Neural Antibodies as Diagnostic Biomarkers of Autoimmune Encephalitis and Inflammatory Demyelinating Disease

Neural antibodies (ie, antibodies targeting neuronal or glial proteins) have emerged as immensely useful biomarkers of central nervous system (CNS) autoimmunity.¹ Detection of these antibodies has revolutionized the diagnosis of autoimmune encephalitis (AE) and played an instrumental role in defining non–multiple sclerosis (MS) inflammatory demyelinating diseases, including neuromyelitis optica spectrum disorder (NMOSD) and myelin oligodendrocyte glycoprotein antibody–associated disease (MOGAD). We review testing for these neural antibodies (summarized in the Table), provide guidance on how to optimize the performance of this testing in clinical practice, and discuss the incorporation of neural antibody testing into recently proposed diagnostic criteria.

Neural Antibodies in AE

AE is an often devastating inflammatory brain disease, and prompt diagnosis is essential to its early treatment. Neural antibodies that are highly specific biomarkers of neurologic autoimmunity have been discovered in people with AE in recent decades. These antibodies can be broadly classified into 2 groups: those against extracellular or intracellular targets. Antibodies against extracellular targets are typically thought to be pathogenic, with generally better prognosis because of their often good response to prompt immunotherapy and low to intermediate risk of underlying malignancy.² Examples include encephalitis mediated by anti–N-methyl-D-aspartate receptor (NMDAR), which is classically observed in younger women with subacute neuropsychiatric symptoms and ovarian teratoma, as well as anti–leucine-rich glioma-inactivated 1 (LG1), which is classically observed in older men with frequent focal-onset seizures, cognitive impairment, and no underlying tumor. Antibodies against intracellular targets are typically thought to be nonpathogenic surrogate markers of cytotoxic T cell–mediated pathology, with generally worse prognosis because of their often poor response to immunotherapy and high risk of underlying malignancy.² Examples include encephalitis associated with anti-Hu/anti-neuronal nuclear antibody type 1 (ANNA-1), which is classically observed in older adults with multifocal nervous system dysfunction and small cell lung cancer, and anti-Yo/Purkinje cell cytoplasmic antibody type 1 (PCA-1), which is classically observed in older women with rapidly progressive cerebellar syndrome and ovarian cancer.

Neural Antibodies in NMOSD and MOGAD

Both NMOSD and MOGAD present as inflammatory demyelinating diseases that are distinct from MS. In 2004, screening for neural antibodies using mouse tissue indirect immunofluorescence led to the discovery of neuromyelitis optica (NMO) immunoglobulin G (NMO-IgG) in people with NMO.³ The antigenic target of NMO-IgG was found to be aquaporin-4 (AQP4), which now serves as a unifying biomarker across clinical presentations that extend beyond optic neuritis and myelitis and thus are better captured by the more inclusive term NMOSD.^4,5 Such additional presentations include area postrema syndrome, acute brainstem syndrome, acute diencephalic syndrome, and symptomatic cerebral syndrome with NMOSD-typical lesions on MRI.⁵

In contrast to anti-AQP4 and NMOSD, antibodies against myelin oligodendrocyte glycoprotein (anti-MOG) were long suspected to be involved in inflammatory demyelinating disease even before recognition of MOGAD as a discrete entity. However, early studies of anti-MOG in MS that used denatured protein yielded conflicting results, suggesting that its detection was of uncertain clinical significance.⁶ In 2007, it was found that detection of anti-MOG using natively folded protein was strongly associated with acute disseminated encephalomyelitis rather than MS, indicating that this antibody might be a biomarker of non-MS inflammatory demyelinating disease presentations.^6,7 Subsequent studies using cell-based assays (CBA) to detect anti-MOG against conformational protein have confirmed this and expanded the spectrum of anti-MOG–associated clinical presentations beyond acute disseminated encephalomyelitis to include optic neuritis, myelitis, brainstem/cerebellar syndromes, and, most recently, meningo-cortical manifestations including FLAMES (fluid-attenuated inversion recovery–hyperintense lesions in anti–MOG-associated encephalitis with seizures)⁸; this spectrum of presentations is referred to as MOGAD.⁹

