Epidemiologic considerations

The prevalence of AIE is increasing, likely related to increasing recognition of the diagnosis. ^{Reference Dubey, Pittock and Kelly1,Reference Hébert, Riche and Vogrig2} AIE occurs in people of all ages; however, there are antibody-specific age predilections. For example, N-methyl-d-aspartate receptor (NMDAR) antibody encephalitis occurs more frequently in children and young women ^{Reference Hébert, Riche and Vogrig2,Reference Gable, Sheriff, Dalmau, Tilley and Glaser3} while leucine-rich glioma-inactivated 1 (LGI1) antibody encephalitis predominates in older men. ^{Reference Gable, Sheriff, Dalmau, Tilley and Glaser3,Reference Irani, Alexander and Waters10} A personal or family history of autoimmune disease may also be a predisposing factor. ^{Reference Mittal, Rabinstein, Hocker, Pittock, Wijdicks and McKeon12} Other known risk factors for AIE include preceding herpes simplex encephalitis and other forms of infectious encephalitis. ^{Reference Armangue, Spatola and Vlagea13–Reference Predkele and Mednieks17} Immune checkpoint inhibitors (ICIs), used in the treatment of a growing number of malignancies, have also been identified as conferring increased risk for diverse neurological autoimmune complications, including AIE. ^{Reference Marini, Bernardini and Gigli18}

Presenting symptoms

AIE typically presents with a subacute onset of symptoms, however, certain neural antibody-associated syndromes may have a more indolent course, such as LGI1 antibody, contactin-associated protein-like 2 (CASPR2) antibody, glutamic acid decarboxylase-65 (GAD65) antibody and dipeptidyl-peptidase-like protein 6 (DPPX) antibody encephalitis. ^{Reference Abboud, Rossman and Mealy19–Reference van Sonderen, Thijs and Coenders22} An infectious prodrome often occurs days to weeks preceding neuropsychiatric manifestations. Other non-specific prodromal symptoms can include headache, fever, fatigue, sleep disturbance, weight loss, and early psychiatric manifestations. ^{Reference Titulaer, McCracken and Gabilondo9}

A detailed history and physical examination are critical to achieving a timely diagnosis of AIE. In one retrospective series examining diagnoses at hospital admission in patients subsequently confirmed as having AIE, only a third were initially thought to have encephalitis; more than 50% of those were initially felt to be of infectious origin. ^{Reference Baumgartner, Rauer and Hottenrott23} This is reflective of the heterogeneity of clinical presentations of AIE. Misdiagnosis is also not uncommon; a recent study of patients referred to AIE specialty clinics reported a misdiagnosis rate of 27% among patients with an original referral diagnosis of AIE. ^{Reference Flanagan, Geschwind and Lopez-Chiriboga4} Factors associated with misdiagnosis include overreliance on isolated serum antibody positivity, and classification of encephalopathy based on non-specific cognitive, psychiatric and/or functional symptoms. ^{Reference Flanagan, Geschwind and Lopez-Chiriboga4}

The most common presenting symptoms of AIE are reflected by diagnostic criteria (see below) and consist of acute mental status changes (including cognitive decline and working memory deficits), seizures, and psychiatric symptoms (commonly paranoia, agitation, personality changes and/or hallucinations). ^{Reference Kayser, Titulaer, Gresa-Arribas and Dalmau24–Reference Uy, Binks and Irani26} Other symptoms may be seen depending on the phenotype/antibody association and common phenotypes of AIE include, but are not limited to, limbic encephalitis, cerebellar or brainstem syndromes, and encephalomyelitis. ^{Reference Uy, Binks and Irani26} While significant overlap exists in the diverse phenotypes of AIE, certain features should prompt suspicion for a particular antibody. Faciobrachial dystonic seizures, for example, are strongly associated with LGI1 antibody encephalitis. ^{Reference Irani, Michell and Lang27}

Diagnostic criteria

Diagnostic criteria for AIE were first proposed by Graus et al. in 2016 (Figure 1). ^{Reference Graus, Titulaer and Balu28} An AIE diagnosis can be classified according to the degree of diagnostic certainty including possible, probable, and definite. While the diagnostic criteria can be helpful in identifying patients in whom a diagnosis of AIE should be considered, they also serve to identify those patients in whom AIE is less likely. In cases where criteria are met only for “possible” AIE after investigations including neural antibody testing are complete, early involvement of a specialist in autoimmune neurology is important. This is especially true if considering a trial of immunotherapy.

Figure 1: Diagnostic criteria for possible autoimmune encephalitis.

Based on the above diagnostic criteria, transition from “possible” to “probable” AIE can be achieved by meeting more specific phenotypic criteria for NMDAR antibody encephalitis or antibody-negative AIE. A diagnosis of definite AIE requires neural specific antibody positivity or meeting specific criteria for limbic encephalitis/acute disseminated encephalomyelitis (ADEM)/Bickerstaff’s encephalitis in addition to the criteria for “possible” AIE (Figure 2).

Figure 2: Diagnostic progression in autoimmune encephalitis (excluding ADEM, Bickerstaff’s). ADEM = Acute disseminated encephalomyelitis; CNS = central nervous system; CSF = cerebral spinal fluid; EEG = electroencephalography; GAD65 = glutamic acid decarboxylase-65; IgG = immunoglobulin G; MRI = magnetic resonance imaging (of the brain); OGBs = oligoclonal bands; PET = positron emission tomography (of the brain).

