Kappa Free Light Chain for Diagnosis and Neurofilament Light Chain for Prognosis in Autoimmune Encephalitis

Abstract

Background and Objectives

Reliable biomarkers for autoimmune encephalitis (AE) are limited, and emerging CSF markers are not incorporated into current diagnostic criteria. Prognostic tools remain insufficient, highlighting the need for biomarkers that support both early diagnosis and assessment of disease severity and prognosis.

Methods

In this multicenter prospective cohort study, we analyzed clinical data and paired CSF-serum samples from adults with definite AE enrolled in the German Network for Research on Autoimmune Encephalitis registry and the CSF biobank of Hannover Medical School. Of 2,330 screened individuals, 92 patients with anti–N-methyl-d-aspartate receptor (NMDAR, n = 53), anti–leucine-rich glioma-inactivated 1 (LGI1, n = 20), or anti–contactin-associated protein-like 2 (CASPR2, n = 19) encephalitis were included and followed longitudinally for a median of 38 months. Control groups comprised patients with relapsing multiple sclerosis, varicella-zoster virus encephalitis, and noninflammatory neurologic conditions (each n = 30), as well as antibody-positive patients without AE (n = 15), with the groups frequency-matched for age and sex. Kappa free light chain (KFLC), neurofilament light chain (NfL), glial fibrillary acidic protein (GFAP), and cytokines were measured in paired CSF-serum samples obtained at baseline and during follow-up. Disease severity and disability were assessed using the Clinical Assessment Scale in Autoimmune Encephalitis (CASE) score and the modified Rankin Scale (mRS).

Results

Intrathecal synthesis of KFLC was detected in 94% of anti-NMDAR, 50% of anti-LGI1, and 53% of anti-CASPR2 encephalitis cases, demonstrating higher diagnostic sensitivity than CSF-restricted oligoclonal bands or pleocytosis. Diagnostic specificity across pooled control groups was moderate at 44% but reached 90% when compared with noninflammatory neurologic controls. CSF NfL z-score levels were strongly associated with baseline disease severity, with each 1-standard deviation increase corresponding to an approximately 10-point higher CASE score, independent of clinical covariates (β = 0.61). Longitudinal changes in NfL concentrations in CSF and serum were associated with disease severity and neurologic disability at follow-up (CASE score: adjusted R² = 0.401; mRS score: adjusted R² = 0.203). GFAP and cytokines showed limited diagnostic or prognostic utility.

Discussion

Intrathecal KFLC synthesis represents a highly sensitive CSF marker that supports early suspicion of autoimmune encephalitis and prompts antibody testing. NfL provides robust biochemical information on baseline disease severity and longitudinal changes that may aid prognostic assessment across AE subtypes.

Introduction

Autoimmune encephalitides (AEs) are immune-mediated CNS disorders with heterogeneous phenotypes.¹ The most common subtypes are associated with antibodies against the N-methyl-d-aspartate receptor (NMDAR), leucine-rich glioma-inactivated 1 (LGI1), and contactin-associated protein-like 2 (CASPR2).² While recognition has improved, diagnosis and prognostication remain challenging; therefore, the search for reliable biomarkers has become a central focus in AE research.^1-3

Kappa free light chains (KFLCs) in CSF, a byproduct of immunoglobulin (Ig) synthesis, are sensitive markers of intrathecal Ig production and have been incorporated into the 2024 revised McDonald criteria for multiple sclerosis (MS).^4,5 Only 2 studies comprising few antibody-positive patients with AE (anti-NMDAR, n = 15; anti-LGI1, n = 15; anti-CASPR2, n = 6; anti-GAD65, n = 3; anti-GABA(a), n = 2; anti-AMPAR, n = 1; anti-IgLON5, n = 1) investigated KFLC, showing higher sensitivity for detecting intrathecal KFLC synthesis than CSF-restricted oligoclonal bands (OCBs) and pleocytosis.^6,7

Neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP) are structural proteins released during neuroaxonal injury and astroglial activation and are being established as prognostic biomarkers in autoimmune diseases.⁸ In AE cohorts, elevated NfL and GFAP in CSF correlated with disease activity and disability but were diagnostically relevant only for differentiating noninflammatory diseases.^8-22 In anti-NMDAR encephalitis, higher NfL levels were associated with severity and, in some studies with long-term outcomes, with delayed peaks paralleling hippocampal atrophy.^9-14,22 In anti-LGI1 encephalitis, serum NfL was consistently elevated and associated with poor outcome, while combining NfL/GFAP improved prognostic accuracy.^15-18,22 In addition, in anti-IgLON5 disease, a clear correlation of serum NfL and mortality was demonstrated.^20,21 A recent meta-analysis underscored both the promise and heterogeneity of NfL as a biomarker across AE subtypes.²²

Cytokines and chemokines provide complementary insights into immune activation. In anti-NMDAR AE, CSF IL-6, IL-17A, and CXCL10 correlated with inflammation and clinical status.²³ In anti-LGI1 and anti-CASPR2 AE, IL-6, IL-8, IL-17A, and CXCL13, as well as adhesion molecules, were associated with relapse or acute severity.^24-27

This multicenter study explored the diagnostic and prognostic potential of KFLC, NfL, GFAP, and cytokine profiles in AE.

Methods

Patients

The prospective registry of the German Network for Research on Autoimmune Encephalitis (GENERATE) was screened for patients with anti-NMDAR, anti-LGI1 and anti-CASPR2 encephalitis presenting between 2014 and 2024. Follow-up visits included at least annual clinical assessments, laboratory analyses, and MRI (1.5–3T) interpreted by local neuroradiologists. Additional patients meeting the criteria were included from the Hannover Medical School CSF biobank.

Inclusion criteria were as follows: (1) definite anti-NMDAR, anti-LGI1, or anti-CASPR2 AE applying currently available diagnostic criteria¹; (2) age >18 years; (3) written informed consent; (4) ≥ 12 months of follow-up; (5) availability of paired CSF/serum samples; and (6) no established immunosuppression at first sampling.

Exclusion criteria were as follows: (1) antibody detection without meeting definite AE criteria; (2) missing paired samples; (3) corticosteroid therapy or plasma exchange performed <4 weeks prior to sampling; (4) prior second-line immunotherapy (rituximab, cyclophosphamide, bortezomib) or IVIg.¹

Controls included the following: (1) patients with relapsing MS per 2017 revised McDonald criteria (n = 30); (2) patients with varicella-zoster virus (VZV) encephalitis confirmed by PCR or antibody index fulfilling the criteria by Venkatesan et al. (n = 30); (3) noninflammatory neurologic controls (NINCs; idiopathic intracranial hypertension or normal pressure hydrocephalus, n = 30); (4) antibody-positive patients not fulfilling the criteria for definite AE (AB, n = 15; anti-NMDAR, n = 8; anti-CASPR2, n = 7).^28,29 Samples obtained from immunologically untreated patients and control groups were frequency-matched for age and sex to the overall AE cohort.

