Abstract
Objective
Depression is the most common psychiatric comorbidity in patients with epilepsy (PWE) but remains frequently underdiagnosed. Identifying objective biomarkers may improve early detection and intervention. Neurofilament light chain (NfL), a marker of neuroaxonal injury, has been linked to both epilepsy‐related neuronal damage and depression, yet its role in comorbidity is unclear.
Methods
We conducted a cross‐sectional study including 152 adult PWE recruited from a large tertiary hospital in Northeast China. Depressive symptoms were assessed using the Hamilton Depression Scale (HAMD), and plasma NfL concentrations were measured by enzyme‐linked immunosorbent assay. Binary logistic regression models were applied to examine the association between plasma NfL and depressive symptoms, and linear regression models with HAMD scores as a continuous outcome were performed as sensitivity analyses. Both approaches were conducted with adjustment for potential confounders. Receiver operating characteristic (ROC) analysis was conducted to evaluate the discriminative performance of plasma NfL, and k‐fold cross‐validation (k = 5) was performed to assess the robustness of the ROC analysis.
Results
Of the 152 participants, 61 (40.1%) had depressive symptoms. Higher plasma NfL levels were independently associated with both increased odds of depressive symptoms and higher HAMD scores. ROC analysis demonstrated good discriminative accuracy (area under the curve [AUC] = .838), with an optimal cutoff value of 44.85 pg/mL yielding a sensitivity of .82 and a specificity of .74. The mean AUC obtained from k‐fold cross‐validation was .842, which was consistent with the overall ROC result.
Significance
These findings suggest that plasma NfL may serve as a candidate biomarker for identifying comorbid depression in epilepsy and provide new insight into shared neurobiological mechanisms.
Keywords: biomarker, comorbidity, depressive symptoms, epilepsy, neurofilament light chain
Key points.
Elevated plasma NfL in PWE was significantly associated with an increased risk of depressive symptoms.
Plasma NfL showed good discriminative ability for distinguishing PWE with depressive symptoms from those without depressive symptoms.
Plasma NfL may serve as a promising candidate biomarker reflecting shared neuroaxonal injury mechanisms underlying epilepsy‐related depression.
1. INTRODUCTION
Epilepsy is among the most prevalent neurological disorders globally, characterized by recurrent seizures resulting from abnormal neural discharges. 1 , 2 It affects more than 70 million individuals worldwide and significantly contributes to the global burden of neuropsychiatric diseases. 1 , 3 Psychiatric comorbidities are increasingly prevalent among patients with epilepsy (PWE), with depression being the most frequent. 4 , 5 There exists a well‐documented bidirectional relationship between epilepsy and depression; approximately 20%–55% of PWE exhibit depressive symptoms, a rate substantially higher than the approximately 5% prevalence observed in the general population. 6 This disparity is partly attributable to seizure‐related adversities and social stigma. Conversely, individuals with depression have a 2.5‐fold increased risk of developing epilepsy, indicating that the two conditions may share underlying neurobiological mechanisms beyond psychosocial factors. 7 Importantly, the risk of suicide is elevated approximately fivefold in PWE with comorbid depression, particularly among those with drug‐resistant epilepsy. 8
Despite its significant clinical implications, depression in individuals with epilepsy is frequently underrecognized and inadequately treated. Approximately one fifth of PWE with depression receive no treatment, and nearly one third of cases remain undiagnosed. 9 This underrecognition is partly attributable to atypical manifestations, such as irritability or cognitive difficulties instead of classic mood symptoms, which often leads clinicians to misinterpret them as a natural and adequate reaction to living with epilepsy. 7 , 10 Given the high prevalence and detrimental outcomes associated with the comorbidity of epilepsy and depression, the identification of sensitive and effective biomarkers to facilitate early detection and timely intervention is critically important.
Although depressive symptoms can be identified through clinical interviews, the detection rate in PWE remains suboptimal due to atypical presentations and the overlap of somatic symptoms with epilepsy or medication side effects. An objective blood‐based biomarker, therefore, holds the potential to serve as a triage or screening tool, prompting clinicians to conduct more targeted psychiatric evaluations even in patients who do not overtly complain of mood disturbances. Ideally, such a biomarker would not only flag “hidden” cases but also provide insights into the underlying neurobiological processes linking epilepsy and depression, thereby moving beyond the limitations of subjective symptom scales.
Neurofilaments are essential structural components of the neuronal cytoskeleton. They strengthen axonal resistance to mechanical stress, determine axonal caliber, indirectly regulate conduction velocity, and serve as scaffolds for organelles and other proteins. 11 , 12 Among them, neurofilament light chain (NfL) is the most abundant and soluble subunit. Because NfL is released into cerebrospinal fluid (CSF) and blood following axonal injury or neuronal death, it has been established as a robust biomarker of neuroaxonal degeneration, widely applied in the early detection and prognostic evaluation of cerebrovascular, neurodegenerative, and neuroimmunological disorders. 13 , 14 In 2022, Giovannini et al. first reported dynamic changes of NfL in PWE; serum NfL levels were significantly higher in those with status epilepticus (SE) compared with drug‐resistant epilepsy and nonepileptic controls. Moreover, NfL dynamics were closely associated with antiseizure medication (ASM) response, treatment duration, and clinical outcomes, highlighting its potential as an indicator of SE‐related neuronal injury, therapeutic response, and functional prognosis. 15 In addition, a recent meta‐analysis suggested that elevated NfL may also represent a promising biomarker for the diagnosis of depression. 16
Despite the recognized role of NfL in both epilepsy‐related neuronal injury and the pathophysiology of depression, its relevance as a biomarker in the context of comorbidity between the two remains poorly understood. Most existing studies attribute elevated NfL in epilepsy to seizure‐related acute or chronic axonal injury, whereas the specific association between NfL and depressive symptoms has been largely overlooked. Therefore, the present study focused on plasma NfL to investigate its potential as a biomarker for identifying depressive symptoms in PWE. Clarifying this association may not only provide an objective biological marker for epilepsy‐related depression but also shed light on shared neurobiological mechanisms, such as neuroinflammation and axonal degeneration.