While psychiatric symptoms are common in AIE ^{Reference Titulaer, McCracken and Gabilondo9} caution must be exercised in cases of isolated psychiatric manifestations. Some studies report the identification of NMDAR antibody in patients with schizophrenia and psychotic disorders, as well as in healthy control subjects; however, antibody detection in these series typically relied on serum testing alone or evaluated antibody subtypes/sub-classes which are not known to be pathogenic. ^{Reference Jernbom Falk, Galletly and Just29–Reference Luo, Li and Jiang31} More recent studies have confirmed that NMDAR antibodies are highly specific for AIE when detected in CSF in addition to serum. ^{Reference Guasp, Giné-Servén and Maudes32} In recognition of this, while balancing the desire to not miss any patient with an autoimmune cause of psychiatric symptoms, it has been suggested that certain factors should prompt further workup in patients with a first psychotic episode. This includes rapid progression, the presence of concurrent neurological symptoms, fluctuating catatonia, refractoriness to antipsychotic medications, or abnormal paraclinical investigations. ^{Reference Herken and Prüss33}

Laboratory testing

Cerebrospinal fluid and serum

Please see Table 1 for an outline of suggested CSF and serum investigations for AIE. ^{Reference Abboud, Probasco and Irani5,Reference Graus, Titulaer and Balu28,Reference Au, Latonas, Shameli, Auer and Hahn34} CSF examination is of vital importance in the workup of AIE. Infectious pathogens should be ruled out. Evidence of inflammation can be found by evaluating the protein, cellular profile, oligoclonal banding, and IgG index. Paired CSF and serum samples must be tested to identify CSF-specific oligoclonal bands. CSF findings most consistent with AIE include mild to moderate lymphocytic leukocytosis (generally <100 white blood cells), presence of CSF-specific oligoclonal bands, increased IgG index, and/or elevated protein. ^{Reference Blinder and Lewerenz35} A significant proportion of people with AIE have normal CSF studies ^{Reference Escudero, Guasp and Ariño36} and inclusion of testing for oligoclonal bands significantly increases sensitivity. ^{Reference Hébert, Gros and Lapointe37} Glucose is typically normal in AIE although low glucose can be seen in autoimmune glial fibrillary acidic protein (GFAP) astrocytopathy. ^{Reference Lan, Li, Ai and Luo38,Reference Zhu, Sun, Yang, Yu and Wang39} Testing for CJD should be sent only if there is high clinical suspicion. Repeat CSF examination can be considered if diagnostic uncertainty remains, as CSF findings may evolve with evolution of AIE. ^{Reference Dalmau, Lancaster, Martinez-Hernandez, Rosenfeld and Balice-Gordon40}

Table 1: Initial laboratory investigations of suspected AIE. Tests in bold are strongly recommended to establish the diagnosis and exclude common mimics. Optional tests are also listed if clinically indicated by presenting history

AFB = acid-fast bacillus; ANA = anti-nuclear antibody; ANCA = anti-nuclear cytoplasmic antibody; C3 = complement component 3; C4 = complement component 4; CMV = cytomegalovirus; CRP = C-reactive protein; CSF = cerebrospinal fluid; dsDNA = double stranded DNA; EBV = Epstein-Barr virus; GAD65 = glutamic acid decarboxylase-65; GFAP = glial fibrillary acidic glycoprotein; HHV-6 = human herpesvirus-6; HIV = human immunodeficiency virus; HSV = herpes simplex virus; IgG = immunoglobulin G; IgLON5 = IgLON Family Member 5; JCV = John Cunningham virus; MOG = myelin oligodendrocyte glycoprotein; PCR = polymerase chain reaction; SSA = Sjögren’s-syndrome-related antigen A; SSB = Sjögren’s-syndrome-related antigen B; TB = tuberculosis; TPO = thyroid peroxidase; TSH = thyroid stimulating hormone; VDRL = venereal disease research laboratory; VZV = varicella zoster virus.

**Hepatitis and tuberculosis testing is important for future treatment planning and should be obtained before use of steroid or IVIg whenever possible.

***Yield of flow cytometry without a history of hematological malignancy nor leptomeningeal gadolinium enhancement has been shown to be extremely low. ^{Reference Au, Latonas, Shameli, Auer and Hahn34}

Serum screening should incorporate a broad differential, excluding infectious, metabolic, and systemic autoimmune causes of encephalopathy. Of note, some systemic clues may exist to suggest underlying cause; for example, hyponatremia can be seen in 65% of cases of LGI1 antibody encephalitis. ^{Reference van Sonderen, Thijs and Coenders22}

Neural antibody testing

Neural antibody testing in patients with suspected AIE is invaluable to patient diagnosis and management. ^{Reference Graus, Titulaer and Balu28,Reference Budhram, Dubey and Sechi41,Reference Ropper, Dalmau and Graus42} While rare clinical features are virtually pathognomonic for an individual neural antibody, many clinical and neuroimaging features of AIE can occur with various neural antibodies (e.g., memory deficits, psychiatric symptoms, seizures, medial temporal lobe T2-hyperintensities on MRI brain). ^{Reference Abboud, Probasco and Irani5,Reference Irani, Michell and Lang27,Reference Graus, Titulaer and Balu28,Reference Budhram, Leung, Nicolle and Burneo43} Because of the potential for phenotypic overlap across antibody-associated presentations, comprehensive panel-based neural antibody testing in patients with suspected AIE is generally recommended over sequential antibody testing to maximize sensitivity and facilitate prompt diagnosis.

Sensitivity of serum is higher than CSF for some antibodies (e.g., LGI1 and CASPR2 antibodies) while sensitivity/specificity of CSF is higher than serum for others (e.g., NMDAR and GFAP antibodies). ^{Reference Budhram, Yang, Bhayana, Mills and Dubey44–Reference Gresa-Arribas, Titulaer and Torrents47} Furthermore, for rare/novel antibodies, the preferred specimen to test may not be well-established. For these reasons, testing both serum and CSF optimizes sensitivity and specificity. Assays commonly used for neural antibody detection include tissue indirect immunofluorescence/immunohistochemistry (TIIF/IHC), immunoblots, cell-based assays, radioimmunoassays and enzyme-linked immunosorbent assays (Table 2). Incorporation of TIIF/IHC is recommended to optimize sensitivity and specificity of neural antibody testing for patients with suspected AIE and should be considered standard of practice. ^{Reference Graus, Vogrig and Muñiz-Castrillo63}

Table 2: Test methodologies employed for neural antibody detection in patients with suspected autoimmune encephalitis

ADEM = acute disseminated encephalomyelitis; AMPAR = α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor; CASPR2 = contactin-associated protein-like 2; CRMP5 = collapsin response-mediator protein 5; DNER = delta/Notch-like epidermal growth factor-related receptor; DPPX = dipeptidyl-peptidase-like protein 6; FLAMES = FLAIR-hyperintense Lesions in Anti-MOG-associated Encephalitis with Seizures; GABA(B)R = γ-aminobutyric acid type B receptor; GAD65 = glutamic acid decarboxylase-65; GlyR = glycine receptor; IGLON5 = IgLON Family Member 5; LGI1 = leucine-rich glioma-inactivated 1; MOG = myelin oligodendrocyte glycoprotein; NMDAR = N-methyl-D-aspartate receptor; PERM = progressive encephalomyelitis with rigidity and myoclonus; SOX1 = SRY-box transcription factor 1; VGKC = voltage-gated potassium channel.

