Clinical manifestations

Presentation and course

• Faciobrachial dystonic seizures can precede the onset of cognitive impairment.

The vast majority of patients with LGI1 antibodies have CNS involvement with limbic encephalitis symptoms (21; 30; 58; 04; 55). Cognitive impairment and seizures are among the most common presentations (53). Faciobrachial dystonic seizures are thought to be pathognomonic for LGI1 antibodies and are present in about 60% of patients with the antibodies (22; 24; 26; 53a; 55). Faciobrachial dystonic seizures are adult-onset, brief (usually 1 to 3 seconds) seizures involving posturing of the hemiface and ipsilateral arm, with a dystonic semiology (24; 26). Events are stereotyped and frequent (8 to 200 per day, median 50 per day) (26; 56). Faciobrachial dystonic seizures are often triggered by kinesigenic or emotive stimuli and can be preceded by a sensory aura, associated with ictal automatisms, and have postictal speech arrest, fear, or agitation following the motor event (26). Interestingly, auditory auras are not associated with faciobrachial dystonic seizures, although they are the prominent features of patients with mutations in LGI1 (39). Faciobrachial dystonic seizures can alternate sides in an individual patient. Associated falls or dropping items are common findings likely due to involvement of the leg and arm, respectively.

Importantly, faciobrachial dystonic seizures can precede the onset of cognitive impairment or, less commonly, can occur concurrently with the cognitive dysfunction (26; 55). The cognitive impairment is frequently florid but can be insidious in onset, which increases the risk of delayed recognition, underdiagnosis, or misdiagnosis (26).

Other seizure semiologies described in around 70% of LGI1 antibody-positive patients include dyscognitive or autonomic focal seizures and motor or gelastic seizures (55; 03). The autonomic involvement can lead to bradycardia and pacemaker-necessitating asystole (38). The most commonly reported sensory events were thermal sensations or body-shuddering, and motor events were most frequently identified as automatisms or vocalizations (03). Generalized tonic-clonic seizures can occur in up to 63% of patients (55; 03). Finally, in one report, paroxysmal dizzy spells also appeared to be characteristic of patients with LGI1 antibodies (17).

The clinical manifestations of LGI1- and CASPR2-antibody encephalitis share significant similarities, including clinical presentations with limbic encephalitis, seizures, sleep disturbances, and neuromyotonia, and investigation findings including serum hyponatremia and medial temporal changes on MRI (07). Therefore, these 2 antibody-mediated syndromes perhaps exist on a spectrum. Faciobrachial dystonic seizures, however, appear exclusive to patients with LGI1 antibodies. Of interest, close to 50 patients who have double-positive antibodies to both LGI1 and CASPR2 have been reported in the literature and lend further weight to the suggestion that these disorders may share an underlying pathogenesis (07). Over 40% of these double-positive patients are found to have a paraneoplastic association, with a third being thymomas (17). Rare phenotypes, including cerebellar ataxia, chorea, parkinsonism, and dystonia, have also been identified in some LGI1- and CASPR2-antibody patients (50; 47; 51).

Sleep disturbances have been identified in LGI1-antibody encephalitis with disruption of both REM and NREM sleep and hyperkinetic movements during the active phase of the disease, and evolution into REM sleep behavior disorder or periodic limb movement syndrome in some patients during the recovery phase (32).

Clinical vignette

A 92-year-old man presented with a 5-month history of right facial grimacing and simultaneous right-arm dystonic posturing lasting seconds. He was previously independent with activities of daily living and at the time of symptom onset had normal cognition. The episodes of right hemifacial and ipsilateral arm twitching could occur up to 8 times a day, and he had no recollection of these episodes.

Within a month following onset, he developed deficits in memory and was noted to be more withdrawn and confused but was still living independently. By 3 months post-onset, he was getting lost within his own home and having difficulty recognizing and naming objects. He had recurrent falls, some resulting in injuries such as wrist fractures. Seizures were suspected, and levetiracetam 250 mg twice daily was commenced without effect.

He was reviewed in a tertiary hospital clinic at 4 months post-onset. On examination, there were no localizing neurologic signs; however, cognitive examination revealed impaired orientation, attention, memory, and visuospatial deficits. The Addensbrooke Cognitive Examination Revised (ACER) score was 64/100. He had mild hyponatremia at 132 mmol/L. Prolonged EEG monitoring recorded stereotyped episodes with rhythmical theta activity in left temporal area, followed by focal sharp waves in the mid temporal area, consistent with left hemispheric focal seizure activity. Serum LGI1 antibodies were positive. MRI demonstrated chronic small vessel disease and normal imaging of the hippocampus. A 3-day course of 1 g intravenous methylprednisolone per day completely abolished seizures. The patient was subsequently commenced on 40 mg daily of prednisone, weaned over a 4-week period.

