Nigral Iron Distribution in Brain of Parkinson’s Disease: A Combined Structural Voxel-wise and ROI-based Study with Quantitative Susceptibility Mapping
Darrell Ting Hung Li1, Edward Sai Kam Hui1, Queenie Chan2, Nailin Yao3, Siew-eng Chua4, Grainne M. McAlonan4,5, Shu Leong Ho6, and Henry Ka Fung Mak1

1Department of Diagnostic Radiology, The University of Hong Kong, Hong Kong, Hong Kong, 2Philips Healthcare, Hong Kong, Hong Kong, 3Department of Psychiatry, Yale University, New Haven, CT, United States, 4Department of Psychiatry, The University of Hong Kong, Hong Kong, Hong Kong, 5Department of Forensic and Neurodevelopmental Science, King’s College London, London, United Kingdom, 6Department of Medicine, The University of Hong Kong, Hong Kong, Hong Kong

Synopsis

Abnormal nigral iron deposition is considered one of the major biomarkers in Parkinson’s disease (PD). Extensive studies had been performed to evaluate iron concentration in substantia nigra using different in vivo imaging methods. Whole structure ROI-based analysis of basal nuclei is a majority approach in similar studies. In this study, we investigated the distribution of iron in substantia nigra with both voxel-wise and split ROI methods. Location of significant higher iron concentration was identified to be around pars compacta of the substantia nigra in PD brain. The two methods adopted in this study agreed with each other.

Purpose

Abnormal iron level in the brain of Parkinson’s disease (PD) was postulated to be a cause of degeneration of dopamine-generating neurons in substantia nigra pars compacta1-3. Studying of in vivo brain iron distribution is essential for the study of prognosis of the disease. In this study, we aimed to investigate the distribution of iron, in terms of measuring magnetic susceptibility with quantitative susceptibility mapping (QSM) technique, in the brain of PD patients particularly in the region of substantia nigra. Both voxel-wise and ROI-based studies were performed to obtain a better understanding to iron deposition in the PD brain.

Methods

55 PD patients (37 males, mean age ± S.D. = 66 ± 9 years, mean illness duration ± S.D. = 9 ± 6 years) and 26 healthy control (15 males, mean age ± S.D. = 62 ± 7 years) were recruited. All QSM raw images were acquired using a 3.0T Philips scanner with the following parameters: 3D-T1FFE sequence, TR/TE = 28/23 ms, flip angle = 15°, NEX = 1, FOV = 230 x 230 x 180 mm3, reconstructed resolution = 0.45 x 0.45 x 1 mm3. QSM images were generated with the following approaches: Lapalcian-based phase unwrap, then background field removal with ReSHARP, followed by total variation regularization-based L1-norm QSM algorithm4,5. The reconstructed QSM images were spatial normalized to MNI152 standard space with the FSL-FNIRT algorithms (fig. 1), and smoothed with a 3D-Gaussian filter before passing to SPM12 for voxel-wise analysis. For structural voxel-wise study, subcortical structures were either segmented by the FSL-FIRST (caudate, putamen, pallidum, thalamus, hippocampus and amygdala) or manually defined on the standard space (substantia nigra, red nucleus and dentate nucleus, fig. 1). The ROIs were applied as explicit masking during statistical analysis. To further analyze the susceptibility distribution in bilateral substantia nigra, slice-by-slice ROI comparison was performed. The measured susceptibility values were normalized to the averaged values of the bilateral frontal white matter. Two-sample t-test, adjusted for age and gender of the subjects, was employed in statistical analysis.

