Diagnostic accuracy of MR planimetry in clinically unclassifiable parkinsonism.

<bold>Introduction: </bold>Quantitative MR planimetric measurements were reported to discriminate between progressive supranuclear palsy (PSP) and non-PSP parkinsonism, yet few data exist on the usefulness of these markers in early disease stages.<bold>Methods: </bold>The pons-to-midbrain area ratio (P/M) and the Magnetic Resonance Parkinsonism Index (MRPI) as well as new indices, termed P/M2.0 and MRPI2.0, multiplying the former by a ratio of the third ventricle (3rdV) width/frontal horns (FH) width, were calculated on T1-weighted images in 84 patients with clinically unclassifiable neurodegenerative parkinsonism (CUP) at the time of imaging. Areas under the curve (AUCs) of these markers for predicting future PSP was determined. The final clinical diagnosis was made after at least 24 months of follow-up.<bold>Results: </bold>Final diagnosis was Parkinson's disease in 55 patients, multiple system atrophy in 12 cases, and PSP in 17. At baseline imaging, patients with a final PSP diagnosis had significantly higher MRPI, P/M, MRPI2.0 and P/M2.0 values compared to the other groups. AUCs in discriminating between future PSP and non-PSP parkinsonism were 0.91 for both the P/M and the MRPI and 0.98 for the P/M2.0 and the MRPI2.0.<bold>Conclusions: </bold>Brainstem-derived MR planimetric measures yield high diagnostic accuracy for separating PSP from non-PSP parkinsonism in early disease stages when clinical criteria are not yet fully met. Consistent with the underlying pathology in PSP, our study suggests that inclusion of 3rdV width makes P/M2.0 and MRPI2.0 more accurate in diagnosing early stage PSP patients than the P/M and MRPI. [ABSTRACT FROM AUTHOR]