Copine‐4 Paraneoplastic Autoimmune Encephalitis

ABSTRACT

We characterize copine‐4 (CPNE4)‐IgG and report clinical associations. Among 844 samples with unclassified neural IgGs identified by mouse‐brain indirect immunofluorescence assay over 2 years, 5 samples (serum, 3; CSF, 2) from 3 patients produced a distinctive cerebrum‐restricted synaptic pattern. Protein microarray disclosed CPNE4 “top hit” in those 3 patients; specificity was confirmed by 3 CPNE4‐specific assays (confocal IFA, cell‐based, and western blot). Median onset age was 67 years (range 60–68); all had encephalitis: a brainstem syndrome during immune checkpoint inhibitor therapy for lung cancer (2) and anti‐NMDA‐receptor encephalitis accompanied by neuroectoderm‐lineage cancer (1). To date, CPNE4‐IgG has paraneoplastic encephalitis significance.

1. Introduction

Timely diagnosis of autoimmune encephalitis (AE) is aided by diagnostic criteria that utilize subacute symptom onset, altered mental status, CSF pleocytosis, and encephalitic MRI findings [1]. Detection of neural IgG autoantibodies (e.g., NMDA‐receptor, LGI1‐IgGs) secures a definite AE diagnosis [1]. Diagnosis may be more challenging in seronegative AE cases and in patients with brainstem encephalitis (BE) who rarely satisfy AE criteria owing to preserved mental status [2]. We recently proposed BE‐specific diagnostic criteria [3]. While definite (antibody positive) BE criteria have high sensitivity and specificity as compared to gold‐standard expert clinician diagnosis, probable (antibody negative) BE criteria have low sensitivity (60%) [3]. Thus, novel disease‐specific neural antibody biomarkers may improve diagnostic certainty. Neural antibodies may also inform cancer risk, histologic subtype, prognosis, and natural history [4]. For example, IgLON5‐IgG is very rarely cancer‐associated, whereas KLHL11‐IgG strongly associates with seminoma [5, 6]. Some novel neural IgGs appear to arise predominantly in the setting of immune checkpoint inhibitor (ICI) therapy [7]. Here, we describe a central nervous system autoantibody targeting copine‐4 (CPNE4) identified in three patients with paraneoplastic AE, two of whom developed BE during ICI therapy for Stage IV non‐small cell lung cancer (NSCLC).

2. Methods

2.1. Standard Protocol Approvals and Registrations

The Mayo Clinic Institutional Review Board (IRB), as a minimum risk research protocol (IRB#, 21‐001297), approved: (1) human waste clinical specimen (serum/CSF) acquisition and testing for antibodies, and (2) retrospective medical record review.

2.2. Screening

Murine tissue (whole brain, kidney and stomach) indirect immunofluorescence assay (IFA) was performed as part of clinical laboratory testing for well characterized and unclassified antibodies [8]. Unclassified antibodies were defined by neural‐restricted murine tissue staining with unknown antigen specificity. Residual serum and CSF samples were screened by human protein microarray (HuProt v4.0, > 21,000 proteins) [9].

2.3. Further IFA and Antigen Confirmation

Samples were retested by tissue IFA to confirm identical staining patterns in the 5 samples. Antigen specificity was confirmed by 3 assays: Supplemental methods [8], Data S1. Confocal IFA used rabbit polyclonal CPNE4 antibody (1:25, Proteintech) alongside patient and control sera. Western blot used recombinant full‐length CPNE4 protein (OriGene). Fixed cell‐based assay (CBA) used GFP‐tagged CPNE4 expressed in HEK293 cells, tested with patient serum/CSF and controls.

2.4. Patients and Controls

Clinical, CSF, imaging, treatment, and outcome data were obtained. Western blot controls included five healthy sera. CBA controls (110 total) included 40 healthy sera, 30 CSF samples from patients with normal pressure hydrocephalus (IFA negative), and 40 (24 sera, 16 CSF specimens) from patients with diverse seropositive autoimmune encephalitides (Data S2: Table).

2.5. Data Access and Availability

Data are available from the corresponding author upon reasonable request.

