Autoimmune Epilepsy

Some seizures and epilepsies are caused by the immune system mistakenly targeting structures within the brain. In these disorders, inflammation, antibodies or immune cells can disturb neuronal signalling and create seizures that may look like more familiar forms of focal or generalised epilepsy.

This is one of the most important potentially treatable causes of new-onset, rapidly worsening or unusually drug-resistant seizures. Recognising an immune cause matters because antiseizure medication alone may not address the underlying disease. In selected patients, treatment directed at the immune process can be central to seizure control and neurological recovery.

But the term autoimmune epilepsy needs careful use. Many people with autoimmune encephalitis have seizures during the acute inflammatory illness and never develop chronic epilepsy. Others are left with an enduring tendency to seizures after the encephalitis has settled. Still others present mainly with recurrent seizures and much less obvious encephalopathy. These groups overlap clinically, but they are not identical.

Where autoimmune epilepsy fits in modern epilepsy classification

The International League Against Epilepsy recognises immune as one of the major aetiological categories of epilepsy. An immune aetiology means there is evidence that immune-mediated inflammation is directly contributing to the person’s epilepsy.

This sits alongside structural, genetic, infectious, metabolic and unknown causes. In practice, more than one category can apply. For example, an immune-mediated encephalitis can injure the temporal lobe and leave structural hippocampal damage that later contributes to chronic epilepsy.

That overlap is one reason autoimmune epilepsy belongs within this site’s Structural & Acquired Epilepsies collection while also linking closely to diagnosis, neuroimmunology and drug-resistant epilepsy.

Autoimmune encephalitis and autoimmune-associated epilepsy are not the same thing

The ILAE Autoimmunity and Inflammation Taskforce has proposed an important distinction between acute symptomatic seizures secondary to autoimmune encephalitis and autoimmune-associated epilepsy.

During active autoimmune encephalitis, the brain is inflamed. Seizures occurring as part of that acute inflammatory state may be considered acute symptomatic seizures. If the immune disease is successfully treated and the inflammation settles, the tendency to seizures may disappear.

Autoimmune-associated epilepsy describes a more enduring predisposition to recurrent unprovoked seizures associated with an autoimmune process. This can occur because inflammation remains active, because the immune disorder itself creates a persistent epileptogenic state, or because previous inflammation has left structural and network changes that continue to generate seizures.

This distinction matters because persistent seizures after encephalitis do not automatically prove that inflammation is still active, and escalating immunotherapy in someone with a stable post-encephalitic epilepsy is not the same decision as treating active autoimmune encephalitis.

How the immune system can generate seizures

The mechanisms differ according to the immune target.

Some antibodies bind to proteins on the neuronal surface or at synapses. These proteins help regulate communication between neurons. Antibodies can alter receptor function, remove receptors from the cell surface or interfere with proteins that stabilise neuronal signalling. The resulting network disturbance can rapidly increase excitability and produce seizures.

Other antibodies are directed against intracellular proteins. In these disorders, the antibody may be more of a marker of the immune response than the main cause of neuronal injury. T cells can play a larger role, and the risk of permanent neuronal damage and later epilepsy may be greater.

These biological differences help explain why some antibody-associated disorders respond dramatically to immunotherapy while others are more difficult to treat.

Which antibodies are associated with seizures?

There is no single autoimmune epilepsy antibody. Several neural antibodies are associated with seizures, and each has a different clinical pattern.

LGI1

Leucine-rich glioma-inactivated 1, or LGI1, antibodies are strongly associated with limbic encephalitis. Seizures can be frequent and may occur before obvious memory or behavioural change.

A particularly distinctive event is the faciobrachial dystonic seizure. These are typically extremely brief, frequent movements involving the face and one arm, sometimes occurring dozens or even hundreds of times per day. Recognising them can be crucial because they may precede the full cognitive syndrome, and early immunotherapy may reduce progression.

LGI1 encephalitis often affects memory because the medial temporal structures, including the hippocampus, are commonly involved. Hyponatraemia can also occur.

NMDA receptor

Anti-NMDA receptor encephalitis can produce seizures alongside psychiatric symptoms, behavioural change, movement disorders, autonomic instability, reduced consciousness and cognitive dysfunction.

The syndrome can affect children and adults. In some patients, seizures are prominent early in the illness, but chronic epilepsy is not inevitable once the encephalitis is controlled.

