ARTICLE OPEN ACCESS

Serum neurofilament light chain is a usefulbiomarker in pediatric multiple sclerosisMarie-Christine Reinert MD Pascal Benkert PhD Jens Wuerfel MD Zuzanna Michalak PhD

Esther Ruberte PhD Christian Barro MD Peter Huppke MD Wiebke Stark MD Harald Kropshofer PhD

Davorka Tomic PhD David Leppert MD Jens Kuhle MD PhD Wolfgang Bruck MD and Jutta Gartner MD

Neurol Neuroimmunol Neuroinflamm 20207e749 doi101212NXI0000000000000749

Correspondence

Dr Reinert

marie-christinereinert

meduni-goettingende

AbstractObjectiveTo investigate serum neurofilament light chain (sNfL) as a potential biomarker for diseaseactivity and treatment response in pediatric patients with multiple sclerosis (MS)

MethodsIn this retrospective cohort study sNfL levels were measured in a pediatric MS cohort (n = 55follow-up 12ndash105 months) and in a non-neurologic pediatric control cohort (n = 301) usinga high-sensitivity single-molecule array assay Association of sNfL levels and treatment and clinicaland MRI parameters were calculated

ResultsUntreated patients had higher sNfL levels than controls (median 190 vs 46 pgmL CI [47326911]) p lt 0001) sNfL levels were significantly associated with MRI activity (+91 percontrast-enhancing lesion CI [1045 1138] p lt 0001 +06 per T2-weighted lesion CI [10011010] p = 0015) Higher values were associated with a relapse lt90 days ago (+511 CI [11841929] p lt 0001) and a higher Expanded Disability Status Scale score (CI [1001 1240] p =0048) In patients treated with interferon beta-1ab (n = 27) sNfL levels declined from 147 to 79pgmL after 6 plusmn 2 months (CI [0339 0603] p lt 0001) Patients with insufficient control ofclinical or MRI disease activity under treatment with interferon beta-1ab or glatiramer acetatewho switched to fingolimod (n = 18) showed a reduction of sNfL levels from 165 to 100 pgmL 6plusmn 2 months after switch (CI [0481 0701] p lt 0001)

ConclusionssNfL is a useful biomarker for monitoring disease activity and treatment response in pediatric MSIt is most likely helpful to predict disease severity and to guide treatment decisions in patients withpediatric MS This study provides Class III evidence that sNfL levels are associated with diseaseactivity in pediatric MS

From the Department of Pediatrics and Adolescent Medicine (M-CR PH WS JG) Division of Pediatric Neurology University Medical Centre Gottingen Georg August UniversityGottingen Germany Clinical Trial Unit (PB) Department of Clinical Research University Hospital Basel University of Basel Medical Image Analysis Centre Basel (MIAC AG) (JW ER)Department of Biomedical Engineering (JW) University Basel Neurologic Clinic and Policlinic (ZM CB DL JK) Departments of Medicine Biomedicine and Clinical ResearchUniversity Hospital Basel University of Basel Novartis Pharma AG (HK DT) Basel Switzerland and Institute of Neuropathology (WB) University Medical Centre Gottingen GeorgAugust University Gottingen Germany

Go to NeurologyorgNN for full disclosures Funding information is provided at the end of the article

The Article Processing Charge was funded by Georg August University Gottingen

This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 40 (CC BY-NC-ND) which permits downloadingand sharing the work provided it is properly cited The work cannot be changed in any way or used commercially without permission from the journal

Copyright copy 2020 The Author(s) Published by Wolters Kluwer Health Inc on behalf of the American Academy of Neurology 1

Pediatric multiple sclerosis (MS) is characterized by a higherrelapse rate but better clinical remission than adult-onsetMS12 Time to secondary progression in pediatric MS islonger but irreversible disability is reached on average at anage 10 years younger3

Chronic disability in patients withMS is assumed to mainly becaused by neuroaxonal damage correlating with functionalworsening and irreversible impairment45 MRI mainly detectsfocal lesions whereas axonal degeneration or involvement ofgray matter as major causes of permanent disability are onlypartially reflected6 Identification of diffuse brain parenchymaldamage subclinical disease activity and neuroaxonal injuryrequires additional new-generation biomarkers

Neurofilament light chain (NfL) has recently been shown to bea promising biomarker in numerous neurologic diseases inadults7ndash12 and children13ndash18 In adult-onset MS NfL is a markerof disease activity and severity with higher serum NfL (sNfL)levels associated with an increased MRI disease activity higherExpanded Disability Status Scale (EDSS) score and recentrelapses19ndash23 sNfL was shown as predictor of disease worseningand brain and spinal cord atrophy192425 and revealed to beprognostic for conversion from radiologically or clinically iso-lated syndrome to definite MS122627 Disease-modifying ther-apies (DMTs) led to sNfL reductions1928

To improve disease monitoring and treatment decisions inpediatric MS a biomarker reflecting subclinical disease activityand neuroaxonal damage is needed9 The aim of this study wasto investigate sNfL as potential biomarker for disease activityand treatment response in pediatric MS We hypothesized el-evated sNfL levels in children with MS compared with controlsand correlation with clinical parameters such as EDSS score andMRI We also hypothesized lower sNfL levels in pediatric thanin adult controls due to age dependency

MethodsResearch questionsWith this study we want to answer the following questions

