Background Neurofilaments (Nf) are major structural proteins that occur exclusively in neurons. In spinal cord injury (SCI), the severity of disease is quantified by clinical measures that have limited sensitivity and reliability, and no blood-based biomarker has been established to further stratify the degree of injury. We aimed to examine a serum-based NfL immunoassay as predictor of the clinical outcome in SCI.
Methods Longitudinal measurement of serum NfL was performed in patients with central cord syndrome (CCS, n=4), motor-incomplete SCI (iSCI, n=10), motor-complete SCI (cSCI, n=13) and healthy controls (HC, n=67), and correlated with clinical severity, neurological outcome, and neuroprotective effect of the drug minocycline.
Results Baseline NfL levels were higher in iSCI (21 pg/mL) and cSCI (70 pg/mL) than in HC (5 pg/mL, p=0.006 and p<0.001) and CCS (6 pg/mL, p=0.025 and p=0.010). Levels increased over time (p<0.001) and remained higher in cSCI versus iSCI (p=0.011) and than in CCS (p<0.001). NfL levels correlated with American Spinal Injury Association (ASIA) motor score at baseline (r=−0.53, p=0.004) and after 24 h (r=−0.69, p<0.001) and 3–12-month motor outcome (baseline NfL: r=−0.43, p=0.026 and 24 h NfL: r=−0.72, p<0.001). Minocycline treatment showed decreased NfL levels in the subgroup of cSCI patients.
Conclusions Serum NfL concentrations in SCI patients show a close correlation with acute severity and neurological outcome. Our data provide evidence that serum NfL is of prognostic value in SCI patients for the first time. Further, blood NfL levels may qualify as drug response markers in SCI.
Statistics from Altmetric.com
Acute spinal cord injuries (SCI) are one of the most devastating accidents affecting a young and active population. Mechanical injury of the spinal cord results in damage to neurons, axons, and glia at the area of impact.1 Over days to weeks, several secondary injury cascades lead to further progressive tissue damage within and adjacent to the primary lesion, on top of the sequelae of exogenous trauma. Other than the trauma itself, secondary injury2 ,3 represents a window for therapeutic interventions to preserve axons and their support structures.4–6 A recent placebo-controlled trial suggested clinical improvement across several outcome measures in patients receiving the drug minocycline,7 using the American Spinal Injury Association (ASIA) exam.8 The ASIA exam can be used to determine the ASIA grade of injury severity. While that is a universally accepted classification, each grade represents a broad category of patients, and the scale lacks sensitivity for longitudinal change making it insensitive as a drug response marker.9 A number of biomarkers have been evaluated for their capacity to be more sensitive and accurate tools to measure neuronal injury but, in part, because these tests are restricted to cerebrospinal fluid (CSF), they have as yet provided limited clinical usefulness.10 ,11
Neurofilaments (Nf) are neuron-specific, major structural proteins that are released into the blood following neuronal damage, and can be quantified using highly sensitive electrochemiluminescence (ECL) techniques.12 Among several activities, minocycline downregulates microglial activation, neuroinflammation and apoptosis, effects that may reduce neuronal injury reflected in lower serum levels of Nf in SCI.4
We analysed levels of Nf light chain (NfL) longitudinally in serum samples derived from subjects enrolled in a phase II clinical trial investigating the usefulness of minocycline to attenuate neurological deficits after spinal injury.7 We report the correlation of serum NfL with acute and long-term clinical outcome. Further, we investigated the potential of serum NfL as drug response marker of the therapeutic effect of minocycline in SCI.
