Neurological complications of varicella zoster virus reactivation: Prognosis, diagnosis, and treatment of 72 patients with positive PCR in the cerebrospinal fluid

Abstract Background VZV infection can involve every level of the neurologic system: from the central nervous system (CNS) to the peripheral nervous system (PNS), including aseptic meningitis. Prognosis seems to differ between these neurological involvements. Prognostic factors remain unknown. Methods This is a retrospective multicenter study including all patients with a positive VZV polymerase chain reaction (PCR) in the cerebrospinal fluid (CSF) from eight centers in Paris (France) between 2011 and 2018. Unfavorable outcome was defined as mortality linked to VZV or incomplete recovery. Modified Rankin Scale (mRS) evaluated disability before and after the infection, with the difference designated as Rankin Delta. Results Seventy‐two patients were included (53% male, median age 51 years, median mRS 0). Immunosuppression was reported in 42%. The clinical spectrum included 26 cases of meningitis, 27 instances of CNS involvement, 16 of PNS involvement, and 3 isolated replications (positive PCR but no criteria for neurological complications from VZV). Antiviral treatment was administered to 69 patients (96%). Sixty‐two patients completed follow‐up. Death linked to VZV occurred in eight cases. Unfavorable outcome (UO) occurred in 60% and was significantly associated with a higher prior mRS (Odd‐ratio (OR) 3.1 [1.4–8.8] p = .012) and the presence of PNS or CNS manifestations (OR 22 [4–181] p = .001, OR 6.2 [1.3–33] p = .03, respectively, compared to meningitis). In the CSF, higher protein level (p < .0001) was also significantly associated with a higher Rankin Delta. Conclusions Neurological complications of VZV with evidence of CSF viral replication are heterogeneous: aseptic meningitis has a good prognosis, whereas presence of CNS and PNS involvement is associated with a higher risk of mortality and of sequelae, respectively.


INTRODUCTION
Varicella zoster virus (VZV) is a double stranded DNA virus from the Herpesviridae family (Davison & Scott, 1986). After primary infection, the virus establishes latency principally in vertebral and cranial sensory ganglia which are the sources of reactivation as cellular immunity decreases (Cohrs & Gilden, 2007;Levin et al., 2003). VZV reactivation may be complicated by either local, neurological, or disseminated involvement. Among many identified risk factors, age and cellular immunodepression have been identified as the most important for VZV reactivation and complications (Kawai et al., 2014;Schmader, 2001).
Neurological complications from VZV are of primary concern. The virus can cause peripheral nerve involvement (cranial nerve including Ramsay-Hunt syndrome, radiculopathy) and central neurological involvement (meningitis, meningoencephalitis, myelitis, stroke, VZV vasculopathy, necrotizing retinitis) Herlin et al., 2021;Nagel & Gilden, 2014;Steiner & Benninger, 2018). VZV vasculopathy, classically present as an ophthalmic zoster followed by acute contralateral hemiplegia or as headache, mental status changes, or focal deficit, may be underestimated in clinical practice because deficit may appear up to one year after herpes zoster (HZ) (Gilden et al., 2009;Sreenivasan et al., 2013). Several studies have reported a potential role for VZV in neurocognitive impairment (Chen et al., 2018;Itzhaki, 2018;Tsai et al., 2017) or giant cell arteritis . Neurological manifestations can precede or even present without skin lesions of HZ as in zoster sine herpete (Blumenthal et al., 2011;Gilden et al., 2010).
Involvement of the central nervous system (CNS), and especially myelitis or cerebral vasculitis, were described as associated with immunosuppression, but data are controversial as it has not been shown by more recent studies (Choi et al., 2014;Corral et al., 2020;Gilden et al., 1994;Nagel et al., 2008). Only a few studies have specifically addressed VZV meningitis, as cases are often reported among other causes of viral meningitis, or among the large dataset of all VZV neurologic complications (Aberle et al., 2005;Corral et al., 2020;Nagel & Gilden, 2014). Follow-up data are scarce to provide insights on potential neurologic and general sequelae of the infection (Becerra et al., 2013;Frantzidou et al., 2008;Grahn & Studahl, 2015;Kaewpoowat et al., 2016;Nowak et al., 2003;Persson et al., 2009;Pollak et al., 2012).
Our hypothesis was that VZV aseptic meningitis, when not associated with other neurological complications, had a favorable prognosis compared to central or peripheral neurologic involvements. Our objectives were to describe the clinical and biological presentation, the therapeutic management (especially of the aseptic meningitis for which it is not codified), the morbimortality of VZV neurological complications, confirmed by positive VZV polymerase chain reaction (PCR) in the cerebrospinal fluid (CSF), and to look for prognostic risk factors. To fulfill these objectives, we conducted a multicenter retrospective study with one follow-up update assessing the neurologic and general outcomes of patients with VZV neurological complications.