1 Do pediatric patients with MS have higher sNfL levelsthan non-neurologic pediatric controls

2 Do sNfL levels in pediatric patients with MS correlatewith clinical disease activity

3 Do sNfL levels in pediatric patients with MS correlatewith MRI disease activity

4 Can sNfL levels in pediatric patients with MS be used tomonitor disease activity and treatment effects

Classification of evidence is Class III evidence

Patients and samplesWe analyzed a cohort of pediatric patients with MS (n = 55)and non-neurologic pediatric controls (n = 301) recruited inthe Department of Pediatrics and Adolescent MedicineUniversity Medical Centre Gottingen Germany

Patients with MS fulfilled the following inclusion criteria (1)confirmed diagnosis of MS according to the McDonald cri-teria 2017 (2) disease onset lt18 years and (3) retrospectiveclinical data and serum samples available for ge12 months offollow-up We defined 2 treatment cohorts

1 Interferon (IFN) group (n = 27) patients treated withIFN beta-1a or -1b during complete follow-up

2 Switching DMT group (n = 28) patients switched fromIFN glatiramer acetate (GA) natalizumab or dimethylfumarate to fingolimod during follow-up As a subgroupwe defined the fingolimod group (n = 18 treatmentswitch from IFNGA to fingolimod)

We collected serum samples at baseline (first contact in ourclinic) and follow-up visits (usually every 6 months and ad-ditional visits due to relapses) between May 2003 and March2018 and stored at minus20degC Study size was determined by thenumber of patients fulfilling the criteria of the switch groupand then completed by a comparable number of IFN patientsLoss of follow-up was due to reaching adulthood or switchingto another treatment center and was accepted if at least 3samples were available for this study

The patients of the control cohort attended the hospital be-cause of non-neurologic diseases between November 2017and May 2018 The blood sample was taken unrelated to thestudy Patients with severelife-threatening diseases or onmedication with chemotherapeutics were excluded Di-agnoses of the controls are shown in table e-1 (linkslwwcomNXIA251)

Standard protocol approvals registrationsand patient consentsPatients and parents or guardians of children younger than 18years provided written informed consent The study was ap-proved by the local ethics committee

Measurements of sNfL with Simoa technologyWe measured sNfL levels using the high-sensitivity single-molecule array (Simoa) NF-Light Advantage Kit (QuanterixLexington MA) according to the manufacturerrsquos instructionsWe analyzes all samples in duplicate within one assay Interassay

GlossaryCEL = contrast-enhancing lesion CV = coefficient of variation DMT = disease-modifying therapy EDSS = ExpandedDisability Status Scale GA = glatiramer acetate IFN = interferon sNfL = serum neurofilament light chain

2 Neurology Neuroimmunology amp Neuroinflammation | Volume 7 Number 4 | July 2020 NeurologyorgNN

coefficients of variation (CVs) for 3 native serum samples werebelow 10 The mean intra-assay CV of duplicate determi-nations for concentration was 50 We did repeat measure-ments for few samples with intra-assay CV above 20 Analyseswere performed blinded to clinical or MRI data

Clinical dataThe EDSS score was rated during the regular clinical follow-up visits by a pediatric neurologist but not during a relapseWe defined relapses as neurologic symptoms that could notbe explained otherwise for lasting at least 24 hours and withthe last relapse at least 30 days ago

Cerebral MRIIn the MS cohort we performed cerebral MRI at baseline andat follow-up visits We quantitatively analyzed MRIs in a spe-cialized imaging clinical research organization (Medical ImageAnalysis Centre Basel Switzerland) in a blinded mannerLesions weremarked and segmented in a standardized fashionusing Amira (Mercury Computer Systems Inc Chelmsford)by expert raters and subsequently confirmed by board certi-fied neuroradiologists We quantified contrast-enhancing le-sion (CEL) in all available cerebral MRIs and T2-weighted(T2w) lesions as a cumulative parameter only once a year foreach patient

Statistical analysisWe described continuous and ordinary variables by medianand interquartile range and categorical variables as counts andpercentages The EDSS score was not reported for some visitsdue to relapses and incomplete data and therefore imputed for51 of 366 visits by using the EDSS value of the previous visitwhen available (otherwise of the subsequent visit)

We did the analyses involving only baseline samples usingordinary linear regression models Thereby the dependentvariable NfL was log transformed In the longitudinal anal-yses we modeled the drop in NfL levels during the follow-upin the IFN and the switching DMT group using lineargeneralized estimating equation models with log(NfL) as thedependent variable To handle repeated measures withinpatients we clustered data points to account for within-subject correlation thereby assuming an exchangeable cor-relation structure With the combined data set using allpatients we investigated 2 separate models (1) a clinicalmodel with EDSS score age sex recent relapse and treat-ment status and (2) an MRI model with age number of T2wlesions and number of CELs (table e-2 linkslwwcomNXIA252) All estimates were back transformed to the originalscale and therefore represent multiplicative effects on thegeometric mean of sNfL

Data availabilityThe anonymized data can bemade available on a research basisInterested scientists can submit a request to the correspondingauthor Requests for access will be reviewed and a data accessagreement will be required

ResultsBaseline characteristicsBaseline demographics and patient characteristics are shown inthe table In the switching DMT group most patients (19 of 28679) switched to fingolimod due to ongoing clinical (5patients) orMRI disease activity (5 patients) or both (9 patients)under previous treatment with IFN GA or dimethyl fumarate

