West Nile virus infection

Methods.

We reviewed the medical records of patients hospitalized with CNS WNV infection at The Cleveland Clinic Foundation in August and September 2002. Diagnosis was based on either of the following criteria: 1) IgM positivity in the CSF using the capture ELISA or immunofluorescence assay techniques (Focus Technologies, Cypress, CA); or 2) high simultaneous serum IgG (>0.9 enzyme immunoassay [EIA]) and IgM (>2.0 EIA) titers. Positive results were confirmed by the Ohio Department of Health. Charts were analyzed for demographic variables and clinical characteristics. A symptom was considered present only if specifically recorded. For each patient, the clinical syndrome was defined as meningitis, encephalitis, weakness, or any combination of the above. Weakness was characterized by reviewing the recorded neurologic examinations. Diagnostic laboratory tests, including blood, CSF findings, available neuroimaging, and electrodiagnostic and pathologic studies of brain and spine, were assessed. Peripheral nerves were not examined.

The χ² contingency test was used to study relations between weakness and the recorded variables.

Results.

Radiologic findings.

In 19 patients who had a CT scan of the brain, only 1 was abnormal with an acute posterior cerebral artery territory infarct. Brain MRI (T1, T2, fluid-attenuated inversion recovery [FLAIR], and diffusion-weighted imaging [DWI]), performed on 18 of 23 cases, was abnormal in six. Table E-3 (at www.neurology.org) summarizes the radiologic findings that included abnormalities either involving cortex or brainstem in isolation or cortex plus underlying subcortical white matter (figure 1, A through D). The location (bilateral temporal and subcortical frontoparietal) and enhancement of the white matter abnormalities were atypical for ischemic lesions.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. MRI of brain showed T2 signal hyperintensities (A, B) that progressed over the course of 1 week in the brainstem of one patient. They also included subcortical white matter changes best seen on T2 (C) and gyriform areas of restricted diffusion (D). MRI of spine showed gadolinium-enhancing lesions involving the cervical cord (E), conus (F), and cauda equina (G), as shown by the arrows.

Eight of 11 patients with weakness had a spine MRI. Three MRI showed signal changes involving the parenchyma (conus or cervical cord), cauda equina, or both (see figure 1, E through G), usually seen on FLAIR and often enhancing.

Pathology results.

Autopsies were performed on brain in three patients and spinal cord in two. Mild diffuse cerebral edema was seen in one. All brains showed perivascular chronic inflammation involving the meninges and parenchymal vessels with microglial cell proliferation, gliosis, and neuronophagia. Preferential involvement of the brainstem was observed in one. Similar inflammatory and reactive changes were observed in the anterior horn cell (AHC) region in the two spinal cords (figure 2). Focal inflammatory changes also involved the lumbosacral nerve roots (figure 3). Details of one case are published elsewhere.3

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Perivascular chronic inflammation and loss of anterior horn cells in the lumbar spinal cord seen in Case 10. Hematoxylin and eosin; original magnification ×200.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Lumbar cord nerve root segment with chronic inflammation seen in Case 10. Hematoxylin and eosin; original magnification ×200.

Discussion.

WNV is predominantly a mosquito-borne disease with peak transmission in late summer and early fall months.1,4⇓ Cases of probable transmission through breastfeeding and organ donation were recently reported.5,6⇓ One of our cases was transfusion related.

Most WNV infections are subclinical, with overt disease occurring in 1 of every 100 to 150 patients.7,8⇓ The early nonspecific symptoms, including fever, gastrointestinal complaints, and back or limb pain, can be seen in many viral infections, including polioviruses. The incidence of rash (26%) in our study is higher than the 19% reported previously.7 Its presence is nonspecific but may point to a viral exanthem and may be compatible with a WNV infection if resolving prior to neurologic manifestations.

The observed hyponatremia is nonspecific and has little diagnostic value late in the infection, but should be looked for as a potential complicating factor.

We noted a stable CSF protein elevation and pleocytosis with initial neutrophil followed by lymphocyte predominance. In polio, the protein concentration also gradually rises to as much as 400 mg/dL and the initial CSF pleocytosis tends to resolve after the first week of paralysis.9 CSF pleocytosis is incompatible with Guillain–Barré syndrome (GBS), a leading differential diagnostic possibility in patients with progressive, diffuse, flaccid weakness. Traditionally, the differential diagnosis of a GBS-like syndrome associated with CSF pleocytosis has included Lyme disease, neoplasia, HIV, and sarcoid meningitis.10 Our data support adding WNV to this list.

Prior reports about WNV infection have emphasized meningoencephalitis with less frequent discussion of the associated weakness.7,8,10,11⇓⇓⇓ The following conclusions can be drawn from reviewing the data in our patients with weakness:

1. Clinically, WNV seems to cause a wide spectrum of motor manifestations ranging from flaccid monoplegia without meningitis or encephalitis to disabling quadriplegia. Rash and abdominal and back pain preceding a proximal asymmetric flaccid weakness with fever may help distinguish the motor syndrome from GBS.

