Neurofilament light chain level is associated with lifetime suicidal behaviors

Abstract

INTRODUCTION

Suicide is a major global public health problem, accounting for approximately 800 000 deaths per year. In 2019, suicides accounted for approximately 1.9% of the global disease burden (World Health Organization, 2019). To date, the suicide rate in Taiwan is higher than the global average (Chang et al., 2020), with an average of 9.8 people committing suicide every day. More than 90% of people who died by suicide had received a diagnosis of mental illness (Arsenault-Lapierre et al., 2004), and mood disorder was identified to be the most related to suicide (Bachmann, 2018). Patients with unipolar depression are >20 times more likely to die by suicide than the general population (Harris and Barraclough, 1998). The suicide rate in patients with bipolar disorder is approximately 10 to 30 times higher than that in the general population (Dome et al., 2019).

Although many scales have been developed to evaluate the risk of suicide in patients with mental illness and many studies have determined the characteristics of individuals more likely to commit suicide, the sensitivity and specificity of these measures are insufficient for their application in clinical settings (Runeson et al., 2017). Suicide risk is mainly evaluated on the basis of the active expression of suicide intention by patients, but people at risk of suicide often do not disclose their suicide intention to others. Thus, studies have been attempting to determine biomarkers that can be used to identify those at risk of suicide, especially those with mood disorders. Biomarkers are defined as “biological molecules found in the blood, other body fluids, or tissues that are a sign of a normal or abnormal process or of a condition or disease” (Califf, 2018). Many biomarkers may be suitable for evaluating suicide risk, including neuroimages (van Heeringen and Mann, 2014), gene variants (Mamdani et al., 2022), proteomics (Peng et al., 2018), and inflammatory markers (Courtet et al., 2016). Until now, no definite biomarker exists for identifying those at suicide risk.

Neurofilament light chain (NfL) is a neuronal cytoplasmic protein. Its function is to maintain the stability of neuronal axons. NfL is abundant in neurons in the brain (Yuan et al., 2017). In case of neuronal damage, NfL is released in large quantities into the cerebrospinal fluid (CSF) and then into blood. The NfL levels in the CSF are approximately 40 times higher than that in serum. The NfL levels both in the CSF and blood have been determined to be closely related to the severity of various neurological diseases, including neurodegenerative, traumatic brain, and cerebrovascular diseases (Khalil et al., 2018).

The neurobiological mechanisms of many psychiatric disorders, such as major depressive disorder (MDD), bipolar disorder, and substance use disorder, have been reported to be related to neuroinflammation and neuroaxonal damage. Findings on the use of NfL as a biomarker in patients with mental illness are inconclusive, with both positive and negative associations being reported. Two studies have reported that patients with bipolar disorder had higher NfL levels than did healthy controls (Jakobsson et al., 2014; Aggio et al., 2022). Another study determined higher NfL levels in patients with ketamine use disorder, especially those with a lifetime history of MDD (Liu et al., 2021). Chen et al. demonstrated that patients with MDD had significantly higher blood NfL levels than did healthy controls, and higher NfL levels were associated with higher tumor necrosis factor-alpha levels and poorer executive function (Chen et al., 2022). By contrast, some studies did not identify differences in the NfL levels between individuals with mental illness and healthy controls (Eratne et al., 2020). Few studies have suggested an association between NfL and depressive symptoms, and a case–control study reported elevated levels of NfL in individuals with MDD (Chen et al., 2022). In patients with Parkinson’s disease, NfL was associated with increased risk of depressive and anxiety symptoms (Yin et al., 2022). As suicide is a particularly severe symptom in mood disorders, we are interested in exploring the relationship between NfL levels and suicidal behavior. To date, only one study has examined the association between the NfL levels and suicidal behavior, and the results revealed significantly higher NfL levels in individuals with suicide attempts than in those without suicide attempts (Ramezani et al., 2022). Given the scant research on the association of NfL with depression and suicide, this study investigated whether the NfL levels differ between patients with MDD and healthy controls and whether NfL is associated with suicidal behaviors.