3. Results

3.1. Case Identification

Between October 1st 2022 and September 30th 2024, 422,000 patient samples were tested on a clinical service basis in the Mayo Clinic Neuroimmunology Laboratory by tissue IFA, and 844 were determined to harbor unclassified neural antibodies (0.20%). Five samples (sera, 3; CSF, 2) from 3 patients (0.0012% of all samples tested) showed identical cerebrum‐predominant synaptic (neuropil) staining (Figure S1). Staining was strongest in hippocampus (Figure S1A) as compared with thalamus, basal ganglia, and cortex (Figure S1B,C), with predilection for CA2/CA3 regions (Figure 1A). Cerebellum, kidney, and gut were negative (Figure S1D). Faint midbrain staining exceeded adjacent negative cerebellum (Figure S1E).

FIGURE 1.

Validation of CPNE4 as antigen. (A) Confocal IFA demonstrated patient IgG (green) and CPNE4 rabbit polyclonal antibody staining (red) and merged (yellow) in mouse hippocampus (CA2/CA3 brighter than other regions), and cerebral cortex. Scale bar = 100 μm. (B) Western blot using recombinant CPNE4 protein produced a ~60‐kDa band when probed with commercial CPNE4 antibody or with patient serums. Healthy control sera were negative. (C) Primary antibody (patient IgG or CPNE4‐specific antibody) bound to HEK‐293 cells transfected with cDNAs encoding green fluorescent protein (GFP)‐tagged CPNE4 (CPNE4‐GFP). Primary antibodies were applied to fixed cells. Top four rows show positive primary antibody (red), CPNE4‐GFP (green) and merged images (yellow). Bottom row shows healthy control serum, negative. Scale bar = 20 μm.

All tested samples produced high‐ranking hits for CPNE4 (62 kDa molecular weight) on protein array (top 5 by AF647 intensity; Figure S1F). None of the remaining 841 unclassified samples ranked CPNE4 in the top 50 (considered negative) [8].

3.2. Confirmation of CPNE4 as Target Antigen

Confocal IFA demonstrated colocalization of patient IgG with the CPNE4 polyclonal antibody signal, including the characteristic hippocampal CA2/CA3 predominance (Figure 1A). Western blot showed an approximately 60 kDa band for all patient sera and CPNE4 polyclonal antibody, but not controls (Figure 1B). Fixed CBA demonstrated reactivity in all five available patient samples (3 sera, 2 CSF, Figure 1C) and none of 110 healthy, autoimmune disease, and nonautoimmune controls. Testing of mock transfected cells and secondary antibody alone on transfected cells was negative.

3.3. Clinical Features

3.3.1. Neurological Phenotype

Two of 3 patients were male and one female; median age at onset was 67 years (range, 60–68). All had subacute symptom onset (Table 1). All presented with AE symptoms, 2 BE predominant, and 1 had anti‐NMDA‐receptor encephalitis phenotype predominant though with atypical MRI findings. Patient 1 developed diplopia, ptosis, ophthalmoplegia, dysarthria, and ataxia over 3 months, concurrent with pembrolizumab treatment every 3 weeks. Patient 2 with anti‐NMDA‐receptor encephalitis presented with psychosis, insomnia, hallucinations, polydipsia, and dysarthria over 4 months, followed by diplopia and proptosis (attributable to left intraorbital mass, Table 1). Patient 3 had diplopia, ptosis, dysarthria, hoarseness, and word‐finding difficulty over 1 month, with concurrent ICI‐related myopericarditis and hepatitis. Neurological symptoms began within 1 week of the last pembrolizumab dose.

TABLE 1.

Clinical characteristics of CPNE4‐IgG positive patients.

N	Neurological	Cancer	MRI	EEG	CSF	Immune therapy	Cancer Tx	Outcomes (duration of follow‐up, months)
1	Diplopia, ptosis, dysarthria, ataxia, ophthalmoplegia	NSCLC, Stage 4	Increase T2/FLAIR signal in pons, thalami, & corona radiata	N/A	WBC, 73 (97% lymphocytes, protein 74)	Steroids, IVIg	Carboplatin/Pemetrexed/Pembrolizumab, TS‐ONE (Tegafur, Giemeracil, Oteracil)	Partial neurological improvement after immuno‐Tx; died from cancer (13)
2	Psychosis a , insomnia, hallucinations, polydipsia, hyponatremia followed by proptosis & diplopia (attributable to ALES)	ALES (left orbit, ethmoid sinuses, left anterior cranial fossa)	Symmetric increase T2/FLAIR signal in bilateral thalami without significant restricted diffusion	Multifocal IEDs (bilateral frontal & right temporal)	WBC, 0; protein normal.	Steroids, PLEX, IVIg, rituximab	N/A	No improvement after immuno‐Tx; wheelchair dependent, chronic encephalopathy (14)
3	Diplopia, ptosis, dysarthria, hoarseness, word‐finding difficulty	NSCLC, Stage 4	Chronic small vessel ischemic disease	N/A	N/A	Steroids, IVIg, azathioprine, tocilizumab	Carboplatin/Pemetrexed/Pembrolizumab	Partial neurological improvement after immuno‐Tx; died from cancer 1 month after neurological presentation (1)

