Introduction

Encephalitis affects 1 in 10,000 individuals annually and results in substantial morbidity and mortality worldwide. Many challenges confront clinicians in the management of patients with encephalitis, including lack of consensus case definitions and standardized diagnostic approaches [1]. Numerous large studies have found that a specific cause of encephalitis is typically identified in fewer than 50 % of cases [2–4], and progress in the field has been slowed by the rarity and heterogeneity of cases. Moreover, the field of encephalitis is in a state of considerable flux. A number of infectious etiologies have either emerged in new geographic areas or reemerged over the past decade. In addition, there has been increased recognition of autoimmune causes of encephalitis, with identification of novel autoantibodies and syndromes [5, 6].

Here, recent advances in the field of encephalitis are described. Four areas of focus will be discussed: (1) the resurgence of West Nile virus (WNV) as a major contributor to encephalitis in the United States in 2012, highlighting the difficulties of predicting outbreaks of viral disease and developing effective vaccines; (2) the increasing recognition of human herpesvirus-associated encephalitis occurring in the setting of immunosuppression; (3) the discovery of a potential link between viral infections and anti-NMDA receptor (anti-NMDAR) encephalitis, an autoimmune cause of encephalitis; and (4) the identification of prognostic factors in encephalitis, which may assist clinicians and families in the management of affected individuals.

Resurgence of West Nile Virus

WNV is the most common cause of acquired arboviral disease in the United States (U.S.). Notably, the demographics of WNV infection have changed dramatically over the past decade. The first identified case of WNV infection in the U.S. occurred during an outbreak of encephalitis in 1999 in New York [7]. Within several years, the virus had spread rapidly across the country, with large regional outbreaks in the central states in 2002 and the mountain states in 2003, resulting in nearly 3,000 cases of neuroinvasive disease each year. From 2004 to 2011, small outbreaks occurred each summer, but incidence generally decreased, with the fewest cases of WNV neuroinvasive disease—under 500 each year—reported in 2009 and 2011 [8•]. In 2012, however, a large multistate outbreak occurred, with more cases reported than in any year since 2003. Indeed, over 5,000 WNV cases and over 2,500 neuroinvasive cases were reported in 48 states, the District of Columbia, and Puerto Rico [9]. Most cases occurred from July to September, and incidence rates of WNV neuroinvasive disease were highest in South Dakota, North Dakota, Mississippi, Louisiana, and Texas. In Texas, although attack rates of WNV were higher in rural counties, the highest numbers of cases were reported from metropolitan cities, suggesting the need for broad surveillance [10]. Notably, WNV outbreaks are not confined to the U.S., since the virus has emerged in several areas and reemerged in others in recent years in central and eastern Europe and in countries surrounding the Mediterranean Sea [11].

Reasons for WNV outbreaks are not well understood, although a firm understanding of the ecology of WNV is likely to be crucial. WNV amplification depends on several factors, including numbers and distributions of reservoir animals (predominantly birds) and vectors (mosquitoes), as well as rate of viral replication in mosquitoes, all of which can be influenced by changes in weather patterns [12]. Notably, the average temperature in the summer of 2012 in the U.S. was the hottest on record for similar periods over the past century, suggesting a potential role for elevated temperatures in the potentiation of WNV. However, WNV outbreaks are not consistently related to heat waves, and many virus-endemic areas that experience heat waves do not have outbreaks [8•]. Overall, a greater understanding of ecological factors underpinning WNV transmission would allow for more effective predictive models of disease outbreaks.

The unpredictability of the epidemiology of WNV has also hampered attempts at further development of human vaccines, and there are currently no FDA-approved human vaccines against WNV. Four equine vaccines for WNV have been licensed, and several phase 1 and phase 2 trials have been conducted in humans [13]. A phase 3 trial with enough power to demonstrate efficacy would necessitate recruitment of a large numbers of patients at considerable, and possibly prohibitive, cost [8•]. Given the low incidence of WNV neuroinvasive disease in otherwise healthy individuals and the unpredictable nature of outbreaks, it is likely that future efforts at vaccine development will need to focus on high-risk groups in order to demonstrate efficacy [13].

Viral Encephalitis in the Immunocompromised Host

Infection by pathogens that cause encephalitis may result in a markedly different presentation and clinical course in immunocompromised, as compared with immunocompetent, individuals. Moreover, immunocompromised individuals are at risk for disease from unique, opportunistic pathogens. Below, we discuss recent findings on encephalitis caused by selected herpesviral infections in the immunocompromised host.

