Brain (1991), 114, 1181-1196
HERPES ZOSTER MYELITIS by ORRIN DEVINSKY, 1 EUN-SOOK C H O , 3 CAROL K. PETITO 2 RICHARD W. PRICE 1 *
and
(From the Departments of 'Neurology and 2Pathology (Neuropathology), The New York HospitalCornell University Medical Center and Memorial Sloan-Kettering Cancer Center, New York, and the ^Department of Pathology, University of Medicine and Dentistry of New Jersey, New Jersey Medical College, Newark, New Jersey, USA) SUMMARY
INTRODUCTION
Herpes zoster (HZ; shingles) is a sporadic illness caused by varicella-zoster virus (VZV) and is characterized by a vesicular dermatomal rash. It results from reactivation of latent infection of sensory ganglion neurons established years earlier during the course of primary infection, that is, chickenpox (varicella) (Hope-Simpson, 1965). Although HZ is characteristically benign, neurological complications are nonetheless common and range from transient pain and paraesthesiae associated with the acute rash to postherpetic neuralgia (Hope-Simpson, 1975; Portenoy et al., 1986), segmental sensory loss or motor paresis (Kendall, 1957; Thomas and Howard, 1972), encephalitis (Jemsek etal., 1983), myelitis (Hardy, 1876), and cerebral vascular occlusion (MacKenzie et al., 1981; Eidelberg et al., 1986). Clinical myelopathy is a rare complication of HZ that usually develops in the immunocompromised host. In large series, the incidence varies from 0 to 0.8% in the general population or in immunosuppressed patients, respectively (Schimpff et al., 1972; Mazur and Dolin, 1978; Ragozzino et al. ,1982). Although HZ myelitis was first reported by Hardy in 1876, most subsequent reports have described only single or a Correspondence to: Dr Orrin Devinsky, Department of Neurology, Hospital for Joint Diseases, NYU Medical Center, 301 East 17th St., New York, NY 10003, USA. * Present address: Department of Neurology, University of Minnesota Medical School, Minneapolis, MN, USA. © Oxford University Press 1991
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
We studied the clinical (10 patients) and pathological (9 patients) findings in 13 patients with herpes zoster myelitis, all of whom had systemic illnesses associated with immunosuppression. The median interval between the onset of the herpes zoster rash and myelopathic symptoms was 12 days, and the subsequent median interval to maximal deficit was 10.5 days. Presenting neurological symptoms were characteristically ipsilateral to the rash, with motor dysfunction predominating, followed by a spinothalamic and, less often, posterior column sensory deficit. Pathological involvement was most severe in the dorsal root entry zone and posterior horn of the spinal cord segment corresponding to the involved dermatome. There was variable spread both horizontally and vertically in the spinal cord. Direct varicella-zoster virus (VZV) infection of neuroectodermal cells, particularly oligodendrocytes, was demonstrated by immunostaining viral antigens (8 cases), and by the presence of Cowdry type A intranuclear inclusions (7 cases) and often was associated with focal demyelination (6 cases). In 4 patients a VZV vasculitis was associated with leptomeningitis and haemorrhagic necrosis (spinal cord in 1; brainstem or cerebellum in 3). The protracted evolution in many cases and the pathologically documented direct viral infection of the spinal cord provide a rational basis for the use of antiviral therapy in preventing or attenuating the evolving myelopathy.
1182
O. DEVINSKY AND OTHERS
few cases (Bruce, 1907; Lhermitte and Nicolas, 1924; Thalhimer, 1924; Bellavitis, 1931; Biggart and Fisher, 1938; Wilson, 1940; Denny-Brown etal, 1944; Kendall, 1957; Boudine/a/., 1958; McAlpine et al., 1959; Harrison, 1964; Rose et al, 1964; Force et al, 1968; McCormick et al, 1969; Nyberg-Hansen, 1972; Hogan and Krigman, 1973; Richmond, 1974; Sachdeva et al, 1975; Gardner-Thorpe et al, 1976; Rao and Ramanamurty, 1976; Cole, 1978; Ruppenthal, 1980; Corston etal, 1981; Cullis et al, 1982; Muder et al, 1983; Mas and Lapresle, 1985; Yeolekar et al, 1985). We reviewed our own experience with 13 patients and additionally analysed the 33 previously reported cases in order to define more clearly the clinical features, pathological characteristics and pathogenesis of this disorder. MATERIAL AND METHODS
RESULTS
Our own series included 8 men and 5 women with a mean age of 49 yrs (median 44, range 19 — 86 yrs). All had predisposing disease which included Hodgkin's disease (3 patients), non-Hodgkin's lymphoma (2 patients), acute lymphocytic leukaemia (1 patient), the acquired immune deficiency syndrome (AIDS) (3 patients), metastatic carcinoma (3 patients) and renal failure (1 patient). The initial zoster rash was rightsided in 5 patients, left-sided in 6, bilateral in 1 and either absent or unnoticed in 1 autopsied case; dissemination of the rash occurred in 5 patients. The dermatonal distribution of the rash was cervical in 1 patient, thoracic in 9 patients, and lumbosacral in 3 patients. The literature series included 13 men, 15 women, and 5 not classified with regard to sex. The mean age was 55 yrs (median 59, range, 16 — 81 yrs). An underlying illness
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
