Detection of Varicella Zoster Virus Reactivation in Cerebrospinal Fluid in Ischemic Stroke or Transient Ischemic Attack

Abstract

Background

Epidemiological data suggest that the risk of ischemic stroke increases after varicella‐zoster virus (VZV) reactivation. The frequency of VZV reactivation in acute ischemic stroke (AIS) is unknown. The risk of recurrent stroke and the antiviral treatment effect, particularly in patients with HIV infection, have yet to be defined. Here, we investigate the proportion of VZV reactivation in the cerebrospinal fluid (CSF) of patients who presented with AIS or transient ischemic attack and underwent VZV testing, along with relevant follow‐up information.

Methods

We retrospectively reviewed medical records of patients who presented with AIS or transient ischemic attack and underwent VZV polymerase chain reaction and anti‐VZV IgG testing in CSF during their workup from January 1, 2014, to December 31, 2021. VZV reactivation was confirmed by a positive VZV polymerase chain reaction result or increased intrathecal anti‐VZV IgG synthesis in CSF. The cause of AIS and transient ischemic attack was classified using the SSS‐TOAST (Stop Stroke Study‐Trial of ORG 10172 in Acute Stroke Treatment) criteria. The occurrence of recurrent ischemic stroke during follow‐up was compared between patients who received antiviral treatment and those who did not, as well as between patients with and without HIV.

Results

Among the 177 patients included, VZV reactivation in CSF was found in 23.2%. VZV reactivation was more common in strokes involving intracranial arteries compared with those that did not (28% versus 3%, P=0.01). Seven (17%) patients with VZV reactivation had recurrent ischemic stroke within 1 year from positive test results. There was no significant difference in recurrent ischemic stroke between patients who received short‐term antiviral therapy and those who did not (14.8% versus 21%, P=0.593). Patients with AIS and HIV had a significantly higher risk of VZV reactivation (57% versus 20%, P=0.002) and recurrent ischemic stroke despite antiviral treatment (42.8% versus 5.0%, P=0.02) compared with patients without HIV.

Conclusions

Among individuals who underwent VZV testing in CSF, VZV reactivation was present in 1 of 5 patients with AIS/transient ischemic attack. Patients with HIV were at particularly high risk of VZV reactivation and recurrent ischemic stroke.

Nonstandard Abbreviations and Acronyms

AIS

acute ischemic stroke

ICAD

intracranial atherosclerotic disease

SVD

small vessel disease

Clinical Perspective.

What Is New?

• Varicella zoster virus reactivation in cerebrospinal fluid was detected in patients with acute ischemic stroke involving small and large cerebral arteries, occurring at a higher rate among those with HIV infection.

• Short‐term antiviral therapy did not reduce the incidence of recurrent ischemic stroke, with a notably high recurrence rate observed among patients who were HIV positive.

What Are the Clinical Implications?

• The data support a clinical trial to evaluate whether recombinant zoster vaccine or antiviral therapy guided by cerebrospinal fluid positivity is effective in preventing recurrent stroke following varicella zoster virus associated acute ischemic stroke, especially in patients with HIV <50 years old.

Varicella‐zoster virus (VZV) is a highly contagious human alphaherpesvirus estimated to have infected 95% of the global population. ¹ After the initial infection, commonly presenting as varicella (chickenpox), the virus becomes latent in the ganglia along the entire neuroaxis. ² Upon reactivation, VZV travels along the peripheral nervous system, leading to herpes zoster. This typically presents as a painful vesicular rash in one or more dermatomes and can also result in cranial neuropathies and cerebritis. Invasion of the intracranial vasculature involving both small and large vessels can also occur. ³ This leads to vascular inflammation and complex pathological remodeling of the vasculature, causing cerebrovascular complications such as ischemic stroke, even without a skin rash.