Data Collection and Clinical Variables

Baseline was defined as the first clinical presentation. Collected data included demographics, tumor screening, disease duration, modified Rankin Scale (mRS) score, CASE score (Clinical Assessment Scale in Autoimmune Encephalitis), NEOS score (anti-NMDAR encephalitis one-year functional status), immunotherapy, intensive care unit (ICU) requirement, and CSF parameters (leucocyte count, lactate, total protein, CSF/serum albumin quotient, OCB).³⁰

Laboratory Analyses

CSF routine analyses, as well as preanalytics, sampling, and biobanking, followed international guidelines and standard protocols.³⁰ Antibodies were tested in commercial laboratories or in-house laboratories of the recruiting local GENERATE centers using commercial cell-based assays (Euroimmun, Lübeck, Germany) and confirmed by rat brain immunohistochemistry.¹ KFLC was measured using the N Latex FLC Kappa Kit (Siemens Healthineers, Marburg, Germany) on a Neph Atellica 630 System (Siemens) at Hannover Medical School, and intrathecal KFLC synthesis was detected using the formulas published by Reiber et al.³¹ In brief, the assessment of intrathecal KFLC synthesis was based on CSF/serum quotient diagrams (Reiber's diagram), which account for blood-CSF barrier function by linking the CSF/serum KFLC concentration quotient to the CSF/serum albumin concentration quotient. A hyperbolic reference range was applied to distinguish between blood-derived and intrathecally synthesized KFLC.³¹ The upper reference limit (Q_lim) was calculated according to an established formula: Qlim(KFLC) = (3.27 × √(QAlb² + 33) − 8.2) × 10⁻³, and values exceeding this limit were considered indicative of intrathecal synthesis. This approach allows a nonlinear correction for blood-CSF barrier permeability and distinguishes intrathecal production from passive diffusion.³¹ NfL and GFAP concentrations were measured using the ELLA automated immunoassay platform (Simple Plex Human Cartridge; ProteinSimple, Bio-Techne, San Jose, CA) and were transformed into age-adjusted z-scores to account for age-dependent reference values.^32-34

Considering potential influencing factors of KFLC and NfL concentrations, information on renal function and ICU treatment at the time of sampling was collected.³⁵ No patient showed evidence of acute renal failure at the time of sampling, and none had markedly reduced renal function (estimated glomerular filtration rate <60 mL/min).³⁵ Furthermore, although a subset of patients required ICU treatment, no cases of cerebral hypoxia were documented at the time of sample collection. Accordingly, CSF and serum NfL concentrations did not differ between patients with and without ICU treatment.

Cytokines were measured using a Luminex-based multiplex assay (Human Th17-plex, Hercules, CA) following standardized preanalytical procedures: interleukin-1β (IL), interleukin-4, interleukin-6, interleukin-10, interleukin-17A, interleukin-17F, interleukin-21, interleukin-22, interleukin-23, interleukin-25, interleukin-31, interleukin-33, interferon gamma (IFNγ), tumor necrosis factor alpha (TNFα), and soluble activation marker CD40 ligand (sCD40L).³⁶ Six CSF (IL-4, IL-6, IL-31, IFNγ, TNFα, sCD40L) and 5 serum (IL-6, IL-17A, IL-23, IFNγ, TNFα) cytokines were consistently detectable.³⁶

A second longitudinal CSF/serum analysis was conducted in 49 patients (32/49 with NMDAR, 8/49 with LGI1, 9/49 with CASPR2) after a median time of 260 days (interquartile range [IQR]: 83–647 days) to evaluate possible ongoing disease activity. As sensitivity analysis, patients with longitudinal CSF samples were stratified according to the interval between first and second lumbar puncture using the 75th percentile as cutoff. Longitudinal changes in CSF and serum NfL and GFAP were compared between patients with shorter and longer sampling intervals.

Statistical Analysis

Statistical analyses were performed using IBM SPSS Statistics 29.0 (IBM Corp., Armonk, NY). Continuous variables were reported as median (IQR) and categorical as counts (percentages). Normality was tested with the Shapiro-Wilk test. Inter-group comparisons used the t test or Mann-Whitney U test, categorical data χ² or Fisher exact test, and longitudinal paired comparisons were performed using non-parametric Wilcoxon signed-rank tests, irrespective of distributional assumptions. To account for multiple testing in exploratory cytokine and chemokine analyses, false discovery rate (FDR) correction was performed using the Benjamini-Hochberg procedure across all comparisons within each table. Receiver operating characteristic (ROC) analysis was performed to assess the diagnostic performance of the KFLC index, with calculation of the area under the curve (AUC) and determination of the optimal cutoff using the Youden index. Associations between biomarkers, disease severity, and outcomes were explored using multivariable linear regression including sex, age, disease duration, CSF lactate concentration, KFLC intrathecal fraction, cytokines, and NfL and GFAP z-scores. AE subtype was included as a categorical covariate with Bonferroni-corrected post hoc pairwise comparisons to assess potential subgroup effects. Although the CASE score is ordinal by definition, it was analyzed as a quasicontinuous variable, consistent with prior clinical biomarker studies, and linearity was supported by comparable Pearson and Spearman correlations, visual inspection, and consistent results using log-transformed NfL values. The dependent variable was CASE score at baseline. In separate models, longitudinal biomarker changes (“delta”: value at second analysis - value at first CSF analysis; NfL, GFAP) were modeled against CASE and mRS scores at second lumbar puncture in patients with paired CSF/serum samples only with and without inclusion of baseline disease severity as covariate. Univariate models assessed the associations of KFLC intrathecal fraction with the NEOS score and serum GFAP z-scores with the use of second-line immunotherapies. Two-sided p < 0.05 was considered significant.

Standard Protocol Approvals, Registrations, and Patient Consents

Data were obtained through the GENERATE network, a multicentric registry of AE competence centers in Germany, Austria, and Switzerland, or from the biobank of the Laboratory for CSF Diagnostics and Neurochemistry at Hannover Medical School. Pseudonymized patient data were collected with approval from the ethics committees of the University of Lübeck (reference 13–162) and all participating sites. Written informed consent was obtained from all patients or their legal representatives. This study followed the STROBE reporting guideline.

