Objectives We evaluated whether the measurement of serum phosphorylated neurofilament heavy chain (pNF-H) titre is likely to be a valid biomarker of axonal injury in multiple sclerosis (MS).
Methods Serum pNF-H concentrations were measured by ELISA in cases with relapsing-remitting (RR)-MS (n=81), secondary progressive (SP) MS (n=13) and primary progressive (PP)-MS; n=6) MS; first demyelinating event (FDE; n=82); and unaffected controls (n=135). A subset of MS cases (n=45) were re-sampled on one or multiple occasions. The Multiple Sclerosis Severity Score (MSSS) and MRI measures were used to evaluate associations between serum pNF-H status, disease severity and cerebral lesion load and activity.
Results We confirmed the presence of pNF-H peptides in serum by ELISA. We showed that a high serum pNF-H titre was detectable in 9% of RR-MS and FDE cases, and 38.5% of SP-MS cases. Patients with a high serum pNF-H titre had higher average MSSS scores and T2 lesion volumes than patients with a low serum pNF-H titre. Repeated sampling of a subset of MS cases showed that pNF-H levels can fluctuate over time, likely reflecting temporal dynamics of axonal injury in MS.
Conclusions A subset of FDE/MS cases was found to have a high serum pNF-H titre, and this was associated with changes in clinical outcome measures. We propose that routine measurement of serum pNF-H should be further investigated for monitoring axonal injury in MS.
- MULTIPLE SCLEROSIS
Statistics from Altmetric.com
In multiple sclerosis (MS), it has been difficult to directly assess the efficacy of current and emerging therapies for reducing axonal injury, the likely pathological substrate of progressive neurological decline. Given the likely association between axonal degeneration and disability progression in MS,1 ,2 a serum biomarker of axonal injury could prove useful as a prognostic or monitoring tool. The three major neurofilament protein subunits, (light, NF-L; medium, NF-M and heavy, NF-H), form the backbone of the axonal cytoskeleton and, following axonal injury or death, can be detected in serum or cerebrospinal fluid (CSF) allowing their potential use as biomarkers of neurodegeneration.3 It has been demonstrated that NF-L and NF-H, released from damaged axons, can be detected in CSF by ELISA methodology and that higher levels of both proteins are associated with poorer clinical outcomes in MS.4–8 Although the results of these studies are promising, the clinical utility of CSF sampling is limited due to risks and discomfort associated with repeated lumbar puncture.
In this study we attempted to detect and measure the level of the heavily phosphorylated axonal form of NF-H (pNF-H) in serum samples of patients with MS, study participants with a first demyelinating event (FDE) and unaffected controls. We selected pNF-H as a target for immunological detection as it is relatively resistant to proteolysis and can be captured with high avidity due to its unusual multiepitope nature.3 We then assessed whether serum pNF-H levels were associated with known clinical and paraclinical measures of disease severity, in order to validate this protein as a likely surrogate marker for axonal injury in MS.
Standard protocol approvals, registrations and patient consents
The core study was approved by the Eastern Health Human Research Ethics Committee, and the Ausimmune Study was approved by nine regional Human Research Ethics Committees as previously described.9 Written informed consent was obtained from all patients participating in the study.
Case and control demographics
Consecutive cases with relapsing-remitting (RR)-MS/secondary progressive (SP)-MS and primary progressive (PP)-MS, scheduled for routine clinical MRI, were recruited at the Eastern Clinical Research Unit, Box-Hill Hospital between 2009 and 2012. A small group of unaffected controls, which comprised Eastern Health Staff (n=19), were also recruited at this site. Initial venous blood sampling and a clinically indicated cerebral MRI were performed on the same day for each patient. Cases with a FDE (n=82) and the majority of unaffected controls (n=116) were participants in the Ausimmune Study. Details of the design of that study are provided elsewhere.9 ,10 Summary study participant demographics are presented in table 1. Unaffected controls had no known neurological diseases. Furthermore, at the time of initial sampling 59/81 RR-MS, 4/13 SP-MS and 1/6 PP-MS cases were receiving disease modifying treatment (ie, Copaxone, Interferon or Tysabri). The median duration from last reported relapse at time of sampling was RR-MS 10 months (IQR 3–29 months) and SP-MS 76 months (IQR 5–232 months); and the number of cases with reported relapse within 3 months of sampling were 20/81 RR-MS and 2/13 SP-MS.