The association between VZV infection and stroke has long been recognized, supported by case studies and epidemiological research. ⁴ , ⁵ Notably, the risk of stroke increases after VZV infection, particularly in cases of herpes zoster ophthalmicus, with reported hazard ratios as high as 4.28. ⁵ Although epidemiological studies indicate that this increased risk persists for up to 1 year, longer periods of VZV‐associated stroke risk have been reported following herpes zoster infection. ⁶

Despite this apparent link, studying VZV in acute ischemic stroke (AIS) remains challenging and limited, leaving many questions unanswered. Diagnosing VZV infection is complicated; positive polymerase chain reaction (PCR) results are reported to be less sensitive than increased intrathecal anti‐VZV IgG synthesis in diagnosing VZV infection. ¹ , ⁷ Consequently, both tests need to be performed and yield negative results to rule out VZV infection. Previous studies have reported rates of VZV reactivation ranging from 9% to 19% in young patients with stroke and patients with cryptogenic stroke and up to 37% in patients with HIV infection. ⁸ , ⁹ However, the duration of increased intrathecal anti‐VZV IgG synthesis in cerebrospinal fluid (CSF) is uncertain. ⁹ Moreover, it remains controversial whether antiviral treatment is necessary or effective when there is an elevation of anti‐VZV IgG titer, even in patients with HIV infection. ⁵ , ⁹ , ¹⁰

To further enhance our understanding of the role of VZV reactivation in cerebrovascular disease, we investigated the proportion of VZV reactivation in CSF among adult patients presenting with AIS or transient ischemic attack (TIA) who underwent VZV testing. We aimed to identify relevant demographic, clinical, and laboratory characteristics associated with VZV reactivation. Additionally, we conducted subgroup analyses based on stroke cause to confirm that elevated anti‐VZV IgG titers were specifically related to stroke involving cerebral arteries. Furthermore, we sought to shed light on the duration of elevated anti‐VZV IgG titers in CSF and the outcomes of antiviral treatment by providing follow‐up information.

METHODS

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request following execution of an appropriate data transfer agreement.

Study Population

This is a retrospective cohort study of patients who presented with AIS or TIA to Clements University Hospital and Parkland Health and Hospital System affiliated with University of Texas Southwestern Medical Center from January 1, 2014 to December 31, 2021. All patients were managed and underwent diagnostic evaluation in accordance with the American Heart Association/American Stroke Association guidelines for AIS or TIA. Patients were eligible for the study if they were evaluated for VZV reactivation in the CSF (Figure 1). This consisted of VZV PCR assay in CSF and anti‐VZV IgG antibody quantitative assay in both CSF and serum samples. The decision to perform a lumbar puncture was based on clinical suspicion at the discretion of the treating physicians. The decision to perform VZV testing was influenced by various factors, including the presence of concurrent herpes zoster, cranial nerve palsy observed during physical examination, and a history of cognitive impairment. Unexplained symptoms such as altered mental status or facial pain upon presentation also prompted testing. Additional considerations were seizures at presentation, imaging findings like arterial ectasia or dolichoectasia on angiogram, progression of arterial stenosis as assessed by the reader, and moyamoya vasculopathy associated with intracranial atherosclerotic disease (ICAD). Other factors included recurrent ischemic stroke despite medical therapy, multiple small acute infarcts in deep or subcortical white matter, an undetermined stroke cause with multiple possible causes, and the need to rule out infectious processes, such as neurosyphilis or neurocysticercosis. In cases with uncertain cause, after VZV tests were ordered, the workup for stroke cause continued during clinic follow‐up, including but not limited to prolonged cardiac monitoring and genetic tests.

Figure 1. Inclusion and exclusion criteria for selecting patients presenting with acute ischemic stroke/transient ischemic attack who underwent VZV reactivation testing with subsequent follow‐up.

AIS indicates acute ischemic stroke; CSF, cerebrospinal fluid; PCR, polymerase chain reaction; TIA, transient ischemic attack; and VZV, varicella zoster virus.

Exclusion criteria included cases where a final diagnosis could not be made due to incomplete or absent diagnostic workup, incomplete VZV diagnostic assay, or inadequate CSF specimen quality (Figure 1). For example, negative VZV PCR without anti‐VZV IgG test in CSF was considered incomplete and not eligible. CSF sample was considered contaminated in cases of traumatic tap or blood brain barrier rupture (defined as serum over CSF albumin ratio <50 or total IgG ratio <100).