Eight patients (286) switched from natalizumab to fingoli-mod 7 due to increasing risk of progressive multifocal leu-koencephalopathy (antibodies against JC virus and natalizumabtreatment duration ge2 years) and 1 due to an allergic reaction tonatalizumab For sNfL analysis under treatment we focused onpatients switching from injectables (IFNGA) to fingolimod(fingolimod group n = 18)

sNfL in a pediatric control cohortIn the control group (n = 301 median age 101 years range 3monthsndash179 years 505 female) median sNfL was 51 pgmL (37 67) Age had a significant effect on sNfL levels (CI[0963 0985] p lt 0001) with higher sNfL levels in youngerchildren (figure 1) In controls covering the age range ofpatients with MS (n = 212 median age 125 years range65ndash179 years 528 female) median sNfL was 46 pgmL(35 60) In these children there was no significant effect ofage on sNfL levels (CI [0998 1026] p = 0092) Percentiles ofcontrols covering the age range of patients with MS are shownin figure 1 We only used these 212 patients and the percentilescalculated from their data for comparison with the patientswith MS

sNfL levels in patients with MS at baselineIn the MS cohort 43 of 55 patients (782) were untreated atbaseline Untreated patients had significantly higher sNfL val-ues at baseline than controls (figure 2) Twenty-six of them(605) had a relapse within the last 90 days associated withhigher sNfL levels (304 pgmL [133 686] vs 159 pgmL[121 247] CI [1089 3711] p = 0027)

Using a logistic regression model to predict future treatmentswitch in patients untreated and recently diagnosed (n = 39)with NfL as a predictor we found an OR of 2596 of need forhigher potent drugs (fingolimod and natalizumab) duringfollow-up in patients with sNfL above 99th percentile of con-trols at baseline (CI [0695 5614] p = 0024)

Association of sNfL and cerebral MRI lesions(MRI model)In the longitudinal analysis sNfL levels were strongly asso-ciated with the number of T2w lesions and with the number ofCELs (figure 3B table e-2 linkslwwcomNXIA252)

Association of sNfL levels and clinical aspects(clinical model)The clinical model (table e-2 linkslwwcomNXIA252 figuree-1 linkslwwcomNXIA250) showed associations of sNfL

NeurologyorgNN Neurology Neuroimmunology amp Neuroinflammation | Volume 7 Number 4 | July 2020 3

and clinical aspects The longitudinal analysis revealed a sig-nificant effect of a relapse le90 days ago on sNfL levels Duringfollow-up higher sNfL levels were associated with higher EDSSscores whereas treatment was associated with lower sNfLlevels There was also an association between sNfL and agewith higher levels in younger children

sNfL under disease-modifying treatmentwith IFN-βIn the IFN group (figure 4A) patients had elevated sNfLlevels at baseline with median sNfL gt99th percentile ofcontrols and sNfL gt90th percentile in 20 of 24 (83)patients sNfL levels decreased significantly already after 6 plusmn

Table 1 Baseline data and patient characteristics

Overall IFN group Switching DMT group

N 55 27 28

Age 149 (127 156) 149 (127 157) 143 (128 156)

Sex (femalemale) 3421 (6238) 1611 (5941) 1810 (6436)

EDSS score 00 (00 10) 00 (00 10) 00 (00 00)

Disease duration (mo) 00 (00 45) 00 (00 10) 20 (00 100)

Relapes lt90 d ago 31 (564) 16 (593) 15 (536)

Follow-up time (mo) 318 (242 489) 308 (264 429) 380 (233 550)

T2w lesion number available 35 (636) 20 (741) 16 (571)

No of T2w lesions

0ndash1 2 (56) 2 (100) 0 (00)

2ndash9 11 (333) 9 (450) 3 (188)

gt9 22 (611) 9 (450) 13 (812)

CEL lesion number available 39 (709) 22 (815) 17 (607)

No of CELs

0 16 (410) 10 (455) 6 (353)

1 5 (128) 4 (182) 1 (59)

2 6 (154) 3 (136 3 (176)

ge3 12 (308) 5 (227) 7 (412)

Treatment at baseline

Untreated 43 (782) 25 (926) 18 (643)

IFNGA 10 (182) 2 (74) 8 (286)

Nat 2 (36) mdash 2 (71)

Treatment sequenceduring follow-up

IFN 27 (491) 27 (100) mdash

IFNGAmdashFTY 18 (327) mdash 18 (643)a

IFNGAmdashDFmdashFTY 1 (18) mdash 1 (36)

IFNGAmdashNatmdashFTY 5 (91) mdash 5 (179)

FTYmdashNat 1 (18) mdash 1 (36)

NatmdashFTY 3 (55) mdash 3 (107)

Abbreviations CEL = contrast-enhancing lesion DF = dimethyl fumarate DMT = disease-modifying therapy EDSS = Expanded Disability Status Scale FTY =fingolimod GA = glatiramer acetate IFN = interferon Nat = natalizumabData are presented as median with interquartile range and as numbers with percent The switching DMT group includes patients with different treatmentsequences as presented in the tablea Marks fingolimod group (n = 18 patients who switched from IFNGA to fingolimod during follow-up)

