Transverse myelitis is a clinical syndrome with many inflammatory and noninflammatory causes. The first step is to rapidly rule out cord compression (MRI), then vascular or connective tissue disease (serology), and then define whether the inflammation is restricted to the cord (brain MRI). The history and exam should exclude connective tissue disease and Behçet disease by focusing on rashes, joint pain, pleuritis, night sweats, shortness of breath, hematuria, anemia, adenopathy, orogenital ulcers, organomegaly, uveitis, and retinitis. Labs should include CBC and differential, CRP, ANA to differentiate disease-related from idiopathic transverse myelitis.
Spinal cord inflammation is a sine qua non for idiopathic acute transverse myelitis, so a Gd-enhanced MRI and CSF analysis strengthen the diagnosis. An MRI of the spinal cord shows increased signal density on T2-weighted images in 50% to 90% of adults (Miller et al 1987; Austin et al 1992; Kalita et al 1998) and 50% to 83% of children (Knebusch et al 1998; Miyazawa et al 2003) presenting with acute transverse myelitis. T1-weighted MRI scans are isointense or mildly hypointense, with swelling of the cord over several segments. Normal MRIs occur in 50% of patients with devastating, localized cord symptoms, suggesting that there are different disease mechanisms or etiologies. Moreover, extensive cord involvement often does not correlate with clinical severity or outcome (Andronikou et al 2003).
Imaging is most fruitful 5 days after onset. The abnormal MRI T2 or FLAIR signal is centrally located or holocord and extends over multiple cord segments (Tartaglino et al 1996; Goh et al 2011). It may reach several segments above the clinically-determined sensory level (Barakos et al 1990; Choi et al 1996). The thoracic cord is typically involved, but some series show cervical predominance (74% of lesions in 20 young French patients) (Dalecky et al 1997). Lupus myelopathy is typically thoracic, Sjögren myelopathy is cervical. Lesions restricted to the conus medullaris can follow virus infections. The spinal cord is sometimes swollen in transverse myelitis (40%), whereas swelling on MRI is rarely seen in multiple sclerosis (Jeffery et al 1993). MRI lesions are all of the same age in acute transverse myelitis, but the signal resolves in some areas before others. More extensive MRI lesions and persistent enhancement after gadolinium infusion herald residual clinical deficits in some cases (Sanders et al 1990; Pardatscher et al 1992), but MRI severity does not always correlate with clinical deficits (Poser 1989). Diffusion tensor imaging shows abnormalities in distal, normal-appearing white matter and may be more sensitive than conventional MRI. Associated brain lesions suggest acute disseminated encephalomyelitis or multiple sclerosis, by definition. However, there is one report of brain MRI abnormalities in 17 of 30 patients and oligoclonal bands in 12 of 25 patients who had "acute transverse myelitis" without clinical signs above the foramen magnum (Miller et al 1987). Patients with CSF abnormalities and brain MRI lesions are more likely to eventually develop multiple sclerosis, as are those with heterogeneous spinal cord involvement on MRI (Simnad et al 1997).
The spinal fluid has elevated protein (often 100 to 120 mg/100 ml; normal less than 50 mg%) and moderate pleocytosis (50 to 100 lymphocytes/mm3) in one third to one half of adult patients (Altrocchi 1963; Lipton and Teasdall 1973; Berman et al 1981; Tippett et al 1991; Austin et al 1992; Jeffery et al 1993) and in up to 80% of children (Miyazawa et al 2003). Glucose and opening pressure are normal (Altrocchi 1963; Ropper and Poskanzer 1978). Seventy percent of children have elevated spinal fluid myelin basic protein (Miyazawa et al 2003). The spinal fluid in transverse myelitis almost always lacks oligoclonal bands and thus differs from the typical picture in multiple sclerosis (Austin et al 1992) (note: with a low-resolution assay) (Jeffery et al 1993). The bands, if present, do not persist (Kesselring et al 1990). IgG (35% to 52%), IgG index (42%, the most specific of these measures of inflammation), and immunoglobulin synthesis rate (33%) are sometimes increased, but not as frequently as in multiple sclerosis (Ropper and Poskanzer 1978; Deuskar et al 1983; Jeffery et al 1993). The axonal 14-3-3 protein is present in only 10% of patients with transverse myelitis and multiple sclerosis (de Seze et al 2002). In 7 patients with acute transverse myelitis, the 4 that had 14-3-3 in their CSF did poorly (Irani and Kerr 2000). In 19 Japanese multiple sclerosis patients, 14-3-3 was present with more damage, progression, and optico-spinal disease (Satoh et al 2003). In another series, none of the 6 patients with transverse myelitis were positive (de Seze et al 2002). High CSF IL-6 predicted recurrent transverse myelitis and disability (Krishnan et al 2004; Kaplin et al 2005). Nonspecific enolase (NSE), myelin basic protein, and S-100 are also potential predictors of severity.