Note: CSF normal values: protein, ≤ 35 mg/dL; white cell count, ≤ 5/μL. CSF‐exclusive oligoclonal bands, IgG index not tested.

Abbreviations: ALES = adamantinoma‐like Ewing sarcoma, IEDs = interictal epileptiform discharges, IVIg = intravenous immunoglobulin, NSCLC = non‐small cell lung cancer, PLEX = plasma exchange, S/Sx = signs/symptoms, Tx = treatment, WBC = white blood cells.

^a Coexisting NMDA‐R Ab in the CSF (by CBA only; tissue IFA negative).

Patients 1 and 3 developed neurological symptoms after carboplatin, pemetrexed, and pembrolizumab for metastatic NSCLC, with onset 9 and 1 month after ICI initiation, respectively. Patient 2 was diagnosed with adamantinoma‐like Ewing sarcoma (ALES) 15 months after neurologic onset.

3.3.2. Imaging and CSF

Patient 1 had T2/FLAIR signal changes in the pons, thalami, and corona radiata without enhancement or diffusion restriction, and also had generalized cerebral atrophy (Figure 2A–C). Patient 2 had symmetric thalamic T2/FLAIR hyperintensities (Figure 2D) atypical for anti‐NMDA‐receptor encephalitis. Patient 3 had nonspecific leukoaraiosis only (not shown). EEG in Patient 2 showed bifrontal and right temporal sharp waves and slowing.

FIGURE 2.

Magnetic resonance imaging (MRI) signal change in CPNE4‐IgG positive patients. T2 fluid‐attenuated inversion recovery (FLAIR) signal changes were observed in Patient 1 in the pons (A, dash arrows), bilateral thalami (B, thin arrows), and bilateral corona radiata (C, empty arrows). T2 FLAIR signal changes were also observed in bilateral thalami in Patient 2 (D, thick arrows).

CSF analysis was available in two patients. Patient 1 had lymphocytic pleocytosis (73 cells/μL) and elevated protein (74 mg/dL). Patient 2 had normal CSF cell counts and protein, and coexisting NMDA‐R IgG.

3.3.3. Treatment and Outcomes

All patients received immunotherapy (Table 1), including corticosteroids (3), IVIg (3), plasma exchange (1), rituximab (1), azathioprine (1), and tocilizumab (1). Patients 1 and 3 had partial neurological improvements but died from cancer 13 and 1 months after BE onset, respectively. Patient 2 did not improve with immunotherapy and was alive at last follow‐up, though was wheelchair bound and demented.

4. Discussion

CPNE4‐IgG was identified in three patients with AE: two with BE during ICI therapy for NSCLC and one with paraneoplastic ALES and coexisting NMDA‐R encephalitis. MRI showed thalamic T2 abnormalities in 2 of 3 patients, including the anti‐NMDA‐receptor encephalitis case. Although rare, this biomarker may improve diagnostic sensitivity in paraneoplastic AE and BE. Only one patient met proposed probable antibody‐negative BE criteria [3]; however, CPNE4‐IgG allowed both BE cases to meet definite BE criteria, while the third met definite AE criteria for anti‐NMDA‐receptor encephalitis [1]. Further validation may support CPNE4‐IgG as a high‐risk paraneoplastic antibody [4].

