Editorial <\/p>\n\n\n\n
November 28, 2022<\/p>\n\n\n\n
The autoimmune encephalitides (AEs) are a group of immune-mediated brain diseases that frequently associate with pathogenic antibodies against neuronal or glial proteins. There are now more than 20 AEs described that individually or collectively fulfill criteria of rare diseases (defined in the US as fewer than 200\u202f000 affected people).1<\/a><\/sup> However, despite being rare, the impact of AEs in neurology, pediatrics, and psychiatry has been extraordinary and they are usually included among the top differential diagnoses for any patient presenting with rapid memory impairment, altered mental functions, or psychiatric symptoms with or without seizures or focal neurologic deficits.2<\/a><\/sup> If first-line studies for infections, psychiatric, and toxicological causes are negative, the AEs inevitably become the frontrunners of the differential diagnosis. Moreover, if any antibody included in the extensive panels of paraneoplastic or autoimmune encephalitis turns out positive, the diagnosis is considered firm and established, overriding the clinical assessment. This path of events, although it may seem derisive, is the main reason for the consultations we currently receive.<\/a><\/p>\n\n\n\n In 2015, the number of consultations for diagnostic errors on AEs was substantial,2<\/a><\/sup> but 2 experiences made us pause and reflect. One was a girl who developed a clear-cut case of narcolepsy but was misdiagnosed and treated for many months as anti\u2013N<\/em>-methyl-d-aspartate receptor (NMDAR) encephalitis based on results of a live cell-based assay (CBA) with serum.3<\/a><\/sup> The other patient was a young girl with unequivocal symptoms of autistic spectrum disorder that was misdiagnosed as autoimmune based on the detection of voltage-gated potassium channel (VGKC) antibodies in serum. For several years, she underwent monthly blood tests (all with borderline results) that led to immunotherapy escalation to include rituximab and cyclophosphamide. Both patients were misdiagnosed at 2 centers of excellence for autoimmune neurology, their antibody findings were not validated with alternative tests, and their symptoms were different from those associated with their supposed AEs, altogether suggesting problems at multiple levels. This experience fostered the drafting of an article on the clinical approach to AEs, subsequently endorsed by a group of experts.4<\/a><\/sup> Although the article is frequently used as a source of stand-alone criteria for AEs, it was aimed to provide a comprehensive clinical algorithm designed to assist in the differential diagnosis of AEs and initiate prompt treatment without waiting for antibody test results. Because of this, the embedded criteria for a multistage disease, such as anti-NMDAR encephalitis, were tailored for a stage in which the syndrome is recognizable, minimizing errors or unnecessary treatments (ie, the criteria are inadequate for the stage with isolated psychosis or seizures).<\/a><\/p>\n\n\n\n Before applying the algorithm, minimal requirements for clinical suspicion of AE (or criteria for possible AE) must be met: (1) rapid progression (\u22643 months) of memory deficits, altered mental status, or psychiatric symptoms, (2) 1 or more of the indicated features (new focal central nervous system deficits, unexplained seizures, cerebrospinal fluid [CSF] pleocytosis, or magnetic resonance imaging [MRI] findings suggesting encephalitis), and (3) exclusion of a specific list of alternative disorders. This exclusion step is extremely important although rarely well implemented. Fulfillment of all 3 criteria suggests that the symptoms may be caused by AE and, therefore, it is worth proceeding with the diagnostic algorithm. Disorders with a more protracted course (ie, Morvan syndrome) were discussed separately.4<\/a><\/sup> Neuronal antibody studies were not at the frontline of this algorithm but they served for later refinement of the diagnosis and treatment.<\/a><\/p>\n\n\n\n In this issue of JAMA Neurology<\/em>, the problem of misdiagnosing AEs is addressed by Flanagan and colleagues5<\/a><\/sup> in a series of 393 adult patients diagnosed with AEs, among whom 107 (27%) did not have AE. The study was conducted in 6 academic centers over a period of 7 years; some patients were misdiagnosed in the same centers and others elsewhere. In 17 cases (16%) the alternative diagnoses were confirmed with biopsy, autopsy, genetic testing, infectious testing, or other, and in the remaining 90 cases (84%), the alternative diagnoses were suggested by laboratory testing and imaging findings. The article offers the possibility to assess the indicated criteria of possible AE,4<\/a><\/sup> whose performance was markedly good: if the first 2 criteria had been applied, 77 misdiagnoses (72%) would have been prevented. Furthermore, considering the final diagnoses, reasons for misdiagnoses, and neuroimaging presented, most of the remaining 30 patients would not have been misdiagnosed if all 3 criteria had been implemented. Altogether, most patients did not meet criteria for possible AE and did not even need the algorithm.