Autoimmune encephalitis is a difficult clinical diagnosis due to the similarities in the clinical, imaging and laboratory findings of many forms of autoimmune and infectious encephalitis. Patients generally have impaired memory and cognition over a period of days or weeks. There may be clues to specific causes on history of physical examination, but often these specific signs are absent. A broad approach to testing for infectious diseases and various neuronal autoantibodies can lead to the correct diagnosis. If a clear autoimmune cause for the symptoms is established, treatment usually involves escalating immune therapies. The process of caring for these patients requires patience and repeated evaluations to determine the proper degree of immune therapy needed at any given time.

Autoimmune encephalitis involves several types of diseases with different pathophysiology. Understanding the pathophysiology of these diseases is helpful in using diagnostic testing and choosing appropriate therapies. The first group includes the classic paraneoplastic disorders associated with antibodies to intracellular antigens, such as anti-Hu. These disorders are strongly cancer associated and involve T-cell responses targeting neurons. The prognosis tends to be poor due to irreversible neuronal killing by these mechanisms, the severity of associated cancers, and the difficulty in controlling these sorts of immune responsess. The antibodies in these disorders are useful tumor markers, and in the appropriate context and titer also useful markers of the paraneoplastic neurological disorders. The antibodies themselves are not directly pathogenic. The second group involves autoantibodies to extracellular epitopes of ion channels, receptors and other associated proteins, such as the NMDA receptor. The cancer associations are variable, and the prognosis tends to be much better. The antibodies in these disorders are thought to be directly pathogenic, causing reversible effects on synaptic function in neurons with relatively little neuronal death. There are also important tumor associations in this group of diseases. For instance, patients with anti-NMDAR encephalitis commonly can recover from a totally unresponsive state to eventually resume a good quality of life. Occupying an intermediate position are diseases with autoantibodies to intracellular synaptic proteins such as GAD65. It is unclear whether this group involves T-cell responses and/or functional effects of antibodies. A final group includes other forms of autoimmune encephalitis in which precise antigens are less clearly established, such as lupus cerebritis or ADEM. Some diseases in this group have systemic manifestations outside the nervous system. This review will focus on the disorders with well-defined brain antibodies.

Autoimmune encephalitis can manifest with several distinct syndromes, complicating its recognition. The classical presentation of encephalitis consists of a subacute (days to a few weeks) progressive decrease in the level of consciousness, often with fluctuations, and altered cognition. Memory, especially retention of new information, may be impaired early in the clinical course. Patients may progress to coma. While many cases of autoimmune encephalitis are indistinguishable from each other or viral encephalitis, there may be clues to specific autoimmune etiologies (Table 1).

Psychiatric manifestations are common early in the course of autoimmune encephalitis. These may include psychosis, aggression, inappropriate sexual behaviors, panic attacks, compulsive behaviors, euphoria or fear. Symptoms may fluctuate rapidly. Although this presentation is well known for anti-NMDAR encephalitis,1 anti-AMPAR and anti-GABA-B-R both may have prominent early psychiatric manifestations2 (Overall, anti-NMDAR encephalitis is more common and should be suspected first, especially in young adults and children, but they could each cause this presentation across a wide range of ages).

Abnormal movements may be the presenting symptom in several types of autoimmune encephalitis. These include anti-NMDAR encephalitis, where movement symptoms may occur early in the disease course, especially in children, who generally have more motor symptoms and fewer psychiatric symptoms than adults.3 These may resemble dystonia or chorea, with writhing and fixed abnormal postures of the limbs. In adults with anti-NMDAR encephalitis, writhing movements of the face and limbs may be most prominent in the comatose phases of the illness. GAD65 and GlyR autoimmunity may present with stiff person syndrome (SPS) or progressive encephalomyelitis with rigidity and myoclonus (PERM).4 A striking feature of PERM with GlyR antibodies is a pathologically exaggerated startle response, resembling hereditary hyperekplexia, a genetic disease caused by GlyR mutations.5 Although there is some degree of overlap, GAD65 is more associated with classical SPS while GlyR antibodies may be seen more with symptoms of hyperekplexia and myoclonus, which are prominent in PERM. Stiffness or exaggerated startle combined with other symptoms of encephalitis should raise concern for GlyR antibodies. Basal ganglia encephalitis has also been reported with D2R antibodies, although this may be very rare.6 Sydenham chorea is a well-recognized autoimmune movement disorder thought to be triggered by streptococcal infections and should be considered in children with this presentation.7