2. MATERIALS AND METHODS
2.1. Study design
Figure 1 depicts the comprehensive study design. This investigation was a cross‐sectional study conducted in Northeast China from March 2023 to December 2024. Participants were consecutively recruited from the Epilepsy Clinic of the Department of Neurology at the First Hospital of Jilin University. Extensive clinical data, neuropsychological assessments, and plasma samples were systematically collected from all participants. The primary objective was to examine the relationship between plasma NfL levels and depressive symptoms in PWE. The study protocol received approval from the Ethics Committee of the First Hospital of Jilin University (approval No. 24K007‐001). Informed consent was obtained in writing from all participants or their legal guardians. All methods were performed in accordance with the relevant guidelines and regulations, including the Declaration of Helsinki.
FIGURE 1.
Overview of the research design in this study. ELISA, enzyme‐linked immunosorbent assay; HAMD, Hamilton Depression Scale; NfL, neurofilament light chain; ROC, receiver operating characteristic.
2.2. Study participants
Between March 2023 and December 2024, adult PWE who were undergoing treatment and follow‐up at the Epilepsy Clinic of the Department of Neurology at the First Hospital of Jilin University were recruited. The diagnosis of epilepsy was made according to the 2017 International League Against Epilepsy (ILAE) classification and confirmed by consensus of at least two experienced neurologists. The inclusion criteria were as follows: (1) a diagnosis of epilepsy based on the ILAE criteria, (2) age range of 18–60 years, (3) seizure‐free for a minimum of 72 h prior to enrollment, (4) availability of blood samples and neuropsychological assessment data, and (5) provision of written informed consent by the patient or their legal representative. The exclusion criteria encompassed the following: (1) the presence of significant neurological disorders, including but not limited to cerebrovascular disease, Parkinson disease, Alzheimer disease, or encephalitis; (2) abnormal brain imaging, such as structural abnormalities (e.g., focal cortical dysplasia or hippocampal sclerosis), central nervous system tumors, or a history of significant head trauma, except for minor lacunar infarctions; (3) the presence of severe systemic illnesses, such as advanced hepatic, renal, or cardiopulmonary diseases, although individuals with early stage or mild systemic conditions, such as essential hypertension, ischemic heart disease, or arrhythmia, were not excluded; and (4) alcohol or substance use disorders, psychotic disorders, bipolar disorder, attention‐deficit/hyperactivity disorder, manic episodes, or other major psychiatric disorders.
2.3. Data collection
Two trained interviewers gathered demographic and clinical data through a structured questionnaire administered during face‐to‐face interviews. The recorded variables included age, sex, educational level, occupation, marital status, place of residence, past medical history, birth history, family history of epilepsy, history of febrile seizures, age at seizure onset, disease duration, average duration of seizures, seizure frequency over the past year, seizure type, and the number of ASMs currently being taken. The 17‐item Hamilton Depression Scale (HAMD‐17) and the 14‐item Hamilton Anxiety Scale (HAMA‐14) were utilized as screening instruments for depressive and anxiety symptoms, respectively, in PWE. Cognitive function was evaluated using the Mini‐Mental State Examination (MMSE). All psychological assessments were conducted by two trained clinicians who had completed standardized training and were blinded to the clinical information of the participants.
2.4. Assessment of depressive symptoms
The HAMD‐17 was used to evaluate depressive symptoms. This instrument comprises 17 items covering domains such as depressed mood, guilt, suicidal ideation, and sleep disturbances. Each item offers up to five response categories—none, mild, moderate, severe, and very severe—scored 0, 1, 2, 3, and 4, respectively. The total score was interpreted as follows: <7, no depressive symptoms; 7–16, mild depression; 17–24, moderate depression; and ≥ 25, severe depression. In this study, PWE who scored ≥ 7 on the HAMD‐17 were classified into the depressive group.
2.5. Assessment of plasma NfL
For each participant, 2 mL of peripheral venous blood was collected in the early morning following an overnight fast and placed into EDTA‐anticoagulated tubes. Plasma was isolated by centrifugation at 3000 rpm for 15 min within 30 min of collection. The resultant supernatant was transferred to EP tubes and stored at −80°C until further analysis. Plasma NfL concentrations were measured using a commercially available sandwich enzyme‐linked immunosorbent assay (ELISA) kit (Jianglai Biological, catalog no. JL32728) according to the manufacturer's instructions. The assay has a detection range of 7.8–500 pg/mL and a sensitivity of 3.7 pg/mL, with intra‐ and interassay coefficients of variation both <10%. All samples and standards were measured in duplicate, and the mean value was used for statistical analysis. Manufacturer‐provided validation data demonstrated acceptable recovery in plasma samples (92%–105%) and good dilution linearity (87%–108%), supporting reliable measurement in plasma matrices. The assay calibration involved standard curves generated from serial dilutions of the provided standards and processed concurrently with the samples. Absorbance was measured at 450 nm using a microplate reader, and concentrations were calculated from the optical density values by curve fitting based on the standard curve. Samples exceeding the upper limit of quantification were diluted with sample diluent and remeasured, and final concentrations were corrected for the dilution factor. Laboratory personnel performing the assays were blinded to the clinical information of the participants, and NfL concentrations were reported in pg/mL.
