Abstract
Objective
This study examined serum brain injury markers and their associations with disease features, cytokines associated with microglial activation in lupus and cognitive dysfunction (CD) in adolescents with childhood-onset SLE (cSLE).
Methods
We used cross-sectional data from cSLE patients (aged 12–17 years) and age-matched, sex-matched healthy controls. Serum levels of brain injury markers (serum neurofilament light, glial fibrillar acidic protein (GFAP), Tau), interferon (IFN)-α, IFN-γ and interleukin-6 (IL-6) were quantified using Simoa assays. cSLE features included disease activity (Systemic Lupus Erythematosus Disease Activity Index 2000), damage (Systemic Lupus International Collaborating Clinics damage index) and glucocorticoid (GC) exposure. A neurocognitive battery assessed executive function, attention and working memory, and CD was determined using standardised scores. We compared brain injury marker levels between cSLE and controls, and those with and without CD using Wilcoxon rank-sum tests. We calculated correlations between injury markers, disease features and cytokines and examined differences in disease features between those with and without high-level brain injury markers (>90th percentile) (using Bonferroni correction).
Results
Participants included 56 cSLE patients (median disease duration=10.6 months (IQR 2.0–14.1), one with neuropsychiatric lupus) and 43 controls. Levels were higher in cSLE versus controls for GFAP (z=−3.97, p<0.001), Tau (z=−2.10, p=0.035), IFN-α (z=−4.80, p<0.001), IFN-γ (z=−2.42, p=0.015) and IL-6 (−3.09, p=0.002). Severe CD (≥2 SD from standardised mean) was present in 31% cSLE versus 9% controls (chi2=6.69, p=0.01), associated with higher Tau levels for cSLE (z=−3.94, p<0.001). High-level brain injury markers were observed in 13 (23%) cSLE patients associated with higher SLEDAI-2K, IL-6 levels and current GC dose.
Conclusion
Brain injury marker levels were high and associated with disease activity and CD in this cSLE adolescent cohort, suggesting a link between systemic inflammation and clinically under-detected neuronal/glial injury. Larger, longitudinal studies should explore the potential clinical utility of brain injury markers for clinical assessment of brain involvement in cSLE.
Keywords: Lupus Erythematosus, Systemic; Autoimmune Diseases; Cytokines
WHAT IS ALREADY KNOWN ON THIS TOPIC
Cognitive dysfunction (CD) affects approximately one third of children with childhood-onset SLE (cSLE) during a critical period of neurodevelopment, yet attribution to neuropsychiatric SLE remains a clinical challenge due to the lack of diagnostic tools.
Brain injury markers such as serum neurofilament light (sNFL), glial fibrillar acidic protein (GFAP) and Tau have been used as diagnostic and monitoring biomarkers in neurologic and neuroinflammatory conditions; however, they have been understudied in cSLE.
WHAT THIS STUDY ADDS
The results of this study show elevated serum levels of GFAP and Tau in children with cSLE compared with healthy controls, with 23% of the cSLE group showing severe CD despite the vast majority without a clinical neuropsychiatric lupus diagnosis.
Tau levels were associated with severe CD in children with cSLE, and those with high-level markers (across sNFL, GFAP and Tau) had higher disease activity, interleukin-6 levels and current glucocorticoid doses.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
These findings indicate that brain injury is occurring in children with cSLE, even in the absence of neuropsychiatric lupus diagnosis, in association with elevated serum cytokines associated with microglial activation, providing mechanistic insight that may guide future development of clinical therapies.
These results suggest the potential utility of brain injury markers as diagnostic and monitoring biomarkers of CD due to cSLE and highlight the need for further study in larger, longitudinal cohorts.
Introduction
SLE is a chronic autoimmune disease with up to 20% of patients diagnosed during childhood. Childhood-onset SLE (cSLE) has typical onset in adolescence with high morbidity and mortality, affecting multiple organs including the nervous system, and a more aggressive course compared with those with adult-onset disease.1 Neuropsychiatric SLE (NPSLE) manifestations affecting the brain are common, affecting approximately 30% of patients with cSLE,2 often striking during a critical period of adolescent brain development.3 Among those with NPSLE manifestations, cognitive dysfunction (CD) affects nearly a third, with significant impacts on education, psychosocial function and overall quality of life.2,7 CD is defined by the American College of Rheumatology NPSLE nomenclature as a deficit detected in any one of eight neurocognitive domains, with executive function, attention, working memory and psychomotor speed among the most impacted domains in cSLE.4 5 Yet the biological changes that underlie CD in cSLE are poorly characterised, resulting in challenges for diagnosis and treatment.3,58
Recent insights have linked neuroinflammation with dysregulated signalling of systemic immune cytokines such as type I interferon alpha (IFN-α), type II interferon gamma (IFN- γ) and interleukin-6 (IL-6), which can influence central nervous system function through disruption of the blood-brain and blood-cerebrospinal fluid (CSF) barriers, activation of microglia in the central nervous system (CNS), neuronal damage and altered synaptic plasticity, leading to CD.8,10 Specifically in cSLE, studies suggest that a chronic systemic inflammatory state mediated by IFNs may exert direct or indirect disturbances to the developing brain in cSLE.11 12 Brain tissue inflammation and damage result in the local release of brain injury markers that can be subsequently detected in the bloodstream.8,10 These markers include serum neurofilament light (sNFL), glial fibrillar acidic protein (GFAP) and Tau, structural proteins associated with neuronal and glial cell damage.13 sNFL is a component of the axonal skeleton in neurons and is released in the setting of axonal injury. GFAP is an intermediate filament protein almost exclusively expressed by astrocytes and is therefore a marker of astrocyte injury. Tau is a microtubule-associated protein and its release from neurons is a marker of neuronal injury and cell death.