METHODS

Patient Recruitment

Patients aged between 20 and 65 years who met the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) criteria for MDD were enrolled in the study group. We excluded patients with the following: (1) major medical disease such as metastatic cancer, brain tumor, decompensated cardiac, hepatic, or renal failure, or myocardial infarction or stroke within the 3 months preceding the study (2) neurological diseases such as delirium, Parkinson’s disease, epilepsy, aphasia, or multiple sclerosis or (3) a history of traumatic brain injury. Age- and sex-matched healthy controls who did not receive a diagnosis of psychiatric disorders or any major medical or neurological disease (ie, epilepsy, stroke, traumatic brain injury, systemic autoimmune disease, or unstable physical diseases were recruited from the community). The Chinese version of the Schedule for Affective Disorders and Schizophrenia-Lifetime (SADS-L), which has high reliability and validity, was used to collect data on psychiatric conditions in all participants and to confirm that patients met DSM-5 diagnostic criteria (Chen et al., 2017).

A trained interviewer obtained information on age, sex, family income, age at symptom onset, psychotic features, number of depressive episodes, years of education, alcohol use, and smoking by using the SADS-L. Comorbidities were evaluated using the SADS-L. In addition, depressive and anxiety symptoms were assessed using the Beck Depression Inventory-II (BDI-II) and Beck Anxiety Inventory (BAI), respectively. This study has been approved by the Institutional Review Board of the Taiwan National Health Research Institutes (NHRI), with the project number NHRI-EX106-10627NI. All data have been verified to have obtained informed consent from the participants through the Institutional Review Board.

Outcome Assessment (Lifetime Suicidal Ideation, Suicide Planning, and Attempt)

In the study population, lifetime suicidal behaviors were measured from 3 aspects: suicidal ideation, suicide planning, and suicide attempts. We assessed lifetime suicidal behaviors by using specific questions from the SADS-L. Suicide ideation was measured using the following question from the SADS-L: “In the past, did you ever seriously think about attempting suicide?” (with a response of yes or no). Suicide planning and suicide attempts were measured using the following question: “In the past, did you make a plan for attempting suicide?” (with a response of yes or no) and “In the past, did you attempt suicide?” (with a response of yes or no), respectively, followed by more detailed questions regarding means and severity. We divided patients on the basis of the history of suicidal behaviors as a dichotomous variable (ie, whether they had suicide ideation, suicide planning, and suicide attempts). Individuals who reported having specific suicidal behaviors or (eg, attempted suicide) were considered as attempters, and those without these behaviors were considered non-attempters.

Measurement of NfL

In patients with MDD and healthy controls, plasma concentration of NfL was measured by quantitative horseradish peroxidase enzyme-linked immunosorbent assay (ELISA) kit (OKCD01380; Aviva Systems Biology). This assay was based on a standard sandwich ELISA method. Antibody specific for NfL was pre-coated on a 96-well plate, then 100 μL of standard or diluted samples (4-fold dilution) were added into different wells and incubated for 1 hour at 37 °C. After removal of the standards and samples, a biotinylated detector antibody specific for NfL was added into each well and incubated at 37 °C for 1 hour. Following 3 times of buffer washing, avidin peroxidase conjugate was then added and incubated at 37 °C for 30 minutes. The unbound conjugate was washed away using a wash buffer for 5 times. An enzymatic reaction was produced through the addition of 3,3′,5,5′ tetramethylbenzidine substrate, which was catalyzed by horseradish peroxidase enzyme and generated a blue color product that changed to yellow after adding 50 μL of acidic stop solution. The NfL concentration was calculated based on a standard curve, which was linearized by plotting the log of the human NfL concentrations between 1.56 and 100 pg/mL vs the log of the optical density, and the best fit line was obtained from regression analysis.

Statistical Analysis

Statistical analyses were conducted in R (version 4.1.2, R Foundation for Statistical Computing). Differences with P < .05 were considered statistically significant. Numerical data are summarized as the mean ± standard deviation, and categorical data are presented as percentages determined using the χ² test. The independent sample t test and χ² test were applied to analyze differences in demographic and clinical data in the case and control groups.

