Table 2.

MRI studies evaluating the relationship between spinal cord atrophy and disability.

Reference	Patients (n)	MRI measure of neurodegeneration [acquisition/quantification methods]	Measure of disability	Results
CROSS-SECTIONAL STUDIES
Rocca et al. (95)	335 with MS	Cervical cord CSA (C2 to C5) [3D T1w/active surface method]	EDSS	Cord CSA correlated with EDSS in patients with RRMS (r = −0.30; p = 0.001), SPMS (r = −0.34; p = 0.001), and PPMS (r = −0.27; p = 0.01), but not in patients with CIS or benign MS
Biberacher et al. (91)	267 with CIS or RRMS	Upper cervical cord CSA at C2/C3 [3D T1w/FSL software]	EDSS	Cord CSA correlated with EDSS (r = −0.131; p = 0.044)
Weier et al. (102)	202 with MS	Whole spinal cord atrophy [T2w/visual assessment]	EDSS	Weak correlation between cord atrophy and EDSS scores (r = 0.30)
Daams et al. (92)	196 with MS	Upper cervical cord CSA [3D T1w/semi-automated method]	EDSS T25FW 9HPT Cord Functional Score	Cord area was independently associated with EDSS (r = −0.296; p < 0.001), T25FW (r = 0.240; p = 0.001), and 9HPT (r = −0.206; p = 0.005)
Bernitsas et al. (90)	150 with MS	Cervical cord CSA (C2) [3D T1w/Losseff semi-automated method (104)]	EDSS	Significant correlation between CSA-C2 and EDSS (r = −0.75; p < 0.0001) Multivariable regression showed that CSA-C2 was a significant predictor of disability independent of disease duration and phenotype (p < 0.0001)
Oh et al. (94)	133 with MS	C3–C4 cord volume [3D T1w/fully automated segmentation protocol (105)]	EDSS MSFC Hip flexion strength Vibration sensation threshold	Correlations between clinical measures (EDSS: r = −0.20, p = 0.02; MSFC: r = 0.16, p = 0.06; hip flexion strength: r = 0.35, p = 0.0001; vibration threshold: r = −0.19, p = 0.03) and cord volume
Yiannakas et al. (99)	120 with MS (40 in longitudinal subgroup; 1-year follow-up)	Cervical cord CSA (two segments: C2/C3 and C2/C5) [3D T1w/Propseg vs semi-automated active surface method]	EDSS MSFC T25FW 9HPT ASIA motor and sensory scores	Baseline CSA was significantly associated with baseline clinical variables (both segments) (p < 0.001 for all) CSA measures at 1 year were significantly associated with ASIA motor and sensory scores only (p = 0.048 to p = 0.001) Baseline CSA for both segments predicted ASIA motor scores at 1 year (p ≤ 0.003)
Schlaeger et al. (96)	113 with MS	Spinal cord WM area (C2/C3) Spinal cord GM area (C2/C3) Upper cervical cord CSA (C2/C3) [2D PSIR/Active surface method]	EDSS T25FW 9HPT	GM, WM, and cord CSA significantly correlated with EDSS (r = −0.60, −0.32, and −0.42, respectively; all p ≤ 0.001) and T25FW (r = −0.50, −0.28, and −0.36, respectively; p < 0.001, p = 0.004 and p < 0.001, respectively) GM area (r = −0.37) and cord CSA (r = −0.22) significantly correlated with 9HPT (p < 0.001 and p = 0.024, respectively) GM area was the strongest correlate of disability in multivariate models
Rocca et al. (106)	77 with MS	Regional cervical cord atrophy (voxel-based) [3D T1w/voxel-based analysis, active surface method]	EDSS FSS	SPMS: cord atrophy at C1/C2 correlated with pyramidal FSS (r = −0.91; p < 0.001) PPMS: cord atrophy at C1/C2 correlated with EDSS (r = −0.68) and pyramidal FSS (r = −0.89) (p < 0.001) No correlation between regional cord atrophy and clinical variables for other MS phenotypes
Valsasina et al. (98)	71 with RRMS or SPMS	Regional cervical cord atrophy [3D T1w/voxel-based analysis, active surface method]	EDSS	Regional cervical cord atrophy was correlated with clinical disability (r = −0.46 to −0.57; p < 0.001)
Benedetti et al. (100)	68 with benign MS or SPMS	Cervical cord CSA [3D T1w/semi-automated method of Losseff (104)]	EDSS	Cord CSA was an independent predictor of EDSS (p = 0.001)
Horsfield et al. (93)	40 with RRMS or SPMS	Cervical cord CSA (C2 and C2–C5) [3D T1w/semiautomatic active surface vs Losseff method (104)]	EDSS Ambulation index	Strong correlations between the EDSS (C2: r = −0.51; C2–C5: r = −0.59) and ambulation index (C2: r = −0.58; C2–C5: r = −0.648) and CSA (p < 0.001)
Healy et al. (101)	34 with MS	C2–3 volume Cervical cord volume Thoracic cord volume Whole cord volume [T2-weighted sequence/JIM software]	EDSS	C2–3 volume and cervical cord volume correlated with EDSS score (p < 0.05)
Song et al. (97)	29 with MS	Upper cervical cord CSA [3D T1w and T2w/semi-automated software (107)]	EDSS	Stronger correlation between EDSS and normalized measurement of cord area vs absolute measurement [r = −0.84 (p < 0.01) vs r = −0.46 (p < 0.05)]
Blamire et al. (103)	11	Spinal cord CSA (C2–C5) [T1w/Jim software]	EDSS 9PHT T25FW	No correlation between cord atrophy and measures of disability
LONGITUDINAL STUDIES
Valsasina et al. (89)	35 with MS (mean follow-up, 2.3 years)	Cervical cord CSA [3D T1w/active surface method vs Losseff method]	EDSS	At baseline, there was a significant correlation between EDSS and both methods used to measure CSA (AS method: r = −0.59; p < 0.001; Losseff method: r = −0.40; p = 0.01) At follow-up, AS cord CSA (but not CSA evaluated using the Losseff method) correlated with EDSS (r = −0.50; p = 0.002)

Studies within each subsection are ordered according to size of patient population.

9HPT, 9-hole peg test; AS, active surface; ASIA, American Spinal Injury Association; CIS, clinically isolated syndrome; CSA, cross-sectional area; EDSS, Expanded Disability Status Scale; FSS, Functional Scale Score; GM, grey matter; MRI, magnetic resonance imaging; MS, multiple sclerosis; MSFC, Multiple Sclerosis Functional Composite; RRMS, relapsing-remitting multiple sclerosis; SPMS, secondary progressive multiple sclerosis; T1w, T1-weighted; T2w, T2-weighted; T25FW, timed 25-foot walk; WM, white matter.