Optimization of Neural Antibody Testing in Clinical Practice

In people with suspected AE for whom neural antibody testing is being pursued, there are numerous aspects of test ordering and interpretation that merit consideration. With respect to which fluid to test, testing for some neural antibodies is more sensitive when using serum (eg, anti-LGI1); for others, testing of CSF is more sensitive and specific (eg, anti-NMDAR).^1,10 Furthermore, there can be substantial phenotypic overlap across the various neural antibody–defined forms of AE (eg, subacute onset of memory deficits, altered mental status, psychiatric symptoms, seizures).^1,10 For these reasons, with few exceptions, comprehensive serum and CSF panel–based testing are recommended to optimize sensitivity and specificity.^1,10

The detection of a neural antibody in a person with suspected AE is highly specific for neurologic autoimmunity, so long as antibody positivity is in line with best testing practices. For most neural antibodies associated with AE, best testing practices refers to a 2-step tissue-based approach to antibody detection. This approach entails the use of a tissue-based assay (eg, tissue indirect immunofluorescence) to identify staining that is characteristic for a particular antibody, along with confirming positivity for that antibody using a second assay.^1,2,10 This rigorous approach to neural antibody detection ensures high specificity and minimizes the risk of false-positive antibody results. However, commercially available kits that were intended to be used alongside tissue-based assays as confirmatory tests (eg, line blots, CBA) are increasingly being used as standalone offerings in clinical laboratories.¹⁰ Whereas this 1-step approach mitigates the cost and expertise required to implement tissue-based assays, it also substantially reduces test specificity and can dramatically increase the proportion of false-positive antibody results encountered in clinical practice.¹⁰ For this reason, clinicians should ensure that their neural antibody testing laboratory follows best testing practices, with appropriate incorporation of tissue-based assays into their testing algorithms. When encountering an individual with a positive neural antibody result but an atypical phenotype for that antibody, the clinician should contact the testing laboratory to review test methodology, discuss the possibility of a false-positive result, and facilitate send-out testing if necessary.

With respect to NMOSD and MOGAD, CBAs are the recommended test methodology for both anti-AQP4 and anti-MOG detection because of their high sensitivity and specificity.^5,6,9 For anti-AQP4, live and fixed CBA have been found to perform comparably well.¹¹ Enzyme-linked immunosorbent assay for anti-AQP4 is clinically available, but its diagnostic performance has been reported to be inferior to CBA, so anti-AQP4 CBA should be ordered over enzyme-linked immunosorbent assay when possible.¹² Recently, a rapid enzyme immunodot assay for anti-AQP4 was found to perform similarly to CBA, suggesting that this time- and cost-efficient assay may be a viable alternative to CBA in resource-limited settings.¹³ Using CBA, serum is more sensitive than CSF for detection of anti-AQP4 and is thus the preferred specimen to submit for testing.^5,6

When testing for anti-MOG, both live and fixed CBAs are suitable. The overall agreement between the 2 has been reported to be high, particularly for people with high antibody titers.^6,14 Live CBAs for anti-MOG performed at specialized centers have been reported to have slightly higher diagnostic accuracy than fixed CBA and may help reconcile unexpected false-positive or false-negative fixed CBA results.^{6, 5,16} However, low titers of anti-MOG by either live or fixed CBA have lower specificity for MOGAD.^6,17 This takes on particular importance in people with reported positivity for anti-MOG and atypical disease phenotypes; in such individuals, anti-MOG titer should be clarified with the testing laboratory, and low-titer antibody results should be viewed critically. With respect to serum vs CSF testing for anti-MOG, serum has overall higher sensitivity when using CBA and is thus the preferred testing specimen. However, isolated CSF positivity for anti-MOG in people with compatible clinical presentations has increasingly been reported, suggesting that CSF anti-MOG testing may be of value in cases with a high index of suspicion for MOGAD and negative serum anti-MOG testing.^6,18

Emergence of Neural Antibody–Based Diagnostic Criteria

Criteria to aid in the diagnosis of CNS autoimmune diseases serve as helpful frameworks that clinicians can apply in routine practice. In 2016, a syndrome-based diagnostic algorithm was proposed to aid in the evaluation of individuals presenting with suspected encephalitis.¹⁹ As per this algorithm, people can only be diagnosed with probable or definite AE if they first meet the criteria for possible AE. To meet these criteria, an individual is required to have subacute onset of working memory deficits, altered mental status, or psychiatric symptoms, which are features that are classic for AE.²⁰ Inclusion of possible AE criteria in the syndrome-based diagnostic algorithm is beneficial to its specificity. However, these criteria may not encompass people who present with formes frustes of AE along with neural antibodies that are highly specific for neurologic autoimmunity, such as people with anti-LGI1 and seizures alone or people with anti–kelch-like protein 11 (KLHL11) and isolated brainstem dysfunction.²¹ To capture such individuals with neural antibody–associated AE who may not meet possible AE criteria and thus fall outside the syndrome-based diagnostic algorithm, the development of complementary neural antibody–based diagnostic criteria for AE has recently been proposed.²²

As per this proposal, the 3 pillars of neural antibody–based diagnostic criteria for a given neural antibody–associated AE would be (1) 1 or more core clinical characteristics, (2) antibody positivity in line with best testing practices, and (3) reasonable exclusion of alternative diagnoses. This 3-pillared approach to neural antibody–based diagnostic criteria (depicted in the Figure) has already been applied to NMOSD and MOGAD,^5,9 indicating the feasibility of its application to neural antibody–defined forms of AE as well.

With respect to criteria for NMOSD and MOGAD, it is noteworthy that NMO was recognized as a distinct entity before the discovery of an associated autoantibody. This is reflected by the fact that the first set of criteria to aid in the diagnosis of NMO was proposed in 1999,²³ before the discovery of anti-AQP4 in 2004. Following this landmark antibody discovery, revised criteria incorporating antibody status were proposed to broadly yet accurately capture the full spectrum of clinical presentations observed with anti-AQP4–seropositive NMOSD.⁵ Notably, in addition to criteria for anti-AQP4–seropositive disease, the most recent criteria to help diagnose NMOSD also retained criteria for anti-AQP4–seronegative disease.⁵ Whereas these criteria aid conceptually in the approach to people who have clinical presentations that are suspicious for NMOSD yet negative for anti-AQP4, it is increasingly recognized that some such anti-AQP4–seronegative individuals have alternative diagnoses, including MOGAD. Furthermore, greater heterogeneity in the pathobiologic markers and immunotherapy responses of people with seronegative NMOSD have been reported,^24,25 and it remains to be seen how this less well-defined entity will be incorporated in future revisions of NMOSD criteria that are likely to emphasize seropositive disease.

In contrast to NMOSD, it was the detection of anti-MOG with assays using conformational protein that spurred descriptions of distinct anti-MOG–associated inflammatory demyelinating disease presentations, which are now encompassed by the term MOGAD. This difference in origins between NMOSD and MOGAD is reflected by their nomenclature: NMOSD is named after the initially recognized clinical syndrome, whereas MOGAD is named after the initially recognized autoantibody. It is also reflected by their proposed criteria: the 2015 NMOSD criteria include criteria to help diagnose seronegative disease, whereas the 2023 MOGAD criteria do not.^5,9 Also unique to these MOGAD criteria is the requirement for supporting features to be present in people with only low antibody titers in serum or only CSF positivity, reflecting their lower level of diagnostic certainty.⁹ Nonetheless, it bears emphasizing that the anti-AQP4–seropositive NMOSD criteria, MOGAD criteria, and recently proposed neural antibody–based diagnostic criteria for AE all share the 3-pillared approach to neural antibody–based diagnosis, signifying a transition from syndrome-based to biomarker-based diagnosis of CNS autoimmune diseases.

Conclusion

Neural antibodies are highly specific biomarkers of AE, NMOSD, and MOGAD. The diagnostic utility of a positive neural antibody relies on confirmation that its method of detection is in line with best testing practices; for this reason, clinicians should maintain an open line of communication with their testing laboratory to review test methodology and discuss unexpected positive or negative neural antibody results. Increasing emphasis on neural antibody positivity in diagnostic criteria for autoimmune neurologic diseases further highlights the importance of ensuring best testing practices are followed to minimize the risk of false-positive antibody results. Neural antibody–based diagnostic criteria are not synonymous with a diagnosis based on neural antibody positivity in isolation; in any individual with neurologic symptoms and a positive neural antibody result, clinical correlation along with exclusion of alternative etiologies remains essential to accurate diagnosis. As the field of autoimmune neurology continues to advance, so will the role of neural antibodies in the diagnosis of CNS autoimmune diseases.