CSF and blood neuronal injury biomarkers in spinal bulbar muscular atrophy and amyotrophic lateral sclerosis 4

Abstract

Spinal and bulbar muscular atrophy (SBMA) and amyotrophic lateral sclerosis 4 (ALS4) are two forms of motor neuron disease characterized by clinically slow disease progression. Based on the current limited human studies, the contribution of central nervous neurodegeneration to these diseases and the rate of clinical progression is unclear. Neuronal proteins glial fibrillary acidic protein (GFAP), neurofilament light (NfL) chain, or Total-tau measured in either cerebrospinal fluid or blood could serve as sensitive markers of neurodegeneration. We studied 56 adult participants (32 SBMA, 7 ALS4, and 17 controls) who were enrolled at the National Institutes of Health, of whom 22 (10 SBMA, 7 ALS4, and 5 controls) underwent paired CSF and serum sampling, and of whom 6 participants were assessed longitudinally up to 24 months from initial visit. An additional 7 controls completed CSF sampling only. CSF GFAP, NfL chain, and Total-tau correlated with corresponding levels in serum (r = 0.74, r = 0.47, and r = 0.70, respectively). CSF GFAP was increased in patients with SBMA (median, 8840 pg/mL, interquartile range (IQR) 5780–10489) as compared to controls (median, 5315 pg/mL, IQR 1822–6657; P = 0.029) but not compared with ALS4 (median, 5015 pg/mL, IQR 3172–9803; P = 0.31). Patients with SBMA had increased concentrations of CSF NfL chain (median, 719 pg/mL, IQR 483–773) as compared to ALS4 (median, 307 pg/mL, IQR 187–629; P = 0.034) or controls (median, 395 pg/mL, IQR 307–497; P = 0.024). In contrast, serum concentrations of either biomarker did not differ significantly between SBMA, ALS4, or controls. Higher CSF GFAP and NfL chain levels were associated with lower SBMA Functional Rating Scale scores (r = −0.49 and r = −0.42, respectively). Over the course of 24 months, the average change in SBMA Functional Rating Scale was −0.83 points, while the changes in CSF GFAP and NfL chain were progressive (increased 1.4-fold and 1.3-fold, respectively). Our data suggest that SBMA patients have increased concentrations of CSF GFAP and NfL chain as compared to ALS4 and controls, and higher levels of these biomarkers are associated with disease severity. Importantly, these results indicate that SBMA is associated with progressive neurodegeneration and that either CSF GFAP or NfL chain may be useful for patient stratification and monitoring treatment effects in clinical trials.

Shahim et al. report that neurofilament light and glial fibrillary acidic protein which are markers of neurodegeneration are elevated in CSF of patients with spinal bulbar muscle atrophy and higher levels of these biomarkers are associated with disease severity scales concluding that spinal bulbar muscle atrophy may be associated with neurodegeneration.

Graphical Abstract

Graphical Abstract.

Introduction

Spinal and bulbar muscular atrophy (SBMA) is an adult-onset neuromuscular disease caused by a CAG-repeat expansion in the androgen receptor gene affecting mainly men.^1,2 SBMA is characterized by slowly progressive weakness and wasting of bulbar and limb muscles.² Preclinical testing in animal models of SBMA has provided a rationale for targeting toxicity within the muscle³ as well as the central nervous system,⁴ and availability of fluid biomarkers to assess tissue-specific engagement will be important in future clinical trials.

Amyotrophic lateral sclerosis type 4 (ALS4) is a rare, adolescent-onset autosomal dominant form of ALS that is due to a gain-of-function mutation in the senataxin (SETX) gene.⁵ The mean onset of ALS4 is 17 years; however, the range is broad (5–63 years).⁶ Unique features of ALS4 include normal life expectancy, absence of bulbar involvement, and symmetric distribution of distal atrophy and weakness. These features therefore, make ALS4 distinct from ALS.⁶ Autopsy in ALS4 patients has demonstrated spinal cord atrophy, loss of anterior horn and dorsal root ganglia cells, and degeneration of corticospinal tracts and posterior columns.⁶ However, clinically the disease is characterized by very slow disease progression.⁶

Currently, there are no therapies available for either SBMA or ALS4. An important limitation of previous trials has been the lack of outcome measures. Biomarkers to measure disease progression and evaluate tissue-specific therapeutic response are strongly needed. Recent advances in the ultrasensitive immunoassay technology field have made it possible to reliably and rapidly quantify neuronally derived proteins in peripheral blood. In particular, neuronal proteins such as glial fibrillary acidic protein (GFAP), neurofilament light (NfL) or total-tau (T-tau) have emerged as sensitive markers of neurodegeneration.^7-11 While these biomarkers have been extensively assessed in other neurodegenerative diseases,^12-14 their role in SBMA and ALS4 has yet to be assessed in detail.

