Vanessa Wiggermann1,2, Enedino Hernandez-Torres2,3, Inga C Ibs4, Stephanie M Schoerner5, Galina Vorobeychik6, Luanne Metz7, David KB Li8,9, Anthony Traboulsee9,10, and Alexander Rauscher2,9,11
1Physics and Astronomy, University of British Columbia, Vancouver, BC, Canada, 2Pediatrics, University of British Columbia, Vancouver, BC, Canada, 3UBC MRI Research Centre, University of British Columbia, Vancouver, BC, Canada, 4University of Osnabrueck, Osnabrueck, Germany, 5Technical University of Dortmund, Dortmund, Germany, 6Fraser Health MS Clinic, Burnaby, BC, Canada, 7Clinical Neurosciences, University of Calgary, Calgary, AB, Canada, 8Radiology, University of British Columbia, Vancouver, BC, Canada, 9Center for Brain Health, University of British Columbia, Vancouver, BC, Canada, 10Medicine (Neurology), University of British Columbia, Vancouver, BC, Canada, 11Child and Family Research Institute, University of British Columbia, Vancouver, BC, Canada
Synopsis
Using magnetic-susceptibility based MR techniques for the assessment of
damage due to multiple sclerosis (MS) has been controversial, in
particular in MS lesions where the underlying pathological changes are not
yet fully understood. Here, we investigated the changes of the MR
frequency and quantitative susceptibility signal during acute MS lesion
formation. We observed that both metrics behave similarly, indicating that
non-local effects have little contribution to the QSM signal increase and
hence dipole inversion might not be required to assess damage during MS
lesion formation accurately.Purpose
Quantitative susceptibility
mapping (QSM) has become an exciting new research field in the past
10 years and is starting to be applied for every day clinical
purposes such as the detection of vessels
1,
differentiation of calcium and iron clusters
2, and more
recently for the assessment of iron accumulation in the brain
3
or other body parts
4. The quantitative assessment of
damage in neurodegenerative diseases is also gaining momentum,
however in particular the characterization of damage in multiple
sclerosis (MS) lesions has been controversial. Studies utilizing
frequency shift (FS) imaging demonstrated a steep signal increase at
the time of MS lesion formation, attributed to microstructural
changes during to the transition of healthy myelin into myelin
debris
5,6. Similarly, in a cross-sectional
study QSM detected susceptibility changes over time, attributing
increases in QSM signal at time of lesion formation to iron
accumulation
7. Given the non-locality of the underlying MR
phase signal, one may argue for the necessity of performing QSM.
However, if the initial FS increase is not due to bulk magnetic
susceptibility changes, but due to structural changes alone, no
non-local field changes will be created and therefore QSM and FS maps
will provide the same information. Here, we tested these hypotheses
by comparing the MR FS and QSM values during the formation of new
acute MS lesions.
Methods
In
a randomized, placebo controlled, double-blind trial of Minocycline,
3T MRI was performed for 38 patients with clinically isolated syndrome
at month 0,3,6,12 and 24, if the patients had not convert to
McDonald definite MS. MR FS maps and QSM maps were calculated on
all time points from the same gradient-echo acquisition (5 echoes,
TR/TE/ΔTE=48/20/6ms, acq.
voxelsize=0.6x0.8x2.4 mm
3, reconstructed to 0.4x0.4x1.2
mm
3). The data was unwrapped using a Laplacian approach8
and weakly high-pass filtered to obtain background field removed FS
maps. For QSM, the unwrapped images were background filtered using
SHARP
9 and susceptibilities were estimated using the
superfast dipole inversion
9 approach with truncated
singular value decomposition and thresholded k-space division at 0.02
and 0.53. We identified enhancing lesions on co-registered
Gadolinium(Gd)-enhanced T
1-weighted image, and defined
their month of appearance as Month 0 on the common time line. All
regions of interest were segmented with an in-house developed
software and average FS and QSM values were determined for all
regions.
Results
9/38 patients had Gd-enhancing
lesions on one of the time points. Mapping them onto a common time
line allowed us to estimate QSM and FS values at -6,-3,0,3,6,9,12 and
21 months of lesion age in 2,23,32,24,5,12,4 and 3 lesions,
respectively. Given the low number of lesions at month -6,12 and 21,
we excluded these time points. Comparing the mean FS and QSM values
in all Gd-enhancing lesions before and after their appearance (Fig.
1), we observed a significant increase for both metrics between month
-3 to month 3 (p<0.004) which remained significantly elevated up
to month 9 (p<0.004). FS maps also showed a significant signal
elevation already at month 0 (p=0.03), while QSM maps did not reach
significance at this time point. Figure 2 qualitatively compares FS
(upper row) and QSM (bottom row) for a lesion that appeared at the
second time point. The lesion (marked with a red circle) showed the
same behaviour on both scans, appearing ring-like at first, followed
by a volume reduction, but consistently elevated FS and QSM values.
Discussion
FS and QSM values depend strongly
on the chosen signal processing pipeline. Therefore, values may not
be comparable between studies. We normalized our FS values to the
Larmor frequency at 3T, however the actual Larmor frequency will vary
from patient to patient. Similarly, normalization of QSM is not yet
standardized
10. Due to white matter loss and brain atrophy
in MS, we decided not to normalize the susceptibility values
potentially causing larger patient-to-patient variations. The low
number of lesions at month 6 may further contribute to the lack of
significance and inconsistent signal reduction on the FS maps.
Conclusion
We
demonstrated in this work that MS lesions appear to show very similar
changes on MR frequency and QSM maps, indicating that the dipole
inversion may not be necessary due to the apparent lack of bulk
magnetic susceptibility changes during early lesion formation.
Acknowledgements
This
study was supported by the Multiple Sclerosis Society of Canada.References
[1] Reichenbach et al., Radiology
204; 1997 [2] Deistung et al., Z Med Phys 16; 2006 [3] Langkammer et
al., NeuroImage 62; 2012 [4] Sharma et al., Magn Res Med 74; 2014 [5]
Wiggermann et al., Neurology 81; 2013 [6] Yablonskiy et al., PNAS 109;
2012 [7] Chen et al., Radiology 271; 2014 [8] Li et al., NeuroImage
55; 2011 [9] Schweser et al., Magn Res Med 69; 2013 [10]
Acosta-Cabronero et al., PLOS One 8; 2013