Classical paraneoplastic antibodies (e.g., anti‐Hu) occur with ICI therapy, while newer antibodies—including PDE10A‐IgG, CAMKV‐IgG, and now CPNE4‐IgG—may be enriched in ICI‐related disease [10, 11]. Although our cohort is small, ICI therapy may broaden the neural antigen repertoire vulnerable to autoimmunity. Patients 1 and 3 partially improved after corticosteroids and IVIg but died from cancer. Patient 2 had paraneoplastic AE with ALES and anti‐NMDA‐receptor encephalitis. ALES is a rare neuroectoderm‐derived tumor with epithelial differentiation that can mimic other head and neck malignancies [12]. In older patients with anti‐NMDA‐receptor encephalitis, coexisting IgGs may guide cancer screening beyond teratoma; for example, CSF neurofilament light chain IgG has coexisted with small cell carcinoma [13].

“Synaptic” tissue IFA staining does not imply pathogenicity, as it may reflect intracellular antigens enriched in nerve terminals, including amphiphysin, GAD65, or CPNE4, as well as classic cell‐surface antigens such as NMDA‐R. CPNE4 is a cytoplasmic neural protein enriched in neurons; thus CPNE4‐IgG is likely a biomarker of T cell‐mediated paraneoplastic AE rather than directly pathogenic. T cell‐directed immunosuppression, such as cyclophosphamide, could be considered in future patients. CPNE4 belongs to the copine family of calcium‐dependent phospholipid‐binding proteins with restricted tissue distribution [14]. Hippocampal CPNE4 expression is upregulated by NMDA receptor activation [15]. Polydipsia may localize to CPNE4‐enriched hypothalamic or mesolimbic regions [16]. CPNE4‐IgG staining paralleled MRI abnormalities and known CPNE4 expression in forebrain and brainstem, but not cerebellum; cerebellar symptoms likely reflect brainstem–cerebellar connectivity [16].

Fixed CPNE4‐specific CBA was positive in all patient samples and negative in controls, supporting confirmatory testing after tissue immunohistochemistry. Given limited cases, further clinical validation is required. Differential diagnoses for cerebrum‐predominant IFA staining include mGluR5‐, AK5‐, and CAMKV‐IgGs [11]. Although CPNE1 and CPNE3 expression has been linked to NSCLC progression, CPNE4 has not previously been associated with cancer [17]. CPNE5‐IgG antibody was recently reported in paraneoplastic encephalitis in 4 patients, all with melanoma, including one post‐ICI case [18].

Despite advances in AE diagnostics, many patients remain seronegative [2]. Sixteen percent of ICI‐associated AE cases do not meet AE criteria [7]. In our laboratory, unclassified antibodies collectively represent the sixth most common neural IgG finding [8]. Characterizing individual rare antibodies may provide diagnostic insight and support mechanistic studies. Limitations include very small case numbers, incomplete CSF data, and lack of tumor immunopathology linking CPNE4 to patient cancers. At this early stage, CPNE4‐IgG appears relevant to paraneoplastic AE, possibly ICI‐related. Testing should be considered in suspected AE, including BE in paraneoplastic or post‐ICI contexts, particularly when the described tissue IgG staining pattern is suspected.

Author Contributions

Conception and design of the study: Andrew McKeon. Acquisition and analysis of data: all authors. Drafting a significant portion of the manuscript or figures: Mimi S. Fang, Andrew McKeon.

Funding

This work was supported by National Institute of Neurological Disorders and Stroke, RO1NS126227.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: IgG staining of cryosectioned mouse tissue by indirect immunofluorescence assay (IFA). Patient serum produced robust synaptic staining of: (A and C) hippocampus (Hi, including stratum lucidum [SL] and dentate gyrus [DG]); (B) thalamus (Thal) and basal ganglia (BG); (A and C) cerebral cortex (Cx), but (C) negative kidney (Kid) and (D) negative cerebellar molecular layer (ML) and granular layer (GL). (E) faint midbrain staining (MB) was also observed. Scale bars = 50 μm. (F) CPNE4 ranked as top neural protein ‘hits’ in each of 3 patient samples by protein microarray as measured by AF647 secondary antibody signal.

Data S1: Detailed methods for protein microarray, indirect immunofluorescence assay (IFA), dual staining by confocal IFA, western blotting and fixed cell‐based immunofluorescence assay (CBA).

Data S2: 40 samples from patients with other neural IgG biomarker defined autoimmune neurological disorders used as controls in CPNE4‐specific cell based assay.

Funding Statement

This work was funded by National Institute of Neurological Disorders and Stroke grant RO1NS126227.

Data Availability Statement

All data pertaining to this article are available upon request.