2.6. Statistical analysis
Categorical data were summarized as n (%), with group differences evaluated using the chi‐squared test. For continuous variables, the Shapiro–Wilk test was applied to examine distributional normality. Variables with normal distributions were reported as mean ± SD and compared using independent samples t‐tests, whereas nonnormally distributed variables were presented as median with interquartile range and analyzed with the Mann–Whitney U‐test. Binary logistic regression models were used to evaluate the association between depressive comorbidity and plasma NfL levels as well as other covariates, with potential confounders adjusted as covariates in the models. Sensitivity analyses were further performed using linear regression models to assess the relationship between HAMD‐17 scores and plasma NfL concentrations after adjustment for confounders. Three progressively adjusted regression models were constructed. Model 1 was adjusted for age and sex; Model 2 was further adjusted for occupation type, educational level, HAMA scores, and MMSE scores; and Model 3 was additionally adjusted for the number of ASMs, seizure frequency during the past year, disease duration, age at seizure onset, average seizure duration, and history of SE. Receiver operating characteristic (ROC) curve analysis was applied to evaluate the discriminative performance of plasma NfL as a potential biomarker for depression. The area under the curve (AUC) was calculated to determine discriminative accuracy, and the optimal cutoff value was identified based on the maximum Youden index. To further evaluate the robustness of the ROC analysis and reduce the potential overfitting associated with a single data split, k‐fold cross‐validation (k = 5) was performed on the entire dataset. The dataset was randomly partitioned into five subsets; in each iteration, four subsets were used for model training and the remaining subset served as the test fold for performance evaluation. The AUC, sensitivity, and specificity were calculated for each test fold, and the mean values across all folds are reported. All statistical analyses were performed using SPSS software (version 29.0) and R software (version 4.5.1). p < .05 was deemed statistically significant, and all statistical tests were two‐sided.
3. RESULTS
3.1. Characteristics of the study participants
Among the 152 participants, 61 (40.1%) were identified as having depressive symptoms based on HAMD‐17 assessment. The mean age of all participants was 38.9 years, and 52% were male; 54.6% had attained an education level above high school. As shown in Table 1, significant differences between patients with and without depressive symptoms were observed in age, educational level, history of SE, HAMA scores, HAMD scores, and plasma NfL levels (all p < .05). In contrast, no significant group differences were found in sex, residence, occupation type, marital status, smoking history, drinking history, birth history, history of febrile seizures, family history of epilepsy, age at seizure onset, disease duration, seizure duration, seizure type, seizure frequency during the past year, number of ASMs used, electroencephalographic abnormalities, or MMSE scores (all p > .05). Figure 2 shows the distribution of plasma NfL levels in PWE with and without depressive symptoms. Patients with depressive symptoms exhibited a right‐shifted distribution, indicating generally higher plasma NfL levels compared with those without depressive symptoms.
TABLE 1.
Demographic and clinical characteristics of patients with epilepsy grouped according to depressive symptoms.
| Variable | All participants, n = 152 | Depressive symptoms | p | |
|---|---|---|---|---|
| Without depressive symptoms, n = 91 | With depressive symptoms, n = 61 | |||
| Age, years, mean± SD | 38.89 ± 9.23 | 37.38 ± 9.52 | 41.13 ± 8.36 | .014* |
| Sex, n (%) | .573 | |||
| Male | 79 (52.0%) | 49 (53.8%) | 30 (49.2%) | |
| Female | 73 (48.0%) | 42 (46.2%) | 31 (50.8%) | |
| Residence, n (%) | .136 | |||
| Urban | 76 (50.0%) | 50 (54.9%) | 26 (42.6%) | |
| Rural | 76 (50.0%) | 41 (45.1%) | 35 (57.4%) | |
| Educational level, n (%) | .025* | |||
| College or above | 56 (36.8%) | 41 (45.1%) | 15 (24.6%) | |
| High school | 27 (17.8%) | 16 (17.6%) | 11 (18.0%) | |
| Middle school or below | 69 (45.4%) | 34 (37.4%) | 35 (57.4%) | |
| Occupation type, n (%) | .261 | |||
| Primarily mental labor | 59 (38.8%) | 39 (42.9%) | 20 (32.8%) | |
| Primarily manual labor | 58 (38.2%) | 30 (33.0%) | 28 (45.9%) | |
| Unemployed | 35 (23.0%) | 22 (24.2%) | 13 (21.3%) | |