Results

Age and gender were included as covariates in the study for adjustment in the statistical tests. Whole-brain voxel-wise study with two-sample t-test showed that possible higher magnetic susceptibility in the bilateral substantia nigra in the PD brain (uncorrected p-value < 0.001, fig.2). In order to test for significance in individual subcortical structures, explicit masking was applied based on the ROI generated. Structural voxel-wise analysis of substantia nigra with two-sample t-test showed that some of the voxels in the structure, which is anatomically corresponded to the substanita nigra pars compacta, indicated a significantly higher magnetic susceptibility in PD patients (FEW-correct p-value < 0.05, fig.3). Statistical tests were also performed in other structures (bilateral thalamus, caudate, putamen, pallidum, hippocampus, amygdala, red nucleus and dentate nucleus) but yet no significance voxels were identified. Further analysis of substantia nigra by slice-by-slice ROI measurement in the standard space was carried out. The average susceptibility value was higher in the PD group, and some of the slices were tested to be statistically significant (p < 0.05, fig. 4). These slices also corresponded to the location of significant voxels in the structural voxel-wise study.

Discussions

Accumulation of mineral iron in substantia nigra of PD patients was extensively studied and reported with both post-mortem and in vivo approaches1-3. The result of this study confirmed that substantia nigra, in particular the pars compacta, is the region with significantly higher iron loading in the PD brain. Both structural voxel-wise analysis and split ROI studies pointed to the same findings and the result agreed with each other. Non-invasive in vivo imaging of mineral iron with the QSM technique could be a potential tool for the prognostic study of the PD. This study, apart from reporting the abnormal iron level in substantia nigra pars compacta, also explored the feasibility of performing voxel-wise analysis. Traditional method to analyze substantia nigra iron concentration includes manual drawing of ROI on subject’s magnitude images or even the QSM maps by researchers. However, such work could be tedious in a large scale study, and introduction of bias and human error would be inevitable. Voxel-wise analysis can be completely automated, which reduces the subjective human error being introduced in the analysis. The method employed in this study, however, assumed accurate deformation of the subcortical structures with FSL. Check of registration accuracy is therefore essential prior to statistical analysis.

Conclusion

Both voxel-wise and ROI-based analyzes confirmed higher iron concentration in substantia nigra pars compacta of PD patients. QSM could be a potential tool to study in vivo iron deposition in PD.

Acknowledgements

No acknowledgement found.

References

1. Du G., et al. Quantitative Susceptibility Mapping of the Midbrain in Parkinson’s Disease. Mov Disord. 2015 Sep 12.

2. Murakami Y., et al. Usefulness of Quantitative Susceptibility Mapping for Diagnosis of Parkinson’s Disease. Am J Neuroradiol 2015 Jun; 36(6): 1102 – 08

3. Barbosa JH., et al. Quantifying brain iron deposition in patients with Parkinson’s disease using quantitative susceptibility mapping, R2 and R2*. Magn Reson Imaging. 2015 Jun;33(5):559-65.

4. Bilgic B, et al. MRI estimates of brain iron concentration in normal aging using quantitative susceptibility mapping. NeuroImage 2012; 59: 2625-2635.

5. Bilgic B, et al. Fast Quantitative Susceptibility Mapping with L1-Regularization and Automatic Parameter Selection. Magn Reson Med. 2014 Nov; 72(5): 1444-59.

Figures

Fig. 1 (Upper row) The average of the 26 magnitude images in the healthy control group with the concerning subcortical structures overlaid. (Lower row) A sample L1-regularized QSM images from one of the patients in the PD group.

Fig. 2 Whole brain voxel-wise analysis to identify any region of possible higher susceptibility in the PD brain. Uncorrected p-value of 0.001 was used with extended threshold of 20 voxels. The result is overlaid on the mean magnitude images of the healthy group.

Fig. 3 Structural voxel-wise analysis of substantia nigra to identify location of higher iron concentration in PD. The yellow pixels on the image indicated higher susceptibility value in PD patients at the FWE-corrected p-value of 0.05. The result was overlaid on the mean magnitude images of the healthy group.

Fig. 4 Comparison of magnetic susceptibility value across the slices of substantia nigra in craniocaudal direction between the healthy and PD groups. Left and right substantia nigra were separately analyzed with two-sample t-test..



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
0690