Studies were included if they were case-series including more than 10 patients, case–control studies, cohort studies or randomized-controlled trials. We included publications that reported on cases with features of encephalitis or meningoencephalitis that were suspected or confirmed to have been caused by HSV or VZV. Diagnosis had to be confirmed by detection of HSV or VZV in the CSF by polymerase chain reaction (PCR) (5, 19). Patients without PCR or serological microbiological confirmation had to be distinguishable in the final analysis. We excluded studies focusing on other causes of infectious encephalitis or meningoencephalitis, autoimmune mediated encephalitis or non-CNS syndromes, such as isolated myelitis or radiculitis. Inclusion criteria were publication in German or English language and the availability of the full text. Studies performed solely in children were also excluded as childhood cases are likely to represent clinical entities that are distinct from adult cases.

A detailed review of each study was conducted by two independent researchers (LA and AD), during which the following details were extracted: number of patients, cause of encephalitis or meningoencephalitis, clinical syndrome, age, abnormal investigation findings including MRI, outcome measures, factors tested for correlation with outcome and study results. We had a particular interest in publications in which MRI findings were used as markers for prognosis. The CSF parameters we considered applicable were those identified by routine testing, such as protein, glucose, white cell count, or differential cell counts. We recorded MRI abnormalities likely due to encephalitis or meningoencephalitis, or any MRI abnormality if these details were not specified.

The studies reviewed are summarised in Table 1 and included 32 retrospective and 8 prospective studies from Europe (France, Denmark, Sweden, Spain, Turkey, Czech Republic, Austria), the United States, Israel, Asia (India, Republic of Korea, Japan, China) and New Zealand. Two multinational studies included data from Arabic countries (Egypt, Iraq, Lebanon) (37, 38). In most studies, the proportion of HSV1 to HSV2 infections was evident. The outcome was generally assessed at discharge and during follow-up periods of 3 to 6 months and/or after 1 year. Eight studies had further follow-up periods of up to 2 years (7, 11, 27, 34, 35, 43, 50) or even up to 11 years (20, 21, 50).

Mortality rates of HSV encephalitis reported in these studies were mostly between 5 and 20% (7, 11, 20, 22–24, 28, 30–38, 42, 46, 47, 49, 50, 52). Tan et al. described significantly increased mortality rates in immunocompromised compared to immunocompetent patients (36 versus 7%) (30). However, three studies reported no fatal cases (40, 53, 54), whereas five other studies found mortality rates of 24–64% (8, 27, 39, 41, 51). It is noteworthy that the studies reporting no mortality are most likely to have included more of the less severely affected patients. Kaewpoowat et al. (54) included 75% patients characterised as having meningitis and Kim et al. (40) reported a mean initial Glasgow Coma Scale (GCS) of 13.2 with altered mental status – a defining criterion for encephalitis – in only 25% of patients. Růžek et al. (53) did not provide further details on the clinical presentation of the study population and only divided the study population retrospectively into two groups: “mild” following successful uncomplicated therapy with good outcome or “severe” describing a severe course accompanied by acute neurological signs. On the other hand, large studies with mortality rates between 10 and 17% (47–49) included admission to the ICU as an inclusion criterion for study participation.

Outcome was reported to be favourable in 29–65% of survivors (8, 20, 22–24, 27, 29, 31–33, 37, 38, 40, 48, 49, 52, 53) and complete recovery was observed in 14–43% (7, 11, 22–24, 32, 33, 40, 51, 52). Interestingly, one prospective treatment study investigated an additional 3-month course of valaciclovir after standard aciclovir treatment. The authors described no or only mild residual neurocognitive deficits after 12 and 24 months in 86 and 90% of patients in the treatment and control group, respectively (34). A nationwide registry cohort study from Denmark noted a significantly increased risk of mortality (19% 1-year absolute excess mortality) in the first year and an increased risk of dementia in the first 5 years after detection of HSV in the CSF (50).

The factors most frequently associated with mortality and morbidity were age (8, 20, 22, 24, 28, 31, 37, 41, 42), pre-existing morbidity (7, 11, 42, 54), fever on admission (31, 47) and duration of fever after start of treatment (28), as well as lower GCS or a higher acute physiology and chronic health evaluation (APACHE) score on admission (8, 22, 24, 41). The following clinical parameters were also found to be associated with a worse outcome: longer interval between onset of main symptoms and hospitalisation (32), pre-existing immunocompromised state (30, 74) and status epilepticus, persistence of impaired consciousness, confusion or aphasia at day 5 of evolution (52), admission-to-MRI delay (52), need for mechanical ventilation (47) and length of stay in the ICU (52). Interestingly, direct admission to the ICU seems to be protective (47).