On review of the clinical and pathological records of patients at the Memorial Sloan-Kettering Cancer Center and the New York Hospital from 1970 to 1986 we identified 13 with a diagnosis of HZ myelitis. The diagnosis was based on clinical grounds: (1) characteristic rash; (2) spinal cord dysfunction; (3) exclusion of alternative causes of myelopathy (e.g., tumour, multiple sclerosis) by clinical and laboratory findings; and (4) absent or only mild supraspinal central nervous system (CNS) signs and symptoms at presentation. The diagnosis of HZ myelitis was based solely on clinical diagnosis in 4 patients, on pathological diagnosis in 2 patients, and on both clinical and pathological diagnoses in the remaining 7 patients. In 9 patients with complete postmortem examination, brains and spinal cords were fixed in 10% formalin for 2 wks, cut into 1 cm or 0.5 cm serial sections; representative sections were embedded in paraffin. The microscopic sections were stained with haematoxylin and eosin, luxol fast blue-haematoxylin and eosin, and Bodian's silver stain. A high-titre herpes zoster convalescent human serum was used as the primary antibody (Horten et al., 1981; Eidelberg et al., 1986) to detect the presence of VZV antigen by the avidin biotin complex (ABC) immunoperoxidase technique using commercially obtained secondary antibody and ABC complex (Vector Laboratories Inc., Burlingame, CA). Paraffin-embedded brain from a known case of VZV encephalitis was used as a positive control. On selected blocks, immunostaining was performed for herpes simplex virus types 1 and 2, and cytomegalovirus with commercially available antibodies (Ortho Laboratories, Bridgewater, NJ). Inclusion of previously reported cases was based on less restrictive criteria: the presence of a characteristic rash and either the development of myelopathy in temporal relation to the rash or pathological findings that excluded other causes and were consistent with herpes zoster myelitis. Twenty-seven cases with clinical evidence of HZ myelitis were identified among the 33 literature cases. The clinical findings in the previously reported cases are described along with our own findings in the results section, while the less complete pathological findings described in the literature are presented in the Discussion.
HERPES ZOSTER MYELITIS
1183
was reported in 7 of 33 patients: neurosyphilis and rheumatoid arthritis each in 2 patients, and lymphoma, ovarian carcinoma, and congestive heart failure each in 1 patient. The initial zoster rash was right-sided in 19 patients, left-sided in 10, and bilateral in 1; dissemination of the rash was not reported. The dermatomal distribution was trigeminal in 1 patient, cervical in 6, thoracic in 21, and lumbosacral in 1. Clinical presentation Symptoms and signs of spinal cord dysfunction usually began within 3 wks of the onset of the rash (fig. 1A). Among our 9 patients with adequate documentation, myelopathic manifestations developed 5 - 2 1 days after the initial rash (median, 12 days). Among documented literature cases (n = 22), onset of spinal cord dysfunction developed from 8 days before to 10 wks after initial rash (median, 14 days).
JLl
1re 0 £L Q.
*o 6 6 4 2
0
n
0_.
2 4 6 8 10 12 14 16 18 20 22 24 26 28 Days
>28
FIG. 1. Timing of myelopathy in relation to the onset of herpes zoster rash. A, interval between the rash and onset of myelopathic symptoms, B, interval between the onset of myelopathic symptoms and maximal neurological deficit. Hatched bars = authors' series; open bars = literature series.
Subacute progression occurred in most patients (fig. 1B). In our own series, maximal deficit usually occurred within 3 wks of initial myelopathy (median, 10.5 days); exceptions included 2 AIDS patients with long intervals (2 and 6 mos) separating the initial and maximal neurological deficit. Among literature cases, maximal neurological deficit occurred within 10 wks (median, 14 days). Leg weakness and sensory abnormalities were the most common presenting symptoms in the 38 patients from the combined series in whom clinical information was available (Table 1). Lower limits weakness occurred in 74% of these patients and was usually unilateral. Sensory abnormalities were present in 53% of all patients. A Brown-Sequard syndrome was noted in 4 patients and bladder dysfunction in 7. The most advanced neurological deficit included lower limb weakness in 35 patients in whom information was available, with paraparesis in 80%. Sensory loss occurred in 26 of the 35 (74%) patients and included a sensory level to all modalities in 12 (34%) patients. Pain and temperature sensation was impaired more often than vibration and
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
4-
2.5 mmol/1 in 15 of 22 patients. In general, the greatest CSF pleocytosis and protein elevation occurred near the time of maximal deficit. CSF glucose was
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
Total
1185
HERPES ZOSTER MYELITIS
normal ( > 2.5 mmol/1) in 15 of 16 patients. The patient with an abnormally low value (1.9 mmol/1) had 1760 WBC/mm3 (40% polymorphonuclear leucocytes) in the CSF and extensive necrosis pathologically (fig. 2, Case 2). Bacterial, viral and fungal culture and cytology were negative in all cases in which they were obtained. Specific VZV cultures were not obtained. • Case 1
Case 3 ,
Case 2 •
Lumbar^F. termmale Case 5
Case 7
Lumban Cowdry type- A inclusion bodies, positive VZV immunoreactivity Partial tissue disintegration
¥x"j Necrotizing vasculitis
•
Haemorrhagic necrosis, petechiae
Shrunken tissue with old damage
Meshwork of denuded axons (fig. 5) FIG. 2. Summary of pathological findings of herpes zoster myelitis of 9 cases of the present study.