In Canada however, many centers rely on commercial laboratories that do not incorporate TIIF/IHC in their testing algorithms. As such, the results reported are not confirmed with 2 testing methodologies, which may result in a high proportion of false-positive results. ^{Reference Budhram, Nicolle and Yang60–Reference Ruiz-García, Martínez-Hernández, Saiz, Dalmau and Graus62,Reference Budhram, Mirian, McFadden, Edmond, Bhayana and Yang65} Similarly, more limited testing can miss antibodies that are not included in, or are sub-optimally detected by, standard commercially available panels; such antibodies may only be detected if TIIF/IHC along with confirmatory assays available at specialized centers are used. ^{Reference McCracken, Zhang and Greene81,Reference Sabater, Saiz, Dalmau and Graus82} As such, it is highly recommended to involve a specialist in autoimmune neurology when a positive neural antibody result is found in a patient with an atypical clinical presentation or when no antibody is detected in a patient in whom AIE is strongly suspected, so they can review results and direct further testing at specialized laboratories.

Importantly, awaiting neural antibody results should not delay consideration of empiric immunotherapy if there is a high index of suspicion for AIE and reasonable exclusion of alternative diagnoses. Once available, a positive neural antibody result can serve to confirm the diagnosis, inform malignancy screening, and support escalation of immunotherapy in cases where there is incomplete response.

Special mention must be given to testing for thyroid antibodies, which are found in 11% of healthy controls and as high as 20% of adults over 60 years. ^{Reference Hollowell, Staehling and Flanders83} Accordingly, testing for thyroglobulin antibodies or thyroid peroxidase antibodies is of limited value in the evaluation of AIE, may contribute to AIE misdiagnosis, and should not be routinely performed. ^{Reference Flanagan, Geschwind and Lopez-Chiriboga4,Reference Mattozzi, Sabater and Escudero84,Reference Valencia-Sanchez, Pittock and Mead-Harvey85} Similarly, detection of voltage-gated potassium channel (VGKC) antibodies without LGI1 or CASPR2 antibodies lacks specificity for neurologic autoimmunity, so testing of VGKC antibodies is not recommended. ^{Reference Lang, Makuch and Moloney79,Reference Michael, Waters and Irani80}

Imaging

Electroencephalography (EEG)

EEG can provide evidence of focality when MRI is normal, occasionally shows findings that may help identifying specific subtypes of AIE and is part of the diagnostic criteria for definite limbic AIE ^{Reference Graus, Titulaer and Balu28,Reference Moise, Karakis and Herlopian106,Reference Wennberg, Steriade, Chen and Andrade107} (Figure 2). Common findings include slowing and epileptiform changes, but these are not specific. ^{Reference Carvajal-González, Leite and Waters70,Reference Elkhider, Sharma, Kapoor, Vattoth and Shihabuddin108} Extreme delta brush is seen in up to 30% of cases of NMDAR antibody encephalitis but is not unique to this disorder. ^{Reference Moise, Karakis and Herlopian106,Reference Nathoo, Anderson and Jirsch109,Reference Baykan, Gungor Tuncer and Vanli-Yavuz110} Hyperventilation induced focal seizures may also be a marker of AIE. ^{Reference Wennberg111}

Other ancillary tests

Biopsy is not typically necessary in AIE, and the presence of inflammatory cells alone is not specific. ^{Reference Dalmau, Lancaster, Martinez-Hernandez, Rosenfeld and Balice-Gordon40} The utility of brain biopsy is for confirmation of alternate pathologies. Nerve conduction studies and electromyography can be helpful in cases of CASPR2 antibody encephalitis, Bickerstaff’s encephalitis, and stiff person spectrum disorders. ^{Reference Alberti, Povedano, Montero and Casasnovas112,Reference De Wel and Claeys113}

Defining severity of AIE

Defining the clinical severity of autoimmune encephalitis can be helpful in guiding treatment. While clinical scales do exist (e.g., Clinical Assessment Scale in Encephalitis ^{Reference Lim, Lee and Moon114} ) we propose a rapid and practical approach to stratify disease severity based on functional impairment.

1. i. Severe – All patients requiring non-elective admission (including ICU) or any patient with significant and/or progressive functional impairment defined as inability to perform basic ADLs.

2. ii. Mild/Moderate – all others (mostly applicable to outpatients).

Recommendations for first-line treatment

The treatment of autoimmune encephalitis (AIE) is predominantly based on case series/retrospective data ^{Reference Nosadini, Eyre and Molteni115–Reference Thaler, Zimmermann and Kammermeier117} and expert consensus apart from one positive randomized controlled trial for intravenous immunoglobulin (IVIg) in LGI1/CASPR2 antibody seropositive patients. ^{Reference Dubey, Britton and McKeon118} A clinical presentation consistent with the diagnosis of AIE with exclusion of an infectious cause are the key elements to initiate first-line treatment without delay. Studies suggest that early immunotherapy is associated with better outcome. ^{Reference Balu, McCracken, Lancaster, Graus, Dalmau and Titulaer7–Reference Titulaer, McCracken and Gabilondo9} Early tumor removal in paraneoplastic encephalitis is crucial to consider in parallel with first-line therapeutic agents.

For patients who have acute presentations (like are typical of viral encephalitides) or subacute/chronic presentations with clinical or ancillary test findings concerning for infection (e.g., bacterial, fungal or parasitic infections of the CNS), antimicrobial treatment for a possible infectious cause of encephalitis as well as infectious disease consultation as appropriate should be pursued prior to empiric administration of immunotherapy.