2. Electrodiagnostically, there is no evidence of demyelination, as has been reported by others.1 In one presentation, severe widespread motor axonal loss with normal sensory amplitudes and evidence of reinnervation on subsequent follow-up (Cases 17 and 19) is most consistent with an AHC process. In the two patients with selective motor involvement, the femoral nerve was more affected than the tibial or peroneal nerves, suggesting a more proximal distribution. As illustrated by Case 17, denervation changes may extend to clinically unaffected contralateral muscles. In a second group of patients (Cases 4 and 14), a uniform, more diffuse reduction of motor response amplitudes was found as well as severely reduced sensory amplitudes compatible with involvement of the dorsal root ganglia (DRG) or peripheral axons. Similar findings have previously been interpreted as suggestive of a mixed sensorimotor polyneuropathy.1,7⇓ Our pathologic data, however, favor considering a myelitis affecting not only the AHC but extending to the sensory roots and perhaps DRG as well.

WNV-related weakness has generally been attributed to a GBS-like syndrome with axonal polyneuropathy.1,7⇓ A case reported in 1979 described an “acute anterior myelitis complicating West Nile fever” in a patient with fever, back pain, and left leg flaccid paralysis,12 and more recent case reports also suggested a poliomyelitis–like syndrome from WNV.13,14⇓ It may therefore be useful to compare the WNV and poliomyelitis based on our data. According to Sabin, “paralytic poliomyelitis can now be regarded as a clinical–pathologic syndrome that is caused by enteroviruses, consisting of the three types of polioviruses and probably 19 other enteroviruses.”9

Clinically, both poliovirus and WNV cause a rapidly progressive febrile paralytic illness with an asymmetric and selective proximal pattern. The encephalopathic changes in WNV are, however, incompatible with poliovirus myelitis.9

Pathologically, the typical changes of poliovirus include hyperemia and edema with perivascular lymphocytic infiltration in brain preferentially involving the precentral gyrus, hypothalamus, reticular formation, quadrigeminal bodies, and spinal cord, compressing the AHC and interstitial tissues mostly at the cervical and lumbar cord levels. This is usually followed by chromatolysis in the AHC and cell death. Pathologic changes affecting the sensory roots and DRG were observed in patients with poliovirus infection and were thought to partially explain back and radicular pain seen early in the disease.15,16⇓ We confirmed similar pathologic findings, especially AHC and radicular inflammation, in our autopsies. Unfortunately, neither of the two patients whose cord was examined had electrodiagnostic testing, and only one had a spine MRI showing an enhancing cauda equina (see figure 1F).

In animal studies, intracerebral inoculation with the WNV causes necrosis in the pyramidal cells of the hippocampus and the dentate gyrus, in the AHC, and less frequently in the olfactory bulbs, cerebral cortex, granular cells of the cerebellum, and habenular nucleus.17 Of the prior human autopsies of WNV patients, scattered microglial nodules and perivascular and perineural inflammation, most marked in the medulla, were observed,3,18⇓ again similar to our findings.

Based on current criteria, electrodiagnostic sensory abnormalities are incompatible with the diagnosis of poliomyelitis. Without further investigations, it cannot be determined whether the two different electrodiagnostic patterns described in our WNV cases actually represent independent involvement of motor and sensory axons or whether they reflect only variability in the severity of the myeloradiculopathy.

Although based on a limited number of patients, we note that the only two patients with flaccid paralysis who did not require long-term mechanical ventilation were those with isolated motor involvement.

The facial nerve was the most commonly affected cranial nerve, correlating with rhombencephalitis and abnormal pontine signal on MRI in one patient. Facial palsy was peripheral in type, a pattern frequently seen with GBS, but less commonly encountered with polioviruses (10 to 15% incidence of cranial nerve dysfunction in general) and other nonpolio paralytic viruses.9,19⇓

Clinical and electrodiagnostic abnormalities were confirmed radiologically. T2-weighted and gradient echo signal changes within the cervical cord and the conus were consistent with myelitis, whereas enhancement of the cauda equina suggested polyradiculitis. Prior reports of patients with possible WNV-associated poliomyelitis described normal MRI studies of the spine and brain.13,14⇓ No information on spine imaging is available from the 1999 New York outbreak review. Based on our results, we recommend imaging the spine in patients with WNV encephalitis who also have weakness. The most commonly noted abnormalities on brain MRI in our series included hyperintense T2 and FLAIR signals, sometimes with associated diffusion abnormalities or gadolinium enhancement.

We did not observe an association between weakness and mortality as previously described, probably owing to the smaller number of patients in our series.7

Many clinical, electrodiagnostic, and neuroimaging findings in WNV infections need further elucidation. Clinical and electrodiagnostic follow-up of WNV patients may be useful in recognizing delayed changes similar to the “postpolio” syndrome. Better understanding of the underlying pathophysiologic mechanisms in WNV infection is needed to find targeted therapy and perhaps prevent a new epidemic of an infectious paralysis.