2797

Quantitative susceptibility mapping of post-mortem ALS brains at 7T with quantitative iron histopathology validation
Chaoyue Wang1, Benjamin Tendler1, Menuka Pallebage-Gamarallage2, Olaf Ansorge2, Sarah Bangerter-Christensen2, Ricarda AL Menke1, Martin R Turner2, Sean Foxley3, and Karla Miller1

1Wellcome Centre for Integrative Neuroimaging, FMRIB, Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom, 2Clinical Neurology, Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom, 3Department of Radiology, University of Chicago, Chicago, IL, United States

Synopsis

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease of the motor system and its wider cortical connections. Progress in therapeutic development in ALS is compromised by a lack of specific biomarkers. In this work, we describe a platform for QSM data acquisition and post-processing protocol for postmortem brains. Preliminary results of 10 brains (validated with quantitative ferritin staining) have shown that ALS brains had significant higher mean susceptibility in motor cortex than control brains, which indicates that QSM has the potential to accurately quantify iron concentration and thus serve as an imaging biomarker for ALS.

Introduction

Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease of the motor system that lacks biomarkers for monitoring disease progression and intervention1. For instance, quantitative susceptibility mapping (QSM) could provide insights into iron concentration in the motor cortex and myelin content in white matter2. In this work, we describe a platform for acquiring and processing QSM data in post-mortem brains. Preliminary results are presented in seven ALS and three control brains, investigating the susceptibility properties of the motor cortex, an area with known pathological changes3 and the visual cortex, a control region. Quantitative iron histopathology validation performed in these postmortem brains was subsequently correlated with the quantitative susceptibility measures.

Methods

Data acquisition

We scanned ten formalin-fixed brains (seven ALS, three control) at 7T (Siemens, 1Tx/32Rx coil). Brain samples were packed in a custom-built container filled with susceptibility-matched liquid (Fluorinert). Scanning used a 3D multi-echo GRE sequence: TR=38ms, TE1=2ms, ΔTE=6.6ms, number of echoes=6, flip angle=15°, pixel bandwidth=651Hz, 0.5*0.5*0.6mm3, four repeats. Diffusion and structural scans were additionally acquired as described previously4.

QSM processing pipeline (Figure 1)

1. Coil combination. Raw data for individual coil channels were exported for custom phase reconstruction. After GRAPPA reconstruction, phase data from the first echo were registered to later echoes using FLIRT5. Coil-specific phase offsets were removed with complex division with the TE=2ms dataset, such that five echoes remained6. Coils were combined via a complex sum for each echo. The four repeats were subsequently co-registered using FLIRT5 before averaging.
2. Phase correction. A spatially linear phase along the read-direction was observed and removed from 2nd to 5th echoes (likely due to eddy currents). A time-varying field offset ($\triangle B$) and constant phase offset ($\phi_{0}$) were then estimated using a voxel-wise fit of the complex data $S(TE,\overrightarrow{r})$ over five echoes by minimizing: $$argmin_{\phi_{0}(\overrightarrow{r}),\triangle B(\overrightarrow{r})}\parallel S(\overrightarrow{r})-\mid S(TE,\overrightarrow{r})\mid \cdot e^{i(\phi_{0}(\overrightarrow{r})+2\pi\cdot\triangle B(\overrightarrow{r})\cdot TE)}\parallel_2^2$$
3. QSM processing. The background field was removed using V-SHARP7, with a maximal kernel size of 12mm and regularisation parameter of 0.02mm-1. The susceptibility ($\triangle \chi$) maps were subsequently generated using STAR-QSM8.

Quantitative analysis

The Fractional Anisotropy maps obtained from the diffusion datasets were used for gray-white segmentation. Masks of the motor cortex (M1) for face, hand and leg sub-areas separated were hand-drawn in the diffusion space. Visual cortex (V2) masks were generated from the Juelich atlas9,10. M1 and V2 masks were then co-registered to the QSM data using FLIRT5. The mean susceptibility values (in parts per billion) in M1 were normalized to the V2 control region, with the aim of accounting for global effects such as post-mortem interval that might vary across specimen. Tissue sections corresponding to the leg area of M1 and V2 for the left hemisphere were processed for ferritin staining. Iron concentration was quantified as the stained area fraction, and M1 values were normalized by V2 values for comparison to QSM.

