Sensorimotor disability in multiple sclerosis (MS) is largely affected by damage to the integrity of the spinal cord¹. However, the ability to evaluate spinal cord lesions is limited. For example, conventional magnetic resonance imaging (MRI) methods, such as T1- and T2- weighted imaging, are only sensitive to necrotic or inflammatory lesions. In MS, however, a myriad of other changes are present including demyelination, axonal loss, and gliosis with additional cellular changes that are often undetected in vivo. This lack of sensitivity to microstructural, biochemical changes that may precede inflammatory lesions with standard MRI techniques requires an improved set of imaging biomarkers for evaluating the spinal cord. Chemical exchange saturation transfer (CEST) is a method that indirectly detects mobile species with exchangeable protons, such as the backbone amide protons of mobile proteins and peptides². This derived CEST index, amide proton transfer (APT), may provide information reflective of molecular changes occurring in normal appearing white matter in MS patients. Here, we apply APT CEST in the cervical spinal cord of healthy and MS cohorts to determine the sensitivity to MS pathology at 3T.

Four healthy controls (28±6 years old, 3M/1F) and five relapsing-remitting MS patients (40±7 years old, 2M/3F) were recruited for this study after signed, informed consent. Imaging was acquired using a 3.0T whole body MR scanner (Philips Achieva, Netherlands). A two-channel body coil in multi-transmit mode was used for excitation and a 16-channel SENSE neurovascular coil was used for reception.

A high-resolution (0.65x0.65x5mm³) three echo, multi-echo gradient echo (mFFE) anatomical was acquired in the axial plane for co-registration (TR/TE1/ΔTE=752/7.1/8.8 ms). The CEST sequence consisted of a 3D gradient echo with multi-shot EPI readout (EPI factor=7, SENSE factor RL=2, 20° flip angle, TR/TE=155/12 ms) covering 14 slices from C2-C5 with a resolution of 1x1x5 mm³ over a FOV of 160x160 mm². CEST saturation was achieved with 3 μT pulse (a single 75 ms RF Gaussian pre-pulse), sampling at 47 asymmetric offsets between +/- 10 ppm; two scans with no saturation were acquired for reference. Total scan duration was 12:35 minutes.

All CEST data from each volunteer was registered to the mFFE using a diffeomorphic algorithm in ANTS³. CEST spectra were normalized voxel-by-voxel using the average of the non-saturated reference scans. The APT CEST concentration was quantified using a Lorentzian fit with a baseline offset⁴, which simultaneously corrects for B₀ inhomogeneity (spectra shift to the minimum of the Lorentzian fit) and baseline saturation. The APT effect was quantified as the integrated area between the fit and the z-spectrum from +3.2-3.8 ppm. Region of interests (ROIs) were manually drawn on the anatomical mFFE to assess different tracts of the spinal cord (dorsal column, lateral column, gray matter, whole cord). To test whether APT CEST is sensitive to pathological changes in the cervical spinal cord, a non-parametric Wilcoxon rank sum test was performed for the different tracts. Additionally, we performed a group analysis by co-registering (diffeomorphic) each cohort to each other and averaging the signals across volunteers to produce one map for MS and one map for healthy controls, highlighting the spatial dependencies of the APT effect in the spinal cord.

Figure 1 shows the mean z-spectra of the healthy and MS cohorts derived from the dorsal column. Differences around the APT frequency (Δω=+3.5 ppm) are observed, marked by the dip in the z-spectrum for the healthy controls (red) and a diminished APT effect in the MS patients (blue, p < 0.05). Figure 2 shows a box plot of the mean APT effect for controls (gray) and MS patients (white) in different white matter columns, gray matter and whole cord at the C3/C4 level. Significant differences between MS and healthy controls were seen in the whole cord (p=0.0159), dorsal column (p=0.0159), and gray matter (p=0.0317); a trend towards decreased APT in the MS cohort was observed in the lateral columns (p=0.0635). In comparison to the healthy controls, the APT map of the MS cohort is profoundly lower (Figure 3) relative to the controls. Despite the difficulty in detecting lesions in the mean mFFE of the MS cohort, the differences, especially in the dorsal column, between the mean APT effect in healthy controls (1.88±0.44%) and MS patients (0.90±0.17) indicate the ability to characterize disparity between cohorts. We demonstrate that APT CEST may provide enhanced characterization of normal appearing white matter of the cervical spinal cord in MS patients at 3T. These findings provide insight on the clinical feasibility of APT CEST and motivate further investigation for examination of this technique for evaluation of disease evolution.