ACER ratings 1 week after steroids were commenced demonstrated a 10-point improvement to 74/100, and within 2 months after treatment commencement this further increased to 84/100. However, 5 months after steroid cessation (10 months after symptom onset), he presented with a recurrence of seizures that once again ceased rapidly on recommencement of 40 mg daily prednisone. This was weaned over a 7-month period, and the patient maintained remission even after cessation of steroids and discontinuation of levetiracetam. ACER 1 year after symptom onset was 93/100, and the patient had returned to his baseline function and cognition.

Biological basis

Etiology and pathogenesis

	• LGI1 and CASPR2 antibodies are primarily of the IgG4 subclass.
	• Mice infused with patient-derived LGI1 antibodies show severe memory deficits that improve after cessation of the infusion.

LGI1 is a secreted protein that is thought to play an important role in transmission between presynaptic and postsynaptic complexes (21; 30). LGI1 interacts with, and acts as a ligand for, 2 synaptic proteins of the disintegrin and metalloproteinase domain (ADAM) family, specifically ADAM22 postsynaptically and ADAM23 presynaptically (15).

LGI1 and CASPR2 antibodies are primarily of the IgG4 subclass. If they are bispecific and express 2 distinct antigen-binding sites, as has been reported for some IgG4 antibodies, they may be less efficient at crosslinking. Furthermore, IgG4 antibodies cannot activate complement (05; 51; 07). Instead, their mode of action is primarily thought to be through direct modulation or functional interference of protein-protein interactions or associated proteins.

In vitro and in vivo studies using patient derived monoclonals and unfractionated patient IgGs have shown that LGI1 antibodies from limbic encephalitis patients inhibit the interaction between LGI1 and ADAM22/23 and reversibly reduce synaptic AMPA receptor clusters in rat hippocampal neurons (39). In addition, both IgG4 and IgG1 LGI1 antibodies may cause internalization of the LGI1-ADAM22 complex (49; 43). Taken together, these findings suggest that LGI1-ADAM22 modulation may reduce and disrupt AMPA receptor function (39; 43), which, in turn, may result in increased excitability and CNS manifestations including seizures. Additionally, genetic LGI1 deficiency has been shown to result in down-regulation of Kv1.1 and Kv1.2 expression, which provides another epileptogenic mechanism (45). The LGI1-ADAM22 complex associates with postsynaptic density protein 95 family membrane-associated granulate kinases (MAGUKs) to organize transsynaptic protein networks such as NMDA and AMPA receptors and Kv1 channels. Mouse models with disruption of LGI1-ADAM22-MAGUK interactions exhibited lethal epilepsy (16).

Moreover, mice infused with patient-derived IgG showed severe memory deficits in novel object recognition, which improved after cessation of the infusion (40). In one study, LGI1 antibodies in sera and CSF were shown to bind to both domains of LGI1 (leucine-rich repeat or LRR; and epitempin repeat or EPTP) (43). However, patient-derived monoclonal antibodies demonstrated mutually exclusive binding to one of these 2 domains, with varied binding strengths and characteristics. LRR-binding monoclonal antibodies demonstrated binding to rodent brain sections, induced internalization of the LGI1-ADAM22/23 complex in HEK293T cells and live hippocampal neurons, and in an in vivo model; those LRR-binding monoclonal antibodies with higher binding strength resulted in memory impairment (43). Some of these effects were mimicked with EPTP binding monoclonals.

Only a few studies evaluate the histopathology of patients with LGI1-antibody encephalitis. One study describes CD45+ and CD8+ T cells in proximity to the hippocampus, with only mild neuronal loss, as well as in the meninges of the lateral temporal lobe and intraparenchymal vessels (44). Bien and colleagues describe immunohistopathological evidence of both immunoglobulin deposition and complement-mediated neuronal destruction in 4 patients with elevated VGKC-complex antibodies (with only 1 identified as having LGI1 antibodies) (06). This may be only part of the answer as it is now understood that most LGI1 antibodies are of the noncomplement-fixing IgG4 subtype. However, patients with faciobrachial dystonic seizures and cognitive impairment have been shown to have higher proportions of minority LGI1-IgG1 antibodies when compared to patients without cognitive impairment, suggesting complement activation may yet have a role in the irreversible cognitive sequelae of LGI antibodies (49).