VZV Test and Diagnostic Criteria for VZV Reactivation in CSF

VZV PCR testing in CSF and anti‐VZV IgG serology in both CSF and serum were performed at the Centers for Disease Control and Prevention (CDC) in Atlanta, Georgia, and at Associated Regional and University Pathologists (ARUP) Laboratories in Salt Lake City, Utah. At the CDC, anti‐VZV IgG serology was measured using a quantitative glycoprotein‐based assay. At ARUP Laboratories, anti‐VZV IgG serology was measured using a semiquantitative chemiluminescent immunoassay. The CDC reports positive VZV DNA results as either wild‐type or vaccine strain, whereas ARUP Laboratories do not specify the strain type in their reporting. VZV reactivation in CSF was diagnosed by either positive VZV PCR in CSF or increased intrathecal anti‐VZV IgG antibody synthesis. The latter was defined as serum over CSF anti‐VZV IgG antibody ratio ≤10. ⁴ , ⁶ A 4‐fold increase in anti‐VZV IgG antibody titers in serum or CSF on 2 separate tests also qualified as VZV reactivation. ¹¹ Due to the absence of biopsy or autopsy studies to elucidate the specific underlying pathology of VZV reactivation in CSF, such as VZV vasculitis or VZV vasculopathy, these terms are not used in this article.

Clinical, Imaging, and Laboratory Data

For eligible patients, we reviewed and collected demographic information, stroke risk factors, history of immunosuppression (including autoimmune diseases, organ transplantation, chemoradiation, or HIV status), concurrent herpes zoster skin lesions (at the time of admission or during hospitalization), and any prior history of herpes zoster (within the past 4 months or earlier). We also collected stroke workup results obtained during hospitalization and clinic follow‐up. Stroke workup data included brain magnetic resonance imaging (MRI) along with angiography of the intra‐ and extracranial vasculature, echocardiogram, cardiac computed tomography/MRI, ECG, and cardiac monitoring findings. Laboratory tests comprised lipid panel, hemoglobin A1C, hypercoagulable workup, autoimmune panel, genetic testing, biopsy results, and other pertinent findings to support the final diagnosis and stroke cause classification.

Patients were followed up in the stroke clinic as standard practice. Follow‐up evaluation was performed by reviewing the electronic medical records of clinic visits, emergency department visits, and hospital admissions. The follow‐up length was defined as the time from the date of the index AIS/TIA onset to the date of recurrent stroke or to the date of the last clinic visit, emergency department visit, or hospital admission. The primary outcome was recurrent ischemic stroke during follow‐up. New ischemic stroke was defined by the presence of new or worsening neurological symptoms with a corresponding positive diffusion‐weighted imaging finding on MRI. Incidental infarcts on MRI without neurological symptoms were not counted as recurrent ischemic strokes.

Acute Ischemic Stroke and Transient Ischemic Attack Diagnosis and Cause Classification

Based on clinical presentation, imaging, and supporting laboratory data obtained during hospitalization and follow‐up, a most likely diagnosis was made for each patient's symptoms, classifying them as AIS or TIA. The cause was categorized based on a modified SSS‐TOAST (Stop Stroke Study‐Trial of ORG 10172 in Acute Stroke Treatment) classification into large vessel atherosclerotic disease, small vessel disease (SVD), cardioembolic, other determined, or undetermined. ¹² (Table S1). The large vessel atherosclerotic disease subgroup was further differentiated as ICAD or extracranial atherosclerotic disease.

These classification processes were independently performed by 3 vascular neurologists (T.S., W.L. and C.R.) who were blinded to CSF VZV data. A consensus was reached by all 3 reviewers in terms of the most likely diagnosis, as well as cause classification of AIS or TIA. The level of confidence (possible, probable, or evident) was not used to describe subcategories.

Statistical Analysis

VZV positive and negative patients were evaluated via categorical and continuous variables comparing demographics, cardiovascular risk factors, immunosuppression, history of herpes zoster, and CSF profile markers. Categorical variables were analyzed using the chi‐square test or Fisher Exact test, and continuous variables were analyzed using 2‐sample t test (assuming equal variance). Multivariable logistic regression analysis was used to determine association of various risk factors with VZV reactivation. For all analyses, P values ≤0.05 considered statistically significant. Given that the nature of this work was hypothesis generating, correction for multiple analyses was not performed. All analyses were performed using SAS 9.2 and IBM SPSS Statistics.