| Marital status, n (%) | .970 | |||
| Married | 85 (55.9%) | 51 (56.0%) | 34 (55.7%) | |
| Single or divorced | 67 (44.1%) | 40 (44.0%) | 27 (44.3%) | |
| Smoking history, n (%) | .293 | |||
| Yes | 38 (25.0%) | 20 (22.0%) | 18 (29.5%) | |
| No | 114 (75.0%) | 71 (78.0%) | 43 (70.5%) | |
| Drinking history, n (%) | .657 | |||
| Yes | 37 (24.3%) | 21 (23.1%) | 16 (26.2%) | |
| No | 115 (75.7%) | 70 (76.9%) | 45 (73.8%) | |
| Birth history, n (%) | .543 | |||
| Vaginal delivery | 117 (77.0%) | 70 (76.9%) | 47 (77.0%) | |
| Cesarean section | 26 (17.1%) | 17 (18.7%) | 9 (14.8%) | |
| Birth injury/asphyxia | 9 (5.9%) | 4 (4.4%) | 5 (8.2%) | |
| History of febrile seizures, n (%) | .390 | |||
| Yes | 22 (14.5%) | 15 (16.5%) | 7 (11.5%) | |
| No | 130 (85.5%) | 76 (83.5%) | 54 (88.5%) | |
| Family history of epilepsy, n (%) | .643 | |||
| Yes | 13 (8.6%) | 7 (7.7%) | 6 (9.8%) | |
| No | 139 (91.4%) | 84 (92.3%) | 55 (90.2%) | |
| Age at onset, years, mean± SD | 31.23 ± 10.04 | 30.04 ± 9.95 | 33.01 ± 10.00 | .075 |
| Disease duration, years, median (Q1–Q3) | 6.0 (3.0–11.0) | 6.0 (3.0–11.0) | 7.0 (3.0–13.0) | .632 |
| Seizure duration, min, median (Q1–Q3) | 2.0 (1.5–3.0) | 2.0 (1.5–3.0) | 2.0 (2.0–3.0) | .205 |
| Seizure type, n (%) | .713 | |||
| FAS | 10 (6.6%) | 5 (5.5%) | 5 (8.2%) | |
| FIAS | 59 (38.8%) | 36 (39.6%) | 23 (37.7%) | |
| FBTCS | 63 (41.4%) | 36 (39.6%) | 27 (44.3%) | |
| GOS | 14 (9.2%) | 9 (9.9%) | 5 (8.2%) | |
| Unknown origin | 6 (3.9%) | 5 (5.5%) | 1 (1.6%) | |
| Seizure frequency in past year, n (%) | .497 | |||
| <1 time/year | 36 (23.7%) | 24 (26.4%) | 12 (19.7%) | |
| ≥1 time/year | 63 (41.4%) | 38 (41.8%) | 25 (41.0%) | |
| ≥1 time/month | 41 (27.0%) | 24 (26.4%) | 17 (27.9%) | |
| ≥1 time/week | 12 (7.9%) | 5 (5.5%) | 7 (11.5%) | |
| Number of ASMs used, n (%) | .387 | |||
| 0 | 19 (12.5%) | 12 (13.2%) | 7 (11.5%) | |
| 1 | 67 (44.1%) | 38 (41.8%) | 29 (47.5%) | |
| 2 | 48 (31.6%) | 27 (29.7%) | 21 (34.4%) | |
| ≥3 | 18 (11.8%) | 14 (15.4%) | 4 (6.6%) | |
| History of status epilepticus, n (%) | .003** | |||
| Yes | 20 (13.2%) | 6 (6.6%) | 14 (23.0%) | |
| No | 132 (86.8%) | 85 (93.4%) | 47 (77.0%) | |
| EEG abnormality, n (%) | .531 | |||
| Mild | 22 (14.5%) | 11 (12.1%) | 11 (18.0%) | |
| Moderate | 89 (58.6%) | 56 (61.5%) | 33 (54.1%) | |
| Severe | 41 (27.0%) | 24 (26.4%) | 17 (27.9%) | |
| HAMA score, median (Q1–Q3) | 5.0 (3.0–8.0) | 3.0 (2.0–5.0) | 8.0 (5.0–11.0) | <.001*** |
| HAMD score, median (Q1–Q3) | 5.0 (3.25–9.0) | 4.0 (2.0–5.0) | 10.0 (8.0–15.0) | <.001*** |
| MMSE score, mean± SD | 27.16 ± 3.17 | 27.52 ± 3.12 | 26.62 ± 3.20 | .088 |
| Plasma NfL, pg/mL, median (Q1–Q3) | 41.0 (32.4–55.6) | 35.8 (29.0–42.8) | 56.5 (43.5–64.5) | <.001*** |
Abbreviations: ASM, antiseizure medication; EEG, electroencephalographic; FAS, focal aware seizures; FBTCS, focal to bilateral tonic–clonic seizures; FIAS, focal impaired awareness seizures; GOS, generalized onset seizures; HAMA, Hamilton Anxiety Scale; HAMD, Hamilton Depression Scale; MMSE, Mini‐Mental State Examination; NfL, neurofilament light chain.
p < .05.
p < .01.
p < .001.
FIGURE 2.
Distribution of plasma neurofilament light chain (NfL) levels in patients with epilepsy with and without depressive symptoms. The upper panel shows patients without depressive symptoms, and the lower panel shows patients with depressive symptoms. Histograms depict the frequency distribution, with overlaid kernel density curves illustrating the distribution of plasma NfL levels in each group.
3.2. Association between depressive symptoms and plasma NfL in PWE: Logistic regression models
Three logistic regression models were applied to examine the independent association between plasma NfL levels and depressive symptoms. Figure 3 presents the odds ratios (ORs), 95% confidence intervals (CIs), and p‐values. In the unadjusted model, higher plasma NfL concentrations, treated as a continuous variable, were significantly related to an increased risk of depressive symptoms (OR = 1.12, 95% CI = 1.08–1.16, p < .001). This relationship remained statistically significant after adjustment for age and sex (OR = 1.12, 95% CI = 1.08–1.16, p < .001). Additional adjustment for further covariates did not attenuate the association; Model 2 produced an OR of 1.11 (95% CI = 1.05–1.16, p < .001), and Model 3 yielded an OR of 1.12 (95% CI = 1.06–1.19, p < .001).
FIGURE 3.
Associations between depression and plasma neurofilament light chain (NfL) levels in patients with epilepsy after adjustment for different covariates. CI, confidence interval; HAMD, Hamilton Depression Scale; OR, odds ratio.