Serum levels of brain injury markers, including sNFL, GFAP and Tau, have been studied as potential biomarkers of CD in traumatic brain injury, neurodegenerative diseases like Alzheimer’s and neuroinflammatory diseases such as multiple sclerosis.13,16 While less is known about brain injury markers in SLE, recent studies suggest potential utility for the diagnosis and monitoring of neuropsychiatric syndromes, mostly focused on sNFL. In adults with NPSLE, one study found significantly elevated levels of sNFL in patients with NPSLE and focal CNS involvement compared with those without NPSLE,17 and another study found an increase of sNFL and GFAP levels during the active NPSLE phase with decline following immunotherapy.18 sNFL levels have been found to correlate with CSF concentrations19 and intrathecal sNFL has been associated with IL-620 21 in adults with SLE. Furthermore, higher sNFL levels were associated with higher SLE disease activity alongside higher IFN-α levels in association with CD in a cohort of adults with SLE in the absence of overt neuropsychiatric symptoms.20 This growing literature supports the potential utility of brain injury markers, even beyond sNFL, for the assessment of neuroinflammation and CD in cSLE, but this is yet to be explored.
In this study, we aimed to examine: (1) differences in serum levels of three brain injury markers (sNFL, GFAP and Tau) between cSLE patients and controls, (2) relationships between brain injury markers and CD in cSLE and controls and (3) relationships between brain injury markers, serum cytokine levels (IFN-α, IFN-γ and IL-6) and cSLE disease features.
Participants and methods
Study design, participants and setting
We used cross-sectional data from an ongoing prospective longitudinal study of brain health in adolescents with SLE (aged 11–17 years) meeting the Systemic Lupus International Collaborating Clinics (SLICC) or European Alliance of Associations for Rheumatology (EULAR)/American College of Rheumatology (ACR) SLE classification criteria22 23 recruited from the lupus clinic at a Canadian tertiary children’s hospital from January 2020 to December 2023. Age-matched and sex-matched healthy controls who met the inclusion criteria were recruited through study advertisements using formal hospital communication channels. cSLE participants were excluded if they had conditions significantly affecting the ability to complete cognitive tests (eg, severe neurodevelopmental or intellectual disorders, hearing loss, vision problems), were on psychotropic medications, had alcohol or drug use within 24 hours of assessment or had a history of significant head trauma. For controls, the same exclusion criteria were applied, in addition to a known diagnosis of chronic medical conditions and use of steroids. Additionally, all participants had to be fluent in English to complete the study measures, which included neuropsychological assessment and a psychiatric interview. The study was approved by The Hospital for Sick Children Research Ethics Board (REB# 1000071306, REB# 1000063027, REB# 1000080072), and all participants and their parents provided written informed consent and/or assent as appropriate.
Demographic variables
Demographic characteristics including age at study visit, biological sex and ethnicity were obtained through questionnaires completed by patients and controls. Self-reported ethnicity, where participants chose from a fixed set of categories, was based on the census categories for country of family origin used by Statistics Canada.24 We categorised ethnicity into five mutually exclusive groups: Asian, black or African, European, Latin America and other (includes Arab/Middle Eastern, Indigenous, Pacific Islander and multiethnic).
Disease variables
Disease characteristics including disease duration (ie, time from diagnosis date to study visit), disease activity (Systemic Lupus Erythematosus25 Disease Activity Index 2000 (SLEDAI-2K)), damage (SLICC/ACR Damage Index (SDI)),25 26 presence of major organ disease (nephritis and/or NPSLE diagnosed by the treating rheumatologist), hypocomplementaemia (low C3/C4 according to clinical lab cut-offs), positive anti-phospholipid antibody status (positive anti-cardiolipin, anti-beta 2 glycoprotein or lupus anticoagulant according to clinical lab cut-offs), presence of abnormal clinical conventional brain MRI (determined by neuroradiologist reading), immunosuppressive medication and glucocorticoid (GC) (prednisone-equivalent dose) use were obtained through electronic medical records. SLEDAI-2K at study visit was calculated as both a continuous variable and binary variable for active disease (defined as a SLEDAI-2K score >4). We reviewed disease activity scores between the diagnosis date and the study visit. To account for varying time intervals between clinic assessments, we calculated the adjusted mean SLEDAI-2K (AMS).27 This was determined by calculating the area under the SLEDAI-2K curve (ie, the length of time between two SLEDAI-2K measurements multiplied by the average of the two measurements) over time by adding the area of each of the blocks of measurement interval and then dividing by the length of time for the whole period.27 For those with only one visit, the AMS was equal to the SLEDAI-2K value for that visit, per the published definition.27 We defined the presence of disease damage as an SDI score of >0 determined at the last clinic visit. GC use was defined as oral and/or intravenous prednisone equivalent dose in mg. Current and cumulative GC dose (expressed in g is a summation of all oral and/or intravenous GCs ever received) was calculated.
Serum brain injury markers and cytokines
Participant blood samples were collected with routine clinical blood draws for patients and during a study visit for controls. To maintain sample integrity, blood samples were stored in a −80 °C freezer after collection and processing and were shipped in batches on dry ice. Serum brain injury marker levels were quantified using ultra-sensitive Simoa assays to detect serum levels of sNFL, GFAP and Tau (Human Neurology 4–Plex B, Quanterix, Billerca, Massachusetts, USA) and IFN-α, IFN-γ and IL-6 (Quanterix, Billerca, Massachusetts, USA), with values expressed in pg/mL.
Cognitive function measures
Cognitive function was measured for the domains of executive function, attention and working memory. Executive function was measured by the Delis-Kaplan Executive Function System Colour Word Interference Test. The test is comprised of four conditions (colour naming, word reading, inhibition and inhibition/switching). We included scores for inhibition (ability to inhibit an automatic verbal response) and inhibition/switching (ability to switch between different task demands), measuring cognitive inhibition and mental flexibility, respectively.28 Performance is measured as completion time in seconds for each condition. Raw scores were converted to age-normed scaled scores (mean=10, SD=3), with lower scores indicating worse executive function. Attention was measured by the computerised Conners Continuous Performance Test third edition.29 Participants were presented with letters on a screen one at a time at varying intervals and instructed to press the space bar when each letter appeared, except for the letter X. Nine age-normed T-scores (mean=50, SD=10) assessed different aspects of attention, with T-scores >60 indicating poorer performance. We included T-scores for omissions (sustained attention), commissions (impulsivity) and hit reaction time (response speed). Working memory was measured by the Wechsler Intelligence Scale for Children fifth Edition/Wechsler Adult Intelligence Scale, fourth Edition (for participants aged 16 years and older) Digit Span Test.30 31 The task required participants to repeat strings of digits across three conditions: forwards, backwards and in sequential order. Raw scores were converted to age-normed scaled scores (mean=10, SD=3), with higher scores indicating better performance.
