Varicella-Zoster Virus Infections of the Nervous System: Clinical and Pathologic Correlates

The recognition of patterns of symptoms and signs in patients is essential to the practice of clinical medicine. Pattern recognition also forms the basis for the visually-based specialty of anatomic pathology. Clinicians selectively obtain corroborating laboratory information after formulating a differential diagnosis. Similarly, in the morphologically based disciplines of autopsy and surgical pathology, the pathologist must recognize patterns of disease to narrow the differential diagnosis and to select the tests or additional stains most likely to be informative.

Varicella-zoster virus (VZV), an alpha herpesvirus found exclusively in humans, can cause a wide spectrum of disorders throughout the lifetime of the individual.1–3 Varicella-zoster virus causes an acute febrile exanthamous illness (varicella or chickenpox), usually in childhood. After chickenpox resolves, VZV becomes latent in neurons of cranial and spinal ganglia of nearly all individuals with no corresponding morphologic consequences. In elderly and immunocompromised individuals, VZV may reactivate to produce a dermatomal rash and radicular pain (zoster or shingles). Rarely, VZV reactivates and causes pain without rash (zoster sine herpete). After zoster resolves, many elderly patients experience severe persistent pain (postherpetic neuralgia). Varicella-zoster virus can also spread to the large cerebral arteries, usually in immunocompetent and occasionally in immunocompromised individuals, to produce a spectrum of large-vessel vascular damage, ranging from vasculopathy to vasculitis, with stroke. In immunocompromised patients, particularly those with cancer or the acquired immunodeficiency syndrome (AIDS), VZV spreads not only to arteries, but also to the central nervous system (CNS), with varying tissue penetration of the spinal cord (myelitis) and brain (small-vessel vasculopathy). Depending on infection of oligodendrocytes, ovoid mixed ischemic and demyelinative lesions may be seen at any level of the neuraxis. Rarely, VZV infection in immunocompromised patients produces lesions that predominantly affect the meninges (meningoencephalitis) or ependyma (ventriculitis).

Detection of the presence of the virus often requires a combination of morphologic, immunohistochemical, in situ hybridization, and polymerase chain reaction (PCR) methods.4–7 Given that the virus may be difficult to document in tissues in association with VZV-associated complications, it is not surprising that the recognition of the patterns of disease caused by VZV has been challenging.4 Patients often come to biopsy, or especially autopsy, without clinical suspicion of VZV-induced disease. Pathologists must recognize these varied patterns of disease caused by the virus and pursue the correct definitive tests.

Patients in the United States usually become infected with VZV in childhood when they develop acute varicella or chickenpox. Humans are the only reservoir for the virus. In temperate climates, this highly contagious virus has a seasonal prevalence in winter and spring.1 Chickenpox is associated with a widespread vesicular exanthema that occurs without respect to dermatomal distribution (Figure 1) and has been well recognized by parents and pediatricians alike for more than 100 years. An effective vaccine has been developed and, if widespread vaccination programs are implemented and continued in the United States, uncomplicated chickenpox may become a rare infectious disease seldom encountered by a new generation of patients and clinicians. Central nervous system complications are estimated to occur in less than 1% of cases of chickenpox, and even this low number may be an overestimate.2 Many children have only mild neurologic sequelae, with headache, photophobia, and neck stiffness, consistent with mild meningitic symptoms. Cerebellar ataxia is one of the most common neurologic abnormalities, and very rarely, transverse myelitis has been reported.2 

The vast majority of patients with chickenpox, even those with neurologic complications, recover from their illness. Thus, neuropathologic reports on chickenpox are limited and are derived from the few children who manifest unusual severity of disease and die. Many cases that were originally clinically classified as having a “cerebral” type of CNS complications8 or “meningoencephalitis” were found to have only cerebral edema at autopsy. These cases were subsequently recognized as encephalopathy of Reye syndrome after varicella, with fatty livers and hepatotoxicity, and were not due to direct viral infection or encephalitis.9 Other “cerebral” cases manifesting with meningeal and perivenous lymphocytic inflammation and limited perivenular demyelination were classified as cases with indirect, presumably immune-mediated effects of the virus, known as perivenous (postinfectious) encephalomyelitis.10 In cases of aseptic meningitis, the CNS findings generally consist only of scattered meningeal chronic inflammatory cells (Figure 2, A). The virus was virtually never isolated from cerebrospinal fluid (CSF) or brain. Only in rare cases, usually in children with cancer, in children receiving steroid therapy, or in neonates, have viral inclusions been seen in the CNS.11,12 Even in children with cancer, viral inclusions in the CNS have been seen in only a very small number of patients with visceral dissemination (2 of 19 patients).12 In fact, most deaths from chickenpox are due to pneumonitis (Figure 2, B).

