Longitudinal follow-up of chronic multiple sclerosis lesions with quantitative MR imaging and partial volume-corrected proton MR spectroscopy
Ivan Kirov1,2, Shu Liu1,2, Assaf Tal3, William E. Wu1,2, Matthew S. Davitz1,2, James S. Babb1,2, Henry Rusinek1,2, Joseph Herbert4, and Oded Gonen1,2

1Center for Advanced Imaging Innovation and Research (CAI2R), New York University School of Medicine, New York, NY, United States, 2Bernard and Irene Schwartz Center for Biomedical Imaging, New York University School of Medicine, New York, NY, United States, 3Chemical Physics, Weizmann Institute of Science, Rehovot, Israel, 4Neurology, New York University Langone Medical Center, New York, NY, United States

Synopsis

We describe the evolution of chronic multiple sclerosis lesions from a quantitative MR imaging and spectroscopy perspective. Metabolite concentrations were obtained along with measures of lesion T1-hypointensity and size. Moderately hypointense lesions were more metabolically active than severely hypointense lesions, driving an increase in the glial marker myo-inositol. Correlational analyses revealed that lesion size is a better predictor of axonal health than T1-hypointensity, with lesions larger than 1.5 cm3 exhibiting terminal axonal injury. A positive correlation between changes in choline and in lesion size in moderately hypointense lesions implied that changes in lesion size are mediated by chronic inflammation.

Introduction

Lesions are the radiological hallmark of multiple sclerosis (MS), but T1- and T2-weighted imaging cannot define their pathophysiology. Proton MR spectroscopy (1H-MRS) can assess tissue metabolism, but it suffers from limited spatial resolution. Large voxel size introduces uncertainty in studying all but the largest lesions, because of contamination by cerebro-spinal-fluid (CSF) and normal-appearing tissue. Another limitation of conventional 1H-MRS is the use of metabolic ratios that mask changes occurring in both the numerator and denominator. Here we use absolute quantification with stringent partial volume correction, combined with 0.75 cm3 spatial resolution 1H-MRS imaging (1H-MRSI), to study MS lesions over a 3-year, semi-annual follow-up. Concentrations of N-acetyl-aspartate (NAA), creatine (Cr), choline (Cho) and myo-inositol (mI) are obtained along with quantitative measures of lesion T1-hypointensity and size. We studied the changes of these metrics over time, as well as their cross-sectional and longitudinal dependence. The goal was to fully characterize the metabolism associated with chronic lesions' evolution and to define the metabolic correlates of the conventional measures of lesion status.

Methods

Subjects: 10 relapsing-remitting MS patients within 6 years from diagnosis, scanned semi-annually for 3 years. Data acquisition: pre- and post-contrast T1-weighted MRI (MP-RAGE), T2-weighted MRI (FLAIR), B0 shimming, 10×8×4.5 cm (AP×LR×IS)=360 cm3 1H-MRS VOI (PRESS TE/TR=35/1800 ms), encoded to 480 voxels, each 1.0×1.0×0.75 cm3 (Fig. 1A-C). Segmentation: Lesion masks were obtained from FLAIR1 and gray, white matter (GM, WM) and CSF masks were segmented from MP-RAGE2. Co-registration: FLAIR was co-registered to the MP-RAGE and the transformation matrix was applied to the lesion mask co-registering it to the MP-RAGE space (Fig. 2B). Mis-registered pixels were deleted (Fig. 2C). All masks, now in MP-RAGE space, were co-registered with the 1H-MRSI, yielding their volume in every 1H-MRSI voxel. Lesion characterization: Based on their MP-RAGE contrast, lesions were quantitatively defined as isointense or hypointense, and within the latter group: as severely hypointense (SH) and moderately hypointense (MH). A T1-contrast ratio was calculated for each lesion as a continuous measure of T1-hypointensity3. Metabolic quantification: Phantom replacement with lesion-specific T1 and T2 relaxation times4,5. 1H-MRSI quality control and partial volume considerations: To maximize lesion inclusion and control partial volume “voxel shifting6” was performed for each lesion (Fig 1D-E). Only voxels with >40% lesion, <30% CSF, <30% GM; metabolite Cramer-Rao lower bounds<20% and 4<linewidths<13 Hz were retained. Metabolite concentrations were corrected for partial CSF volume. Statistics: ANCOVA, random coefficients regression and Pearson correlations with 2-sided p values. Analyses controlled for variable lesion volume in the 1H-MRS voxels.

