Neurological and cognitive outcomes after antibody-negative autoimmune encephalitis in children

Machine learning for the early prediction of long-term cognitive outcome in autoimmune encephalitis

BACKGROUND AND OBJECTIVE

Autoimmune encephalitis (AE) is an immune-mediated disease. Some patients experience persistent cognitive deficits despite receiving immunotherapy. We aimed to develop a prediction model for long-term cognitive outcomes in patients with AE.

METHOD

In this multicenter cohort study, a total of 341 patients with AE were enrolled from February 2014 to July 2023. Cognitive impairment was identified using the telephone Mini-Mental State Examination (t-MMSE). Six machine learning (ML) algorithms were used to assess the risk of developing cognitive impairment.

RESULTS

The median age of the patients with AE was 30.0 years (23.0-48.25), and 48.90 % (129/264) were female in the training cohort.77 (29.2 %) patients were identified with cognitive impairment after a median follow-up of 49 months. Among 16 features, the following six features were finally selected to develop the model: Cognitive Reserve Questionnaire (CRQ), Clinical Assessment Scale for Autoimmune Encephalitis (CASE), status epilepticus (SE), age, MRI abnormalities, and delayed immunotherapy. Compared to other ML models, the random forest (RF) model demonstrated superior performance with an AUC of 0.90. The accuracy, sensitivity, and specificity in the testing cohort were 0.87, 0.79, and 0.90, respectively.

CONCLUSION

The RF model based on CRQ, CASE scores, SE, age, MRI abnormalities and delayed immunotherapy demonstrates superior predictive performance and shows promise in predicting the risk of long-term cognitive outcomes in patients with AE in clinical settings.

Interpretable machine learning model for predicting the prognosis of antibody positive autoimmune encephalitis patients

OBJECTIVE

The objective was to utilize nine machine learning (ML) methods to predict the prognosis of antibody positive autoimmune encephalitis (AE) patients.

METHODS

The encephalitis data from the Global Burden of Disease (GBD) study is analyzed to reflect the disease burden of encephalitis. This study included 187 patients with AE. 121 patients as training set and 67 patients as validation set. Decision trees (DT), random forest (RF), extreme gradient boosting (XGBoost), k-nearest neighbor (KNN), support vector machine (SVM), naive bayes (NB), neural network (NN), light gradient boosting machine (LGBM), and logistic regression (LR) are ML methods used to construct predictive models. The constructed models were validated for discrimination, calibration and clinical applicability using validation set data. Shapley additive explanation (SHAP) analysis was used to explain the model.

RESULTS

The number of encephalitis worldwide deaths, incidence and prevalence is increasing every year from 2010 to 2021. The training set included 121 patients with AE. Univariate analysis and LASSO screening identified six variables. The results of constructing models using 9 ML methods showed RF had the highest accuracy (0.860), followed by XGBoost (0.826), with F1 scores of 0.844 and 0.807, respectively. Validation set data showed good discrimination, calibration and clinical applicability of the model. The SHAP values of infection, CSF monocyte percentage, and prealbumin were 0.906, 0.790, and 0.644, respectively.

LIMITATIONS

As a rare disease, the sample size of this study is relatively small.

CONCLUSION

The model constructed using RF and XGBoost has good performance, good discrimination, calibration, clinical applicability, and interpretability.

Impact of autoantibodies against myelin oligodendrocyte glycoprotein in paediatric acquired demyelinating disease: Intellectual functioning and academic performance