The pathogenesis of acute cerebellitis associated with chickenpox is unknown. Although originally thought to be an immune-mediated form of postinfectious encephalomyelitis restricted to the cerebellum, magnetic resonance imaging (MRI) only rarely reveals the focal cerebellar or brain stem lesions characteristic of postinfectious encephalomyelitis.13 However, PCR detection of VZV DNA in the CSF suggests that disease is directly related to the presence of the virus.14 The unusual CNS complications associated with chickenpox include transplacental transmission to the fetus after maternal varicella infection, with viral inclusions demonstrable in the fetal lung and chorionic villi,15 and pseudotumor cerebri in a child with varicella, the pathologic substrate of which is uncertain.16 

After chickenpox resolves, VZV becomes latent in the peripheral nervous system (PNS) ganglia of virtually all infected individuals and persists throughout the lifetime of the host.3 Numerous ganglia are affected, more often the trigeminal cranial ganglia at the skull base but also multiple dorsal root ganglia adjacent to the spinal cord.17 During latency, VZV is not infectious and does not transcribe most of its genetic material, thereby apparently escaping detection and clearance of the virus by the immune system.18 Intense research continues to focus on understanding the physical state of the virus during latency, including the abundance and configuration of viral DNA, the identification of cells that harbor the virus in latently infected ganglia, and the extent of viral transcription and translation.19–21 Quantitative PCR studies indicate a very low copy number of the VZV genome, with 6 to 31 copies per 100 000 latently infected ganglionic cells.19 Unlike retroviruses, which incorporate their nucleic acid into host DNA, VZV remains extrachromosomal but in a noninfectious form. The present understanding of latency derives largely from postmortem studies of trigeminal and dorsal root ganglia rapidly harvested at autopsy. Most studies have shown that the virus is present predominantly, if not exclusively, in neurons of ganglia (Figure 3).20,21 Despite the presence of latent virus within these neurons, little morphologic change and no inflammatory response has been noted during latency. Again, however, relatively few pathologic reports exist detailing any subtle changes that may occur in ganglia during VZV latency when the patient develops intercurrent systemic diseases (such as malignancy) or undergoes transplantation. Four VZV genes (genes 21, 29, 62, and 63) are known to be transcribed in latently infected human ganglia, but it is not yet clear which of these genes are translated into proteins during latency.22 The protein product of VZV gene 63 has been identified immunohistochemically in latently infected ganglionic neurons (Figure 4).23 A better understanding of exactly which genes are transcribed and translated during VZV latency may lead to strategies that prevent viral reactivation. This prevention is important since zoster and its attendant complications are difficult to treat, especially if patients develop postherpetic neuralgia or if the virus spreads to the CNS or cerebral arteries. Although VZV vaccination prevents chickenpox, the effect of vaccination on the incidence of zoster in the elderly will not be known for decades. Varicella-zoster virus vaccine is a live attenuated virus that becomes latent after vaccination.24 

Varicella-zoster virus appears to become latent in nearly all infected individuals, and the likelihood of viral reactivation to shingles (zoster) increases with each advancing decade of age. Zoster is common, with an incidence of more than 300 000 cases annually in the United States, albeit without a seasonal predilection like chickenpox.1–3 Zoster usually develops in elderly individuals and is 8 to 10 times more frequent after age 60 years than before. Immunocompromised patients are also at high risk.