4 Neurology Neuroimmunology amp Neuroinflammation | Volume 7 Number 4 | July 2020 NeurologyorgNN

2 months of treatment During follow-up under IFN treat-ment sNfL levels stayed decreased but did not reach valuesof controls ie patientsrsquomedian sNfL levels remained abovethe 80th percentile of controls up to 30 plusmn 2months of follow-up Figure 5 A and B are examples of 2 individual diseasecourses

sNfL during treatment switch to fingolimodIn the fingolimod group (figure 4B) patients had elevated sNfLlevels with a median sNfL gt99th percentile of controls andsNfL gt90th percentile in 94 of patients before switch fromIFNGA to fingolimod After treatment switch median sNfLlevels decreased significantly after 6 plusmn 2 months of fingolimod

Figure 1 sNfL in a control cohort

Data for sNfL vs age in neurologically healthycontrols (n = 301) Vertical lines denote the agerange covered by the MS cohort Specific percen-tiles were calculated from these samples withinthe MS age range (n = 212) and are shown ashorizontal lines with sNfL percentile values in pgmL A nonparametric smoothing line (loess) isshown in blue There is an age dependency of sNfLwith lower levels in younger children (CI [09630985] p lt 0001) that is not shown for the controlscovering the age range of the MS cohort (CI [09981026] p = 0092) Percentiles are not stratified bysex because there was no significant effect onsNfL sNfL = serum neurofilament light chain

Figure 2 Untreated pediatric patients with MS show elevated sNfL levels compared with controls

Untreated patients at baseline (n = 43) reveal elevated sNfL levels compared with controls of the same age range (n = 212 CI [4732 6911] p lt 0001) withamedian sNfL of 190 pgmL [117 438] vs 46 pgmL [35 60] Data are shown as boxplots withmedian and interquartile range sNfL = serumneurofilamentlight chain

NeurologyorgNN Neurology Neuroimmunology amp Neuroinflammation | Volume 7 Number 4 | July 2020 5

treatment but remained gt80th percentile of controls during 12months of follow-up Figure 5 C and D shows 2 individualdisease courses

There were no differences in the sNfL levels between patientswith (9 patients sNfL 165 pgmL [104 217]) or withouta relapse within 90 days before switch from IFNGA to

Figure 3 sNfL and MRI data

(A) sNfL levels correlate with the number of T2-weighted (T2w) lesions MRI data vs sNfL levels are shown stratified by the number of T2w lesions MostMRIs showmore than 9 lesions sNfL levels are strongly correlated with the number of T2w lesions with an average increase in sNfL of 06 per lesion(CI [1001 1010] p = 0015) (B) sNfL levels correlate with the number of contrast-enhancing lesions (CELs) MRI data vs sNfL levels are shownstratified by the number of CELs Most MRIs do not show CELs sNfL levels are strongly correlated with the number of CELs with an average increase insNfL of 91 per lesion (CI [1045 1138] p lt 0001) Data are shown as boxplots with median and interquartile range sNfL = serum neurofilamentlight chain

6 Neurology Neuroimmunology amp Neuroinflammation | Volume 7 Number 4 | July 2020 NeurologyorgNN

Figure 4 sNfL during DMT

(A) sNfL levels decrease under DMT with IFN In the IFN group patients with IFN treatment show elevated sNfL levels at baseline (median 147 pgmL [95430]) sNfL levels decreased significantly under DMT with IFN already after 6 plusmn 2months of treatment (79 pgmL CI [0339 0603] p lt 0001) but mediansNfL levels stays above the 80th percentile of controls during follow-up (B) sNfL levels decrease after switch from IFNGA to fingolimod Patients switchingfrom IFN or GA to fingolimod during follow-up (fingolimod group) mostly had sNfL levels above the 99th percentile of controls before treatment switchAfter treatment switch sNfL levels decreased significantly from 165 pgmL (128 267) to 10 pgmL (95 430) after 6 plusmn 2months of fingolimod treatment(CI [0481 0701] p lt 0001) but stayed above the 80th percentile of controls during follow-up All data are shown as boxplots with median andinterquartile range Percentiles of controls aremarked by lines DMT = disease-modifying therapy GA = glatiramer acetate IFN = interferon sNfL = serumneurofilament light chain

NeurologyorgNN Neurology Neuroimmunology amp Neuroinflammation | Volume 7 Number 4 | July 2020 7

Figure 5 Individual disease courses demonstrate potential prognostic value of sNfL as a biomarker

(A) Patient 61 IFN group Diagnosed at age 153years treatment with IFN follow-up for 30months No relapses and only 1 CEL weredetected after 12 months sNfL levels neverdroppedbelow the 90th percentile and only once(after 12 months) below the 99th percentile (B)Patient 78 IFN group Diagnosed at age 75 yearsIFN treatment follow-up for 105 months Thepatient had 1 relapse 6 months after treatmentstart and after that no clinical or MRI diseaseactivity anymore sNfL levels were increased be-fore treatment start and dropped under DMTlevels at follow-up were always below the 80thpercentile of controls (C) Patient 14 fingolimodgroup Diagnosed at age 72 years initial IFNtreatment for 6 years with a relapse rate of 05per year and in the first 3 years each 1 CEL at 3time points sNfL levels were always above the90th percentile apart from 1 measurement after35 months After treatment switch to fingolimodafter 66 months due to ongoing clinical and MRIdisease activity there was no relapse or cranialCEL during 12 months of follow-up sNfL levelsdropped below the 90th percentile after 6months and below the 80th percentile after 12months of fingolimod treatment (D) Patient 33fingolimod group Diagnosed at age 143 yearsinitial IFN treatment At age 156 years treatmentswitch to fingolimod due to ongoing clinical andMRI disease activity Follow-up under treatmentwith fingolimod was 22 months without any re-lapse or CEL detection sNfL levels were elevatedat the time point of switch and decreased underfingolimod treatment levels below the 90thpercentile were reached 22 months later sNfLlevels are shown as (x) connected with a brokenline EDSS levels are shown as blue numbersnumbers of CELs are marked with red asterix ()green lines show the 50th and 90th percentiles ofcontrols and red arrows (uarr) mark relapses CEL =contrast-enhancing lesion DMT = disease-mod-ifying therapy EDSS = Expanded Disability StatusScale sNfL = serum neurofilament light chain