Standard Protocol Approvals, Registrations, and Patient Consents

This study was approved by the Institutional Review Boards at University of Texas Southwestern Medical Center (STU 052017‐041). As a retrospective cohort study, the need for individual informed consent was waived. We have adhered to Strengthening the Reporting of Observational Studies in Epidemiology guidelines for observational studies.

RESULTS

Characteristics of Study Population and VZV Reactivation

A total of 220 patients were identified for this retrospective study. VZV PCR and anti‐VZV IgG testing in CSF and serum were performed in 202 patients (92%) at the CDC in Atlanta, and in 18 patients (8%) at ARUP Laboratories in Utah. Seventeen patients were excluded due to incomplete or inadequate CSF sample quality (n=14), unclear diagnosis (n=2), or incomplete imaging workup (n=1). Out of the remaining 203 patients, we identified 177 (87%) who had either AIS (n=164, 93%) or TIA (n=13, 7%) and were eligible for the study. The mean age was 57.6±14.6 years and 54% were male (Table S2).

Diagnosis of VZV reactivation in CSF was confirmed in 38 patients based on either a positive VZV PCR test in CSF (n=13) or elevated intrathecal anti‐VZV IgG antibody synthesis (n=25). An additional 3 patients were diagnosed based on a more than 4‐fold increase in anti‐VZV IgG antibody levels in serum (n=1) or CSF (n=2) on 2 separate tests, coupled with clinical symptoms. In total, 41 out of 177 cases (23.2%) were diagnosed with VZV reactivation in the CSF.

There were a total of 14 patients with HIV, 17 patients with other types of immunosuppression, and 146 patients without any immunosuppression. VZV reactivation was significantly higher among patients with HIV compared with patients without HIV (n=8, 57% versus n=33, 20%, P=0.002), and compared with patients without any immunosuppression (n=8, 57% versus n=27, 18%, P=0.0008). Additionally, patients who were HIV positive were younger than those without HIV (41.8 ± 12 years versus 58.8 ± 14 years, P < 0.001). There was no significant difference between patients with HIV and those with other forms of immunosuppression, nor between patients with other immunosuppression and those without any immunosuppression (Figure 2). Among patients who had neither any form of immunosuppression nor a recent history of herpes zoster, there were 23 who were VZV positive (16%).

Figure 2. VZV reactivation rates by immunosuppression status.

VZV reactivation was observed in 57% of patients with HIV, 35% of patients with other forms of immunosuppression (excluding HIV), and 18% of patients without any immunosuppression. Among patients without HIV, the overall VZV reactivation rate was 20%. An asterisk symbol (*) denotes a significant difference between patients with HIV and patients without any immunosuppression (P=0.001). A hash symbol (#) indicates a significant difference between patients with HIV and those without HIV (P=0.002). VZV indicates varicella zoster virus.

Among the 12 positive PCR tests conducted at the CDC—including 2 cases where PCR was repeated on CSF samples—all were identified as the wild‐type VZV strain. The 3 positive PCR tests performed at ARUP Laboratories did not report whether the strain was wild‐type or vaccine‐type.

Risk Factors Associated With VZV Reactivation in AIS or TIA

Out of 177 patients, 41 (23.2%) were identified as VZV positive. Compared with patients who were VZV negative (n=136), there was no statistically significant difference regarding demographics (sex, age) or cardiovascular risk factors—including hypertension, hyperlipidemia, diabetes, smoking, heart failure, atrial fibrillation, and history of ≥2 ischemic strokes (Table 1). However, the VZV‐positive group demonstrated a significantly higher percentage of immunosuppression (34% versus 13%, P=0.0014), as well as a history of herpes zoster within 4 months (20% versus 1.5%, P < 0.001) or before 4 months (32% versus 16%, P=0.005). Other clinical features of patients who were VZV positive and VZV negative were provided in Table S3. Compared with the group who were VZV negative, the group who were VZV positive included more patients presenting with cranial nerve palsy, altered mental status, and seizures. Although there was no overall difference in headache or facial pain between the 2 groups, a higher number of patients with facial pain or itchiness was noted in the VZV‐positive group (3, 7% versus 0, 0%; P=0.001). No differences were observed between the 2 groups regarding the history of cognitive impairment or monocular vision loss. Myeloradiculitis was not identified in any patient within this cohort.

Table 1.

Demographic and Clinical Characteristics Between Patients Who Were VZV Positive and Negative Who Experienced Ischemic Stroke or Transient Ischemic Attack

Demographics	VZV positive	VZV negative	P value
	n=41	n=136
Age, y (mean±SD)	58.4 ±15.2	57.3 ±14.5	0.68
Sex
Male	25 (61%)	70 (51%)	0.284
Female	16 (39%)	66 (49%)
Race or ethnicity
White	12 (29%)	44 (32%)	0.057
Black	24 (58%)	48 (35%)
Hispanic	5 (12%)	40 (29%)
Asian	0	3 (2%)
Non‐Hispanic/Latino	0	1 (0.7%)
Past medical history
Hypertension	39 (95%)	124 (91%)	0.673
Hyperlipidemia	34 (83%)	106 (78%)	0.474
Diabetes	21 (51%)	70 (51%)	0.978
Smoking	20 (49%)	58 (43%)	0.488
Congestive heart failure	5 (12%)	11 (8%)	0.646
Atrial fibrillation	5 (12%)	19 (14%)	0.771
History of >2 ischemic strokes	8 (20%)	24 (18%)	0.786
Current immunosuppression	14 (34%)	17 (13%)	0.0014
Autoimmune	5	7
HIV	8	6
Chemotherapy for malignancy/radiation	1	1
Other	0	1
Multiple	0	2
None	27	119
Remote history of herpes zoster (>4 mo)	13 (32%)	22 (16%)	0.005
Recent history of herpes zoster (within 4 mo)	8 (20%)	2 (1.5%)	<0.001
Active herpes zoster skin lesions during stroke/transient ischemic attack	7 (17%)	0	N/A
CSF white blood cell=or >5	17 (41%)	23 (17%)	0.001
CSF protein=or >42 mg/dL	32 (78%)	68 (50%)	0.001

CSF indicates cerebrospinal fluid; and VZV, varicella zoster virus.

There were 7 patients (17%) among the 41 who were VZV positive who exhibited active (concurrent) herpes zoster skin lesions, either at the time of admission (n=5) or during hospitalization (n=2). The dermatomal locations of the zoster in these patients were right T5, right C6, left C5–T1, left T12, right T1–T2, right T12, and right V1 herpes zoster ophthalmicus.

In terms of CSF analysis, patients who were VZV positive demonstrated significantly higher rates of pleocytosis—defined as an elevated white blood cell count of ≥5—with 41% of cases compared with 17% in patients who were VZV negative (P=0.001). They also exhibited elevated protein levels—defined as ≥42 mg/dL—in 78% of cases versus 50% in patients who were VZV negative (P=0.001) (Table 1).

In a multivariable logistic regression analysis, the elevated protein levels and pleocytosis in the CSF and cranial nerve palsy at presentation emerged as the factors associated with VZV reactivation in the CSF (Table S4). Factors analyzed included immunosuppression, history of herpes zoster (including remote or recent herpes zoster), elevated protein levels and pleocytosis in the CSF, seizure, altered mental status, and cranial nerve palsy at presentation. These model covariates were selected based on significance level on univariate analysis.

VZV Reactivation Among Different AIS/TIA Causes

To confirm the specificity of VZV reactivation in stroke, we aimed to investigate the proportion of VZV reactivation across different stroke subtypes. The rationale is that if VZV reactivation is indeed a risk factor for stroke, it would most likely affect strokes involving cerebral arteries based on the pathophysiology of VZV infection. It is unlikely to be implicated in strokes caused by noncerebral artery causes, such as cardiac embolism, or other known cause, such as genetic diseases.

Patients with AIS and TIA (n=177) were categorized using modified SSS‐TOAST criteria, which revealed ICAD to be the most common cause (n=68, 38.4%), followed by undetermined (n=39, 22%), and SVD (n=33, 18.6%) (Table 2). There were 4 patients (2.3%) whose causes were attributed to VZV reactivation because they had skin lesions consistent with herpes zoster on admission while blinded to VZV test results. All 4 patients were confirmed positive based on CSF studies. These 4 cases were not included in further analysis among different causes.

Table 2.

Breakdown of Stroke Cause Based on SSS‐TOAST Mechanism With Corresponding Number of VZV Reactivations per Cohort Based on CSF Analysis

SSS‐TOAST mechanism	Patients (% of stroke/TIA)	VZV positive
Intracranial atherosclerotic disease	68 (38.4%)	18 (26.5%)
Extracranial atherosclerotic disease	3 (1.7%)	0
Small vessel disease	33 (18.6%)	11 (33.3%)
Cardioembolism	10 (5.6%)	0
Other determined	20 (11.3%)	1 (5.0%)
Undetermined	39 (22.0%)	7 (17.9%)
VZV (clinical diagnosis)	4 (2.3%)	4 (100%)

CSF indicates cerebrospinal fluid; SSS‐TOAST, Stop Stroke Study‐Trial of ORG 10172 in Acute Stroke Treatment; TIA, transient ischemic attack; and VZV, varicella zoster virus.

When compared with the combined stroke cause group consisting of extracranial atherosclerotic disease, cardioembolism, and other determined causes (n=1, 3%), VZV reactivation in CSF was significantly higher in the group with SVD (P=0.001), the group with ICAD (P=0.005), and the undetermined group (P=0.045) (Figure 3). Overall, VZV reactivation in CSF was observed in 28.7% (n=29) of individuals within the intracranial artery disease groups (either ICAD or SVD), which was significantly higher compared with the combined group of extracranial atherosclerotic disease, cardioembolism, and other determined causes (3%, n=1, P=0.01). Notably, the single patient categorized under “other determined” cause due to antiphospholipid syndrome was also found to have VZV reactivation.

Figure 3. VZV reactivation rate per each stroke mechanism group.

Combined extracranial atherosclerotic disease, cardioembolic, and other determined (3%, 1/33); intracranial atherosclerotic disease (26.5%, 18/68, P=0.005^#); small vessel disease (33.3%, 11/33, P=0.001^#); and undetermined (17.9%, 7/39, P=0.045^#). Note: # indicates statistically significant rate of VZV reactivation compared with combined group. ECAD indicates extracranial atherosclerotic disease; ICAD, intracranial atherosclerotic disease; SSS‐TOAST, Stop Stroke Study‐Trial of ORG 10172 in Acute Stroke Treatment; SVD, small vessel disease; and VZV, varicella zoster virus.

Follow‐Up of Cohort of VZV Reactivation in Patients With AIS/TIA

Among the 41 patients in the group with AIS/TIA with VZV reactivation, 27 received treatment, which varied based on their CSF results. For patients with a positive PCR in the CSF, intravenous acyclovir was administered for 2 weeks, sometimes followed by several months of oral valacyclovir. For those with increased intrathecal anti‐VZV IgG synthesis, treatment included intravenous acyclovir for 14 days followed by oral valacyclovir for 3 to 6 months or oral valacyclovir alone for the same duration without preceding intravenous therapy.

Fourteen patients did not receive treatment. The reasons for not treating these patients included patient refusal or noncompliance, inability to reach the patient or personal health proxy, uncertainty about the significance of results due to multiple stroke causes at presentation, decisions made by the treatment team, and early death of the patient.

First, we aimed to determine the duration of increased intrathecal anti‐VZV IgG synthesis. We identified 10 patients who underwent repeated CSF studies, totaling 22 CSF analyses including the initial CSF tests. Two patients had 2 repeated CSF tests each due to positive VZV PCR results in the CSF on their first repeat tests. Interestingly, both patients had negative VZV PCR in CSF on the initial tests. Among these 10 patients, 8 received treatment for varied times, and 2 did not receive treatment. Seven CSF samples were repeated and remained positive at an average of 205±67 days. Five CSF samples were repeated at an average of 370±199 days, and the increased IgG titers had resolved.