Clinical outcome measures
Neurological outcome was assessed using the Kurtzke Expanded Disability Status Scale (EDSS). For MS patients, the EDSS score was used in combination with prospectively collected patient data (IMED database) on disease duration from the time of the first reported symptom, to generate Global Multiple Sclerosis Severity Scores (MSSS).11
For MS patients and Box-Hill controls, venous blood samples (9 mL) were drawn into an additive free plastic blood collection tube (Becton Dickinson, Australia) and allowed to clot for a minimum of 30 min and a maximum of 120 min at room temperature. Samples were centrifuged at 2000 g for 10 min, and serum supernatant removed, aliquoted and stored at −80°C. For Ausimmune Study participants, venous blood was drawn into an serum separation tube (SST) tube and allowed to clot at room temperature. Serum was removed after centrifuging, aliquoted and stored at −80°C.
Immediately following venous blood sampling, RR-MS, SP-MS and PP-MS cases were scanned with a 1.5 T MRI scanner (Sonata, Siemens Medical Systems, Erlangen, Germany). T2, fluid-attenuated inversion recovery, diffusion weighted imaging, pregadolinium and postgadolinium (Gd) enhancing T1 images were acquired using the following parameters: Turbo Spin Echo (T2 weighted imaging. Axial. TR/TE=4000/80 ms, FA=90°C, FOV=256×208 mm2, Matrix=256×208, NEX=2. Thickness=5 mm, Slices=20, Interslice=1 mm), Spin Echo (T1 weighted imaging. Axial, sagittal. TR/TE=500/8 ms, FA=90°C, FOV=256×208 mm2, Matrix=256×208, NEX=2, Thickness=5 mm, Slice=20, Interslice=1), Single-Shot Spin-Echo-Echo Planar Imaging (Diffusion weighted imaging, Axial. TR/TE=6000/87 ms, FA=90°, FOV=256×256 mm2, Matrix=128×128, NEX=3, Thickness=5 mm, Slice=20, Interslice=1 mm, b=100 s/mm2) and Fluid-Attenuated Inversion Recovery (Axial. TR/TE/TI=9000/107/2500 ms, FA=180°, FOV=256×208 mm2, Matrix=256×208, Thickness=5 mm, Slice=20, Interslice= 1 mm). The MRI data were assessed for the presence or absence of gadolinium enhancing and T2 lesions by an experienced neuroradiologist (YL). T2 lesion volume was calculated using MRIcro software, with manual validation. MRI data was not available for four RR-MS cases, one PP-MS case and one FDE case.
ELISA for pNF-H detection
The pNF-H ELISA used a pure mouse monoclonal capture antibody clone AH1 (1 µg/mL) EnCor Biotechnology Inc, Gainesville, Florida, USA, with methods slightly modified from those previously described.12 ,13 In brief, a 96-well plate was incubated with 100 µl/well of pure mouse monoclonal capture antibody clone MCA-AH1 (1 µg/mL) EnCor Biotechnology Inc, Gainesville, Florida, USA in sodium bicarbonate buffer (50 mM, pH 9.5) overnight at 4°C. For assessment of experimental samples, 40 µL of serum sample in 10 µL blocking solution (2% skim milk in 0.1% TBST) was added per well, and the plate was incubated at 4°C overnight. Bovine pNF-H protein standards were serially diluted down the plate in blocking solution from 12.5 ng/mL, with blocking solution used as a blank (0 ng/mL). The plate was then incubated for 1 h with antigen affinity purified rabbit anti-pNF-H (RPCA-NF-H, ∼1 microgram/mL, EnCor), followed by goat anti-rabbit alkaline phosphatase (1:500, Vector Laboratories) for 1 h at room temperature. Then 100 µL of developing solution (5 mg of p-nitrophenyl phosphate (Sigma-Aldrich, St Louis, Missouri, USA) in 5 mL of 0.1 M glycine, 1 mM MgCl2 and 1 mM ZnSO4 solution) was added to each well. Optical densities were measured after 2 h development using a Biorad Benchmark Plus Microplate Spectrophotometer at 405 nm absorbance. Samples were tested in duplicate on independent plates.