We recommend that patients with severe AIE be treated with a combination of high dose steroids (1 g IV methylprednisolone daily for 5 days, or equivalent), AND IVIg (2 g/kg over 2–5 days) or plasma exchange (PLEX) (5 exchanges every other day over 10 days) (Figure 3). Treatment with steroid monotherapy can be considered in mild/moderate cases, such as in patients with anti-LGI1 encephalitis who are assessed in the outpatient setting with seizures in relative isolation, but input from a specialist in autoimmune neurology is recommended. If an associated neoplasm is found, treatment of the neoplasm must be part of the first-line protocol.

Figure 3: Proposed treatment algorithm for AIE. PLEX = plasma exchange; IVIg = intravenous immunoglobulin; GAD = glutamic acid decarboxylase; GFAP = glial fibrillary acidic protein.

Although some studies suggest that PLEX might induce a more rapid effect by removing autoantibodies and other inflammatory substances in the plasma. ^{Reference Zhang, Liu, Jiang, Chen and Su119} there is currently no definite evidence to suggest the superiority of IVIg versus PLEX. Additionally, there are some concerns that antibody mediated diseases associated with pathogenic IgG4 antibodies (e.g., LGI1, CASPR2) may not respond as well to IVIg. ^{Reference Dalakas120} It should be noted, that treatment with IVIg can affect autoantibody testing in serum and may result in a CSF pleocytosis and increased CSF IgG, ^{Reference Shayan, Adamiak, Choe, Relkin and Lee121,Reference Yelehe-Okouma, Czmil-Garon, Pape, Petitpain and Gillet122} and so completion of investigations is necessary prior to initiation of IVIg.

In addition to the recommendation for early and appropriate treatment, we recommend early consideration for the transfer of any patient with moderate to severe AIE to a center with experience in the management of such patients and access to PLEX.

Use of corticosteroids after initial treatment

Current evidence for or against the use of corticosteroid taper is limited, although use of a taper may help prevent early relapse. ^{Reference Abboud, Probasco and Irani5,Reference Alkabie and Budhram123,Reference Nosadini, Eyre and Giacomini124} We recommend that corticosteroid taper be considered in most patients treated with any first-line treatment protocol which includes high dose IV steroids, particularly in patients with severe disease, unless a clear medical contraindication exists (including the risk of exacerbating neuropsychiatric symptoms). This is in keeping with other expert recommendations. ^{Reference Abboud, Probasco and Irani5,Reference López-Chiriboga and Flanagan125}

If a steroid taper is used, we recommend 0.5–1 mg/kg/day of oral prednisone, tapering over 4–12 weeks depending on risk of side effects and comorbidities. An alternate (and possibly better tolerated regimen) of pulse steroids (1 g IV methylprednisolone or PO equivalent) weekly initially for 6 weeks followed by every 2 weeks for the next 6 weeks may also be considered.

Longer tapers may be helpful in some diseases (e.g., LGI1 antibody encephalitis). ^{Reference Alkabie and Budhram123} For NMDAR antibody encephalitis, if ongoing corticosteroids are administered in the first months of disease then IV pulses rather than oral taper may be preferable, although guidance for adult cases is limited. ^{Reference Nosadini, Thomas and Eyre126} Input from a specialist in autoimmune neurology may be helpful in guiding taper.

Treatment trials in patients meeting criteria only for “Possible” AIE

Patients meeting only possible AIE criteria after completion of investigations merit special consideration. Importantly, these were not intended to be used as standalone criteria, but rather as the minimum requirements to suspect AIE. ^{Reference Dalmau and Graus127} In patients who only meet criteria for possible AIE but in whom alternative diagnoses have reasonably been excluded, a controlled immunotherapy trial with predefined distinct and objective treatment measures of efficacy can be considered with involvement of a specialist in autoimmune neurology.

Defining treatment failure

There is no widely accepted definition of treatment failure or refractory disease in autoimmune encephalitis. Second-line therapies are felt to be relatively safe and effective in managing AIE. ^{Reference Titulaer, McCracken and Gabilondo9} They are also associated with better outcomes when compared to those who receive additional first-line treatment or no further treatment. ^{Reference Titulaer, McCracken and Gabilondo9} Accordingly, we suggest a low threshold to escalate immunotherapy, particularly in patients with severe AIE.

We define treatment failure in patients with AIE as lack of improvement or worsening 5–10 days after initiation of first-line treatment in severe AIE and 2–4 weeks after initiation of first-line treatment in mild/moderate AIE. The shorter time period in severe AIE is to encourage earlier consideration of second-line therapy (discussed below). Whenever possible, lack of improvement or worsening of objective measures (e.g., seizure frequency, cognitive testing scores) should be used to define treatment failure.

In patients who experience treatment failure, particularly those who are antibody-negative, care should be taken to ensure that the diagnosis of autoimmune encephalitis is correct.

Acute second-line treatment in patients with cell surface antibodies or negative antibodies

Patients with autoimmune encephalitis who have antibodies targeting cell surface antigens (Table 3) should preferentially receive rituximab as second-line therapy due to possible better efficacy and more favorable safety profile ^{Reference Abboud, Probasco and Irani5,Reference Shin, Lee and Park128} (Figure 3). In antibody-negative AIE, rituximab is preferable over cyclophosphamide due to its more favorable safety profile. ^{Reference Abboud, Probasco and Irani5,Reference Shin, Lee and Park128,Reference Ledingham, Plant, Mustafa and Patil129} In general, however, input from a specialist in autoimmune neurology is recommended prior to administering second-line immunotherapies given their prolonged immunosuppressive effect and unique toxicities (e.g., infertility with cyclophosphamide, discussed later on). For ease of administration, rituximab may be administered as two intravenous infusions of 1 g, two weeks apart, rather than four 375 mg/m² infusions for 4 weeks. ^{Reference Shin, Lee and Park128} Administration of monthly cyclophosphamide infusions of 600–1000 mg/m² for 3–6 months have commonly been described. ^{Reference Shin, Lee and Park128} Myeloablative cyclophosphamide protocols can be considered in severe disease with input from a specialist in autoimmune neurology with experience in their use.

