Column-specific spinal cord magnetization transfer, not T2w, imaging relates to sensorimotor function in multiple sclerosis

Background: Magnetic resonance imaging (MRI), such as T1 and T2*w, assists in diagnosing multiple sclerosis (MS) and identifying new lesions. However, the correlation with neurological deficits is imperfect, perhaps because it poorly characterizes underlying microstructural damage. Magnetization transfer (MT) MRI allows quantification of white matter columns of the spinal cord. Objective: To demonstrate that column-specific MT MRI correlates better with sensorimotor dysfunction in MS than T2*w MRI. Methods: Forty MS subjects underwent 3 Tesla MRI and quantitative measurements of sensorimotor function. From axial MTw cervical spinal cord (C2-C6) images, we calculated the cerebrospinal fluid (CSF)-normalized MT (MTCSF) signal intensity as well as the T2*w signal in the dorsal column (DC) and lateral column (LC) white matter. We evaluated sensorimotor function using walking, ankle strength, standing balance and vibration sensation tests. Results: The MTCSF in the LC was significantly correlated with strength (R=-.34, p=0.04) and walking speed (R=-.43, p=.004); MTCSF in the DC was significantly correlated with sensation (R=.48, p=0.002) and balance (R=.39, p=.02). By contrast, T2*w in the LC and DCs were far less correlated with these impairments (LC: R=.03, p=.86, R=-.08, p=.65; DC: R=.26, p=.11, R=-.27, p=.08). Progressive MS subtype was a strong predictor of sensorimotor impairment. When subtypes were pooled, a stepwise multiple regression showed MTCSF of the LC to be the significant predictor of strength (R2=0.13), and the DC to be the significant contributor of sensation (R2=0.23). Conclusions: Column-specific spinal cord MTCSF, but not T2*w, imaging results were significantly related to sensorimotor impairments in individuals with MS. We suspect this is because MTCSF is sensitive to myelin health, whereas T2*w MRI reports on inflammation. This suggests that it is the myelin damage that correlates with neurological deficits. Since the spinal cord is somatotopically organized, the correlations presented here further indicate that these sensorimotor measures may be used as surrogate markers of spinal cord damage.