The interassay/intra-assay coefficient of variability was calculated from the optical densities of plate standards. The standards were tested in duplicate, and taken from six independent assay plates over a period of 10 days. The average interassay coefficient of variance was 10%, and the average intra-assay coefficient of variance was 14%.
Logistic regression analyses were used to investigate associations between pNF-H status and other factors, in IBM SPSS Statistics for Windows (V.21.0. Armonk, New York, USA: IBM Corp). Comparisons between group medians were conducted using a Mann–Whitney Rank sum test or Kruskal-Wallis ANOVA.
Serum pNF-H levels in participant groups
To determine the cut-off values for normal serum pNF-H concentrations, we first assayed sera from 135 unaffected controls (figure 1A). The serum pNF-H concentration was found to be median 0, IQR 0–0.09 ng/mL. We conservatively defined a ‘positive’ serum pNF-H level as ≥97th centile for controls, that is, 0.57 ng/mL. The comparison group, that is, participants with serum pNF-H levels below the cut-off, were designated as negative. A total of 20/182 (11%) patients with either FDE or MS were serum pNF-H positive, compared with 3/135 unaffected controls (2.2%) (figure 1A). The median serum pNF-H concentration for the groups was: FDE 0 (IQR 0–0.18), RR-MS 0 (IQR 0–0.16), SP-MS 0.18 (IQR 0.04–0.63) and PP-MS 0.12 (IQR 0.02–0.71). The median serum pNF-H concentration for SP-MS cases was significantly higher than healthy controls (p=0.011), FDE (p=0.041) and RR-MS cases (p=0.048). No statistically significant differences were observed between other groups.
Serum pNF-H status in relation to disease severity in MS patients
Patients with SP-MS were more likely to have a high serum pNF-H titre compared to RR-MS and FDE participants (38.5% vs 9%). As permanent disability in MS is thought to be largely attributable to central nervous system (CNS) axonal injury, we hypothesised that a serum pNF-H positive status would be associated with more severe disease, as indicated by a high MSSS score. Indeed, the pNF-H-positive RR-MS/SP-MS patients had higher median MSSS scores than pNF-H negative patients (figure 1B; 5.9 IQR 4.2–6.5 vs 4.0 IQR 1.4–6.2, p<0.05). Furthermore, univariate logistic regression analyses showed that MSSS was predictive of pNF-H status (positive/negative) (p=0.028). Area under the curve analysis for MSSS vs NFH status also showed that high MSSS is a fair predictor of a positive NFH status (AUC=0.683, SE 0.066; asymptotic significance=0.041).
Serum pNF-H status in relation to MRI measures
The pNF-H status was not associated with the presence of Gd-enhancing lesions, which were uncommon in this cohort (p=0.9). Specifically, 8/100 patients had evidence of at least one Gd enhancing lesion, and one of these was pNF-H positive. The pNF-H positive MS patients had a higher median T2 lesion volume compared to negative patients (figure 1C; 9947 mm3 IQR 2445–19638 vs 3560 mm3 IQR 1003–11624, p=0.07). Univariate logistic regression analyses showed that T2 lesion volume was predictive of pNF-H status (p=0.011)
Using univariate logistic regression analyses, we showed that in RR-MS/SP-MS patients, age, sex, time from last reported relapse, contrast lesion number or treatment status were not independently predictive of serum pNF-H status (p>0.2 for all comparisons). Only 1/12 pNF-H positive cases reported relapse within 3 months of sampling, and 7/12 were receiving disease modifying treatment.
In the FDE cohort, 80% of cases had their disease status recorded at 5-year follow-up. Interestingly, of the 32 cases with clinically confirmed MS at 5-year follow-up, two were serum pNF-H positive at FDE diagnosis. These cases had EDSS scores of 3 and 8 at 5 years, compared to a median 5-year EDSS score of 2.0 (IQR 1–2.5) for the remainder of the group. It should be noted that five additional cases were positive, but remained stable over 5 years. Based on these observations, it is possible that high serum pNF-H at time of FDE, although uncommon, might predict more severe disease at 5 years, but this needs to be verified in a larger data set. Due to the relatively small group of FDE patients, it was not possible to ascertain if FDE serum pNF-H titre or status was associated with risk of conversion to MS.