Twelve patients were treated for herpes zoster myelitis and received glucocorticoids and/or antiviral agents (vidarabine or acyclovir). Eight received glucocorticoids alone: 3 improved, 3 deteriorated, and 2 showed no change. Two of the 3 patients who improved during steroid treatment were unusual cases from the literature who exhibited a relapsing course and in whom neuropathological examination was not performed (McAlpine et al., 1959; Nyberg-Hanson, 1972). Vidarabine monotherapy was associated with clinical improvement in both patients treated, including 1 who had deteriorated during a previous trial of glucocorticoids. In another patient vidarabine was used with glucocorticoids and the progressive disease stabilized. Acyclovir monotherapy was used in 1 non-AIDS patient who improved, and in 2 AIDS patients. One of the latter improved during treatment but deteriorated several weeks later. The other was begun on treatment 2 wks after the onset of neurological
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
Thoracic and lumbar • •
1186
O. DEVINSKY AND OTHERS
Neuropathology Pathological findings in our 9 patients with postmorten examination are summarized in Table 2 and fig. 2. The most severely affected spinal cord levels were within two segments of the initial rash in all 6 patients for whom information was available. Pathological changes were limited to one region of the spinal cord in 6 patients and were multiple in 3 (Cases 2, 4, 7). Six of the 9 patients had bilateral lesions and in 2 of these (Cases 2, 9) single level lesions involved nearly the entire transverse section. Dorsal root ganglia supplying the initially involved dermatomes were available in only 2 cases. They showed extensive haemorrhagic necrosis associated with necrotizing vasculitis and thrombosis. Cowdry type A intranuclear inclusions were seen in a majority of the ganglion and satellite cells, and were also observed in many endothelial and adventitial cells. Eight of the 9 patients had abnormalities in the posterior nerve roots. These included mild lymphocytic infiltration (2 patients), focal areas of pallor (2 patients), axonal spheroids (1 patient), and haemorrhagic necrosis with acute inflammation (2 patients). In all 4 of the cases tested, VZV immunoreactivity was present and in 3, Cowdry type A intranuclear inclusion bodies were found in epineural fibroblasts and cells within the endoneurium, the morphology of which was consistent with their being Schwann cells. One patient (Case 9), who survived 6 mos after clinical myelopathy, had atrophic and fibrotic posterior nerve roots. The posterior horns were abnormal in all 9 cases and in 6 of the 9 cases there was bilateral involvement. Pathological changes included focal or diffuse necrosis (5 patients), mild inflammation (2 patients), and only mild changes of capillary proliferation and
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
symptoms because of delay in seeking medical attention, and showed no apparent response. Information on neurological outcome, based on follow-up examination 1—24 mos after maximal neurological illness, was available in 25 patients: recovery was complete in 3 and partial in 16, while 6 had little or no improvements. Those who improved appreciably were generally patients with a milder deficit; of the 3 patients with 'complete' recovery, none had paraplegia or a dense sensory level, and sphincter disorder had been absent (2 patients) or mild (1 patient with urinary frequency). Improvement in the 16 patients with partial recovery included motor, sensory and sphincter function without apparent preference. At least 3 patients with paraparesis became ambulatory. Persistent pain was described in 2 patients. Neurological improvement was not clearly related to the location of the rash, the time period from the rash to initial CNS disease or from initial to maximal neurological deficit, or involvement of specific spinal cord pathways (e.g., motor or sensory). Patients in whom the neurological deficit was static for at least 2 days were unlikely to progress further. Recovery usually commenced within 2 wks of maximal deficit, and the greatest improvement occurred during the subsequent 2 mos, although some continued to improve after 1 yr. Sixteen of 45 patients in the combined series (11 of our 13 patients) died within 4 mos of initial rash. The HZ myelitis and its secondary complications (e.g., pulmonary embolism and decubitus ulcers causing sepsis) contributed to the death of 14 patients. Follow-up information was available for 14 patients with immunosuppressive disorders; 11 died within 4 mos of rash and only 3 showed partial recovery.
1187
HERPES ZOSTER MYELITIS TABLE 2. SUMMARY OF CLINICOPATHOLOGICAL FINDINGS Duration from neurological symptoms to death I day
Maximal neurological deficit Confusion, R LE; weakness
VZVCNS lesion
Age (yrs) sex
Clinical level
1. Hodgkin's disease
19 M
RT12 Disseminated
2. Breast carcinoma
61 F
LT10 Disseminated
3 days (11 days)
Flaccid paraplegia; extensor plamars; T7 sensor level for PS and TS; T9 sensory level for VS, TPS
Focal necrosis (medulla), ependymitis (4th)
3. Hodgkin's disease
25 M
LT8-9 Disseminated
1 wk
Lethargic, L hemiparesis (4/5); extensor plantars; L UE dysmetria; decreased sensation to all modalities (no clear level)
Focal necrosis (medulla, pons)
4. Hodgkin's disease
74 F
L L3 Disseminated
4 wks
Stupor, meningism, bilateral extensor plantars
Focal necrosis, vasculitis (medulla 4th vent.)
