This study used the same cohort of patients in our previous study (11). Briefly, inpatients with AIS from the Department of Neurology, Sichuan Provincial People's Hospital and Ya'an People's Hospital during May 1, 2016 and Dec 31, 2019, were screened for eligibility for this study. Patients were excluded if they have one of the following conditions: (1) Have previously diagnosed depression, anxiety, and insomnia before AIS onset; (2) have other psychological disorders, such as schizophrenia and bipolar disorders et al. before AIS onset; (3) Cannot complete psychological tests due to hearing, language, or communicating disabilities; (4) Other severe neurological diseases which may affect circulating NfL levels, such as Parkinson's disease, Alzheimer's disease, and traumatic brain injury; (5) Refused to participate in this study. Written informed consent for participation was obtained from patients or their legal relatives. This study conformed with the principles of the Declaration of Helsinki and was approved by the investigational review board of the Sichuan Provincial People's Hospital.

The demographic information, including age, sex, education level, body mass index (BMI), smoking history, medical history, including oral anticoagulion or antiplatelet drug use, comorbidities such as hypertension, diabetes mellitus, hypercholesteremia, and arial fibrillation were collected from the medical records. AIS was diagnosed according to the World Health Organization (WHO) Multinational Monitoring of Trends and Determinants in Cardiovascular Disease (WHO-MONICA) criteria and was verified by magnetic resonance imaging (MRI) performed within 24 h since symptom onset. The neurological deficits of patients were examined with the National Institutes of Health Stroke Scale (NIHSS) upon admission (12), performed by a certified stroke neurologist. AIS subtype was determined with the TOAST criteria.

We focused on symptoms of depression, anxiety, and insomnia for all participants 6 months after AIS onset, using Chinese versions of validated measurement tools. Accordingly, the 9-item Patient Health Questionnaire (PHQ-9; range, 0–27) (13), the 7-item Generalized Anxiety Disorder (GAD-7) scale (range, 0–21) (14), the 7-item Insomnia Severity Index (ISI; range, 0–28) (15), were used to assess the severity of symptoms of depression, anxiety, and insomnia, respectively. The total scores of these measurement tools were interpreted as follows: PHQ-9, normal (0–4), mild (5–9), moderate (10–14), and severe (15–27) depression; GAD-7, normal (0–4), mild (5–9), moderate (10–14), and severe (15–21) anxiety; ISI, normal (0–7), subthreshold (8–14), moderate (15–21), and severe (22–28) insomnia. These categories were based on cutoff values established in the literature (13–15). The cutoff value for detecting symptoms of major depression, anxiety, and insomnia distress were 10, 7, and 15, respectively. Participants with scores above the cutoff threshold were characterized as having major depression, anxiety and insomnia, respectively.

Blood was sampled within 2 h since admission, and serum was separated within 30 min after sampling and stored at −80°C until further analysis. Serum NfL was determined using the single-molecule (Simoa) array according to manufacturer's instructions (16). Each test was done in duplicates and the means of each test were used for statistical analysis. Monoclonal antibodies and purified bovine NFL were used as calibrators.

Continuous variables were tested for normality, and if they were normally distributed, an independent t-test was used, but if they were not normally distributed, a Mann-Whitney U-test was used. For categorical data, two-sample tests of proportions were used to compare proportions. Logistical regression models were utilized to investigate the association between serum NfL concentrations at baseline and mental health outcomes. We first fitted univariate models with a single candidate variable at one time. The potential risk factors as determined by a p-value < 0.2 were included in the final multivariate regression models. The receiver operating characteristic (ROC) curve analysis were utilized to test the predictive effects of baseline serum NfL on mental health outcomes at follow-up. Optimal sensitivity and specificity were determined via a non-parametric approach. The Youden index was calculated for the cutoff value to determine the cutoff value that maximized the discriminating power of the test. Statistical analyses were conducted using SPSS statistical package version 24 (IBM SPSS Statistics for Windows, Armonk, NY, USA) and a p-value < 0.05 was regarded as statistically significant.

We categorized the patients into two groups according to serum NfL concentrations, the high NfL group and the low NfL group. There was no significant difference in the mean age, median BMI, the frequency of family history of stroke, frequencies of antiplatelet drug use and anticoagulation drug use between the high NfL and low NfL group. Frequencies of hypertension, diabetes mellitus, hypercholesteremia and arial fibrillation between the high NfL and low NfL group were also not significantly different between groups. No significant difference was not observed in the distribution of infarction region and stroke etiology between the high and low NfL group. No significant difference was also not observed in the incidences of hemorrhagic transformation and recurrent AIS during follow-up between groups. The high NfL group had significantly higher incidences of PSD, PSA, and PSI than the low NfL group during follow-up (Table 1).

High NfL group had significantly higher PHQ-9, GAD-7, and ISI scores than low NfL group (Figure 1). Serum NfL concentrations were positively associated with PHQ-9, GAD-7, and ISI scores (Figure 2). Furthermore, serum NfL predicted major depression with a sensitivity of 70.27%, a specificity of 67.79% and an AUC of 0.694 (Figure 3A). Serum NfL predicted anxiety with a sensitivity of 69.23%, a specificity of 64.02% and an AUC of 0.683 (Figure 3B). Serum NfL predicted insomnia with a sensitivity of 75.00%, a specificity of 66.43% and an AUC of 0.723 (Figure 3C).

We utilized three logistic regression models to investigate risk factors of major depression, anxiety and insomnia. In univariate analyses, family history of stroke [ORs (95% CI): 3.053 (1.021, 9.125)], DWI hyperintensity volume [ORs (95% CI): 1.037 (0.998, 1.077)], cerebral lobe infarction [ORs (95% CI): 1.984 (0.894, 4.402)], and high NfL level [ORs (95% CI): 4.244 (1.871, 9.625)] were found to be potential risk factors of major depression. Male sex was found to be a protective factor against major depression [ORs (95% CI): 0.526 (0.262, 1.053)]. However, only high NfL level remained to be a significant risk factor of major depression [ORs (95% CI): 4.427 (1.918, 10.217)] in the final multivariate model (Table 2).

In univariate analyses, family history of stroke [ORs (95% CI): 2.502 (0.929, 6.736)], thalamus infarction [ORs (95% CI): 2.762 (0.847, 9.008)], and high NfL level [ORs (95% CI): 3.665 (1.993, 6.742)] were found to be potential risk factors of anxiety. Male sex was found to be a protective factor against anxiety [ORs (95% CI): 0.522 (0.300, 0.908)]. However, only high NfL level [ORs (95% CI): 3.063 (1.939, 6.692)] remained to be a significant risk factor of anxiety and male sex [ORs (95% CI): 0.514 (0.288, 0.917)] remained to be a protective factor against that in the final multivariate model (Supplementary Table 1).

In the analysis of risk factors of insomnia, only high NfL level [ORs (95% CI): 4.200 (1.526, 11.562)] was found to be a risk factor in both univariate and multivariate analyses. Collectively, these findings indicate that high serum NfL level might be a potential risk factor of major depression, anxiety, and insomnia (Supplementary Table 2).