In the context of SBMA, a recent study found no elevations in the serum concentrations of NfL compared with controls,¹⁵ although mild elevation in early-stage cases has been reported.¹⁶ Several studies have found increased concentrations of NfL in cerebrospinal fluid (CSF) of patients with ALS compared with controls.^17-21 ALS4 has several distinct features from ALS, and the role of NfL as a biomarker of disease progression is yet to be examined. Based on the current limited human studies, the contribution of central nervous neurodegeneration in SBMA is unclear. ALS4 may be associated with mild progressive neurodegeneration given the slow rate of clinical progression.

We previously reported the clinical features of SBMA.² In this study, we examined whether SBMA and ALS4 are associated with CSF and serum biomarkers of neuronal injury and astrogliosis and whether higher levels of these biomarkers in CSF or blood correlate with functional outcome measures. We specifically assessed cross-sectional and longitudinal differences in CSF and blood GFAP, NfL and T-tau across patients with SBMA, ALS4 and controls hypothesising that patients with SBMA and ALS4 would have higher concentrations of these biomarkers in CSF and serum. We also assessed the relationship between these biomarkers and functional outcome measures, hypothesising that biomarkers at baseline would predict changes in functional outcome measures.

Materials and methods

Study design and participants

A detailed study protocol and inclusion and exclusion criteria can be found on (ClinicalTrials.gov Identifier: NCT04944940 and NCT04394871). In summary, men over the age of 18 years with and without SBMA diagnosis, and ALS4 patients over 5 years were enrolled in two natural history studies of SBMA and ALS4 at the National Institute of Neurological Disorders and Stroke (NINDS). All participants had a medical history, physical exam, and blood collection at enrolment. Participants were also offered lumbar puncture (LP). Participants with SBMA were followed up over 2 years (every 6 months). Participants without SBMA were assessed only once.

Fluid, clinical, and functional outcome measures

The main outcome measures were differences in CSF or serum GFAP, NfL, and T-tau in relation to the diagnoses of SBMA or ALS4, and SBMA Functional Rating Scale (SBMAFRS).²² We also assessed the relationship between CSF biomarkers and timed up and go (TUG),²³ total adult myopathy assessment tool (AMAT),²⁴ and 6-min walk test (6MWT).²⁵

Blood and CSF handling and processing

Blood samples were collected by venipuncture into gel-separator tubes for serum and centrifuged within 20–60 min. Serum samples were aliquoted and stored at −80°C pending biochemical analysis. LP was performed through L3-L4 or L4-L5, between 10:00 a.m. and 3:00 p.m. A total of 10–20 mL CSF was collected in a single polypropylene tube from each participant. The CSF was gently mixed, and a cell count was performed to exclude blood contamination. Thereafter, the CSF was centrifuged (2000 g at 4°C for 10 min), and the supernatant was aliquoted in 0.5 mL portions in polypropylene screw cap cryo tubes that were stored at −80°C pending analysis. The participants completed a neurological examination before LP.

Quantification of GFAP, NfL and T-tau

GFAP, NfL and T-tau concentrations in CSF and serum were measured using the Neurology 4-plex assay kit (Quanterix Corporation, Lexington, MA, USA) on a Single molecule array HD-1 Analyzer (Quanterix Corporation, Bellerica, MA, USA). The average coefficient of variation of measurement of GFAP, NfL and T-tau in either CSF or serum were <4%, <5% and <10%, respectively. All samples were processed and analyzed at the NINDS, Bethesda, MD, USA, using the same batch of reagents by certified laboratory technicians blinded to clinical information.

Statistical analyses

We used χ² test to examine differences in categorical variables between the SBMA versus the control group. Continuous variables were inspected visually with histograms and Q-Q plots. GFAP, NfL and T-tau concentrations in CSF and serum were non-normally distributed in patients with ALS4; therefore, non-parametric tests were used unless specified otherwise. For group-wise comparison, we used Kruskal-Wallis analysis of variance performed for the multiple group comparisons followed by the Mann-Whitney U test. The correlations between the CSF and serum values as well as the correlation between SBMA functional outcome measures and the fluid biomarkers were assessed using either Pearson or Spearman rank correlation. All statistical analyses were performed in GraphPad Prism v.10 (GraphPad Inc., San Diego, CA) and R (v.4.4.1, The R Foundation for Statistical Computing).