While there is not a standardly accepted definition for CD in cSLE, we used definitions in keeping with previous studies in this population which enabled capture of the range of cognitive difficulties across domains.3 We defined overall CD as (1) performance on any of the cognitive domain tests of ≥2 SD poorer than the mean standardised score or (2) performance on at least two tests (from different cognitive domains) of at least one SD poorer than the mean standardised score. We defined severe CD as performance on any of the cognitive domain tests of ≥2 SD poorer than the mean standardised score. Definitions for overall CD and severe CD were therefore not mutually exclusive. We also defined domain-specific CD as performance at least one SD poorer than the mean standardised score.
Statistical analysis
Descriptive statistics were used to summarise participants’ demographic and disease characteristics. We compared serum brain injury marker and cytokine levels between the cSLE and control groups using the Wilcoxon rank-sum test and analysed group differences in proportion of participants with CD using a χ2 test. P values of <0.05 were considered statistically significant.
We conducted three exploratory analyses. First, we examined differences in brain injury marker levels between those with and without overall CD using Wilcoxon rank-sum tests. Second, we examined relationships between brain injury markers, cSLE disease features and cytokines using Spearman correlations. Third, we examined differences in cSLE features between those with and without high levels of brain injury markers (any of the three marker levels falling above the 90th percentile). In these analyses, we accounted for multiple comparisons with Bonferroni correction.
There was no missing data. One cSLE patient with low IQ <80 was omitted from the CD analyses due to previously undiagnosed developmental delay. All analyses were performed using Stata/BE V.18.
Results
Participant demographic and disease characteristics
Participants included 56 children with cSLE (86% female, mean age=15.1 years ± SD 1.8) and 43 healthy controls (81% female, mean age=15.1 years ± SD 1.7). cSLE patients had a median disease duration of 10.6 months (IQR 2.0–14.1); 15 (27%) had active disease. Disease damage was present in five patients (9%), comprised of six items in renal (n=3), neuropsychiatric (n=1, seizures), skin (n=1, extensive scarring/panniculum) and ocular (n=1, cataracts) domains. Considering major organ involvement, 21 (38%) had lupus nephritis and one patient had an NPSLE diagnosis. Conventional clinical brain MRI abnormalities were present in 10 (18%) of cSLE patients; these included T1 and T2 hyperintensities, T1 hypointensities and other abnormalities (eg, low-lying and/or rounded cerebellar tonsils, empty sella/flattened pituitary gland, cystic changes in pineal gland), none of which were attributed to cSLE. 23 (41%) were taking GCs, and the median cumulative GC dose was 2.9 grams (IQR 0.6–6.9). Additional participant characteristics are shown in table 1.
Table 1. Participant characteristics.
| cSLE group (n= 56) |
Control group (n= 43) |
|
|---|---|---|
| Demographic characteristics | ||
| Age at study visit, years | 16.0 (13.5–16.5) | 16.0 (14.0–17.0) |
| Age at SLE diagnosis, years | 14 (12–15) | – |
| Female sex | 48 (86) | 35 (81) |
| Ethnicity | ||
| Asian | 27 (48) | 7 (16) |
| European | 15 (27) | 24 (56) |
| Black or African | 9 (16) | 4 (9) |
| Latin American | 2 (4) | 3 (7) |
| Other* | 3 (4) | 5 (12) |
| Disease characteristics | ||
| Disease duration, months | 10.6 (2.0–14.1) | – |
| Disease activity at visit (SLEDAI-2K) | 2.5 (2.0–5.5) | – |
| Active disease at visit (SLEDAI-2K >4) | 15 (27) | – |
| Adjusted mean SLEDAI-2K, mean (SD) | 4.7 (3.8) | - |
| Disease damage (SDI >0) | 5 (9) | – |
| NPSLE diagnosis | 1 (2) | – |
| Lupus nephritis | 21 (38) | – |
| Low C3/C4 | 39 (70) | |
| Anti-phospholipid antibody positive | 12 (21) | – |
| Abnormal conventional brain MRI | 10 (18) | – |
| DMARD use† (current) | 30 (54) | – |
| Hydroxychloroquine use (current) | 56 (100) | – |
| Rituximab exposure (ever) | 10 (18) | – |
| Intravenous cyclophosphamide exposure (ever) | 1 (2) | – |
| Current glucocorticoid use | 23 (41) | – |
| Current prednisone-equivalent dose, mg | 2.1 (0.6–7.0) | |
| Cumulative prednisone-equivalent dose, g | 2.9 (0.6–6.9) | – |
| Intravenous steroid use (ever) | 14 (25) | – |
Continuous variables are listed as median (IQR) and categorical variables are listed as n (%), unless otherwise specified.
Includes Arab/Middle Eastern, Indigenous, Pacific Islander and multiethnic.
DMARD includes mycophenolate analogues or azathioprine or methotrexate.
cSLE, childhood-onset SLE; DMARD, disease-modifying antirheumatic drugs; NSPLE, neuropsychiatric SLE; SDI, Systemic Lupus International Collaborating Clinics/American College of Rheumatology (SLICC/ACR) Damage Index; SLEDAI-2K, Systemic Lupus Erythematosus Disease Activity Index 2000.
Comparison of serum brain injury markers and cytokine levels between cSLE and control groups
Figure 1 depicts group differences for cSLE versus controls showing higher levels of GFAP (median 114 pg/mL, IQR 69–140 vs 74 pg/mL, IQR 55–92, p<0.0001) and Tau (median 3.6 pg/mL, IQR 2.3–4.8 vs controls 2.6 pg/mL, IQR 2–3, p=0.036). sNFL levels did not differ between groups (median 5.1 pg/mL, IQR 4.2–7.8 vs 4.8 pg/mL, IQR 3.5–7.3, p=0.154). Table 2 shows higher levels of cytokines in the cSLE group versus controls, for IFN-α (median 0.28 pg/mL, IQR 0.02–1.23 vs 0.10 pg/mL, IQR 0.06–0.23, p<0.0001), IFN-γ (median 0.07, IQR 0.05–0.10 vs 0.02 pg/mL, IQR 0.01–0.03, p=0.015) and IL-6 (median 1.20, IQR 0.72–2.65 vs 0.67, IQR 0.40–1.60, p=0.0018).