With reactivation, the virus spreads transaxonally to the skin, causing a rash with a dermatomal distribution; the rash is characterized by vesicles on an erythematous base. In immunocompetent individuals, the zoster rash is usually confined to 1 or 2 dermatomes25 and is associated with severe radicular pain. Zoster is nearly always monophasic, with recurrence occurring in less than 5% of zoster patients. Because the rash can occur anywhere on the trunk, face, ear, mouth, extremities, or genitalia, patients are likely to seek medical attention from various medical specialists. Any level of the neuraxis can be involved, but thoracic zoster is most often seen clinically, followed by lesions on the face (Figure 5). The latter usually involves the ophthalmic division of the trigeminal nerve and may be associated with zoster keratitis, a potential cause of blindness. Involvement of the seventh cranial nerve results in weakness of the ipsilateral facial muscles and a vesicular rash in the external ear (zoster oticus) or on the hard palate (Figure 6). Such lesions are easily overlooked or misinterpreted by clinicians or autopsy pathologists.

In some cases, the zoster rash must be distinguished from lesions caused by trauma or sexual abuse.26 The latter distinction is of special interest to forensic pathologists. Although zoster is relatively uncommon in children, occasional cases of a vesicular perianal zoster rash, a rash with a sacral nerve distribution, have been mistaken clinically for genital herpes simplex virus (HSV) lesions. The presence of sexually transmitted diseases in young children, such as genital HSV, should always prompt suspicion of sexual abuse.27,28 The phenomenon of zoster after trauma is also of interest to forensic and autopsy pathologists. Atypical varicella exanthems have been associated with childhood cutaneous injuries such as antecedent wasp stings,29 and classic dermatomal zoster has been reported in adults after spinal surgery30 and even after trivial external musculoskeletal trauma.31 

Zoster in immunocompromised patients develops after radiation therapy; bone marrow transplantation; or in association with lymphoma, leukemia, cancer, AIDS, or even prolonged steroid use. In contrast to the course of zoster in immunocompetent patients, zoster is often recurrent or protracted in the immunocompromised population, especially in patients with AIDS.3 In immunocompromised patients, VZV is also considerably more likely to disseminate after zoster to multiple cutaneous sites and internal organs and to spread to the CNS causing complications that will be described further.

In all cases, the pathologic substrate for zoster is varying degrees of ganglionic hemorrhage and inflammation. In 1900, Head and Campbell32 published their classic and detailed autopsy studies on 21 patients with shingles; their monograph was recently reviewed by Oaklander,25 who noted the near impossibility of duplicating such extensive CNS and PNS dissections today. More recently, Nagashima and coworkers33 and Ghatak and Zimmerman34 also reported hemorrhage, necrosis, and inflammation within affected spinal ganglia of patients with zoster. Collectively, these reports and our own experience suggest considerable variability among patients and among multiple ganglia from a single patient in the degree of changes present in individual ganglia; this variability underscores the need for extensive dissection of the ganglia at autopsy. The histologic features may include mononuclear and lymphocytic inflammation, neuronal degeneration, neuronal phagocytosis by satellite cells, and empty neuronal cell beds, with eventual fibrous scarring of the ganglia 4,25 (Figure 7). Vasculitis in the adjacent nerve results in the loss of myelin and axons,4 which can result in limb paresis.35 Interestingly, in contrast to latent VZV during which the virus is found only in the neurons of ganglia,20,21 zoster is characterized by viral inclusions and viral particles detectable by light and electron microscopy, respectively, in neurons, satellite cells, and fibroblast-like cells of ganglia.33,34 

Rarely, VZV reactivates from latency in dorsal root ganglia but fails to reach the skin or cause zoster rash. This entity, known as zoster sine herpete, is characterized by severe radicular pain that can be documented only by sophisticated virologic testing and may well be underrecognized.36 Virtually nothing is known about the pathologic correlates of zoster sine herpete. Presumably, however, this syndrome represents more restricted inflammation at the level of the ganglion, with unsuccessful transaxonal spread to skin because of a more effective host response.