GABA-B receptor

GABA-B receptor antibodies are strongly associated with seizures and limbic encephalitis. Seizures can be severe or refractory, including status epilepticus.

This antibody can also be associated with cancer, particularly small-cell lung cancer, making tumour investigation an important part of the assessment when the clinical context supports it.

CASPR2

CASPR2 antibodies can be associated with limbic encephalitis, seizures, cognitive change, peripheral nerve hyperexcitability and autonomic symptoms. The clinical presentation is variable and antibody results need to be interpreted within the whole syndrome.

AMPA receptor

AMPA receptor antibodies can cause limbic encephalitis with memory impairment, confusion and seizures. Some cases are paraneoplastic.

GAD65

High-titre GAD65 antibodies are associated with several neurological autoimmune syndromes, including a form of chronic temporal lobe epilepsy. This pattern differs from many neuronal-surface antibody syndromes.

GAD65-associated epilepsy can be persistently drug-resistant and may respond less predictably to immunotherapy, particularly when epilepsy has been established for a long time. Because low-level GAD antibodies can also occur in other autoimmune conditions, including type 1 diabetes, the antibody result cannot be interpreted in isolation.

Intracellular paraneoplastic antibodies

Antibodies such as Hu, Ma2 and CRMP5 may indicate a T-cell-driven paraneoplastic neurological syndrome. Seizures can occur, but these conditions are biologically different from classic neuronal-surface antibody encephalitides and may carry a higher risk of persistent neurological injury.

Not every positive antibody test proves autoimmune epilepsy

This is a critical diagnostic principle.

Neural antibody tests vary in specificity, and some antibodies are clinically meaningful only when detected by the right method, at the right concentration and in the right biological sample. A low-level serum result without a matching clinical syndrome may be incidental or misleading.

For some antibodies, cerebrospinal fluid testing is particularly important. For others, serum can be more sensitive. Many specialist centres therefore use paired serum and CSF testing when autoimmune encephalitis or autoimmune-associated epilepsy is strongly suspected.

The diagnosis should integrate the clinical picture, EEG, MRI, CSF findings and antibody result rather than treating a laboratory result as a stand-alone diagnosis.

Autoimmune epilepsy can be antibody-negative

Failure to identify a known neural antibody does not completely exclude an immune-mediated neurological disorder.

New antibodies continue to be characterised, currently available panels do not cover every immune target, and some inflammatory disorders may be predominantly T-cell mediated rather than antibody mediated.

For this reason, modern diagnostic criteria for autoimmune encephalitis were deliberately designed so that clinicians can recognise probable disease using the clinical syndrome and conventional investigations without waiting for a positive antibody result.

At the same time, antibody-negative epilepsy should not automatically be labelled autoimmune. Infectious, genetic, metabolic, toxic, structural and other causes must still be considered.

When should an autoimmune cause be suspected?

There is no single feature that identifies autoimmune epilepsy, but certain combinations raise suspicion.

Important clues include:

  • new-onset frequent focal seizures without a clear previous epilepsy history;
  • rapid escalation in seizure frequency;
  • new drug-resistant seizures developing over days, weeks or months rather than years;
  • seizures accompanied by subacute memory loss;
  • new psychiatric or behavioural change;
  • confusion or altered consciousness that is disproportionate to individual seizures;
  • new movement disorder or abnormal involuntary movements;
  • autonomic instability;
  • faciobrachial dystonic seizures;
  • very frequent temporal-lobe seizures;
  • new-onset status epilepticus or NORSE;
  • inflammatory CSF findings;
  • medial temporal abnormalities on MRI;
  • a personal or family history of autoimmunity in the right clinical context;
  • recent or concurrent cancer when a paraneoplastic syndrome is possible.

None of these features alone proves autoimmunity. Their value lies in the pattern they create together.

Why a sudden change in an established epilepsy deserves attention

Autoimmune disease is most often discussed in new-onset seizures, but immune-mediated neurological disease can also appear in someone who already has epilepsy.

A major unexplained change in seizure type, frequency or treatment response may require reassessment rather than assuming the person’s underlying epilepsy has simply worsened. Clinicians may consider new structural lesions, infection, medication interactions, metabolic disturbance, sleep disruption and other explanations as well as immune disease.

Faciobrachial dystonic seizures are a major diagnostic clue

Faciobrachial dystonic seizures are among the best-known seizure signatures in autoimmune neurology. They are most strongly associated with LGI1-antibody disease.