Results

Demographic and clinical characteristics

A total of 56 adult participants (32 SBMA, 7 ALS4 and 17 controls) were enrolled, of whom 22 (10 SBMA, 7 ALS4 and 5 controls) underwent paired CSF and serum sampling. An additional seven age and sex matched control CSF samples to the SBMA patients were purchased from PrecisionMed. Patients with SBMA were assessed with paired CSF and serum sampling, of whom six underwent longitudinal sampling (2 at 12 months and 4 at 12 and 24 months from initial visit). One patient with ALS4 underwent longitudinal sampling 12 months after initial assessment. One of the control participants was excluded from the analysis due to abnormally high concentrations of NfL (1478 pg/mL) in CSF. The demographic and clinical characteristics of the participants at baseline are shown in Table 1. The demographic and clinical characteristics of participants who underwent LP for CSF analysis are shown in Supplementary Table 1. Age and sex did not differ significantly between SBMA, ALS4 or controls (Table 1 and Supplementary Table 1). The associations between baseline levels of the fluid biomarkers and demographic and clinical characteristics are shown in Supplementary Fig. 1. Serum GFAP correlated with age and sex hormone binding globulin, while serum NfL correlated with age and creatine kinase (Supplementary Fig. 1). No significant correlations were observed between CSF levels of these measures and age or other clinical measures except for T-tau and age, (Supplementary Fig. 1).

Table 1.

Demographic and clinical characteristic of study participants

Characteristics	SBMA	ALS4	Controls
N	32	7	16
Age, y	56 (10)	51 (20)	55 (10)
Sex, no. (%)
Female	0 (0.0)	2 (29)	0 (0.0)
Male	32 (100)	5 (71)	17 (100)
Disease duration, y	16 (10)	37 (27)	n/a
CAG repeats	46 (3.5)	n/a	n/a
SBMAFRS score	41 (7)	n/a	n/a
Creatinine, mg/dL	0.69 (0.22)	0.65 (0.24)	0.94 (0.19)
Creatine kinase, U/L	692 (460)	235 (115)	179 (82)
Myoglobin, ng/mL	161 (79)	n/a	52 (12)
SHBG, nmol/L	49 (14)	33 (8)	47 (11)
Testosterone, ng/dL	509 (226)	248 (171)	500 (137)
FSH, U/L	8.3 (6.1)	n/a	6.4 (3.5)
LH, U/L	5.8 (3.9)	n/a	5.6 (3.8)

Data are shown mean (SD) for the baseline assessment unless otherwise indicated. 7 of 16 controls had data on creatinine, creatine kinase, myoglobin, SHBG, testosterone, FSH, and LH.

Abbreviation: FSH, follicle-stimulating hormone; LH, luteinizing hormone; SHBG, sex hormone binding globulin.

Correlation between CSF and serum biomarkers

CSF GFAP, NfL and T-tau correlated with corresponding levels in serum (r = 0.74, r = 0.47 and r = 0.70, respectively; Fig. 1).

Figure 1.

Correlation between CSF and serum biomarkers. Scatter plots showing the pairwise relationship between CSF and serum level of (A) GFAP, (B) NfL, and (C) T-tau in all participants (SBMA, ALS4 and controls). Each data point represents a study participant (n = 23). All correlations were calculated using the Pearson method. CSF, cerebrospinal fluid; GFAP, glial fibrillary acidic protein; NfL, neurofilament light chain; T-tau, total-tau.

CSF GFAP and NfL are increased in SBMA compared with ALS4 or controls

CSF GFAP was increased in patients with SBMA (median, 8840 pg/mL, IQR 5780–10489) as compared to controls (median, 5315 pg/mL, IQR 1822–6657; P = 0.029) but not compared with ALS4 (median, 5015 pg/mL, IQR 3172–9803; P = 0.31) (Fig. 2A). Patients with SBMA had increased concentrations of CSF NfL (median, 719 pg/mL, IQR 483–773) as compared to ALS4 (median, 307 pg/mL, IQR 187–629; P = 0.034) or controls (median, 395 pg/mL, IQR 307–497; P = 0.024) (Fig. 2B). CSF T-tau did not differ between the groups (Fig. 2C). In contrast, serum concentrations of NfL or GFAP did not differ between SBMA, ALS4 or controls (Fig. 2D–F).

Figure 2.