Seizures are common in autoimmune encephalitis and may be a presenting symptom. In anti-NMDAR encephalitis seizures may occur at any stage of the illness. Autoantibodies to two important inhibitory receptors in the brain, GABA-B and GABA-A receptors (at high titer) convey a high risk of severe seizures and intractable status epilepticus.89 GAD65 antibodies may present with epilepsy, perhaps also with memory impairment, but with few other symptoms to suggest an autoimmune etiology. GAD65 autoimmunity may therefore resemble other forms of treatmentresistant epilepsy. Fasciobrachial dystonic seizures (FBDS) are brief seizures consisting of rapid jerks of the face and/or ipsilateral arm and shoulder.10 Seizures may be partial or associated with temporary disruptions in consciousness and may be multifocal and variable on EEG. FBDS are characteristic of LGI1 autoimmunity and may precede other symptoms of the disease by weeks or months. Patients may have hundreds of these seizures per day. These seizures may have only limited response to seizure medications but respond well to immune therapies.

Cerebellitis is a distinct syndrome of ataxia of gait, limb movements, eye movements, voice, and/or swallowing. The precise mixture of symptoms varies from patient to patient. Vertigo and nystagmus are common. Cerebellitis may occur with infectious causes, but the presentation of a subacute cerebellar syndrome portends a good probability a specific autoimmune etiology and also a significant risk of tumors. Paraneoplastic cerebellar degeneration is associated with conventional onconeuronal autoantibodies such as Yo, but also with cell surface autoantibodies targeting mGluR1, DNER, and other antibodies. GAD65 antibodies are perhaps the most common finding in this phenotype in my experience. Autoimmune cerebellitis may result in the irreversible loss of Purkinje neurons, and the prognosis is recovery may be poorer than with other types of autoimmune encephalitis.

Certain types of autoimmune encephalitis may precede or follow neuromuscular manifestations, particularly acquired neuromyotonia (Isaacs syndrome). Isaacs syndrome presents with muscle spasms, cramps and fasciculations due to peripheral nerve hyper-excitability.1112 Morvan syndrome (Morvan's fibrilllary chorea) consists of peripheral nerve hyper-excitability with encephalitis and severe insomnia. Some cases of Isaacs syndrome are associated with autoantibodies to Caspr2 or other, often undefined, members of the voltage-gated potassium channel (VGKC) complex.1314 Caspr2 antibodies are even more likely in patients with Morvan syndrome, especially patients with thymoma, who may have multiple autoimmune disorders during their disease course. Encephalitis with these neuromuscular manifestations (myasthenia gravis, neuromyotonia) therefore suggests a specific autoimmune etiology.

As discussed below, most of the autoimmune causes of encephalitis are paraneoplastic, each conveying a risk profile for various tumors. Detecting particular tumors may therefore also suggest particular autoimmune causes. For instance, the likelihood of anti-NMDAR encephalitis is increased in a young woman with ovarian teratoma, and the likelihood of anti-DNER is higher in patients with cerebellar degeneration and Hodgkin lymphoma.

In addition to the antibody-mediated and paraneoplastic forms of encephalitis, there are other autoimmune diseases that may present with encephalitis. In the case of ADEM, encephalitis is a common presentation. The characteristic brain lesions, and sometimes involvement of the optic nerves or spinal cord, are an important clue to diagnosis.

Multiple sclerosis (MS) is generally easier to distinguish from autoimmune encephalitis due to more focal symptoms and characteristic brain imaging findings.

