Axonal damage markers in cerebrospinal fluid are increased in ALS

ALS is the most common form of motor neuron disease, characterized by progressive degeneration of spinal and bulbar innervating motor neurons as well as the pyramidal motor neurons.¹ Studies on histopathology^2,3 as well as transgenic mouse models^4–6 indicate that cytoskeletal components may be key factors contributing to neurodegeneration in ALS.⁷

Previous studies established cytoskeletal proteins such as neurofilaments, the microtubule-stabilizing tau protein, and protein 14-3-3 as biochemical markers for monitoring neurodegeneration, especially of the axonal component in vivo.^8–11 These proteins are released into the interstitial fluid compartment during neuroaxonal disintegration and diffuse into the CSF, where they can be quantified. Increased levels were found in a range of diseases characterized by neurodegeneration, including ALS.^9,12–15 Neurofilaments and other cytoskeletal components may therefore provide surrogates, closely linked to pathology, that allow monitoring of disease activity and progression in ALS.

In a previous study in a limited number of patients with ALS, we demonstrated a relationship between CSF biomarkers and disease severity.¹³ To what extent disease progression or clinical subtypes contribute to the elevation of CSF markers has not yet been investigated.

Extending our previous study,¹³ we obtained clinical and electrophysiologic data at baseline and follow-up to test whether CSF levels of two biomarkers for axonal damage (tau and NfH) would 1) distinguish clinical subgroups of ALS, 2) correlate with disability on clinical scales^16,17 and 3) be of prognostic value.

Methods.

This study was approved by the local ethics committee. Informed consent was obtained from all patients. Paired serum and lumbar CSF samples were collected by the Department of Neurology, University of Ulm (Germany) from 69 patients with probable or definite ALS according to revised El Escorial criteria (table 1).¹⁸ The patients with ALS included 67 patients with sporadic ALS and 2 patients with familial ALS (SOD1). The disease presented as classic (Charcot) ALS in 56 patients and as bulbar onset in 13 patients.¹⁹ Disability was rated using the Medical Research Council sumscore (MRCS)¹⁶ and Amyotrophic Lateral Sclerosis Functional Rating Scale (ALSFRS)¹⁷ by two experienced neurologists in our department (S.D.S. and A.C.L.), each blinded to the other’s assessment and unaware of any results on the CSF biomarkers. At time of lumbar puncture, 30 patients (43.5%) were being treated with riluzole (50 mg twice a day).

In 49 patients, follow-up data were available, with clinical scores recorded at time of lumbar puncture and after follow-up of up to 51 months (median 8 months). Disease progression was evaluated according to the monthly change in the MRCS between baseline and follow-up. The median was taken as cutoff for statistical purposes. The top 50% were classified as rapidly progressive, and the bottom 50% were classified as slowly progressive.

The control group consisted of 33 age-matched patients who presented with tension-type headache and showed no evidence of a structural, hemorrhagic, or inflammatory lesion. All CSF samples were labeled to ensure anonymity and stored at −80 °C until analysis. As disease controls, 73 patients with AD according to National Institute of Neurologic and Communicative Disorders and Stroke–Alzheimer’s Disease and Related Disorders Association criteria²⁰ were included (see table 1).

CSF leukocyte count (cells/mm³); total protein (g/L); lactate (mmol/L); CSF_alb/serum_alb concentration ratio (Q_alb); immunoglobulin (Ig) G, A, and M; and oligoclonal IgG bands as well as serum creatine kinase were obtained as previously described.¹³

A standard ELISA technique was used to quantify the CSF levels of tau.²¹ The neurofilament heavy chain phosphoform recognized by the monoclonal antibody SMI35 is labeled as NfH^SMI35, with the detecting antibody indicated in the superscript as suggested in a previously published nomenclature.⁹ CSF NfH^SMI35 levels were determined using an in-house ELISA technique based on commercially available antibodies.⁹

Motor and sensory nerve conduction velocity studies as well as EMG were performed to support the diagnosis of ALS and to rule out other neurogenic disorders. Determination of motor evoked potentials (MEPs), including calculation of central motor conduction time (CMCT), followed standard techniques.²²

Data analysis was performed using SPSS (version 12.0, SPSS Inc., Chicago, IL). Because of nonnormal data distribution, the medians and interquartile ranges are shown. All correlations were studied using the Spearman rank correlation coefficient. Multiple correlations were corrected using the Bonferroni method. Differences between groups were compared using the two-sided Wilcoxon two-sample test. Trend analysis was conducted using the Mantel–Haenszel chi-square test. P values less than 0.05 were considered significant. The association between severity of disease as measured by clinical scores and CSF tau or NfH^SMI35 concentrations was investigated by comparing proportions of patients with high values (increased above the top value of the control group, NfH^SMI35 > 1.77 ng/mL, tau > 350 pg/mL). To examine the accuracy of biomarkers to differentiate ALS from controls and AD, we used receiver operating characteristic (ROC) analysis, calculating the area under the ROC curve (AUROCC). The Youden index was calculated for each cutoff value as corresponding to [(sensitivity + specificity)] − 1] to find the cutoff values that maximize discriminating accuracy of the tests.²³

Results.