8 Neurology Neuroimmunology amp Neuroinflammation | Volume 7 Number 4 | July 2020 NeurologyorgNN

fingolimod (8 patients 174 pgmL [131 294] CI [03652080] p = 0738)

sNfL in patientswithMSwithout clinical orMRIdisease activityThere were 191 samples from patients withMSwith noCEL incranial MRI at the time of sampling and no recent relapsewithin 90 days before sampling These patients without clinicalor MRI disease activity had a median sNfL level of 72 pgmL[55 119] with a range from 30 to 655 pgmL Hence thereare patients without clinical orMRI disease activity but elevatedsNfL levels (figure 5A)

DiscussionIn adult-onset MS and other neuroinflammatory and neuro-degenerative diseases sNfL appears to be a promising bio-marker for disease activity and disability prognosis9 In thisstudy we showed that sNfL may be a useful biomarker fordisease activity and treatment monitoring in pediatric MS

In healthy children we revealed lower sNfL levels than those inadult cohorts described in the literature19252829 It has beendemonstrated that NfL levels in healthy adults are age de-pendent with an annual increase in sNfL of 22192528 In ourstudy we also showed an age dependency but with higher sNfLlevels in younger healthy children (figure 1) This could reflectcell migration and cell differentiation including neuronalremodeling processes in the developing brain as recently shownfor neurofilament heavy chain in infants30 Furthermore a cor-relation with the development of the blood-brain barrier andCSF flow rate is conceivable The albumin quotient a well-known CSF diagnostic marker shows a similar dynamic withhigh levels after birth a decrease in childhood and increasinglevels in older individuals3132 In the age range covering the MScohort there were no age-dependent differences in the controlsFurther analysis involving young adults will identify the age atwhich the age-associated increase in NfL levels may begin

Because of high comparability offered by studies with the NF-Light Advantage Kit (Quanterix) our percentiles for sNfL inpediatric controls (figure 1) are usable for future sNfL inves-tigations not only in pediatric MS but also for other childhood-onset neuroinflammatory and neurodegenerative diseases

We showed associations of sNfL levels with MRI and clinicaldisease activity (figures 2 and 3 and figure e-1 linkslwwcomNXIA250) However because of low EDSS values in pediatricMS in general the EDSS data need careful interpretation andthe scoremay only roughly reflect the clinical status Alternativescores or methods to record clinical status in pediatric MS mayimprove future studies

Whereas we did not see an age dependency in the controlscovering the MS age range in the MS cohort we observed anassociation between sNfL and age with higher sNfL levels in

younger patients (figure e-1 linkslwwcomNXIA250) Thisfinding is consistent with histologic observations for pediatricMS lesions where the amount of acutely damaged axons in-versely correlated with the patientsrsquo age33 In addition earlierstudies on CSF biomarkers showed age correlations of NfLwith highest levels in younger children with neurologic dis-eases16 On the other hand phenomena such as regression tothe mean or function of treatment effect could be involvedhere For investigating whether there is higher disease activityin children with very early MS diagnosis (age lt10 years) ingeneral a larger cohort has to be analyzed

In clinical settings treatment decisions and the questionwhether the effect of a DMT is sufficient are one of the greatchallenges In pediatric MS IFN and GA continue to be thestandard first-line immunomodulatory treatments34 Accordingto the high level of inflammation and due to the large pro-portion of gt40 of children with highly active MS these first-line therapies are often not sufficient35 However treatmentdecisions especially switching to higher potent second-linedrugs are complicated by the fact that most of the new and highpotent immunomodulatory drugs are not approved for chil-dren and long-term efficacy and safety data are missing Nev-ertheless the introduction of higher-efficacy drugs such asfingolimod or natalizumab has improved the clinical course ofpediatric patients with highly active MS35ndash37 Decision criteriapreferably based on signs of current disease activity andexpected disease course will help to weigh the benefits of morepotent therapy and the risks of potential side effects for in-dividual patients and improve individualized treatment Ourstudy shows that sNfL levels decrease significantly under DMT(figure 4A) Because an untreated pediatricMS control group ismissing and cannot be investigated for ethical reasons wecannot exclude the possibility that sNfL levels will decreaseover time even in untreated patients In patients with ongoingclinical or MRI disease activity under DMT with IFNGAreflected by elevated sNfL values switching to a more potentdrug (from IFNGA to fingolimod) led to a significant declinein sNfL (figure 4B) The same was already shown for adultpatients with MS28 In addition our data show a tendencytoward a later escalation of therapy with higher NfL values atthe time of diagnosis