Lupus may affect diverse areas of the nervous system, causing neuropathy, vasculitis, myelitis, venous sinus thrombosis, stroke, and other manifestations.15 Neuropsychiatric lupus may manifest with seizures, psychosis, or neurovascular disease. These manifestations are most common with severe disease affected other organ systems such as the gastrointestinal, renal and hematological systems. In one large series, one forth of deaths from lupus were related to CNS involvement and 16% were due to CNS infection, suggesting vigilance for both autoimmune and infectious encephalitis is warranted for these patients.16

Vasculitis affecting the CNS may rarely present with symptoms resembling encephalitis. When this is suspected, imaging of the cerebral vessels, search for other evidence of vasculitis (such as serologies for lupus and other rheumatologic diseases) may be useful.

Bickerstaff encephalitis and Miller Fisher syndrome enter into the differential diagnosis of autoimmune encephalitis due to the presence of altered mental status and/or cranial neuropathies. These diseases may at first resemble the brain-stem syndrome associated with anti-Ri, but the loss of reflexes is an important clue suggesting Miller Fisher syndrome. Detection of the GQ1b antibody may be helpful in securing these diagnoses.17

It is typical for patients with autoimmune encephalitis to have testing for various infectious etiologies and they are frequently covered with antibiotic and/or antiviral therapies (such as acyclovir) empirically as infectious causes are excluded. Some of the relevant risk factors for various causes of encephalitis are listed in Table 2 and the major infections are listed in Table 3.

Most cases of infectious encephalitis are viral. In the USA, leading viral infections include are HSV, VZV, enterovirus, and West Nile virus (WNV).18 Japanese encephalitis (JE) was once the leading cause of viral encephalitis in East Asia, but has declined dramatically in Korea and other nations due to successful vaccination programs.19 Bacterial causes include listeria, atypical presentations of streptococcus, syphilis, Lyme disease, and tuberculosis. Fungal causes such as Cryptococcus or aspergillis are particularly likely in immunocompromised patients.

A detailed travel and exposure history is essential. For instance, West Nile encephalitis has been reported only once in Korea, in a patient who had recently traveled to Guinea.20 Travel to areas endemic for malaria or Lyme disease may be similarly relevant. Exposure to persons with tuberculosis or other infectious agents may be a clue to diagnosis.

Since the risk of various infections depends on the status of the immune system, it is important to know patients' HIV status and consider whether other medical conditions (transplantation, cancer) may have weakened immunity. Certain types of tumors, such as lymphomas, may weaken the immune system and also increase the risk of paraneoplastic/autoimmune diseases. A detailed review of encephalitis in immunocompromised patients highlighted HSV and VZV as leading causes, along with a host of rarer entities.21 Infections have been transferred by the organ donor to an organ transplant recipient, including rabies, LCMV, Westnile, CMV, and EBV;22 these infections are a consideration in encephalitis in the weeks after transplantation.

Human herpesvirus 6 (HHV-6), and less often HHV-7, may cause encephalitis and their reactivation may be an important cause of encephalitis in transplant patients.232425 HHV-6 and HHV-7 are diagnosed by PCR, although this may not be positive initially.26 There are no approved treatments, but ganciclovir, foscarnet, and cidofovir have been used.27 HHV-6 may be more likely in cases whether the donor and/or recipient of a stem cell transplant carry a particular HLA type (HLA-B^*40:06).28

Due to the diversity of potential infections, using a separate test for each pathogen may not be practical. To address this issue, advanced genetic techniques (e.g., metagenomic deep sequencing of cerebrospinal fluid) have been proposed to screen for thousands of pathogens rapidly and efficiently. Early successes with this approach have included diagnosing cases of encephalitis due to leptospirosis, neurotropic astrovirus infection, Balamuthia mandrillaris, and squirrel bornavirus.29303132 This approach should assume a wider role in diagnosing unusual CNS infections in the coming years.