3.3. Association between HAMD scores and plasma NfL in PWE: Linear regression models
As a sensitivity analysis of the logistic regression, we applied linear regression with HAMD score as a continuous outcome to assess its association with plasma NfL concentration. The results are shown in Figure 3. Plasma NfL levels were positively associated with HAMD scores across all models. In the unadjusted model, there was a positive association (β = .23, 95% CI = .18–.27, p < .001). In Model 1, the effect size was virtually identical (β = .23, 95% CI = .18–.28, p < .001). In Models 2 and 3, the association remained significant but was substantially attenuated after adjusting for additional covariates, with the β coefficient decreasing from .23 in the unadjusted model to .12 in the fully adjusted models (Model 2: β = .12, 95% CI = .08–.17, p < .001; Model 3: β = .12, 95% CI = .07–.17, p < .001).
3.4. Discriminative value of plasma NfL for depression in PWE: ROC analysis
As shown in Figure 4, an ROC curve was plotted to evaluate the discriminative value of plasma NfL levels for depressive symptoms in PWE. The AUC was .838 (95% CI = .772–.904, p < .001), indicating excellent discriminatory ability. The optimal cutoff value, determined by the maximum Youden index (.562), was 44.85 pg/mL, corresponding to a sensitivity of .82 and a specificity of .74. Fivefold cross‐validation showed consistent discriminative performance. The AUC values across the five folds ranged from .792 to .912, with a mean AUC of .842. The mean sensitivity and specificity were .83 and .79, respectively (Table 2). These results were broadly consistent with the overall ROC analysis, supporting the stability of the discriminative performance of plasma NfL.
FIGURE 4.
Receiver operating characteristic curve of plasma neurofilament light chain for the predictive diagnostic value of depressive symptoms in patients with epilepsy. AUC, area under the curve; CI, confidence interval; FPR, false positive rate; Se, sensitivity; Sp, specificity; TPR, true positive rate.
TABLE 2.
Results of fivefold cross‐validation.
| Fold | N‐test set | AUC | Sensitivity | Specificity |
|---|---|---|---|---|
| Fold 1 | 31 | .825 | .92 | .67 |
| Fold 2 | 30 | .833 | .75 | .83 |
| Fold 3 | 31 | .912 | .92 | .79 |
| Fold 4 | 30 | .847 | .75 | .89 |
| Fold 5 | 30 | .792 | .83 | .78 |
| Mean | 152 | .842 | .83 | .79 |
Abbreviation: AUC, area under the curve.
4. DISCUSSION
To our knowledge, this is the first study to investigate the association between NfL levels and depressive symptoms in PWE. In our cohort, depressive symptoms were initially associated with older age, lower educational attainment, a history of SE, anxiety symptoms, and higher plasma NfL levels. After adjusting for confounding factors, higher plasma NfL levels remained independently correlated with depressive symptoms and higher HAMD scores. Moreover, ROC curve analysis demonstrated that plasma NfL has good discriminative value for identifying depressive symptoms in PWE, suggesting that it may serve as a candidate biomarker for detecting depression in this population.
In this study, PWE with depressive symptoms exhibited a higher mean age (41 years), which may reflect a cumulative effect associated with longer epilepsy duration, although the difference in disease duration between groups did not reach statistical significance. Roussos et al. have suggested that individuals around the age of 40 years are more vulnerable to depressive symptoms, partly because they may experience declining capacity to perform previously manageable social or functional roles. 17 We also found that PWE with an educational level of middle school or below were more likely to report depressive symptoms. Limited access to health care resources and less effective stress‐coping strategies in this population may contribute to greater social isolation and reduced psychosocial adaptability, thereby increasing vulnerability to depression. 17 , 18 In addition, a history of early life SE was associated with a significantly higher risk of depression. A plausible explanation is that recurrent seizure activity may induce structural and functional remodeling within temporal lobe regions, which has been implicated in the development of psychiatric disorders. 19
Elevated levels of NfL in the blood of PWE reflect excitotoxicity‐induced neuronal injury. This process is primarily mediated by excessive glutamatergic neurotransmission, leading to intracellular calcium overload and ultimately resulting in neuronal apoptosis, axonal degeneration, and synaptic dysfunction. 20 , 21 As a major structural component of the axonal cytoskeleton, NfL is released into the extracellular space following neuronal damage and subsequently enters both the CSF and peripheral circulation.
Concurrently, persistent neuroinflammation has been established as a critical pathogenic mechanism of depression. Chronic stress or peripheral immune activation induces sustained microglial activation in the central nervous system. 22 , 23 Activated microglia secrete proinflammatory cytokines such as interleukin‐6 and tumor necrosis factor‐α, which not only disrupt synaptic function but also contribute to neuronal and axonal injury through the induction of oxidative stress and excitotoxicity, thereby promoting the release of NfL. 24 In addition, both epileptic seizures and depressive states can compromise blood–brain barrier integrity, facilitating further leakage of NfL from the CSF into the peripheral blood. Hyperactivation of the hypothalamic–pituitary–adrenal (HPA) axis represents one of the earliest recognized neurobiological abnormalities in depression. 25 Chronic stress associated with depression can alter glucocorticoid levels, which are key hormones regulated by the HPA axis, and exert widespread negative effects on cellular physiology. 26 Prolonged stress reduces the mitochondrial transport of glucocorticoid receptors, leading to decreased expression of oxidative phosphorylation‐related genes and alterations in mitochondrial DNA transcription. The accumulation of reactive metabolites, reductions in mitochondrial membrane potential, and impaired calcium‐buffering capacity across various brain regions further contribute to mitochondrial dysfunction. 27 Ma et al. reported that corticosterone treatment in PC12 cells caused marked mitochondrial ultrastructural damage, including mitochondrial swelling and cristae fragmentation. This was accompanied by increased cytochrome‐c release and caspase‐3 activation, indicating apoptosis through mitochondrial pathways. Corticosterone has also been shown to impair autophagic flux by disrupting AMP‐activated protein kinase (AMPK) / mechanistic target of rapamycin (mTOR) signaling in neurons, which reduces the clearance of damaged or senescent organelles and misfolded proteins and ultimately results in neuronal injury. 28 These biological processes provide a plausible explanation for the elevated NfL levels observed in individuals with depressive symptoms.