Data Availability

Anonymized data not published within this article will be made available by request from any qualified investigator.

Results

Patients

A total of 92 patients with definite AE were included in the final analysis: 53 with anti-NMDAR (58%), 20 with anti-LGI1 (22%), and 19 with anti-CASPR2 (21%; eFigure 1). Patients with anti-LGI1 and anti-CASPR2 antibodies were significantly older (p = 0.0034), while mRS (p = 0.0078) and CASE (p = 0.0006) scores at disease onset and at their worst were significantly higher in individuals with anti-NMDAR antibodies compared with those with anti-LGI1 antibodies. Immunotherapy was initiated earlier in patients with anti-NMDAR (p = 0.0028) and anti-LGI1 (p = 0.0036) AE compared with those with anti-CASPR2 AE. Second-line immunotherapies were more frequently used in patients with anti-NMDAR (p = 0.0041) and anti-LGI1 (p = 0.0225) AE than in those with anti-CASPR2 AE. Patients with AE were followed over a median time of 1,164 days (IQR: 627–2,258). Clinical, demographic, and treatment data are given in Table 1.

Table 1.

Demographic, Clinical, and Therapeutic Data

Characteristic	NMDAR-E (n = 53)	LGI1-E (n = 20)	CASPR2-E (n = 19)	p Value
Demographic data
Age at first presentation [y], median (IQR)	25 (21–39)	66 (58–72)	60 (44–70)	<0.0001
Symptom onset to first CSF analysis [d], median (IQR)	26 (10–79)	52 (12–150)	136 (42–317)	0.0577
Women, n (%)	36 (68)	10 (50)	1 (5)	0.0788
Follow-up since diagnosis [d], median (IQR)	994 (448–2075)	895 (509–2,411)	725 (395–2,293)	0.7491
Clinical characteristics
mRS score at first presentation, median (IQR)	3 (2–3)	2 (1–2)	2 (2–3)	0.0150
mRS score at second CSF analysis, median (IQR)a	2 (1–4)	2 (1–2)	2 (1–4)	0.5677
mRS score at last follow-up, median (IQR)	1 (0–2)	1 (1–2)	1 (0–2)	0.7482
CASE score at first presentation, median (IQR)	6 (3–14)	3 (2–5)	4 (3–6)	0.0011
CASE score at second CSF analysis, median (IQR)	3 (1–7)	4 (2–4)	2 (1–4)	0.9766
CASE score at last follow-up, median (IQR)	1 (0–3)	1 (1–3)	1 (0–2)	0.7975
Underlying tumor or teratoma identified, n (%)	22 (42)	0	2 (11)	0.3114
Therapeutic data
ICU treatment, n (%)	17 (32)	1 (5)	0	0.0348
Symptom onset to immunosuppressive treatment [d], median (IQR)	27 (11–92)	54 (10–214)	181 (55–356)	0.0100
IVMP applied, n (%)	46 (87)	19 (95)	17 (89)	0.6027
IVIG applied, n (%)	22 (42)	4 (20)	8 (42)	0.2064
PE/IA applied, n (%)	28 (53)	5 (25)	5 (26)	0.0324
SLI applied, n (%)	42 (79)	16 (80)	8 (42)	0.0056
Immunosuppressive combination therapy, n (%)	47 (89)	17 (85)	13 (68)	0.1201

Abbreviations: CASE = clinical assessment scale in autoimmune encephalitis; E = encephalitis; IA = immunoadsorption; ICU = intensive care unit; IQR = interquartile range; IVIG = IV immunoglobulin; IVMP = IV methylprednisolone; IQR = interquartile range; LGI1 = leucine-rich glioma inactivated 1, CASPR2 = contactin-associated protein-like 2; mRS = modified Rankin Scale; NMDAR = N-methyl-d-aspartate receptor; PE = plasma exchange; SLI = second-line immunotherapy (rituximab, cyclophosphamide, bortezomib).

^aData for second CSF analysis for n = 49. p Values for comparison between 3 groups (ANOVA) are provided.

As provided in Table 2, CSF leukocyte count, pleocytosis, total protein, albumin quotient, and OCB differed significantly among subtypes at both CSF analyses. In anti-NMDAR AE, significant lower values or frequencies of pathologic results at the second CSF analysis during follow-up were observed for cell count (p = 0.0272), pleocytosis (p = 0.0015), albumin quotient (p = 0.0019), and OCBs (p = 0.0405) compared with the first CSF analysis at disease onset. In anti-LGI1 AE, significantly lower CSF cell counts were in the second CSF analysis compared with the first analysis (p = 0.0464). Patients with AE showed 8.7-fold lower CSF leucocyte counts and a 2.2-fold lower frequency of pleocytosis than patients with VZV encephalitis, as well as 2-fold fewer cases with OCBs than patients with MS. By contrast, patients with AE showed higher values for these parameters compared with NINCs. Comparison of CSF routine analysis parameters with matched control groups is given in eTable 1.

Table 2.

CSF Routine Analyses

CSF characteristic	NMDAR-E (n = 53)	LGI1-E (n = 20)	CASPR2-E (n = 19)	p Value
CSF analysis at first presentation
Leukocyte count [n/μL], median (IQR)	11 (2–38)	2 (1–3)	2 (1–5)	0.0004
Pleocytosis, n (%)	33 (62)	2 (10)	5 (26)	<0.0001
Total protein [mg/L], median (IQR)	377 (282–482)	460 (378–550)	495 (386–706)	0.0179
Total protein elevated [>500 mg/L], n (%)	11 (21)	7 (35)	9 (47)	0.0753
Albumin quotient (CSF/serum) x103, median (IQR)	4.8 (3.7–6.4)	6.9 (5.6–9.0)	8.2 (5.6–12.0)	0.0011
Albumin quotient elevated according to age-adjusted reference ranges >4 + (age/15 × 103), n (%)	51 (96)	17 (85)	17 (89)	0.2354
CSF-restricted oligoclonal bands, n (%)	36 (68)	1 (5)	8 (42)	<0.0001
Lactate concentration [mmol/L], median (IQR)	1.8 (1.5–2.0)	1.8 (1.6–2.1)	1.6 (1.5–1.9)	0.6343
Second CSF analysis (during follow-up)a
Time from symptom onset to second CSF analysis [d], median (IQR)	392 (119–950)	278 (200–721)	717 (265–1,649)	0.2602
Leukocyte count [n/μL], median (IQR)	2 (1–7)	1 (1–2)	2 (2–8)	0.0208
Pleocytosis, n (%)	8 (25)	0 (0)	3 (33)	0.2179
Total protein [mg/L], median (IQR)	340 (265–462)	520 (460–610)	710 (330–874)	0.0066
Total protein elevated [>500 mg/L], n (%)	2 (6)	4 (50)	5 (56)	0.0009
CSF/serum albumin quotient, median (IQR)	5.0 (3.9–7.0)	8.4 (7.3–11.5)	12.7 (5.6–17.3)	0.0013
CSF/serum albumin elevated, n (%)	23 (72)	4 (50)	5 (56)	0.4037
CSF-restricted oligoclonal bands, n (%)	14 (44)	0	2 (22)	0.0470

Abbreviations: E = encephalitis; IQR = interquartile range; LGI1 = leucine-rich glioma inactivated 1, CASPR2 = contactin-associated protein-like 2; NMDAR = N-methyl-d-aspartate receptor.