<\/a><\/p>\n\n\n\n Considering the alternative diagnoses and reasons for diagnostic errors provided by the authors,5<\/a><\/sup> the high frequency of misdiagnoses (approximately 1 of 4 patients) comes as no surprise. In 53 of 105 patients (50%), the misdiagnoses were caused by serum testing (overinterpretation or nonspecific positive result). In contrast, CSF testing caused misdiagnosis in only 7 of 91 cases (8%). Of these, 4 samples considered NMDAR antibody positive did not react with brain tissue (which in our experience should be considered antibody negative),6<\/a><\/sup> another 2 samples had antibodies that should not be used for diagnosing AE (1 VGKC, 1 uncharacterized antigen), and 1 sample with GAD antibodies was from a patient with Alzheimer disease\/vascular dementia, which are not clinical features of these antibodies.7<\/a><\/sup><\/a><\/p>\n\n\n\n A recurrent problem over the years has been the unintelligible effort of some investigators in favoring serum over CSF antibody testing in diseases of the brain, even at the expense of errors.3<\/a><\/sup> The concept of greater serum sensitivity has been overemphasized, paying much less attention to the syndrome association, specificity, or clinical response to immunotherapy. These approaches have been fertile ground for stoking concepts, such as autoimmune psychosis, autoimmune epilepsy, autoimmune movement disorders, or autoimmune depression, which when devoid of antibodies in CSF, the clinical implications of serum antibodies are unclear.2<\/a><\/sup> In this setting, the response to treatment should also be taken with caution, as demonstrated by the history of VGKC antibodies. There was a time that these antibodies were the emperor of many maladies (ie, pain, limbic encephalitis, epilepsy, occupational neuropathies, or dementia)8<\/a><\/sup>,9<\/a><\/sup> frequently showing good correlations between immunotherapy, decrease of titers, and clinical improvement. Currently, VGKC antibodies are considered a frequent source of misdiagnosis. In a study of 72 patients considered VGKC-antibody positive (but LGI1 and CASPR2 negative), most did not have autoimmune diseases, the serum did not react with cultured neurons (suggesting intracellular epitopes), and the response to immunotherapy was limited. The authors concluded that double-negative VGKC complex antibodies are not pathogenic and do not support the use of immunotherapy.10<\/a><\/sup><\/a><\/p>\n\n\n\n Another persistent theme in any study on AE misdiagnoses are those caused by thyroid peroxidase (TPO) antibodies as biomarkers of Hashimoto encephalitis. This includes the challenging association of a nonspecific antibody (TPO) with an ill-defined syndrome.11<\/a><\/sup> GAD65 antibodies are also a persistent cause of diagnostic errors, particularly if the clinical context, sample examined, and titers are not carefully assessed.7<\/a><\/sup><\/a><\/p>\n\n\n\n Twenty-seven (25%) of the 107 misdiagnosed patients had psychiatric or functional disorders. The absence of a cure for many psychiatric diseases encourages a search for alternative options. Additionally, the extensive literature describing serum neuronal antibodies in countless disorders (eg, schizophrenia, psychosis, stroke, Parkinson, amyotrophic lateral sclerosis)12<\/a><\/sup> makes matters worse. These publications are catchy, but the clinical consequences are not being considered. Most do not confirm the results nor examine CSF (or if done, is negative) and the presence of serum antibodies rarely associates with responses to immunotherapy.13<\/a><\/sup><\/a><\/p>\n\n\n\n How can we prevent misdiagnoses? (1) Become familiar with the syndromes and favor clinical reasoning over antibody results. (2) Use CSF and serum testing; if forced to pick one, use CSF. (3) Be aware that a positive test result does not always mean presence of antibodies; thus, request antibody confirmation by at least 2 techniques (brain tissue immunohistochemistry and CBA). (4) Doubt any positive serum result that in the context of suspected AE is accompanied by negative CSF antibodies. An exception is the antibodies against myelin oligodendrocyte glycoprotein (MOG) that are more frequently detected in serum and can associate with symptoms of AE.14<\/a><\/sup> Even for anti-LGI1 encephalitis, the concept that antibodies occur more frequently in serum is inaccurate; virtually all patients have LGI1 antibodies in CSF if properly studied with brain immunohistochemistry and CBA expressing LGI1 with one of its natural ligands (ADAM23).15<\/a><\/sup> Therefore, diagnostic laboratories should strongly consider including a warning message in their reports, indicating that a positive serum result for antibodies against neuronal surface proteins mandates the confirmation of antibodies in CSF, as the risk for clinical misdiagnosis of AE dramatically increases if the CSF status is negative or unknown.<\/a><\/p>\n\n\n\n