Development of autoimmune encephalitis (AE) within 3 months after HSV encephalitis has been described in up to 27% of cases (74% N-methyl-D-aspartate (NMDAR), 26% unknown antigens) (36, 43). Risk factors were younger age (≤4 years) and shorter interval between HSV encephalitis and detection of AE antibodies (43). Early detection of anti-NMDAR antibodies was associated with an overall increase in inflammatory CSF response and worse outcome (35, 36).

Regarding laboratory parameters, lower CSF cell count and initially negative HSV PCR were found to be associated with worse outcome in immunocompromised patients (30). Interestingly, another study noted that of 273 HSV encephalitis patients, 11 had negative HSV PCR in the first lumbar puncture performed 1 day after symptom onset (49). An initial negative HSV PCR was associated with worse outcome. In-hospital mortality was 27% and modified Rankin Scale (mRS) ≥4 at 3 months in 73% of PCR-negative patients compared to 14 and 33%, respectively in PCR-positive HSV encephalitis patients. This difference was only partially explained by delayed start of aciclovir treatment (49). Controversially, Mulatero et al. reported that 13% (11/76) of patients had negative HSV PCR in the first lumbar puncture performed – mean 1.8 ± 3 days (range 0–17 days) – after admission (52). These patients had less severe disease, but no difference was seen in the outcome (52). Overall, studies reported between 4 and 13% initial false negative HSV PCR results (41, 47, 49, 52).

Levels of neurofilament (NFL) in CSF (36) and serum albumin levels on admission (28) have been associated with outcome. No direct correlation between viral load in CSF and clinical outcome has been found (8, 23, 27, 53). Xanthochromia (haem degradation products in the CSF leading to a yellowish appearance) is a rare condition in HSV encephalitis (38) and is associated with poor outcome (8). Most likely, xanthochromia reflects advanced brain infection with tissue necrosis (34).

Several studies mention imaging findings (33), of these, nine studies described them in more detail or focused on MRI findings and their prognostic value (32, 39–41, 44, 47, 48, 52), as summarised in Table 2. Extent of brain involvement seen on MRI at admission, especially bilateral temporal lobe involvement, has been described as a factor associated with poor prognosis (32). The study by Sili et al. (32) has limitations due to missing information on time from hospital admission to MRI, lack of detailed description of abnormal MRI sequences and the high proportion of “suspected” HSV encephalitis (48% of the study population). However, it has been confirmed that fluid-attenuated inversion recovery (FLAIR) MRI signal abnormalities affecting more than three brain lobes, as well as the presence of diffusion-weighted MRI signal abnormalities in the left thalamus, were independently associated with poor outcome (47, 48). The multicentre studies by Jaquet et al. (47) and Sarton et al. (48), analysed large cohorts of patients with a well-described study population. Both these studies (47, 48) analysed the same cohort of HSV encephalitis patients requiring ICU treatment; however, Sarton et al. (48) included fewer patients in their analysis and focused exclusively on MRI and functional outcomes after HSV encephalitis. In these two studies, the MRI acquisition took place a median of 3 days after hospital admission and 1 day after ICU admission and it was abnormal in 99.3% of patients, with FLAIR hyperintensities as the most important finding (48). Singh et al. also showed that restricted diffusion on MRI was associated with poor outcome (mRS ≥3) in elderly people (median age 66 years) at hospital discharge as well as 6–12 months later (41). In contrast, other studies showed no association between MRI findings and outcome (40, 55). The studies by Kim et al. (40) and Jordan et al. (55) both included a retrospective analysis, of 25 and 15 patients, respectively. In the study by Kim et al. (40) there was a surprisingly low number of abnormal MRI (64% FLAIR and 59% diffusion-weighted imaging (DWI)) compared to larger studies (32, 48). However, the authors did not explicitly report the proportion of normal and abnormal MRI in HSV patients and the value of the study is therefore limited.

Delay of aciclovir initiation significantly worsens clinical outcome (20–22, 30, 32, 37, 41). However, a dosage of aciclovir that is higher than the recommended standard dose of 10 mg per kg body weight every 8 h together with an additional course of oral valaciclovir therapy after aciclovir treatment did not improve outcome (29, 34). Very recently Mulatero et al. described an association between worse outcome, body weight and aciclovir dosage and suggested a weight-adjusted dose regimen, increasing the dose for patients with lower body weight (of <79 kg) up to 15 mg/kg body weight, especially for patients with a body weight below 57 kg (52). The question whether additional treatment with corticosteroids is beneficial for long-term outcome has yet to be answered (24, 45, 75). In one prospective randomized, double-blind, placebo-controlled treatment trial that had to be stopped prematurely due to slow recruitment, adjunctive steroid treatment did not affect mortality or neurological sequelae (45).