Genetic data show that patients with LGI1 antibodies have human leukocyte antigen DRB1*07:01 through DQB1*02:02 haplotypes, present in 90% of LGI1-antibody limbic encephalitis patients from East Asian and Caucasian cohorts (28; 52; 07). The identification of human leucocyte antigen susceptibilities provides the impetus to study the relevance of T cell interactions in an autoantibody-driven disease process. It further raises the possibility of an underlying genetic predisposition activated by a second hit such as an infection, environmental trigger, or tumor to trigger the clinical manifestations of the condition (07).

Persistent cognitive impairment has been described in a few LGI1-antibody patients with hippocampal atrophy and elevated flortaucipir retention on PET scanning, suggesting tau neuropathology may associate with cognitive impairment in stable LGI1 patients (10). Hence, some neurodegeneration may coexist with an autoimmune pathology in some patients with permanent deficits following this autoimmune encephalitis.

Epidemiology

LGI1-antibody encephalitis most commonly affects middle-aged and older adults with a broad age range excluding children; the median age of onset is around 63 years of age (21; 54). There is a 2 to 1 male predisposition (21; 54), and the annual United Kingdom incidence is around 1 million per year (36).

Differential diagnosis

A diagnosis of possible autoimmune encephalitis applies to patients with a subacute onset of memory deficits, altered mental state, or psychiatric symptoms plus at least one of the following: new focal CNS findings; otherwise unexplained seizures; CSF pleocytosis; MRI features suggestive of encephalitis; and reasonable exclusion of alternative causes. These presentations should prompt appropriate testing for antibodies to neuronal surface antigens (19). Clinical features suggestive of specific autoimmune encephalitic syndromes, such as the stereotyped movement disorders in NMDAR-antibody encephalitis or the faciobrachial dystonic seizures associated with LGI1 antibodies, may help direct testing. In the absence of these clinical pointers, screening investigations for the range of available neuronal surface antibody syndromes may be reasonably performed.

Autoimmune encephalitis associated with antibodies targeting onconeural intracellular antigens, such as cytoplasmic or nuclear proteins (eg, Ma, Hu, and Ri), are often paraneoplastic with a progressive course, typically despite immunotherapies, and with a typically guarded prognosis (06; 42). Treatment should be primarily directed to the underlying neoplasia, and neuronal cell death is primarily attributed to T-cell-mediated neurotoxicity, suggesting that these antibodies are not primarily pathogenic but, rather, biomarkers of associated tumors (06). This antibody panel should be performed if there are concerns of a paraneoplastic neurologic presentation with rapidly progressive neurologic deficits, detected tumor, or lack of response to immunotherapy.

Other CNS inflammatory syndromes, including stiff person syndrome, progressive encephalomyelitis with rigidity and myoclonus, Bickerstaff encephalomyelitis, and Hashimoto encephalopathy, should also be considered in the presence of specific neurologic features associated with these syndromes, and testing for antibodies to glutamic acid decarboxylase, the glycine receptor, GQ1b, and thyroid antibodies can be pursued as deemed clinically relevant (19). Morvan syndrome and Rasmussen encephalitis generally have a more chronic course but may be considered in the differential diagnosis of atypical presentations of encephalopathy (19).

An important differential diagnosis for autoimmune encephalitis in a patient presenting with acute or subacute CNS involvement and cognitive impairment is Creutzfeldt-Jacob disease (18). The faciobrachial dystonic seizures associated with LGI1 antibodies may be misdiagnosed as myoclonus associated with Creutzfeldt-Jacob disease. Although the prevalence is low, evaluation for LGI1 antibodies and potentially other neuronal surface antibodies should be considered in patients with suspected Creutzfeldt-Jacob disease. A fascinating case report highlights a patient with faciobrachial dystonic seizures, limbic encephalitis, and LGI1 antibodies plus consistent radiology who demonstrated a complete response to immunotherapy and who subsequently experienced a rapidly progressive dementia with postmortem confirmation of Creutzfeldt-Jacob disease. This case draws attention to a rare example of these 2 conditions occurring in the same individual (27).

Diagnostic workup

	• Electroencephalography during faciobrachial dystonic seizures is often normal.
	• Brain MRI demonstrates increased T2 signal and swelling in the medial temporal lobe in 50% to 74% of patients.