For an additional benefit over MRI and clinical evaluationalone a serum biomarker should detect subclinical disease ac-tivity We found 3 aspects supporting that sNfL is able to do so(1) Patients under DMT did not reach sNfL levels observed incontrols even when effective clinical and MRI disease controlwas observed In the IFN group with satisfying clinical andMRIdisease control median sNfL levels stayed above the 80thpercentile during 30 months of follow-up (figure 4A) Weshowed the same for patients in the fingolimod group up to 12months after switch from IFNGA to fingolimod (figure 4B)with the limitation that long-term data are missing for thisgroup (2) In patients switching from IFNGA to fingolimodsNfL levels at the time point of switch were elevated but notsignificantly influenced by a relapse less than 90 days ago (3) In

NeurologyorgNN Neurology Neuroimmunology amp Neuroinflammation | Volume 7 Number 4 | July 2020 9

patients with no acute clinical or MRI disease activity themedian sNfL level was above the 80th percentile of controlswith a range from 30 to 655 pgmL These results underlinethat there is subclinical disease activity with ongoing neuro-axonal damage reflected by sNfL measurements

Nevertheless it is possible to normalize sNfL values underDMT Figure 5 describes individual disease courses and givesan idea how sNfL could be used in clinical practice It ispossible to detect treatment responder (figure 5 B and D)and nonresponder (figure 5C) and especially patients withsubclinical disease activity (figure 5A)

One limitation of our study is the retrospective design andtherefore a missing standardized treatment and follow-up Inaddition as a national center for pediatric MS we probablysee a disproportionally high number of patients with moresevere disease courses potentially leading to a bias exagger-ating differences between controls and MS populationMoreover cohorts for highly active MS treated with natali-zumab rituximab alemtuzumab and others are not or onlyinsufficiently included and should be analyzed in furtherstudies To investigate the potential long-term prediction ofsNfL for disease course brain atrophy cognition and EDSSworsening as shown for adult patients1921242538 long-termstudies covering the transition from pediatric to adult medicalcare are needed

Recently van der Vuurst de Vries et al27 showed that CSFNfL is a promising predictive marker for disease course inchildren with CIS and later MS diagnosis In addition to thesefindings the results of this study highly suggest that sNfL isalso a useful biomarker for monitoring disease activity andtreatment response in pediatric MS Access via blood samplesmade possible by the Simoa technology3940 with high cor-relations to CSF measurements19284142 is an important steptoward everyday clinical practice implementation especiallyin the pediatric setting sNfL has the potential to guidetreatment decisions to an individualized treatment regimeespecially in patients with highly active disease course and thenecessity of change in therapy due to ongoing or recurringdisease activity A treatment goal of reaching sNfL levels egbelow the 90th percentile could be a possible strategy forfuture individualized treatment decisions yet the clinical rel-evance of a certain threshold should first be evaluated in long-term studies In addition in the case of particularly high sNfLvalues at disease onset this biomarker might be the basis todirectly start a highly potent immunomodulatory therapy

AcknowledgmentThe authors thank Ellen Kramer for excellent technicalassistance

Study fundingThis study was financially supported by Novartis Pharma AGBasel Switzerland The authors acknowledge support by theOpen Access Publication Funds of the Gottingen University

DisclosureH Kropshofer and D Tomic are full-time employees andstock holder of Novartis Pharma AG whichmanufactures oneof the drugs investigated in the study D Leppert has been anemployee of Novartis until January 2019 M-C Reinertreports grants from Novartis Pharma AG during the conductof the study J Gartner reports grants from Novartis PharmaAG during the conduct of the study and personal fees fromBayer Vital Biogen and Novartis outside the submittedwork W Bruck reports grants and personal fees fromNovartis during the conduct of the study and personal feesfrom Bayer Vital Merck Serono and Rewind grants andpersonal fees from Biogen Teva Pharma Sanofi-GenzymeandNovartis and grants fromMedDay outside the submittedwork C Barro reports conference travel grant from Novartisoutside the submitted work J Kuhle reports grants fromBiogen Novartis Roche Teva the Swiss MS Society Gen-zyme the University of Basel Swiss National ResearchFoundation Bayer Vital Merck and Celgene outside thesubmitted work Peter Huppke reports personal fees fromNovartis Bayer Vital and Merck Serono outside the sub-mitted work P Benkert J Wuerfel Z Michalak E RuberteandW Stark report no disclosures Go to NeurologyorgNNfor full disclosures

Publication historyReceived by Neurology Neuroimmunology amp NeuroinflammationAugust 20 2019 Accepted in final form April 9 2020

Appendix Authors

Name Location Contribution

Marie-ChristineReinert MD

University MedicalCentre GottingenGermany

Designed the study majorrole in the acquisition ofdata interpreted the dataand drafted the manuscriptfor intellectual content

PascalBenkertPhD

University of BaselSwitzerland

Analyzed the dataand revised the manuscriptfor intellectual content

JensWuerfel MD

Medical Image AnalysisCentre Basel (MIAC AG)Basel Switzerland

Analyzed and interpretedthe data and revised themanuscript for intellectualcontent

ZuzannaMichalakPhD

University of BaselSwitzerland

Analyzed the data

EstherRubertePhD

Medical Image AnalysisCentre Basel (MIAC AG)Basel Switzerland

Analyzed the data

ChristianBarro MD

University of BaselSwitzerland

Analyzed the data

PeterHuppke MD

University MedicalCentre GottingenGermany

Major role in the acquisitionof data and revised themanuscript for intellectualcontent