Table 3: Anti-neural antibodies associated with encephalitis

AMPAR = α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor; ANNA = antineuronal nuclear antibody; CASPR2 = contactin-associated protein-like 2; CRMP5 = collapsin response-mediator protein 5; DNER = Delta/notch-like epidermal growth factor-related receptor; DPPX = dipeptidyl-peptidase-like protein; GABA_AR = gamma-aminobutyric acid-a receptor; GABA_BR = gamma-aminobutyric acid-b receptor; GAD = glutamic acid decarboxylase; GFAP = glial fibrillary acidic protein; GlyR = glycine receptor; LGI1 = leucine-rich glioma-inactivated protein 1; LKLHL11 = Kelch-like protein 11; mGluR1 = metabotropic glutamate receptor type 1; mGluR5 = metabotropic glutamate receptor type 5; NMDAR = N-methyl-D-aspartate receptor; PCA-1 = Purkinje cell cytoplasmic antibody type 1; PCA-2 = Purkinje cell cytoplasmic antibody type 2; References: 63,130.

Consideration should also be given to repeated high dose pulse steroids, concurrent steroid taper, and repeated PLEX/IVIg (if initial PLEX/IVIg was effective) or initiation of PLEX/IVIg (if one of these were not applied during first-line treatment). ^{Reference Abboud, Probasco and Irani5} This should be strongly considered when there is potential for delays in treatment, for example if insurance/administrative approval is needed for use of rituximab.

Acute second-line treatment in patients with intracellular antibodies

In confirmed or probable paraneoplastic cases, treatment and/or tumor resection is strongly recommended in all patients in consultation with oncology and/or neuro-oncology (Figure 3). Paraneoplastic encephalitis with autoantibodies against intracellular targets may be immunotherapy-resistant, in contrast to autoimmune encephalitis with autoantibodies against extracellular targets. ^{Reference Ledingham, Plant, Mustafa and Patil129} However, this is not universal and so trials of immunotherapy (both first line and second line if initial treatment fails) should still be considered in these cases. ^{Reference Vedeler, Antoine and Giometto130} Patients with encephalitis associated with antibodies targeting intracellular antigens, and in particular those with high-risk paraneoplastic antibodies (e.g., anti-Hu, anti-Yo), should preferentially receive cyclophosphamide due to its broader immunosuppressive effect. ^{Reference Binks, Uy, Honnorat and Irani131,Reference Madondo, Quinn and Plebanski132} There is little evidence regarding optimum cyclophosphamide therapy or dosing regimens in this setting.

Despite B-cell depletion being the main mechanism of action for rituximab, there is also indirect suppression of T-cell activity through changes in B- and T-cell signaling pathways. ^{Reference Ramwadhdoebe, van Baarsen and Boumans133} Rituximab can be considered in circumstances where contraindications to cyclophosphamide exist.

For patients with encephalitis associated with antibodies targeting intracellular antigens that have a lower cancer association (e.g., anti-GFAP, anti-GAD65), either cyclophosphamide or rituximab may be considered for second-line immunotherapy.

Additionally, if they have not been given previously, trial of IVIg or PLEX may be considered when other immunotherapies are unavailable or contraindicated, given their relatively low risk and rare reports of benefit attributed to their use even in patients with AIE who harbor antibodies against intracellular targets. ^{Reference Fumal, Jobe and Pepin134–Reference Hu, Shang, Liu and Yuan136} It should be acknowledged, however, that in general their efficacy is likely to be lower than in those with AIE in the setting of extracellular antibodies, which are considered to be pathogenic, as opposed to intracellular antibodies, which are not considered to be pathogenic.

Second-line therapy in patients with neoplasm

Second-line therapy can be undertaken in concert with treatment of the underlying neoplasm. Guidance from oncology in cases where surgery is not expected to be curative is essential and delaying second-line therapy may or may not be necessary based on oncology input.

Choice of second-line therapy should be guided by antibody type as above.

Treatment failure after second-line therapy

Defining treatment failure after second-line therapy is challenging due to variations in onset of action of second-line therapies depending on dose/regimen used as well as the co-administration of PLEX/IVIg/repeat IV steroids. As a result, we suggest the determination of treatment failure after second-line therapy be done in consultation with a specialist in autoimmune neurology.

Third-line/alternative immunotherapy

There are a number of third-line and experimental immunotherapies (tocilizumab, bortezomib) which have been employed in cases refractory to second-line immunotherapy. ^{Reference Abboud, Probasco and Irani5,Reference Shin, Lee and Park128,Reference Lee, Lee and Moon137,Reference Turnbull, Siegel, Becker, Stephens, Lopez-Chiriboga and Freeman138} Involvement of a specialist in autoimmune neurology is essential to discuss additional therapeutic options in cases when a patient fails second-line therapy.

Special considerations for refractory NMDAR-IgG AIE

NMDAR antibody encephalitis is associated with underlying teratoma in 11–36% of cases. ^{Reference Titulaer, McCracken and Gabilondo9,Reference Bost, Chanson and Picard139–Reference Xu, Lu and Huang141} Microteratomas (occult or imaging negative) are rarely reported ^{Reference Boeck, Logemann and Krauß142,Reference Cirkel, Cirkel and Royl143} and true incidence is unknown.

In cases of severe refractory NMDAR antibody encephalitis, evidence for bilateral “blind” oophorectomy for removal of potential occult ovarian teratoma is limited. ^{Reference Iyengar, Hébert and Climans144} The existing evidence is predominantly based on case series/case reports, many of which were published before options for third-line treatment were available. Given the permanent effects on fertility/iatrogenic menopause and the fact that most patients with refractory NMDAR antibody encephalitis would be unable to provide informed consent due to their underlying disease, such procedures should only be undertaken as a last resort and with informed consent through a surrogate decision maker.

Accordingly, we strongly recommend these procedures be limited to patients with persistently life-threatening disease who fail second (rituximab and/or cyclophosphamide), and third-line therapy (tocilizumab or bortezomib) and who have had multiple imaging modalities to investigate for teratoma including endovaginal ultrasound and MRI pelvis as well as CSF confirmation of NMDAR antibodies. Given the poor level of evidence for “blind” oophorectomy and permanent effects on fertility some authors involved in this guideline strongly believe that this procedure should not be offered.

