Transverse lesions of the spinal cord were attributed to thrombosis caused by arteriosclerosis, syphilis, and other infections by Bastian in the 1880s (Follis and Netsky 1970). Foix and Alajouanine also believed that transverse myelitis had a vascular etiology (Foix and Alajouanine 1926). They described 2 patients with subacute necrosis in the sacral or thoracolumbosacral cord associated with massive dilatation and endomysial hypertrophy of the extramedullary veins and similar but less marked changes in the intramedullary vessels. The lumens were not obliterated, and the arteries were not involved. The term "angiodysgenetic necrotizing myelopathy" is sometimes applied to this condition (Follis and Netsky 1970). A similar syndrome is believed to be caused by spinal dural arteriovenous fistulae, although Mirich and colleagues found no arteriovenous malformations in 4 typical cases of subacute necrotizing myelopathy (Mirich et al 1991). Paine and Byers were the first to report a large clinical series with follow-up exams and used the term “transverse myelopathy” (Paine and Byers 1953). They also postulated a vascular cause but did not present any pathological evidence. In most of these early reports, 1 or 2 patients were studied, and the autopsies were performed months to years after the original illness (Jaffe and Freeman 1943).
In more recent series, autopsies from patients with acute transverse myelitis seldom show significant vascular alterations. Lipton's series of 34 contained only 2 cases with infarcts of unknown etiology and one with hemorrhage plus telangiectasias (Lipton and Teasdall 1973).
Other investigators assumed the condition was an inflammatory myelitis from an antigen-antibody reaction (Plum and Olson 1973) or a hypersensitivity reaction to infection or vaccination (Moersch and Kernohan 1934). Histology showed demyelination, variable destruction of axons, and differing degrees of mononuclear cell infiltration. (There was loss of myelin in the cases of Foix and Alajouanine, but this does not prove that demyelination was the primary event.)
Several large series appeared in the 1960s and 1970s (Altrocchi 1963; Lipton and Teasdall 1973; Ropper and Poskanzer 1978; Berman et al 1981). The consensus that evolved was that acute transverse myelitis was an inflammatory demyelinating disease of the cord. It sometimes followed infections, but usually it had no antecedent. In a series of 33 cases of acute or subacute noncompressive myelopathy, 46% were parainfectious, 12% were associated with cord ischemia, 21% were multiple sclerosis exacerbations, and 21% were idiopathic (Jeffery et al 1993). The underlying infection is seldom diagnosed.
Many types of cord lesion or systemic disease can cause a transverse spinal cord syndrome or myelopathy. Etiologies include infectious, neoplastic, compressive, traumatic, embolic, degenerative, or vitamin-substrate deficiency. However, acute transverse myelitis, as defined here, is an idiopathic disorder caused by inflammation of the spinal cord, associated with marked demyelination and often with significant axonal loss. The process that instigates the inflammation in transverse myelitis is unknown. (This includes the third of transverse myelitis patients with antecedent virus infections.)
Specific diagnostic criteria for transverse myelitis have been proposed (Group TMCW 2002). These include (1) neurologic symptoms attributable to the spinal cord, ie, negative brain MRI and no clinical brain symptoms, (2) bilateral signs or symptoms, (3) clearly defined sensory level, (4) exclusion of compression by MRI, (5) inflammation in the spinal cord (CSF cells or IgG index, or MRI gadolinium enhancement) seen at onset or within 7 days, (6) progressive worsening to maximum from 4 hours to 21 days after onset, and (7) no other etiology. With these strict criteria, only 16% of 288 patients with acute transverse myelitis from all causes have idiopathic acute transverse myelitis (de Seze et al 2005). Scott and colleagues argue that evidence of inflammation is less common in transverse myelitis than in multiple sclerosis and should not be used as a criterion (Scott et al 2005). Acute transverse myelitis can be split into complete or partial transverse myelitis, each with a different outcome (Scott 2007).
These definitions are undergoing revision at this time. A subset of transverse myelitis has been associated with antibodies against aquaporin-4 (“NMO-IgG”), and another subgroup is associated with Sjögren disease (Javed et al 2008).
In the following discussion, idiopathic transverse myelitis is assumed not to have a known infectious etiology.