5. Lymphoma
53 M
LT8
4 mos
L Homer's syndrome; L LE spastic weakness; (4/5); L extensor plantar; T8 sensory level to PS and TS; minimal distal VS loss, NL JPS
6. Renal failure
81 F
RT4-5
7. Lymphoma
44 M
C5-6
8. Oat cell carcinoma of lung
67 F
9. AIDS
36 M
Case and diagnosis
Focal necrosis (medulla, pons MB, CBL)
Meningeal vasculitis (brainstem) 6 mos
Flaccid paraplegia; no report of sensory exam
R = right; L = left; UE = upper extremity; LE = lower extremity; CSF = cerebrospinal fluid; NL = normal; C = cervical; T = thoracic; L = lumbar; VS = vibration sense; JPS = joint position sense; PS = pain sensation; TS = temperature sensation; MB = midbrain; CBL = cerebellum.
parenchymal rarefaction (2 patients) (fig. 3). Inclusion bodies were present in 5 patients. The posterior horn in 1 patient (Case 5) was atrophic and gliotic. Wedge-shaped regions of demyelination were present in 6 of the 9 patients. These were located near the root entry zone corresponding to the clinically involved dermatomes (fig. 4A, B). Axons were relatively well preserved although occasionally swollen and beaded (fig. 5). Scattered lipid-laden macrophages and scanty fibrillary astrocytes were seen within the demyelinated regions. Oligodendrocytes were markedly decreased.
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
Perivascular lymphocytes (medulla)
1188
O. DEVINSKY AND OTHERS
Intranuclear Cowdry type A inclusion bodies and cytoplasmic VZV immunoreactivity were seen in occasional macrophages and astrocytes within and adjacent to the demyelinated lesions (fig. 6). Wallerian degeneration was not observed. Two patients had extensive transverse involvement of thoracic spinal cord at approximately one level. In 1 (Case 9), a large region of recent demyelination with sheets of lipid-laden macrophages and preserved axons involved more than half of a transverse section of thoracic spinal cord (fig. 7A, B). A focal area of chronic demyelination was present in the posterior columns within this lesion. The second patient (Case 2) had necrotizing transverse myelitis with vasculitis at this thoracic level and a small area of demyelination at a lumbar level. In both patients, these areas contained abundant inclusion bodies and VZV reactivity. Three of 9 patients did not have demyelination in the spinal cord. One (Case 3) had numerous inclusion bodies and VZV positivity in the posterior columns with little evidence of tissue disruption. The second (Case 5) showed cavitation near an anterior root entry zone with a narrow rim of partial necrosis in which inclusion bodies and VZV reactivity were detected. The third (Case 4) had no white matter lesions. Acute necrotizing vasculitis, found in 4 patients, was present around the involved segments of the spinal cord and brainstem in 3 patients and found only in the medulla in the fourth. The vasculitis was associated with haemorrhagic necrosis in 2 of the 3 patients with brainstem vasculitis and in 1 patient with spinal cord vasculitis. However, in 1 patient (Case 2) with complete transverse myelitis there was acute haemorrhagic necrosis, involving the entire cross-section of the spinal cord, accompanied by numerous
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
FIG. 3. Posterior horn with multiple intranuclear inclusion bodies of Cowdry type A in oligodendrocytes (arrow head) and a probable neuron (arrow). Note disintegrating neuropil with oedema. (Case 8). Hematoxylin-eosin. Bar = 100 /im.
HERPES ZOSTER
MYELITIS
inclusion bodies and cytoplasmic VZV immunoreactivity in adventitial and endothelial cells of the involved vessels and in adjacent leptomeningeal arachnoid cells. Four patients had an associated focal VZV encephalitis characterized by necrosis and demyelination associated with inclusion bodies and VZV immunoreactive cells. The involved regions included the nuclei gracilis and cuneatus in 3 patients (fig. 8), fourth ventricular ependyma in 2 patients, upper brainstem in 2 patients, and cerebellar cortex in 1 patient. Although all 4 patients with VZV encephalitis had leptomeningeal or parenchymal vasculitis, only a few foci were haemorrhagic and appeared to be related to vasculitis. Additional neuropathological changes included vascular myelopathy (Petito et al.,
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
FIG. 4. Large wedge-shaped lesion involving both posterior and lateral columns including the posterior root entry zone, A, total absence of myelin on Luxol fast blue staining. B, preserved meshwork of demyelinated axons on Bodian's silver stain. (Case 8, thoracic spinal cord).
1190
O. DEVINSKY AND OTHERS
FIG. 6. Higher magnification of demyelinated meshwork, illustrated in fig. 4, with immunocytochemica! staining of VZV antigen in a few astrocytes (arrow). Note total absence of oligodendrocytic nuclei in the field. (Case 8). Bar = 100 /im.
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
FIG. 5. Higher magnification of demyelinated meshwork, illustrated in fig. 4, with relatively well preserved axons. Note beaded appearance in some axons (arrows). (Case 8). Bodian's silver impregnation. Bar = 100 /im.
