Edited by: Zsolt Illes, University of Southern Denmark, Denmark
Reviewed by: Iain Comerford, University of Adelaide, Australia; Clara Ballerini, University of Florence, Italy
This article was submitted to Multiple Sclerosis and Neuroimmunology, a section of the journal Frontiers in Neurology
†These authors have contributed equally to this work
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Autoimmune encephalitis (AE) is an important cause of unexplained movement disorders, rapidly progressive cognitive decline and refractory epilepsy (
A further clinical challenge are patients who clinically appear to have AE, but no associated autoantibody is identified (
Novel surrogate markers including CSF light chains and CSF cytokines have been associated with CNS inflammation (
A superior biomarker which more reliably differentiates CNS autoinflammation from other causes will assist clinicians commence treatment earlier. Our study examines a cohort of patients with high clinical suspicion of AE to identify biomarkers that might indicate this disease.
This is an exploratory study aiming to identify conventional or novel surrogate biomarkers of neurological inflammation associated with AE which provides class III evidence for the potential of the CSF cytokines IL21 and IP10 as biomarkers for AE. This study was approved by the Ethics Committee of Westmead Hospital (LNR/16/WMED/192) and written informed consent was obtained by all participants.
Patients at a single quaternary referral center in Western Sydney, Australia were prospectively recruited over 18 months between 2016 and 2018 (
The decision for recruitment was based on clinical grounds, prior to lumbar puncture and before knowledge of subsequent results of investigations. If investigations results revealed an alternative diagnosis, patients were reclassified to the appropriate group i.e., infectious. Patients with an identified autoantibody associated with AE were classified as antibody positive (AbPAE) while those without antibodies were classified as antibody negative (AbNAE). To prevent information bias, negative results from CSF analysis did not exclude enrolled patients.
Any enrolled patients diagnosed with a CNS viral infection through polymerase-chain-reaction (PCR) were included in the infectious control cohort. In addition, stored CSF samples classified as viral infection based on positive PCR results were included as viral controls. These samples were supplied as deidentified aliquots.
Non-inflammatory (NI) control CSF samples were obtained from patients undergoing large-volume lumbar puncture for “non-inflammatory” neurological disease (NIND) and from patients undergoing routine spinal anesthesia. Patients in the NIND group had a diagnoses of simple headache, idiopathic intracranial hypertension (IIH), or normal pressure hydrocephalus (NPH).
A disease control group consisting of patients with neuropsychiatric lupus, cerebral vasculitis, and multiple sclerosis were also included (OAND group).
CSF samples for AE and NI controls were collected in standard 10 mL CSF tubes. CSF for light chains and cytokine analysis were aliquoted from these samples and frozen at −80 degrees Celsius. Assays for CSF light chains and CSF cytokines were batched for analysis to minimize analytical variation.
All investigations unless otherwise stated, were performed at ICPMR (NSW Health Pathology, Australia).
Patients underwent conventional investigations for AE including blood tests and lumbar puncture for collection of CSF (
The following tests were also performed on serum: anti-thyroid antibody (Siemens Munich, Germany) and thyroglobulin antibody (Siemens Munich, Germany); and CSF: microscopy and culture; protein (Siemens Vista Erlangen, Germany), anti-glutamic acid decarboxylase (GAD) antibodies (ELISA, SEALS Pathology, Prince of Wales Hospital, NSW Australia; RSR Cardiff, UK), and polymerase chain reaction (PCR) for viral infections: HSV (Artus Hamburg, Germany), VZV (in-house PCR assay), ENTV (in house PCR assay), and EBV by PCR (Ellitech Paris, France).
All conventional investigations were collected according to current practice and performed according to the usual procedures available at the receiving diagnostic laboratory.
CSF studies performed purely on a research basis were: CSF light chains (Freelite assay; Binding Site, Birmingham United Kingdom) and a broad panel of CSF cytokines (Milliplex; Merk Millipore Darmstadt Germany) using the magnetic multi bead array kits (MPHSCTMAG28SK17; MPHCYP3MAG63K01; MPHCYTOMAG60K03; MPHCYTP2MAG62K02. Cytokines tested were: IFN-γ, ITAC/CXCL11, IL-12p70, TNFα, CXCL9, IP-10/CXCL10 (Th1 cytokines); IL-13, IL4, IL5, TARC/CCL17, Eotaxin/CCL11 (Th2 cytokines); IL17a, IL-6, IL-8 (Th17 cytokines); and IL-1β, IL-21, IL-2, IL-23, IL-7, IL-10, BCA-1/CXCL13, GMCSF, GCSF (other cytokines). Kits were chosen based on maximum sensitivity for cytokine detection. Lower limits of detection of the cytokine assay are detailed in
Any additional clinically necessary investigations for diagnosis or management including MRI was performed. MRI reports included in our data as suggestive of neuroinflammation had features of hyperintensity, hippocampal swelling, or other signs of oedema. Reports including cortical dysplasia, mild involutional change or atrophy, and bleeding were not considered positive. MRI results were not available for NI and viral controls.
Clinical details for AE patients, NI, and OAND cohorts were collected by interviewing treating clinicians and verified through medical records. Clinical data for viral samples were not available.