Notably, NfL itself may exacerbate neurodegeneration through a feed‐forward mechanism. Recent studies have shown that although human leukocyte antigen–DR isotype‐expressing microglia can phagocytose axonal debris, this process may also trigger aberrant microglial proliferation and immune dysregulation. 29 , 30 , 31 A recent in vivo study using single‐cell RNA sequencing analyzed brain cells after recombinant NfL was injected into the hippocampus of mice and revealed a robust macrophage and microglial response. In the SOD1 mouse model of amyotrophic lateral sclerosis, NfL knockout consistently attenuated microgliosis and delayed symptom onset. 32 Furthermore, Gong et al. reported that NfL pretreatment induces microglia to release exosomes containing ferritin heavy chain, which may aggravate neuronal ferroptosis through noncanonical pathways. 33
Several studies have explored the potential mechanisms underlying the impact of depression and anxiety on cognitive function. Chronic anxiety and depression are linked to dysregulation of the HPA axis, leading to persistently elevated cortisol levels. 34 , 35 Elevated cortisol exerts neurotoxic effects on the brain. 36 NfL, a marker of axonal injury, reflects shared neurobiological processes underlying both anxiety and depression, such as chronic neuroinflammation and HPA axis dysfunction, which contribute to cumulative brain damage. In the current study, the association between plasma NfL levels and depressive symptoms remained significant even after adjusting for HAMA scores in the statistical models. This finding suggests that NfL‐indicated axonal injury may be more directly and specifically related to depressive symptoms, rather than being solely mediated by anxiety.
Taken together, these findings suggest that elevated plasma NfL in PWE may reflect the combined effects of excitotoxicity, neuroinflammation, HPA axis dysregulation, and blood–brain barrier dysfunction, all of which are also implicated in depression. Therefore, investigating plasma NfL as a biomarker for depressive symptoms in epilepsy may provide novel insights into the shared neurobiological mechanisms underlying this comorbidity and facilitate early identification and intervention.
Our findings may be influenced by several potential confounders, such as the effects of age and neurodegenerative diseases on NfL levels. NfL increases with increasing age and is markedly elevated in neurodegenerative diseases. However, individuals with major neurological disorders were excluded at enrollment. Moreover, the mean age of our cohort was relatively young (38.9 years), well below the typical onset period for clinically significant dementia. Age and global cognitive function were rigorously adjusted for in all multivariable regression models. Therefore, differences in age or cognitive status are unlikely to fully account for the associations we observed.
However, several limitations of this study should be acknowledged. First, the cross‐sectional design of our study only allows us to identify a potential association between plasma NfL levels and depressive symptoms in PWE, suggesting that NfL may serve as a candidate biomarker for detecting comorbid depression in epilepsy. However, causality cannot be inferred, nor can the prognostic value of NfL be established. Longitudinal cohort studies are needed to clarify the temporal relationship and to determine whether elevated plasma NfL can predict the onset or worsening of depressive symptoms in PWE. Second, the relatively small sample size, drawn from a single center, may limit the generalizability of our results. Third, this study did not include a healthy control group. Therefore, the findings only reflect relative differences in plasma NfL levels within the PWE population, and it cannot be determined whether NfL levels were pathologically elevated. Accordingly, the optimal cutoff value derived from ROC analysis should be interpreted with caution, and its generalizability to other populations requires further validation. Fourth, plasma NfL levels were measured using a conventional ELISA assay in this study. Compared with ultrasensitive platforms such as single molecule array (Simoa), ELISA has lower analytical sensitivity for detecting low NfL concentrations. However, the detection range and sensitivity of the assay used were sufficient to cover the NfL concentrations observed in our cohort. Future studies using ultrasensitive assays may further validate these findings. Fifth, NfL is a nonspecific biomarker of neuroaxonal injury and cannot specifically reflect depressive symptoms. Although we adjusted for potential confounders such as age, sex, and history of SE, residual confounding cannot be completely excluded. Unmeasured factors, such as cumulative seizure burden or age‐related subclinical neurodegeneration, may also influence NfL levels. Future studies incorporating more detailed epilepsy severity measures and longitudinal designs are needed to further clarify these relationships.
5. CONCLUSIONS
The findings of this study indicate that elevated plasma NfL levels are significantly associated with depressive symptoms in PWE. Plasma NfL holds as a candidate biomarker for identifying comorbid depression in epilepsy. Future research should employ multicenter prospective cohort designs and in‐depth mechanistic studies to further validate these results and facilitate their translation into clinical practice and risk management.
AUTHOR CONTRIBUTIONS
Zhiqing Chen conceptualized and designed the study, performed the research, and drafted the initial manuscript. Huaiyu Sun, Jingqi Lin, and Jiaai Li contributed to the statistical analyses. Jingyi Yao, Wuqiong Zhang, and Shuai Hou assisted with data collection. Hongmei Meng critically reviewed and revised the manuscript. All authors actively contributed to the study, reviewed the manuscript, and approved the final version for submission.
FUNDING INFORMATION
This study was supported by the National Natural Science Foundation of China (81871008 and 81801145), Jilin Provincial Scientific and Technological Development Program (20240304165SF), the Department of Education of Jilin Province (2025KC102), and a Jilin Province Medical and Health Talent Special Project (JLSWSRCZX2023‐88).
CONFLICT OF INTEREST STATEMENT
The researchers confirm that the investigation was conducted without any potential conflicts of interest related to business or financial affiliations. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.