Figure 1. Comparison of brain injury markers between cSLE and controls. Legend: boxplots showing group comparison of brain injury marker levels between patients and controls. P value <0.05 was considered statistically significant (Wilcoxon rank-sum test). cSLE, childhood-onset SLE; GFAP, glial fibrillar acidic protein; sNFL, serum neurofilament light.
Table 2. Comparison of serum-IFN and IL-6 levels between cSLE and controls.
| cSLE (n=56) | Controls (n=43) | Z-statistic, P value | |||
|---|---|---|---|---|---|
| Median (IQR) | Range | Median (IQR) | Range | ||
| IFN-α (pg/mL) | 0.28 (0.02–1.23) | 0–104.0 | 0.10 (0.06–0.23) | 0.03–15.90 | −4.80, <0.0001 |
| IFN-γ (pg/mL) | 0.07 (0.05–0.10) | 0–4.32 | 0.02 (0.01–0.03) | 0.02–14.90 | −2.42, 0.015 |
| IL-6 (pg/mL) | 1.20 (0.72–2.65) | 0.24–173.30 | 0.67 (0.40–1.60) | 0.10–19.40 | −3.09, 0.0018 |
Shown are group comparisons of cytokine levels using the Wilcoxon rank-sum test; p values <0.05 are in bold and were considered statistically significant.
cSLE, childhood-onset SLE; IFN-α, interferon alpha; IFN-γ, interferon gamma; IL-6, interleukin-6.
CD in cSLE versus control participants
Figure 2 depicts proportions of overall CD and domain-specific CD for cSLE and control participants. Overall CD was present in n=22 (38%) cSLE versus n=10 (23%) controls (χ2=3.08, p=0.08), and severe CD was present in n=17 (31%) cSLE versus n=4 (9%) controls (χ2=6.69, p=0.01). For domain-specific measures, CD for attention on the hit-reaction time task was present in n=18 (33%) cSLE versus n=5 (12%) controls (χ2=5.98, p=0.01). There were no significant differences in the presence of CD for the other cognitive domains.
Figure 2. Comparison of brain injury markers between cSLE and controls. Shown are proportions of participants with cognitive impairment, compared between cSLE and control groups using the χ2 test with significant p value <0.05 indicated by *. Cognitive function domains were measured as follows: (1) executive function by the Delis-Kaplan Executive Function System (DKEFS) Colour Word Interference Test; (2) attention by the Conners Continuous Performance Test third edition (CPT-3); (3) working memory by the Wechsler Intelligence Scale for Children fifth Edition (WISC-V)/Wechsler Adult Intelligence Scale, fourth Edition (WAIS-IV) Digit Span Test. Overall CD: (a) performance on any of the cognitive domain tests of ≥2 SD poorer than the mean standardised a score or (b) performance on at least two tests (from different cognitive domains) of at least one SD poorer than the mean standardised score. Severe CD: performance on any of the cognitive domain tests of ≥2 SD poorer than the mean standardised score (definitions for overall CD and severe CD were therefore not mutually exclusive). Domain-specific CD: performance at least one SD poorer than the mean standardised score. CD, cognitive dysfunction; cSLE, childhood-onset SLE.
Relationships between brain injury markers and CD
Table 3 shows comparisons of brain injury marker levels by CD status for cSLE and control participants. Tau levels were higher for cSLE patients with overall CD versus not (median 4.58 pg/mL, IQR 3.44–7.07 vs 2.64 pg/mL, IQR 1.97–3.93, p<0.001) and with severe CD versus not (median 4.58 pg/mL, IQR 3.44–7.07 vs 2.74 pg/mL, IQR 1.97–4.10, p<0.001), both statistically significant for a corrected p value <0.004.
Table 3. Brain injury marker levels by level of cognitive dysfunction for cSLE and controls.
| Moderate-severe CD | Severe CD | |||||
|---|---|---|---|---|---|---|
| Absent, median (IQR) |
Present, median (IQR) |
Z-statistic, P value | Absent, median (IQR) |
Present, median (IQR) |
Z-statistic, P value | |
| cSLE (n=55) | ||||||
| sNFL | 4.92 (3.60–6.53) | 6.05 (4.68–16.3) | −1.91, 0.056 | 4.99 (3.93–7.29) | 6.05 (4.65–8.59) | −1.13, 0.264 |
| GFAP | 90.05 (64.70–127.0) | 131.0 (114.0–216.0) | −2.17, 0.029 | 93.75 (64.70–132.0) | 125.0 (114.0–151.0) | −1.61, 0.109 |
| Tau | 2.64 (1.97–3.93) | 4.58 (3.44–7.07) | −3.60, <0.001* | 2.74 (1.97–4.10) | 4.58 (3.44–7.07) | −3.94, <0.001* |
| Controls (n=43) | ||||||
| sNFL | 4.81 (3.65–7.32) | 3.89 (3.41–5.46) | 0.78, 0.449 | 4.77 (3.58–7.32) | 3.99 (2.67–6.87) | 0.63, 0.556 |
| GFAP | 74.30 (57.80–91.40) | 73.80 (34.70–102.0) | 0.09, 0.944 | 74.30 (56.90–91.60) | 60.25 (33.50–94.90) | 0.50, 0.643 |
| Tau | 2.43 (1.97–3.08) | 3.33 (2.65–5.01) | −2.13, 0.033 | 2.56 (2.01–3.21) | 5.15 (3.58–5.65) | −1.84, 0.068 |
Shown are group comparisons of brain injury marker levels (median, IQR in pg/mL) by cognitive dysfunction group for cSLE and control participants using the Wilcoxon rank-sum test; p values <0.05 are in bold and those with an asterisk are significant after Bonferroni correction (p<0.004).
Moderate-severe CD: performance on any of the cognitive domain tests of 2 SD poorer than the mean standardised score and/or performance on at least two tests (from different domains) of one SD poorer than the mean standardised score. Severe CD: performance on any of the cognitive domain tests of 2 SD poorer than the mean standardised score.