The research protocol was approved by the University of Calgary Conjoint Health Research Ethics Board. All patients and healthy controls provided written informed consent. Serum samples from 67 healthy controls (HC) and 27 SCI patients (with sufficient amount of serum samples available7) were included: 13 patients with a motor-complete SCI (cSCI; ASIA A or B; due to lack of material no 96 hours (h) and 108 h samples for two patients and no 84 h and 120 h samples for one patient each), 10 patients with a motor-incomplete SCI (iSCI; ASIA C or D; no 36 h sample for one patient) and 4 patients with a central cord syndrome (CCS; ASIA C or D with more selective injury to the centrally located motor tracts and disproportionately greater motor impairment in the upper compared to lower extremities (mean lower extremity motor scores>upper extremity)13) (table 1).7 Surgical decompression and stabilisation was performed within 24 h of injury, and subjects were not treated with corticosteroids. Patients were randomised (1:1) to receive intravenous minocycline (Wyeth Pharmaceuticals either 200 mg of minocycline twice daily (low dose), or a loading dose of 800 mg tapered by 100 mg every 12 h until 400 mg was reached (high dose), or placebo. More detailed procedures and inclusion and exclusion criteria have been described previously.7
Clinical assessment and analytical procedure
Neurological function was assessed using the ASIA standardised neurological examination, including the motor (score ranging from 0 to 100, with higher score representing better motor function) and sensory (pinprick and light touch ranging from 0 to 112 each, with higher score representing better sensory function) composites.8 These examinations were performed at days 1 (time of enrollment), 4, 5 and 7; weeks 3, 6 and 12; and months 6 and 12. Motor function observed in the study population plateaued after 3 months.7 Motor and sensory outcome was defined as the mean of the motor or sensory scores at time points 3, 6 and 12 months.7
We examined blood samples drawn within 12 h of the SCI prior to treatment randomisation and every 12 h thereafter for 7 days (15 time points, including the baseline sample). Samples were spun at 2000 g for 10 min, aliquoted in polypropylene tubes and stored at −80°C within 2 h.7
An immunoassay developed in-house for NfL (NfLUmea47:3) was used for quantification of NfL in serum. The immunoassay is described in detail elsewhere.12 Briefly, 96-well plates (Multi-Array plates, Meso Scale Discovery, Gaithersburg, Maryland, USA) were coated with the capture monoclonal antibody 47:3 (UmanDiagnostics, Umea, Sweden). Adhering to a previously proposed nomenclature, the soluble fraction of NfL measured is specified with the capture antibody in the superscript (NfLUmea47:3).14 Samples were diluted 1:1 with Tris-buffered saline containing 1% BSA, and 0.1% Tween 20. 25 µL of standard, control or bovin serum albumin sample was then added in duplicates and the plate incubated at room temperature for 2 h. After washing, 25 µL of the secondary antibody (0.5 µg/mL, mAB 2:1) diluted in TBS containing 1% BSA, 0.1% Tween 20 was added to each well and the plate was incubated for 1 h at room temperature (UmanDiagnostics). After washing, MSD SULFO-TAG labelled streptavidin (0.25 µg/mL, MSD, Gaithersburg, Maryland, USA) diluted in TBS containing 1% BSA, and 0.1% Tween 20 was added and incubated for 1 h at room temperature. Following a final wash, 150 µL of ECL read buffer (MSD) diluted 1:2 with distilled water was added and the ECL signal, detected by photodetectors, measured using the MSD Sector Imager 2400 plate reader. A four-parameter weighted logistic fit curve was generated, and sample concentrations extrapolated and analysed using the Discovery Workbench V.3.0 software (MSD). If required, samples were appropriately diluted to fall in the range of the standard curve. Bovine NfL was obtained from UmanDiagnostics and standards ranged from 10 000 pg/mL to 0 pg/mL. The mean intra-assay and interassay coefficients of variation were 4.3% and 9.7%, respectively. The investigators who conducted the measurements had no access to the clinical data.
Baseline characteristics were compared between groups using Kruskal Wallis test for continuous variables, or Fisher's exact test for categorical variables. Continuous variables are described by their median and IQR, and categorical variables by numbers and percentages. The NfL area under the curve (AUC) for the 7 days sampling period was calculated using the trapezoidal rule. NfL values were log-transformed to achieve a Gaussian distribution. The longitudinal data of the groups (twice daily NfL levels) were analysed using mixed-effects linear regression, which allows for repeated measurements and missing values. One-way ANOVA with Bonferroni adjustment if significance was reached was used to compare log NfL between groups at individual time points. Additionally, mixed-effects linear regression was used to test for the effect of minocycline on log NfL, motor score, pinprick or light touch (baseline values were subtracted from all values at individual time points for this analysis). Because the number of patients per group was small, a pooled analysis (placebo vs both minocycline dosing schemes) was performed. Motor outcome was classified as ‘good’ or ‘poor’ considering the median value as cut-off. All correlation analyses were performed with Spearman's r. A two-sided p value<0.05 was considered as significant. All statistical analyses were performed using Stata V.12 (StataCorp, College Station, Texas, USA) and Graph Pad Prism V.5.02 for Windows (GraphPad Software, San Diego, California, USA).