PATIENT CONSENT STATEMENT
Written informed consent was obtained from all participants, who voluntarily agreed to participate and provided relevant clinical information and blood samples.
ACKNOWLEDGMENTS
We thank the Department of Biobank, Division of Clinical Research, First Hospital of Jilin University, for providing human tissue samples.
Contributor Information
Shuai Hou, Email: houshuai@jlu.edu.cn.
Hongmei Meng, Email: menghm@jlu.edu.cn.
DATA AVAILABILITY STATEMENT
The datasets generated and analyzed in this study are all available in the article. Contact the corresponding author for additional information.
REFERENCES
- 1. Thijs RD, Surges R, O'Brien TJ, Sander JW. Epilepsy in adults. Lancet. 2019;393(10172):689–701. 10.1016/s0140-6736(18)32596-0 [DOI] [PubMed] [Google Scholar]
- 2. Chen Z, Sun H, Zhang W, Hou S, Yang X, Lin J, et al. Exploring correlations between immune cell phenotypes and the risk of epilepsy: a bidirectional mendelian randomization study. Epilepsy Behav. 2024;157:109896. 10.1016/j.yebeh.2024.109896 [DOI] [PubMed] [Google Scholar]
- 3. Ouriques Martins SC, Sacks C, Hacke W, Brainin M, de Assis Figueiredo F, Marques Pontes‐Neto O, et al. Priorities to reduce the burden of stroke in Latin American countries. Lancet Neurol. 2019;18(7):674–683. 10.1016/s1474-4422(19)30068-7 [DOI] [PubMed] [Google Scholar]
- 4. Scott AJ, Sharpe L, Hunt C, Gandy M. Anxiety and depressive disorders in people with epilepsy: a meta‐analysis. Epilepsia. 2017;58(6):973–982. 10.1111/epi.13769 [DOI] [PubMed] [Google Scholar]
- 5. Zhong R, Chen Q, Li M, Li N, Chu C, Li J, et al. A cross‐sectional study on the Association of Serum Uric Acid Levels with depressive and anxiety symptoms in people with epilepsy. BMC Psychiatry. 2021;21(1):17. 10.1186/s12888-020-03019-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Bølling‐Ladegaard E, Dreier JW, Kessing LV, Budtz‐Jørgensen E, Lolk K, Christensen J. Directionality of the association between epilepsy and depression: a Nationwide register‐based cohort study. Neurology. 2023;100(9):e932–e942. 10.1212/wnl.0000000000201542 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Shokr MM, Eladawy RM, Azar YO, Al Raish SM. Probiotics and the gut‐brain Axis: emerging therapeutic strategies for epilepsy and depression comorbidity. Foods. 2025;14(17):2926. 10.3390/foods14172926 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Kanner AM. Major depression, anxiety disorder and suicidality in epilepsy: what should neurologists do? Epilepsy Behav Rep. 2025;30:100758. 10.1016/j.ebr.2025.100758 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Boylan LS, Flint LA, Labovitz DL, Jackson SC, Starner K, Devinsky O. Depression but not seizure frequency predicts quality of life in treatment‐resistant epilepsy. Neurology. 2004;62(2):258–261. 10.1212/01.wnl.0000103282.62353.85 [DOI] [PubMed] [Google Scholar]
- 10. Elger CE, Johnston SA, Hoppe C. Diagnosing and treating depression in epilepsy. Seizure. 2017;44:184–193. 10.1016/j.seizure.2016.10.018 [DOI] [PubMed] [Google Scholar]
- 11. Thebault S, Booth RA, Freedman MS. Blood neurofilament light chain: the Neurologist's troponin? Biomedicine. 2020;8(11):523. 10.3390/biomedicines8110523 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Chen Z, Guo Y, Yao J, Lin J, Sun H, Li J, et al. Blood neurofilament light chain as a biomarker for cognitive impairment in adults with epilepsy: integrated evidence from clinical cohorts in Northeast China and European Gwas data. Epilepsia. 2026;67(1):341–356. 10.1111/epi.18659 [DOI] [PubMed] [Google Scholar]
- 13. Gaetani L, Blennow K, Calabresi P, Di Filippo M, Parnetti L, Zetterberg H. Neurofilament light chain as a biomarker in neurological disorders. J Neurol Neurosurg Psychiatry. 2019;90(8):870–881. 10.1136/jnnp-2018-320106 [DOI] [PubMed] [Google Scholar]
- 14. Jung Y, Damoiseaux JS. The potential of blood neurofilament light as a marker of neurodegeneration for Alzheimer's disease. Brain. 2024;147(1):12–25. 10.1093/brain/awad267 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Giovannini G, Bedin R, Ferraro D, Vaudano AE, Mandrioli J, Meletti S. Serum neurofilament light as biomarker of seizure‐related neuronal injury in status epilepticus. Epilepsia. 2022;63(1):e23–e29. 10.1111/epi.17132 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Habibzadeh A, Ostovan VR, Ghezel MA, Kavari K, Kardeh S, Tabrizi R. Neurofilament light chain as a promising biomarker for depression diagnosis: a systematic review and meta‐analysis. BMC Psychiatry. 2024;24(1):617. 10.1186/s12888-024-06051-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Roussos S, Gerogianni G, Vasilopoulos G, Koutelekos I, Pavlatou N, Kalogianni A, et al. Factors associated with anxiety and depression in persons with epilepsy (Pwe). Cureus. 2024;16(3):e56401. 10.7759/cureus.56401 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Engidaw NA, Bacha L, Kenea A. Prevalence of depression and associated factors among epileptic patients at Ilu Ababore zone hospitals, south West Ethiopia, 2017: a cross‐sectional study. Ann Gen Psychiatry. 2020;19:19. 