HERPES ZOSTER MYELITIS
1191
**•%
w
A
'
HERPES ZOSTER MYELITIS by ORRIN DEVINSKY, 1 EUN-SOOK C H O , 3 CAROL K. PETITO 2 RICHARD W. PRICE 1 *
and
(From the Departments of 'Neurology and 2Pathology (Neuropathology), The New York HospitalCornell University Medical Center and Memorial Sloan-Kettering Cancer Center, New York, and the ^Department of Pathology, University of Medicine and Dentistry of New Jersey, New Jersey Medical College, Newark, New Jersey, USA) SUMMARY
INTRODUCTION
Herpes zoster (HZ; shingles) is a sporadic illness caused by varicella-zoster virus (VZV) and is characterized by a vesicular dermatomal rash. It results from reactivation of latent infection of sensory ganglion neurons established years earlier during the course of primary infection, that is, chickenpox (varicella) (Hope-Simpson, 1965). Although HZ is characteristically benign, neurological complications are nonetheless common and range from transient pain and paraesthesiae associated with the acute rash to postherpetic neuralgia (Hope-Simpson, 1975; Portenoy et al., 1986), segmental sensory loss or motor paresis (Kendall, 1957; Thomas and Howard, 1972), encephalitis (Jemsek etal., 1983), myelitis (Hardy, 1876), and cerebral vascular occlusion (MacKenzie et al., 1981; Eidelberg et al., 1986). Clinical myelopathy is a rare complication of HZ that usually develops in the immunocompromised host. In large series, the incidence varies from 0 to 0.8% in the general population or in immunosuppressed patients, respectively (Schimpff et al., 1972; Mazur and Dolin, 1978; Ragozzino et al. ,1982). Although HZ myelitis was first reported by Hardy in 1876, most subsequent reports have described only single or a Correspondence to: Dr Orrin Devinsky, Department of Neurology, Hospital for Joint Diseases, NYU Medical Center, 301 East 17th St., New York, NY 10003, USA. * Present address: Department of Neurology, University of Minnesota Medical School, Minneapolis, MN, USA. © Oxford University Press 1991
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
We studied the clinical (10 patients) and pathological (9 patients) findings in 13 patients with herpes zoster myelitis, all of whom had systemic illnesses associated with immunosuppression. The median interval between the onset of the herpes zoster rash and myelopathic symptoms was 12 days, and the subsequent median interval to maximal deficit was 10.5 days. Presenting neurological symptoms were characteristically ipsilateral to the rash, with motor dysfunction predominating, followed by a spinothalamic and, less often, posterior column sensory deficit. Pathological involvement was most severe in the dorsal root entry zone and posterior horn of the spinal cord segment corresponding to the involved dermatome. There was variable spread both horizontally and vertically in the spinal cord. Direct varicella-zoster virus (VZV) infection of neuroectodermal cells, particularly oligodendrocytes, was demonstrated by immunostaining viral antigens (8 cases), and by the presence of Cowdry type A intranuclear inclusions (7 cases) and often was associated with focal demyelination (6 cases). In 4 patients a VZV vasculitis was associated with leptomeningitis and haemorrhagic necrosis (spinal cord in 1; brainstem or cerebellum in 3). The protracted evolution in many cases and the pathologically documented direct viral infection of the spinal cord provide a rational basis for the use of antiviral therapy in preventing or attenuating the evolving myelopathy.
1182
O. DEVINSKY AND OTHERS
few cases (Bruce, 1907; Lhermitte and Nicolas, 1924; Thalhimer, 1924; Bellavitis, 1931; Biggart and Fisher, 1938; Wilson, 1940; Denny-Brown etal, 1944; Kendall, 1957; Boudine/a/., 1958; McAlpine et al., 1959; Harrison, 1964; Rose et al, 1964; Force et al, 1968; McCormick et al, 1969; Nyberg-Hansen, 1972; Hogan and Krigman, 1973; Richmond, 1974; Sachdeva et al, 1975; Gardner-Thorpe et al, 1976; Rao and Ramanamurty, 1976; Cole, 1978; Ruppenthal, 1980; Corston etal, 1981; Cullis et al, 1982; Muder et al, 1983; Mas and Lapresle, 1985; Yeolekar et al, 1985). We reviewed our own experience with 13 patients and additionally analysed the 33 previously reported cases in order to define more clearly the clinical features, pathological characteristics and pathogenesis of this disorder. MATERIAL AND METHODS
RESULTS
Our own series included 8 men and 5 women with a mean age of 49 yrs (median 44, range 19 — 86 yrs). All had predisposing disease which included Hodgkin's disease (3 patients), non-Hodgkin's lymphoma (2 patients), acute lymphocytic leukaemia (1 patient), the acquired immune deficiency syndrome (AIDS) (3 patients), metastatic carcinoma (3 patients) and renal failure (1 patient). The initial zoster rash was rightsided in 5 patients, left-sided in 6, bilateral in 1 and either absent or unnoticed in 1 autopsied case; dissemination of the rash occurred in 5 patients. The dermatonal distribution of the rash was cervical in 1 patient, thoracic in 9 patients, and lumbosacral in 3 patients. The literature series included 13 men, 15 women, and 5 not classified with regard to sex. The mean age was 55 yrs (median 59, range, 16 — 81 yrs). An underlying illness
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
