Clinical approach to the diagnosis of autoimmune encephalitis in the pediatric patient

Methodology

At the 2014 Autoimmune Encephalitis Alliance (AEalliance.org) conference in North Carolina, the Autoimmune Encephalitis International Working Group was formed and initiated discussions around developing diagnostic criteria for AE. A subcommittee of pediatric neurologists and rheumatologists identified that adult-focused criteria may not apply well to children. As a result, this subcommittee collaborated through conference calls and email correspondence to consider the pediatric-specific approach to AE. The subcommittee reviewed the literature on relevant AE studies and sought additional input from other experts. The first author (T.C.) developed a draft based on the preceding discussions that was subsequently reviewed and modified by all authors.

Existing diagnostic criteria for AE

The International Encephalitis Consortium 2013 diagnostic criteria for encephalitis of presumed infectious or autoimmune etiology require patients to have altered mental status lasting more than 24 hours with no alternative cause identified.²⁸ Confirmation of this diagnosis requires at least 3 minor criteria, including fever within 72 hours of presentation; new onset focal neurologic findings; CSF leukocytosis; acute new neuroimaging abnormality suggestive of encephalitis; or EEG abnormalities consistent with encephalitis.²⁸ These criteria do not differentiate autoimmune from infectious encephalitis.

More recently, an international group developed diagnostic criteria for early diagnosis of AE in adults, which require (1) subacute onset over less than 3 months of working memory deficits, altered mental status, or psychiatric symptoms; (2) at least one of the following: new focal CNS findings, seizures not explained by a preexisting disorder, CSF pleocytosis, and/or MRI features suggestive of encephalitis; and (3) reasonable exclusion of alternative causes.²⁷ Specific neurologic syndromes were given criteria, including limbic encephalitis, anti-NMDAR encephalitis, and autoantibody-negative AE.²⁷

These AE criteria required modification to be applied to children. For example, deficits in working memory are challenging to identify in younger children. Also, children are less likely to present with a well-defined neurologic syndrome and, even in anti-NMDAR encephalitis, the sequence of symptom development may differ from adults.^5–7 Furthermore, the differential diagnosis for a child presenting with temporal lobe seizures and cognitive slowing is broad, whereas this presentation in adults suggests limbic encephalitis or acquired temporal pathology.

Clinical features distinguishing adults and children with AE

Typically, children with AE are previously healthy and present with rapid onset of neuropsychiatric symptoms. Prodromal symptoms including fever occur in over 50% of patients.^2,4–6 Between disease onset and initiation of therapy, symptoms typically persist over time. This distinguishes AE from pediatric acute-onset neuropsychiatric syndrome (PANS), where patients often experience a relapsing-remitting course with rapid progression to maximum symptom severity and rapid return to previous function over hours or days, sometimes without therapy.

Neurologic manifestations of AE include altered level of consciousness, confusion, disturbed sleep, movement disorders and seizures. Seizures are the most common feature in AE and may be the predominant manifestation.^{4–7,10–21} Seizures may be focal or generalized and are often multifocal.^{4–7,10–21} Over one third of patients with AE have abnormal movements, such as ataxia, chorea, dystonia, myoclonus, or tremor.^4–7,13,15 Both seizures and movement disorders can be highly refractory to standard treatments in children with AE.^10,14,16,24 Some degree of cognitive impairment is seen in the overwhelming majority of AE patients and is considered a cardinal symptom.^{4,5,13,14,16,19,21} As such, a diagnosis of AE would be highly questionable in patients with documented normal cognition, again differentiating AE from PANS where cognition is often preserved. Assessing memory deficits in young children may be challenging; however, developmental regression, language loss or speech impairments may be presenting features of pediatric AE.^5–7,29

Behavioral changes, such as repetitive or stereotypical behaviors, irritability, hyperactivity, hypersexuality, insomnia and anger outbursts, are common in pediatric AE.^4–7 Psychiatric symptoms may range from mood swings and mild personality changes to fulminant psychosis and occur in over 50% of AE patients.^4–7 New-onset psychosis in children younger than 13 years is uncommon and considered a red flag for an underlying medical, rather than primary psychiatric, condition. It is critical to assess for cognitive changes, seizures, movement abnormalities, or other neurologic symptoms in children with acute psychiatric symptoms, as these symptoms are suggestive of AE.

