Studying serum neurofilament light chain levels as a potential new biomarker for small fiber neuropathy

Abstract Background and purpose Diagnosing small fiber neuropathies can be challenging. To address this issue, whether serum neurofilament light chain (sNfL) could serve as a potential biomarker of damage to epidermal Aδ‐ and C‐fibers was tested. Methods Serum NfL levels were assessed in 30 patients diagnosed with small fiber neuropathy and were compared to a control group of 19 healthy individuals. Electrophysiological studies, quantitative sensory testing and quantification of intraepidermal nerve fiber density after skin biopsy were performed in both the proximal and distal leg. Results Serum NfL levels were not increased in patients with small fiber neuropathy compared to healthy controls (9.1 ± 3.9 and 9.4 ± 3.8, p = 0.83) and did not correlate with intraepidermal nerve fiber density at the lateral calf or lateral thigh or with other parameters of small fiber impairment. Conclusion Serum NfL levels cannot serve as a biomarker for small fiber damage.

To overcome these diagnostic difficulties, but also in order to get an insight into the pathophysiology of SFN, it was investigated whether thin nerve fiber degeneration in SFN can increase serum neurofilament light chain (sNfL) levels, thereby making it a potential biomarker.
Neurofilaments confer structural stability and are present in dendrites and neuronal soma, as well as in axons, where their expression is particularly high [15][16][17].NfL is consistently released at lower levels from axons, probably in an age-dependent fashion, with elevated NfL release observed in older individuals [17].Importantly, NfL is also found to be correlated with neuronal damage in the cerebrospinal fluid (CSF) and serum of patients with conditions such as traumatic brain injury [18], neurodegenerative diseases like amyotrophic lateral sclerosis [19,20] and chronic inflammatory conditions like multiple sclerosis [21][22][23].Notably, there is a growing body of evidence indicating that blood NfL levels are elevated in polyneuropathies [24][25][26][27][28][29][30][31][32][33] and correlate with neuropathy acuteness and progression [26,32,[34][35][36].As there is damage/degeneration of small fiber axons [37][38][39][40] and neurofilaments are important components of the axon, it was hypothesized that NfL, which is broadly investigated and significantly increased in other polyneuropathies, could be a surrogate for fiber loss in SFN.Age-and sex-matched healthy controls (HCs) without any known acute (at the time of recruitment) or chronic disease and in particular absence of any signs of neuropathy were also investigated.Due to the positive correlation between age and sNfL [43], it was sought to create homogeneous age-matched study groups.Participants (both patients with SFN and HCs) with comorbidities that could affect sNfL (e.g., other neurological disorders, malignancies) were excluded.

Toronto Clinical Neuropathy Scale and symptom evaluation
The Toronto Clinical Neuropathy Scale (TCNS) [44], which assesses symptoms and objective sensory-motor signs, was used as a clinical tool to define neuropathy severity.A TCNS score from 0 to 5 represents no/very mild neuropathy, 6-8 mild neuropathy, 9-11 moderate neuropathy, and a score ≥12 represents severe neuropathy [44].Pain intensity between 1 and 3 on the 11-point numerical rating scale was regarded as mild, 4-6 as moderate and 7-10 as severe pain, in accordance with previous publications [45][46][47].

Serum NfL measurement
Serum NfL levels were measured using the same protocol described previously [22].In brief, whole blood was collected from all patients and controls in 7.5 mL S-Monovette® Serum Gel (Sarstedt, Germany).Blood was allowed to clot for about 5 min after sampling.
Next, samples were spun at 1400 g at room temperature for 10 min.
Directly after centrifugation, the serum was evenly transferred (1 mL/tube) to 1 mL polypropylene tubes and locally stored at −80°C.sNfL was measured in several rounds by SiMoA HD-1 (Quanterix, USA) using the NF-Light Advantage Kit (Quanterix) from the same batch according to the manufacturer's instructions.Resorufinβd- galactopyranoside was incubated at 33°C for 60 min prior to running the assay.Samples were measured in duplicate.The coefficient of variation (as a percentage) of each sample was obtained by dividing the standard deviation of both replicates by the mean of both replicates multiplied by 100.sNfL measurements were performed in a blinded fashion without information on clinical data.

Electrophysiological studies
All patients and controls underwent NCS.Ulnar and tibial motor nerve conduction velocity (NCV), compound muscle action potential and antidromic ulnar and sural sensory NCV and sensory nerve action potential (SNAP) were performed (right side of the body, or predominant side of symptoms) under controlled conditions using standard methods [48].

Quantitative sensory testing
Quantitative sensory testing (QST) was performed according to the established protocol of the German Research Network on Neuropathic Pain (DFNS) [49] at the lateral calf (test area) and ipsilateral cheek (control area).Thermal and mechanical detection and pain thresholds, paradoxical heat sensations with alternating thermal stimuli, dynamic mechanical allodynia, wind-up ratio for painful pinprick stimuli and the vibration disappearance threshold were determined.Since 5% of healthy individuals can present with at least one pathological test [49], determination of pathological QST required the presence of at least two pathological tests, with the exception of the presence of paradoxical heat sensations or dynamic mechanical allodynia, since they do not normally occur in healthy humans.Reference values are according to the DFNS [49].