As mentioned above, VZV causes a wide range of clinical manifestations of infection of the nervous system. Most frequently VZV infection or reactivation affects the PNS causing ganglionitis and dermatomal rash or facial nerve palsy (64). Less frequently, patients present with encephalitis, meningitis, cerebellitis, myelitis, or stroke/vasculopathy (59, 60, 64, 76, 77). Therefore, VZV infection can usually be discriminated clinically from HSV infection by the typical rash – if present – and the clinical presentation. However, no clinical sign or symptom can discriminate clinically between VZV and HSV encephalitis in the very early phase.

In this review, we focus on studies investigating outcome and prognosis after an encephalitic or meningitic course of VZV infection. Studies exclusively investigating outcome after PNS infection, herpes zoster or after vasculopathy are beyond the scope of this review. We included 22 studies (16 retrospective, 6 prospective) from 12 countries worldwide into the review, as summarised in Table 1. The outcome was generally assessed at discharge, some studies had follow-up periods of 1 to 6 months and/or after 1–3 years.

Encephalitis and meningitis due to VZV infection has a mortality rate of 0–15%, with fatal cases more likely during an encephalitic disease course (11, 34, 43, 53, 54, 57, 59, 62–64, 67–70). Only two studies reported mortality rates as high as 33% (72) and 36% (70). The first of these included only patients with severe encephalitis requiring intensive care (median GCS at admission 12, and mechanical ventilation in 84%) (72). In the second study, all the patients who died had a meningoencephalitic course plus stroke and/or myelitis (70).

A precise estimation of clinical outcome of survivors is difficult because of the varying definitions of outcome between the studies, different time points of evaluation (from discharge to follow-up after 3 years). Often outcome is reported combining various clinical manifestations of VZV infection of the CNS, sometimes even including PNS infection. However, the largest prospective study of VZV encephalitis, which included 92 patients, reported full recovery in 49% of patients after 3 months (68). Another study, prospectively investigating various infectious causes and outcomes of encephalitis described complete recovery in 41% of VZV encephalitis patients after 3 years (59). Interestingly, in another publication from the same group, which investigated the overall long-term outcome in patients from the same cohort study on infectious encephalitis, only 33% of VZV patients were found to have made a complete recovery after 3 years (7). A third study, with a retrospective design, worth mentioning here, showed a favourable 1-year outcome (i.e., mRS 0–2) in 36% of the whole study population and in 48% (20/41, excluding patients who died) of ICU survivors (72).

Most studies on VZV meningitis observe a good overall outcome in 70–100% of patients (54, 56, 60, 62, 63, 65, 66, 70, 71), although persisting neurological sequelae in 0 (60, 66) up to 82% (70) of patients have been described in some studies. In a small case–control study on 14 patients with VZV CNS infection (4 with meningitis, 6 with encephalitis and 4 with radiculitis or polyneuropathy) mild cognitive deficits were seen more frequently in a follow-up examination after 3–4 years than in a control group (60).

Prognostic factors for a severe acute disease course are controversial: whereas three studies found that higher VZV DNA load in the CSF was associated with disease severity (57, 58, 63), this was not confirmed in another study (53). In the acute phase, skin rash has been reported in 43–91% of patients (54, 56–58, 63, 65, 67, 68, 70). Only a few studies report herpes zoster in less than 60% of patients (56, 62, 70), occurring in 30–70% before (57, 58, 68), at or after onset of neurological signs and symptoms (57, 58). A shorter interval between appearance of herpes zoster and onset of neurological signs and symptoms has been described as a negative prognostic factor for death or sequelae (65). Older age (11, 67, 68, 70, 72) and pre-existing comorbidities (11, 54, 69, 70), as well as an encephalitic course of disease (54, 56, 68), need for mechanical ventilation (72) or signs of vasculitis (68) were associated with a worse outcome.

A large Danish cohort study analysing data from the national health registry showed that immunosuppressive state and comorbidities (Charlson Comorbidity Index >1) were a risk factor for detecting VZV DNA in the CSF (69). Mortality was increased in this VZV cohort, especially in the first year of observation and in patients with immunosuppressive or comorbid conditions (69). An increased risk of dementia and epilepsy, but not psychiatric disease, was found in the same cohort during the observation period of 12 years (69). Immunosuppression was a risk factor for more severe disease, but was not associated with worse outcome as found in three other studies (65, 70, 72).

MRI findings in patients with neurological VZV infections have been mixed, with pathological findings in 5% up to 70% of meningitis and encephalitis patients during the acute phase (54, 56, 61, 65, 67, 68, 70–72). To our knowledge, MRI findings have not so far been evaluated for their potential to serve as prognostic parameters, most likely due to incomplete data sets and mainly nonspecific MRI findings.

Interestingly, contrary to one prospective study (68) and two retrospective studies (67, 72), Yan et al. recently identified delayed time to aciclovir treatment as an independent risk factor for worse outcome (71). In the study by Le Bot et al., a higher dose of intravenous aciclovir (15 mg/kg every 8 h) was not found to be protective (67).