On review of the clinical and pathological records of patients at the Memorial Sloan-Kettering Cancer Center and the New York Hospital from 1970 to 1986 we identified 13 with a diagnosis of HZ myelitis. The diagnosis was based on clinical grounds: (1) characteristic rash; (2) spinal cord dysfunction; (3) exclusion of alternative causes of myelopathy (e.g., tumour, multiple sclerosis) by clinical and laboratory findings; and (4) absent or only mild supraspinal central nervous system (CNS) signs and symptoms at presentation. The diagnosis of HZ myelitis was based solely on clinical diagnosis in 4 patients, on pathological diagnosis in 2 patients, and on both clinical and pathological diagnoses in the remaining 7 patients. In 9 patients with complete postmortem examination, brains and spinal cords were fixed in 10% formalin for 2 wks, cut into 1 cm or 0.5 cm serial sections; representative sections were embedded in paraffin. The microscopic sections were stained with haematoxylin and eosin, luxol fast blue-haematoxylin and eosin, and Bodian's silver stain. A high-titre herpes zoster convalescent human serum was used as the primary antibody (Horten et al., 1981; Eidelberg et al., 1986) to detect the presence of VZV antigen by the avidin biotin complex (ABC) immunoperoxidase technique using commercially obtained secondary antibody and ABC complex (Vector Laboratories Inc., Burlingame, CA). Paraffin-embedded brain from a known case of VZV encephalitis was used as a positive control. On selected blocks, immunostaining was performed for herpes simplex virus types 1 and 2, and cytomegalovirus with commercially available antibodies (Ortho Laboratories, Bridgewater, NJ). Inclusion of previously reported cases was based on less restrictive criteria: the presence of a characteristic rash and either the development of myelopathy in temporal relation to the rash or pathological findings that excluded other causes and were consistent with herpes zoster myelitis. Twenty-seven cases with clinical evidence of HZ myelitis were identified among the 33 literature cases. The clinical findings in the previously reported cases are described along with our own findings in the results section, while the less complete pathological findings described in the literature are presented in the Discussion.
HERPES ZOSTER MYELITIS
1183
was reported in 7 of 33 patients: neurosyphilis and rheumatoid arthritis each in 2 patients, and lymphoma, ovarian carcinoma, and congestive heart failure each in 1 patient. The initial zoster rash was right-sided in 19 patients, left-sided in 10, and bilateral in 1; dissemination of the rash was not reported. The dermatomal distribution was trigeminal in 1 patient, cervical in 6, thoracic in 21, and lumbosacral in 1. Clinical presentation Symptoms and signs of spinal cord dysfunction usually began within 3 wks of the onset of the rash (fig. 1A). Among our 9 patients with adequate documentation, myelopathic manifestations developed 5 - 2 1 days after the initial rash (median, 12 days). Among documented literature cases (n = 22), onset of spinal cord dysfunction developed from 8 days before to 10 wks after initial rash (median, 14 days).
JLl
1re 0 £L Q.
*o 6 6 4 2
0
n
0_.
2 4 6 8 10 12 14 16 18 20 22 24 26 28 Days
>28
FIG. 1. Timing of myelopathy in relation to the onset of herpes zoster rash. A, interval between the rash and onset of myelopathic symptoms, B, interval between the onset of myelopathic symptoms and maximal neurological deficit. Hatched bars = authors' series; open bars = literature series.
Subacute progression occurred in most patients (fig. 1B). In our own series, maximal deficit usually occurred within 3 wks of initial myelopathy (median, 10.5 days); exceptions included 2 AIDS patients with long intervals (2 and 6 mos) separating the initial and maximal neurological deficit. Among literature cases, maximal neurological deficit occurred within 10 wks (median, 14 days). Leg weakness and sensory abnormalities were the most common presenting symptoms in the 38 patients from the combined series in whom clinical information was available (Table 1). Lower limits weakness occurred in 74% of these patients and was usually unilateral. Sensory abnormalities were present in 53% of all patients. A Brown-Sequard syndrome was noted in 4 patients and bladder dysfunction in 7. The most advanced neurological deficit included lower limb weakness in 35 patients in whom information was available, with paraparesis in 80%. Sensory loss occurred in 26 of the 35 (74%) patients and included a sensory level to all modalities in 12 (34%) patients. Pain and temperature sensation was impaired more often than vibration and
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
4-
2.5 mmol/1 in 15 of 22 patients. In general, the greatest CSF pleocytosis and protein elevation occurred near the time of maximal deficit. CSF glucose was
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
Total
1185
HERPES ZOSTER MYELITIS
normal ( > 2.5 mmol/1) in 15 of 16 patients. The patient with an abnormally low value (1.9 mmol/1) had 1760 WBC/mm3 (40% polymorphonuclear leucocytes) in the CSF and extensive necrosis pathologically (fig. 2, Case 2). Bacterial, viral and fungal culture and cytology were negative in all cases in which they were obtained. Specific VZV cultures were not obtained. • Case 1
Case 3 ,
Case 2 •
Lumbar^F. termmale Case 5
Case 7
Lumban Cowdry type- A inclusion bodies, positive VZV immunoreactivity Partial tissue disintegration
¥x"j Necrotizing vasculitis
•
Haemorrhagic necrosis, petechiae
Shrunken tissue with old damage
Meshwork of denuded axons (fig. 5) FIG. 2. Summary of pathological findings of herpes zoster myelitis of 9 cases of the present study.