^aSamples from a second CSF analysis were available from n = 32 with anti-NMDAR, n = 8 with anti-LGI1, and n = 9 with anti-CASPR2 encephalitis. p values for comparison between 3 groups (ANOVA) are provided.

KFLC Analysis

An intrathecal KFLC synthesis in Reiber's diagram was more frequent in patients with anti-NMDAR (50/53, 94%) than anti-LGI1 (10/20, 50%) and anti-CASPR2 (10/19, 53%) AE in the first CSF analysis (Figure 1A). In the second CSF analysis during follow-up, an intrathecal KFLC synthesis remained more frequent in patients with anti-NMDAR AE than in those with anti-LGI1 and anti-CASPR2 AE (Table 3). An intrathecal KFLC synthesis was more frequent in patients with MS but less common in patients with antibodies without AE and in NINCs (eTable 2, eFigure 2A). Compared with pleocytosis and OCBs, evidence of an intrathecal KFLC synthesis revealed a significantly higher diagnostic sensitivity in patients with definite anti-NMDAR and anti-LGI1 AE (anti-NMDAR: 94% (95% confidence interval (CI)): 85–98%; anti-LGI1: 50%, 95% CI 30–70%; anti-CASPR2: 53%, 95% CI 30%–75%). Inclusion of an intrathecal KFLC synthesis led to a higher yield of inflammatory CSF findings than oligoclonal bands (Figure 1B). The diagnostic specificity of intrathecal KFLC synthesis, calculated against the pooled control cohort, was 44% (95% confidence interval (CI): 34%–54%) and was comparable across AE subtypes. When analyzed by control category, specificity was highest vs NINCs (90%, 95% CI 74%–98%) but was lower vs VZV encephalitis (37%, 95% CI 20%–56%), antibody-positive non-AE cases (53%, 95% CI 27%–79%), and MS (0%, 95% CI 0%–12%). For comparison, the diagnostic specificity of CSF-restricted OCBs, calculated against the same pooled control cohort, was 51% (95% CI 42%–61%). Similarly, specificity was highest vs NINCs (90%, 95% CI 74%–98%), but decreased in VZV encephalitis (57%, 95% CI 39%–74%) and antibody-positive non-AE cases (67%, 95% CI 38%–88%) and was absent in MS (0%, 95% CI 0%–12%). Isolated CSF pleocytosis without oligoclonal bands or intrathecal KFLC synthesis was observed in a small subset of patients. At baseline, this pattern was most frequent in anti-NMDAR AE (n = 5), followed by anti-CASPR2 AE (n = 3) and anti-LGI1 AE (n = 1). Among patients with follow-up CSF analysis, isolated pleocytosis was rare and observed in only one patient each with anti-NMDAR and anti-CASPR2 AE and in none with anti-LGI1 AE.

Figure 1. Kappa Free Light Chain and NfL in Autoimmune Encephalitis.

Shown are (A) the proportion of patients with anti-NMDAR (blue), anti-LGI1 (red), and anti-CASPR2 (black) AE exhibiting intrathecal KFLC synthesis according to Reiber's diagram. Colors correspond to the different AE subtypes as indicated in the legend within the panel. Shown are patients with only a diagnostic CSF analysis available (baseline only), patients with available longitudinal CSF analysis (baseline), and the corresponding follow-up CSF analyses. Numbers indicate the number of patients per subgroup. p values above the lines indicate statistical significance of group comparisons between AE subtypes. (B) Distribution of inflammatory CSF findings across AE subtypes, stratified by CSF pleocytosis with CSF-restricted oligoclonal bands (light blue), CSF-restricted oligoclonal bands only without CSF pleocytosis (dark gray), intrathecal KFLC synthesis with CSF pleocytosis (green), and intrathecal KFLC synthesis only without CSF pleocytosis (light gray), with white indicating no inflammatory CSF findings. Colors correspond to the different patterns of inflammatory CSF findings as indicated in the legend within the panel. Isolated CSF pleocytosis without oligoclonal bands or intrathecal KFLC synthesis was observed in a small proportion of patients (n = 11 overall), most frequently in anti-NMDAR AE at baseline (n = 5; anti-CASPR2 AE n = 3; anti-LGI1 AE n = 1), with only isolated cases at follow-up (anti-NMDAR n = 1; anti-CASPR2 n = 1). (C) Association between baseline CSF NfL z-scores and disease severity at first presentation assessed using the CASE (Clinical Assessment Scale in Autoimmune Encephalitis) score, with the regression line shown for illustrative purposes as part of a multivariable regression analysis. (D) Association between longitudinal changes in serum NfL concentrations (delta serum NfL = second − first CSF analysis) and CASE score at follow-up, analyzed using multivariable regression. For panel C, Pearson (r = 0.15, p = 0.17) and Spearman (ρ = 0.21, p = 0.06) correlation coefficients were comparable, and visual inspection of scatterplots showed no marked deviation from linearity. The association remained significant after log-transformation, supporting an approximately linear, monotonic relationship between CSF NfL and CASE score. AE = autoimmune encephalitis; CASE = Clinical Assessment Scale in Autoimmune Encephalitis; CASPR2 = contactin-associated protein-like 2; KFLC = kappa free light chain; LGI1 = leucine-rich glioma-inactivated 1; NfL = neurofilament light chain; NMDAR = N-methyl-d-aspartate receptor.

Table 3.