CSF NfL and GFAP are increased in patients with SBMA compared with ALS4, or controls. Scatterplots show the differences in GFAP, NfL and T-tau measured in CSF (A–C) and serum (D–F) across SBMA, ALS4 and controls. Each data point represents a study participant. Values are presented as medians; error bars indicate interquartile range. P values are from the Kruskal-Wallis test of variance followed by Mann-Whitney U test. Significant group differences are displayed, except for CSF NfL that was also significantly higher in SBMA compared with ALS4 (P = 0.034). CSF, cerebrospinal fluid; GFAP, glial fibrillary acidic protein; NfL, neurofilament light chain; T-tau, total-tau. 11 age-and sex matched controls to SBMA from NIH blood bank.

The diagnostic accuracies of the fluid biomarkers in differentiating SBMA from ALS4 or controls are shown in Supplementary Fig. 2. In summary, CSF NfL had the highest diagnostic accuracy in separating SBMA from controls and ALS4 with area under the receiver operating characteristic curve (AUROC) of 0.79 and 0.77, respectively (Supplementary Fig. 2A, B). In serum, the AUROCs were not statistically significant, except for T-tau, where it was 0.94 in separating SBMA from controls and 0.89 for SBMA versus ALS4 (Supplementary Fig. 2D–F).

Association between fluid biomarkers and SBMA outcome measures

CSF GFAP, NfL and T-tau correlated with SBMAFRS, TUG, AMAT function and total AMAT scores (Fig. 3), with the strongest correlations between GFAP and TUG (r = 0.71, P < 0.0001; Fig. 3E). Similar significant correlations were also seen between serum measurements of these biomarkers and SBMAFRS, TUG, AMAT function, and total AMAT scores (Fig. 3A). Correlations calculated using non-parametric method are shown in Supplementary Fig. 3.

Figure 3.

Association between fluid biomarkers and SBMA outcome measures. Panel (A) shows the correlation between fluid biomarkers and functional outcome measures. Plots (B–F) show examples of correlations indicated in the heatmap. Warmer colours indicate positive correlation, while cooler colours indicate negative correlations. Bold outlines, P < 0.01; dashed outlines P < 0.05. Each data point represents a study participant sample. All correlations were calculated using the Pearson method. 6MWT, 6-minute walk test; AMAT, total adult myopathy assessment tool; SBMAFRS, spinal bulbar muscular atrophy functional rating scale; TUG, timed up and go. SBMA, n (CSF) = 30 (23). TUG, n (CSF) = 28 (22). AMAT, n (CSF) = 27 (21). MWT, n (CSF) = 27 (21).

Longitudinal changes in CSF biomarkers and SBMA functional outcome measures

Over the course of 24 months, the average change in SBMAFRS was −0.83 points, while the changes in CSF GFAP, NfL and T-tau were 1.40-fold, 1.25-fold and 1.16-fold, respectively (Fig. 4A). In contrast, the changes in serum were incremental except for T-tau (1.2-fold) (Fig. 4A). Similarly, the average changes over time in the SBMA functional outcomes were small (Fig. 4B). The individual changes in the fluid biomarkers and SBMA outcome measures over time are provided in Supplementary Figs 4–5.

Figure 4.

Longitudinal changes in fluid and SBMA functional outcome measures. Plot (A) shows the absolute fold changes over 24 months for the fluid biomarkers measured in CSF and serum. Plot (B) shows the absolute fold changes over 24 months for the SBMA outcome measures for comparison. Data in plot (A) and (B) are based on the 22 patients with SBMA who underwent paired CSF and serum sampling; of whom six underwent longitudinal sampling (2 at 12 months and 4 at 12 and 24 months from initial visit). 6MWT, 6-minute walk test; AMAT, total adult myopathy assessment tool; CSF, cerebrospinal fluid; GFAP, glial fibrillary acidic protein; NfL, neurofilament light chain; SBMAFRS, spinal bulbar muscular atrophy functional rating scale; T-tau, total-tau; TUG, timed up and go.

Discussion

The main findings of this study were: (i) patients with SBMA have increased concentrations of CSF GFAP and NfL as compared to ALS4 and controls; (ii) although CSF and serum levels of these biomarkers were correlated, the serum levels did not differ significantly between the groups; (iii) CSF levels of GFAP and NfL correlated with SBMA clinical outcome measures; and (iv) longitudinally, the levels of GFAP and NfL increased while the SBMAFRS was stable. This study provides CSF evidence of neuronal degeneration in SBMA and evidence that either GFAP or NfL could be used as sensitive outcome measures in future trials of SBMA.