They are often extremely brief and can easily be mistaken for a movement disorder, startle, muscle spasm or behavioural event. Scalp EEG may not show a clear ictal pattern during every event, which makes recognition of the clinical semiology especially important.

The frequency of these attacks can be extraordinary. Their identification matters because they can occur before severe memory impairment develops and because immune treatment, rather than antiseizure medication alone, is often central to control.

What EEG can show

EEG findings in autoimmune encephalitis and autoimmune-associated epilepsy are variable. They may include focal epileptiform discharges, focal slowing, diffuse slowing, electrographic seizures or status epilepticus.

Some disorders have suggestive EEG patterns, but there is no universal ‘autoimmune EEG’. A normal routine EEG therefore does not exclude an immune cause.

Prolonged video-EEG can be particularly useful when events are frequent, when non-convulsive seizures are suspected or when altered behaviour could represent either seizure activity or encephalopathy.

See also: Global Access to EEG.

What MRI can show

MRI may show swelling or high signal within the medial temporal lobes in limbic encephalitis. The hippocampus and amygdala can be particularly affected.

However, MRI can also be normal, especially early in the illness or in some antibody syndromes. A normal MRI therefore does not rule out autoimmune encephalitis.

Over time, severe or prolonged inflammation can leave hippocampal atrophy or sclerosis, providing one mechanism by which an acute immune illness can evolve into chronic structural epilepsy.

See also: Global Access to MRI and Neuroimaging and Hippocampal Sclerosis.

Why cerebrospinal fluid matters

Lumbar puncture can provide evidence of central nervous system inflammation. Clinicians may examine white blood cell count, protein, oligoclonal bands, IgG index, infectious studies and neural antibodies.

CSF results vary widely. Some patients with proven autoimmune encephalitis have only mild inflammatory abnormalities, and some have apparently normal routine CSF indices.

CSF is also essential because infections such as herpes simplex encephalitis can mimic autoimmune encephalitis and require urgent antiviral treatment. In some cases, autoimmune encephalitis can even occur after viral encephalitis, further complicating the differential diagnosis.

Why infection must be excluded

Seizures, confusion, fever, behavioural change and temporal-lobe MRI abnormalities can occur in both infectious and autoimmune encephalitis.

Because untreated infectious encephalitis can be rapidly dangerous, clinicians must consider infection urgently. The diagnostic process may therefore involve parallel testing and, where appropriate, empirical antimicrobial treatment while autoimmune investigations continue.

Autoimmune treatment suppresses immune activity, so overlooking an active infection can have serious consequences.

Could it be cancer-related?

Some autoimmune neurological syndromes are paraneoplastic: the immune system responds to a tumour and cross-reacts with proteins in the nervous system.

The strength of the cancer association differs between antibodies. Some neural antibodies have a strong link to particular malignancies, while others are commonly found without cancer.

When a paraneoplastic syndrome is suspected, clinicians may investigate for an underlying tumour using age-, sex- and antibody-appropriate imaging and screening. In some syndromes, treating the tumour is an important part of controlling the immune process.

What are APE2 and other autoimmune epilepsy scores?

Because broad neural-antibody testing is expensive and can generate difficult-to-interpret results, clinical scoring systems have been developed to identify patients in whom testing is more likely to be useful.

The APE2 score — Antibody Prevalence in Epilepsy and Encephalopathy — incorporates clinical features associated with neural autoantibodies. Other tools include ACES for focal epilepsy and newer scores designed for particular diagnostic settings.

These tools can support clinical judgement, but they do not replace specialist assessment. Different scores were developed in different populations, and a low score cannot rule out every immune-mediated syndrome.

NORSE and FIRES

New-onset refractory status epilepticus (NORSE) describes a clinical presentation in which a person without active epilepsy or another clear relevant neurological disorder develops refractory status epilepticus without an immediately obvious structural, toxic or metabolic cause.

Autoimmune encephalitis is one of the important causes clinicians investigate in NORSE. Some cases remain cryptogenic despite extensive testing.

Febrile infection-related epilepsy syndrome (FIRES) is a subcategory of NORSE in which a febrile illness precedes refractory status epilepticus. FIRES is most recognised in children but can occur in adults.

These conditions are medical emergencies and are considerably broader than autoimmune epilepsy alone.

How are seizures treated during autoimmune encephalitis?