To evaluate the effects of NfL levels on suicidal behaviors, we conducted multivariate logistic regressions, using suicidal behaviors as the response variable and NfL as the predictor variable. The regressions included the following relevant covariates: demographic characteristics (age and gender) and lifestyle factors (alcohol use and smoking). We used stepwise regression to select and evaluate the best-fitting models for 3 suicide outcomes. For the outcomes of suicide behavior and suicide attempts, the optimal model included the variables NfL, age, sex, and smoking, while excluding alcohol. For suicide ideation, the preferred model included only the variable smoking. To facilitate comparison and ensure consistency across the 3 suicide outcomes, we elected to use a final regression model that included NfL, age, sex, and smoking to report association findings. Using the best-fitting model, we additionally conducted logistic regression analyses including BDI-II and BAI as variables. This allowed us to evaluate whether current depression and anxiety symptoms might influence the association between suicide outcomes and NfL.

RESULTS

Comparison of Characteristics Between Patients With MDD and Healthy Controls

Overall, 73 patients with MDD and 40 age- and sex-matched controls were enrolled. Table 1 presents the characteristics of the cases and controls. The majority of the patients with MDD and controls were women. No significant differences were noted in demographic variables, namely age, sex, and body mass index (BMI). A significantly higher proportion of the patients with MDD than did the controls engaged in smoking and alcohol use. Regarding lifetime suicidal behaviors, a significantly higher proportion of the patients with MDD had suicide ideation, suicide planning, and suicide attempts. More than half of the patients with MDD reported suicide ideation (84.93%), 67.12% reported suicide planning, and 43.80% reported suicide attempts. Among the healthy controls, 25% reported suicide ideation, only 2.5% reported suicide planning, and none of them reported suicide attempts.

NfL Levels and Lifetime Suicidal Behavior

The patients with MDD had a higher mean NfL levels (17.60 pg/mL) than did the controls (14.82 pg/mL), but the difference did not reach statistical significance (Table 1). Table 2 presents the multivariate logistic regression results for the association between NfL and lifetime suicidal behaviors. In the adjusted model, the NfL levels were not associated with suicide ideation (odds ratio [OR] = 1.32 with every 10 pg/mL increase in the NfL levels, P = .260, pseudo R² = 0.097). The NfL levels were significantly associated with suicide planning (OR = 1.64 with every 10 pg/mL increase in the NfL levels, P = .040, pseudo R² = 0.080) and suicide attempts (OR = 1.94 with every 10 pg/mL increase in the NfL levels, P = .014, pseudo R² = 0.136). Higher OR estimates were observed for more severe suicidal behaviors. The results revealed that increased NfL levels were associated with higher risks of lifetime suicide planning and suicide attempts. Similar results were observed in the initial models that included only demographic characteristics age and gender and lifestyle factors (alcohol use and smoking) (Table S1A and B). However, if we restricted the analysis to the patients with MDD only, the similar effect size for suicide attempts did not reach statistical significance either (OR = 1.06 with every 10 pg/mL increase in the NfL levels, P = .062, pseudo R² = 0.102; Table S2).

We included BDI-II and BAI one at a time in the multivariable logistic regression model. For suicide ideation and suicide plan, adjusting for BDI-II/BAI reduced the OR of NfL, resulting in a nonsignificant association between NfL levels and these suicidal behaviors (Table S3A and B). This suggests that current depression or anxiety symptoms have a substantial effect on the risk of suicide ideation and plan. Conversely, adjusting for BDI-II/BAI did not change the OR and P-values of NfL for suicide attempts (Table S3C), indicating that the association between NfL and suicide attempts is not influenced by current depression or anxiety symptoms.

DISCUSSION

In the present study, we observed higher serum NfL levels in the patients with MDD than in the controls, but the difference did not reach statistical significance. Second, the findings of multivariate logistic regression revealed that the NfL levels were significantly associated with higher odds of lifetime suicide planning and suicide attempts.