WiebkeStark MD

University MedicalCentre GottingenGermany

Major role in the acquisitionof data

10 Neurology Neuroimmunology amp Neuroinflammation | Volume 7 Number 4 | July 2020 NeurologyorgNN

References1 GormanMP Healy BC Polgar-Turcsanyi M Chitnis T Increased relapse rate in pediatric-

onset compared with adult-onset multiple sclerosis Arch Neurol 20096654ndash592 Huppke P Gartner J A practical guide to pediatric multiple sclerosis Neuropediatrics

201041157ndash1623 Renoux C Vukusic S Mikaeloff Y et al Natural history of multiple sclerosis with

childhood onset N Engl J Med 20073562603ndash26134 Singh S Dallenga T Winkler A et al Relationship of acute axonal damage Wallerian

degeneration and clinical disability in multiple sclerosis J Neuroinflammation 20171457

5 Compston A Coles A Multiple sclerosis Lancet 20083721502ndash15176 Petrova N Carassiti D Altmann DR Baker D Schmierer K Axonal loss in the

multiple sclerosis spinal cord revisited Brain Pathol 201828334ndash3487 Wang H Wang C Qiu W Lu Z Hu X Wang K Cerebrospinal fluid light and heavy

neurofilaments in neuromyelitis optica Neurochem Int 201363805ndash8088 Mattsson N Andreasson U Zetterberg H Blennow K for the Alzheimerrsquos Disease

Neuroimaging I Association of plasma neurofilament light with neurodegeneration inpatients with alzheimer disease JAMA Neurol 201774557ndash566

9 Weston PSJ Poole T Ryan NS et al Serum neurofilament light in familial Alzheimerdisease A marker of early neurodegeneration Neurology 2017892167ndash2175

10 Meeter LH Dopper EG Jiskoot LC et al Neurofilament light chain a biomarker forgenetic frontotemporal dementia Ann Clin Transl Neurol 20163623ndash636

11 Kuhle J Gaiottino J Leppert D et al Serum neurofilament light chain is a biomarkerof human spinal cord injury severity and outcome J Neurol Neurosurg Psychiatry201586273ndash279

12 Gaetani L Blennow K Calabresi P Di Filippo M Parnetti L Zetterberg H Neuro-filament light chain as a biomarker in neurological disorders J Neurol NeurosurgPsychiatry 201990870ndash881

13 Toorell H Zetterberg H Blennow K Savman K Hagberg H Increase of neuronalinjury markers Tau and neurofilament light proteins in umbilical blood after intra-partum asphyxia J Maternal-Fetal Neonatal Med 2018312468ndash2472

14 Shah DK Ponnusamy V Evanson J et al Raised plasma neurofilament light proteinlevels are associated with abnormal MRI outcomes in newborns undergoing thera-peutic hypothermia Front Neurol 2018986

15 Pranzatelli MR Tate ED McGee NR Verhulst SJ CSF neurofilament light chain iselevated in OMS (decreasing with immunotherapy) and other pediatric neuro-inflammatory disorders J Neuroimmunology 201426675ndash81

16 Shahim P Darin N Andreasson U et al Cerebrospinal fluid brain injury biomarkersin children a multicenter study Pediatr Neurol 20134931ndash39e32

17 Boesen MS Jensen PEH Magyari M et al Increased cerebrospinal fluid chitinase3-like 1 and neurofilament light chain in pediatric acquired demyelinating syndromesMult Scler Relat Disord 201824175ndash183

18 Kristjansdottir R Uvebrant P Rosengren L Glial fibrillary acidic protein and neu-rofilament in children with cerebral white matter abnormalities Neuropediatrics200132307ndash312

19 Disanto G Barro C Benkert P et al Serum neurofilament light a biomarker ofneuronal damage in multiple sclerosis Ann Neurol 201781857ndash870

20 Varhaug KN Barro C Bjornevik K et al Neurofilament light chain predicts diseaseactivity in relapsing-remitting MS Neurol Neuroimmunol Neuroinflamm 20185e422 doi101212NXI0000000000000422

21 Kuhle J Nourbakhsh B Grant D et al Serum neurofilament is associated withprogression of brain atrophy and disability in early MS Neurology 201788826ndash831

22 Khalil M Teunissen CE Otto M et al Neurofilaments as biomarkers in neurologicaldisorders Nat Rev Neurol 201814577ndash589

23 Kuhle J KropshoferHHaeringDA et al Blood neurofilament light chain as a biomarkerof MS disease activity and treatment response Neurology 201992e1007ndashe1015

24 Siller N Kuhle J Muthuraman M et al Serum neurofilament light chain is a bio-marker of acute and chronic neuronal damage in early multiple sclerosis Mult Scler(Houndmills Basingstoke England) 201925678ndash686

25 Barro C Benkert P Disanto G et al Serum neurofilament as a predictor of diseaseworsening and brain and spinal cord atrophy in multiple sclerosis Brain 20181412382ndash2391

26 Matute-Blanch C Villar LM Alvarez-Cermeno JC et al Neurofilament light chainand oligoclonal bands are prognostic biomarkers in radiologically isolated syndromeBrain 20181411085ndash1093

27 van der Vuurst de Vries RMWong YYMMescheriakova JY et al High neurofilamentlevels are associated with clinically definite multiple sclerosis in children and adultswith clinically isolated syndrome Mult Scler (Houndmills Basingstoke England)201925958ndash967

