Autoimmune Encephalitis

Definition

Autoimmune encephalitis is a clinical label applied to a group of disorders of the brain associated with a self-reactive immune response causing brain dysfunction. The group of disorders includes cellular-immunity mediated syndromes—such as acute disseminated encephalomyelitis (1), Rasmussen’s encephalitis (2), and intracellular-paraneoplastic antibody–associated encephalomyelitis (3)—but also disorders associated with autoantibodies to synaptic proteins (“autoimmune synaptopathies” [4]), the poststreptococcal neurological syndromes (5), encephalitis lethargica (6), and a combination of mechanisms (antibody, cytokine, vascular) in syndromes including neuropsychiatric systemic lupus erythematosus (7), neuro-Behçet’s disease (8), and neurosarcoidosis (9). Antibody-associated encephalitis often presents with a syndrome of psychiatric and cognitive features, which are often but not exclusively accompanied by seizures and movement disorders. This article focuses on antibody-associated synaptic disorders of the brain, which are of the most relevance to psychiatrists.

Assessment and Differential Diagnosis

Mental status change, fever, and seizures are hallmarks of classic encephalitis, and the identification and empirical treatment of viral infection of the brain, commonly herpes-simplex virus (HSV), is the first action required of physicians to avoid the devastating brain destruction that occurs from viral encephalitis and to identify bacterial causes of illness, including Listeria meningitis. Patients with autoimmune encephalitis may have a similar clinical presentation, often associated with fever, which in itself is not reliable to distinguish between infective and autoimmune encephalitis (10). Computed tomography and magnetic resonance imaging of the brain are often abnormal in infectious encephalitis, although often after a delay of 72 hours, and cerebrospinal fluid (CSF) examination for pleocytosis and HSV polymerase chain reaction may be falsely negative if done fewer than 72 hours into the illness; therefore, a strong index of suspicion is required to ensure that a diagnosis of infective encephalitis is not missed. Extensive investigation and consideration of alternative diagnoses is necessary, which can range across most differential diagnoses in neurology and psychiatry (drug-induced states, hereditary disease; e.g., C9ORF72-related frontotemporal dementia, intravascular and primary central nervous system [CNS] lymphoma, low-grade tumors of the brain, HIV infection).

Autoimmune disorders of brain synapses often produce symptoms that result in review by a psychiatrist first, particularly the N-methyl-d-aspartate receptor (NMDAR) antibody encephalitis: psychosis, mania, amnesia, and subcortical cognitive symptoms, catatonia-like movement disorders, and affective symptoms (4). Alternative symptoms, including seizures, headache, and cognitive dysfunction with prominent anterograde amnesia and frontal dysexecutive dysfunction, may be the prominent features at presentation. The nature of symptoms may be subacute or acute, often worsening over a period of weeks or months. The identification of different autoantibodies has been useful in allowing classification into “cell-surface antibody syndromes,” although there is considerable overlap (Table 1). Older clinical diagnoses—including Hashimoto’s encephalopathy (an encephalopathy first described by Lord Brain in 1966, associated with thyroid autoimmunity and rapid corticosteroid treatment responsiveness) (11) and Bickerstaff’s brainstem encephalitis (an encephalopathy associated with GQ1b ganglioside antibodies and the peripheral nerve Fisher syndrome [areflexia, ophthalmoplegia, and ataxia]) (12)—are likely to represent disorders on a clinical spectrum, and much work is under way to try to reorganize the classification on the basis of disease mechanisms and hence allow tailored therapy and the discovery of novel therapies.