Cerebellitis may occur as an infectious or post-infectious phenomenon. Leading viral causes include VZV, JC virus, influenza, EBV, and enterovirus. Numerous other bacterial, fungal, malignant, and other causes have also been reported.33

Wernicke encephalitis, due to thiamine deficiency, is most recognized in alcoholics but may also affect patients with gastric bypass or other causes of insufficient nutrition.34 A history of weeks to months of rapid weight loss is common in the later group, who do not have alcohol as a supplemental source of calories (The complex nutritional deficiencies after gastric surgery may present with several distinct syndromes. Of greatest relevance to this review, Wernicke encephalitis may occur along with or following an acute painful lower-extremity-predominant neuropathy.).35 Prompt and thorough repletion with thiamine, often along with other nutrients, may be life saving and should not wait on laboratory confirmation of the diagnosis.

Intoxications such a neuroleptic malignant syndrome and serotonin syndrome may often present with similarities to autoimmune encephalitis. Conversely, patients with anti-NMDAR encephalitis may develop psychosis as an initial symptom and be treated with neuroleptics, then later show catatonia, rigidity, autonomic instability and altered level of consciousness; this pattern of findings may be mistaken for neuroleptic malignant syndrome. Autoimmune encephalitis therefore should enter into the differential diagnosis of any case of suspected neuroleptic malignant syndrome (Patients with anti-NMDAR encephalitis may be particularly sensitive to strong dopamine antagonists, and our group attempts to avoid using these medications).

Lymphoma or carcinomatous meningitis may present similarly to autoimmune encephalitis.36 Particularly, symptoms may have a subacute onset and involve multiple cranial neuropathies. Both may occur in patients with or without known tumors. Repeating CSF cytology, awareness of known tumors, and screening for malignancy may all help lead to recognition of these causes.

Treatment for suspected autoimmune encephalitis is often given empirically prior to specific antibody test results. This may include steroids and/or IVIG. If a cell-surface/synaptic antibody disorder is diagnosed, initial treatments may include IVIG, plasmapheresis, and/or steroids. Steroids may be beneficial in a range of autoimmune disorders but could potentially create problems with the diagnosis of certain disorders such as CNS lymphoma. IVIG offers an important advantage of being unlikely to make infectious encephalitis worse. Plasmapheresis is also unlikely to significantly worsen infectious encephalitis.

If a synaptic/cell-surface antibody is detected and the patient has any significant symptoms, first-line therapy should be given if it has not already been tried. In general, prompt treatment, and escalation of treatment in patients who remain ill, is associated with better outcomes. Although there are not randomized treatment trials, protocols have been proposed for anti-NMDAR encephalitis,1 and these approaches have been applied to other diseases in the cell-surface/synaptic autoantibody category. Our group often uses IV solumedrol (1 gram daily for 3-5 days then a taper over several weeks) and IVIg (0.4 g/kg/day for 5 days). Other groups have advocated plasmapheresis instead of IVIg, and so far there is not convincing evidence of superiority for either approach.

If the patient remains significantly impaired after first-line therapy, second-line treatments are typically used. Some groups might wait 2 weeks or longer to allow first-line therapies time to work, but our group often proceeds more quickly to second line therapy sooner is patients who are very ill, for instance comatose patients with anti-NMDAR encephalitis. Second line therapies include rituximab (often 375 mg/m² weekly for 4 weeks) or cyclophosphamide (750 mg/m² IV monthly until improvement is noted), or both. Rituximab is a monoclonal antibody targeting CD20, so plasmapheresis generally should not be done after it is administered. Rituximab depletes CD19+/CD20+ B-cells, and circulating levels of these cells typically become undetectable for several months after treatment. Due to its relatively favorable safety profile, rituximab is more often used as monotherapy in children. Rituximab is thought to be generally effective against neurological diseases where the autoantibodies are of the IgG4 subtype.81 Since IgG4 responses predominate in LGI1 and Caspr2 encephalitis this provides an additional theoretical support for using rituximab in those diseases. Cyclophosphamide has several important toxicities, including a risk of infertility, especially in young women who received repeated doses (The risk cumulatively increases, potentially up to 40% after 12 doses). This risk be can reduced with use of a GnRH agonist in women,82 or addressed with egg/sperm collection.