Most zoster in immunocompetent patients resolves without sequelae. However, many elderly patients have prolonged, often debilitating pain, known as postherpetic neuralgia.3,37 This intractable pain occurring in a dermatomal distribution can persist for months or even years. Postherpetic neuralgia, defined as pain persisting more than 4 to 6 weeks after the rash resolves, is the most frequent neurologic complication of zoster.37 Age is the most important predictor of postherpetic neuralgia, with the risk for postherpetic neuralgia as high as 45% in patients who develop zoster after age 60 years. The incidence is also slightly higher in women and in individuals with zoster that has a trigeminal distribution. The pain of postherpetic neuralgia is often described by patients as excruciating and debilitating, perhaps best appreciated visually by the drawing furnished to us by an artist who suffered from postherpetic neuralgia and who pictured his head skewered with innumerable nails (Figure 8)! Despite numerous clinical trials of various antiviral agents, steroids, and pain relievers, there is no proven cure to date.

The pathology of postherpetic neuralgia is just beginning to be understood, and much less morphologic information is available for this condition than for zoster. Unfortunately, the monograph by Head and Campbell32 did not address the pathology of postherpetic neuralgia.25 Postmortem microscopic analysis of ganglia from 2 patients with postherpetic neuralgia revealed the presence of inflammatory cells, often around dying neurons, that were still present 1 to 2 years after acute zoster.38,39 Smith38 found gross cystic distortion of thoracic sensory ganglia surgically removed 2.5 months after the onset of a zoster rash, persistent chronic inflammatory cells in a ganglion from a patient who had experienced zoster 1 year previously, and ghosts of sensory neurons in the ganglia of an elderly woman 2 years after the patient had experienced postherpetic neuralgia. He hypothesized that the “altered structure of surviving cells” (ie, the neuron ghosts) may have contributed to the pathophysiology of her intractable pain. Zacks et al40 found no morphologic differences in the nerves from patients with or without postherpetic neuralgia but did not have access to the corresponding ganglia for study. Oaklander et al41 and Rowbotham and colleagues42 recently demonstrated a greater loss of small cutaneous nerve endings in skin biopsies obtained from patients with zoster who developed postherpetic neuralgia than in those who did not develop this complication.

Additional pathologic studies of ganglia obtained at autopsy and of nerves and skin obtained at biopsy from individuals with postherpetic neuralgia are needed. However, sophisticated virologic studies at the level of the ganglia in these patients are especially desirable. Proof of a greater viral burden and viral reactivation in the neurons of ganglia from such patients would provide a rationale for aggressive treatment with antiviral reagents. On the other hand, such therapy implies the intravenous use of drugs over several weeks, at great cost and inconvenience to the patient.

Uncommonly, VZV spreads in the CNS in both immunocompetent and immunocompromised patients, although this spread is more likely in the latter group of patients.43–48 The virus can spread with a centripetal pattern (ie, toward the spinal cord or brain), a centrifugal pattern (ie, toward the skin), or in both directions. When the spread is exclusively centripetal, CNS complications may develop without concomitant zoster rash, making clinical recognition and pathologic suspicion even more challenging.49 

In immunocompetent patients with reactivated virus, the virus spreads into the proximal nerve roots adjacent to dorsal root ganglia, causing neuritis or plexitis.35 The affected nerves show lymphocytic inflammation, vasculitis, demyelination, and axonal loss. Neuritis at the cranial level produces various symptoms,3 including the Ramsay Hunt syndrome. When VZV spreads centrally along peripheral nerves toward the spinal cord, the virus may enter the spinal cord at levels contiguous with the affected dorsal root ganglia and nerves, causing myelitis. Patients with this condition demonstrate the clinical features common to any myelitis (Figure 9, A). If the cervical spinal cord is involved, patients exhibit weakness of all 4 extremities and impaired sphincter function. Bladder and bowel dysfunction may also be seen with zoster that has a sacral distribution, even in the absence of myelopathy. Furthermore, zoster that has a cervical or thoracic distribution may produce diaphragmatic paralysis, which should be kept in mind as a potential cause of death in patients with VZV cervical radiculitis or myelitis. The clinical recognition and diagnosis of zoster myelitis is low, but pathologic involvement may be greater. Before death, the diagnosis of myelitis can be corroborated by radiographic studies, usually MRI (Figure 9, B). Pathologically, inflammation of the spinal cord meninges is coupled with varying degrees of underlying spinal necrosis, demyelination, and microglial influx (Figure 9, C). Although clinical recovery is variable, many immunocompetent patients improve significantly. Those with persistent weakness presumably have greater degrees of spinal cord necrosis and neuronal loss. The extent to which zoster myelitis in immunocompetent patients is due to a direct viral cytopathic effect is still unknown. Further clinicopathologic correlation and virologic studies of autopsy cases are needed.

