Send to

Choose Destination
Neurology. 2011 Sep 13;77(11):1042-7. doi: 10.1212/WNL.0b013e31822e55d0. Epub 2011 Aug 10.

MRI measurements predict PSP in unclassifiable parkinsonisms: a cohort study.

Author information

Institute of Neurology, University Magna Graecia, Germaneto, Catanzaro, Italy.



Magnetic resonance parkinsonism index (MRPI) has been proposed as a powerful tool to discriminate patients with progressive supranuclear palsy (PSP) from those with Parkinson disease (PD) or other parkinsonisms, on an individual basis. We investigated the usefulness of MRPI in predicting the clinical evolution in PSP of patients with clinically unclassifiable parkinsonism (CUP), i.e., parkinsonism not fulfilling the established clinical diagnostic criteria for any parkinsonian disorders, using a cohort study.


Forty-five patients with CUP underwent baseline clinical evaluation and MRI with calculation of MRPI. All patients were divided in 2 groups according to MRPI values. A group included 30 patients with CUP with normal MRPI values while the other group included 15 patients with CUP with MRPI values suggestive of PSP (higher than 13.55). A clinical follow-up was performed in all patients.


Duration of clinical follow-up in these 2 groups was 28.4 ± 11.7 months (mean ± SD). None of the patients with CUP with normal MRPI values fulfilled established clinical criteria for PSP (follow-up ranging from 24 to 60 months). By contrast, 11 of 15 patients with CUP with abnormal MRPI values (higher than 13.55) developed during the follow-up (range from 6 to 48 months) additional clinical features characteristic of probable (1 patient) or possible (10 patients) PSP. MRPI showed a higher accuracy in predicting PSP (92.9%) than clinical features, such as vertical ocular slowness or first-year falls (61.9% and 73.8%, respectively).


Our findings suggest that MRPI is more powerful than clinical features in predicting the evolution of CUP toward PSP phenotypes.

[Indexed for MEDLINE]

Supplemental Content

Full text links

Icon for HighWire
Loading ...
Support Center