Twelve patients were treated for herpes zoster myelitis and received glucocorticoids and/or antiviral agents (vidarabine or acyclovir). Eight received glucocorticoids alone: 3 improved, 3 deteriorated, and 2 showed no change. Two of the 3 patients who improved during steroid treatment were unusual cases from the literature who exhibited a relapsing course and in whom neuropathological examination was not performed (McAlpine et al., 1959; Nyberg-Hanson, 1972). Vidarabine monotherapy was associated with clinical improvement in both patients treated, including 1 who had deteriorated during a previous trial of glucocorticoids. In another patient vidarabine was used with glucocorticoids and the progressive disease stabilized. Acyclovir monotherapy was used in 1 non-AIDS patient who improved, and in 2 AIDS patients. One of the latter improved during treatment but deteriorated several weeks later. The other was begun on treatment 2 wks after the onset of neurological
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
Thoracic and lumbar • •
1186
O. DEVINSKY AND OTHERS
Neuropathology Pathological findings in our 9 patients with postmorten examination are summarized in Table 2 and fig. 2. The most severely affected spinal cord levels were within two segments of the initial rash in all 6 patients for whom information was available. Pathological changes were limited to one region of the spinal cord in 6 patients and were multiple in 3 (Cases 2, 4, 7). Six of the 9 patients had bilateral lesions and in 2 of these (Cases 2, 9) single level lesions involved nearly the entire transverse section. Dorsal root ganglia supplying the initially involved dermatomes were available in only 2 cases. They showed extensive haemorrhagic necrosis associated with necrotizing vasculitis and thrombosis. Cowdry type A intranuclear inclusions were seen in a majority of the ganglion and satellite cells, and were also observed in many endothelial and adventitial cells. Eight of the 9 patients had abnormalities in the posterior nerve roots. These included mild lymphocytic infiltration (2 patients), focal areas of pallor (2 patients), axonal spheroids (1 patient), and haemorrhagic necrosis with acute inflammation (2 patients). In all 4 of the cases tested, VZV immunoreactivity was present and in 3, Cowdry type A intranuclear inclusion bodies were found in epineural fibroblasts and cells within the endoneurium, the morphology of which was consistent with their being Schwann cells. One patient (Case 9), who survived 6 mos after clinical myelopathy, had atrophic and fibrotic posterior nerve roots. The posterior horns were abnormal in all 9 cases and in 6 of the 9 cases there was bilateral involvement. Pathological changes included focal or diffuse necrosis (5 patients), mild inflammation (2 patients), and only mild changes of capillary proliferation and
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
symptoms because of delay in seeking medical attention, and showed no apparent response. Information on neurological outcome, based on follow-up examination 1—24 mos after maximal neurological illness, was available in 25 patients: recovery was complete in 3 and partial in 16, while 6 had little or no improvements. Those who improved appreciably were generally patients with a milder deficit; of the 3 patients with 'complete' recovery, none had paraplegia or a dense sensory level, and sphincter disorder had been absent (2 patients) or mild (1 patient with urinary frequency). Improvement in the 16 patients with partial recovery included motor, sensory and sphincter function without apparent preference. At least 3 patients with paraparesis became ambulatory. Persistent pain was described in 2 patients. Neurological improvement was not clearly related to the location of the rash, the time period from the rash to initial CNS disease or from initial to maximal neurological deficit, or involvement of specific spinal cord pathways (e.g., motor or sensory). Patients in whom the neurological deficit was static for at least 2 days were unlikely to progress further. Recovery usually commenced within 2 wks of maximal deficit, and the greatest improvement occurred during the subsequent 2 mos, although some continued to improve after 1 yr. Sixteen of 45 patients in the combined series (11 of our 13 patients) died within 4 mos of initial rash. The HZ myelitis and its secondary complications (e.g., pulmonary embolism and decubitus ulcers causing sepsis) contributed to the death of 14 patients. Follow-up information was available for 14 patients with immunosuppressive disorders; 11 died within 4 mos of rash and only 3 showed partial recovery.
1187
HERPES ZOSTER MYELITIS TABLE 2. SUMMARY OF CLINICOPATHOLOGICAL FINDINGS Duration from neurological symptoms to death I day
Maximal neurological deficit Confusion, R LE; weakness
VZVCNS lesion
Age (yrs) sex
Clinical level
1. Hodgkin's disease
19 M
RT12 Disseminated
2. Breast carcinoma
61 F
LT10 Disseminated
3 days (11 days)
Flaccid paraplegia; extensor plamars; T7 sensor level for PS and TS; T9 sensory level for VS, TPS
Focal necrosis (medulla), ependymitis (4th)
3. Hodgkin's disease
25 M
LT8-9 Disseminated
1 wk
Lethargic, L hemiparesis (4/5); extensor plantars; L UE dysmetria; decreased sensation to all modalities (no clear level)
Focal necrosis (medulla, pons)
4. Hodgkin's disease
74 F
L L3 Disseminated
4 wks
Stupor, meningism, bilateral extensor plantars
Focal necrosis, vasculitis (medulla 4th vent.)
