Myelitis is a frequent manifestation of aquaporin-4 immunoglobulin G–seropositive (AQP4-IgG+) neuromyelitis optica spectrum disorder (NMOSD). A characteristic imaging feature is longitudinally extensive transverse myelitis (LETM). Additional MRI observations reported in NMOSD include bright spotty lesions (BSLs) and a novel imaging pattern described as the double-contoured LETM sign. The prevalence and diagnostic value of these established and newer spinal MRI signs had not been fully characterized.
This retrospective study sought to describe spinal cord MRI findings across disease stages in adults with AQP4-IgG+ NMOSD myelitis and to evaluate the diagnostic performance (sensitivity, specificity, likelihood ratios) of several specific MRI signs compared with other causes of myelopathy.
Researchers reviewed spinal MRI scans from a university-based hospital cohort of adults with AQP4-IgG+ NMOSD. The overall dataset included 217 patients (median age at onset 44.5 years; 95.9% women), of whom 170 (78.3%) experienced myelitis. In total, 331 NMOSD spinal MRI scans were analyzed. Scans were categorized by timing relative to clinical attack: acute (<30 days), subacute (1–3 months), and remission (>3 months).
A comparison group comprised 72 spinal MRI scans from patients with other causes of myelopathy; the most common diagnoses in that group were spinal cord infarction (39 scans) and idiopathic transverse myelitis (14 scans).
Acute scans were specifically assessed for a set of MRI patterns previously described in NMOSD or proposed as novel imaging markers. These included:
The study performed descriptive analyses of MRI features across disease stages and calculated diagnostic test characteristics for the acute-phase signs when comparing NMOSD myelitis with the control myelopathies.
When analysis was restricted to first-ever myelitis episodes (53 acute scans), LETM was observed in 88.7% of cases. For other acute MRI signs in this subgroup, the study reported the following diagnostic metrics:
Combining the above signs increased overall specificity further but reduced sensitivity to approximately 30%, indicating that combined criteria are more useful to rule in NMOSD myelitis than to screen it out.
When acute scans from both first-ever and recurrent myelitis episodes were analyzed together (161 scans), the prevalence of the three key signs decreased compared with the first-ever only analysis. Reported prevalences in this combined acute group were:
These lower prevalences when including recurrent attacks suggest that some MRI features are more prominent early in disease or during initial presentations.
Serial imaging across stages showed that radiologic outcomes evolved over time. Among 129 scans obtained during remission (>3 months), only 12.4% demonstrated complete radiologic resolution. Conversely, spinal cord atrophy was common in the remission setting, found in 56.6% of these scans. This indicates a high rate of persistent structural change after NMOSD myelitis.
The control cohort used for specificity estimates was enriched for noninflammatory myelopathies, most notably spinal cord infarction and idiopathic transverse myelitis. Because the comparison group contained a high proportion of noninflammatory causes, specificity estimates for the MRI signs reflect differentiation primarily from those conditions. The authors note this enrichment as a factor that may limit the generalizability of specificity values to broader clinical populations with inflammatory myelopathies.
Key conclusions from the study are that BSLs, the double-contoured LETM sign, and sagittal ring enhancement each demonstrate high specificity for AQP4-IgG+ NMOSD myelitis and can aid differentiation from other causes of longitudinally extensive myelopathy. The double-contoured LETM sign is presented as a novel imaging feature, particularly relevant in first-ever acute myelitis.
However, sensitivities were moderate to modest, and combining signs increases specificity at the cost of sensitivity. The enriched comparison cohort (many noninflammatory myelopathies) may limit how broadly the specificity estimates apply; the authors state that further validation in independent and more diverse cohorts is warranted.
Overall, the findings support the value of several spinal MRI signs for increasing diagnostic confidence when NMOSD is suspected, while underscoring that absence of these signs does not exclude disease and that imaging evolution over time commonly results in residual cord atrophy.