Hyponatremia is identified in up to 60% of patients and provides a helpful paraclinical clue that is easily available in all clinical settings (30). It may be secondary to LGI1 antibodies binding the hypothalamic ADH-secreting neurons (25).

Electroencephalography during faciobrachial dystonic seizures is often normal, and ictal localization has been noncontributory in most patients. In a minority of faciobrachial dystonic seizures, there is evidence of preceding focal rhythmic delta wave activity and focal discharges or slowing immediately following the motor event (26; 03). EEG may additionally show multifocal interictal epileptiform discharges during other focal seizure types or focal slowing interictally; extending beyond the expected temporal localization to also encompass discharges arising from the frontal and parietal regions (55; 03). Multiple seizure semiologies, subclinical seizures, and interictal epileptiform discharges appear to be characteristic of LGI1-antibody encephalitis (03).

MRI may demonstrate changes including increased T2 signal and swelling in the medial temporal lobe in 50% to 74% of patients (21; 30; 54; 55). Signal changes in the basal ganglia are almost uniquely present in patients with faciobrachial dystonic seizures, typically contralateral to the side of limb and hemiface involvement (26; 14; 17).

CSF is normal in up to 75% of patients, and inflammatory changes, such as lymphocytic pleocytosis and elevated protein levels, are not present in the majority of patients (55). Only a minority have intrathecal oligoclonal bands. Serum testing for LGI1 antibodies additionally appears to be more sensitive and specific when compared with CSF (55; 51; 17). Indeed, patients with CSF negative for LGI1-antibodies were clinically indistinguishable from those with both serum and CSF LGI1-antibodies (37).

Management

	• Faciobrachial dystonic seizures are often refractory to multiple antiepileptic drugs.
	• Steroids are the main treatment. Although seizure response may be rapid within days to weeks, improvement in cognition is typically slower and may take months.

Overall, a third of patients with faciobrachial dystonic seizures experience these in isolation, often with normal sodium levels, CSF analysis, and MRI. Those who go on to develop limbic encephalitis with associated cognitive impairment may demonstrate other seizure semiologies, including more typical mesial temporal lobe seizures as well as hyponatremia and hippocampal T2 hyperintensity on imaging (24; 26; 49). Hence, again, clinical diagnosis is at the forefront of management.

Faciobrachial dystonic seizures are often refractory to multiple antiepileptic drugs, despite use of a median of 3 agents (range 1 to 10). Also, 29% to 50% of patients have serious cutaneous drug reactions to antiepileptic drugs (24; 26; 49). Only 10% of patients with faciobrachial dystonic seizures responded to antiepileptic drugs alone, and these patients tended to have a milder phenotype (49).

In contrast, the majority of patients with faciobrachial dystonic seizures respond well, often within weeks, to the initiation of immunotherapy. Many cases responded even within 2 weeks. This immunotherapy is used in addition to antiepileptic drugs or in isolation (24; 26). A study evaluating 103 patients with faciobrachial dystonic seizures identified that 51% of patients responded with cessation of faciobrachial dystonic seizures to immunotherapy within 1 month of initiation (26; 49). This rapid response suggests that steroids may have antibody-independent effects on seizure control (26). The benefit of immunotherapy over anticonvulsant therapy in seizure cessation has been reproduced in other cohorts (11). In the first randomized controlled trial within the field, intravenous immunoglobulin was superior to placebo in reducing seizure frequency by 50% in patients with LGI1-antibody encephalitis (12). However, this study was limited by a relatively small number of patients.

Although seizure response may be rapid within days to weeks, improvement in cognition is typically slower and may take months to years (51). The need for ongoing immunotherapy may be met with oral corticosteroids, alone or in combination with steroid-sparing agents such as azathioprine or mycophenolate (51). However, the value of such slow-acting agents is unclear. Response to second-line therapies, including cyclophosphamide or rituximab, has only been evaluated in small numbers of patients; although effective in a few patients, the impact on others has been, perhaps surprisingly, limited (23; 54; 05). This likely reflects the fact that outcome measures in these patients are confounded by delay in initiation of second-line immunotherapy and that selection bias dictates that more severely affected patients receive second-line therapy. Further prospective studies with larger numbers of patients on these agents are required to provide empirical evidence for their utility and efficacy. Although seizures are common in the early stage of LGI1-antbody encephalitis, once the disease course is stabilized with immunotherapy, patients appear to be unlikely to go onto develop epilepsy. In seizure free patients, withdrawal of anticonvulsants in the medium to long-term does not appear to result in ongoing seizures (11; 20).