NfH^SMI35 and tau in ALS, AD, and controls.

The median CSF NfH^SMI35 concentrations were more than 5-fold higher in patients with ALS compared with the control group and more than 10-fold higher compared with AD patients (p < 0.001 for each comparison, table 1 and figure 1A. Also, the median CSF tau concentrations were approximately 2-fold higher in the ALS compared with the control patients (p = 0.02, see table 1 and figure 1B, whereas they were lower than in AD patients. To discriminate between ALS and controls, the AUROCCs were 0.87 and 0.65 for NfH^SMI35 and tau, respectively. To discriminate between ALS and AD, they were 0.94 and 0.84 for NfH^SMI35 and tau. The sensitivity and specificity were based on the previously published data,^9,24 and values for optimal cutoff values (our logistic regression analysis) are presented in table 2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. (A) Box plots show CSF NfH^SMI35 concentrations in patients with ALS, patients with Alzheimer disease (AD), and controls (CTRL). The box represents the 25th to 75th quartile, the whiskers represent the range, and the horizontal line in the box represents the median. (B) Box plots show CSF tau concentrations in patients with ALS, patients with AD, and controls. The box represents the 25th to 75th quartile, the whiskers represent the range, and the horizontal line in the box represents the median.

View this table:

• View inline
• View popup
• Download powerpoint

Table 2 Accuracy of CSF NfH^SMI35 and tau levels in diagnosis of ALS vs AD and controls

CSF NfH^SMI35 levels were 2-fold higher in ALS patients with predominantly clinical evidence of upper motor neuron (UMN) involvement¹ (2.4 ng/mL) compared with those with predominantly lower motor neuron (LMN) pathology¹ (1.2 ng/mL, p = 0.02, figure 2). Accordingly, CSF NfH^SMI35 levels in patients showing a prolongation of CMCT in MEP (n = 22, 2.5 ng/mL) were higher compared with patients whose CMCT was not prolonged (n = 10, 0.6 ng/mL, p = 0.02). Interestingly, the seven ALS patients with a full MRCS (60/60) also had CSF NfH^SMI35 concentrations higher (median 1.5 ng/mL) than those of controls (p = 0.002). CSF tau (but not NfH^SMI35) increased with age in the ALS patients (p = 0.03, R = 0.25) and the control population (p = 0.03, R = 0.42). There was no significant difference in either CSF tau or CSF NfH^SMI35 levels between patients with classic ALS and bulbar onset. Likewise, no difference was seen between patients with or without treatment with the neuroprotective agent riluzole (50 mg twice a day). CSF findings in patients with ALS were normal except for mild increases of Q_alb in 27.5% (n = 19) of all cases. No significant correlation of either CSF tau or NfH^SMI35 with Q_alb could be detected.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. Box plots show CSF NfH^SMI35 concentrations in ALS with predominantly lower motor neuron signs (LMN, n = 36) and ALS with predominantly upper motor neuron signs (UMN). The box represents the 25th to 75th quartile, the whiskers represent the range, and the horizontal line in the box represents the median.

NfH^SMI35, tau, and disease progression.

The median disease progression on the MRCS was 1.0 points/month in this cohort. Those patients whose disease progressed more quickly had higher CSF NfH^SMI35 concentrations than those who deteriorated by less than 1.0 points/month (p < 0.001, table 3 and figure 3). Although CSF tau concentrations tended to be slightly higher in patients with rapid disease progression, this did not reach significance (p = 0.13).

View this table:

• View inline
• View popup
• Download powerpoint

Table 3 CSF NfH^SMI35and tau in ALS with rapid and slow progression of disease

• Download figure
• Open in new tab
• Download powerpoint

Figure 3. Box plots show CSF NfH^SMI35 concentrations in ALS with rapid progression of disease (loss in Medical Research Council sumscore [MRCS] over follow-up > median of 1.0/month) and ALS with slow progression of disease (loss in MRCS over follow-up < median of 1.0/month). The box represents the 25th to 75th quartile, the whiskers represent the range, and the horizontal line in the box represents the median.

Discussion.

