Spinal cord (SC) attacks in neuromyelitis optica (NMO) are often associated with incomplete improvement, even after plasma exchange (PLEX). MRI parameters may correlate with nadir and recovery EDSS.

We studied the correlation of easily available MRI parameters with nadir and recovery EDSS. Study was focused on index attack in a large monocentric cohort.

This retrospective study was based on West French Indies and included all Afro-Caribbean NMO-SD patients in their index SC attack. Exclusion criteria were: unavailable MRI or clinical data; SC relapse in the next 6 months. Since the study was dedicated to SC, optical system was excluded from EDSS calculation. MRI scans were performed on ≥1T magnet, included sagittal T1, T2 and post-contrast T1 images. Lesion length was approximated to number of adjacent vertebral bodies.

In bivariate analysis, nadir and residual EDSS correlated with lesion length, oedema and hypo-T1 lesion. No correlation was found with gadolinium uptake or PLEX treatment. In multivariate analysis, nadir EDSS correlated only with lesion length (P=0.02) and œdema (P=0.01). Residual EDSS correlated strongly with nadir EDSS (P<0.001) and short PLEX delay (≤10 days; P=0.003) but not with MRI parameters. After exclusion of PLEX from model, high residual EDSS correlated only with hypoT1 (P=0.03).

Expectedly, nadir EDSS was the main predictor of residual EDSS. Simple MRI parameters correlated with nadir and recovery EDSS, but in multivariate models, hypo-T1 lesion was associated to poor residual outcome only after exclusion of PLEX. Early PLEX were still associated with an improved outcome although most PLEX were started late (median 17 days).

MRI features for severe SC lesion are associated with clinical severity, but residual impairment is mainly predicted by nadir EDSS and could be lowered by early PLEX.