Immunotherapy in the context of antibody seropositivity in unselected patients with new-onset seizures was shown to benefit patients with some subtle features of encephalitis, such as attentional deficits and affective disturbances, providing a method to select patients presenting to first-seizure clinics who will benefit from immunotherapy (35).

Although in some patients, clinical course appears to be broadly reflected in antibody levels, this correlation is imperfect and should not be used to guide therapy (19). Indeed, serum LGI1 antibodies may remain positive for prolonged durations despite clinical remission and recovery (02). Taken together, this suggests that management should focus on the clinical status of the patient, hence, the value in better understanding clinical trajectories and outcomes.

Outcomes

There are infrequent reports of patients with LGI1-antibody limbic encephalitis where the diagnosis was made retrospectively, 36 and 53 months after symptom onset, and patients did not receive immunotherapy (48). Both patients in this well-described report achieved seizure freedom and a monophasic natural history with recovery without relapses. However, in other cases, multiple relapses were noted over the course of a decade (01). Although the possibility of spontaneous resolution of seizures suggests a more benign course, it is of note that follow-up imaging revealed global brain atrophy and hippocampal sclerosis in one patient and either mild or moderate cognitive impairment, particularly in domains of memory and verbal fluency, in both patients (48).

If faciobrachial dystonic seizures are treated adequately and proactively with immunotherapy, there is a significant reduction in the development of cognitive impairment (26; 49), whereas, patients with ongoing seizures show a poorer cognitive outcome. In a cohort of 103 patients with faciobrachial dystonic seizures, 56% of patients who had faciobrachial dystonic seizures as their first presentation developed cognitive impairment after 90 days of active seizures; compare this to only one patient who developed cognitive impairment after cessation of faciobrachial dystonic seizures (49). This suggests that the effective termination of faciobrachial dystonic seizures may prevent the onset of cognitive impairment, which typically defines the encephalitis.

Relapses of seizures and cognitive impairment have been documented in patients with LGI1 antibodies, particularly in the context of steroid tapering and withdrawal, with amelioration of symptoms with reinstitution of steroids (26). Patients with faciobrachial dystonic seizures, irrespective of the presence of hippocampal changes in the acute phase, may go on to have smaller whole-brain volumes compared to controls (26). Longer duration of faciobrachial dystonic seizures has been additionally shown to be associated with a reduction in pallidal volume (13). Even in patients who have received immunotherapy, cohort studies have identified mesial temporal sclerosis in up to 44% (55).

Although up to 80% of patients who receive immunotherapy have a favorable response with a rapid benefit in terms of seizure control, improvement in cognition may be slower and incomplete, with many patients reporting lasting cognitive deficits (55; 17). Retrograde amnesia for the period of the acute illness is a common sequela (55). Verbal and visuospatial memory deficits, with associated hippocampal atrophy and interruption of hippocampal microstructural integrity, is common (13). In addition, apathy is noted in many patients (55; 13). Delayed initiation of immunotherapy was associated with worse verbal and visuospatial episodic memory performance (13). Patients with increased clinical severity, often reflected in failure to respond to second line therapy, had more severe hippocampal structural damage, which in turn was associated with worse memory performance (13).

Overall, expedited immunotherapy and normal cognition at baseline are the best predictors of reduced disability at 24 months (49). Those without an initial response to immunotherapy and those who experienced relapses had a worse prognosis (02). Telehealth follow up in 37 patients identified that premorbid cognitive reserve and MRI with bilateral hippocampal hyperintensity were predictors for long-term cognitive impairment (46). Furthermore, high seizure frequency may be associated with more limited recovery rates (03). In one study, relapses have been reported in up to 35% of patients and may implicate utility of chronic immunotherapy in a subgroup of patients. The impact of impaired cognition, but more so fatigue and mood disorders, may result in insidious long-term residual deficits in patients, impairing the employment status and quality of life of patients with LGI1-antibody encephalitis (08).

Taken together, these findings provide a compelling argument that early diagnosis and the rapid institution of appropriate immunotherapy in LGI1-antibody encephalitis is essential to prevent irreversible long-term sequelae. Refinement of outcome measures is critical as functional scales such as the mRS may underestimate the impact of factors such as cognitive impairment and fatigue on quality of life (46; 08).