Analysis of the surrogate markers examined in this study was performed using StataMP 13 and scatterplot figures of results were prepared using GraphPad Prism.
For continuous independent variables, univariate analysis using the Mann-Whitney
Significant findings were then combined in a multivariate logistic model to determine markers that were significantly and independently associated with disease group classification (AE vs. viral, NI, and OAND groups). These markers were then fitted to a predictive model and a receiver operating characteristic (ROC) curve created.
A total of 32 patients with a high clinical probability of AE were recruited. These were subdivided into nine AbPAE patients and 23 AbNAE patients. Ten NI controls, 24 viral controls, and five OAND were also included in the analysis (
Patient Demographic Details.
Number of patients | 10 | 4 | 9 | n/a | 23 | n/a | n/a |
Median age | 54 | 46.5 | 37 | 0.98 | 44 | 0.4 | 0.2 |
Age range | 17–81 | 19–60 | 15–58 | n/a | 18–73 | n/a | n/a |
Gender (M:F) | 4:6 | 1:3 | 3:6 | 0.4 | 16:7 | 0.1 | 0.02 |
Most investigations were analyzed in over 90% of samples from the AE groups. Exceptions were CSF GAD (78%) and serum VGKC (59%).
Of the 10 NI controls, five were perioperative patients where only 1 mL of CSF was able to be collected. These samples were reserved for assessment of novel markers.
Two of the OAND controls had commenced immunosuppression at time of lumbar puncture: one cerebral lupus (methotrexate and mycophenolate) and one with cerebral vasculitis (pulsed methylprednisolone). However, both these patients required intensification of immunosuppression prior to remission.
Twenty-five PCR positive viral controls were included in this study. One was a recruited patient diagnosed with herpes simplex virus (HSV) positive on PCR. Twenty-four others were obtained from aliquoted stored samples and consisted of nine enterovirus (ENTV) positive samples, four HSV positive, three Epstein-Barr virus (EBV) positive, and eight varicella-zoster virus (VZV) positive samples. All ENTV samples were 500 μL in volume and were used for both CSF cytokine and CSF light chain analysis. There were only 200 μL of CSF for VZV, EBV, and HSV samples. Therefore, 5 HSV, 3 EBV, and 4 VZV samples were used for cytokine analysis and a further 4 VZV samples were used CSF light chain analysis. Thirteen viral samples also underwent IIF on primate brain.
There was a trend for increased proportions of positive results in some conventional CSF markers of CNS inflammation in patients with AE and OAND compared to NI controls (
Conventional Markers of CNS inflammation.
CSF monocytosis > 5 (2) | 0 (0%) | 1 (20%) | 2 (22%) | n/a | 4 (17%) | n/a |
CSF oligoclonal bands | 0 (0%) | 3 (60%) | 6 (67%) | n/a | 5 (22%) | n/a |
CSF protein |
3 (60%) | 2 (40%) | 4 (44%) | 0.5 | 9 (32%) | 0.6 |
Neuronal IIF (any staining) | 1 (20%) | 2 (40%) | n/a | n/a | 8 (35%) | 0.4 |
MRI changes | n/a | 4 (80%) | 3 (33%) | n/a | 5 (22%) | n/a |
Two of 9 (22%) of AbPAE and 4/23 (17%) of AbNAE patients had evidence of CSF monocytosis >5 monocytes. Six of 9 (67%) of the APAE and 4/23 (17%) of AbNAE high-risk patients had evidence of CSF oligoclonal bands. None of the NI group had CSF monocytosis >5 or oligoclonal bands and these markers were not able to be statistically analyzed.
Raised CSF protein (>0.45 g/L) was seen in 4/9 (44%) of the AbPAE and 9/23 (32%) of AbNAE groups but also in 3/5 (60%) of the NI group.
CSF neuronal IIF staining was observed in 8/23 (35%) of the AbNAE group but was also observed in 1/5 of the NI group, 1/13 (8%) of the viral controls with sufficient sample for testing (enterovirus only) and 2/5 (40%) of OAND controls. The viral sample with non-specific IIF staining was EBV positive on PCR however this was supplied as a deidentified aliquot and further clinical details could not be verified.
Three of 9 (33%) of the AbPAE group and 5/23 (22%) of the AbNAE group had non-specific changes on MRI indicative of neuroinflammation. No MRI results were available for viral or NI controls.
Therefore, whilst markers such as CSF oligoclonal bands, monocytosis or presence of MRI changes may indicate an autoimmune process, these are not sensitive or specific (
Heatmap cluster analysis revealed differential profiles of cytokine concentrations in patients with viral infections and NI controls compared to the combined AbPAE, AbNAE group, and OAND groups (
CSF cytokines.
