Comparison of neurofilament light and heavy chain in spinal muscular atrophy and amyotrophic lateral sclerosis: A pilot study

Abstract Background Spinal muscular atrophy (SMA) and amyotrophic lateral sclerosis (ALS) were two major motor neuron diseases with similar symptoms and poor outcomes. This study aimed to identify potential biomarkers in disease monitoring and differential diagnosis of adult SMA patients with sporadic ALS patients. Methods This was a pilot study with ten adult SMA patients and ten ALS patients consecutively enrolled during hospitalization. Serum and cerebrospinal fluid (CSF) samples were collected for assessment of neurofilament light (NFL) and phosphorylated neurofilament heavy chain (pNFH). Serum creatine kinase (CK) and creatinine (Cr) were also compared between groups. The receiver operating characteristic (ROC) curves were used to identify differentiated values among ALS and SMA patients. Results Serum Cr, CSF NFL, and CSF pNFH levels of ALS patients were significantly higher than those of the adult SMA patients (p < .01). Serum CK and Cr were strongly correlated with baseline ALSFRS‐R scores in SMA patients (p < .001). The ROC curves revealed an area under the curve (AUC) of 0.94 in serum Cr with a cut‐off value of 44.5 μmol/L (Sensitivity 90%, Specificity 90%). AUC from the ROC curve of CSF NFL and CSF pNFH were 1.0 and 0.84, with cut‐off values of 1275 pg/mL and 0.395 ng/mL, respectively (Sensitivity and Specificity of 100% and 100% in CSF NFL; Sensitivity and Specificity of 90% and 80% in CSF pNFH). Conclusion CSF NFL and pNFH might be useful biomarkers for differential diagnosis of adult SMA and ALS.


Participants
This is a pilot study consecutively enrolled patients hospitalized in the

Sample collection and measurement
Blood and CSF samples were collected during the first admission.
Serum aliquots were obtained after being centrifuged (3000 × g for 10 min) of within 2 h of blood collection, which were stored with CSF samples at −80 • C for further analysis. Additionally, serum samples were taken from ten healthy volunteers for comparison of biomarkers. Serum neurofilament light (NFL) concentrations were analyzed by single molecular array (AstraBio, Suzhou, China) assay and the fully automated instrument AST-DX90 Analyzer (AstraBio, Suzhou, China) following the manufacturer's instructions. The mean interassay and intraassay CV were both less than 10%. We measured the concentrations of NFL in CSF using the commercially available enzymelinked immunosorbent assay (ELISA) kits (IBL, Hamburg, Germany), and the mean intraassay coefficient of variation (CV) was less than 20%. For measurements of phosphorylated neurofilament heavy chain (pNFH) concentrations, another ELISA kit (EUROIMMUN AG, Lübeck, Germany) was used with an intraassay of less than 5%. All the measurements were performed by research assistants who were blinded to the diagnosis and clinical data.

Statistical analysis
Statistics were carried out using SPSS 22.0, and graphs were drawn with Graphpad Prism 7. Values of NFL and pNFH that are below the lower limit of detection (LLOD) were approximated to the lowest concentration of detection (0.01 pg/mL of NFL, 0.01 ng/mL of pNFH), and those that exceed the higher limit of measurement were

Clinical characteristics
A total of 20 patients completed the study. Table 1

Relationship between biomarkers with clinical features
The age at enrollment was not correlated with ALSFRS-R scores or any laboratory parameters in SMA and ALS patients, but the onset age was positively correlated with the CK and Cr levels in SMA patients and CSF total protein in ALS patients (p < .05, Supplementary Tables S1 and S2). The ALSFRS-R scores were also strongly correlated with the serum levels of CK and Cr in SMA patients (p < .001, r = 0.927; p < .001, r = 0.924, respectively), while significant correlation of ALSFRS-R scores in ALS patients was only found with serum pNFH levels (p = .015, r = −0.735). The rate of disease progression was inversely correlated with Cr levels in SMA patients, while no significant correlation between DPR and biomarkers was found in ALS patients. For the pulmonary function in SMA patients (Supplementary Table S1), we found FVC was positively correlated with onset age and ALSFRS-R scores (p < .01), and inversely correlated with DPR (p < .01). Moreover, the CK and Cr levels were also positively correlated with FVC in SMA patients. plementary Table S2), we found the NFL levels in CSF were significantly correlated with pNFH levels in CSF among ALS patients (p = .005, r = 0.804), while no correlation was found between NFL and pNFH levels in serum (p > .05). We also did not find a correlation between NFL and pNFH levels both in serum and CSF among SMA patients (p > .05).

For the correlation between different laboratory parameters (Sup
In addition, the NFL levels in serum were strongly correlated with NFL levels in CSF among ALS patients (p = .006, r = 0.791). However, no correlation between NFL levels in serum and CSF was found in SMA patients (p > .05).