Indicates significance of p<0.004 after Bonferroni correction.
CD, cognitive dysfunction; cSLE, childhood-onset SLE; GFAP, glial fibrillar acidic protein; sNFL, serum neurofilament light.
Relationship between brain injury markers and cSLE disease features
Table 4 shows Spearman correlations between brain injury markers and disease features in the cSLE group. Higher Tau levels correlated with higher SLEDAI-2K at study visit (r=0.40, p=0.002) and AMS over the disease duration (r=0.49, p=0.002), although these did not reach statistical significance for a corrected p value <0.002.
Table 4. Correlations between brain injury markers and disease characteristics in cSLE.
| sNFL | GFAP | Tau | |
|---|---|---|---|
| SLEDAI-2K at study visit |
r=0.30
p=0.027 |
r=0.30
p=0.026 |
r=0.40
p=0.002 |
| AMS, over 1 year |
r=0.31
p=0.021 |
r=0.38
p=0.004 |
r=0.49
p=0.002 |
| SDI |
r=0.27
p=0.04 |
r=0.22 p=0.10 |
r=0.33
p=0.012 |
| IFN-α |
r=0.31
p=0.023 |
r=0.27
p=0.048 |
r=0.05 p=0.701 |
| IFN-γ | r=0.25 p=0.070 |
r=0.28
p=0.034 |
r=0.04 p=0.70 |
| IL-6 | r=0.20 p=0.151 |
r=0.28 p=0.042 |
r=0.18 p=0.194 |
| Current GC dose |
r=0.30
p=0.025 |
r=0.30
p=0.027 |
r=0.27
p=0.045 |
| Cumulative GC exposure | r=0.13 p=0.348 |
r=−0.14 p=0.315 |
r=0.15 p=0.273 |
Shown are relationships between brain injury markers and disease characteristics in patients with cSLE.
Spearman correlations in bold are significant for p<0.05.
AMS, adjusted mean SLEDAI-2K; cSLE, childhood-onset SLE; GC, glucocorticoids; GFAP, glial fibrillar acidic protein; IFN-α, interferon alpha; IFN-γ, interferon gamma; IL-6, interluekin-6; SDI, Systemic Lupus International Collaborating Clinics/American College of Rheumatology (SLICC/ACR) Damage Index; SLEDAI-2K, Systemic Lupus Erythematosus Disease Activity Index 2000; sNFL, serum neurofilament light.
Disease features associated with high-level brain injury marker cSLE subgroup
High-level markers were present in 13 (23%) patients with cSLE. Of these, six were elevated for one brain injury marker, six were elevated for two brain injury markers and one was elevated for all three brain injury markers. Table 5 shows comparisons of disease features between the high-level marker group and the rest of the cSLE group. The high-level marker group had a higher SLEDAI-2K level at study visit (median 6.0, IQR 4.0–8.0 vs 2.5, IQR 2.0–5.3, p<0.001), AMS over the disease duration (median 9.8, IQR 6.8–11.0 vs 2.5, IQR 1.2–4.9, p<0.001), serum IL-6 (median 3.16 pg/mL, IQR 1.20–10.70 vs 1.00 pg/mL, IQR 0.62–1.73, p<0.001) and current GC dose at visit (median 10 mg, IQR 7.5–17.5 vs 0 mg, IQR 0–3.0, p<0.001), all statistically significant for a corrected p value <0.0025.
Table 5. Comparison of characteristics between cSLE patients with versus without high-level brain injury markers.
| High-level marker group (n=13) |
Rest of cSLE group (n=43) |
P value | |
|---|---|---|---|
| Age at study visit, years | 15.0 (14.0–16.0) | 16.0 (13.0–17.0) | 0.409 |
| Female sex | 13 (100) | 35 (81) | 0.093 |
| Disease duration, months | 2.0 (0.9–6.7) | 9.2 (3.0–16.8) | 0.006 |
| Disease activity at visit (SLEDAI-2K) | 6.0 (4.0–8.0) | 2.5 (2.0–5.3) | <0.001* |
| Adjusted mean SLEDAI-2K | 9.8 (6.8–11.0) | 2.5 (1.2–4.9) | <0.001* |
| Damage (SDI >0) | 3 (23) | 2 (5) | 0.041 |
| IFN-α, pg/mL | 0.86 (0.22–5.20) | 0.20 (0.02–0.90) | 0.056 |
| IFN-γ, pg/mL | 0.20 (0.08–0.30) | 0.10 (0.06–0.20) | 0.033 |
| IL-6, pg/mL | 3.16 (1.20–10.70) | 1.00 (0.62–1.73) | <0.001* |
| NPSLE diagnosis | 0 (0) | 1 (2) | 0.579 |
| Lupus nephritis | 9 (69) | 12 (26) | 0.007 |
| Anti-phospholipid antibody positive | 2 (15) | 10 (23) | 0.544 |
| Low C3/C4 | 12 (92) | 27 (63) | 0.043 |
| DMARD use† (current) | 9 (69) | 21 (49) | 0.196 |
| Rituximab exposure (ever) | 5 (39) | 5 (12) | 0.027 |
| Current prednisone-equivalent dose, mg | 10 (7.5–17.5) | 0 (0–3.0) | <0.001* |
| Cumulative prednisone-equivalent dose, g | 7.0 (1.0–12.8) | 1.6 (0–4.7) | 0.107 |
| Intravenous steroid use (ever) | 5 (38) | 9 (21) | 0.201 |
| Abnormal conventional brain MRI | 1 (8) | 9 (21) | 0.275 |
Shown are exploratory comparisons of disease characteristics for cSLE patients with high-level brain injury markers (above the 90th percentile) versus those without high levels. Wilcoxon rank-sum tests were used for continuous variables (listed as median (IQR), unless otherwise specified) and χ2 tests were used for categorical variables (listed as n (%)). P values <0.05 are shown in bold.
Indicates significance after Bonferroni correction (p<0.0025).
DMARD includes mycophenolate analogues or azathioprine or methotrexate.
cSLE, childhood-onset systemic lupus erythematosus; IV, intravenous; NSPLE, neuropsychiatric SLE; SDI, Systemic Lupus International Collaborating Clinics/American College of Rheumatology (SLICC/ACR) Damage Index; SLEDAI-2K, Systemic Lupus Erythematosus Disease Activity Index 2000.