Table 1 shows baseline characteristics of the SCI subgroups and healthy controls. The vast majority (24/27) of patients had a cervical level of injury. Across all subgroups, motor vehicle accidents represented two-thirds of causes of injury. cSCI, iSCI and CCS patients differed significantly for overall injury severity (ASIA score, p<0.0001), pinprick (p=0.020) and light-touch scores (p=0.028), while differences in motor scores were borderline significant (p=0.053) at baseline. Otherwise, there were no significant differences with regard to gender, age, or delay to surgery.
Fourteen patients (52%) were treated with either low dose (n=5) or high dose (n=9) minocycline, and 13 patients received placebo.
Levels of NfL at baseline and during follow-up
Baseline NfL levels were different between groups (F (3,90)=9.49, p<0.001); iSCI (21 (15–90) pg/mL, p=0.006) and cSCI (70 (17–134) pg/mL, p<0.001) had higher levels of serum NfL than HC (5 (2–11) pg/mL). Similarly, serum NfL levels in iSCI and cSCI were higher than in CCS (6 (0.3–18) pg/mL, p=0.025 and p=0.010, respectively) (figure 1).
There was an increase of serum NfL levels over time (p<0.001, mixed-effects model). Levels increased in all three groups from baseline (p<0.001) and were higher in cSCI versus iSCI (p=0.011) and CCS (p<0.001) (p=0.045 for iSCI vs CCS). Differences for individual time points, especially between cSCI and CCS, were strong (figure 2). Accordingly, the NfL AUC was significantly higher in cSCI (4614 (3486–7701)) versus CCS (1260 (420–2257), p=0.012; F (2,24)=5.9, p=0.008).
Baseline NfL and motor, pinprick and light-touch scores
NfL levels correlated with the motor score at baseline (r=−0.53, p=0.004; figure 3A and table 2), but not with the pinprick or light-touch scores (table 2). Levels also correlated with several of the postbaseline motor scores, less so with the sensory scales (table 2).
Interestingly, serum NfL levels determined after 24 h showed a stronger association with the baseline motor score (r=−0.69, p<0.001) and throughout all other sampling time points, including also the sensory scores (figure 3A and table 2). These correlations were of similar frequency and strength for the NfL AUC (ASIA score: r=0.60, p=0.001, figure 3A and table 2).
Correlation of NfL with clinical outcomes
Motor outcome plateaued at 3 months after SCI (median ASIA score: 60 points), regardless of whether the injury was motor-complete, motor-incomplete or of central cord type. As expected, the motor outcome was better in CCS (89 points, p<0.05) and iSCI (86.0 points, p<0.01) versus cSCI (25 points), and correlated strongly with respective baseline values (r=0.83, p<0.0001).7 Similar correlations between baseline and outcome scores were seen for the pinprick and light touch (r=0.63, p<0.001 and r=0.58, p=0.001).
During the period of 24 to 168 h after injury, NfL levels were increasingly higher (figure 4) in patients with a poor outcome (as defined by below median motor score, n=13) compared to those with a better outcome (median motor score above median, n=14, p=0.001). This was corroborated by an AUC analysis of NfL (better outcome: 1878 (1058–3464) versus worse outcome: 5234 (3017–8584), p=0.002). In line with the correlation of clinical scores at baseline with those of outcome, NfL levels at baseline correlated with motor outcome (r=−0.43, p=0.026, figure 3B). Again, this correlation became stronger over time for NfL measurements after 24 h (12 h: r=−0.56, p=0.003; 24 h: r=−0.72 (figure 3B); 48 h: r=−0.82; 72 h: r=−0.81; 120 h: r=−0.79; 144 h: r=−0.83; 168 h: r=−0.82, p<0.0001 for all) and for the AUC analysis (figure 3B). Similar correlations were noted for the outcome of pinprick and light touch (data not shown).
Correlation of NfL with minocycline treatment
Treatment was not evenly allocated across CCS, iSCI and cSCI (table 1); the median motor score at baseline was 30 for patients receiving placebo (n=13, 54% cSCI patients), 46 for those receiving low dose (n=5, 20% cSCI patients), and 26 for those receiving high dose minocycline (n=9, 56% cSCI patients) (p=0.9050). Compared to placebo, treatment with low or high dose of minocycline (median baseline motor score: 27 points) versus placebo did not have a significant effect on the longitudinal profile of serum NfL levels in this group of patients (p=0.67, mixed-effects model, figure 5A). Likewise, minocycline treatment had no effect on the longitudinal profile of motor (p=0.495), pinprick (p=0.324) or light-touch scores (p=0.264) in an all-patients analysis. A more comprehensive analysis of the effect of minocycline treatment on neurological outcome including these patients was previously published7.