The difference in the NfL level between the patients with depression and healthy controls did not reach statistical significance. Previous studies have reported inconsistent findings regarding the NfL level in patients with mood disorders. A recent study observed significantly higher NfL levels in patients with MDD than in healthy controls (Chen et al., 2022), while another study found no significant differences in NfL levels among patients with stress-induced exhaustion disorder, MDD, and healthy controls (Wallensten et al., 2022). The possible explanation is that NfL is a quantifiable indicator of neuronal damage, and its sensitivity may be affected by many clinical variables, such as age, smoking, cardiovascular disease, and head trauma (Barro et al., 2020). These factors might be associated with increased NfL levels. By contrast, increasing BMI may lead to increased blood volume, causing decreases in serum NfL levels. The differences in clinical variables among these 3 studies might have resulted in inconsistent findings across these studies (Chenet al., 2022; Wallensten et al., 2022). Furthermore, the different treatment effects may explain the inconsistent findings regarding NfL levels between patients with MDD and controls among the 3 studies. In patients with multiple sclerosis, treatment with disease-modifying drugs reduced the NfL levels (Piehl et al., 2018; Delcoigne et al., 2020). A possible explanation is that treatment of antidepressants may reduce the NfL levels, leading to nonsignificant differences in the NfL levels between patients with MDD and healthy controls, which requires further studies to provide empirical evidence for this possibility. A case–control study comparing the NfL levels between medication-naïve patients with MDD and healthy controls revealed significantly higher NfL levels in patients with MDD, supporting that the treatment of depression may affect the NfL levels in patients with MDD. Another possible reason is the varying age distribution in the study samples. The average age of patients varied among the 3 studies of MDD. In the MDD case–control study conducted by Chen et al, the average age was 20 to 30 years (Chen et al., 2022), and the average age was 40 years in our study and another study (Wallensten et al., 2022). Although patients with MDD in the 3 studies were compared with age-matched controls, the effect of age should have been corrected originally. However, one study reported higher variability in the NfL levels in the older age group than in the younger group (Khalil et al., 2020), and with age, the relative contribution of the primary disease to the overall NfL levels may be concealed by silent damage from coexisting comorbidities (Barro et al., 2020). Therefore, although the patients of the 3 studies were all matched by age, the different age distribution of the study sample may lead to increased variability in the NfL levels, thus reducing the effect of primary diseases on NfL.

Our study found a significant association between blood NfL levels and an increased likelihood of lifetime suicidal behavior, including both suicide planning and suicide attempts. This suggests that elevated serum NfL levels may indicate a greater severity of lifetime suicidal behavior. We observed a trend of increased ORs for suicide planning (1.64) and suicide attempts (1.94), as shown in Table 2. Our findings align with those of Ramezani et al. (2022), who also reported significantly higher NfL levels in individuals with suicide attempts compared to those without. Moreover, the NfL level may serve as a pertinent trait marker for lifetime suicidality. In this study, we also assessed current suicidal behaviors, further inquiring only if there was a history of past suicidal behavior. Consequently, focusing solely on current suicidal symptoms would reduce the overall sample size and statistical power of the study. Among MDD cases, 46 had current suicidal ideation, 34 had suicidal plans, and 15 had attempted suicide. The healthy control group reported no current suicidal symptoms. With this limited sample size, we conducted regression analyses on current suicidal behaviors and found similar trend but no significant association between NfL levels and current suicidal behaviors (OR range: 1.20-1.64, P-value range:.09-.44). It is possible that NfL levels require long-term environmental stimuli and accumulation, and current suicidal behaviors may not reflect immediate changes in NfL levels.

On the other hand, a study focusing on late-life depression found that higher NfL levels were associated with more depressive symptoms and an elevated risk of incident depressive events over time (Schuurmans et al., 2024). Additionally, a study in the general population reported that increased serum NfL level was linked to a higher risk of depressive symptoms, even after adjusting for confounding factors (Zhang et al., 2025). Collectively, these studies suggest that NfL concentrations may be indicative of the severity of specific depressive symptoms or a higher burden of depressive symptoms. Similarly, our study shows that the association between NfL and suicide becomes nonsignificant for suicide ideation and suicide planning after accounting for current depression and anxiety symptoms (Table S3A and B). Only the association with suicide attempts remains unaffected by these symptoms. This finding further supports that NfL concentration is related to more severe suicidal behaviors.

A previous study reported that individuals who attempt suicide had more white matter hyperintensity (WMH) compared with non-suicidal individuals(van Heeringen and Mann, 2014). A meta-analysis reported that patients with unipolar depression who were suicide attempters had 1.9 times more WMH than those who were non-attempters (Grangeon et al., 2010). One neuropathological mechanism of WMH is the demyelination and degeneration of white myelinated axons (Wardlaw et al., 2015). Studies have determined a strong correlation between NfL and WMH. First, NfL in the CSF was identified as a biomarker of changes in the white matter (Osborn et al., 2018). Second, the plasma NfL levels may reflect the integrity of WM, and the blood NfL levels may predict WM damage in the brain (Walsh et al., 2021). These studies suggest a possible association between suicidal behavior and alterations in brain structure or neural damage.