NMDAR Antibody Encephalitis

Dalmau in 2005 described a group of young women with ovarian teratoma with an encephalitis characterized by a psychosis prodrome, rapid descent into profound coma, and rhythmic movement disorder, with antibodies to the NMDAR described in this group in 2007 (13, 14). The original patients had ovarian teratoma, likely leading to a paraneoplastic phenomenon, because the teratoma express neuronal tissue; but subsequently, it has been shown that most patients do not have an underlying tumor. People of all ages, from infants to the elderly, may be affected, although young adults are most commonly affected, and the disease is as common, if not more common, than infective encephalitis among those under age 30, for example, as seen in the California Encephalitis Project (15). In NMDAR encephalitis, a hyperkinetic movement disorder may be present more often than a parkinsonian syndrome; dysautonomia with sweating changes, tachycardia, and syncope may be present, although more often among patients with severe disease resulting in intensive care admission (16–18). Brain imaging may be normal, reflecting the disease mechanism of antibody-mediated neuronal dysfunction rather than gross cellular infiltration. Untreated, the NMDAR encephalitis syndrome is associated with spontaneous remission and frequent relapse (18), as well as with long-term cognitive dysfunction and atrophy, likely as a result of damage induced by prolonged disease activity, although there is little research to date on long-term outcomes, particularly related to the other antibody associations. Electroencephalogram is often abnormal, although it may be normal for 20% of patients. CSF examination is useful for excluding mimics of autoimmune encephalitis and may reveal a cellular pleocytosis, often lymphocytic, or oligoclonal bands; but a normal spinal fluid examination does not exclude the disease. Low levels of NMDAR antibodies are seen perhaps as a secondary phenomenon in degenerative diseases such as motor neuron disease, prion disease, and Alzheimer’s disease; the clinical relevance of these antibodies is not clear (19). The NMDAR antibodies are likely pathogenic, reducing NMDAR currents in vitro and inducing appropriate effects in animal models with passive transfer. The likely mechanism is that of internalization of NMDAR antibodies and disruption of trafficking of the NMDAR at synapses (4). Patients with the full syndrome may have a protracted course, with some requiring ventilation in the intensive care unit for many months, but then making excellent recovery. A range of therapies have been tried—including intravenous immunoglobulins, plasmapheresis, oral immunosuppressant agents, cyclophosphamide, and B cell depletion with rituximab (18)—though clinical trials are required to identify the most effective and safe treatment regimens.

We and others have looked for NMDAR and related antibodies in patients with psychosis for three main reasons: two-thirds of patients with NMDAR encephalitis first see a psychiatrist (11, 17), NMDAR hypofunction is a good model of psychosis and schizophrenia (20), and the neuropathology of schizophrenia (21) has similarities to the neuropathology of NMDAR encephalitis (22). One caveat with patients experiencing psychosis is that they often have clinical or subclinical movement disorders and autonomic and cognitive dysfunction, although not at a level that would normally prompt referral to a neurologist. We and others have found a small percentage (2%−7%) of patients with relatively pure acute psychosis to have NMDAR or related antibodies (23–25). These patients often have a psychosis refractory to conventional therapies, although they are highly sensitive to immunotherapies (26). Different prevalence rates in different epidemiological studies looking for these antibodies in psychosis are likely related to the methodology of antibody measurement as well as to the demographic characteristics of those tested (27). In particular, caution is called for in the interpretation of prefixed cell-based assay slides and in the high prevalence of IgA and IgM (unlikely to be pathogenic) found, for instance, in a large number of disease controls and in dementias (28). The possibility that antibody-associated autoimmunity underlies the symptoms of some patients with psychotic illness, who may improve with available therapies, is very exciting, although further study and treatment require care, particularly with the risks of immunotherapies and immunosuppressives, and well-designed randomized clinical trials are now required.

A further discrete group is younger patients, often infants, with HSV-1 encephalitis who recover with treatment but relapse with a hyperkinetic movement disorder and who have an absence of HSV-1 in CSF by polymerase chain reaction and a presence of NMDAR antibodies (29). The association of HSV-1 infection with cognitive symptoms in schizophrenia is a potential link here and deserves further study (30). Further detailed reviews are provided by Graus et al. (31), Irani et al. (32), and Crisp et al. (4).

LGI1, Caspr2, and Antibodies to the Voltage-Gated Potassium Channel (VGKC) Complex