Discussion
In this cSLE cohort, we examined relationships between serum brain injury markers, measures of systemic inflammation and cognitive function. We found that adolescents with cSLE had significantly higher serum levels of GFAP, Tau, IFN-α, IFN-γ and IL-6 and a high proportion with CD compared with age-matched and sex-matched controls, despite the majority lacking a clinical NPSLE diagnosis. In addition, Tau was associated with CD in patients with cSLE, and those with the highest levels of brain injury markers (across sNFL, GFAP and Tau) had higher disease activity, serum IL-6 levels and GC dose. Our novel findings in cSLE add to the growing literature indicating a link between systemic inflammation, brain injury and CD in SLE, even in those without recognised NPSLE.
The brain injury markers assessed in this study represent different aspects of cellular brain injury, and our findings shed light on potential underlying mechanisms for brain involvement in cSLE. While sNFL, a marker of axonal injury that tracks closely with CSF levels and has been the most implicated in neuropsychiatric features of SLE,17,1921 we did not find differences in sNFL. Rather, GFAP and Tau were significantly elevated in cSLE compared with controls. GFAP is a specific marker of injury to astrocytes, which are abundant in the brain to provide neuro-supportive functions; however, under pathological inflammatory conditions in the setting of activated microglia, astrocytes change into a neurotoxic phenotype, reflected by high GFAP levels.13 32 Unlike sNFL, which is highly expressed in large-calibre myelinated neurons that extend subcortically, Tau is predominantly expressed in unmyelinated neurons in the cortex and released in the setting of injury to these vulnerable neurons.12 Our findings suggest that these neurons are particularly affected in adolescents with cSLE.
During adolescence, the brain undergoes marked changes in neurogenesis and cortical synaptic remodelling or ‘pruning’ during which unmyelinated neurons are actively changing, making this period susceptible to neurodevelopmental insults such as trauma and inflammation.4 33 Supporting this, we found the proportion with CD to be higher in adolescents with cSLE compared to their healthy peers, and CD was associated with higher Tau levels in cSLE. Although serum Tau is not as closely linked to CSF levels as sNFL, these findings parallel studies correlating elevated serum Tau to CD in patients with traumatic brain injury and neurodegenerative conditions.13 14 Together, the high GFAP and Tau levels and associated CD in our cSLE cohort indicate that brain injury occurs during the adolescent period of neurodevelopmental vulnerability, early in their disease course (median disease duration <1 year) and in patients largely (98%) without clinically diagnosed neuropsychiatric involvement. Although this finding needs to be replicated in other cohorts, the implication is that previously under-detected brain injury and dysfunction occurs in adolescents with cSLE in the early stages when disease activity is typically the highest.
Our analysis provides insight into the role of disease features in brain injury in adolescents with cSLE. While the correlations between brain injury markers and disease activity did not reach statistical significance, our results show weak associations between sNFL, GFAP and Tau and SLEDAI-2K at study visit, and slightly stronger associations with AMS representing averaged disease activity over the disease course. Furthermore, SDI was weakly associated with sNFL and Tau, supporting a potential link between overall disease damage and neuronal damage in the CNS. As expected, IFN-α, IFN-γ and IL-6 were higher in the cSLE group compared with controls, and sNFL and GFAP showed weak associations with these cytokines. It is notable, however, that the median SLEDAI-2K was low at 2.5 (IQR 2.0, 5.5), only 27% of the cohort had active disease and 9% had damage, which, along with our small sample size, may have affected our ability to detect associations between disease activity, damage and individual brain injury markers.
Our exploratory analysis of cSLE patients with the highest levels across all three brain injury markers found statistically significant relationships with higher disease activity at study visit and averaged over the disease course, and with higher IL-6 levels. These findings are in keeping with studies showing elevated IL-6 in patients with NPSLE.10 34 Additionally, shorter disease duration and the presence of lupus nephritis neared statistical significance, which is aligned with higher disease activity. It is worth noting that serum Tau can also be expressed from non-CNS tissues such as kidneys, liver and testes13; however, the elevation of both GFAP and Tau and associated CD in the cSLE cohort is supportive of a CNS source of serum Tau. We also note that we did not find associations between brain injury markers and other potential contributors to brain inflammation and injury such as complement-mediated factors and anti-phospholipid antibody status,8 although we were under-powered to detect these. Similarly, there was no association between high-level markers and abnormalities on conventional brain MRI, but more sensitive neuroimaging may be needed to detect related changes, and this is an area for future study, as combining brain injury markers and neuroimaging has yielded diagnostic advances in traumatic brain injury and multiple sclerosis.35 36
Regarding the relationship between immunosuppressive treatment in brain injury markers, we found that the current GC dose was weakly correlated with sNFL, GFAP and Tau levels and was associated with having high-level markers in the cSLE group. In contrast, cumulative GC dose over the disease course was not correlated to individual brain injury markers. As disease activity is likely paralleled by GC use in this early disease cohort, we think these findings suggest cSLE disease activity to be the likely driver rather than GC effects. Nevertheless, it is possible that GCs can exert neurotoxic effects on the brain,37 and this warrants further study as there may be implications for further steroid-sparing efforts to support neuroprotection. We did not find a difference in high-level brain injury marker status by disease-modifying antirheumatic drug use, although we were not powered for it. This is of interest for future study, given that a previous small study of youth and adults with NPSLE found that sNFL and GFAP serum levels declined with immunosuppressive therapy.18
The limitations of our study include the cross-sectional design and relatively small sample size with only one NPSLE participant. We assessed overall CD across executive function, attention and working memory, but other cognitive domains may also be affected. Although we looked at associations between brain injury markers and conventional clinical brain MRI abnormalities, we did not examine advanced neuroimaging such as diffusion tensor imaging, which provides more detail on microstructural brain tissue changes. We were also not able to look at the impact of different types of GCs (eg, intravenous vs oral) and specific immunosuppressive medications, and that is a future direction for study in a larger cohort. Additionally, we did not look at additional cytokines (eg, IL-1 beta and TNF-alpha) and individual complement cascade proteins, which have been implicated in microglial activation and brain injury. Larger, longitudinal studies will enable further investigation of these important gaps. Nevertheless, the strengths of our study lie in its investigation of multiple brain injury markers in a well-characterised cSLE cohort and inclusion of age-matched and sex-matched controls.