Variation in serum pNF-H levels over time
Repeat serum samples were collected from 45 of the RR-MS/SP-MS cases at various time points over a 2-year period (n=8 SP-MS and n=37 RR-MS), at least 6 months apart, to assess the dynamics of serum pNF-H levels over time (n=26 re-sampled once and n=19 re-sampled twice). These re-samples included 7/12 cases that were originally found to have higher than normal serum pNF-H levels. Interestingly, all re-sampled cases with high initial serum pNF-H titre had at least one other positive serum pNF-H sample over a 2-year re-sampling period. Of the cases with low serum pNF-H titre at the time of initial sampling, 9/38 (23.6%) were found to have a high serum pNF-H titre on at least one occasion over a 2-year study period. Comparison of cases with at least one high serum pNF-H sample relative to those with persistently low serum pNF-H, again revealed that these cases had higher median MSSS scores (high serum pNF-H (n=16): 5.5 IQR 0.6–6.4 vs low pNF-H (n=29): 3.5 IQR 1.1–5.5; p=0.04). Collectively, these observations suggest that serum pNF-H levels do fluctuate over time in some patients and that intermittent or persistently high titre is associated with more severe disease.
We have demonstrated that 11% of FDE/MS cases have higher than normal serum pNF-H levels at a single time point. Further, ‘High pNF-H’ status can fluctuate over time, and approximately 30% of MS cases will show a positive test at some point over 2 years with repeat sampling. Importantly, more SP-MS patients than RR-MS/FDE patients were found to be serum pNF-H positive; and serum pNF-H positive status was associated with more severe disease and larger T2 lesion volumes, consistent with the hypothesis that the presence of this protein in serum is likely to reflect higher levels of injury to CNS axons.
In previous studies, NF-H and NF-L levels have been shown to be increased in the CSF of MS cases and possibly associated with disease severity. Initially, a moderate association between CSF NF-L concentration and EDSS score was reported in RR-MS cases,4 and this was later confirmed in progressive MS cases.7 Other work has shown a moderate correlation between CSF pNF-H concentrations and EDSS scores and noted that higher levels are found in progressive MS cases.5 Conversely, assessments of neurofilament levels in serum are limited, although a few, preliminary studies suggest that this is likely to be feasible.3 Recently, it was suggested that a higher serum to CSF ratio in SP-MS cases could be prognostic for EDSS worsening at 3 years.14 In that study, however, the prognostic value of serum pNF-H levels were not assessed independently of CSF levels. In the present study, we use the global MSSS, which provides a cross-sectional measure of comparative disease severity for patients with different disease durations. It should be noted that the MSSS has not been validated for long-term outcome and should not be implied as a prognostic measure. We show that pNF-H positive RR-MS/SP-MS cases have a higher mean MSSS compared to pNF-H negative patients, consistent with a more severe disease course. Our preliminary findings are therefore consistent with the premise that serum pNF-H titres, measured by ELISA, are likely to be reflecting CNS axonal injury, one of the main pathological mechanisms underlying accumulation of permanent disability in MS.1 ,2
Several MS cases with high MSSS did not show a high serum pNF-H titre at baseline sampling. However, our repeated sampling studies showed that serum pNF-H levels fluctuate over time. These findings suggest that pNF-H may be released periodically rather than continuously and that following release, it is degraded in serum over time. It is, therefore, likely that cross-sectional sampling will underestimate the proportion of patients with intermittently high pNF-H serum titre. Hence, we suggest that follow-up studies should incorporate a repeated sampling strategy at planned intervals, to determine whether routine monitoring of serum pNF-H levels could be useful for predicting the clinical course of MS, or therapeutic response.