Prior case reports and small series have suggested that herpes simplex infections of the central nervous system (CNS) may result in an atypical presentation or more aggressive clinical course in the immunocompromised population [14–16]. In order to better characterize the clinical and neuroimaging features of herpes simplex encephalitis (HSE) in immunocompromised individuals, we performed a retrospective case–control review of patients diagnosed with HSE. We found that immunocompromised patients with HSE presented with fewer prodromal symptoms and with imaging evidence of more diffuse brain involvement extending beyond the temporal lobes and cingulate cortex [17•]. In addition, the absence of cerebrospinal fluid (CSF) pleocytosis was not uncommon. These findings likely reflect the inability of the host’s immune system to mount an adequate inflammatory reaction and to control the infection, and it is anticipated that this will translate into a worse prognosis. Indeed, we found that lack of CSF leukocytosis was a poor prognostic factor, and overall the odds ratio of death was 6 times higher in immunocompromised patients [17•]. These findings are particularly notable in the current setting of increased utilization of potent immunomodulatory and immunosuppressive therapies for a broad range of systemic and CNS autoimmune disorders [18] and highlight the need for clinicians to adopt a high index of suspicion for HSE in individuals on immunsuppression.

Human herpesvirus 6 (HHV6) comprises two variants, HHV6A and HHV6B, the latter of which has been firmly associated with human disease. HHV6B is the primary cause of the childhood disease roseola infantum and has also been associated with febrile seizures [19, 20]. Viral reactivation appears to occur preferentially in immunocompromised individuals, and there is a well-documented association with posttransplant acute limbic encephalitis. In this syndrome, anterograde amnesia and seizures typically begin several weeks after transplantation. MRI demonstrates bilateral nonenhancing T2 hyperintensities in the medial temporal lobes; there may be a mild CSF pleocytosis, and HHV-6 DNA can be amplified from the CSF of most patients [21, 22]. Guidelines recommend use of ganciclovir or foscarnet sodium either singly or in combination in these individuals [23].

Several studies have also noted that HHV6 may account for a substantial percentage of cases of focal encephalitis in immunocompetent adults and children [24–29]. In these cases, it remains unclear whether HHV-6 is causative or simply a “bystander,” since HHV6 reactivation frequently occurs during acute infections with other viruses and also in the setting of other neurologic conditions, including multiple sclerosis and temporal lobe epilepsy [30]. Indeed, HHV6 DNA has also been found in normal brain tissue and CSF, raising questions about the specificity of this finding [31]. Further confounding the issue of detection of HHV6 DNA in human samples is the recognition that HHV6 can stably integrate into the chromosomes of human cells. Thus, a positive PCR may either indicate actively replicating virus or chromosomal integration. Newer assays to distinguish chromosomal integration from reactivated virus may aid in the determination of whether a finding of HHV6 DNA is likely to be pathogenic [32, 33••].

Links Between Infections and Anti-NMDAR Encephalitis?

Infectious causes of autoimmune conditions have been proposed for over 100 years. A number of potential mechanisms, including molecular mimicry whereby a microbial antigen shares epitopes or structural similarities with self-antigens, the unmasking of cryptic epitopes in the setting of tissue damage, or breaking of immune tolerance, may trigger autoimmune disease in the setting of infection [34, 35]. Several well-established relationships, such as the association between Group-A-beta-hemolytic streptococcal infections and Sydenham’s chorea and that of Campylobacter jejuni infection and Guillain-Barre syndrome [36, 37], have strengthened the view that infections can contribute to autoimmune disease.

Recent evidence points to a potential relationship between infections and anti-NMDAR encephalitis. Increasingly recognized as an important cause of encephalitis worldwide, anti-NMDAR encephalitis is characterized by psychiatric changes, seizures, abnormal movements, and autonomic dysfunction and results from an antibody-mediated immune response against extracellular epitopes of the NMDA receptor [38]. In one study, the frequency of anti-NMDAR encephalitis in young individuals surpassed those of individual viral etiologies, including enterovirus and HSE [39]. In up to half of cases, an ovarian teratoma is present, and the syndrome is considered to develop as a paraneoplastic manifestation. In the remaining cases, however the etiology of anti-NMDAR encephalitis is unclear. Notably, many patients have prodromal symptoms that include headache, fever, nausea, vomiting, diarrhea, or upper respiratory symptoms, raising the possibility that the autoimmune process may be triggered by an infection [38]. Indeed, isolated cases of patients with positive mycoplasma or EBV serology or confirmed influenza infection developing anti-NMDAR encephalitis have been reported [40–42].

Recent evidence has linked the development of antibodies to the NMDAR to preceding or concurrent HSE. In a study of 44 patients with PCR-proven HSE, serum and CSF were analyzed for a large panel of autoantibodies. Surprisingly, NMDAR antibodies were detected in 30 % of patients (13 of 44) and comprised IgA, IgG, and IgM subclasses [43••]. Importantly, all three subclasses of NMDAR antibodies have been demonstrated to have pathogenic effects on neurons [44–46]. In some patients, NMDAR titers were undetectable at onset but increased over subsequent weeks, suggesting the initiation of a new B-cell response against the NMDAR in the setting of HSE. Although these findings suggest the possibility that HSE may trigger NMDAR encephalitis, in the relatively short follow-up time, none of the antibody-positive patients developed a syndrome compatible with NMDAR encephalitis. Thus, an alternate hypothesis is that neuronal damage in the setting of HSE results in the development of NMDAR autoantibodies as an epiphenomenon, having no relationship to the well-recognized clinical syndrome of anti-NMDAR encephalitis. At present, the data linking HSE and anti-NMDAR encephalitis are not conclusive, and this remains an area of intense investigation.