Anti-NMDAR encephalitis is by far the most common type of antibody-mediated encephalitis in children. The age distribution of other types of synaptic autoimmune disorders either skews much older (the median age for LGI1 or Caspr2 antibodies is about 60 years) or the disorders are much less common (GABA-A antibodies) or both. In a study by the California encephalitis project, anti-NMDAR encephalitis was more common any single viral etiology.83

Anti-NMDAR encephalitis in children may present differently than in adults.84 Children are more likely to have abnormal movements (chorea, incoordination) early in the disease course and also may have atypical motor symptoms such as ataxia or hemiparesis. Children more often have seizures than adults. The classic symptoms of psychosis seen in adults are less common, but behavioral regression is frequently noted. Patients may have prominent speech difficulties. Treatment strategies are similar in children and adults, but physicians may be more reluctant to use cyclophosphamide, relying more on rituximab as a second line treatment (As with adults, the optimal treatment strategies are not known). Responses to treatment are similar in children and adults, with about half failing first line therapies. Ovarian teratoma is less likely in female children before puberty, so tumors are uncommon in young children.

Patients with encephalitis may recover, completely or partially, and then experience worsening symptoms. In autoimmune encephalitis, relapse tends to follow a similar clinical course to the initial attack. In anti-NMDAR encephalitis, these relapse tend to be milder than the initial attack and manifest with confusion, worsening memory, personality change, hallucinations or new seizures (In my experience, seizures in my cases of autoimmune encephalitis remit with appropriate treatment, and new seizure should always raise concern for relapse). The risk of relapse in anti-NMDAR encephalitis in approximately 12% over two years (but continues beyond that) and is highest in untreated patients, intermediate in patients who had only first-line therapy, and lowest in patients treated with second-line therapies.84 Relapsed patients are usually treated with second-line therapies, possibly after first line therapies. These patients may be treated for longer periods of time with second line therapy, especially rituximab, but the optimal duration of treatment has not been established. In other types of autoimmune encephalitis, the risk of relapse is less clearly established. LGI1 antibodies and Caspr2 antibodies may associate with milder encephalitis, compared with NMDAR antibodies, than is chronic or relapsing. Similar treatment strategies may be used with these antibodies.

An important recent advance has been the recognition that patients with HSV encephalitis may rarely develop anti-NMDAR encephalitis several weeks later as a post-infectious complication.85 In these patients CSF from the initial attack has no NMDAR antibodies with positive PCR for HSV, but CSF from the second attack now has NMDAR antibodies with negative PCR for HSV. This phenomenon may be due to exposure of CNS antigens to the immune system in the presence of a powerful infectious stimulus. Patients who worsen after infectious encephalitis should therefore be carefully evaluated for both infectious and autoimmune etiologies. Similarly, patients who have been treated for autoimmune encephalitis may be immunosuppressed and at risk for diverse infections. However, opportunistic CNS infections after autoimmune encephalitis are probably very rare compared to worsening of the autoimmune disease.

The proper diagnosis and management of autoimmune encephalitis requires an organized approach. Evaluation should begin with a detailed history and physical examination to detect clues to specific causes. A diverse range of infections should be considered, and appropriate testing should be done to exclude relevant pathogens. Ancillary testing with MRI, EEG, and lumbar puncture may further support a diagnosis of encephalitis and potentially suggest particular causes. A broad group of autoantibody tests may be used to diagnose or exclude particular autoimmune causes, but these tests are complex and not every positive result is definite evidence of an autoimmune disorder. The risk of neoplasm should always be considered during initial treatment and follow-up visits, both in terms of diagnosing serious cancers and because specific tumors may suggest particular autoimmune causes. Treatment should depend on the pathophysiology of the disorder (e.g., T-cell-mediated or antibody mediated) and the clinical situation of the patient. Patients may relapse and should receive appropriate follow-up care from a physician familiar with the diseases.