The virus may also spread to the large blood vessels at the base of the brain, causing a process very different from meningitis. The most common form of cranial zoster is zoster with a trigeminal distribution; this form of zoster usually develops in immunocompetent individuals. For unknown reasons, the ophthalmic division of the trigeminal nerve (herpes zoster ophthalmicus) is disproportionately affected as compared with the maxillary and mandibular divisions.50 The virus is thought to spread via afferent trigeminal ganglionic fibers to the blood vessels at the base of the brain. Days to weeks after herpes zoster ophthalmicus, patients develop stroke with the onset of hemiplegia contralateral to the zoster. Angiography and pathology examination at autopsy usually reveal involvement of the internal carotid artery or its branches ipsilateral to the zoster, with resultant vessel thrombosis, varying degrees of vessel wall inflammation, and large, ipsilateral brain infarctions.4–6,50–53 This large-vessel involvement by VZV is known variously as herpes zoster ophthalmicus, granulomatous angiitis, VZV vasculitis, or simply stroke associated with VZV. Depending on the time elapsed since the onset of viral-mediated damage, and possibly other as yet unknown factors, there is great variability in the extent of inflammation in affected blood vessels in cases of VZV meningoencephalitis, myelitis, neuritis, and large-vessel vasculopathy at the time of autopsy.4–6, 50–53 A spectrum of vascular involvement exists, ranging from necrotizing arteritis to modest, chronic vascular inflammation, to thrombosis without inflammation, to remote vascular occlusion resembling atherosclerosis at the level of small blood vessels in the brain, spinal cord, peripheral nerves, and large vessels in the circle of Willis.

Recognition of the ability of VZV to directly infect blood vessel walls has been slow to occur4–6,50–54 despite a seminal report by Linnemann and Alvira54 in 1980 demonstrating the virus in the outer layers of vessel walls in a patient with Hodgkin disease and granulomatous angiitis after ophthalmic zoster. The use of immunohistochemistry, in situ hybridization, and PCR techniques has confirmed the presence of the virus in cranial blood vessels.5,6 Like Linnemann and Alvira,54 our laboratory has usually found evidence of virus in the outer vessel wall and not in the endothelium. Interestingly, the role of direct vascular involvement by VZV as the underlying cause of disease has been underscored by the demonstration of viral inclusions in abdominal blood vessels in an elderly cancer patient with intestinal necrosis.55 In more remote cases, particularly in cases of childhood AIDS associated with VZV, the vessel wall may develop a fibrotic, intimal proliferative lesion without inflammation, and the virus may or may not be demonstrable56 (Figure 10).

In AIDS patients, VZV tends to reactivate from multiple dorsal root ganglia levels, and the disease is often disseminated. Skin lesions extend beyond 1 or 2 dermatomes, and the viscera, spinal cord, and brain are often involved, with or without concomitant rash. AIDS patients and other severely immunocompromised individuals often manifest unduly dramatic and sometimes protracted infections.3,4,45–49 Such patients present clinically with headache, confusion (Figure 11, A), focal weakness, or combinations thereof. AIDS patients with very low CD4 counts often develop a greater extent of tissue penetration by VZV than do immunocompetent patients. In the brain, VZV involvement of small blood vessels or ependymal cells results in small-vessel vasculopathy (Figure 11, B, D, and E) (multifocal leukoencephalopathy) or ependymitis (ventriculitis), respectively.4–6,45–49 Both MRI (Figure 11, B) and PCR of the CSF for VZV DNA (Figure 11, C) are helpful in the diagnosis.