SBMA is primarily thought to be a lower motor neuron disease with minimal or no upper motor neuron involvement. However, a recent study found white matter abnormalities as measured by diffusion tensor MRI.²⁶ Both GFAP and NfL have been shown to be highly sensitive markers of neuronal degeneration where they have been shown to detect even mild forms of CNS injury and may detect neurodegeneration before it becomes visible on imaging.^27,28 In the context of SBMA, a previous study found no relationship between serum NfL and diagnosis of SBMA,¹⁵ although testing in a cohort of early-stage patients showed a mild elevation of serum NfL.¹⁶ Consistent with aforementioned studies, herein, we did not observe significant differences in the levels of serum NfL between SBMA, ALS4, or controls. In contrast to the previous studies, we measured NfL as well as GFAP and T-tau in paired CSF and serum samples of SBMA, ALS4, and controls. We found that GFAP and NfL measured in CSF are significantly elevated in patients with SBMA compared with controls and CSF NfL is also elevated in SBMA compared with ALS4. CSF T-tau did not differ significantly between SBMA, ALS4, and controls. We have previously observed similar results where CSF GFAP and NfL were more sensitive markers of neurodegeneration, especially in cases of mild or early neurodegeneration when compared with CSF T-Tau.^29-31 There are a number of plausible explanations to why the levels of serum GFAP, NfL or T-tau in serum did not perform as well as CSF despite the moderate to strong correlations between the two fluid sources, including faster clearance from blood or intact function of the blood-brain barrier. The mechanism of how these biomarkers, especially NfL are released into the blood is yet to investigated. Patients with SBMA had lower serum T-tau compared with ALS4, and controls, which could be due to minimal contributions of T-tau in disorders with less severe CNS degeneration.³² Another plausible explanation could be that T-tau measured in peripheral blood may also increase with physical exertion and patients with SBMA are less ambulatory compared with controls.³³ Furthermore, we found that higher levels of both CSF and serum NfL, GFAP and T-tau correlated with several SBMA clinical outcome measures including SBMAFRS. Notably, we had a larger serum sample size than CSF, which could impact the correlation analysis with other disease measures. Together, these results suggest that SBMA is associated with CNS degeneration and that neuronal degeneration may underlie clinical worsening.

ALS4 may mimic ALS or SBMA; however, there is mild clinical disease progression with potentially less CNS involvement.⁶ In support of this hypothesis, we found no differences in the levels of either CSF NfL, GFAP or Tau between ALS4 and controls.

Longitudinal sampling allows us to assess neuronal degeneration over time. We found that in patients with SBMA both GFAP and NfL measured in CSF increased over 24 months. In contrast, SBMAFRS remained mostly stable over the same period, decreasing by only 0.83 points during the same time. In direct comparison, a recent meta-analysis found that SBMAFRS decreased with 0.9 points per year.³⁴ Similar results have been observed in other neurodegenerative diseases where changes in GFAP and NfL as well MRI-measured neurodegeneration preceded functional and cognitive outcome measures.³⁵ Together, these results indicate that SBMA may be associated with progressive neurodegeneration.

Furthermore, given the stability of a common outcome measure of SBMA demonstrating stability over time, the dynamic range of biomarkers such as GFAP or NfL may provide suitable options as outcome measures. The opportunity to distinguish and measure both CNS and peripheral contributions to degeneration is important in complex disorders that involve multiple organ systems.^12,23 Biomarkers indicative of CNS injury would be useful to follow in future trials of SBMA in which the CNS tissue is a therapeutic target and impact on the CNS would be anticipated as part of an early pharmacodynamic effect.

The main limitation of this study was the modest sample size for the CSF part of the study; however, both SBMA and ALS4 are rare diseases and obtaining CSF samples is challenging. The correlations we detected between CSF and serum levels of CNS injury markers would suggest that blood levels may reflect CNS damage in the disease. Further characterization of these peripheral measurements will be needed to understand their connection with neuronal damage in SBMA and establish their relationship with disease progression. Our analysis was restricted to measurement of GFAP, NfL, and T-tau and evaluation of additional markers from the CSF may provide additional insight.

The results suggest that SBMA patients have increased concentrations of CSF GFAP and NfL as compared to ALS4 and controls, and higher levels of these biomarkers are associated with disease severity. Importantly, these results indicate that SBMA is associated with progressive neurodegeneration and that either CSF GFAP or NfL may be useful for patient stratification and monitoring treatment effects in trials of disease-modifying therapies. Future directions include replication and validation of the CSF findings in a larger longitudinal cohort.

Supplementary material

Supplementary material is available at Brain Communications  online.

Funding

The study was funded by the Intramural Research Program of the National Institute of Neurological Disorders and Stroke (ZIA-NS009455-01). The funders of the study had no role in the design, data collection, data analysis, data interpretation, or writing of the report.

Competing interests

The authors report no competing interests.

Data availability

The data supporting the findings are available from the corresponding author upon reasonable request.