Conclusion
In our investigation of brain injury markers sNFL, GFAP and Tau in adolescents with cSLE, we found elevated levels of GFAP and Tau and almost 25% had very high levels across all three markers. Brain injury in this early disease cohort, at a critical neurodevelopmental period and largely without clinical NPSLE diagnosis, was associated with active disease, systemic pro-inflammatory cytokines and CD, suggesting inflammation-driven brain injury impacting function. Our study highlights the value of examining sNFL, GFAP, Tau and other neurological biomarkers collectively to broaden our understanding of mechanisms underlying brain involvement in cSLE and guide advances in the diagnosis and treatment of NPSLE. Importantly, our findings suggest that CD in cSLE may be under-detected and under-treated by the current standard of care, indicating a potential role for neuroprotective strategies and targeted cognitive interventions. Further studies in larger, longitudinal cohorts are needed to confirm our findings and determine trajectories of brain injury as well as temporal relationships to cSLE disease activity, damage and treatment.
Acknowledgements
We thank all the families, patients, parents and caregivers who participated in the study and allowed us to use their samples and data for this work. Parts of this work have previously been presented as meeting abstracts at the American College of Rheumatology Convergence 2024, International Neuropsychological Society 2025, Canadian Rheumatology Association Meeting 2024 & 2025, Lupus International Congress 2025 and Paediatric Rheumatology European Society 2025.
Footnotes
Funding: This work was supported by the Lupus Research Alliance (Novel Research Grant No. 481569 and Empowering Lupus Research Career Development Award No. 935840 to AMK), US Department of Defense (Lupus Research Program Impact Award No. W81XWH-20-1-0560 to AMK), Lupus Foundation of America (Gina M. Finzi Memorial Student Summer Fellowship Award to GR), Canada Institutes of Health Research (Canada Research Chair Tier 2 Award No. PCS-190986 to AMK) and SickKids Garry Hurvitz Centre for Brain & Mental Health (Capitalize for Kids Pediatric Mental Health Fellowship Award to OM).
Data availability free text: The datasets generated and analysed during the current study are available from the corresponding author on reasonable request. Due to patient confidentiality, the datasets will not be publicly shared.
Patient consent for publication: Not applicable.
Ethics approval: This study involves human participants and was approved by The Hospital for Sick Children Research Ethics Board (REB# 1000071306, REB# 1000063027, REB# 1000080072). Participants gave informed consent to participate in the study before taking part.
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient and public involvement: Patients and/or the public were not involved in the design, conduct, reporting or dissemination plans of this research.
Data availability statement
Data are available upon reasonable request.
References
- 1.Tucker LB, Uribe AG, Fernández M, et al. Adolescent onset of lupus results in more aggressive disease and worse outcomes: results of a nested matched case-control study within LUMINA, a multiethnic US cohort (LUMINA LVII) Lupus (Los Angel) 2008;17:314–22. doi: 10.1177/0961203307087875. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Santos F, Nascimento BR, Calderaro DC, et al. Neuropsychiatric syndromes in childhood-onset systemic lupus erythematosus: a systematic review and meta-analysis. J Clin Rheumatol. 2021;27:206–14. doi: 10.1097/RHU.0000000000001029. [DOI] [PubMed] [Google Scholar]
- 3.van der Heijden H, Rameh V, Golden E, et al. Implications of inflammatory processes on a developing central nervous system in childhood-onset systemic lupus erythematosus. Arthritis Rheumatol. 2024;76:332–44. doi: 10.1002/art.42736. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Kammeyer R, Ogbu EA, Cooper JC, et al. Cognitive dysfunction in pediatric systemic lupus erythematosus: current knowledge and future directions. Child Neuropsychol. 2024;30:818–46. doi: 10.1080/09297049.2023.2273573. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Williams TS, Aranow C, Ross GS, et al. Neurocognitive impairment in childhood-onset systemic lupus erythematosus: measurement issues in diagnosis. Arthritis Care Res. 2011;63:1178–87. doi: 10.1002/acr.20489. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Teixeira Santos FPS, Ferreira GA, de Paula JJ, et al. Cognitive dysfunction in patients with childhood-onset systemic lupus erythematosus may impact treatment. Adv Rheumatol . 2023;63:18. doi: 10.1186/s42358-023-00300-8. [DOI] [PubMed] [Google Scholar]
- 7.El Tal T, Jaffan J, Ledochowski J, et al. Examining self-reported executive function and its relationship to patient-reported outcomes and disease-related factors in youth with childhood-onset lupus: a cross-sectional study. ACR Open Rheumatology. 2025;7:e70102. doi: 10.1002/acr2.70102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Schwartz N, Stock AD, Putterman C. Neuropsychiatric lupus: new mechanistic insights and future treatment directions. Nat Rev Rheumatol. 2019;15:137–52. doi: 10.1038/s41584-018-0156-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Aw E, Zhang Y, Carroll M. Microglial responses to peripheral type 1 interferon. J Neuroinflammation. 2020;17:340. doi: 10.1186/s12974-020-02003-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Wang J, Yang C, Zhao Q, et al. Microglia activation induced by serum of SLE patients. J Neuroimmunol. 2017;310:135–42. doi: 10.1016/j.jneuroim.2017.07.010. [DOI] [PubMed] [Google Scholar]
- 11.Liang MH, Corzillius M, Bae SC, et al. The American College of Rheumatology nomenclature and case definitions for neuropsychiatric lupus syndromes. Arthritis & Rheumatism. 1999;42:599–608. doi: 10.1002/1529-0131(199904)42:4<599::AID-ANR2>3.0.CO;2-F. [DOI] [PubMed] [Google Scholar]
