Spinal Cord Neuropathology in Human West NileVirus Infection

West Nile virus (WNV) is a flavivirus within the Japanese encephalitis serocomplex, which includes St Louis encephalitis, yellow fever, and dengue fever. Most human infections with WNV are clinically inapparent.1,2 Approximately 1 in 5 infected persons develops mild febrile illness, and 1 in 150 develops more severe neurologic disease, including meningitis or encephalitis.1–4 Muscle weakness is a prominent feature in patients with WNV infection. In the New York City outbreak of 1999, more than half of the patients with confirmed WNV encephalitis displayed severe muscle weakness as a cardinal clinical manifestation.4 Weakness was also an apparent risk factor predicting death in patients with WNV encephalitis.4,5 In the 2002 WNV epidemic in the United States, muscle weakness remained a well-recognized feature associated with increased morbidity and mortality. Several case series initially attributed the complication of muscle weakness to a peripheral nerve process, namely Guillain-Barré syndrome, motor axonopathy, or severe axonal polyneuropathy.4,6,7 However, closer scrutiny of such cases, in conjunction with recent clinical, laboratory, and electrophysiologic findings in additional patients with WNV-associated acute flaccid paralysis, has suggested that weakness was due to involvement of anterior horn cells of the spinal cord, causing a poliolike syndrome.8–13 

Pathologic data now suggest that poliomyelitis is a cause of WNV-associated muscle weakness and acute flaccid paralysis.14–16 However, pathologic investigation of the spinal cord in WNV infection remains in its infancy, particularly with respect to clinicopathologic correlations. This article describes the detailed spinal cord pathology in 4 confirmed fatalities from WNV infection that occurred during the epidemic of 2002 in Mississippi.

Among the 193 confirmed cases of WNV in the state of Mississippi during the epidemic year of 2002, 12 deaths were recorded. Four of these patients had postmortem examinations at the University of Mississippi Medical Center; all were men, ranging in age from 62 to 83 years (Table 1). Presenting signs and symptoms varied, but each man developed muscle weakness and respiratory distress that required intubation and assisted ventilation. The intervals from onset of symptoms to death in these 4 patients were 16, 60, 30, and 90 days.

Selected areas of brain from these 4 cases were routinely processed and embedded in paraffin, and 4-μm sections were stained with hematoxylin-eosin. In each case, these sections included frontal cortex, basal ganglia, thalamus, hippocampus, midbrain, pons, medulla, and cerebellar hemisphere at the level of the dentate nucleus. Sections of cervical, thoracic, and lumbar spinal cord were also submitted, along with dorsal root and sympathetic ganglia. Selected sections were incubated with the following antibodies: antibody to a pan-leukocyte antigen (anti- CD45 or leukocyte common antigen, Ventana Medical Systems, Tucson, Ariz), to B cells (CD20, Ventana), to T cells (CD45RO or CD3, Cell Marque Corporation, Hot Springs, Ark), and to macrophages/microglial cells (CD68, Cell Marque). The leptomeninges and parenchyma were inspected for perivascular inflammation and microglial cells, proceeding from rostral to caudal regions of the central nervous system. The relative intensity of the inflammatory reaction was graded using a 3-point scale: mild (+), indicated by a sparse perivascular leptomeningeal lymphocytic infiltrate or widely separated microglial nodules in the parenchyma; moderate (++), showing increased density of perivascular lymphocytes or microglial nodules in brain tissue; and severe (+++), equivalent to confluent sheets of microglia and chronic inflammatory cells. Supratentorial and infratentorial sites were compared in terms of the inflammatory reaction, as well as cervical, thoracic, and lumbosacral spinal cord levels.

No gross lesions were detected in the 4 formalin-fixed brains and spinal cords.

In general, signs of focal or diffuse inflammation superior to the spinal cord were mild. Lesions in the cerebral hemispheres and brainstem included perivascular lymphocytes in leptomeninges and parenchyma, with scattered microglial nodules in gray and white matter. In cases 3 and 4, the inflammatory nodules were more focal and were concentrated in the brainstem. The cerebellum showed only sporadic, low-grade meningeal lymphocytosis in 3 of the 4 cases. In case 3, however, we noted Purkinje neuronal drop-out, gliosis, and proliferation of microglial nodules in the cerebellar folia.