Results

Representative QSM images are shown in Figure 2 for an ALS brain, demonstrating the overall quality of both the data and performance of the processing pipeline.

Mean susceptibility ($\triangle \chi_{mean}$) values of different M1 regions (normalized to V2) are given in Figure 3 with one-tailed t-tests. Normalized $\triangle \chi_{mean}$ of ALS brains in the face and hand M1 regions are significantly higher than that of controls (p=0.003 and p<0.0001, respectively). There was no significant difference for normalized $\triangle \chi_{mean}$ between ALS and control brains in leg M1 region (p=0.18). Examined as a whole, the normalized mean susceptibility of M1 in ALS brains is significantly greater than that of controls (p<0.0005).

Quantitative iron histopathology (ferritin staining) were performed for the left leg M1 region and V2, with. M1 results normalized by V2. Correlation between susceptibility and ferritin staining results is R2=0.2727 (p=0.061). These results are close to significance, but also demonstrate considerable variation across both ALS and controls.

Discussion

Our preliminary results suggest the proposed protocol is promising for post-mortem QSM. ALS brains were found to have significantly greater mean susceptibility in M1 areas than control brains, which agrees with previous in-vivo study3. This susceptibility difference could be related to iron-related disease pathology. QSM has the potential to accurately quantify iron concentration and thus serve as an imaging biomarker for ALS. Correlation between QSM-extracted susceptibility values and quantified ferritin stains identified a near-significant correlation. However, ferritin stains were only available in the leg M1 region, where ALS and control groups were less clearly differentiated than other M1 regions. These results may reflect a situation in which QSM in gray matter is predominantly reflecting iron content, but pathology exhibits considerable heterogeneity across ALS cases consistent with variable disease presentation. Future work will aim to extend ferritin staining to other M1 regions where QSM already exhibits differences between patients and controls.

Acknowledgements

No acknowledgement found.

References

1. Kiernan, Matthew C., et al. "Amyotrophic lateral sclerosis." The Lancet 377.9769 (2011): 942-955.

2. Haacke, E. Mark, et al. "Quantitative susceptibility mapping: current status and future directions." Magnetic resonance imaging 33.1 (2015): 1-25.

3. Costagli, M., et al. "Magnetic susceptibility in the deep layers of the primary motor cortex in amyotrophic lateral sclerosis." NeuroImage: Clinical 12 (2016): 965-969.

4. Pallebage-Gamarallage, Menuka, et al. "Dissecting the pathobiology of altered MRI signal in amyotrophic lateral sclerosis: A post mortem whole brain sampling strategy for the integration of ultra-high-field MRI and quantitative neuropathology." BMC neuroscience 19.1 (2018): 11.

5. Jenkinson, M., Bannister, P., Brady, J. M. and Smith, S. M. Improved Optimisation for the Robust and Accurate Linear Registration and Motion Correction of Brain Images. NeuroImage, 17(2), 825-841, 2002.

6. Robinson, Simon Daniel, et al. "Combining phase images from array coils using a short echo time reference scan (COMPOSER)." Magnetic resonance in medicine 77.1 (2017): 318-327.

7. Özbay, Pinar Senay, et al. "A comprehensive numerical analysis of background phase correction with V‐SHARP." NMR in Biomedicine 30.4 (2017).

8. Wei, Hongjiang, et al. "Streaking artifact reduction for quantitative susceptibility mapping of sources with large dynamic range." NMR in Biomedicine 28.10 (2015): 1294-1303.

9. Jenkinson, Mark, et al. "Fsl." Neuroimage 62.2 (2012): 782-790.

10. Eickhoff, Simon B., et al. "A new SPM toolbox for combining probabilistic cytoarchitectonic maps and functional imaging data." Neuroimage 25.4 (2005): 1325-1335.

Figures

Figure 1. Demonstration of QSM processing pipeline for postmortem brain data.

Figure 2. Representative images from an ALS brain. (a) and (c) are three orthogonal views of GRE magnitude and QSM images. (b) and (d) are magnified views of an axial slice. Red boxes indicate motor cortex and adjacent sensory cortex region where motor cortex shows high susceptibility.

Figure 3. Mean susceptibility difference between two groups and group-wise statistical analysis.

Figure 4. Correlation analysis between normalized mean susceptibility and normalized ferritin staining results in left leg M1 area.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
2797