5. Lymphoma
53 M
LT8
4 mos
L Homer's syndrome; L LE spastic weakness; (4/5); L extensor plantar; T8 sensory level to PS and TS; minimal distal VS loss, NL JPS
6. Renal failure
81 F
RT4-5
7. Lymphoma
44 M
C5-6
8. Oat cell carcinoma of lung
67 F
9. AIDS
36 M
Case and diagnosis
Focal necrosis (medulla, pons MB, CBL)
Meningeal vasculitis (brainstem) 6 mos
Flaccid paraplegia; no report of sensory exam
R = right; L = left; UE = upper extremity; LE = lower extremity; CSF = cerebrospinal fluid; NL = normal; C = cervical; T = thoracic; L = lumbar; VS = vibration sense; JPS = joint position sense; PS = pain sensation; TS = temperature sensation; MB = midbrain; CBL = cerebellum.
parenchymal rarefaction (2 patients) (fig. 3). Inclusion bodies were present in 5 patients. The posterior horn in 1 patient (Case 5) was atrophic and gliotic. Wedge-shaped regions of demyelination were present in 6 of the 9 patients. These were located near the root entry zone corresponding to the clinically involved dermatomes (fig. 4A, B). Axons were relatively well preserved although occasionally swollen and beaded (fig. 5). Scattered lipid-laden macrophages and scanty fibrillary astrocytes were seen within the demyelinated regions. Oligodendrocytes were markedly decreased.
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
Perivascular lymphocytes (medulla)
1188
O. DEVINSKY AND OTHERS
Intranuclear Cowdry type A inclusion bodies and cytoplasmic VZV immunoreactivity were seen in occasional macrophages and astrocytes within and adjacent to the demyelinated lesions (fig. 6). Wallerian degeneration was not observed. Two patients had extensive transverse involvement of thoracic spinal cord at approximately one level. In 1 (Case 9), a large region of recent demyelination with sheets of lipid-laden macrophages and preserved axons involved more than half of a transverse section of thoracic spinal cord (fig. 7A, B). A focal area of chronic demyelination was present in the posterior columns within this lesion. The second patient (Case 2) had necrotizing transverse myelitis with vasculitis at this thoracic level and a small area of demyelination at a lumbar level. In both patients, these areas contained abundant inclusion bodies and VZV reactivity. Three of 9 patients did not have demyelination in the spinal cord. One (Case 3) had numerous inclusion bodies and VZV positivity in the posterior columns with little evidence of tissue disruption. The second (Case 5) showed cavitation near an anterior root entry zone with a narrow rim of partial necrosis in which inclusion bodies and VZV reactivity were detected. The third (Case 4) had no white matter lesions. Acute necrotizing vasculitis, found in 4 patients, was present around the involved segments of the spinal cord and brainstem in 3 patients and found only in the medulla in the fourth. The vasculitis was associated with haemorrhagic necrosis in 2 of the 3 patients with brainstem vasculitis and in 1 patient with spinal cord vasculitis. However, in 1 patient (Case 2) with complete transverse myelitis there was acute haemorrhagic necrosis, involving the entire cross-section of the spinal cord, accompanied by numerous
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
FIG. 3. Posterior horn with multiple intranuclear inclusion bodies of Cowdry type A in oligodendrocytes (arrow head) and a probable neuron (arrow). Note disintegrating neuropil with oedema. (Case 8). Hematoxylin-eosin. Bar = 100 /im.
HERPES ZOSTER
MYELITIS
inclusion bodies and cytoplasmic VZV immunoreactivity in adventitial and endothelial cells of the involved vessels and in adjacent leptomeningeal arachnoid cells. Four patients had an associated focal VZV encephalitis characterized by necrosis and demyelination associated with inclusion bodies and VZV immunoreactive cells. The involved regions included the nuclei gracilis and cuneatus in 3 patients (fig. 8), fourth ventricular ependyma in 2 patients, upper brainstem in 2 patients, and cerebellar cortex in 1 patient. Although all 4 patients with VZV encephalitis had leptomeningeal or parenchymal vasculitis, only a few foci were haemorrhagic and appeared to be related to vasculitis. Additional neuropathological changes included vascular myelopathy (Petito et al.,
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
FIG. 4. Large wedge-shaped lesion involving both posterior and lateral columns including the posterior root entry zone, A, total absence of myelin on Luxol fast blue staining. B, preserved meshwork of demyelinated axons on Bodian's silver stain. (Case 8, thoracic spinal cord).
1190
O. DEVINSKY AND OTHERS
FIG. 6. Higher magnification of demyelinated meshwork, illustrated in fig. 4, with immunocytochemica! staining of VZV antigen in a few astrocytes (arrow). Note total absence of oligodendrocytic nuclei in the field. (Case 8). Bar = 100 /im.
Downloaded from http://brain.oxfordjournals.org/ by guest on April 11, 2016
FIG. 5. Higher magnification of demyelinated meshwork, illustrated in fig. 4, with relatively well preserved axons. Note beaded appearance in some axons (arrows). (Case 8). Bodian's silver impregnation. Bar = 100 /im.
HERPES ZOSTER MYELITIS
1191
**•%
w
A
'