10.1186/s12991-020-00268-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Gom RC, George AG, Harris SA, Wickramarachchi P, Bhatt D, Acharjee S, et al. Emotional comorbidities in epilepsy result from seizure‐induced corticosterone activity. Neurobiol Stress. 2024;33:100678. 10.1016/j.ynstr.2024.100678 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Barker‐Haliski M, White HS. Glutamatergic mechanisms associated with seizures and epilepsy. Cold Spring Harb Perspect Med. 2015;5(8):a022863. 10.1101/cshperspect.a022863 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Sears SM, Hewett SJ. Influence of glutamate and Gaba transport on brain excitatory/inhibitory balance. Exp Biol Med (Maywood). 2021;246(9):1069–1083. 10.1177/1535370221989263 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Sălcudean A, Bodo CR, Popovici RA, Cozma MM, Păcurar M, Crăciun RE, et al. Neuroinflammation‐a crucial factor in the pathophysiology of depression‐a comprehensive review. Biomolecules. 2025;15(4):502. 10.3390/biom15040502 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Sarapultsev A, Gusev E, Chereshnev V, Komelkova M, Hu D. Neuroimmune interactions in stress and depression: exploring the molecular and cellular mechanisms within the neuroinflammation‐depression nexus. Curr Med Chem. 2024. 10.2174/0109298673320710240920055041 [DOI] [PubMed] [Google Scholar]
- 24. Saraste M, Matilainen M, Vuorimaa A, Laaksonen S, Sucksdorff M, Leppert D, et al. Association of Serum Neurofilament Light with microglial activation in multiple sclerosis. J Neurol Neurosurg Psychiatry. 2023;94(9):698–706. 10.1136/jnnp-2023-331051 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Wang Y, Liu L, Gu JH, Wang CN, Guan W, Liu Y, et al. Salt‐inducible kinase 1‐Creb‐regulated transcription coactivator 1 Signalling in the paraventricular nucleus of the hypothalamus plays a role in depression by regulating the hypothalamic‐pituitary‐adrenal Axis. Mol Psychiatry. 2024;29(6):1660–1670. 10.1038/s41380-022-01881-4 [DOI] [PubMed] [Google Scholar]
- 26. Choi GE, Han HJ. Glucocorticoid impairs mitochondrial quality control in neurons. Neurobiol Dis. 2021;152:105301. 10.1016/j.nbd.2021.105301 [DOI] [PubMed] [Google Scholar]
- 27. Vyas S, Rodrigues AJ, Silva JM, Tronche F, Almeida OF, Sousa N, et al. Chronic stress and glucocorticoids: from neuronal plasticity to neurodegeneration. Neural Plast. 2016;2016:6391686. 10.1155/2016/6391686 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Ma RD, Zhou GJ, Qu M, Yi JH, Tang YL, Yang XY, et al. Corticosterone induces neurotoxicity in Pc12 cells via disrupting autophagy flux mediated by Ampk/Mtor signaling. CNS Neurosci Ther. 2020;26(2):167–176. 10.1111/cns.13212 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Huizinga R, van der Star BJ, Kipp M, Jong R, Gerritsen W, Clarner T, et al. Phagocytosis of neuronal debris by microglia is associated with neuronal damage in multiple sclerosis. Glia. 2012;60(3):422–431. 10.1002/glia.22276 [DOI] [PubMed] [Google Scholar]
- 30. Rajda C, Galla Z, Polyák H, Maróti Z, Babarczy K, Pukoli D, et al. Cerebrospinal fluid neurofilament light chain is associated with kynurenine pathway metabolite changes in multiple sclerosis. Int J Mol Sci. 2020;21(8):2665. 10.3390/ijms21082665 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Parbo P, Madsen LS, Ismail R, Zetterberg H, Blennow K, Eskildsen SF, et al. Low plasma neurofilament light levels associated with raised cortical microglial activation suggest inflammation acts to protect prodromal Alzheimer's disease. Alzheimer's Res Ther. 2020;12(1):3. 10.1186/s13195-019-0574-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Kahn OI, Dominguez SL, Glock C, Hayne M, Vito S, Sengupta Ghosh A, et al. Secreted neurofilament light chain after neuronal damage induces myeloid cell activation and neuroinflammation. Cell Rep. 2025;44(3):115382. 10.1016/j.celrep.2025.115382 [DOI] [PubMed] [Google Scholar]
- 33. Gong L, Yu Q, Wang H, Xu C, Dou Y, Mao B, et al. Neurofilament light chain (Nf‐L) stimulates lipid peroxidation to neuronal membrane through microglia‐derived ferritin heavy chain (Fth) secretion. Oxid Med Cell Longev. 2022;2022:3938940. 10.1155/2022/3938940 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Mikulska J, Juszczyk G, Gawrońska‐Grzywacz M, Herbet M. Hpa Axis in the Pathomechanism of depression and schizophrenia: new therapeutic strategies based on its participation. Brain Sci. 2021;11(10):1298. 10.3390/brainsci11101298 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Nandam LS, Brazel M, Zhou M, Jhaveri DJ. Cortisol and major depressive disorder‐translating findings from humans to animal models and Back. Front Psychiatry. 2019;10:974. 10.3389/fpsyt.2019.00974 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Qiu H, Wang Z, Wang Y. Influence of comorbid anxiety and depression disorder on cognition in older adults with epilepsy. Front Psychiatry. 2025;16:1598767. 10.3389/fpsyt.2025.1598767 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The datasets generated and analyzed in this study are all available in the article. Contact the corresponding author for additional information.