Teratomas are rare in patients younger than 18 or older than 45 ^{Reference Titulaer, McCracken and Gabilondo9} and blind oophorectomy is not recommended outside of this age range.

Involvement of specialists in autoimmune neurology and gynecology is essential in such cases given permanent impacts to fertility and iatrogenic menopause.

Immune checkpoint inhibitor related autoimmune encephalitis

A detailed discussion of treatment of ICI related AIE is outside the scope of this article and oncology guidelines exist elsewhere. ^{Reference Schneider, Naidoo and Santomasso145} In general, first-line treatment remain similar to other forms of AIE. Rituximab may be used in antibody positive cases although concerns exist around persistent immunosuppression in active malignancy, and oncology input is essential. ^{Reference Schneider, Naidoo and Santomasso145,Reference Haanen, Carbonnel and Robert146}

Seizure management

The International League Against Epilepsy nomenclature of “acute symptomatic seizures secondary to autoimmune encephalitis” (ASSAE) and “autoimmune-associated epilepsy” are helpful conceptually when classifying patients AIE with seizures, ^{Reference Steriade, Britton and Dale158} although the modified term “autoimmune encephalitis-associated epilepsy” (AEAE) serves to make explicit the link between AIE and seizures in all cases. ^{Reference Budhram and Burneo159} The proportion of patients with AIE who develop seizures has been estimated at 50–70%. ^{Reference Budhram and Burneo159} AIE associated with autoantibodies targeting extracellular antigens – especially GABA_AR, ^{Reference Petit-Pedrol, Armangue and Peng59} GABA_BR, ^{Reference van Coevorden-Hameete, de Bruijn and de Graaff161} NMDAR, ^{Reference Dalmau, Gleichman and Hughes162} and LGI1 ^{Reference Gadoth, Pittock and Dubey46} – are a more commonly encountered cause of seizures than AIE associated with those targeting intracellular antigens. GAD65 antibody is a notable exception. ^{Reference Budhram and Burneo159,Reference Budhram, Sechi and Flanagan77,Reference Steriade, Gillinder and Rickett163} There is a paucity of data on the overall incidence and risk factors associated with the development of AEAE. Relatively small retrospective cohort studies of patients treated for seizures related to AIE reported that, at a two-year follow-up, 2-42% of patients developed AEAE. ^{Reference Casciato, Morano and Fattouch164–Reference Steriade, Moosa, Hantus, Prayson, Alexopoulos and Rae-Grant166}

We recommend a combined initial treatment approach with anti-seizure medications (ASMs) and immunosuppression for the treatment of seizures related to AIE. Some patients (in particular those with antibodies against extracellular targets) may respond to immunosuppression alone, indicating ASSAE. ^{Reference Dubey, Britton and McKeon118,Reference Abboud, Probasco and Irani148,Reference de Bruijn, van Sonderen and van Coevorden-Hameete165,Reference Feyissa, Lamb and Pittock167} Patients who receive immunosuppression, and who receive it quickly, have a higher chance of achieving seizure freedom. ^{Reference Dubey, Britton and McKeon118,Reference de Bruijn, van Sonderen and van Coevorden-Hameete165,Reference Feyissa, Lamb and Pittock167,Reference Bozzetti, Rossini and Ferrari168} Sodium-channel blockers are possibly more effective for the treatment of seizures related to AIE, especially in patients with LGI1 antibody encephalitis, but are associated with higher risks of side effects. ^{Reference Gong, Chen and Liu140,Reference Feyissa, Lamb and Pittock167,Reference Cabezudo-García, Mena-Vázquez, Villagrán-García and Serrano-Castro169} Enzyme-inducing ASMs should be avoided in patients treated with chemotherapy due to risk of lowering drug levels. ^{Reference Gonzalez Castro and Milligan170} In the patient who is no longer in the acute phase of AIE and has become seizure-free a gradual wean from ASMs can be considered after discussing the importance of vigilance for signs or symptoms of seizure recurrence that would merit prompt clinical re-assessment. ^{Reference Abboud, Probasco and Irani148,Reference de Bruijn, van Sonderen and van Coevorden-Hameete165}

Neuropsychiatric symptoms

Neuropsychiatric symptoms are common in AIE and are expected to improve with first- and second-line treatments. ^{Reference Hébert, Day, Steriade, Wennberg and Tang-Wai8,Reference Loughan, Allen, Perna and Malkin171} To our knowledge, there are no randomized controlled trials investigating treatment of neuropsychiatric symptoms in patients with AIE; management therefore relies on observational studies and expert opinion, usually with conventional psychiatric treatments. ^{Reference Oldham172} Psychiatric symptoms in patients with AIE improve less frequently with symptomatic treatments compared to patients with psychiatric conditions, and patients are more likely to experience side effects. ^{Reference Marinas, Matveychuk, McCombe, Tymchuk, Dursun and Baker173–Reference Zhang, Cai and Zheng175}

Regarding acute agitation and psychosis, special caution is required with antipsychotic medications. ^{Reference Marinas, Matveychuk, McCombe, Tymchuk, Dursun and Baker173,Reference Dale, Gorman and Lim176} Patients face a higher risk of drug induced movement disorders, worsening agitation, rhabdomyolysis, and neuroleptic malignant syndrome, especially with medications with high affinity for the D2 dopamine receptor, such as haloperidol. ^{Reference Liu, Tian and Guo177} If required, atypical antipsychotics are preferred. ^{Reference Oldham172,Reference Coffey and Cooper178} Options are detailed in Table 5 including strategies to target sleep disturbance.