By contrast, in cSCI (6 patients receiving minocycline, and 7 patients receiving placebo) minocycline-treated patients showed lower longitudinal NfL levels than placebo (p=0.048) (figure 5B). The effect of minocycline was more pronounced after removing the one patient on the low dose regimen (p=0.006), or the two patients with a thoracic level of the injury (p=0.030).
This difference of NfL levels was not paralleled by clinical findings, as scores between treatment groups were not different (motor: p=0.234; after removing one patient on low dose minocycline: p=0.132, or the two patients with the thoracic injury level: p=0.686; pinprick: p=0.681, and p=0.827, p=0.417 or light touch: p=0.185, and p=0.186, p=0.312) and failed to indicate a treatment effect of minocycline.
Similarly, in the 13 patients who had a motor score below the median motor score (28) at baseline (1 patient with CCS, 3 with iSCI and 9 cSCI), minocycline-treated patients (n=7) showed a trend for lower NfL levels over time in comparison to the placebo-treated group (n=6, p=0.085, figure 5C, after removing 2 patients on low dose minocycline: p<0.001). Again, clinical scores were not different (motor: p=0.567, and p=0.972; pinprick: p=0.551, and p=0.656; light touch: p=0.636, and p=0.320) between minocycline and placebo-treated patients.
This is the first study exploring the usefulness of NfL in serum as a marker of injury severity and outcome in patients with SCI, using a high-sensitivity ECL-immunoassay.12 NfL is the most abundant and also most soluble Nf subunit, factors that likely contribute to the superior sensitivity of NfL over neurofilament heavy chain (NfH) assays.15 The assay system is validated for CSF and serum, the latter allowing the acquisition of longitudinal measures in routine clinical settings.
In the course of acute and chronic neuronal damage, disruption to the axonal cell membrane releases Nf into the interstitial fluid, that eventually reaches the CSF and blood compartments.16 ,17 In SCI, trauma causes direct acute neuronal necrosis, followed by secondary injury mechanisms that increase neuronal loss by apoptosis or necrosis, a process that may last from days to weeks.18 ,19
At baseline, NfL levels in iSCI and cSCI were increased 4.2-fold and 14-fold compared to healthy controls, and 3.5-fold and 11.7-fold compared to CCS, respectively. Over the 7-day follow-up, NfL levels steadily increased in all three disease groups, with maximum levels (CCS: 230 pg/mL, iSCI: 384 pg/mL, cSCI: 796 pg/mL) being markedly higher than in more chronic diseases like Alzheimer's disease (37 pg/mL), Guillain-Barré syndrome (102 pg/mL) or amyotrophic lateral sclerosis (120 pg/mL).12 Only limited human data is available beyond this time point: three of four ASIA A and one of two ASIA C patients reached highest NfH plasma levels after 10 days in a recent pilot study,20 leaving the question of peak and duration of release open.
Only a few studies have investigated Nf in CSF or blood in SCI so far. First evidence for the usefulness of NfL as a marker of neuronal damage in SCI arose from a study investigating CSF in acute spinal cord disease: all 6 patients with SCI and 3 of the 17 with whiplash injury showed increased concentrations of NfL.21 In a rat model of SCI, NfH levels in blood22 were correlated with the extent of damage and were reduced by treatment with minocycline: however, this difference did not reach significance.23 The same group of investigators performed the first study in humans with acute cervical SCI: NfH was detectable in plasma of 11 of the 14 included patients, and ASIA A showed higher levels than ASIA C patients; however, correlation with clinical subscores, or outcome was not presented.20
The prognostic value of the ASIA grading system is limited by its lack of dynamic change over time on individual grounds, and its susceptibility to interference due to other injuries such as head or multisystem traumas, and drug effects.9 The consequence of these constraints for clinical studies is large patient numbers to achieve adequate statistical power, to observe differences in treatment regimens.24 The inclusion of motor and sensory subscores may increase the accuracy of the clinical grading.20 Our results show a high correlation between NfL and motor scores, at baseline, during follow-up, and for long-term outcome, indicating that serum NfL may be a reliable quantitative biomarker of the degree of SCI. This may be of specific value in the context of clinical trials where inter-rater variability of clinical assessments may increase the threshold to detect treatment effects. Furthermore, serum NfL may also allow for better evaluation of injury severity than the ASIA grade. In particular, it may allow further stratification and prognostication of the large population of ASIA A injuries.