Paediatric acquired demyelinating syndromes (pADS) attack white matter pathways in the brain during an important period of development. Affected children can experience poor functional outcomes, including deficits in specific cognitive domains. Understanding risk factors for poor outcome will guide clinical management of these children. One clinical phenotype which may differentially impact cognitive outcomes is the presence of autoantibodies to myelin oligodendrocyte glycoprotein (MOG). Preliminary research has suggested that cognitive difficulties exist in paediatric patients who test positive for MOG antibodies or MOGAD (Myelin Oligodendrocyte Glycoprotein Associated Disease) however, they experience a less severe profile compared to seronegative counterparts. The current study assesses children diagnosed with pADS who tested positive or negative for MOG-ab using standardised assessments of both intellectual functioning and academic ability. The results show that a subset of MOGAD patients experience clinically significant sequalae in intellectual functioning and academic ability. The neuropsychological profile also differed between children with and without MOG-ab positivity, with seronegative patients more likely to show a clinically relevant difficulties at the individual patient level. Whilst no differences existed at the group-level; the current study demonstrates the relative additional risk of intellectual/academic difficulty associated with MOG-ab seronegativity. This research further supports the growing perspective that MOG-positivity confers a more favourable neuropsychological outlook than is the case for their seronegative counterparts. This broadening consensus offers reassurance for clinicians, families, and patients.

Clinical features and outcomes in children with seronegative autoimmune encephalitis

AIM

To characterize the presenting features and outcomes in children with seronegative autoimmune encephalitis, and to evaluate whether scores at nadir for the Modified Rankin Scale (mRS) and Clinical Assessment Scale for Autoimmune Encephalitis (CASE) or its paediatric-specific modification (ped-CASE) are predictive of outcomes.

METHOD

This observational study included children younger than 18 years of age with seronegative autoimmune encephalitis. Demographics and clinical data were collected. The mRS and CASE/ped-CASE scores were used to evaluate disease severity. Descriptive statistics and logistic regression were used for data analysis and to evaluate associations between scale scores and outcomes.

RESULTS

Sixty-three children were included (39 [62%] females, median age 7 years, interquartile range [IQR] 4 years 1 months-11 years 6 months), with follow-up available for 56 out of 63 patients (median follow-up 12.2 months, IQR 13.4-17.8). The most frequent presenting neurological manifestation was encephalopathy (81%). Median CASE/ped-CASE and mRS scores at nadir were 12.0 (IQR 7.0-17.0) and 1.0 (IQR 0-2.0) respectively. Thirty-three patients (59%) had persistent neurological deficits at follow-up. Both scoring systems suggested good functional recovery (mRS score ≤2, 95%; CASE/ped-CASE score <5, 91%). CASE/ped-CASE score was more likely than mRS to distinguish children with worse outcomes.

INTERPRETATION

Children with seronegative autoimmune encephalitis are likely to have neurological deficits at follow-up. CASE/ped-CASE is more likely to distinguish children with worse outcomes than MRS.

WHAT THIS PAPER ADDS

Encephalopathy was the most common reason for presentation in children with seronegative autoimmune encephalitis. Children with seronegative autoimmune encephalitis are likely to have neurological deficits at follow-up. The Clinical Assessment Scale for Autoimmune Encephalitis or its paeditric version at nadir was better at distinguishing children with poor outcomes than the Modified Rankin Scale.

Higher incidence of acute symptomatic seizures in probable antibody-negative pediatric autoimmune encephalitis than in major antibody-positive autoimmune encephalitis

Purpose

To delineate the characteristics of probable antibody-negative pediatric autoimmune encephalitis (probable Ab-negative AE), we compared the clinical features of probable Ab-negative AE to those of major antibody-positive AE.

Methods

We retrospectively reviewed the clinical features of 18 patients with probable Ab-negative AE, 13 with anti-N-methyl-D-aspartate receptor encephalitis (NMDARE), and 13 with myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD). Clinical characteristics, neuroimaging findings, treatments, and outcomes were analyzed.

Results

The age of onset and length of hospital stay were significantly higher in the NMDARE group than in the other groups (p = 0.02 and p < 0.01). Regarding initial neurological symptoms, acute symptomatic seizures in the probable Ab-negative AE group (67%) were significantly more frequent than in the NMDARE (15%) and MOGAD (23%) groups (p < 0.01). Paraclinical evidence of neuroinflammation within 1 month of disease onset revealed that single-photon emission computed tomography (SPECT) detected abnormal alterations in 14/14 (100%), cerebrospinal fluid (CSF) analysis in 15/18 (83%), and magnetic resonance imaging (MRI) in 11/18 (61%) in patients with probable Ab-negative AE. In the probable Ab-negative AE group, seven patients (39%) developed autoimmune-associated epilepsy, whereas one patient (8%) had both NMDARE and MOGAD (not statistically significant, p = 0.07).