Skin biopsies
According to the consensus paper by the European Federation of Neurological Societies [1], skin punch biopsies were obtained 10 cm proximal to lateral malleolus and 20 cm distal to spina iliaca with a disposable 6-mm punch biopsy after subcutaneous local anesthesia.
All skin samples were processed to estimate IENFD according to a previously published protocol [50].Skin biopsies were gathered and sent to the histology laboratory of the Department of Neurology, University of Würzburg, Germany.The IENFD was determined following standardized counting rules [1] by an investigator blinded to subject allocation.Results were then evaluated by an experienced clinician also blinded to the subject.Regarding QST data, raw data were transformed into z values as previously described [51] allowing comparison between different sexes and ages.For individual assessments, values are regarded as pathological if the individual results at the test area lie outside the 95% confidence interval (CI) of the age-adapted reference values [51].
All recordings from NCS of both patients and the control group were within the normal limits of age-controlled normative values.

Assessment of the small fibers and diagnosis of SFN
Quantitative sensory testing (QST) revealed predominant loss of thermal perception, that is, warm detection threshold (WDT), cold detection threshold (CDT) and thermal sensory limen (TSL) (Figure 1).Individually, 11/29 patients with SFN were below the normal limits (−1.96 standard deviations) of CDT, 8/29 of WDT and 8/29 of TSL.QST data from one patient are missing because of no-show at the QST test site.
All patients fulfilled the criteria for a definite SFN [41]; 17 had both pathological IENFD and abnormal QST thermal thresholds (Figure 3).

Serum NfL in patients with SFN
Serum NfL could be measured in all participants; levels did not differ (p = 0.83) between patients with SFN (9.1 ± 3.9) and HCs (9.4 ± 3.8).The distribution of individual results broadly overlaps (Figure 4).Although mean age was similar in the two groups, the correlation of age and sNfL was calculated separately for the two groups and they both revealed a strong positive Pearson correlation (for HCs p = 0.001, r = 0.70; for patients with SFN p < 0.001, r = 0.69).
In the final step of our analysis, whether there was a correlation between sNfL and tests for small fiber integrity or a sensitive parameter for large fiber impairment, namely sural NCS, was examined (Table 2).In the SFN group, a significant correlation was found between sNfL and sural SNAP (r = −0.43,p = 0.02) although

DISCUSS ION
Neurofilaments contribute to the growth and stability of axons in both central and peripheral neurons and maintain mitochondrial stability and microtubule content [56].NfL is expressed in axons and is shed into the peripheral blood following axonal injury in patients with a variety of neurological diseases in the central and peripheral nervous system [57].sNfL levels are known to be increased in peripheral sensorimotor neuropathies, particularly if they are significantly progressing [24][25][26][27][28][29][30][31][32][33].The availability of highly sensitive assays for NfL in serum samples [17] and the need for a quick and reproducible biomarker for nerve fiber damage particularly in clinically ambiguous cases motivated our measurements of sNfL levels in patients with pure SFN.
Our results indicate that sNfL levels of patients with SFN were not different from HCs.Furthermore, no correlation was found between any parameters reflecting small nerve fiber function and integrity and sNfL levels in our patients.That is, sNfL levels do not reflect damage to small diameter axons in SFN although ongoing axonal degeneration and regeneration might happen [37][38][39][40].sNfL are therefore not suited as a biomarker for SFN.Our negative results indicate that either the NfL content of small intraepidermal nerve fibers is too low [58] or the axonal damage in SFN is generally too subtle to detect altered NfL in serum samples.It has already been discussed before that CSF as well as blood NfL remains high for 2-3 months after a relapse and then drops to lower levels in multiple sclerosis patients [15,17] and that serum levels in chronic inflammatory demyelinating polyneuropathy significantly decrease after 1 month of treatment and in remission periods [32,34]; this could also affect the diagnostic accuracy of sNfL in patients with SFN, as there are no clear signs of relapses or the exact timing of fiber loss and degeneration.
Patients and HCs were homogeneous, especially in terms of sex and age.Age is an important positive predictor for sNfL [43,59].Our analysis confirmed this finding by presenting a strong correlation of sNfL with age in both study cohorts.Furthermore, an ultra-sensitive fourth-generation (single-molecule array) immunoassay was used that can reliably measure blood levels of sNfL and detect subtle longitudinal changes in disease and in healthy controls [57,60].All values from patients and controls were above the detection limit.
One strength of our study is that our patients adhered to the definition of "pure" SFN.Large fiber neuropathy could not be detected in any of the patients, and QST profiles were typical for patients with small fiber damage [12,61].However, the negative correlation of sural SNAP with sNfL in the patient group, even though SNAP was in the normal range, might indicate some preclinical impairment of large sensory fibers.Obviously, this impairment is too subtle to cause an absolute increase of sNfL as has been shown in more progressive sensory-motor neuropathies [25,62,63].Indirectly, this interpretation is supported by the lack of correlation between sNfL and these parameters in HCs.
It was observed that a reduction in IENFD from the distal or proximal leg biopsy site was not sufficient to increase sNfL.This is in contrast to a recent study showing increased phosphorylated heavy chain neurofilaments (NfH) in plasma samples of patients with diabetic SFN [63].In that study [63], patients had a significantly lower sural SNAP compared to their HCs which might be an indication for at least some sensory fiber damage leading to an increase in serum neurofilaments, as discussed in the previous paragraph.In contrast to those results [63], in our study sural SNAP did not differ between the two groups.NfL and NfH are equally increased in the serum and CSF of patients with amyotrophic lateral sclerosis [64] and are both equally stable for measurement [65], but fewer data are available on NfH [66], particularly when peripheral neuropathies are concerned.Moreover, the SiMoA method is a highly sensitive technique compared to the enzyme-linked immunosorbent assay or the electrochemiluminescence-based assay for CSF and serum samples [67][68][69] to sensorimotor neuropathy in due course, despite the clinical similarity.
The major limitation of our study is that the sample size is limited, precluding comparisons of different etiological SFN subgroups.Whilst the majority of our patients have idiopathic SFN, whose pathology might be confined to the epidermal nerve fibers, future investigations should include a sufficient number of subjects with SFN as a result of an ongoing (e.g., diabetes or thyroid dysfunction) or terminated systemic disease (e.g., drug-related), in order to further address the role of sNfL as biomarkers for peripheral neuropathies.