Kappa Free Light Chains in Patients With Autoimmune Encephalitis

	NMDAR-E (n = 53)	LGI1-E (n = 20)	CASPR2-E (n = 19)	p Value
CSF analysis at first presentation
KFLC IF [Reiber's diagram], median (IQR)	78.1 (47.3–91.8)	5.3 (0–36.4)	5.8 (0–65.4)	<0.0001
Intrathecally synthesized KFLC [local concentration (mg/L), according to Reiber's diagram], median (IQR)	0.71 (0.16–2.67)	0.01 (0–0.15)	0 (0–0.52)	<0.0001
CSF KFLC concentration [mg/L], median (IQR)	0.90 (0.34–3.25)	0.31 (0.17–0.51)	0.43 (0.20–0.97)	0.0007
Serum KFLC concentration [mg/L], median (IQR)	10.7 (8.2–14.6)	11.9 (7.9–41.3)	11.2 (10.0–18.4)	0.3801
Second CSF analysis (during follow-up)a
KFLC IF [Reiber's diagram], median (IQR)	18.4 (0–41.5)	0	0 (0–34.2)	0.0208
Intrathecally synthesized KFLC [mg/L, according to Reiber's diagram], median (IQR)	0.02 (0–0.39)	0	0 (0–0.15)	0.0201
CSF KFLC concentration [mg/L], median (IQR)	0.13 (0.08–0.61)	0.12 (0.08–0.14)	0.32 (0.19–1.07)	0.1193
Serum KFLC concentration [mg/L], median (IQR)	8.6 (6.9–11.1)	10.4 (8.0–15.2)	15.3 (6.6–25.0)	0.2426

Abbreviations: CASPR2 = contactin-associated protein-like 2; E = encephalitis; IF = intrathecal fraction; IQR = interquartile range; KFLC = kappa free light chain; LGI1 = leucine-rich glioma inactivated 1; NMDAR = N-methyl-d-aspartate receptor.

^aSamples from a second CSF analysis were available from n = 32 with anti-NMDAR, n = 8 with anti-LGI1, and n = 9 with anti-CASPR2 encephalitis. p Values for comparison between 3 groups (ANOVA) are provided.

In addition, the linear KFLC index was evaluated as an alternative marker of intrathecal immunoglobulin synthesis. Applying the recommended cutoff of >6.1 for diagnosing MS, elevated KFLC index values were observed less frequently than intrathecal KFLC synthesis defined by Reiber's diagram.⁵ At baseline, an increased KFLC index was present in 40 (75%) of 53 patients with anti-NMDAR AE, 5 (25%) of 20 with anti-LGI1 AE, and 6 (32%) of 19 with anti-CASPR2 AE. During follow-up, elevated KFLC index values were less common (anti-NMDAR: 6/32, 19%; anti-LGI1: 0/8, 0%; anti-CASPR2: 1/9, 11%). ROC analysis of the KFLC index showed limited diagnostic performance, with an AUC of 0.52. The optimal cutoff determined by the Youden index was 3.3, yielding a sensitivity of 80% and a specificity of 40%. Given the lack of discriminative performance, deriving an AE-specific cutoff does not appear clinically meaningful. Overall, the diagnostic performance of the KFLC index was inferior to intrathecal KFLC synthesis assessed by Reiber's diagram.

NfL and GFAP Analyses

CSF NfL and serum GFAP concentrations at the first and second analyses were lower in patients with anti-NMDAR compared with patients with anti-LGI1 and anti-CASPR2 encephalitis (Table 4, eTable 3, eFigure 3). Patients with anti-CASPR2 AE showed higher CSF GFAP at follow-up (p = 0.0156). Although the interval between first and second CSF analysis was numerically longer in anti-CASPR2 AE compared with other subtypes, this difference was not statistically significant. Moreover, sensitivity analyses stratifying patients according to the interval between lumbar punctures (≤ vs >75th percentile) did not reveal differences in longitudinal changes of CSF or serum NfL or GFAP, indicating that these findings were not driven by variability in follow-up timing. Testing continuous CSF and serum NfL z-scores against the reference value of zero (Wilcoxon signed-rank tests) revealed significantly higher values for each AE subtype and for the pooled cohort (p < 0.0001). Patients with AE had higher concentrations and z-scores than NINCs and higher serum GFAP than the MS group (eTable 4). Elevated CSF GFAP and serum NfL z-scores >2 were more frequent in patients with VZV encephalitis compared with patients with AE, while CSF GFAP was lower in AB patients without AE.

Table 4.

Neurofilament Light Chains and Glial Fibrillary Acidic Protein in Patients With Autoimmune Encephalitis

	NMDAR-E (n = 53)	LGI1-E (n = 20)	CASPR2-E (n = 19)	p Value
CSF analysis at first presentation
CSF NFL concentration [pg/mL], median (IQR)	565 (197–1,065)	1,152 (722–2,212)	1,232 (592–3,353)	0.0146
Serum NFL concentration [pg/mL], median (IQR)	25.6 (9.8–61)	51.4 (17.9–63)	23 (17.2–59)	0.5167
CSF GFAP concentration [mg/L], median (IQR)	186 (68–403)	294 (138–453)	279 (137–551)	0.0665
Serum GFAP concentration [pg/mL], median (IQR)	5.1 (5.0–10.8)	12.4 (7.7–21.9)	8.8 (5.0–17)	0.0062
Second CSF analysis (during follow-up)a
CSF NFL concentration [pg/mL], median (IQR)	306 (212–1,024)	1,099 (625–1,654)	1,406 (809–3,098)	0.0157
Serum NFL concentration [pg/mL], median (IQR)	12 (7–68)	24 (17–40)	34 (20–91)	0.1488
CSF GFAP concentration [mg/L], median (IQR)	183 (117–346)	346 (238–720)	652 (389–1,507)	0.0022
Serum GFAP concentration [pg/mL], median (IQR)	5.04 (5.0–10.3)	7 (5.04–27)	12.1 (5.5–37.3)	0.1027

Abbreviations: CASPR2 = contactin-associated protein-like 2; E = encephalitis; GFAP = glial fibrillary acidic protein; IQR = interquartile range; LGI1 = leucine-rich glioma inactivated 1; NFL = neurofilament light chain; NMDAR = N-methyl-d-aspartate receptor.

^aSamples from a second CSF analysis were available from n = 32 with anti-NMDAR, n = 8 with anti-LGI1, and n = 9 with anti-CASPR2 encephalitis. p Values for comparison between 3 groups (ANOVA) are provided.