28 Piehl F Kockum I Khademi M et al Plasma neurofilament light chain levels inpatients with MS switching from injectable therapies to fingolimod Mult Scler(Houndmills Basingstoke England) 2018241046ndash1054

29 Hyun J-W Kim Y Kim G Kim S-H Kim HJ Longitudinal analysis of serum neu-rofilament light chain a potential therapeutic monitoring biomarker for multiplesclerosis Mult Scler J 2019 Epub 2019 Mar 26

30 Darras BT Crawford TO Finkel RS et al Neurofilament as a potential biomarker forspinal muscular atrophy Ann Clin Transl Neurol 20196932ndash944

31 Statz A Felgenhauer K Development of the blood-CSF barrier Dev Med ChildNeurol 198325152ndash161

32 Reiber H Proteins in cerebrospinal fluid and blood barriers CSF flow rate andsource-related dynamics Restor Neurol Neurosci 20032179ndash96

33 Pfeifenbring S Bunyan RF Metz I et al Extensive acute axonal damage in pediatricmultiple sclerosis lesions Ann Neurol 201577655ndash667

34 Ghezzi A Amato MP Makhani N Shreiner T Gartner J Tenembaum S Pediatricmultiple sclerosis conventional first-line treatment and general management Neu-rology 201687S97ndashs102

35 Huppke P Huppke B Ellenberger D et al Therapy of highly active pediatric multiplesclerosis Mult Scler (Houndmills Basingstoke England) 20192572ndash80

36 Chitnis T Arnold DL Banwell B et al Trial of fingolimod versus interferon beta-1a inpediatric multiple sclerosis N Engl J Med 20183791017ndash1027

37 Gartner J Chitnis T Ghezzi A et al Relapse rate and MRI activity in young adultpatients with multiple sclerosis a post hoc analysis of phase 3 fingolimod trials MultScler J Exp Transl Clin 201842055217318778610

38 Chitnis T Gonzalez C Healy BC et al Neurofilament light chain serum levelscorrelate with 10-year MRI outcomes in multiple sclerosis Ann Clin Transl Neurol201851478ndash1491

39 Rissin DM Kan CW Campbell TG et al Single-molecule enzyme-linked immuno-sorbent assay detects serum proteins at subfemtomolar concentrations Nat Bio-technol 201028595ndash599

40 Kuhle J Barro C Andreasson U et al Comparison of three analytical platforms forquantification of the neurofilament light chain in blood samples ELISA electro-chemiluminescence immunoassay and SimoaClin ChemLabMed 2016541655ndash1661

41 Gaiottino J Norgren N Dobson R et al Increased neurofilament light chain bloodlevels in neurodegenerative neurological diseases PLoS One 20138e75091

42 Kuhle J Barro C Disanto G et al Serum neurofilament light chain in early relapsingremittingMS is increased and correlates withCSF levels andwithMRImeasures of diseaseseverity Mult Scler (Houndmills Basingstoke England) 2016221550ndash1559

Appendix (continued)

Name Location Contribution

HaraldKropshoferPhD

Novartis Pharma AGBasel Switzerland

Conceptualization of thestudy interpreted the dataand revised the manuscriptfor intellectual content

DavorkaTomic PhD

Novartis Pharma AGBasel Switzerland

Interpreted the data andrevised the manuscript forintellectual content

DavidLeppert MD

University of BaselSwitzerland

Conceptualization of thestudy interpreted the dataand revised the manuscriptfor intellectual content

Jens KuhleMD PhD

University of BaselSwitzerland

Conceptualization of thestudy analyzed andinterpreted the data andrevised the manuscript forintellectual content

WolfgangBruck MD

University MedicalCentre GottingenGermany

Conceptualization of thestudy interpreted the dataand revised the manuscriptfor intellectual content

JuttaGartner MD

University MedicalCentre GottingenGermany

Design andconceptualization of thestudy interpreted the dataand revised the manuscriptfor intellectual content

NeurologyorgNN Neurology Neuroimmunology amp Neuroinflammation | Volume 7 Number 4 | July 2020 11

DOI 101212NXI000000000000074920207 Neurol Neuroimmunol Neuroinflamm

Marie-Christine Reinert Pascal Benkert Jens Wuerfel et al Serum neurofilament light chain is a useful biomarker in pediatric multiple sclerosis

This information is current as of May 13 2020

ServicesUpdated Information amp

httpnnneurologyorgcontent74e749fullhtmlincluding high resolution figures can be found at

References httpnnneurologyorgcontent74e749fullhtmlref-list-1

This article cites 41 articles 2 of which you can access for free at

Subspecialty Collections

httpnnneurologyorgcgicollectionmultiple_sclerosisMultiple sclerosis

httpnnneurologyorgcgicollectionall_pediatricAll Pediatricfollowing collection(s) This article along with others on similar topics appears in the

Permissions amp Licensing

httpnnneurologyorgmiscaboutxhtmlpermissionsits entirety can be found online atInformation about reproducing this article in parts (figurestables) or in

Reprints

httpnnneurologyorgmiscaddirxhtmlreprintsusInformation about ordering reprints can be found online

Academy of Neurology All rights reserved Online ISSN 2332-7812Copyright copy 2020 The Author(s) Published by Wolters Kluwer Health Inc on behalf of the AmericanPublished since April 2014 it is an open-access online-only continuous publication journal Copyright

is an official journal of the American Academy of NeurologyNeurol Neuroimmunol Neuroinflamm