CON CLUS ION
Even if measured with a very sensitive assay, sNfL levels are not suited to objectify or to monitor loss of small epidermal nerve fibers in SFN.Future research has to clarify whether the amount of NfL in epidermal nerve fibers is too low, whether other filament proteins like the recently investigated peripherin [72] are better suited to monitor small nerve fiber damage, or whether the pathophysiology of SFN is not related to an ongoing axonal damage.

A
cross-sectional observational study was conducted and included patients with SFN referred to the Neurology Department at the University Medical Center of the Johannes Gutenberg University Mainz (Mainz, Germany), from March 2019 to December 2020.The study was conducted according to the Declaration of Helsinki and was approved by the Ethics Committee of the Rhineland-Palatinate Medical Association (837.437.17).A written informed consent was obtained from each participant.During their first visit, patients received a comprehensive examination by an experienced neurologist.Patients diagnosed with SFN of any etiology were included.The diagnosis SFN was made according to guidelines and landmark papers [1, 41].More specifically, only patients who met the Diabetic Neuropathy Study Group of the European Association for the Study of Diabetes (NEURODIAB) criteria [41, 42] for a definite SFN diagnosis, which means clinical evidence of small fiber damage, normal sural nerve conduction studies (NCS), abnormal thermal thresholds in QST at the foot and/or reduced IENFD at the ankle were included.
Statistical analyses were performed using IBM SPSS 23 Statistics, version 23.0, and GraphPad Prism 9 for Windows.The level of statistical significance was set at p < 0.05.All data were tested for normal distribution by the D'Agostino−Pearson test and by visual inspection of the distribution.Comparison of data between groups was performed with t tests.Pearson correlation analyses were performed to explore associations between sNfL levels and clinical and histological parameters, as well as neurophysiological data.Categorical data were analyzed with χ 2 tests.
/30 SFN patients had a pathological low distal IENFD but also 2/19 HCs presented with an asymptomatic low TA B L E 1 Summary of clinical characteristics and results of NCS of the study population.

1
Quantitative sensory testing of all patients.QST z scores of the SFN patients (mean/SD at the lateral calf).The zero line denotes the mean value of the control group, the dashed lines in each case 1.96 times the standard deviation from the mean value of the control group, i.e., the range in which the values of the control group lie within a 95% probability (95% confidence interval).
Pearson correlations between sNfL and small fiber parameters as well as sural nerve NCS.There were no significant correlations between sNfL and the other results of SFN assessment in the patient cohort (n = 30), but there was a strong negative correlation with sural nerve SNAP.Abbreviations: CDT, cold detection threshold; IENFD, intraepidermal nerve fiber density; MDT, mechanical detection threshold; MPT, mechanical pain threshold; NCS, nerve conduction studies; NCV, conduction velocity; SFN, small fiber neuropathy; SNAP, sensory nerve action potential; sNfL, serum neurofilament light chain; TSL, thermal sensory limen; VDT, vibration disappearance threshold; WDT, warm detection threshold.
[4,70,71] investigations are needed to explore whether sNfL could differentiate between SFN of different etiologies.It is yet speculative, but axonal damage (and sNfL) in idiopathic SFN, prevalent in 81.8% of our study cohort, which could remain stable over a long period[4,70,71], might be different from SFN in systemic diseases like diabetes leading TA B L E 2 *significant p value.