Results

Eighteen unique lesions providing 89 measurements were studied. All were T1-hypointense (6 SH, 12 MH) (Fig. 3A-B).

Cross-sectional findings:

Fig. 3C: Cr was higher in MH than in SH lesions (6.0 millimolar±1.3 vs. 5.0±1.4, p=0.05). There was a statistical trend of lower NAA in SH versus MH lesions (p=0.06).

Fig. 4: NAA concentration was inversely related to lesion size (r=-0.3, p=0.04). A trend for such relationship was also found for Cr (r=-0.3, p=0.06).

Longitudinal findings:

Fig. 5A: An increase in mI (0.03 mM/month, p=0.02), driven by its change in MH lesions (statistical trend: 0.03 mM/month, p=0.06). 5B: A decrease in T1-contrast ratio (-0.001/month, p=0.03), also driven by its change in MH lesions (-0.001/month, p=0.04).There were trends for increasing NAA in MH lesions (0.03 mM/month, p=0.07) and for size increase in the SH lesions (0.01 cm3/month, p=0.06). Rates of change in lesion size were strongly correlated with changes in Cho, for MH lesions (r=0.8, p=0.01).

Discussion

Cross-sectional findings: The findings among SH and MH lesions (Fig. 3) are in line with previous 1H-MRS3,7,8 and histopathology9,10, indicating that very hypointense lesions suffer more neuronal damage than less hypointense ones. Here we show, however, that NAA and Cr decrease as a function of T1-hypointensity only within SH lesions (Fig. 4), and that this relationship is mediated by lesion size, which is a better predictor of low NAA than T1-hypointensity. Additionally, lesions larger than 1.5 cm3 exhibited very narrow (low) NAA concentration range, indicating terminal axonal injury.

Longitudinal findings: The increase in mI (Fig. 5A) indicates astrogliosis, a histopathological hallmark of the MS lesion11. The correlation between changes in Cho and in lesion size implies lesion size is mediated by chronic inflammation. The trend for NAA increase may indicate medication effects, as hypothesized previously12-14. The observed changes in T1-hypointensity (Fig. 5B) and size indicate that decrease in lesion T1-hypointensity precedes enlargement.

Conclusion

The results shed light on the metabolism of lesion evolution, and inform about the metabolic correlates of lesion appearance on conventional MRI, as encountered in clinical practice.

Acknowledgements

This work was supported by NIH grants NS050520, NS29029, EB01015 and the Center for Advanced Imaging Innovation and Research (CAI2R, www.cai2r.net), a NIBIB Biomedical Technology Resource Center (NIH P41 EB017183). Assaf Tal acknowledges the support of the Monroy-Marks Career Development Fund, the Carolito Stiftung Fund, the Leona M. and Harry B. Helmsley Charitable Trust and the historic generosity of the Harold Perlman Family.

References

1. Rusinek H, Glodzik L, Mikheev A, Zanotti A, Li Y, De Leon M. Fully automatic segmentation of white matter lesions: error analysis and validation of a new tool. Intern J Computer Assisted Radiology & Surgery 2013; 8: 289-291.

2. Ashburner J, Friston K. Multimodal image coregistration and partitioning--a unified framework. Neuroimage 1997; 6: 209-17 .

3. van Walderveen MA, Barkhof F, Pouwels PJ, van Schijndel RA, Polman CH, Castelijns JA. Neuronal damage in T1-hypointense multiple sclerosis lesions demonstrated in vivo using proton magnetic resonance spectroscopy. Ann Neurol 1999; 46: 79-87

4. Brief EE, Vavasour IM, Laule C, Li DK, Mackay AL. Proton MRS of large multiple sclerosis lesions reveals subtle changes in metabolite T(1) and area. NMR Biomed 2010; 23: 1033-7.