Several mechanisms may explain the link between suicidal behavior, neuronal loss, and neurodegeneration. One mechanism involves inflammation and kynurenine pathway dysregulation. Suicidal patients display elevated markers of inflammation in the central nervous system and peripheral tissues (Courtet et al., 2016), where inflammatory mediators elevate neurotoxic quinolinic acid, contributing to neural dysfunction and degeneration. Another mechanism is the low level of brain-derived neurotrophic factor in suicidal patients (Eisen et al., 2015), which is crucial for neuronal survival; its deficiency leads to cell death and neurodegeneration, particularly in conditions like Parkinson’s and Alzheimer’s diseases (Bathina and Das, 2015). A third mechanism involves the hypothalamic-pituitary-adrenal axis, where hyperactivity and elevated cortisol levels, observed in suicide attempters (Mann, 2003), cause hippocampal neuron apoptosis (Lupien et al., 2018). These mechanisms suggest that the neurobiology of suicide may be linked to axonal injury, leading to increased NfL levels. It is noteworthy that the aforementioned potential mechanisms are largely inferred from other studies and further research is needed to validate these findings.

This study has some limitations that should be addressed. First, the study design was cross-sectional; thus, we could not investigate changes in the NfL levels over time, and the directionality of the observed association could not be inferred, warranting longitudinal studies to elucidate the causal inference. Second, the small sample size indicates that the results should be interpreted with caution. A prospective study enrolling a larger sample is necessary to elucidate the association between NfL and suicide. Third, the association between NfL and suicide was only observed for lifetime suicide but not for current suicidal behavior. This may be due to the limited sample size, suggesting that future studies should include more cases with current suicidal behavior to further explore the relationship between NfL and current suicidal behavior. Fourth, because our study mainly focused on patients with MDD, whether our results can be generalized to patients with other psychiatric diagnosis should be further investigated. Fifth, peripheral NfL levels do not represent total NfL levels in the central nervous system.

CONCLUSION

In summary, we determined that the serum NfL levels were significantly associated with higher odds of lifetime suicide planning and suicide attempts; NfL levels may serve as a relevant trait marker for lifetime suicidality. Although a single biomarker is not sufficient for identifying and assessing suicide risk, our finding regarding the association between serum NfL levels and suicide in our study may provide a crucial direction for future studies. Future studies should integrate biological and clinical markers for the identification of individuals at high suicide risk and thereby identify targets for intervention.

Acknowledgments

This manuscript was edited by Wallace Academic Editing. We thank Chiao-Erh Chang for their valuable contributions to data management.

Additionally, this study was conducted with the support of China Medical University Taiwan Hsinchu Hospital (CMUHCH-DMR-114-017).

Author contributions

Po-Hsiu Kuo (Conceptualization [equal], Data curation [equal], Funding acquisition [equal], Investigation [equal], Project administration [equal], Resources [equal], Supervision [equal]), Ying-Chih Cheng (Data curation [equal], Formal analysis [equal], Investigation [equal], Software [equal], Writing—original draft [equal]), Yu-Li Liu (Data curation [supporting], Formal analysis [equal]), Wen-Yin Chen (Data curation [equal], Methodology [equal], Project administration [equal], Resources [equal]), Chih-Chiang Chiu (Resources [equal]), and Ming-Chyi Huang (Supervision [supporting], Writing—review & editing [supporting])

Funding

This work was supported by the Ministry of Science and Technology (MOST 108-2314-B-002-136-MY3), National Health Research Institutes Project (NHRI-EX108-10627NI), National Taiwan University Career Development Project (110L7860), and China Medical University Taiwan Hsinchu Hospital (CMUHCH-DMR-114-017). The funders played no role in the study design; collection, analysis, and interpretation of data; manuscript writing; or the decision to submit the paper for publication. The authors declare that they have no competing interests.

Conflicts of interest

None declared.

Data availability

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

References