Predictors of Outcome in Encephalitis

In large studies, mortality rates in encephalitis range from 5 % to 10 %, with significant morbidity among survivors. Little, however, is known of prognostic indicators of outcomes in encephalitis. Some attempts have been made to address prognostic factors in subgroups of patients with encephalitis. In HSE, for example, it is well documented that a delay of greater than 2 days between hospital admission and initiation of acyclovir therapy is associated with worse prognosis, as are older age and poor level of consciousness at the time of initiation of therapy [47, 48]. In a prospective study of patients with acute infectious encephalitis, several factors, including older age, immunosuppression, and mechanical ventilation, were associated with death during hospitalization [49]. In patients with encephalitis involving the brainstem, elevated CSF protein and glucose were associated with a poorer outcome. Importantly, elevated CSF glucose was tightly associated with elevated serum glucose, suggesting that poor systemic glycemic control may drive increased CSF glucose in such patients, potentially providing a rationale for aggressive glucose control in affected patients [50].

More recently, prognostic factors in anti-NMDAR encephalitis were addressed in an observational study that included over 500 patients. Almost half had no improvement in the first month after initiation of immunotherapy or tumor removal, but improvements from severe to slight disability occurred within the first 24 months of treatment in 81 % of patients [51•]. Predictors of good outcome included early treatment with steroids and/or immunosuppression and lack of admission to an intensive care unit (ICU).

We recently examined outcomes in patients with acute encephalitis admitted to two hospitals affiliated with the Johns Hopkins Medical Institutions. This study differs from others in that all causes of encephalitis were included; moreover, only those patients who were admitted to the ICU during their hospital stay were studied. Among almost 500 patients identified with acute encephalitis, 103 had ICU stays of greater than 48 h. Of these patients, 28 had viral encephalitis, 10 had bacterial or fungal encephalitis, 17 had autoimmune causes of encephalitis, and the remaining 48 patients had encephalitis of unknown cause [52]. Multivariate regression analysis demonstrated that of the survivors, patients with autoimmune etiology were more likely to have a poorer outcome on hospital discharge, as assessed by modified Rankin Scale, than those with other etiologies. Our results are consistent with that seen in anti-NMDAR encephalitis, in which almost half of patients had no improvement in the first month after initiation of immunotherapy or tumor removal [51•]. Thus, both our study and that of Dalmau and colleagues suggest that patients with autoimmune encephalitis may experience substantial delays before any meaningful functional recovery. We also addressed predictors of in-hospital mortality and found an association with several conditions, including cerebral edema (odds ratio [OR] 18.6), status epilepticus (OR 8.16), and thrombocytopenia (OR 6.28), all of which are potentially reversible and modifiable. Importantly, these findings suggest that aggressive supportive therapies utilized in the critical care setting may improve outcomes in those with acute encephalitis.

Recently, we have begun examining predictors of outcome in encephalitis using hospital administrative data from the United States Agency for Healthcare Research and Quality (AHRQ) Healthcare Costs and Utilization Project (HCUP) Nationwide Inpatient Sample (NIS). We have found that complications including intubation, acute respiratory failure, aspiration pneumonia, and sepsis are strong predictors of in-hospital mortality (Venkatesan & George, unpublished), consistent with a study in France of patients with infectious encephalitis [49]. Moreover, extremes of age and comorbidities, including HIV/AIDS, cancer, and other medical diseases, were also associated with increased mortality, in agreement with other studies [17•, 47, 49, 53–55]. In addition to identifying and substantiating a number of predictors in encephalitis, these findings lend support to the further exploration of large administrative data sets to define hospitalization rates, trends in etiologies, and outcomes in encephalitis.

Conclusions

Encephalitis continues to impose a significant burden on patients and society. A number of recent studies have furthered our understanding of demographics, clinical spectra, etiologies, and prognosis. However, novel approaches are clearly needed in order to advance diagnosis and treatment of encephalitis. New tools to uncover etiologies, including pathogen discovery efforts and the application of sophisticated methodologies to identify humoral and cell-mediated autoimmune disorders, will need to be brought to bear upon this important syndrome. Further therapeutic approaches will also need to be explored, with focus not only on the etiologic agent, but also on accompanying clinical manifestations that may be modifiable. Given the heterogeneity and unpredictability of disease, it is likely that future advances in encephalitis will involve collaborative, multicenter groups.