In contrast to the mild inflammation rostral to the spinal cord, the gray matter of the cervical and lumbar regions was most severely affected in all 4 patients (Table 2). The spectrum of neuronal damage included occasional chromatolytic neurons (Figure 1), neuronophagia (Figure 2), and a feltwork of proliferating astrocytes that indicated places where neurons had dropped out. Cuffs of lymphocytes surrounded blood vessels in spinal gray matter, sometimes extending into the adjacent white matter (Figure 3). Immunohistochemical preparations, especially CD68 and CD45, highlighted the magnitude of the inflammatory response around vessels and within the substance of the gray matter (Figure 4). Antibodies to B and T cells indicated the predominance of T cells in the immunologic reaction to the virus (Figure 5). Microglial cells, often grouped, were stained by the CD68 marker as they permeated the ventral gray matter of the lumbar cord.

Sections of dorsal root and sympathetic ganglia documented the focal loss of ganglionic neurons. Satellite cells in the dorsal root ganglion aggregated at sites of neuronal disappearance (Nageotte nodules, case 2, Figure 6). Microglial nodules clustered around the eosinophilic husk of once-viable ganglion cells in cervical sympathetic ganglia (case 4, Figure 7). The cellular reaction in the ganglia was patchy, and typically the magnitude of the inflammatory response correlated with the severity of the poliomyelitis. Secondary demyelination was not observed in spinal nerve roots.

Recent clinical and electrodiagnostic evidence has suggested that most patients developing profound muscle weakness in the setting of acute WNV infection suffer from damage to spinal anterior horn cells, resulting in a poliomyelitis-like syndrome.8–11 Although many health professionals feel that the term poliomyelitis should refer exclusively to disease caused by the poliovirus, the term poliomyelitis, as well as the clinical and pathologic features of the disease, existed long before the discovery of the poliovirus.17 Today we recognize that many viruses, including enteroviruses, echoviruses, Coxsackie viruses, and flaviviruses can cause the clinical syndrome of poliomyelitis. The electrophysiologic, spinal fluid, and histopathologic findings may also be identical. Differentiation must therefore be made on the basis of serological studies or virus isolation. Thus, when we describe spinal cord lesions, inflammation (itis) of the spinal cord (myelos) gray matter (polios) is correctly referred to as poliomyelitis, irrespective of the cause. The current autopsy series provides pathologic support for this concept and demonstrates that the spinal cord gray matter is a major site of involvement in patients with WNV-associated muscular weakness and respiratory distress. In our series, all 4 patients developed poliomyelitis and experienced profound muscular weakness and respiratory distress requiring endotracheal intubation and mechanical ventilation. As in previous pathologic reports of WNV infection,7,16 we found no exudates in the leptomeninges and no gross lesions on external examination of the brain. Similarly, no gross lesions were found in the spinal cord, in which gray and white matter were clearly demarcated. However, by light microscopy, profound alterations in the ventral gray matter of all 4 spinal cords confirmed the tropism of WNV for anterior horn cells. Pathologic findings included cell loss, gliosis, vacuolization, microglial cell proliferation, and perivascular lymphocytic infiltrates; the most severely affected regions were the gray matter of the cervical and lumbar cord. In our series, all patients were in their seventh decade of life, or older, supporting the concept that elderly patients are at greatest risk for morbidity and mortality from WNV.4–6 

The current autopsy series extends previous descriptions of lesions associated with WNV infection and illustrates that neuronophagia, neuronal disappearance, and satellite cell proliferation may not be limited to anterior horn cells or spinal gray matter. Pathologic alterations were also evident in sympathetic and dorsal root ganglia of all 4 patients, with the magnitude of the inflammatory reaction roughly correlating with the severity of the poliomyelitis. However, the degree of inflammation was generally less than that in gray matter of the cervical and lumbar cord. Since the sympathetic and dorsal root ganglia lie outside the confines of the spinal cord, the boundary of gray matter involvement affected by WNV must be enlarged. Based on our observations, we might predict that, in a subset of patients, dorsal root ganglia lesions may be severe enough to produce sensory deficits. Although sensory loss attributed to WNV infection has not been a prominent clinical finding to date, there have been reports of reduced sensory nerve action potentials elicited by objective electrophysiologic tests.2,13,16,18 Damage to dorsal horn neurons by WNV was previously observed only in vitro.19 In addition, the relative disappearance of sympathetic neurons, as reported here, could lead to autonomic instability. Such instability has been observed in a minority of WNV patients, including labile vital signs, hypotension, and potentially lethal cardiac arrhythmia (A.A.L., unpublished data, 2002).20 

The authors gratefully acknowledge the inspiration and assistance of Theresa Harrington, MD, MPH, who is currently working at the Mississippi State Department of Health, Jackson, Miss.