The spread of VZV beyond the small vessels and into oligodendrocytes adds a demyelinative component to the lesions of multifocal leukoencephalopathy; these lesions are often small, located deep in the white matter or at junctions of the gray and white matter (Figure 11, B and D), and are less coalescent than the demyelinative lesions of progressive multifocal leukoencephalopathy.45,46 The greater abundance of virus in the lesions, apparently due to an inability of the severely immunocompromised host to mount an effective immune response, results in readily visible intranuclear “owl's eye” Cowdry type A viral inclusion bodies (Figures 11, E and 12, B). The lesions often lack inflammation that might hint at the infectious etiology of the small-vessel vasculopathy.46 Some AIDS patients develop overwhelming ventriculitis that may be radiographically demonstrable (Figure 12, A), with innumerable Cowdry A viral inclusions in the ependymal cells lining the ventricles47, 57 (Figure 12, B). Such patients usually clinically manifest not only headache and mental status changes, but also gait disturbances and hydrocephalus. In an unusual case, severe meningoencephalitis (Figure 12, C) has been reported to be associated with a profound elevation of CSF protein levels and vasculitis with an extensive fibrinoid necrosis (Figure 12, D).58 Myelitis is well documented in AIDS patients59 and is often insidious, progressive, and sometimes fatal. Autopsy studies in these cases have shown frank spinal cord invasion by the virus, sometimes with more extensive necrosis and inflammation than that seen in immunocompetent patients.4 In situ hybridization studies in immunocompromised patients have demonstrated large numbers of cells in CNS lesions that contain VZV DNA, especially in cases with ventriculitis (Figure 13) or small-vessel vasculopathy (multifocal leukoencephalopathy).

Polymerase chain reaction is now the standard method of diagnosis for many viral CNS infections.60,61 Compared with viral isolation of VZV by culture, PCR detection is more rapid and sensitive, and the methodology can be adapted to a wide variety of fluid and tissue specimens to detect infection.62 Cerebrospinal fluid PCR for VZV DNA has considerably broadened our understanding of the neurologic complications due to VZV. In one study of the prevalence of CNS disease due to VZV, PCR was used to test CSF specimens from 84 consecutive patients infected with the human immunodeficiency virus (HIV) who had new neurologic symptoms.63 Six patients had positive results for VZV DNA in CSF, and all had clinical presentations consistent with ocular or other CNS VZV-associated diseases, indicating a tight correlation between positive results and disease. Importantly, 4 patients did not have a zoster rash at the time, underscoring the critical role of CSF PCR testing in cases that may be clinically confusing.63 In another study, CSF PCR results from 514 consecutive HIV-infected patients with neurologic disease were positive for VZV DNA in 13 (2.5%) of the patients.64 Four of the 13 patients had VZV encephalitis or meningoencephalomyelitis and received appropriate antiviral therapy; in 2 patients, viral DNA could no longer be detected in the CSF after therapy, and the patients' clinical conditions improved. In the other 2 patients, PCR analysis showed the persistence of VZV DNA in the CSF; these patients deteriorated clinically and died.64 Thus, CSF PCR may be useful in monitoring therapeutic response. In that same study, 9 of the 13 patients with neurologic symptoms who had positive results on CSF PCR analysis were thought to have symptoms caused by complications of HIV other than VZV meningoencephalomyelitis and were subsequently diagnosed with subclinical reactivation of VZV.64 Thus, in some cases, positive results on CSF PCR testing may antedate clinically recognizable disease. The results of CSF PCR became negative in patients who received appropriate antiviral therapy, suggesting a prognostic utility for testing patients in whom disease might be averted by prophylactic therapy. Together, these studies indicate that CSF PCR testing for VZV DNA can influence therapeutic decision making.