We investigated different CSF biomarkers for axonal degeneration in a large cohort of patients with ALS compared with another neurodegenerative disease, AD. Consistent with previous studies,^9,13,25,27 the current data showed that CSF NfH^SMI35 and tau concentrations were increased in patients with ALS. This finding supports the notion of ongoing destruction of axons in the course of the disease leading to an increased release of cytoskeletal tau and NfH^SMI35 in CSF. This is consistent with postmortem human studies and with evidence derived from animal models demonstrating axonal and neuronal accumulation of neurofilament proteins.^2–7,11 Importantly, CSF NfH^SMI35 was superior to tau with concentrations fivefold higher in ALS patients than in controls.

This finding was reinforced by the markedly higher levels of sensitivity of CSF NfH^SMI35 compared with tau for distinguishing between ALS and another neurodegenerative disease (AD) or controls, based on previously published upper reference limits (80% vs 7%).^9,24 In an attempt to improve sensitivity and specificity, we used ROC analysis. Because these new, optimized reference limits differed considerably in distinguishing patients with ALS from either patients with AD or controls (see table 2), we recommend using the previously published reference limit for CSF NfH^SMI35 of 0.73 ng/mL.⁹ For tau, our data revealed a considerable overlap between patients with ALS and controls, a finding that may explain a report of normal CSF tau concentrations in patients with ALS in a smaller cohort of patients.¹⁵ This may impair the use in clinical practice.

To evaluate which pathologic features may give rise to the observed increase of CSF biomarkers for axonal degeneration, we compared disease subtypes, disease severity (MRCS and ALSFRS), and progression rate (between baseline and follow-up). There was no association between CSF markers of axonal damage and severity of disease as measured by clinical scores (MRCS and ALSFRS). However, CSF NfH^SMI35 concentrations in patients with a normal MRCS were significantly higher compared with those of controls. Our findings suggest that biomarkers such as CSF tau and NfH^SMI35 could enable us to detect axonal damage in ALS patients at a biologically early, probably subclinical stage. This is important because the patient may not yet have accumulated enough axonal degeneration to result in a functional deficit detectable on clinical scales such as the MRCS or the ALSFRS. We believe this finding potentially to be important for the subgroup of genetically identified familial ALS patients in whom a biomarker may help to identify the period preceding phenotype conversion, enabling targeted treatment options in the future.

In addition, we observed CSF NfH levels to be higher in ALS patients showing clinical evidence of damage predominantly to the UMN¹ as well as in patients showing neurophysiologic evidence of UMN involvement. Damage to the UMN would result in wallerian degeneration of a large-caliber axon along its entire spinal cord pathway, resulting in the release of a large amount of NfH into the CSF. In contrast, degeneration of the LMN would only release NfH into the CSF from adjacent axons between the anterior horn cell and the exit of the motor root. This would be proportionally less important. The CSF NfH levels observed in patients with predominantly LMN involvement (1.2 ng/mL) were comparable to those we found in patients with Guillain–Barré syndrome who had long-term disability thought to be caused by proximal axonal degeneration affecting the nerve roots.²⁶ These findings are also consistent with a previous report demonstrating approximately seven times more neurofilament light chain (NfL) in patients with UMN compared with LMN pathology.²⁷ The problem we and others encountered with NfL was its susceptibility to proteases, which makes it unstable, as opposed to the more heavily phosphorylated NfH.^28–30 Indeed, 30% to 40% of CSF NfL are lost within 3 to 4 days when stored at 4 °C.³¹

Our data showed CSF NfH^SMI35 to be increased in patients with rapidly progressive disease as measured by loss of motor function (MRCS) over follow-up (see figure 3 and table 3). We conclude that very high concentrations of CSF NfH^SMI35 and tau represent extensive axonal damage and point toward a rapid progression of disease that may be associated with a poorer prognosis. This is in accord with studies on other neurodegenerative diseases, where increases of CSF tau and 14-3-3 protein are characteristic findings and correlate with rapid progression of disease.^32,33 However, we believe the current data to be still too weak to allow any prognostic conclusions based on CSF markers of axonal damage in clinical practice. Follow-up data in a larger group of patients over a longer period of time may be necessary to prove this point.

In contrast to previous findings by our group on a smaller cohort of patients with ALS,¹³ the current data showed no significant association between CSF tau levels and duration of disease. The release of biomarkers for axonal damage into CSF seems to be primarily influenced by UMN involvement and a faster clinical progression of disease.

Acknowledgment

The authors thank Miss Donna Grant, Mrs. Dagmar Vogel, Mrs. Refika Aksamija, and Mrs. Christa Ondratscheck for technical assistance.