There were no significant differences in cytokine levels between the AbPAE and AbNAE groups. Therefore, for statistical analysis, the AE patients were analyzed as one group when compared to NI and viral controls. The cytokines IL1b and IL12p70 were raised in the AE group when compared to the OAND group. Results of univariate analysis between individual groups are detailed in
Univariate analysis found that levels of IL21 (
A multivariate logistic regression model was used to compare the combined cohort with a combined group of normal, viral, and OAND controls. Only IL-21 (
Only CSF lambda light chains were higher in both AE (
The diagnosis of antibody-negative AE remains largely one of exclusion (
The most promising surrogate marker for AE is IL21 which was raised in both AbPAE and AbNAE groups. While the detected levels of IL-21 in the CSF were in the low range of the assay (0–6 pg/ml) they were significantly increased compared to viral and normal controls. Considering the short serum half-life of IL-21 (1–3 h) (
IL21 has many roles in B, CD8 T, and NK cell activation. In B cells, IL-21 acts as both an inhibitor and activator (
Antibodies associated with AE are continually being described. It is possible that patients diagnosed with antibody-negative AE may have an antibody that is yet to be discovered. Another consideration is that the finding of higher IL21 indicates a role in non-antibody mediated inflammation. IL21 down regulates FOXP3+ regulatory T cells leading to enhanced autoimmunity (
The main mimic of AE is viral encephalitis which is an important consideration in the context of potential immunosuppression. The main cytokine indicative of viral infection in our study was IP10/CXCL10. IP10/CXCL10 is secreted in response to interferon gamma (
Translating these findings into routine clinical practice, IL-21 and IP10 may contribute to the diagnostic armamentarium in the investigation of encephalitis, possibly helping to differentiate AE from conditions presenting in a similar fashion where immunosuppression may be harmful. A pragmatic way of comparing these values may be through an IL-21/IP10 ratio. In our cohort, this ratio had an excellent AUC on ROC curve analysis when AE was compared to NIND and viral controls, but this needs to be further validated.
Data available on CSF cytokines in this disease setting are limited and comparisons between studies are difficult because of heterogeneity in disease definitions and differences between cytokine detection platforms and their lower limit of detection, as well as kit manufacturers.
There is only one other study to our knowledge that examines CSF cytokine profiles in adults. This study examined CSF cytokines in 78 patients using a different platform manufacturer (Bio-Rad), including 20 with an autoimmune neurological disease (
There is more literature available about CSF cytokines associated with AE in children, but it is unclear if these data are applicable to an adult population. One study (of children aged 28 days-14 years old) examined CSF cytokines (Bio-Rad kit) in viral encephalitis compared with NMDAR (four patients) encephalitis and found significant elevations in IL-6, IL7, and IL13 in the viral encephalitis group compared to the NMDAR encephalitis group (
Our study did not find other novel potential novel markers useful in differentiating AE from NI or viral controls. In examining the literature, raised CSF free light chains (FLC) levels have been associated with neuroinflammation, however, the normal range is not well-established (
Currently utilized conventional markers are neither sensitive (
This study was limited by small patient numbers and limited CSF volumes, reflecting the rarity of this disease. There was difficulty in obtaining sufficient normal and viral control samples of CSF and there were pre-analytical collection issues in this study. The use of deidentified viral PCR positive CSF aliquots for controls meant clinical correlation was not possible. We did not check serum cytokines for patients in our cohort and comparison between CSF and serum cytokine levels in these patients should be a focus of further study.
Nevertheless, this is the first study, to our knowledge, to prospectively examine both conventional and novel markers of neuroinflammation in these groups of adult patients prior to immunosuppression. We have demonstrated the CSF cytokines CXCL10/IP-10 and IL-21 are potential differentiators of AE from viral encephalitis, particularly when there is no CNS specific autoantibody detected. These novel markers could have a future role to help expediate the decision to commence immunosuppression in this group of patients warranting their prospective validation in separate cohorts of AE patients.
The datasets generated for this study are available on request to the corresponding author.
This study was approved by the Ethics Committee of Westmead Hospital (LNR/16/WMED/192) and written informed consent was obtained by all participants.
JJ: study design, patient recruitment, data extraction, cytokine assays, statistical analysis, drafting, and editing of manuscript. NF: cytokine assays, heatmap production, and editing of manuscript. SD: cytokine assays and editing of manuscript. AF, SRD, and MS: patient recruitment and editing of manuscript. SR and AB: initial study design and editing of manuscript. SB: assistance with assays and editing of manuscript. BJ: recruitment of controls and editing of manuscript. DB and M-WL: study design, patient recruitment, and editing of manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We thank the clinicians of the Department of Neurology at Westmead Hospital: Victor Fung, Chong Wong, Andrew Duggins, Andrew Henderson, Neil Mahant, Steve Vucic, and Maori Wijayath. We thank the scientists of the immunopathology department at ICPMR Pathology West, Westmead hospital for their expertise in the assays: Enny Soesilowati, Suzanne Cullican, David Macdonald, David Campbell, Fakria (Hila) Kakkar, and Hong Wing. We thank the Infectious Disease and Microbiology department at Westmead Hospital: Jen Kok, John Iredell, and Rita Baini for provision of the viral control samples. We thank Louise Weinholdt on suggestions for investigations examined in this study. Data acquisition was performed in the Flow Cytometry Core Facility that was supported by Westmead Institute, Westmead Research Hub, Cancer Institute New South Wales, and National Health and Medical Research Council.
The Supplementary Material for this article can be found online at: