Evaluation of neurofilament light chain as a biomarker in dogs with structural and idiopathic epilepsy

Abstract

Background

Neurofilament light chain (NfL) is a frequently used biomarker in humans for both diagnostic and therapeutic monitoring purposes in various neurologic diseases.

Hypothesis/Objectives

It was hypothesized that dogs with diagnosed structural epilepsy (SE) would have a significantly higher serum NfL concentrations compared to dogs with idiopathic epilepsy (IE). The secondary hypothesis was that dogs would have a significantly higher serum NfL concentrations when measured within 7 days after a seizure compared to being seizure‐free for at least 30 days.

Animals

Fifty client‐owned dogs presented to the neurology service for evaluation of seizures were enrolled. Fourteen dogs had SE and 36 dogs had IE.

Methods

Prospective cohort study performed on 52 serum samples obtained for NfL concentration measurement using single molecule array technology.

Results

The median serum concentration of NfL in dogs with SE was significantly higher (109 pg/mL; range, 11.4‐741.3 pg/mL) than in dogs with IE (17.7 pg/mL; range, 5.8‐188 pg/mL; Wilcoxon rank sum test, P = .001). No significant relationship was found between serum NfL concentration and time of sampling in relation to the most recent seizure in dogs with IE.

Conclusions and Clinical Importance

Serum NfL may serve as an adjunctive biomarker for the differentiation of SE and IE.

Abbreviations

AO

atlanto‐occipital

CSF

cerebrospinal fluid

EEG

electroencephalogram

GRE

gradient echo

IE

idiopathic epilepsy

MRI

magnetic resonance imaging

MUE

meningoencephalitis of unknown etiology

NfL

neurofilament light chain

SE

structural epilepsy

Simoa

single molecule array

TNCC

total nucleated cell count

1. INTRODUCTION

Neurofilaments are crucial proteins of the neuronal cytoskeleton that increase in serum and cerebrospinal fluid (CSF) when central nervous system (CNS) disease is present. Neurofilaments can be measured and used as a biomarker for the diagnosis and therapeutic monitoring of neurologic disease in people and animals. ¹ , ² , ³ , ⁴ Neurofilaments consist of 4 subunits which are heavy, medium, and light chains as well as either alpha‐internexin in the CNS or peripherin in the peripheral nervous system. ⁵

Neurofilament light chain (NfL) has been evaluated as a biomarker in aging dogs with and without cognitive dysfunction as well as in dogs diagnosed with meningoencephalitis of unknown etiology (MUE). ³ , ⁴ In dogs with MUE, NfL concentrations were not different between those MUE dogs with and without seizures. ⁴ This finding was presumed to be a consequence of the NfL concentration already being so markedly increased because of structural brain disease, which masked the ability to detect any increase in serum or CSF concentration of NfL solely attributable to seizure activity. Neurofilament light chain also was shown to be significantly increased in dogs with neurodegenerative disease such as cognitive dysfunction. ³

Currently, idiopathic epilepsy (IE) in dogs is a diagnosis of exclusion based on a 3‐tier confidence level system requiring a normal interictal neurological examination, CBC, and serum biochemistry profile at the lowest level, in addition to normal bile acid concentrations, brain magnetic resonance imaging (MRI), CSF analysis, and electroencephalography (EEG). ⁶ These additional diagnostic tests have some limitations such as prohibitive cost, availability, and risk of general anesthesia. Furthermore, accurately determining seizure frequency in dogs can be difficult, because most pet owners cannot constantly and directly supervise their dogs, and the actual frequency of seizures often is underestimated when based on subjective, owner‐witnessed events compared with EEG recordings. ⁶ , ⁷ , ⁸ It is also not always possible to discriminate dogs with IE from those with structural epilepsy (SE) based solely on historical and clinical examination findings, because dogs with SE caused by numerous diseases may have a normal interictal neurological examination and age of first seizure onset that falls within expected range for IE. ⁶ Therefore, a clinical need exists for readily available and non‐invasive serum biomarkers that could assist with discrimination of genetic vs structural causes of epilepsy or identification of epileptic patients that recently have experienced seizures. Serum NfL concentrations are increased in humans experiencing status epilepticus compared to both healthy people and those with epilepsy who experience occasional isolated seizures. ⁹ , ¹⁰ Furthermore, serum NfL concentrations in humans correlate well with CSF concentrations in numerous neurological diseases, and thus serum NfL concentrations generally are considered reliable biomarkers for CNS disease. ¹⁰ , ¹¹

Our objective was to evaluate NfL as a serum biomarker for epilepsy in dogs. Our primary hypothesis was that serum NfL concentrations would be significantly higher in dogs with SE compared to those with IE. Our secondary hypothesis was that NfL concentrations would be significantly higher if measured within 7 days after a seizure compared to being seizure‐free for at least 30 days before blood sampling.

2. MATERIALS AND METHODS

2.1. Animals

Our study was a prospective cohort study. Client‐owned dogs with epilepsy that presented to the Virginia Tech Veterinary Teaching Hospital were eligible for enrollment. Owners provided written, informed consent for their dogs to participate in the study. The study was approved by the Institutional Animal Care and Use Committee (Protocol Number #23‐071). Epilepsy was defined as ≥2 unprovoked seizures, with no known cause for reactive seizures. Dogs were diagnosed with IE and included in the study if they met criteria for at least a tier I level of evidence as defined by the International Veterinary Epilepsy Task Force. ⁶ A tier I diagnosis includes a history of ≥2 unprovoked epileptic seizures occurring at least 24 hours apart, an age of seizure onset between 6 months and 6 years of age, no clinically relevant abnormalities on minimum database blood and urine tests, and normal interictal physical and neurologic examinations. ⁶ Dogs with SE were included when they were diagnosed with a structural lesion on brain MRI; CSF was not required for inclusion into this group. Dogs were excluded from the SE if they had concurrent signs localized to the spinal cord. Dogs with SE and IE were further divided into acute and chronic seizure groups. The acute group included dogs that experienced a seizure within 7 days of serum sampling for NfL analysis. The chronic group included dogs with no documented seizure activity within at least 30 days of serum sampling for NfL analysis. Dogs were not sampled for NfL analysis if a seizure occurred between 7 and 30 days, and owners were asked to return their dogs for serum sampling on or after Day 30 or after the last known seizure. Timing of sampling in acute and chronic groups was based on NfL having a previously reported half‐life in vivo of 3 weeks in the brain. ¹²

2.2. Diagnostic evaluation

Dogs included in the study received physical and neurologic examinations performed by a neurology resident and board‐certified neurologist. Neuroanatomic localization and individual neurologic deficits were recorded for each dog. Dogs were included if they had minor clinicopathologic changes on their serum biochemical profiles that could be explained by either recent seizure activity or administration of anti‐seizure drugs. Magnetic resonance imaging was performed in some dogs with IE based on the owner's preference. Interpretation of MRI was performed by a board‐certified radiologist, board‐certified neurologist, and a neurology resident. In IE dogs in which brain MRI was performed, results must have been interpreted as normal or shown evidence of bilaterally symmetrical T2/Fluid attenuated inversion recovery (FLAIR) hyperintense, T1 isointense to hypointense, minimally enhancing or non‐enhancing intraparenchymal lesions consistent with post‐ictal change. ¹³ Dogs were included in the SE group if they had brain MRI evidence of an intra‐axial or extra‐axial brain mass with imaging characteristics compatible with a primary intracranial tumor or multifocal T2/FLAIR hyperintense intraparenchymal brain lesions and CSF pleocytosis (presumptive MUE). ⁴ , ¹⁴ Other MRI characteristics including mass effect, peritumoral edema, degree of contrast enhancement, and presence of signal voids on a gradient echo sequence (GRE) were recorded when present. Electroencephalography was not required, but was performed in 1 dog with IE when nonconvulsive status epilepticus was suspected. Histopathologic confirmation of structural brain diseases was not required for inclusion.

2.3. Serum NfL assay

Blood samples from the jugular vein, peripheral vein, or arterial catheter were collected into serum (plain) tubes within 7 days of the most recent seizure in the acute group and after at least 30 days from the most recent seizure in the chronic group. Serum was separated by centrifugation (2000g, 10 minutes) at room temperature and stored at −80°C within 1 hour of collection until assayed for NfL. Serum NfL concentrations were analyzed in duplicate using single molecule array (Simoa) technology. The assay was performed using a Simoa HD‐1 Analzyer (Quanterix, Billerica, Massachusetts) with the NfL kit designed for humans using an anti‐NfL monoclonal antibody (UmanDiagnostics, Umea, Sweden). Two control samples of human recombinant antigen were provided with the kit and analyzed in each run for quality control purposes. The Canis lupus familiaris (GeneID: 100856220) and Homo sapiens (GeneID: 4747) neurofilament light chain (NEFL) genes are orthologs as determined using The National Center for Biotechnology Information (NCBI) HomoloGene database. ¹⁵ Using NCBI Basic Local Alignment Search Tool (BLAST) protein analytical suite (Query_47245), the canine NfL polypeptide has 98.35% sequence homology with the human NfL (100% sequence coverage). Thus, this gene and protein are conserved across species and this assay has been used previously to measure serum NfL concentrations in dogs. ³ , ⁴ , ¹⁶

2.4. Statistical methods

Statistical analyses were performed using JMP statistical software (JMP, Cary, North Carolina). The results are expressed as median and range. The data set was not normally distributed, and thus a nonparametric Wilcoxon rank sum test was used to compare serum NfL concentrations between dogs with IE and SE. A 2‐sided P value <.05 was considered significant. To determine the optimal NfL concentration for the differentiation of SE and IE, the area under the receiver operating characteristic curve (AUC) was analyzed. A multivariate analysis with a Wilcoxon rank sum test for each pair was used to assess comparisons between dogs with acute or chronic seizures. For correlation between continuous variables, non‐parametric Spearman's correlation coefficient was used. An r value <0.3 was considered a very weak correlation, 0.3 < r < 0.5 was considered weak, 0.5 < r < 0.7 was considered moderate, and an r > 0.7 was considered strong. The statistics performed were verified by a statistician (SW).

3. RESULTS

3.1. Study population

Fifty dogs were included in the study (36 IE, 14 SE), from which a total of 52 serum samples were collected. One dog in each of the IE and SE groups had a second serum sample obtained when they were seizure‐free for at least 30 days after initially being enrolled in the acute group. Breeds included in the IE group were mixed breed dog (7), Australian Shepherd (5), Staffordshire Terrier (3), Border Collie (2), Golden Retriever (2), Siberian Husky (2), Labrador Retriever (2), Dogue De Bordeaux (2), Cane Corso (1), Chesapeake Bay Retriever (1), Dachshund (1), English Bulldog (1), Great Dane (1), Italian Greyhound (1), Bloodhound (1), Miniature Poodle (1), Schnauzer (1), and terrier (1). Breeds included in the SE group were French Bulldog (2), Boxer (2), Boston Terrier (2), Beagle (1), American Bulldog (1), Jack Russell Terrier (1), Miniature Schnauzer (1), Staffordshire Terrier (1), Golden Retriever (1), Weimaraner (1), and mixed breed (1). The median age of dogs included was 4 years (range, 1‐8 years) and 8 years (range, 3‐14 years) for IE and SE, respectively. Dogs in the IE group were significantly younger than dogs in the SE group (P = .01). No significant correlation was found between age and serum NfL concentration (P = .11). In the IE group, there were 12 spayed females, 2 intact males, and 22 neutered males. In the SE group, there were 6 spayed females, 1 intact male, and 7 neutered males.

3.2. Diagnostic evaluation

In the SE group, 2/14 dogs had an abnormality on physical examination including generalized muscle atrophy in 1 dog and subcutaneous masses previously diagnosed as lipomas by fine needle aspirates in the other dog. In the SE group, all dogs had a forebrain neuroanatomic localization, which was lateralized in 7/14 dogs. The most common neurologic examination abnormalities noted were proprioceptive deficits in ≥1 limb (11/14), menace deficits (4/14), mentation changes (3/14), and cervical pain (3/14). Two dogs were suspected to have concurrent brainstem lesions because of vertical nystagmus (1/14) and severe mentation change (1/14).

In the IE group, 13/36 dogs had increased ALP activity and 2/36 had increased ALT activity. The median increased ALP activity for dogs with IE was 171 IU/L (range, 73‐1400 IU/L; reference range, 8‐70). In the SE group, 4/14 dogs had increased ALP activity and 2/14 dogs had a mild inflammatory leukogram characterized by a mature neutrophilia. The median increased ALP activity for dogs with SE was 197 IU/L (range, 150‐351 IU/L; reference range, 8‐70 IU/L).

Brain MRI was performed in 12/36 (33%) dogs in the IE group. In 10 dogs, MRI was considered normal. The other 2 dogs had bilaterally symmetrical T2/FLAIR hyperintensities located in the cingulate gyrus and hippocampus that were minimally contrast‐enhancing and were attributed to post‐ictal changes secondary to recent seizure activity. An EEG was obtained in 1 dog in the IE group. This EEG disclosed nonconvulsive epileptiform activity characterized by intermittent spike and wave activity and intermittent phase reversal in the right frontal lobe, which resolved with phenobarbital treatment.

In the SE group, 12/14 dogs were diagnosed with intracranial neoplasia and 2/14 were diagnosed with MUE based on MRI findings. Of the dogs diagnosed with intracranial neoplasia, 5/12 had a histopathologic diagnosis of glioma (2 astrocytomas, 1 oligodendroglioma, 2 undefined glial tumors). These 5 dogs were histopathologically diagnosed by stereotactic brain biopsy. Of the 2 dogs diagnosed with MUE, 1 dog had histopathologically confirmed necrotizing meningoencephalitis. This dog had a normal brain MRI except for symmetrical T2/FLAIR hyperintense, minimally contrast‐enhancing lesions throughout the cerebrum that were suspected to represent post‐ictal change.

3.3. Serum NfL concentration in dogs with IE

Neurofilament light chain concentration was measured in 37 serum samples for dogs with IE, which included 28 acute samples and 9 chronic samples. The median serum concentration of NfL in all dogs with IE was 17.7 pg/mL (range, 5.8‐188 pg/mL). The median serum concentration of NfL in dogs with IE in the acute group was 19.0 pg/mL (range, 6.30‐188 pg/mL). The median serum concentration of NfL in dogs with IE in the chronic group was 15.7 pg/mL (range, 5.8‐154 pg/mL). The median time from seizure to sampling of serum in IE dogs in the acute group was 1 day (range, 0‐4 days). The median time from seizure to sampling of serum in dogs with IE in the chronic group was 55 days (range, 30‐730 days).

3.4. Serum NfL concentration in dogs with SE

Fifteen serum samples were available from dogs with SE, including 13 acute samples and 2 chronic samples. The median serum concentration of NfL in all dogs with SE was 109 pg/mL (range, 11.4‐741.3 pg/mL). For dogs with brain tumors, the median serum concentration of NfL was 109 pg/mL (range, 12.9‐277.2 pg/mL). The median serum concentration of NfL in dogs with SE in the acute group was 129 pg/mL (range, 11.4‐741.3 pg/mL). The median serum concentration of NfL in dogs with SE in the chronic group was 33.6 pg/mL (range, 29.2‐38.0 pg/mL). The median time from seizure to sampling of serum in dogs with SE in the acute group was 1 day (range, 0‐5 days). The median time from seizure to sampling of serum in dogs with SE in the chronic group was 41.5 days (range, 30‐64 days).

3.5. Relationship of serum NfL concentration to seizure etiology and sampling time

Serum NfL concentration (Figure 1) was significantly higher in dogs with SE (n = 15; median, 109 pg/mL; range, 11.4‐741.3 pg/mL) compared to dogs with IE (n = 37; median, 17.7 pg/mL; range, 5.8‐188 pg/m; P = .001). When excluding dogs with MUE (n = 2), a significant difference in serum NfL concentration still existed between dogs with brain tumors and those with IE (P = .01). No significant difference was found between serum NfL concentrations when evaluating sampling time in relation to the most recent seizure in dogs with IE or SE (<7 days vs >30 days; Figure 2). Using a cut‐off of 71.17 pg/mL for serum NfL concentration, the specificity was 89% and sensitivity was 67% for differentiating SE and IE (AUC, 0.8; Figure 3).

FIGURE 1.

Serum concentrations of neurofilament light chain (NfL) in dogs with structural epilepsy (n = 15 samples; 13 acute, 2 chronic) were significantly higher than in dogs with idiopathic epilepsy (n = 37 samples; 28 acute, 9 chronic). Box represent the interquartile range with the center line representing the median. The error bar represents the minimum and maximum quartile ranges, and the dots represent the outliers.

FIGURE 2.

Serum neurofilament light chain (NfL) concentrations in dogs with idiopathic (n = 37 samples; 28 acute, 9 chronic) and structural epilepsy (n = 15 samples; 13 acute, 2 chronic) with seizures occurring within the 7 days before sample collection (acute) and dogs with no known seizures within the last 30 days before sample collection (chronic). Significantly higher NfL concentrations were observed when comparing acute structural epilepsy (SE) to acute idiopathic epilepsy (IE) (P = .002) and acute SE to chronic IE (P = .03). When comparing acute SE to chronic SE (P = .3), acute IE to chronic IE (P = .99), chronic IE to chronic SE (P = .72), and acute IE to chronic SE (P = .29) no significant difference was identified. Boxes represent the interquartile range with the center line representing the median. The error bar represents the minimum and maximum quartile ranges, and the dots represent the outliers.

FIGURE 3.

Receiver operating characteristic curve graph showing the area under the curve (AUC) when determining the optimal serum neurofilament light chain (NfL) concentration for the differentiation of structural epilepsy (SE) and idiopathic epilepsy (IE). Using a cut‐off of 71.17 pg/mL for serum NfL concentration, the specificity was 89% and the sensitivity 67% for differentiating SE and IE (AUC, 0.8).

4. DISCUSSION

We identified a significant increase in serum NfL concentration in dogs with SE caused by brain tumors and MUE compared with dogs with IE. This information may be useful in differentiating SE and IE in a clinical setting when MRI is unavailable. Obtaining an MRI diagnosis is not always feasible for a variety of reasons such as accessibility, cost, and anesthetic safety, and brain MRI also does not always provide a definitive diagnosis. ¹⁷ Because we did not evaluate NfL concentrations in dogs with SE caused by structural anomalies, vascular, or neurodegenerative conditions, our results may not be generalizable to all dogs with SE. However, because brain tumors and meningoencephalitides are the 2 most common causes of SE in dogs, our results are clinically applicable. ¹⁸

No significant relationships were identified between serum NfL concentration and timing of sample collection relative to the last seizure. Different time frames for defining and measuring NfL concentrations in acute and chronic seizures may be required, because the time‐time intervals selected for our study were based on the half‐life of NfL in mice. ¹² Our study is also limited by small sample sizes, particularly in the chronic seizure groups, which may have impacted our ability to detect changes in NfL concentration in relation to seizure timing. Another possibility is that owners failed to accurately observe or log recent seizure activity resulting in a misrepresentation of the true interictal interval. ⁸