Related syndromes have a different MRI and CSF picture. In acute necrotizing hemorrhagic leukoencephalomyelitis, there is fever and peripheral blood neutrophilia. The CSF is xanthochromic, contains variable amounts of protein and often red blood cells as well as up to 2000 polymorphonuclear lymphocytes. In milder cases the cell count is lower and largely mononuclear. In progressive necrotizing myelopathy, from spinal vascular malformations (Foix and Alajouanine 1926; Mirich et al 1991) or from a remote effect of cancer (Follis and Netsky 1970), the cord is swollen on myelography and the CSF is xanthochromic with high protein (Rowe and Gorman 1986), a normal cell count or a few lymphocytes or polymorphonuclear lymphocytes, and no oligoclonal bands. In acute disseminated encephalomyelitis, MRI lesions are all of the same age and are widespread in brain and cord, often with bilateral optic nerve or basal ganglia lesions. These disappear with the passage of time (Ziegler 1966; Kesselring et al 1990). There is moderate pleocytosis, moderately increased protein, and occasional but transient oligoclonal bands. In chronic progressive myelopathy, 44% have oligoclonal bands, and 44% have abnormal visual evoked potentials (Paty et al 1978), suggesting that many of these patients have multiple sclerosis. In 20 patients with "myelopathy" developing over days to years, 65% had T2-weighted MRIs compatible with multiple sclerosis, and an additional 25% had abnormal visual evoked potentials or CSF oligoclonal bands (Miska et al 1987).
MRI and lumbar puncture, which are sensitive and specific, have largely obviated other tests. Myelograms are usually normal (Lipton and Teasdall 1973; Berman et al 1981) although cord swelling may be present on CT myelography (Rowe and Gorman 1986). CT provides less information than MRI. MRI sometimes (2 of 5 cases) shows cord swelling when myelograms are normal (Barakos et al 1990).
Electrophysiologic tests of spinal cord function are sensitive. Visual and brainstem evoked potentials help exclude CNS dissemination (which would suggest multiple sclerosis) and nerve conduction velocity and in some cases excludes damage to the peripheral nervous system. In parainfectious and idiopathic transverse myelitis, one-fourth of patients have evidence of peripheral damage (eg, abnormal motor unit action potentials, sensory nerve action potentials, and nerve conduction velocities) (Harzheim et al 2004), unlike multiple sclerosis.
There is slowed central motor conduction time to the lower limbs (abnormal in 90%) after magnetic stimulation. Electrically induced motor evoked potentials are even more sensitive. This test can be painful and will be supplanted by painless magnetically evoked potentials. Somatosensory evoked potentials are abnormal in up to 85% (Ropper et al 1982; Wulff 1985; Kalita et al 1998); normal potentials suggest a better prognosis. Absent or reduced sensory action potentials could also indicate peripheral nervous system damage. F waves may be absent on electrodiagnostic testing, and this raises the possibility of Guillain-Barré syndrome. Electromyography of the lower limb muscles showing neuropathic potentials suggests demyelination in the ventral root zone. These indicate a poor outcome (Misra and Kalita 1998), as do abnormalities of peripheral nervous function (Harzheim et al 2004).
NMO-IgG and other serum tests for connective tissue disease are important, especially if cord lesions are contiguous or extend for more than 2 segments.