Conclusion

Patients with probable Ab-negative AE exhibited acute symptomatic seizures as initial neurological symptoms significantly more frequently. They developed autoimmune-associated epilepsy more frequently than those with NMDARE and MOGAD, which was not statistically significant. SPECT within 1 month of disease onset might be a valuable surrogate marker of ongoing neuroinflammation and neuronal dysfunction, even in patients with negative MRI findings.

Immune-mediated encephalitis

A neurological deterioration in a child presents a significant worry to the family and often a diagnostic challenge to the clinician. A dysregulated immune response is implicated in a wide and growing spectrum of neurological conditions. In this review we consider the current paradigms in which immune-mediated encephalopathies are considered; the development of paediatric specific diagnostic criteria that facilitate early consideration and treatment of immune-mediated conditions and the limitations and potential developments in diagnostic testing. We consider the expanding phenotype of myelin oligodendrocyte glycoprotein antibody, the spectrum of virus-associated encephalopathy syndromes, and the strategies that have been employed to build an evidence base for the management of these rare conditions. Looking forward we explore the potential for advanced molecular investigations to improve our understanding of immune-mediated encephalitides and guide future treatment strategies. Recently characterized immune-mediated central nervous system disorders include new antibodies causing previously recognized phenotypes. Aggregation of conditions with similar clinical triggers, and characterization of unique imaging features in virus-associated encephalopathy syndromes. Immune treatment iscurrently guided by meta-analysis of individualized patient data and/or multi-national consensus.

Diagnostic criteria for autoimmune encephalitis: utility and pitfalls for antibody-negative disease

Increased awareness of autoimmune encephalitis has led to two unintended consequences: a high frequency of misdiagnoses and the inappropriate use of diagnostic criteria for antibody-negative disease. Misdiagnoses typically occur for three reasons: first, non-adherence to reported clinical requirements for considering a disorder as possible autoimmune encephalitis; second, inadequate assessment of inflammatory changes in brain MRI and CSF; and third, absent or limited use of brain tissue assays along with use of cell-based assays that include only a narrow range of antigens. For diagnosis of possible autoimmune encephalitis and probable antibody-negative autoimmune encephalitis, clinicians should adhere to published criteria for adults and children, focusing particularly on exclusion of alternative disorders. Moreover, for diagnosis of probable antibody-negative autoimmune encephalitis, the absence of neural antibodies in CSF and serum should be well substantiated. Neural antibody testing should use tissue assays along with cell-based assays that include a broad range of antigens. Live neuronal studies in specialised centres can assist in resolving inconsistencies with respect to syndrome-antibody associations. Accurate diagnosis of probable antibody-negative autoimmune encephalitis will identify patients with similar syndromes and biomarkers, which will provide homogeneous populations for future assessments of treatment response and outcome.

Biomarkers in autoimmune diseases of the central nervous system

The autoimmune diseases of the central nervous system (CNS) represent individual heterogeneity with different disease entities. Although clinical and imaging features make it possible to characterize larger patient cohorts, they may not provide sufficient evidence to detect disease activity and response to disease modifying drugs. Biomarkers are becoming a powerful tool due to their objectivity and easy access. Biomarkers may indicate various aspects of biological processes in healthy and/or pathological states, or as a response to drug therapy. According to the clinical features described, biomarkers are usually classified into predictive, diagnostic, monitoring and safety biomarkers. Some nerve injury markers, humoral markers, cytokines and immune cells in serum or cerebrospinal fluid have potential roles in disease severity and prognosis in autoimmune diseases occurring in the CNS, which provides a promising approach for clinicians to early intervention and prevention of future disability. Therefore, this review mainly summarizes the potential biomarkers indicated in autoimmune disorders of the CNS.

