Differentiation of Parkinsonism-Predominant Multiple System Atrophy from Idiopathic Parkinson Disease Using 3T Susceptibility-Weighted MR Imaging, Focusing on Putaminal Change and Lesion Asymmetry

Study Population

This retrospective cohort study was approved by the institutional review board of our hospital, and the requirement for informed consent was waived. The investigators searched all patients who were referred from the Movement Disorder Center to undergo brain MR imaging, including SWI, for further evaluation of parkinsonism between April 2010 and May 2012 (n = 207). Among those patients, we enrolled subjects who had been initially, clinically diagnosed with either MSA-p (n = 33) or IPD (n = 50). All routine brain MR imaging protocols performed during the study period for the evaluation of movement disorders included SWI, except in 1 patient. Neurologists experienced with movement disorders (H.-J.K. and B.S.J.) made the final clinical diagnosis by reviewing the initial and follow-up clinical data in December 2013. MSA-p was diagnosed according to the “Second Consensus Statement on the Diagnosis of Multiple System Atrophy”¹ and was categorized as probable or possible. IPD was diagnosed on the basis of the UK Parkinson's Disease Society Brain Bank criteria.¹⁴

Five initial patients with MSA-p were excluded for the following reasons: an old intracerebral hemorrhage involved the basal ganglia (n = 1), diagnoses were changed to cerebellar dysfunction type multiple system atrophy (MSA-c) (n = 3), and final diagnosis found no movement disorder (n = 1). One initial patient with IPD was excluded due to lack of SWI by a technical error. One initial patient with MSA-p had the final diagnosis changed to IPD. Finally, there were 27 patients with MSA-p and 50 patients with IPD. Twenty-seven age-matched subjects who met the following criteria were enrolled as the control group: 1) They underwent brain MR imaging with SWI between September 2012 and November 2012, 2) had no definite focal lesions in the brain parenchyma, and 3) had no clinical evidence of neurodegenerative diseases or movement disorders. The control group subjects were age-matched 1:1 with the subjects with MSA-p. Finally, our study included 27 patients with MSA-p (17 probable and 10 possible), 50 patients with IPD, and 27 healthy control subjects.

MR Imaging Protocol

All subjects underwent brain MR imaging by using a 3T scanner (Magnetom Verio; Siemens, Erlangen, Germany) with a 32-channel head coil. A set of images was acquired by using a 3D fully flow-compensated gradient-echo SWI sequence along the transverse plane parallel to the anterior/posterior commissure lines. The imaging parameters were as follows: TR = 28 ms, TE = 20 ms, flip angle = 15°, section thickness = 2 mm, FOV = 178 × 220 mm, matrix size = 364 × 448, and number of excitations = 1. Therefore, the voxel size was 0.49 × 0.49 × 2.0 mm³. The magnitude and phase images were processed to create final SWI on the MR imaging console workstation (syngo MR B17; Siemens). The corrected phase images and final processed SWI were used for further analyses.

Image Interpretation

Two radiologists (I.H. and C.-H.S., with 4 and 17 years of experience in neuroradiology, respectively), who reached a consensus, reviewed the processed SWI. The investigators qualitatively evaluated the imaging findings for putaminal atrophy and signal intensity of the posterolateral putamen compared with adjacent structures in the basal ganglia (eg, globus pallidus). In particular, loss of lateral convexity of the posterolateral putaminal border was also indicative of putaminal atrophy.² In addition to the SWI findings, cerebellar and brain stem findings were also reviewed with T2-weighted axial images for the following: 1) hot-cross bun sign of the pons, 2) high signal intensity in the middle cerebellar peduncle, and 3) cerebellar atrophy. The putaminal signal intensity was graded by using a relative 4-step scale introduced by Kraft et al,¹⁵ in which the signal of the putamen was rated as higher = 0, equal = 1, hypointense = 2, or markedly hypointense = 3. Other image findings were rated as negative = 0, suspicious = 1, or definitely positive = 2. Laterality was also recorded if a finding of putaminal atrophy was rated as suspicious or definitely positive or the signal intensity was rated as marked hypointensity. During the interpretation, images were presented randomly and investigators were blinded to the patients' diagnoses.