Cytokine and Chemokine Analyses

In serum, IL-31 and sCD40L were higher in patients with anti-LGI1 encephalitis at first analysis, but these differences did not remain statistically significant after FDR correction (eTable 5, eFigure 4). In patients with anti-NMDAR encephalitis, serum IL-6 concentrations decreased at second CSF analysis (p = 0.0395). Compared with patients with VZV encephalitis, patients with AE showed lower CSF IL-17A, IL-23, IFNγ, and TNFα concentrations, although only IL-17A remained significant after FDR correction (eTable 6). Compared with patients with MS, higher serum TNFα and CSF IL-6 levels remained significant after FDR correction, whereas other differences did not. NINCs showed lower cytokine levels overall, but after correction, significant differences were retained mainly for serum TNFα and CSF IL-6.

Associations With Disease Severity, Treatment, and Outcome

In multivariate regression (adjusted R² = 0.71 for the overall model, p = 0.002; eTable 7), higher CASE scores at first presentation were associated with higher CSF NfL z-scores (p < 0.001), serum NfL z-scores (p = 0.013; eFigure 2B), and IL-31 (p = 0.012); lower serum sCD40L (p = 0.026) and CSF lactate (p = 0.023); and younger age (p = 0.036). CSF NfL showed the strongest association with baseline CASE scores (unstandardized regression coefficient B = 10.48, 95% CI 5.87–15.09; standardized regression coefficient β = 0.61; Figure 1C). Although AE subtype appeared as a significant covariate in the initial model, post hoc analyses revealed that this effect was not independent after adjustment for biomarker and demographic covariates.

In univariate models, KFLC IF correlated with NEOS scores in the anti-NMDAR AE group (β = 0.34, R² = 0.095, p = 0.014) and serum GFAP z-scores were associated with use of second-line immunotherapies (β = 0.22, R² = 0.036, p = 0.039).

Models considering longitudinal changes in CSF and serum NfL and GFAP (delta NfL, delta GFAP) as predictors were significantly associated with CASE scores (unadjusted R² = 0.204, p = 0.032) and mRS scores at second analysis (unadjusted R² = 0.177, p = 0.033), and these associations remained consistent across AE subtypes (p = 0.152, p = 0.708, respectively). To address potential confounding by baseline severity, multivariate models including baseline CASE and baseline mRS scores were applied, improving model fit substantially (CASE: adjusted R² = 0.401, p < 0.001; mRS: adjusted R² = 0.283, p = 0.004). Considering variability in follow-up time, delta of CSF NfL concentrations (CASE score: β = −0.43, p = 0.047; mRS: β = −0.51, p = 0.015; eFigure S2) and serum NfL concentrations (CASE score: β = −0.59, p = 0.009, Figure 1D; mRS: β = −0.73, p < 0.001, eFigure 2D) were significantly associated, while delta GFAP concentrations were not (delta CSF GFAP: p = 0.497 and 0.437, for CASE scores and mRS scores, respectively; delta serum GFAP: p = 0.936 and 0.336, for CASE scores and mRS scores, respectively).

Discussion

This multicenter study within the GENERATE network systematically evaluated the role of KFLC, NfL, GFAP, and cytokine profiles in anti-NMDAR, anti-LGI1, and anti-CASPR2 encephalitis. Routine CSF parameters align with previous studies from the network, with slight differences compared with other cohorts.³⁷

The first key result is that intrathecal KFLC synthesis demonstrated markedly higher sensitivity than the established CSF markers (pleocytosis, OCBs) currently included in diagnostic criteria.¹ This is especially important in the clinical phase before an antibody-based diagnosis has been reached. In our cohort, an intrathecal KFLC synthesis was detected in approximately half of patients with anti-LGI1 and anti-CASPR2 AE and in 94% of patients with anti-NMDAR AE, significantly outperforming CSF pleocytosis and OCBs in anti-NMDAR and anti-LGI1 AE. This mirrors findings in patients with MS, in whom intrathecal KFLC synthesis demonstrated a very high diagnostic sensitivity by reflecting an immunoglobulin production of all classes compared with OCBs, which reflect intrathecal IgG synthesis only.^4,38,39 Especially in anti-LGI1 encephalitis, a distinct difference between a low prevalence of OCBs and high prevalence of intrathecal KFLC synthesis was observed, which is in line with the literature.⁴⁰ This high prevalence likely reflected the elevated anti-LGI1-specific IgG indices observed in 54% of patients, a frequency comparable to that of intrathecal KFLC synthesis.⁴⁰ It is also consistent with the presence of clonally expanded LGI1-specific plasmablasts in the CSF of patients.^41,42 Not surprisingly, the specificity of intrathecal KFLC synthesis was limited in our study (44%) because KFLC positivity was also frequent in MS (100%) and VZV encephalitis (63%) and still occurred in 10% of NINCs. This illustrates that specificity depends on the control cohort because KFLC reflects intrathecal Ig synthesis of all classes.^4,43 By contrast, the linear KFLC index demonstrated limited diagnostic utility in our cohort. Using the currently recommended cutoff of 6.1, it showed lower diagnostic sensitivity, poor discriminative performance in ROC analysis, and inferior sensitivity compared with intrathecal KFLC synthesis assessed using Reiber's diagram, emphasizing the use of the latter. Furthermore, NfL, GFAP, and cytokine profiles did not provide additional diagnostic value. Cytokine analysis revealed only minor differences, which did not remain statistically significant after correction for multiple testing and should, therefore, be interpreted as exploratory, consistent with previous reports.^23-27 Therefore, intrathecal KFLC synthesis as additional diagnostic criterion could substantially improve the sensitivity of AE diagnosis, particularly in anti-NMDAR AE, while overall specificity remains modest.

Second, longitudinal CSF analyses revealed a decline in the absolute level of intrathecal KFLC synthesis in anti-NMDAR and anti-LGI1 AE. These findings likely reflect decreased B-cell activity under second-line immunotherapies, mainly rituximab, consistent with data from MS.^4,44,45 This reduction was more pronounced compared with MS, possibly reflecting the monophasic nature of AE.^1,46 Because intrathecal KFLC synthesis was not an independent predictor of disease severity, it is suggestive that KFLC dynamics rather represent treatment response on a group level, similar to specific antibody titers or CXCL13, and might serve as a potential marker of therapeutic response, although validation is needed.^47,48 Given the short interval between CSF analyses (260 days) in our study, a treatment-related effect appears more likely than a spontaneous decline of specific antibody synthesis.^49,50 By contrast, NfL and GFAP did not decrease and CSF GFAP in anti-CASPR2 AE even increased, possibly indicating ongoing neuronal and astroglial injury, suggestive of a neurodegenerative process after inflammation, as observed before.^13,14,20,51 Cytokine levels remained largely stable, consistent with the limited and nonrobust differences observed compared with antibody-positive non-AE cases. Prior studies also reported more pronounced cytokine shifts in infectious than in AE.^23-27