Determination of Clinical, Electrophysiological, and Radiological Characteristics of Pediatric Autoimmune Encephalopathy

Autoimmune encephalopathy (AE) is a group of diseases with subacute onset, that represents a wide clinical spectrum, manifested by complex neuropsychiatric symptoms and signs. In this study, the data of 27 patients diagnosed and followed up in our clinic with the diagnosis of AE between 2011 and 2021 were evaluated retrospectively. Out of 27 patients, 6 were definite seropositive AE, 2 of them met the diagnostic criteria for limbic encephalitis, and the remaining 19 were probable AE. Nowadays, we see AEs with increasing frequency. While there is a generally established approach in the diagnosis and treatment of seropositive patients, there are still hesitations and diagnostic difficulties in seronegative AEs. In this study, clinical, radiological, and prognostic features of definite and probable AE patients diagnosed in a tertiary pediatric neurology clinic were documented. It is thought that pediatric neurologists have an important responsibility to increase awareness about AE in pediatricians. In the future, it is predicted that AE will be diagnosed more frequently with new antibodies and one has to differentiate it from viral encephalitis and neuropsychiatric syndromes and diseases.

Seizure Evolution and Outcome in Pediatric Autoimmune Encephalitis

BACKGROUND

Our study aimed to characterize seizure incidence and seizure outcome of pediatric autoimmune encephalitis (AE) focusing on subgroup analysis based on antibody (Ab).

METHODS

Among 110 pediatric patients with AE, we compared seizure characteristics and outcomes in 68 patients with seizure, who satisfied the proposed criteria of pediatric AE. Accordingly, patients were classified into three groups, anti-myelin oligodendrocyte glycoprotein (anti-MOG) AE, anti-N-methyl-D-aspartic acid receptor (anti-NMDAR) AE, and Ab-negative AE. Univariate and multivariate analyses were performed to evaluate the risk factors for postencephalitic seizures, defined as persisting seizures six months after onset.

RESULTS

Seizure incidence in the anti-NMDAR (88.9%) and Ab-negative (71.1%) groups differed from anti-MOG group (37.8%). Median seizure frequency within six months was higher in the Ab-negative group (6.0, interquartile range [IQR] 3.0 to 13.0) than in the anti-NMDAR group (3.0, IQR 2.0 to 4.5) and anti-MOG group (2.0, IQR 1.0 to 5.0). Patients in the Ab-negative group tended to develop postencephalitic seizures more frequently and have a lower seizure freedom rate than those in the anti-NMDAR and anti-MOG groups. Ab-negative status, high seizure frequency within six months, and the presence of status epilepticus were associated with the development of postencephalitic seizures on univariate analysis. On multivariate analysis, Ab-negative status remained the only significant variable linked with postencephalitic seizure (odds ratio, 4.17; 95% confidence interval, 1.02 to 18.05).

CONCLUSIONS

We delineated the seizure incidence, evolution, and outcome of pediatric patients with Ab-positive and Ab-negative AE. Ab-negative status is predictive of higher seizure burden, more frequent development of postencephalitic seizures, and less favorable seizure outcome than anti-NMDAR and anti-MOG Ab-positive status.

Evaluation of multiple consensus criteria for autoimmune encephalitis and temporal analysis of symptoms in a pediatric encephalitis cohort

Objective

To evaluate the sensitivity and specificity of current criteria for the diagnosis of autoimmune encephalitis (AE) and the temporal onset of neuropsychiatric symptoms (NP) in a pediatric encephalitis cohort.

Background

Multiple criteria for AE have been developed, including the Graus and pediatric-focused Cellucci consensus criteria, and the Determining Etiology in Encephalitis (DEE) score for patients with encephalitis. Early identification and treatment of AE is crucial to improve outcomes, but this can be difficult given the frequent overlap of clinical presentation between AE and infectious encephalitis (IE).