Comparisons of laboratory parameters
We compared laboratory parameters among different groups (Table 1).
The serum Cr levels were significantly higher in ALS patients than those in SMA patients (p < .001), while no significant difference in serum CK levels was found between ALS patients and SMA patients (p > .05). The serum NFL and pNFH levels in ALS patients were significantly higher than those in normal controls (p < .05, Figure 1A and B). However, there was no significant difference in serum NFL and pNFH levels between the SMA and controls, as well as between SMA and ALS patients ( Figure 1A and B). It should be noted that the value of pNFH was lower than LLOD in all the controls (n = 10) and most patients (ALS: n = 6; SMA: n = 4), and those values below the LLOD were approximated to the lowest concentration of detection (0.01 ng/mL).
For all CSF samples from patients and controls, WBC was within the normal range (0-8) and did not differ in groups (Table 1). CSF total protein was slightly higher in ALS patients than in SMA patients (p = .086).
Both CSF NFL and pNFH levels were significantly different between SMA patients and ALS patients (p < .01). The CSF levels of NFL in ALS patients were significantly higher than in SMA patients (p < .001), and similarly, the CSF pNFH levels in ALS patients were significantly higher than in SMA patients (p < .01).
Then, we tested whether these biomarkers could differentiate SMA patients from ALS patients using the ROC curves. Serum Cr showed a significant value of differentiating SMA patients from ALS patients and the cut-off value was > 44.5 µmol/L (Sensitivity = 90%, Specificity = 90%, AUC = 0.94, p < .001, Figure 1C), while serum CK did not ( Figure 1D). Serum NFL and pNFH also did not show significant value in the differential diagnosis of SMA and ALS patients. The AUC of CSF NFL was 1 with a cut-off value of < 1275 pg/mL in the differential diagnosis of two groups (Sensitivity = 100%, Specificity = 100%, p < .001, Figure 1E). We also observed that CSF pNFH < 0.395 ng/mL could differentiate SMA patients from ALS patients with 90% sensitivity and 80% specificity (AUC = 0.84, p = .01, Figure 1F).

DISCUSSION
Traditionally, SMA and ALS are two distinct motoneuron diseases with progressive symptoms and severe disability. With the emergence of gene therapies, SMA patients achieve promising improvements in motor function and lifespan (Panagiotou et al., 2022;Waldrop et al., 2020). Moreover, early diagnosis and treatment are critical to therapeutic efficacy in SMA patients (Kong et al., 2021). In clinical practice, however, it is often difficult to distinguish adult SMA patients and ALS patients with the predominant involvement of lower motor neurons by clinical symptoms or electromyography. This pilot study compared the levels of NFs between SMA and ALS patients, and found significantly lower CSF levels of NFL and pNFH in SMA patients than in ALS patients, indicating the potential diagnostic value of NFs in motoneuron diseases. The distinct differences in NFs levels may result from different pathology and patterns of motor neuron degeneration. The main pathogenesis mechanisms of SMA were reduced levels of the survival of motor neuron (SMN) protein resulting from mutation of the SMN1 gene, which was essential for the maintenance of cellular homeostasis in motor neurons (Lefebvre & Sarret, 2020). It has been suggested that SMN protein would be located in the dendrites and axons of motor neurons during development and played an essential role in correct axonal growth (Chaytow et al., 2018). Thus, SMN protein deficiency leads to disruption of normal cytoskeleton development resulting in shorter axons in SMA patients, which may partly explain the lower NFs levels in SMA patients than in ALS patients. Moreover, the timepoints and disease duration of neuromuscular disruption and motor neuron death were different in the two diseases (Comley et al., 2016;Nash et al., 2016). Autopsy and animal studies identified active degeneration of motor neuron axons that began during embryogenesis and kept progressing in infancy in SMA (Kong et al., 2021), which were different from ALS with symptoms that occurred later for some unknown reason.
Inconsistent with results from the CSF cohort, serum NFL and pNFH were not significantly different among SMA and ALS patients in this pilot study. Moreover, we found a significant association between serum and CSF levels of NFL in ALS patients, which is in line with previous reports (Benatar et al., 2020), but a similar association did not show in SMA patients. Wurster et al. (2020) (Nitz et al., 2021). The possible explanation for these controversial findings could be the different extents of blood-brain barrier dysfunction in SMA patients with various types and duration. It is also likely that the circulating clearance of NFL probably differs in different ages of patients and also differs from clearance of pNFH. Further studies are needed to identify the possible value of serum NFL and pNFH in diagnosis and distinguishing SMA from ALS.

F I G U R E 1
Comparisons of serum levels of NFL and pNFH, and receiver operating curves (ROC) for classifying SMA and ALS. Serum levels of NFL (a) and pNFH (b) were compared between SMA patients, ALS patients, and normal controls. ROC curves of Cr (c), CK (d), CSF NFL (e), and CSF pNFH (f) were generated for discriminating between SMA and ALS patients. *p < .05, **p < .01. CK, creatine kinase; Cr, creatinine; CSF, cerebrospinal fluid; NFL, neurofilament light chain; pNFH, phosphorylated neurofilament heavy chain.
Our preliminary results also showed that serum levels of pNFH were negatively associated with ALSFRS-R scores in ALS patients, suggesting that pNFH may be a biomarker for disease severity. However, no correlation between NFs and motor function was found in SMA patients. Contrary to our findings, previous studies (Darras et al., 2019;Wurster et al., 2020) reported higher serum pNFH and NFL levels in SMA patients with more severe motor function. Differences in the selection of study subjects and instruments for motor function measurement may contribute to inconsistent results. In the present study, we enrolled adult patients with SMA type 2-3 rather than infants/children patients, extending the knowledge of different pathological axonal degeneration and metabolism in older SMA patients.
The widely available biomarkers, CK and Cr, have been increasingly reported for diagnosis and prognostic prediction in neurodegenerative diseases (Ceccanti et al., 2020;Lombardi et al., 2019). In this pilot study, we also found significantly lower Cr in SMA patients than in ALS patients, and higher CK and Cr were strongly associated with higher ALSFRS-R scores. These findings are consistent with previous observations (Freigang et al., 2021;Milella et al., 2021), suggesting that CK and Cr are potential markers of disease severity in SMA. No association of NFs was found with CK or Cr in SMA patients, these findings indirectly corroborated the hypothesis that degeneration of motor neurons and muscle was independently involved in the pathogenesis of SMA (Habets et al., 2022;Kim et al., 2020). The underlying mechanism is still unknown, and further studies are needed to explore the possible