Table 5: Recommended agents for neuropsychiatric symptoms

Electroconvulsive therapy has been used for severe psychosis and catatonia in AIE ^{Reference Loughan, Allen, Perna and Malkin171–Reference Marinas, Matveychuk, McCombe, Tymchuk, Dursun and Baker173,Reference Dhossche and Wachtel179} with some success, especially when the catatonia is treatment resistant, and immunotherapy has failed. ^{Reference Tanguturi, Cundiff and Fuchs180} Catatonia can be treated with benzodiazepines, with caution in patients with hypoventilation and decreased arousal. ^{Reference Coffey and Cooper178}

Even with treatment, recovery can take many years and is often incomplete with cognitive, and neuropsychiatric symptoms persisting in 50% of patients. ^{Reference Zhang, Cai and Zheng175,Reference Huang, Wang, Feng, Wang and Li181–Reference Seery, Butzkueven, O’Brien and Monif183} Behavioral agitation, emotional lability, posttraumatic stress disorder symptoms, anxiety, and depression are common. Selective serotonin reuptake inhibitors, pindolol and valproic acid have been used for agitation and disinhibition ^{Reference Marinas, Matveychuk, McCombe, Tymchuk, Dursun and Baker173} in those with chronic symptoms.

Persistent cognitive symptoms include deficits in attention, memory, executive function, and processing speed, and are associated with poor psychosocial outcomes. ^{Reference Marinas, Matveychuk, McCombe, Tymchuk, Dursun and Baker173} Supervision, rehabilitation, and family-centered care are important. ^{Reference Loughan, Allen, Perna and Malkin171,Reference Scheer and John184} Non-pharmacological management borrowed from the TBI literature includes reducing overstimulation, clustering care, minimizing physical restraint as well as assessing the environment and sleep wake cycles. ^{Reference Vova185}

Rehabilitation/prognosis for recovery

Prognostication in AIE is challenging due to lack of long-term evidence, and variability in outcomes based on specific AIE subtype and associated cancer. ^{Reference Deng, Qiu, Liu, Wu, Lei and Lu186,Reference Broadley, Seneviratne and Beech187} For example, 75–85% of patients with NMDAR antibody encephalitis achieve good functional outcomes (modified Rankin Score of 0-2) ^{Reference Gong, Chen and Liu140,Reference Dalmau, Gleichman and Hughes162} compared with only 43% of patients with GAD65 antibody encephalitis. ^{Reference Budhram, Sechi and Flanagan77} Approximately half of patients with AIE may continue to experience cognitive impairment after the acute onset of disease. ^{Reference Hébert, Day, Steriade, Wennberg and Tang-Wai8} Earlier initiation of treatment has been associated with improved long-term functional, ^{Reference Deng, Qiu, Liu, Wu, Lei and Lu186,Reference Broadley, Seneviratne and Beech187} cognitive ^{Reference Hébert, Day, Steriade, Wennberg and Tang-Wai8} and psychiatric ^{Reference Nguyen, Yang, Goyal, Irani and Graves188} outcomes across a variety of AIE subtypes. Once patients are no longer in the acute phase of AIE, they should be routinely screened for residual neuropsychiatric impairment and be offered a combination of psychiatric care ^{Reference Blum, Tomlinson, Jetté, Kwon, Easton and Yeshokumar189} and neuropsychological rehabilitation. ^{Reference Abboud, Probasco and Irani148,Reference Blum, Tomlinson, Jetté, Kwon, Easton and Yeshokumar189}

Relapse in autoimmune encephalitis

Relapse in AIE is not uncommon, and characteristics demonstrate significant variability between AIE subtypes. ^{Reference Titulaer, McCracken and Gabilondo9,Reference van Sonderen, Ariño and Petit-Pedrol21,Reference van Sonderen, Thijs and Coenders22,Reference Ropper, Dalmau and Graus42,Reference Shin, Lee and Park128,Reference Smith, Dubey, Liebo, Flanagan and Britton190} Retrospective observational studies of NMDAR, LGI1, and CASPR2 antibody encephalitis indicate relapse rates ranging from 10 to 41% and symptoms may be identical to the index episode (e.g., LGI1 antibody encephalitis re-presenting with seizures), similar but milder (e.g., NMDAR antibody encephalitis presenting with behavioral symptoms after previous severe ICU admission) or a different distinct core clinical syndrome (e.g., CASPR2 antibody encephalitis presenting with cerebellar ataxia after previous seizure presentation). Reliable data is however hard to obtain as rates and characteristics are confounded by small cohort sizes with limited follow-up duration and a lack of consensus on the definition of relapse in autoimmune encephalitis. Resultantly, relapse attributes are not fully known for all AIE subtypes (e.g., the relapse rate for antibody-negative AIE is unknown).

For the purposes of these guidelines, we define relapse in AIE as a clear objective worsening of symptoms after initial improvement or plateau, preferably supported by evidence of inflammation on MRI or in CSF, and usually occurring at a minimum of 2–3 months from original presentation. PET scan findings are often non-specific ^{Reference Graus and Dalmau102} and isolated PET abnormalities should be interpreted with caution if they are the only objective marker of relapse. Antibody titers are typically not helpful in defining relapse due to their imperfect correlation with disease activity. ^{Reference Gresa-Arribas, Titulaer and Torrents47,Reference Di Giacomo, Deleo and Pastori191}

For acute treatment of relapse, we suggest following the first-line and/or second-line treatment algorithm as appropriate unless a patient is known to have failed a particular therapy in the past. Long-term immunosuppression may be appropriate and consultation with a specialist in autoimmune neurology is recommended in cases of relapse both to confirm relapse and guide long-term treatment.

Long-term immunosuppression

Data to guide long-term immunosuppression decisions in AIE are lacking. Given uncertainty and variability of relapse rates, long-term therapy is not routinely recommended for all patients at first presentation of AIE, although we acknowledge substantial variability in practice even among experts. One to two years of immunotherapy may be justified for patients with more severe disease. In relapsing forms of AIE, three or more years of maintenance immunotherapy has been proposed. ^{Reference Abboud, Probasco and Irani148} We recommend involvement of a specialist in autoimmune neurology when considering initiation of long-term immunosuppression.

Latent infection screening

Although risks of recrudescent infection vary by pathogen and immunotherapy, early broad testing to expedite appropriate prophylactic or definitive treatment is advised. Screening should routinely include testing for VZV, HIV, Hepatitis B/C, and tuberculosis. When abnormalities are found, immunotherapy can often be started soon after implementing appropriate monitoring/treatment (Table 6). When active or latent infections are identified, consultation with an infectious disease specialist is advised.

