Accuracy of magnetic resonance parkinsonism index for differentiation of progressive supranuclear palsy from probable or possible Parkinson disease

Abstract

Background: Combined measurements on conventional magnetic resonance imaging (MRI), such as midbrain area/pons area or magnetic resonance parkinsonism index (MRPI) (pons area/midbrain area × middle cerebellar peduncle width/superior cerebellar peduncle width), have been proposed as powerful tools in the differential diagnosis between progressive supranuclear palsy (PSP) and Parkinson disease (PD). In this study, we evaluated the accuracy of MRPI, compared with midbrain/pons ratio, in distinguishing PSP from probable and possible PD.

Methods: Forty-two PSP patients, 170 probable PD patients, 132 possible PD patients, and 38 control subjects underwent MRI and, for each patient, midbrain/pons ratio and MRPI were calculated.

Results: Midbrain/pons ratio showed low accuracy in distinguishing PSP patients from those with probable PD (92.9% sensitivity; 85.3% specificity; 86.8% diagnostic accuracy) or those with possible PD (88.1% sensitivity, 88.3% specificity, and 88.2% diagnostic accuracy) and control subjects (97.6% sensitivity, 92.1% specificity, and 95% diagnostic accuracy). By contrast, MRPI showed higher accuracy to distinguish PSP from probable PD (100% sensitivity, 99.4% specificity, and 99.5% diagnostic accuracy), from possible PD (100% sensitivity, 99.2% specificity, and 99.4% diagnostic accuracy), and from control subjects (sensitivity, specificity, and diagnostic accuracy of 100%).

Conclusions: Our study confirms that MRPI is a more accurate measure than midbrain/pons ratio for differentiation of patients with PSP from those with probable and possible PD.