Finally, in this study, an elevated NfL CSF z-score was associated with approximately 10-point higher CASE scores at baseline, highlighting the potential utility of NfL as a surrogate marker of biochemical disease severity and a later prognostic marker. In line with previous studies in autoimmune encephalitis, elevated NfL (and to a lesser extent GFAP) was associated with disease severity and disability across different AE subtypes, including anti-NMDAR and anti-LGI1 encephalitis, as well as other AEs such as anti-IgLON5 disease.^8-22

This correlation echoed findings of patients with MS, where NfL rises during inflammatory attacks.^8,52 Notably, this association was consistent across all AE subtypes and not primarily driven by anti-NMDAR AE, underscoring the robustness of NfL as a biomarker of disease severity. Furthermore, higher delta NfL values in CSF and serum were weakly associated with higher CASE and mRS scores during follow-up, providing additional predictive value beyond baseline clinical severity, thereby underlining the importance of repeated measures at clinically meaningful intervals in patients with AE.^11,13,14 It is important to note that variability in the interval between CSF analyses did not significantly influence longitudinal biomarker changes, supporting the robustness of these associations. GFAP revealed weaker associations, limited to elevated serum GFAP z-scores correlating with second-line immunotherapy use. The cytokines IL-31 and sCD40L showed only minor associations with disease severity (CASE scores) at baseline, which should be interpreted with caution, given the lack of robust group differences after correction for multiple testing. This aligns with prior publications highlighting IL-6, IL-8, IL-17A, or IL-17F as more consistent indicators of disease activity in AE.^23-27 Overall, NfL was the most reliable biomarker of disease activity in AE, while KFLC and GFAP added limited prognostic value.

Despite the multicenter design and the prospective analytic approach including diverse control groups, sample size remained restricted to the most common AE subtypes. Furthermore, inclusion of patients with AE who underwent repeated CSF analyses might bias toward more severe cases. Because the CASE score was originally developed for anti-NMDAR AE, evaluation in patients with anti-LGI1 AE and anti-CASPR2 AE might not fully reflect disease activity, although, in this study, correlations between CASE scores and biomarker levels were consistent across all AE subtypes and confirmed by additional outcome parameters.

In conclusion, this study identifies distinct biomarker patterns with diagnostic and prognostic implications across AE subtypes. KFLC constitute a sensitive yet moderately specific marker of intrathecal humoral immune activation, particularly in anti-NMDAR AE, supporting their inclusion in future diagnostic frameworks. When intrathecal synthesis of KFLC is assessed in patients with AE, Reiber's diagram should preferably be used because of its higher diagnostic sensitivity and specificity compared with the KFLC index. NfL represents a robust and reproducible biomarker of disease activity and seems to provide prognostic information across AE subtypes, warranting further validation in larger, antibody-defined cohorts.

Glossary

AE

autoimmune encephalitis

AUC

area under the curve

CASE

Clinical Assessment Scale in Autoimmune Encephalitis

FDR

false discovery rate

GFAP

glial fibrillary acidic protein

IFNγ

interferon gamma

IQR

interquartile range

KFLC

kappa free light chain

LGI1

leucine-rich glioma-inactivated 1

mRS

modified Rankin Scale

NfL

neurofilament light chain

NMDAR

N-methyl-d-aspartate receptor

ROC

receiver operating characteristic

TNFα

tumor necrosis factor alpha

Author Contributions

F.F. Konen: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data. E. Bucak: drafting/revision of the manuscript for content, including medical writing for content. F. Bachhuber: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. J. Dargvainiene: drafting/revision of the manuscript for content, including medical writing for content. M. Guasp: drafting/revision of the manuscript for content, including medical writing for content. J. Brenner: drafting/revision of the manuscript for content, including medical writing for content. A.L. Streichert: drafting/revision of the manuscript for content, including medical writing for content. D. Hudasch: drafting/revision of the manuscript for content, including medical writing for content. L.M. Albers: drafting/revision of the manuscript for content, including medical writing for content. S. Räuber: drafting/revision of the manuscript for content, including medical writing for content. J. Wickel: drafting/revision of the manuscript for content, including medical writing for content. K. Siebenbrodt: drafting/revision of the manuscript for content, including medical writing for content. M. Khalil: drafting/revision of the manuscript for content, including medical writing for content. K.-P. Wandinger: drafting/revision of the manuscript for content, including medical writing for content. M. Süße: drafting/revision of the manuscript for content, including medical writing for content. J. Lewerenz: drafting/revision of the manuscript for content, including medical writing for content. C. Geis: drafting/revision of the manuscript for content, including medical writing for content. C.S. Falk: drafting/revision of the manuscript for content, including medical writing for content. N. Melzer: drafting/revision of the manuscript for content, including medical writing for content. S.G. Meuth: drafting/revision of the manuscript for content, including medical writing for content. P. Schwenkenbecher: drafting/revision of the manuscript for content, including medical writing for content. T. Skripuletz: drafting/revision of the manuscript for content, including medical writing for content. H. Tumani: drafting/revision of the manuscript for content, including medical writing for content. F.S. Thaler: drafting/revision of the manuscript for content, including medical writing for content. T. Kümpfel: drafting/revision of the manuscript for content, including medical writing for content. M.J. Titulaer: drafting/revision of the manuscript for content, including medical writing for content. J.D. Lünemann: drafting/revision of the manuscript for content, including medical writing for content. F. Leypoldt: drafting/revision of the manuscript for content, including medical writing for content. K.-W. Sühs: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; study concept or design; analysis or interpretation of data.

Study Funding

The authors report no targeted funding.