It remains unknown in dogs if isolated focal or generalized epileptic seizures, in the absence of status epilepticus or temporally dense cluster seizures, will cause substantial increases in serum NfL concentrations. In humans, duration of status epilepticus is positively correlated with the magnitude of serum increase of NfL, and serum NfL concentrations are significantly higher in people who have recently experienced status epilepticus compared to isolated seizures. ⁹ , ¹⁰ Thus, serum NfL increases may result from the presence of seizures only in those animals that have experienced seizures of sufficient severity and duration to result in neuronal death, and experimental studies of status epilepticus suggest that this putative time window is 20‐30 minutes of continual seizure activity. ⁹ , ¹⁰ It is possible that increased serum NfL concentrations observed in dogs with SE may reflect neuronal injury induced by the presence of the tumor or inflammation, independent of the presence of any seizure activity. This observation is supported by a previous study that also demonstrated that dogs with MUE have increased serum NfL concentrations, although no differences were observed in serum NfL concentrations between those MUE dogs with and without epileptic seizures. ⁴

Plasma NfL concentration increases in a linear fashion in healthy dogs as they age. ³ Although NfL concentrations in dogs with SE were significantly increased compared with dogs with IE, the serum NfL concentrations in dogs with IE in our study were similar to results previously reported in healthy adult dogs, and thus it remains to be determined if dogs with IE have serum NfL concentrations that differ from healthy controls. ³ Although dogs in the SE group were significantly older compared to dogs in the IE group, we did not identify a significant correlation between age and serum NfL concentration in our study. Therefore, age likely had a minimal effect on interpretation of the data, and may reflect the inherent age differences between dogs with IE and SE. ¹⁶ Within the IE group, 8 dogs were between 1 and 6 years of age at the time of seizure onset, but were 7‐8 years old at the time of sampling for NfL measurement. We also did not examine body weight or body mass index in our study, which have been previously shown to contribute to NfL concentration variability in humans. ¹⁹

Our study had some limitations. Although our goal was to include an equal number of dogs with IE and SE, doing so was not possible given our patient population, However, we believe this situation reflects the clinical setting because IE is the most common cause of seizures in dogs. Additionally, most SE dogs in our study were diagnosed with brain tumors, and only 2 were diagnosed with MUE. These findings decrease the generalizability of our data, because several anomalous, degenerative, infectious, and vascular causes of SE were not included. We elected also to include statistical analysis on dogs with brain tumors because this information has not been published previously. Another limitation of our study is the possible misclassification of dogs with IE, because dogs with IE were not required to have brain MRI or CSF analysis. Although this situation replicates a common clinical scenario, some of these dogs may have had underlying structural brain lesions. ⁶ In addition, based on their neurologic examination findings, we assumed that dogs in the SE group did not have additional neurologic disease localized to the spinal cord or peripheral nervous system that possibly could cause or contribute to increased serum NfL concentrations. One dog in the SE group did have mild generalized muscle atrophy on examination. This dog had no other evidence of neurological disease, but it cannot be entirely excluded.

5. CONCLUSION

Our study provides evidence that serum NfL concentration has potential value as a biomarker for the diagnosis of epilepsy in dogs by serving as a supplemental diagnostic tool for differentiating IE and SE. We found that serum NfL concentration is significantly increased in dogs with SE compared with dogs with IE. Additional research is needed to assess serum NfL concentrations in additional causes of SE and its utility as a specific biomarker for the detection of seizures.

CONFLICT OF INTEREST DECLARATION

Authors declare no conflicts of interest.

OFF‐LABEL ANTIMICROBIAL DECLARATION

Authors declare no off‐label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION

This study was approved by IACUC at Virginia‐Maryland College of Veterinary Medicine.

HUMAN ETHICS APPROVAL DECLARATION

Authors declare human ethics approval was not needed for this study.

Fowler KM, Shinn RL, Rossmeisl JH, Parker RL. Evaluation of neurofilament light chain as a biomarker in dogs with structural and idiopathic epilepsy. J Vet Intern Med. 2024;38(3):1577‐1582. doi: 10.1111/jvim.17033