5. Kirov II, Liu S, Fleysher R, Fleysher L, Babb JS, Herbert J, et al. Brain metabolite proton T2 mapping at 3.0 T in relapsing-remitting multiple sclerosis. Radiology 2010; 254: 858-66.

6. Bracewell RN. The Fourier transform and its applications. New York: McGraw-Hill, 1978.

7. Brex PA, Parker GJ, Leary SM, Molyneux PD, Barker GJ, Davie CA, et al. Lesion heterogeneity in multiple sclerosis: a study of the relations between appearances on T1 weighted images, T1 relaxation times, and metabolite concentrations. J Neurol Neurosurg Psychiatry 2000; 68: 627-32.

8. He J, Inglese M, Li BS, Babb JS, Grossman RI, Gonen O. Relapsing-remitting multiple sclerosis: metabolic abnormality in nonenhancing lesions and normal-appearing white matter at MR imaging: initial experience. Radiology 2005; 234: 211-7.

9. van Waesberghe JH, Kamphorst W, De Groot CJ, van Walderveen MA, Castelijns JA, Ravid R, et al. Axonal loss in multiple sclerosis lesions: magnetic resonance imaging insights into substrates of disability. Ann Neurol 1999; 46: 747-54.

10. van Walderveen MA, Kamphorst W, Scheltens P, van Waesberghe JH, Ravid R, Valk J, et al. Histopathologic correlate of hypointense lesions on T1-weighted spin-echo MRI in multiple sclerosis. Neurology 1998; 50: 1282-8.

11. Frohman EM, Racke MK, Raine CS. Multiple sclerosis--the plaque and its pathogenesis. N Engl J Med 2006; 354: 942-55.

12. Wiebenga OT, Klauser AM, Schoonheim MM, Nagtegaal GJ, Steenwijk MD, van Rossum JA, et al. Enhanced axonal metabolism during early natalizumab treatment in relapsing-remitting multiple sclerosis. AJNR Am J Neuroradiol 2015; 36: 1116-23.

13. Khan O, Shen Y, Bao F, Caon C, Tselis A, Latif Z, et al. Long-term study of brain 1H-MRS study in multiple sclerosis: effect of glatiramer acetate therapy on axonal metabolic function and feasibility of long-Term H-MRS monitoring in multiple sclerosis. J Neuroimaging 2008; 18: 314-9.

14. Narayanan S, De Stefano N, Francis GS, Arnaoutelis R, Caramanos Z, Collins DL, et al. Axonal metabolic recovery in multiple sclerosis patients treated with interferon beta-1b. J Neurol 2001; 248: 979-86.

Figures

Figure 1. Positioning of the 1H-MRSI VOI (A, B, C) and voxel shifting (D, E)

Note the spectra before (F) and after (G) voxel shifting of the lesion which originally fell between two voxels (D). Fitted model functions are shown in gray.


Figure 2. Segmentation and co-registration

Original FLAIR and MP-RAGE (A), lesion segmentation and image and lesion co-registration (B), and tissue segmentation (C) showing deletion of mis-registered voxels (red arrow).


Figure 3. Moderately hypointense (MH) and severely hypointense (SH) lesions

An example of each lesion type (A) and baseline T1-contrast ratio distributions of all MH and SH lesions (B), compared to the signal intensity of normal-appearing GM/WM (NAGM/NAWM). Metabolite distributions in MH and SH lesions (C) and example spectra (D).


Figure 4. Relationship between lesion metabolism and MRI

Lesion metabolite concentrations (in millimolar, mM) as a function of lesion T1-contrast ratio (left), and size (right). The statistical analyses showed a correlation between NAA and lesion size (dotted line), and a trend for correlation between Cr and lesion size.


Figure 5. 1H-MRSI and MRI changes in chronic lesions

Changes in mI concentration (A) and T1-contrast ratio (B) among all chronic lesions, with dotted lines indicating their rates of change as statistically significant.

T1/T2-weighted MRI and spectra at each corresponding timepoint of a MH (C) and a SH (D) lesion.




Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
4066