We were unable to show an association between serum pNF-H levels and the presence of cerebral Gd-contrast enhancing lesions. On the other hand, we found that serum pNF-H positive patients showed higher average cerebral T2 lesion volumes. These findings are potentially concordant with the hypothesis that active lesions contribute less to the rate of axonal injury than macrophage-mediated injury at the edge of chronic active lesions and low-level inflammation in perilesional white matter, both occurring in the absence of ongoing blood-brain barrier breakdown.15 Overall lesion volume could be a better proxy for the extent of these events than Gd-enhancing lesion load. It should, however, be noted that alternative MRI measures including the extent of cerebral atrophy16 and T1 black holes,17 not examined here, could have utility in future studies of disease severity in relation to serum pNF-H status. Further, we did not scan the spinal cords at time of sampling, so that the relationship of acute cord lesions with serum pNF-H was not evaluated.
In our exploratory analyses, we were unable to determine whether high serum pNF-H titre at time of FDE onset was predictive of later conversion to MS due to the small group sizes and suggest that this should be a focus of future studies. We were also unable to confirm that serum pNF-H concentration or status was associated with age or recent relapse, in contrast to previous studies using CSF samples from MS cases.6 Although the association between serum and CSF levels were not investigated here, it has been previously found that CSF and serum NF-H levels were not strongly associated within individuals with a range of inflammatory and non-inflammatory neurological conditions.14 Based our observations that NF-H level and status are not associated with the presence of inflammatory lesions on MRI or relapse activity, we propose that some pNF-H could be released into serum directly from chronic lesions. Thus, it may be necessary to assess the utility of CSF and serum blood pNF-H levels as independent biomarkers of axonal injury in MS.
We have shown that high serum pNF-H titres can be detected in a proportion of MS patients and are more likely to be detected in SP-MS than FDE/RR-MS. A ‘pNF-H positive’ status is associated with higher disease severity scores and T2 cerebral MRI lesion load, supporting our hypothesis that serum pNF-H levels are likely to reflect CNS axonal injury. Our results suggest that the routine measurement of serum pNF-H should be further investigated as a prognostic indicator of disease outcome and therapeutic response in MS.
Collaborators The following authors are part of the Ausimmune Consortium; K-J Lazarus, MP Marriott, O Skibina, and The Ausimmune Consortium: BV Taylor, RM Lucas, TJ Kilpatrick, K Dear, MP Pender, I van der Mei, C Chapman, A Coulthard, T Dwyer, AJ. McMichael, P C Valery, D Williams, A-L Ponsonby.
Contributors MG: Contributed to the study design, responsible for optimising and conducting ELISAs, data analyses and manuscript preparation. LFD and AP: Contributed to experimental design and manuscript preparation. GS: Inventor of pNF-H ELISA, assisted with optimising this methodology for testing of human samples; contributed to manuscript preparation. YL and HP: Radiologists responsible for conducting and analysing MRI scans. BVT, RML: Assisted with analysis of the Ausimmune data set, and contributed to revision of the manuscript. JH, FP, AP, LL, DM: Involved in study design or assisted with data interpretation and manuscript revision. AV: Assisted with statistical analyses. HB: Primary investigator responsible for study design and collection of clinical data, and also assisted with data analysis and manuscript preparation. Statistical analyses were conducted by MMG, HB, Department of Medicine, University of Melbourne, Parkville, Australia; and AV, Royal Melbourne Hospital, Parkville, Australia.
Funding This work was supported by Multiple Sclerosis Research Australia (Incubator grant to MG and HB; Postgraduate Research Scholarship awarded to LFD); CASS Foundation (Medical research grant to MG and HB); CharityWorks for MS (fellowship to MG); The National Health and Medical Research Council Australia (Project grant 628699 to HB, GS, TK; Career development Award 628856 to HB and 1024898 to RL; and AWP is an NHMRC Senior Research Fellow); NHMRC Centre for Research Excellence Grant 1001216 (fellowship support for MG); Beijing Nova Programme (xx2013045 to YL) and the National Natural Science Foundation of China (Nos. 81101038 to YL).
Competing interests The authors declare that GS holds equity in EnCor Biotechnology Inc, a company commercialising the Neurofilament pNF-H ELISA used in this study, and may benefit by receiving royalties of equity growth. AP holds a patent for Neurofilament-H cleavage products (WO 2012005588 A2). All other authors have nothing to declare.
Ethics approval Eastern Health Human Research Ethics Committee.
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.