The utility of quantitative MRI parameters in discriminating progressive supranuclear palsy from Parkinson's disease

We sought to re-evaluate the utility of all the quantitative neuroimaging parameters attributed to progressive supranuclear palsy (PSP) in discrimination between PSP and Parkinson' s disease (PD) subjects in our cohort. We aimed to propose some practical clinical remarks in this field.

In our retrospective study, 19 patients with 'probable' PSP and 37 patients with PD were enrolled. The radiological measurements of PSP, described in the previous reports, have been calculated in all subjects. The comparisons between the groups were performed and the measures regarding the accuracy of these parameters in the differentiation of PSP from PD subjects were analyzed.

We found that the values of magnetic resonance parkinsonism index-2 (MRPI-2), pontine-to-midbrain area (P/M) ratio, P/M 2 ratio, and 3rdV/bifrontal width ratio had high AUC values and very good discriminative powers. The analyses revealed that; for the discrimination of PSP from PD subjects, a 3rdvent/bifrontal width cut-off value of 0.30 had 42.1 % sensitivity and 97.3 % specificity; a P/M cut-off value of 6.03 had a 52.6 % sensitivity and 97.3 % specificity; and an MRPI-2 cut-off value of 7.43 had 57.9 % sensitivity and 97.3 % specificity. Remarkably, we also found that the presence of high values for both P/M and 3rdV/bifrontal ventricle rate had a positive predictive value of 100% for the diagnosis of PSP.

Our study results support the utility of previously defined neuroimaging parameters in distinguishing PSP and PD subjects. Besides, combined use of a high P/M ratio and 3rdV/bifrontal width may be practical and present strictly high evidence for the diagnosis of PSP.