Children with AE likely differ from adults in their clinical presentations due to evolution of neuronal circuits, neuroreceptor densities and myelination during normal development. Children with AE are more likely to present with multifocal neuropsychiatric symptoms, rather than isolated clinical syndromes. For example, children with GAD65 antibodies may not present with the classic stiff-person syndrome or cerebellar degeneration seen in adults.^11,12,22 Children with anti–NMDAR-associated encephalitis are more likely to present with movement abnormalities, agitation, insomnia, seizures, speech deficits, ataxia, and/or hemiparesis, whereas memory deficits, psychiatric manifestations, and central hypoventilation are more common in adults with the same antibody.^5–7 Pediatric AE is less associated with tumors compared with adults.^4–7

Diagnostic evaluation of children and teenagers with suspected AE

Although no single investigation is diagnostic of pediatric AE, the presence of a suggestive clinical phenotype and supportive paraclinical testing is essential to diagnose an underlying inflammatory process and to exclude alternative diagnoses. Initial investigations to be considered for any child with suspected AE are listed in table 1, although diagnostic workup should be tailored to the individual.

Blood tests are helpful to assess for systemic inflammatory changes, autoantibodies associated with systemic autoimmune diseases, vitamin B12 deficiency, markers of infection, elevated lactate due to metabolic conditions, and recreational drug use. Erythrocyte sedimentation rate, C-reactive protein, leukocyte counts, and platelet counts may be normal in children with AE.^1,4–21

CSF pleocytosis and/or elevated protein levels may be seen at diagnosis or during disease course, but are not uniformly present.^1,4–21 Recommended tests to assess for infectious encephalitis were based on population-based studies in California and England (table 1).^30,31 However, workup for infectious etiologies varies depending on the season and region where the patient lives or has traveled. A recent report suggests that anti-NMDAR encephalitis may be more common in children than any specific infectious encephalitis, further highlighting the importance of considering AE when evaluating for infectious encephalitis.³² CSF neopterin is a useful but not rapidly accessible biomarker that is frequently elevated in anti-NMDAR encephalitis and other encephalitides, but normal in PANS.³³ There is evidence that AE (particularly anti-NMDAR encephalitis) may be triggered by herpes simplex virus encephalitis and Japanese encephalitis.³⁴

All patients should have a brain MRI with and without gadolinium. Over half of patients with AE will have a normal brain and spine MRI at diagnosis.^{4–7,16,21,22} Inflammatory lesions (high signal on T2 and fluid-attenuated inversion recovery sequences) may develop over time, and cerebral atrophy may occur months later.^4,6,7,15 MRI lesions are most likely to be present in those with antibodies to MOG or the gamma-aminobutyric acid-A receptor (GABA_AR).^9,14,15 Neuroimaging findings are not limited to the temporal lobe or cortex.^1,5–21 A normal MRI lessens suspicion for CNS vasculitis, demyelinating diseases, infections, and malignancies.¹ In contrast, restriction on diffusion-weighted imaging reduces the likelihood of pediatric AE and should prompt consideration of other etiologies, such as infection-associated encephalopathies and vasculitis.¹ Small retrospective adult AE studies have proposed that functional PET and SPECT studies may demonstrate brain dysfunction, but experience is limited in pediatric AE.^35,36

A normal EEG is unusual in children with AE during active disease, although prolonged EEG may be needed for improved sensitivity. Therefore, focal or generalized seizures, epileptiform discharges, and encephalopathic changes, such as diffuse or focal slowing, may help to distinguish AE from primary psychiatric disorders or PANS. Adults with AE are more likely to have EEG changes predominantly involving the temporal lobes, whereas EEG findings in children may be more generalized.^{4–7,14–21} Specific EEG features, such as the “delta brush” pattern and extreme spindles, have been linked to anti-NMDAR encephalitis, but sensitivity is low.^6,22,23

Neurocognitive testing may identify deficits in memory, attention, problem solving, language, and processing speed, particularly in younger children. A change in neurocognitive function supports a diagnosis of pediatric AE and may differentiate these patients from those with primary psychiatric disorders. However, interpretation of neurocognitive testing at diagnosis should be undertaken with caution, as there is often no premorbid testing for comparison.

