Introduction

Neurodegeneration, featured by myelin and axonal damage, is a key element of multiple sclerosis (MS) pathology and a primary source of patients’ functional deterioration¹. The magnetization transfer ratio (MTR)² provides an assay of this myelin damage³, but is also sensitive to non-physiological parameters (i.e., hardware, RF power, and sequence timings), which makes it challenging to deploy for large-scale, multiple-site clinical trials^4-5. By removing these confounding features, quantitative MT (qMT) methods⁵ provide indices with improved tissue-specificity, such as the macromolecular-to-free water pool-size-ratio (PSR) and the R₁ of the free water pool (R_1f). We previously developed^6-9 a selective inversion recovery (SIR) qMT sequence that: i) simplifies the complicated acquisition/analysis associated with standard qMT methods (e.g., need for independent measures of T₁, B₁, and B₀) and ii) reduces scan times⁹. In this study, we evaluated the relationship between the resulting SIR-derived indices (PSR and R_1f) in the corticospinal tract (CST) and clinical disability. The CST was chosen for study herein because: i) previous work¹⁰ has shown that tract-specific quantitative MRI is more predictive of disability than whole-brain analyses and ii) the CST plays a significant role in motor function.

Methods

Subjects/Study Overview: A retrospective cohort study of multiple sclerosis patients (n=18) and age- and sex-matched healthy controls (n=9) was performed. MRI was acquired in all subjects as described in the following section. In addition, clinical data, including the Expanded Disability Status Scale (EDSS) and Timed 25-Foot Walk Test (T25FW) test, was acquired in all patients for comparison to MRI findings. A summary of the demographic, clinical, and anatomical MRI characteristics of the study cohort are given in Fig. 1. Together, these data suggest that cohort studied herein is representative of the typical MS phenotype.

MRI Acquisitions: All scans were acquired using a whole-body 3T Achieva MRI scanner (Philips Healthcare, Cleveland, OH) equipped with a volume transmit and 32-channel receive head coil (NOVA Medical, Wilmington, MA). The scanning protocol included conventional anatomical scans (T₁-weighted MP-RAGE, T₂-weighted TSE, and T₂-weighted FLAIR) along with DTI and SIR acquisitions. SIR data were acquired using an inversion recovered prepared TSE sequence at four different optimization combinations of TI={10,10,278m1007} ms an TD={684,4171,2730,10} ms. Additional parameters included: resolution=2x2x4 mm³, TE=80 ms, TSE factor=26, echo-spacing=5.9 ms, SENSE factor=2.2, and scan time=15-30 minutes (depending on the number of independently acquired slices). To identify the CST, whole-brain DTI data were acquired with multi-slice PGSE sequence and the following parameters: TR/TE=13.5 s/74 ms, b-values={0,1000} s/mm², 45 diffusion directions, SENSE factor=2, resolution=2x2x2 mm³, and scan time=10 mins.

Analysis: All volumes were co-registered to the MP-RAGE volume via FSL (FMRIB Software Library v6.0). SIR parameter maps (PSR and R_1f) were then estimated in MATLAB (2017a) by fitting SIR data with a two-pool model of the MT effect as previously described⁹. The corticospinal tract was manually defined using color-coded FA maps generated with the Medical Image Processing, Analysis, and Visualization (MIPAV v7.3) software package. The resulting CST masks were then overlaid onto the co-registered SIR parameter maps as shown in Fig. 2. Manually defined white matter lesions and black holes were subtracted from these CST masks and mean PSR/R_1f values in normal-appearing CST were estimated for each subject. Linear regressions were performed between SIR parameters and disability scores, and these relationships were assessed via Pearson’s correlation coefficient. In addition, unpaired student t-tests were performed to test for differences in PSR/R_1f between patients and controls.

Results and Discussion

As shown in Fig. 3, significant correlations were observed between SIR-derived R_1f values in the CST and both EDSS (p=0.0016, r²=0.52) and T25FW (p=0.016, r²=0.37). Stronger correlations were observed between PSR values EDSS (p=0.0002, r²=0.72) and T25FW (p=0.001, r²=0.68), suggesting PSR may be a more specific measure of myelin content that R_1f. This is consistent with our comparisons of SIR parameters between multiple sclerosis patients and controls, which found significant differences in PSR values (p=0.049) between patients (10.3 ±1.1%) and controls (11.3 ±1.1%), but not for R_1f values (p=0.26) between patients (0.86 ±0.05 s-1) and controls (0.88 ±0.03 s-1). Note that the reported findings were independent of the exclusion/inclusion of white matter lesions and/or black holes in the CST due to the small volumes of these lesions relative to the overall volume of the normal-appearing CST. The small volume of lesions within the CST of this cohort precluded any asymmetry analysis.

Conclusions

The results herein suggest that SIR-derived measures of PSR in the CST may be viable biomarkers of myelin content and patient disability in MS. R_1f provides complementary information on myelin content, although R_1f is additionally sensitive to inflammation/edema as well as iron accumulation within lesions and macrophages. Future work includes deploying these SIR protocols in larger longitudinal studies to determine the responsiveness of PSR to change over time. If successful, SIR-derived measures, especially PSR, may be able to detect white matter damage and repair earlier than is currently achievable through conventional MRI.

Acknowledgements

R01 NS097821for funding and Brian Thurber for assistance in data analysis.