Acute seizures are treated with appropriate antiseizure medication, but in a true immune-mediated encephalitis this may be only one part of treatment.

If inflammation is driving the seizures, controlling that inflammation can be crucial. First-line immunotherapies commonly include high-dose corticosteroids, intravenous immunoglobulin and plasma exchange, selected according to the syndrome and individual clinical circumstances.

When disease remains active or severe, specialist teams may use second-line therapies such as rituximab or cyclophosphamide. Other immune treatments can be used in particular refractory syndromes.

These are specialist treatments with substantial risks. The choice, timing and combination depend on the suspected immune mechanism, antibody, infection risk, cancer status, comorbidities and severity of neurological illness.

Why antiseizure medication alone may fail

In active autoimmune encephalitis, the brain can remain intensely hyperexcitable because the underlying inflammatory process continues. Adding more antiseizure medication may reduce seizures but not remove the driver.

This is particularly evident in some neuronal-surface antibody syndromes where seizures improve dramatically once effective immunotherapy begins.

That does not make antiseizure medicines unimportant. They remain essential for immediate seizure control and status epilepticus. The point is that immune-mediated seizures may require treatment of both the seizure and its cause.

Does everyone need long-term antiseizure medication?

No. The long-term seizure risk differs substantially between autoimmune syndromes.

Many patients with seizures during neuronal-surface antibody encephalitis become seizure-free after the acute disease is controlled and do not develop enduring epilepsy. In these circumstances, antiseizure medication may eventually be reduced under specialist supervision.

Other patients have a much greater risk of chronic epilepsy, particularly when there has been severe structural injury, prolonged status epilepticus, persistent epileptiform EEG abnormalities or an immune disorder more strongly associated with enduring neuronal damage.

There is therefore no single duration of treatment that applies to everyone.

Why GAD65-associated epilepsy is different

GAD65-associated epilepsy illustrates why autoimmune epilepsies cannot all be treated as one disease.

Unlike some rapidly reversible neuronal-surface antibody syndromes, GAD65-associated temporal lobe epilepsy can become a chronic drug-resistant epilepsy. Immunotherapy responses are less consistently dramatic, especially after epilepsy has been present for a long period.

Some patients develop mesial temporal sclerosis. The relationship between active immune inflammation and fixed epileptogenic structural change can become difficult to separate.

This is one reason early recognition of potentially immune-mediated epilepsy matters: there may be a window before chronic network and structural changes become established.

Can epilepsy surgery still be considered?

In selected patients with chronic autoimmune-associated drug-resistant focal epilepsy, epilepsy surgery may be discussed. The decision is complex because an immune disease can affect networks beyond the apparent MRI lesion or seizure-onset zone.

Surgical outcomes vary by immune syndrome. A stable, well-localised structural epilepsy after a past inflammatory insult is biologically different from ongoing active multifocal inflammation.

Presurgical assessment must therefore consider whether the immune process is active, whether seizures are truly focal and whether the proposed epileptogenic zone can be treated safely.

See also: Global Access to Epilepsy Surgery.

Can autoimmune epilepsy relapse?

Yes. Some autoimmune encephalitis syndromes can relapse after an initial recovery.

A return of seizures can sometimes be part of relapse, but a seizure recurrence does not automatically mean the immune disease has returned. People can also have seizures because of medication changes, sleep deprivation, intercurrent illness or established post-encephalitic epilepsy.

Clinicians may therefore repeat MRI, EEG, CSF or antibody investigations when relapse is suspected rather than assuming that every seizure represents renewed inflammation.

What happens to memory and cognition?

Autoimmune disorders involving the limbic system frequently affect memory because the hippocampus and connected temporal networks are central to forming and retrieving memories.

Patients may have difficulty retaining new information, become disoriented, repeat questions or lose chunks of recent memory. Attention, executive function, language, mood and behaviour can also be affected depending on the syndrome.

Cognitive recovery can continue after seizures and inflammation are controlled, but some people have lasting impairment, particularly after severe or prolonged disease.

Psychiatric symptoms do not mean the seizures are psychological

Autoimmune encephalitis can present with dramatic psychiatric or behavioural symptoms: agitation, paranoia, hallucinations, anxiety, mood disturbance, disinhibition or catatonia.

This can lead to initial psychiatric assessment before the neurological syndrome becomes clear. The presence of psychiatric symptoms does not make seizures non-epileptic, and the presence of genuine epileptic seizures does not exclude coexisting psychiatric illness or functional seizures.