Disclosure

F.F. Konen received honoraria for lectures and travel compensation from Alexion, Argenx, Merck, Novartis, and Takeda. He received research support as fellow of the German Research Foundation (DFG)–funded and Hannover Medical School (MHH)–funded Clinician Scientist Program (PRACTIS) at MHH and from Merck, Siemens, and Erwin-Röver-Foundation. E. Bucak, FB, J. Dargvainiene, and M. Guasp declare no conflict of interest considering the present work to report. J. Brenner on behalf of Erasmus MC has filed a copyright for the PROSE (Patient-Reported Outcome Scale in Encephalitis). A.L. Streichert, D. Hudasch, and L.M. Albers declare no conflict of interest to report. S. Räuber received travel grants from Merck Healthcare Germany GmbH, Alexion Pharmaceuticals, Jazz Pharmaceuticals, and Bristol Myers Squibb; she served on a scientific advisory board from Merck Healthcare Germany GmbH and received honoraria for lecturing from Roche and Merck Healthcare Germany GmbH; her research was supported by Novartis Pharma GmbH, Sanofi-Aventis Deutschland GmbH, “Stiftung zur Förderung junger Neurowissenschaftler,” and 'Else Kröner-Fresenius-Stiftung.' J. Wickel received honoraria for lectures and travel compensations from Alexion, Argenx, and Neuraxpharm. K. Siebenbrodt has no conflict of interest to report. M. Khalil has received travel funding and speaker honoraria from Bayer, Biogen, Novartis, Merck, Sanofi, Roche, and Teva; he serves on scientific advisory boards for Biogen, Bristol-Myers Squibb, Gilead, Merck, Neuraxpharm, Novartis, Alexion, Amgen, and Roche and as a consultant for Roche; he received research grants from the Austrian MS Society, Biogen, Novartis, and Roche. K.P. Wandinger has no conflict of interest to report. M. Süße reports personal fees and grants from Merck, Bayer Vital GmbH, Biogen, and Siemens. J. Lewerenz has no conflict of interest to report. C. Geis received speaker fees and compensation for advisory boards from Alexion, Roche, Sobi, Argenx, Kyowa, and Astra Zeneca. C.S. Falk has no conflict of interest to report. N. Melzer received funding from the Bundesministerium für Bildung und Forschung, CONNECT-GENERATE 2.0 - Forschungsverbund zur Aufklärung, and Kategorisierung und Behandlung autoimmuner Hirnentzündungen und verwandter Erkrankungen (01GM2208). S.G. Meuth received honoraria for lecturing and consulting, as well as travel expenses for attending meetings and research support, from Academy 2, Argenx, Alexion, Almirall, Amicus Therapeutics Germany, AstraZeneca, Bayer Health Care, Biogen, BioNTech, BMS, Celgene, Datamed, Demecan, Desitin, DiaMed, Diaplan, DIU Dresden, DPmed, Gen Medicine and Healthcare products, Genzyme, Hexal AG, IGES, Impulze GmbH, Janssen Cilag, KW Medipoint, MedDay Pharmaceuticals, Medmile, Merck Serono, MICE, Mylan, Neuraxpharm, Neuropoint, Novartis, Novo Nordisk, ONO Pharma, Oxford PharmaGenesis, QuintilesIMS, Roche, Sanofi, Springer Medizin Verlag, STADA, Chugai Pharma, Teva, UCB, Viatris, Wings for Life International, and Xcenda; his research is funded by the BMBF; German Federal Institute for Risk Assessment (BfR); DFG; Else Kröner Fresenius Foundation; Gemeinsamer Bundesausschuss (G-BA); German Academic Exchange Service; Hertie Foundation; IZKF Munster; German Foundation for Neurology; Ministry of Culture and Science of the State of North Rhine-Westphalia; the Daimler and Benz Foundation; DSMG; Peek & Cloppenburg Düsseldorf Foundation; Hempel Foundation for Science, Art and Welfare; German Alzheimer Society e.V. (Dementia self-help), and Alexion, Almirall, Amicus Therapeutics Germany, Argenx, Bayer Vital GmbH, BGP Products Operations (Viatris company), Biogen, BMS, Demecan, DiaMed, DGM e.V., Fresenius Medical Care, Genzyme, Gesellschaft von Freunden und Förderern der Heinrich-Heine-Universität Düsseldorf e.V., HERZ Burgdorf, Hexal, Janssen, Merck Serono, Novartis, Novo Nordisk Pharma, ONO Pharma, Roche, and Teva. P. Schwenkenbecher has no conflict of interest to report. T. Skripuletz reports research support from Alnylam, CSL Behring, Merck, Novartis, and Siemens and honoraria for lectures, travel support for meeting attendance, and/or consultancy fees from Alexion, Alnylam, Amgen, Argenx, Bayer, Biogen, Bristol Myers Squibb, Centogene, CSL Behring, Grifols, Hexal AG, Horizon, Janssen, Merck, Novartis, Pfizer, Roche, Sanofi, Siemens, SOBI, Teva, and Viatris; H. Tumani received honoraria as a consultant/speaker and/or for events sponsored by Alexion, Bayer, Biogen, Bristol-Myers Squibb, Celgene, DiaMed, Fresenius, Fujirebio, GlaxoSmithKline, Horizon, Janssen-Cilag, Merck, Novartis, Roche, Sanofi-Genzyme, Siemens, Teva, and Viatris. F.S. Thaler received speaker honoraria from Alexion. T. Kümpfel has received honoraria for lectures and/or advisory board participation from Novartis Pharma, Roche Pharma, Alexion/Astra Zeneca, Horizon Therapeutics/Amgen, Merck, Chugai Pharma, and Biogen; the entity for which she works received compensation for membership on the Roche Steering Committee. T. Kümpfel is the principal investigator on several randomized clinical trials (Novartis Pharma, Roche Pharma, BMS, and Sanofi Genzyme) and on a randomized clinical trial supported by the BMBf (funding code: 01 GM1908E); her institution has received compensation for clinical trials, all of which are outside the present work. M.J. Titulaer has no conflict of interest considering the present work to report. J.D. Lünemann received speaker fees, research support, travel support, and/or served on advisory boards by Abbvie, Alexion, Adivo, Amgen, Argenx, Biogen, CSL Behring, Janssen-Cilag, Merck, Moderna, Novartis, Octapharma, Roche, Sanofi, Takeda, and UCB Pharma, and he is a member of the medical advisory board of the German Myasthenia Gravis Society. F. Leypoldt is supported by E-Rare Joint Transnational Research Support (ERA-Net, LE3064/2-1), European Joint Program for Neurodegenerative Diseases (EJPRD) IGNITEMIND (01ED2506B), ERA-Net MICE-AE (01 EW2507B), Stiftung Pathobiochemie of the German Society for Laboratory Medicine, and HORIZON MSCA 2022 Doctoral Network (101119457, IgG4-TREAT) and discloses speaker honoraria from Grifols, Teva, Biogen, Bayer, Roche, Novartis, and Fresenius; travel funding from Merck, Grifols, and Bayer; and service on advisory boards for Roche, Biogen, and Alexion. K.W. Sühs reports honoraria for lectures or travel reimbursements for attending meetings from Bavarian Nordic, Biogen, Bristol-Myers, Merck, Mylan, Novartis, Roche, Viatris, and Bristol-Myers Squibb as well as research support from Bristol-Myers Squibb, all outside the submitted work. Go to Neurology.org/NN for full disclosures.