Study design
We conducted a retrospective multicenter observational study that involved 20 medical departments from four French tertiary hospitals: Hôpital

Ethics
With the agreement of the heads of the concerned departments, medical files were de-identified and consulted in each center.

Data collection
Clinical data were collected from the electronic medical records.

Definitions of neurologic complications
Among all patients with positive VZV PCR in the CSF, complications were defined as detailed in Table 1 (Gilden et al., 1994;Persson et al., 2009;Robillard et al., 1986;Steiner & Benninger, 2018;Venkatesan et al., 2013). Meningitis was defined by the association of a pleocytosis (≥5 nucleated cells/mm 3 in the CSF) and suggestive symptoms of meningitis, with negative CSF bacterial cultures and without focal central or peripheral symptoms. After a careful review of medical charts, patients with alternative diagnoses not responding to the VZV neurological complications classification were excluded.

Medical care and follow-up
Hospital stay data (location, duration, intensive care unit requirement), treatments (oral or intravenous antiviral therapy, dosage, delay, and duration of treatment), and discharge data were collected. At followup contact, sequelae were assessed by the patient and the main investigator. The patient determined the recovery date (defined as the time of full recovery or when symptoms were stable). An unfavorable outcome was defined by an incomplete recovery (any persistent symptom or sequelae) and/or death related to VZV infection. A favorable outcome was defined by a full recovery (no sequelae, no persistent symptom in comparison to the status before infection). Level of disability at recovery date was assessed by the mRS. Change in mRS was defined by the difference from baseline mRS (mRS delta = mRS at recovery date -prior to infection mRS). The type of sequelae was specified by binary questions and free comments by the patients.

Statistical analysis
An analysis was conducted to identify clinical or biological prognostic factors that were associated with: a) unfavorable outcome (mortality and/or incomplete recovery) and b) increase of mRS delta.
Median delay between onset of headache, zoster, fever, and the lumbar puncture were 3, 2, and 2 days, respectively.

DISCUSSION
We report a large series of VZV neurological complications with positive CSF VZV PCR, describing 72 patients, with long-term follow-up available in 86% of cases. The observed neurological complications were heterogeneous and involved every level of the neurologic system from CNS to PNS with also milder forms of aseptic meningitis. The current study adds greatly in the effort to describe the clinical presentation of these types of patients and also to identify prognostic risk factors associated with a higher morbimortality.
With the routine measurement of VZV PCR in the CSF, more cases are being identified and VZV neurological complications are increasingly being reported. For example, VZV is now considered among the three first causes (along with enterovirus and HSV) of viral meningitis (Kupila et al., 2006). More data about outcome and prognosis are needed to guide therapeutic strategies, especially for VZV aseptic meningitis. Indeed, these forms are strongly felt to be milder than CNS involvements, but, to date, no study gives sufficient reassuring information to lighten their treatment compared to meningoencephalitis therapeutic recommendations (Stahl et al., 2017). A major limitation of our knowledge of aseptic meningitis outcomes is the paucity of data specif-  ( Corral et al., 2020;De La Blanchardiere et al., 2000;Haanpaa et al., 1998;Ihekwaba et al., 2008;Kaewpoowat et al., 2016;Pahud et al., 2011;Persson et al., 2009;Skripuletz et al., 2018;Tabaja et al., 2020), but follow-up data were available for only 12 patients at 6 months (Corral et al., 2020;Persson et al., 2009). Other studies have included meningoradiculitis with meningitis cases (Aberle et al., 2005;Echevarria et al., 1994;Kim et al., 2017), which in our opinion, reflects a different pathophysiological process leading to different sequelae. To date, there is also poor evidence about the role of immunosuppression in the outcome of VZV neurological complications.
Our most notable finding was that patients with meningitis had a favorable outcome (in terms of mortality and sequelae) compared to patients with complications involving the PNS or CNS (of note, 24 out of 26 patients with meningitis received antiviral treatment).
Most patients with meningitis were young, had no significant past medical history, and had a typical aseptic meningitis presentation (fever, headache, and meningeal signs). HZ was present in two-thirds of patients, but as previously described, its absence did not rule out the diagnosis of VZV meningitis. Inflammatory syndrome was low, and the most frequent CSF abnormality was a lymphocytic pleiocytosis. Presence of polynuclear neutrophils or low glucose in the CSF did not rule out the diagnosis. Thus, an aseptic meningitis with hypoglycorrhachia could evoke a VZV infection as described in 40% of the patients in our cohort. In the meningitis group, no patient died from VZV infection and only six (24%) reported an incomplete recovery which was significantly lower than in the CNS and PNS groups.
In contrast, members of the CNS group had poorer outcomes than previously described in published cohorts (Corral et al., 2020;Grahn & Studahl, 2015) with a higher mortality (36%) and a high rate of sequelae (45% with VZV encephalitis having positive VZV PCR in the CSF or intrathecal anti-VZV IgG. As the latter are only rarely used in practice in France, they were not part of our inclusion criteria or data collection, which may constitute a selection bias. Follow-up data permitted us to assess two well-defined outcomes (full recovery and mRS delta) and to conduct a multivariate analysis that identified risk factors associated with a poorer prognosis: higher age, higher prior-to-infection mRS, CNS or PNS involvement compared to aseptic meningitis, and CSF protein level. Older age is a well-known risk factor for VZV reactivation and complications (Kaewpoowat et al., 2016;Nagel & Gilden, 2014;Schmader, 2001). Elevated CSF protein level had not been previously described as associated with a poorer prognosis. This could reflect the alteration of the blood brain barrier by the viral infection (or the intensity of the inflammation associated with the viral infection and/or immune response). Viral load was not associated with prognosis, which is in accordance with one (Rottenstreich et al., 2014), out of two studies (Persson et al., 2009) that assessed this parameter. Interestingly, immunosuppression was not associated with mortality or incomplete recovery, as reported in another recent study (Corral et al., 2020).
Our study was limited by the potential for misclassification bias.
Patients in our cohort were not classified in groups according to file The other biases related to our study design were minimized by 1) the inclusion of all patients with positive CSF VZV PCR during the study period in the participating centers, limiting recruitment bias; 2) the stringent judgment criterion-full recovery is subjective but clinically relevant and subject to overestimation of sequelae rather than underestimation; and 3) updated follow-up data by direct patient contact.
This study was not designed to investigate VZV vasculopathy, as VZV PCR in the CSF is not routinely prescribed through stroke management and can be negative in such cases. Nevertheless, our follow-up could have detected the occurrence of a new neurological deficit, but none of our patients' evolution suggested the presence of VZV vasculopathy.
Our study allowed us to examine therapeutic practices in French hospitals, especially the therapeutic management of VZV aseptic meningitis. Physicians tend to treat these patients and prefer the IV antiviral therapy at high dose for at least 1 week. In the absence of guidelines on this matter in the medical literature, physicians seem to follow the management guidelines for VZV encephalitis instead.
Considering the design of the study, we did not test the association between therapeutic regimen and outcome. Given the favorable outcome of VZV meningitis, a prolonged high dose IV treatment may not be necessary for young patients with no significant medical history and rapid improvement. Indeed, IV antiviral therapy induces longer hospital stays, greater risk of line-related infection, and can lead to acute renal insufficiency. In our study, only two patients did not receive any antiviral therapy. They presented with typical aseptic meningitis and were discharged after symptomatic treatment. At follow-up, they reported a full recovery. Although definite conclusions cannot be drawn from those isolated cases, this emphasizes the need for a prospective randomized trial in order to provide guidelines for the treatment of VZV meningitis specifically. Patients' selection could be guided by the risk factors of poorer prognosis that we identified. Patients with older age or higher CSF protein level should not be included because they might need more aggressive treatment. Given the relative rarity of VZV meningitis, another interesting project might be to open a prospective registry that would capture VZV neurological complications and management.

CONCLUSION
In this retrospective study, neurological complications from VZV were diverse with heterogeneous outcomes: aseptic meningitis without other neurological complications had a good prognosis whereas CNS involvements presented a higher risk of mortality and PNS involved a higher risk of functional sequelae. Age, higher prior-to-infection Rankin score, and higher CSF protein level were risk factors of poorer prognosis. While these data need to be confirmed by prospective studies, they provide key information to guide future registries or controlled trials.