So far, no drug has been shown to ameliorate the course of SCI, despite several candidate compounds showing promising results in animal models.25–27 Apart from the larger heterogeneity of human disease when compared to experimental SCI,28 this failure may also be attributed, in part, to the lack of sensitive biomarkers with a broad dynamic measuring range.
Minocycline is a tetracycline antibiotic that has shown neuroprotective properties in a variety of models of degenerative and acute neurological diseases, including SCI,29–33 by pathways unrelated to its antimicrobial activity.4 In the minocycline trial, an intravenous loading and maintenance dose to achieve serum levels similar to those efficacious in animal models of SCI was used.7 ,34 In the subgroups of cSCI, despite the small sample size, treated patients showed lower NfL levels at every time point beyond 24 h postinjury, whereas the clinical scales were insensitive to detect a difference between treatment groups. This reduction of NfL levels was more pronounced after excluding the patients with a thoracic SCI or those on low dose minocycline. These findings are in line with the clinical scores of the core study7 in which thoracic SCI patients did not benefit from minocycline treatment, and the comparison of the low dose and high dose minocycline groups suggested a greater effect with higher doses.
In summary, our data provide new evidence that serum NfL may represent a useful indicator of acute severity and long-term outcome of neuronal injury, especially in cases where accurate clinical assessment is not possible. Further studies are warranted to increase the evidence for NfL as drug response marker in SCI.
GG and SC contributed equally.
Contributors All coauthors fulfil the criteria for authorship: 1. Substantial contributions to the conception or design of the work; or the acquisition, analysis, or interpretation of data for the work; 2. Drafting the work or revising it critically for important intellectual content; 3. Final approval of the version to be published; 4. Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
Funding This study was supported by an unrestricted grant from Genzyme and institutional funding. The study sponsors had no involvement in the study design; in the collection, analysis and interpretation of the data; in the writing of the report; and in the decision to submit the paper for publication.
Competing interests J. Kuhle has received research support from the Swiss MS Society, Swiss ALS Society, Protagen AG, Roche, Novartis and Genzyme and served in scientific advisory boards for Genzyme/Sanofi-Aventis, Merck Serono and Novartis Pharma. His work is supported by an ECTRIMS Research Fellowship Programme and by the ‘Forschungsfonds’ of the University of Basel, Switzerland. J. Gaiottino reports no disclosures. D. Leppert is an employee of F. Hoffmann-La Roche Ltd. J.P. Bestwick reports no disclosures. A. Petzold has received research support from the Dutch MS research foundation and is named inventor on a patent on the use of neurofilaments as a biomarker for neuronal loss (WO2012005588A3). A. Malaspina reports no disclosures. R. Dobson reports no disclosures. C Lu reports no disclosures. G Disanto reports no disclosures. N. Norgren is employed by UmanDiagnostics AB, Sweden. A. Nissim reports no disclosures. L. Kappos reports, the University Hospital Basel as employer of Dr. Kappos has received and dedicated to research support fees for board membership, consultancy or speaking, or grants, in the last 3 years from Actelion, Advancell, Allozyne, Bayer, Bayhill, Biogen Idec, BioMarin, CSL Behring, Eli Lilly, European Union, GeNeuro, Genmab, Gianni Rubatto Foundation, Glenmark, Merck Serono, MediciNova, Mitsubishi Pharma, Novartis, Novartis Research Foundation, Novonordisk, Peptimmune, Roche, Roche Research Foundation, Sanofi-Aventis, Santhera, Swiss MS Society, Swiss National Research Foundation, Teva, UCB, and Wyeth. RJ Hurlbert reports no disclosures. VW Yong reports no disclosures. G. Giovannoni has received research grant support from Bayer–Schering Healthcare, Biogen–Idec, GW Pharma, Merck Serono, Merz, Novartis, Teva and Sanofi–Aventis. He has received personal compensation for participating on Advisory Boards in relation to clinical trial design, trial steering committees and data and safety monitoring committees from: Bayer–Schering Healthcare, Biogen–Idec, Eisai, Elan, Fiveprime, Genzyme, Genentech, GSK, Ironwood, Merck–Serono, Novartis, Pfizer, Roche, Sanofi–Aventis, Synthon BV, Teva, UCB Pharma and Vertex Pharmaceuticals. S Casha reports no disclosures.
Patient consent Obtained.
Ethics approval University of Calgary Conjoint Health Research Ethics Board.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.