Other diagnostic tests may be considered. Most children with AE do not require brain biopsy. However, a _targeted brain biopsy of MRI abnormalities may be needed when the diagnosis remains uncertain after initial workup. The diagnostic yield of brain biopsy is higher in pediatric patients than in adults.³⁷

Antibody testing and interpretation in children and teenagers with suspected AE

Antibodies associated with pediatric AE are listed in tables 2 and 3. Each antibody is associated with characteristic symptoms, seizure types, and other clinical findings. However, there is significant overlap between the different disorders and so testing a panel of neural autoantibodies is recommended for any child with suspected AE. The most common autoantibodies identified in children _target NMDAR, MOG, GAD65, and GABA_AR. Given the rarity of other autoantibodies, further testing should be considered only if antibodies to these _targets are negative and suspicion of AE persists (table 3).

Antibody testing should be performed in both CSF and serum to avoid false-negative and false-positive results. For example, testing for NMDAR antibodies typically has higher sensitivity in CSF compared with serum, with up to 15% of patients having negative serum results.^5–7 In contrast, MOG autoantibodies have higher sensitivity in serum.⁹

Interpretation of antibody test results should carefully consider the child's clinical presentation, especially when more than 1 antibody is identified. For example, GAD65 antibodies tend to be associated with personal or familial autoimmunity and low titers, such as those seen in type 1 diabetes mellitus, are not neurologically relevant.²² The presence of more than 1 antibody in some patients with AE has been recognized and may be associated with overlapping syndromes. Antibody specificity is also important when interpreting antibody test results. For instance, only IgG isotype antibodies to the GluN1 subunit of the NMDAR on a cell-based assay are specifically associated with AE.^5,38

In adults with AE, most antibodies to the voltage-gated potassium channel complex (VGKCC) do not bind to the channel, but to proteins in the complex, particularly leucine-rich glioma-inactivated protein 1 (LGI1) and contactin-associated protein-like 2 (Caspr2).³⁹ In children, VGKCC antibodies rarely _target LGI1 or Caspr2.^40,41 It has been argued that VGKCC antibodies without specific binding to LGl1 or Caspr2 have limited clinical significance.⁴⁰

Proposed classification criteria and algorithm for diagnosis of pediatric AE

We modified the criteria for adult AE and propose provisional classification criteria for possible pediatric AE, probable antibody-negative pediatric AE, and definite antibody-positive pediatric AE in table 4.²⁷ A diagnostic algorithm is also provided in figure. The provisional criteria and algorithm should be assessed prospectively in future cohorts.

A diagnosis of pediatric AE should be considered in previously healthy children who present with acute or subacute (less than 3 months) onset of new focal or diffuse neurologic deficits, cognitive difficulties, developmental regression, movement abnormalities, psychiatric symptoms, and/or seizures. Although children with preexisting developmental delay or chronic behavior/psychiatric abnormalities may develop AE, alternative diagnoses, such as genetic, metabolic, or neurodegenerative etiologies, should be considered in these patients.

Children with a clinical presentation suggestive of AE should have serum and CSF examined for neuronal antibodies, undergo paraclinical testing for neuroinflammation, and have disease mimics excluded (tables 1 and 4). EEG is not included as paraclinical evidence of neuroinflammation because EEG cannot differentiate AE from other encephalopathies. However, EEG encephalopathic features are allowable as an alternative for clinical features of encephalopathy. If a patient fulfills criteria for possible pediatric AE (table 4) and is functionally impaired, therapy may be started while awaiting the results of antibody and other testing, given the importance of early treatment to improve outcomes.^4,25,26 If a patient with possible AE subsequently does not have positive antibodies or paraclinical testing for neuroinflammation, a diagnosis of AE is not supported. For these children, careful further consideration of the differential diagnosis is warranted, and additional immune therapy should only be undertaken with caution (table 5, figure).