Four patients were lost to follow‐up. The remaining 37 patients were each followed for >1 year, with a mean follow‐up duration of 1108±986 days. During the follow‐up period, a total of 10 ischemic strokes and 1 subarachnoid hemorrhage occurred, and 11 deaths were recorded. Because the increased intrathecal anti‐VZV IgG synthesis appeared to resolve around 1 year after onset, and epidemiological data indicate a heightened stroke risk within 1 year following infection, ⁵ we evaluated the incidence of recurrent ischemic stroke within 1 year from the first CSF study in 41 patients. Among these patients, ischemic stroke occurred in 7 (17%) individuals within 1 year. In an intention‐to‐treat analysis, there was no significant difference in the incidence of recurrent ischemic stroke between patients who received antiviral treatment and those who did not (n=4, 14.8% versus n=3, 21%; P=0.593). However, patients with HIV had a significantly higher risk of recurrent ischemic stroke compared with those without HIV, even after treatment (3 out 7, 42.8% versus 1 out of 20, 5%; P=0.015). There was no significant difference in clinical characteristics and treatment strategy between patients with HIV with and without recurrent stroke (Table 3). The status of recombinant zoster vaccine was not documented.

Table 3.

Clinical Characteristics of Patients With HIV Receiving Antiviral Therapy, With and Without Recurrent Stroke

Clinical characteristics	Recurrent stroke	No recurrent stroke
	n=3	n=4
Age (y, SD)	39.7±11.8	33.5±12.2
Sex (M)	1 (33%)	3 (75%)
CD4 cells counts at treatment (cells/mL)	64 (±69)	210 (±235)
Treatment duration (IV and oral) (d)	25 (±14)	15 (±5)
Pulse prednisone treatment	0	1
Suppressive therapy	3 (100%)	1 (25%)
Time for index stroke to recurrent stroke (d)	100±68	NA

P values were not calculated due to small sample sizes.

DISCUSSION

In our study, we found that VZV reactivation in the CSF was present in 23.2% of patients with AIS or TIA who underwent VZV testing in CSF per treating team. Patients with AIS and HIV had a significantly higher rate of VZV reactivation compared with those without HIV (57% versus 20%; P=0.002), and those without any immunosuppression (57% versus 18%, P=0.001). Previous studies reported VZV reactivation rates of 37% in patients with stroke and HIV, 9% in young patients with stroke without HIV, and 19% in cryptogenic stroke. ⁸ , ⁹ The rates observed in our study are higher than those reported in previous studies for patients with HIV and without any immunosuppression, which may be attributed to differences in patient populations and study methodologies.

Testing for VZV reactivation in patients with AIS has been controversial. ⁷ , ¹³ There are concerns that VZV reactivation is not specific to stroke and cannot be definitively attributed as a causative factor. Our study showed a higher rate of VZV reactivation in CSF (28.7%) in patients with AIS/TIA with causes secondary to intracranial artery disease of large or small vessels (ICAD or SVD). In contrast, VZV reactivation was found in only 3% of AIS/TIA patients with causes not involving intracranial arteries including extracranial atherosclerotic disease, cardioembolism, hypercoagulable states, and other causes believed not to be associated with VZV infection, such as primary central nervous system vasculitis, reversible cerebral vasoconstriction syndrome, and cerebral autosomal recessive arteriopathy with subcortical infarcts and leukoencephalopathy. These findings are consistent with anatomical models of VZV reactivation and invasion of intracranial arteries. ¹⁴ , ¹⁵ , ¹⁶ It suggests that increased intrathecal anti‐VZV IgG synthesis is not merely incidental, reinforcing the current understanding of VZV reactivation in AIS.

Another potential pathophysiological mechanism of VZV reactivation in ischemic stroke involves acceleration of underlying atherosclerosis and lipohyalinosis in the setting of immune‐mediated inflammation. A recent study involving an ethnically diverse Manhattan population have linked chronic viral infections to the development of intracranial large artery stenosis. ¹⁷ VZV has been shown to trigger the activation of NLRP3 (NOD‐like receptor family pyrin domain containing 3) inflammasome complex, which leads to atherosclerotic plaque progression and instability, as well as formation of proinflammatory cytokines. ¹⁸ , ¹⁹ Our CSF studies of patients with AIS/TIA have demonstrated significantly higher rates of pleocytosis and elevated protein among the cohort with VZV reactivation compared with the cohort with negative VZV (Table 1). Although we did not perform a cytokine analysis of the CSF samples from the patients with stroke and VZV reactivation, prior cytokine profiling of CSF from meningitis following VZV reactivation has shown elevated levels of interferon gamma and IL‐6 (interleukin‐6), IL‐8, and IL‐10 as well as chemokines CXCL‐9 (chemokine ligand 9) and CXCL‐10. ²⁰