Design/methods

A retrospective review was conducted of patients seen at our institution from 2000 to 2021 with a final diagnosis of AE or IE. These were narrowed through multiple exclusions to etiology-confirmed IE or antibody-positive/negative AE. Time of onset or results of all symptoms and diagnostics were recorded. Sensitivity and specificity of each criterion under various clinical scenarios were calculated over the first month after initial NP symptom onset.

Results

A total of 23 antibody-positive AE, 9 antibody-negative AE and 23 IE patients were included in final analysis. Under an idealized scenario with rapid initial diagnostic evaluations, the sensitivity for pediatric AE by day 28 after onset of NP symptoms approached 90% for both Cellucci and Graus criteria. Specificity within these 28 days was low without infectious testing results, increasing the greatest with rapid PCR testing and second with infectious antibody testing-reaching ~90% with both. A DEE score of 3 provided a specificity of 100% in identifying IE, but low sensitivity (29%). Symptoms were noted to cluster within several days of onset in IE, but in AE were spread out. Personality/behavioral change, speech change, affective disorder, and sleep disturbance were noted more often in AE, while fever, elevated C-reactive protein or CSF protein, and abnormal MRI-Brain occurred more often in IE.

Conclusion

In this study, we provide the first evaluation of the Cellucci criteria and the first validation of the DEE score in the differentiation of pediatric AE and IE. Further refinement of AE criteria is needed to improve early detection and treatment of pediatric AE.

Clinical Characteristics and Prognosis of Antibody-Negative Autoimmune Encephalitis in Children: A Single-Center Retrospective Study

BACKGROUND

Autoimmune encephalitis (AE) is a group of immune-mediated brain diseases. However, new diagnostic criteria for AE in children indicate that partial pediatric patients with AE may be diagnosed without evidence of positive autoantibodies. Therefore, the clinical characteristics and prognosis of children with antibody-negative but probable AE require further investigation.

METHODS

Forty-one children with AE admitted to our hospital from April 2014 to January 2021 were retrospectively enrolled in this study. Children were divided into two groups according to positive or negative antibody tests. Clinical characteristics, cerebrospinal fluid, video electroencephalography, brain magnetic resonance imaging, and prognosis were analyzed, and the correlation between modified Rankin scale (mRS) and neutrophil-to-lymphocyte ratio (NLR) was examined.

RESULTS

Of 41 children, 16 cases tested positive for autoantibodies. The main features were psychiatric symptoms, cognitive disturbances, speech disturbances, movement disorders, and seizures. All the children were given a combination of intravenous methylprednisolone pulses with intravenous immunoglobulin therapy; 26 cases (63%) had a good outcome, and 15 cases (37%) had a poor outcome. Antibody-positive and antibody-negative but probable AE were analyzed by univariate analysis and showed lower lymphocyte counts and higher NLR and mRS scores in the antibody-negative group (P < 0.05). The Spearman rank correlation analysis showed a positive correlation between NLR level and mRS scores (P < 0.05).

CONCLUSIONS

Antibody-negative but possible AE is frequent in children who may have a more severe neurological impairment and higher NLR than antibody-positive AE. Aggressive immunotherapy in antibody-negative AE is essential to achieve a good prognosis.

Autoimmune Encephalitis in Children

Autoimmune encephalitis is a group of central nervous system (CNS) inflammatory disorders that most commonly affect young adults and children. These disorders are closely associated with antibodies against neuronal cell-surface proteins, receptors, and ion channels; however, some forms of the disorder have no known antibody at this time. In children, neurological manifestations such as seizure, movement disorders, and focal neurological deficits are more prominent at initial presentation than psychiatric or behavioral symptoms. When psychiatric symptoms do occur, they often manifest as temper tantrums, aggression, agitation, and rarely psychosis. Prompt diagnosis and early treatment can lead to improved outcomes and decreased relapses. First-line therapies include intravenous steroids, intravenous immunoglobulin, and plasmapheresis, whereas rituximab and cyclophosphamide are utilized for refractory or relapsing disease. This review highlights the different forms of this disorder, discusses approach to diagnosis and treatment, and reviews the outcome and prognosis of children diagnosed with different forms of autoimmune encephalitis.