Rapid behavioural change combined with new seizures, cognitive decline, abnormal movements or autonomic instability is a neurological red flag that deserves careful medical assessment.

Children can present differently

Autoimmune encephalitis occurs in children as well as adults. Younger patients may show behavioural regression, loss of language or developmental skills, sleep disturbance, movement disorders or new school difficulties alongside seizures.

Age also influences the probability of particular antibodies and associated tumours. Paediatric assessment therefore requires clinicians experienced in childhood epilepsy and neuroimmunology.

Why diagnosis can be delayed

Several features make autoimmune epilepsy easy to miss. Seizures may initially look like ordinary focal epilepsy. MRI can be normal. Routine EEG can be non-specific. Psychiatric symptoms can dominate. Antibody results can take time, and some patients have no identified antibody at all.

Conversely, indiscriminate antibody testing can create false-positive or clinically irrelevant results.

The safest approach is therefore neither ‘test everyone’ nor ‘only test obvious encephalitis’. It is targeted investigation based on clinical pattern, supported by specialist interpretation.

Global access is unequal

Diagnosis of autoimmune-associated epilepsy can require MRI, prolonged EEG, lumbar puncture, specialised antibody assays, cancer screening and expert neuroimmunology interpretation. These resources are not evenly distributed worldwide.

In health systems where basic EEG and MRI are already difficult to obtain, autoimmune epilepsy can remain completely unrecognised. Antibody panels may need to be sent internationally and can be prohibitively expensive.

Because some immune-mediated seizure disorders are highly treatable when recognised early, this diagnostic inequality can directly affect long-term neurological outcome.

Important questions clinicians are trying to answer

When autoimmune epilepsy is suspected, the key questions are not simply ‘Is an antibody positive?’ They include:

  • Is there evidence of active brain inflammation?
  • Are the seizures acute symptomatic events during encephalitis or has enduring epilepsy developed?
  • Does the antibody fit the patient’s neurological syndrome?
  • Is the antibody likely pathogenic, a disease marker or incidental?
  • Could infection explain the presentation?
  • Could an underlying tumour be driving the immune response?
  • Is ongoing seizure activity due to active inflammation, structural injury or both?
  • Would immunotherapy still be expected to help?
  • What long-term seizure and cognitive risks remain after the acute illness?

The central lesson

Autoimmune epilepsy shows why identifying the cause of seizures matters.

Two people can have apparently similar focal seizures but require fundamentally different treatment. One may have a stable structural epilepsy managed primarily with antiseizure medication or surgery. Another may have an active immune disorder in which rapid diagnosis and immunotherapy can alter the course of the disease.

At the same time, not every seizure during autoimmune encephalitis becomes lifelong epilepsy, and not every positive neural antibody proves that immunity is the cause.

The diagnosis depends on putting the entire neurological picture together.

What this page does — and does not — cover

This page focuses on seizures and epilepsy caused by immune-mediated neurological disease. It does not replace a full guide to autoimmune encephalitis, individual antibody syndromes, NORSE, cancer-associated neurological disease or the technical interpretation of neural-antibody assays. Those areas require their own detailed discussion.

Sources and further reading

International League Against Epilepsy — Classification of the Epilepsies and immune aetiology
Steriade C et al. Acute symptomatic seizures secondary to autoimmune encephalitis and autoimmune-associated epilepsy: conceptual definitions. Epilepsia. 2020.
Smith KM et al. Autoimmune-associated seizure disorders. Epileptic Disorders. 2024.
Graus F et al. A clinical approach to diagnosis of autoimmune encephalitis. Lancet Neurology. 2016.
Geis C et al. Autoimmune seizures and epilepsy. Journal of Clinical Investigation. 2019.
Husari KS, Dubey D. Autoimmune Epilepsy. Neurotherapeutics. 2019.
Seizures, Epilepsy, and NORSE Secondary to Autoimmune Encephalitis: A Practical Guide for Clinicians. 2023.
Autoimmune Encephalitis. Continuum. 2024.
Seizures Associated with Autoimmune Disorders — Current Treatment Approaches. 2025.
Autoimmune-associated epilepsy: definitions and diagnosis — a narrative review. 2026.

Evidence checked: September 2026.

Page information · Published:

If revised after publication, the update date appears below:

Dates reflect website publication and editing, not independent medical review. See our sources and editorial policy, or report a correction.