Quantitative Image Analysis

We quantitatively measured the putaminal width and phase-shift values in corrected phase images. Image analyses were also performed by 2 radiologists who performed qualitative image interpretation and reached a consensus. The most representative section demonstrating the largest width and most well-demarcated border of the posterior putamen was selected for further analysis. The images were analyzed by using ImageJ 1.46r software (National Institutes of Health, Bethesda, Maryland) with 400% magnification to more easily determine the boundaries of the basal ganglia structures. Investigators drew 3 straight lines with widths of 3 pixels crossing the midportion, posterior half, and far posterior portion of the putamen in the right-to-left direction, as shown in Fig 1A. Using the Plot Profile function in ImageJ, we plotted phase-shift values along the straight line, and putaminal positive phase-shift values were easily recognized (Fig 1A). The width and the mean phase-shift value on each side of the putamen were recorded by using the 3 lines. The phase-shift value was recorded in Siemens Phase Units, which range from 0 to 4096 corresponding to −π to +π radian.¹⁶ To normalize the putamen width by brain size, we also measured the largest biparietal diameter of the brain parenchyma on the same image plane. The normalized putaminal width (putaminal width/biparietal diameter × mean of biparietal diameter of all subjects) was calculated and used for the statistical analysis. Dominant-side values (shorter putamen width or higher mean phase-shift value) and the mean of the values of both sides were calculated and used for the statistical analysis. Finally, to evaluate the asymmetry of the measured values, we also calculated the dominant/nondominant side ratios of the measured values. All analyses were performed for each straight line drawn at the mid-, posterior half, and far posterior portion of the putamen; then, the level that showed the most differentiation was selected and further analyzed.

• Download figure
• Open in new tab
• Download powerpoint

Fig 1.

Measurement of putaminal width and mean phase-shift values from corrected phase images by using ImageJ software. A, Three lines crossing the mid-, posterior half, and far posterior putamen are drawn to allow measurements. B, The corresponding plot profile of the line crossing the far posterior portion of the putamen demonstrates increased phase-shift values in both putamina (arrows).

Neurologic Assessment

Experienced neurologists (H.-J.K. and B.S.J.) assessed the final clinical diagnoses of IPD or MSA-p by reviewing the initial and follow-up clinical data. Those neurologists also assessed initial modified Hoehn and Yahr scale scores. In addition, the clinical-onset side of resting tremors or bradykinesia was recorded if the patient presented with asymmetric symptoms at the time of disease onset. If the symptom onset was not asymmetric or information regarding the side of onset was unclear, the more severe symptomatic side was recorded.

Statistical Analyses

All statistical analyses were performed by using MedCalc for Windows, Version 14.10.2 (MedCalc Software, Mariakerke, Belgium). For all statistical analyses, a P value < .05 was considered a statistically significant difference. For the qualitative assessment, the prevalence of each finding was calculated. In addition, the sensitivity and specificity for differentiating MSA-P from IPD were also investigated. To assess interrater agreement, we calculated the weighed κ coefficient for 2 major qualitative imaging findings (putaminal atrophy and signal intensity of the posterolateral putamen) on the basis of the initial review data before the 2 reviewers arrived at a consensus. A χ² test, Mann-Whitney test, 1-way analysis of variance test, and Kruskal-Wallis test were used to compare the clinicopathologic characteristics and quantitatively measured values for the MSA-p, IPD, and control subjects, as appropriate. A pair-wise comparison was used for the post hoc analysis. Furthermore, a receiver operating characteristic (ROC) curve was drawn for the variables that were significantly different by the Kruskal-Wallis test for differentiating MSA-p from IPD. The area under the ROC curve was calculated to evaluate the diagnostic performance of those variables.