subtracting the age-associated activity from the measured activity and then normalized with standard deviation of the longitudinal template. An SVM-based error-correcting output codes (ECOC) method was applied to build a multi-class classifier for the differential diagnosis. The classifier based on the age-adjusted Z score was compared with the classification based on metabolic activities [2] and a spatial covariance analysis (SCA) [3] using one-way ANOVA. 500 bootstraps of training dataset (20 IPD, 20 MSA and 20 PSP) were applied to investigate the stability of the classifier under different training data.

Results: The age-adjusted Z score achieved on average 95.6% sensitivity, 92.4% specificity, 77.4% PPV and 98.8% NPV for the differentiation of PD, 95.6% sensitivity, 92.6% specificity, 98.8% PPV and 78.8% NPV for the classification of MSA and 97.2% sensitivity, 97.3% specificity, 99.8% PPV and 73.9% NPV for the classification of PSP. It produced a significantly (p<0.05) higher sensitivity and specificity than classification based on metabolic activity and SCA scores.

Conclusion: The proposed age-adjusted Z score improves the early differential diagnosis of Parkinsonism by adjusting for the age-related metabolic changes. Testing of various longitudinal models may further improve the results. Research Support: Sino-German Center for Brain Functional Imaging

• Download figure
• Open in new tab
• Download powerpoint