Serum neurofilament to Magnetic Resonance Imaging lesion area ratio differentiates spinal cord infarction from acute myelitis

Background and purpose: The diagnosis of spontaneous spinal cord infarction (SCI) is limited by the lack of diagnostic biomarkers and MRI features that often overlap with those of other myelopathies, especially acute myelitis. We investigated whether the ratio between serum neurofilament light chain levels and MRI T2-lesion area (neurofilament light chain/area ratio-NAR) differentiates SCI from acute myelitis of similar severity. Methods: We retrospectively identified Mayo Clinic patients (January 1, 2000-December 31, 2019) with (1) SCI, (2) AQP4 (aquaporin 4)-IgG or MOG (myelin oligodendrocyte glycoprotein)-IgG-associated myelitis at disease clinical presentation, or (3) idiopathic transverse myelitis from a previously identified population-based cohort of patients seronegative for AQP4-IgG and MOG-IgG. Serum neurofilament light chain levels (pg/mL) were assessed at the Verona University (SIMOA, Quanterix) in a blinded fashion on available stored samples obtained ≤3 months from myelopathy presentation. For each patient, the largest spinal cord lesion area (mm2) was manually outlined by 2 independent raters on sagittal T2-weighted MRI images, and the mean value was used to determine NAR (pg/[mL·mm2]). Results: Forty-eight patients were included SCI, 20 (definite, 11; probable, 6; possible, 3); acute myelitis, 28 (AQP4-IgG-associated, 17; MOG-IgG-associated, 5; idiopathic transverse myelitis, 6). The median expanded disability status scale score (range) at myelopathy nadir were 7.75 (2-8.5) and 5.5 (2-8), respectively. Serum neurofilament light chain levels (median [range] pg/mL) in patients with SCI (188 [14.3-2793.4]) were significantly higher compared with patients with AQP4-IgG-associated myelitis (37 [0.8-6942.9]), MOG-IgG-associated myelitis (45.8 [4-283.8]), and idiopathic transverse myelitis (15.6 [0.9-217.8]); P=0.01. NAR showed the highest accuracy for identification of SCI versus acute myelitis with values ≥0.35 pg/(mL·mm2) yielding 86% specificity and 95% sensitivity (area under the curve=0.93). The positive and negative likelihood ratios were 6.67 and 0.06, respectively. NAR remained independently associated with SCI after adjusting for age, gender, immunotherapy before sampling, and days from myelopathy symptoms onset to sampling (P=0.0007). Conclusions: NAR is a novel and promising clinical biomarker for differentiation of SCI from acute myelitis.