Children may have AE caused by antibodies that have not yet been identified and may meet criteria for probable antibody-negative pediatric AE (table 4). These patients will have 1 or more positive paraclinical tests for neuroinflammation, but negative antibody testing. Children who meet the criteria for definite antibody-positive pediatric AE will have positive antibody testing. If CSF antibodies are present (e.g., NMDAR and GAD65), no other paraclinical evidence of neuroinflammation is required for a diagnosis of definite AE (table 4). If only serum antibodies are present, 1 or more paraclinical tests of neuroinflammation must be abnormal. There should be caution in diagnosing AE when only serum antibodies (particularly NMDAR, GABA_AR, and glycine receptor) are found in the absence of paraclinical evidence of neuroinflammation.

The proposed pediatric AE criteria differ from the adult criteria in several ways (table 4, table e-5, links.lww.com/nxi/A184).²⁷ First, the pediatric criteria include both acute and subacute time frames for symptom onset, reflecting the range in disease course observed in children. Adult AE criteria were developed for several well-defined syndromes (i.e., limbic encephalitis, acute disseminated encephalomyelitis [ADEM], and anti-NMDAR encephalitis) and the associated algorithm focuses on whether patients meet criteria for these syndromes.²⁷ In contrast, many pediatric patients with AE do not present with a well-defined syndrome and so the pediatric criteria were devised to capture the breadth of clinical and paraclinical findings reported in children. Similarly, the pediatric AE algorithm (figure) does not focus on syndrome identification, but is intended to guide a clinician in assessing clinical features and in paraclinical and antibody testing, so as to determine whether an AE diagnosis is appropriate. The adult AE criteria group clinical and paraclinical markers together, whereas the pediatric criteria distinguish clinical evidence of neurologic dysfunction from paraclinical evidence of neuroinflammation.

Patients with definite AE may benefit from continued or advanced immunosuppressive therapy, although specific protocols are not yet validated. Identification of an antibody associated with AE may facilitate counseling regarding expected course and outcomes. Timing of clinical responses to immunotherapy in children with AE may vary from immediate to months after starting.^5–7,24,42 Therefore, using response to therapy as confirmatory support for a diagnosis of AE may be misleading.

Approach to clinically recognizable syndromes

Approach to probable antibody-negative pediatric AE

Children with a clinical phenotype of AE and paraclinical findings of neuroinflammation, but negative testing for neural antibodies, may meet criteria for probable antibody-negative pediatric AE (table 4). It is well recognized that not all neural autoantibodies have been identified. Having CSF and serum testing in a research laboratory may identify patients who have antibodies against neural cell surface antigens of yet unknown identity and who may respond to immunotherapy.

Probable antibody-negative AE is one of the most challenging clinical scenarios. It is appropriate that a child presenting with new onset encephalopathy, neuropsychiatric features, and changes in function be investigated for possible AE. However, the differential diagnosis in children is arguably broader than in adults, and so it is important to ensure that other diagnoses have been excluded before giving an AE diagnosis. Pathologic entities that often cause diagnostic difficulty are cortical dysplasias and genetic epilepsies presenting with fever-provoked symptomatic focal seizures, infection-provoked encephalopathy and PANS. In these syndromes, CSF pleocytosis or oligoclonal bands are usually absent, and MRI is either normal or demonstrates alternative pathology. Therefore, critical examination of paraclinical tests for evidence of CNS inflammation is mandatory to avoid unnecessary immune suppression. A diagnosis of probable antibody-negative pediatric AE should also be reassessed in children with atypical features.

Differential diagnosis of AE

The spectrum of inflammatory brain diseases in children has rapidly expanded as new diseases and new etiologies for existing conditions have been described. The underlying pathogenic mechanisms that lead to CNS inflammation may involve vessel wall inflammation, demyelination, or an immune response directed against neurons and supporting structures.^1,3 Inflammation may also occur secondary to infection, malignancy, or a systemic inflammatory disease. Diagnosing pediatric AE is especially challenging because of the clinical overlap between conditions in the differential diagnosis (table 5) and the clinical heterogeneity within patients having the same disease.

Discussion

Proposed pediatric AE criteria are intended to address differences in clinical presentations, paraclinical findings, and autoantibody profiles between children and adults. The accompanying algorithm aims to guide diagnostic workup and facilitate earlier initiation of therapy.

Study funding

There was no external funding for this manuscript.

Disclosure

The authors have no conflicts of interest relevant to this article to disclose, with the exception of Dr. Dalmau being the editor of Neurology: Neuroimmunology & Neuroinflammation. Go to Neurology.org/NN for full disclosures.