Cerebrospinal fluid PCR testing for VZV DNA has also been critical in identifying VZV as a causative factor in cases of CNS infections without skin manifestations,49,65 cases of arteritis,6 and cases of Ramsay Hunt syndrome (with testing of both CSF and the fluid from vesicles in the ears).66 This testing has also been critical in excluding VZV as a causative factor in multiple sclerosis67 and in most cases of Bell's palsy (with testing of the endoneurial fluid).68 Polymerase chain reaction testing for VZV DNA on tissues obtained at biopsy or autopsy also has diagnostic utility. Unlike other members of the herpesvirus family such as HSV-1, which have been detected in some brain tissues obtained at autopsy from patients without neurologic diseases,69 VZV has not been found incidentally in such specimens. This absence of detectable VZV in the brain suggests that, after chickenpox or zoster, the virus becomes latent only in the PNS, at the level of the cranial or dorsal root ganglia, and that the virus does not normally gain access to the CNS.

Few studies have directly addressed the incidence of VZV DNA detection in autopsy tissues from healthy individuals. Liedtke et al70 used PCR to determine the presence of the DNA of HSV-1 and VZV in trigeminal ganglia and olfactory bulbs of 109 forensic postmortem cases. HSV-1 DNA was found in 15.5% of olfactory bulbs, but VZV DNA was found in only 1% of olfactory bulbs. Several studies that have evaluated the role of VZV in various neurologic diseases provide additional evidence that VZV DNA is not incidentally, widely detectable in postmortem brain tissues. Analysis by PCR failed to reveal VZV DNA in temporal lobe cortex from 8 schizophrenic patients, 8 nonschizophrenic suicide victims, and 8 healthy controls.71 Similarly, no VZV DNA was demonstrated by PCR in brain tissues from epileptic children with Rasmussen chronic encephalitis72 and in brain tissues from healthy elderly individuals and patients with Alzheimer disease.73 Only a single study reported detection of VZV DNA in a significant proportion of brains from patients with multiple sclerosis and control individuals,74 but those results have never been confirmed. Polymerase chain reaction testing of various tissue specimens has excluded VZV as the cause of giant cell arteritis75 and multifocal encephalomalacia76 but has verified the role of the virus in some congenital infections.77 

Testing of CSF for VZV antibodies helps to confirm the role of VZV in producing the varied clinical syndromes of the PNS and CNS. In the appropriate clinical setting (ie, acute or subacute spinal cord disease, acute or chronic progressive encephalitis, or chronic radicular pain with or without rash), the presence of VZV antibody in the CSF is strong presumptive evidence of VZV infection. The diagnosis of VZV infection of the nervous system is supported by the detection of VZV antibody in the CSF, even in the absence of PCR-amplifiable VZV DNA.78 The analysis of serum anti-VZV antibody is of no value since VZV antibodies persist in the serum of nearly all adults.79 

In other acute viral encephalitides, the viruses are detectable in the CSF early in disease, and an antibody response follows days to weeks later; in contrast, VZV infection of the nervous system is often protracted, especially in immunocompromised patients. Because it is impossible to predict from clinical information whether CSF testing will reveal VZV DNA or antibody, clinicians should request both PCR and antibody analysis. Collectively, these tests have confirmed the diagnosis of VZV myelitis, large- or small-vessel encephalitis, and zoster sine herpete.3 

Varicella-zoster virus can infect a wide variety of cells in the CNS and PNS, including neurons, oligodendrocytes, meningeal cells, ependymal cells, and cells of the blood vessel wall. The wide range of susceptible cells explains the diversity of the clinical and pathologic nervous system manifestations of VZV. Few other infectious agents induce such protean clinical neurologic syndromes. The spectrum of VZV infection can range from asymptomatic latent infection to syndromes characterized by neuritis with rash (zoster); postherpetic neuralgia; and a greater extent of tissue penetration of the virus resulting in vasculitis or vasculopathy with stroke, meningoencephalitis, and/or myelitis. The results of PCR analysis suggest that, in contrast to the pattern noted for several other members of the herpesvirus family including HSV-1, the detection of VZV DNA in tissues is indeed indicative of true viral presence and in most cases correlates with neurologic disease. The PCR analysis of CSF remains a specific and sensitive test for VZV detection and is the mainstay for diagnosing the neurologic complications of VZV infection during the life of the patient.

This work was supported in part by Public Health Service grants AG06127 and NS32623. We thank Marina Hoffman for editorial review and Cathy Allen and Virginia McCullough for preparing the manuscript.