- 12.Frittoli RB, Pereira DR, Lapa AT, et al. Axonal dysfunction is associated with interferon-γ levels in childhood-onset systemic lupus erythematosus: a multivoxel magnetic resonance spectroscopy study. Rheumatology (Oxford) 2022;61:1529–37. doi: 10.1093/rheumatology/keab530. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Zetterberg H, Blennow K. Fluid biomarkers for mild traumatic brain injury and related conditions. Nat Rev Neurol. 2016;12:563–74. doi: 10.1038/nrneurol.2016.127. [DOI] [PubMed] [Google Scholar]
- 14.Koníčková D, Menšíková K, Tučková L, et al. Biomarkers of neurodegenerative diseases: biology, taxonomy, clinical relevance, and current research status. Biomedicines. 2022;10:1760. doi: 10.3390/biomedicines10071760. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Restuccia G, Susinna C, Marafioti G, et al. Serum and cerebrospinal fluid biomarkers as predictors of cognitive impairment in multiple sclerosis: a systematic review of longitudinal studies. J Neurol. 2025;273:36. doi: 10.1007/s00415-025-13499-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Abdelhak A, Foschi M, Abu-Rumeileh S, et al. Blood GFAP as an emerging biomarker in brain and spinal cord disorders. Nat Rev Neurol. 2022;18:158–72. doi: 10.1038/s41582-021-00616-3. [DOI] [PubMed] [Google Scholar]
- 17.Engel S, Boedecker S, Marczynski P, et al. Association of serum neurofilament light chain levels and neuropsychiatric manifestations in systemic lupus erythematosus. Ther Adv Neurol Disord. 2021;14:17562864211051497. doi: 10.1177/17562864211051497. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kammeyer R, Chapman K, Furniss A, et al. Blood-based biomarkers of neuronal and glial injury in active major neuropsychiatric systemic lupus erythematosus. Lupus (Los Angel) 2024;33:1116–29. doi: 10.1177/09612033241272961. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Zervides KA, Janelidze S, Nystedt J, et al. Plasma and cerebrospinal fluid neurofilament light concentrations reflect neuronal damage in systemic lupus Erythematosus. BMC Neurol. 2022;22:467. doi: 10.1186/s12883-022-02998-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Zervides KA, Grenmyr E, Janelidze S, et al. The impact of disease activity and interferon-α on the nervous system in systemic lupus erythematosus. Arthritis Res Ther. 2025;27:60. doi: 10.1186/s13075-025-03539-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Balajková V, Prokopcová A, Elisak M, et al. Elevated serum neurofilament light chain levels in patients with neuropsychiatric systemic lupus erythematosus: a cross-sectional study. Lupus Sci Med. 2025;12:e001309. doi: 10.1136/lupus-2024-001309. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Rodrigues Fonseca A, Felix Rodrigues MC, Sztajnbok FR, et al. Comparison among ACR1997, SLICC and the new EULAR/ACR classification criteria in childhood-onset systemic lupus erythematosus. Adv Rheumatol . 2019;59:20. doi: 10.1186/s42358-019-0062-z. [DOI] [PubMed] [Google Scholar]
- 23.Aringer M, Costenbader K, Daikh D, et al. 2019 European League Against Rheumatism/American College of Rheumatology classification criteria for systemic lupus erythematosus. Arthritis Rheumatol. 2019;71:1400–12. doi: 10.1002/art.40930. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Statistics Canada . Ethnic origin reference guide, census of population, 2016. 2017. p. 5. [Google Scholar]
- 25.Gladman DD, Ibañez D, Urowitz MB. Systemic lupus erythematosus disease activity index 2000. J Rheumatol. 2002;29:288–91. [PubMed] [Google Scholar]
- 26.Gladman D, Ginzler E, Goldsmith C, et al. The development and initial validation of the Systemic Lupus International Collaborating Clinics/American College of Rheumatology damage index for systemic lupus erythematosus. Arthritis Rheum. 1996;39:363–9. doi: 10.1002/art.1780390303. [DOI] [PubMed] [Google Scholar]
- 27.Ibañez D, Urowitz MB, Gladman DD. Summarizing disease features over time: I. Adjusted mean SLEDAI derivation and application to an index of disease activity in lupus. J Rheumatol. 2003;30:1977–82. [PubMed] [Google Scholar]
- 28.Delis DC, Kaplan E, Kramer JH. Delis-Kaplan Executive Function System (D-KEFS) Examiner’s Manual. San Antonio, TX: The Psychological Corporation; 2001. [Google Scholar]
- 29.Conners Continuous Performance Test 3rd Edition (Conners CPT 3) North Tonawanda, NY: Multi-Health Systems; 2014. [Google Scholar]
- 30.Wechsler D. Wechsler Adult Intelligence Scale–Fourth Edition (WAIS-IV) San Antonio, TX: The Psychological Corporation; 2008. [Google Scholar]
- 31.Wechsler D. WISC-V: Technical and Interpretive Manual. NCS Pearson Incorporated; 2014. [Google Scholar]
- 32.Fan Z, Hai W, Xianyi Z, et al. Advancements in understanding neuropsychiatric lupus: deciphering immune heterogeneity with single-cell resolution. Front Med. 12:1732805. doi: 10.3389/fmed.2025.1732805. n.d. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Crews F, He J, Hodge C. Adolescent cortical development: a critical period of vulnerability for addiction. Pharmacol Biochem Behav. 2007;86:189–99. doi: 10.1016/j.pbb.2006.12.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Hirohata S, Kikuchi H. Role of serum IL-6 in neuropsychiatric systemic lupus erythematosus. ACR Open Rheumatol. 2021;3:42–9. doi: 10.1002/acr2.11217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Duerksen J, Lopez RCT, Tappia PS, et al. Efficacy of biomarkers and imaging techniques for the diagnosis of traumatic brain injury: challenges and opportunities. Mol Cell Biochem. 2025;480:2797–814. doi: 10.1007/s11010-024-05176-w. [DOI] [PubMed] [Google Scholar]
- 36.Willard C, Puglisi L, Ravi D, et al. Combined magnetic resonance imaging and serum analysis reveals distinct multiple sclerosis types. Brain (Bacau) 2025;148:4578–91. doi: 10.1093/brain/awaf331. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Sapolsky RM. Glucocorticoids and hippocampal atrophy in neuropsychiatric disorders. Arch Gen Psychiatry. 2000;57:925–35. doi: 10.1001/archpsyc.57.10.925. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
Data are available upon reasonable request.