Table 6: Overview of screening response by pathogen

†Rituximab and cyclophosphamide can be used to treat the immunological complications of Hepatitis C but a risk of severe hepatitis exists beyond risk posed by hepatitis C.

‡VZV testing is to ascertain if sufficient immunity remains to reduce the risk of emergent singles.

Ab: Antibody, Ag: Antigen, cART: combined antiretroviral therapy, CD: Cell differentiation factor, CNS: Central Nervous System, CSF: Cerebrospinal Fluid, DNA: Deoxyribonucleic Acid, esp: especially, HB: Hepatitis B, HC: Hepatitis C, HIV: Human Immunodeficiency Virus, HBs Ag: Hepatitis B surface antigen, HBc Ab: Hepatitis B core antibody, ID: Infectious disease.

Vaccination

Age appropriate and risk context appropriate vaccinations (e.g., meningococcal vaccination in military recruits) should be considered for all patients embarking on immunosuppression (Table 7). ^{Reference Mejia and Nutman200,203–206} Although delaying immunotherapy 4–6 weeks after a vaccine series is completed is typically recommended, this is not practical in acute autoimmune encephalitis presentations. Often vaccination may be delayed pending treatment conclusion (Table 8) ^{207,Reference Rubin, Levin and Ljungman208} or, if therapy is prolonged, be taken during treatment with non-live vaccines, recognizing efficacy of vaccines may be reduced.

Table 7: Vaccination recommendations – adapted from CDC 2022 and Canadian guidelines (Rubin et al. 2014; Government of Canada 2021; CDC 2022)

HPV: Human Papilloma Virus; InV: Inactivated vaccine; LV: Live vaccine; MMR: Measles, Mumps, Rubella; PCV15: 15-valent pneumococcal conjugate vaccine; PCV20: 20-valent pneumococcal conjugate vaccine; Pneumococcal polysaccharide vaccine (PPSV23); RIV4: recombinant influenza vaccine; LAIV4: live attenuated influenza vaccine; RV: Recombinant Vaccine; RZV: Recombinant Varicella Vaccine; SRPCV: Subunit, recombinant, polysaccharide, and conjugate vaccines; TD: Tetanus, Diphtheria; Tdap: Tetanus, Diphtheria, Acellular Pertussis, Tox: Toxoid; IIV4: inactivated influenza vaccine.

Table 8: Recommendations for treatment related vaccine delays

Reproductive health/pregnancy risks

First-line immunotherapies are generally safe in pregnancy. ²¹¹ Steroids can be administered, including high dose IV pulse therapy. ^{Reference Bandoli, Palmsten, Forbess Smith and Chambers212–Reference Hviid and Mølgaard-Nielsen214} When used, decision makers should be advised of mild increased infection risk including those more relevant to pregnancy (e.g., endometritis). Treatment with IVIg and PLEX are considered safe in pregnancy although additional hypercoagulability should be discussed. ^{Reference Achiron, Kishner and Dolev215,Reference Watson, Katz and Bowes216}

Some second-line agents can also be used safely. Guidelines for rituximab vary with regards to advisable delay of 6 to 12 months between treatment and conception/delivery. ^{Reference Chakravarty, Murray, Kelman and Farmer217} Reports have demonstrated safety when treatment occurs within 6 months of delivery even if an ability to cross the placenta routinely leads to transient CD19/20 depletion in the infant. ^{Reference Das, Damotte and Gelfand218,Reference Smith, Hellwig, Fink, Lyell, Piehl and Langer-Gould219} Cyclophosphamide should never be used in pregnancy and its use in men or women requires discussion of infertility risk, and potential gamete harvesting/storage (acuity permitting) should be explored. Involvement of a specialist in autoimmune neurology in such cases is strongly recommended.

Pneumocystis jirovecii (PJP) prophylaxis

PJP prophylaxis is crucial but precise application is debated for those on immunotherapy. Guidelines exist for those with rheumatologic/hematologic conditions but relevance in neurological illness is unknown.

Trimethoprim-sulfamethoxazole (TMP-SMX) is the gold standard for PJP prophylaxis. If a prior minor reaction to TMP-SMX is reported (i.e., no respiratory or mucosal membrane involvement or fever) then desensitization and subsequent use is preferable to therapy with atovaquone, dapsone, or pentamidine. ^{Reference Cooley, Dendle and Wolf220,Reference Urbancic, Ierino, Phillips, Mount, Mahony and Trubiano221}

Overall, we recommend providing PJP prophylaxis in the following scenarios ^{Reference Cooley, Dendle and Wolf220,Reference Park, Curtis and Jun222} .

1. 1. All patients on high dose corticosteroids (20 mg of prednisone for 2 weeks or more, alone or in combination with other medications).

2. 2. Overlapping rituximab/cyclophosphamide therapy until the resolution of either CD19/20 cell counts (if rituximab is discontinued) or lymphopenia/neutropenia (if cyclophosphamide is discontinued).

3. 3. Patients on 5 mg prednisone equivalent or more in combination with cyclophosphamide.

Complications of corticosteroid use

Corticosteroids infrequently cause lasting side effects during brief courses (e.g., gastric ulcer/bleeding, avascular necrosis), but the consequences of long-term use can undermine therapeutic efforts. Common issues include worsening metabolic syndrome (i.e., increased body mass index, hypertension, hyperlipidemia, glucose intolerance), osteoporosis, infections, and impaired wound healing/skin thinning. ^{Reference Miloslavsky, Naden and Bijlsma223–Reference Kersey and Broadway225} Neuropsychiatric manifestations are perhaps the most problematic as they can overlap with encephalitic symptomatology. If side effects are extreme, more rapid tapering/discontinuation may be required however several mitigation strategies exist (Table 9). ^{Reference Liu, Ahmet and Ward226,Reference Liu, Reidy, Saatee and Collard227}

Table 9: Steroid side effect mitigation ^{Reference Liu, Ahmet and Ward226}

BMD: Bone mineral density; IOP: Intraocular pressure; FRAX: Fracture risk assessment tool; MRI: Magnetic resonance imaging.