There is no consensus whether treatment is necessary after VZV reactivation is detected in the CSF in ischemic stroke, particularly when there is increased intrathecal anti‐VZV IgG synthesis without a positive PCR result. ⁹ , ¹⁰ , ²¹ , ²² If treatment is initiated, the optimal duration remains uncertain. In our study, there was no significant difference in recurrent ischemic stroke after short‐term antiviral treatment. Most important, we showed that patients with HIV and AIS were at a significantly higher risk of recurrent ischemic stroke despite antiviral treatment (42.8% versus 5%, P=0.02) compared with patients without HIV. Of note also, patients with HIV and AIS are typically younger than those without HIV. Future studies could explore the use of CSF positivity as a guide for determining the appropriate duration of antiviral therapy, especially among the population with HIV. Another modality to be considered is immunization of patients with stroke and VZV reactivation using a zoster vaccine. Analyses of 2 large data sets of the general population (Medicare and Veterans Administration) have shown a lower incidence of stroke in vaccinated populations. ²³ , ²⁴ The first study included data from only live zoster vaccination, but the second study included data from both live and recombinant zoster vaccination. Live zoster vaccine should no longer be administered to patients with HIV because of serious adverse events, including acute retinal necrosis and even 1 death. ²⁵ , ²⁶ Instead, the CDC recommends a 2‐dose series of recombinant zoster vaccine for all adults aged ≥18 with HIV. ²⁷ Therefore, consideration should be given to a large‐scale and long‐term study of recombinant zoster vaccination of patients with stroke and VZV reactivation, including patients with HIV, with a goal of reducing the incidence of stroke.

There are several limitations to this retrospective study. First, there was likely a large interprovider variability in terms of threshold for ordering CSF VZV tests as part of stroke workup, leading to sampling bias for patients eligible for our study. As a result, our rate of VZV reactivation did not accurately reflect the true prevalence. The small number of patients with HIV is also a limiting factor. We did not find documentation on whether patients with VZV reactivation had received the live zoster vaccine (Zostavax Merck), which has been reported to cause VZV reactivation and herpes zoster. ²⁸ However, in our cases tested at the CDC in Atlanta, only wild‐type VZV DNA was reported; no vaccine strains were detected. For eligible patients, key clinical histories and comorbidities may have been omitted by providers who had a low index of suspicion for VZV reactivation as a form of confirmation bias for other causes. Furthermore, the modified SSS‐TOAST classification for AIS or TIA cause was based on independent review of imaging and medical chart by 3 vascular neurologists. Although disagreements were resolved by a consensus group discussion, potential interpretation biases exist. Finally, we did not review MRI vessel wall imaging, its pattern of enhancement would have been useful in elucidating atherosclerotic versus viral inflammation as the cause of ischemic stroke. To surmount some of these limitations, more detailed investigations of CSF samples of patients with ischemic stroke will be required, including the increased use of multiplex diagnostic pathogen assays and multiplex immunoassays, to elucidate the magnitude of the role of VZV infection in stroke and the optimal treatment strategy.

Sources of Funding

There is no funding for this retrospective stroke study. Research by Charles Grose on varicella virus pathogenesis is supported by National Institutes of Health grant AI153817.

Disclosures

Wenyang Li, Chintan Rupareliya, Suriya Subramanian, Hisham Salahuddin, Khalil S. Husari, William Moore, Mark Johnson, Alejandro Magadan, and Ty Shang have no disclosures.

Peter Sguigna received research support from the National Multiple Sclerosis Society/International Progressive MS Alliance, Patient‐Centered Outcomes Research Institute, Genentech, Clene Nanomedicine, the National Institutes of Health, and the Department of Defense/Congressionally Directed Medical Research Programs. He has received consulting fees from EMD Serono, Genentech, Horizon Therapeutics, and Bristol Myers Squibb.

Charles Grose received funding for research on varicella virus pathogenesis from National Institutes of Health grant AI153817.

Ank E. Nijhawan received research funds from Gilead Sciences.

Supporting information

Tables S1–S4