In 2001, before the NMDAR antibody discovery, Angela Vincent and colleagues at Oxford University described anterograde amnesia and seizures with high-intensity signal change in mesial temporal lobes (overall, a “limbic” picture) in two patients with high levels of antibodies to the VGKC determined by radioimmunoprecipitation assay (33). One of the original patients had coexistent myasthenia gravis. The radioimmunoprecipitation assay identifies antibodies to proteins complexed to the VGKC, commonly leucine-rich glioma-inactivated 1 (LGI1) (34) and caspr2 (35). Low levels of antibodies to the VGKC measured by radioimmunoprecipitation assay may therefore detect intracellular binding and reflect an epiphenomenal finding or be found at low levels in healthy aging (36). A serum level of 400 pM or above is more likely to be disease relevant in the appropriate clinical setting, although in the absence of LGI1 or caspr2 antibodies it should be treated with caution. Caspr2 antibodies are more likely to be associated with an occult underlying malignancy, and a thorough search, including with positron emission tomography imaging, is recommended. Caspr2 is expressed in the juxtaparanode, and patients with this antibody are more likely to experience muscle cramps and sweating, reflecting peripheral nerve hyperexcitability (the Isaac syndrome in isolation [37], or Morvan’s syndrome when in combination with a “limbic” encephalitis [35]). Brief stereotyped focal dystonic seizures (faciobrachial dystonic seizures [FBDS]) involving the face and arm, and sometimes leg, can occur from tens to hundreds of times a day, refractory to anticonvulsant therapy (38, 39) in patients with LGI1 antibodies. FBDS can precede the amnesia in this syndrome, and corticosteroid treatment is associated with a better response and probably prevention of atrophy and damage in retrospective series (40), although no randomized trials have been undertaken in this rare disorder. The amnesia may be subtle, and the study of these patients has furthered insight into basic memory mechanisms in the temporal lobe, for example, demonstrating impaired binding together of disparate features of an object in memory (41). Isolated psychosis may occur, often in relapse of patients with LGI1 antibodies (39), and psychiatric symptoms, including affective and psychotic and cognitive symptoms, are common (unpublished data). Cardiac arrhythmias occur in NMDAR and LGI1 antibody-associated encephalitis, and ECG monitoring is recommended, particularly in the severe stages of the illness (42).

Less Common Antibody Associations

Novel antibodies are being discovered at a rapid pace, some with clues to clinical syndromes (4). Glycine receptor antibodies, the target highly expressed in the spinal cord and brainstem, are associated with progressive encephalomyelitis, rigidity, and myoclonus and with opsoclonus and encephalitis with seizures among some patients (43). Some of these patients have antibodies to the intracellular protein glutamic acid decarboxylase, an antibody usually associated with stiff-person syndrome and type 1 diabetes (44). Dopamine D2-receptor antibodies are associated with a brainstem encephalitis with prominent psychosis and hypo- and hyperkinetic movement disorders (45); antidipeptidyl-peptidase–like protein-6 antibodies are associated with profuse diarrhea prior to an encephalitic illness (46); the gamma-aminobutyric acid (GABA)-(A) and GABA(B) receptor antibodies are associated with refractory status epilepticus (47). Descriptions of these antibody associations are few and require replication and description in larger populations. Table 1 summarizes these and related disorders, including poststreptococcal neuropsychiatric disorders, CNS lupus, and encephalitis lethargica (Box 1).

Future Directions

The disease-causing mechanisms of antibodies to brain proteins are under study in this novel field, including passive-transfer experiments (injection of antibodies into the blood or spinal fluid in mice or rats to determine the effects of the antibodies, followed by a washout period to see whether the effects can be reversed) (48), in-depth analysis of the origins of the immunoglobulins and immunizing mechanisms (49), and biomarker discovery (e.g., protein markers measurable in blood or spinal fluid that can help track disease activity with treatments, or predict disease activity states). A hunt is under way for the discovery of novel antigen targets that may help categorize subtypes of illness and aid rapid clinical diagnosis, through extensive testing of possible cases but also with neurological, immunological, and healthy controls to independently judge the specificity of an antibody test. Gold-standard tests, such as live cell-based assays, will have advantages over rapid fixed immunoassays and enzyme-linked immunosorbent assay tests, although there will be a role for the latter in terms of providing rapid results to clinicians—within a day of blood sampling, which is a situation seen in rheumatology in the antinuclear antibody testing for suspected systemic lupus erythematosus. The field is in need of in-depth clinical and phenotypic analysis (50) to provide independent biomarkers to aid in clinical diagnosis and research studies, given the nonspecific nature of many of the phenotypes—for instance, advanced brain imaging or CSF biomarkers of inflammation or damage. Greater integrative working between psychiatrists and neurologists is required to take forward work in synaptic neuropsychiatric disorders, which heavily relies on collaboration, sample collection, and engagement of patients and charities in research, including natural history studies and clinical trials. Immune mechanisms are increasingly implicated in the pathogenesis of degenerative diseases, including Alzheimer’s disease and Parkinson’s disease. The ability to modify and target immune mechanisms through the repurposing of existing therapies to allow modification of the disease course in neurodegenerative disease is a topic of great interest.