Serum Neurofilament Light Chain in the Prognosis of Disease Activity in Relapsing Multiple Sclerosis


Relapsing multiple sclerosis (RMS) is characterized by recurrent episodes of inflammatory demyelination in the brain and spinal cord. Despite recommendations to consider disease activity and severity levels when determining the right treatment strategy, no consensus exists on how to classify patients with RMS into high- or low-risk groups to prioritize certain treatment options. Given that neurofilament light chain (NfL) may represent a prognostic and monitoring biomarker in MS that may be capable of assessing disease activity, a study was conducted to confirm the prognostic value of serum NfL (sNfL) for on-study disease activity and worsening in individuals with RMS.

This Journal Club discussion will focus on a pooled analysis of 2 phase 3 trials by Ziemssen and colleagues that evaluated and sought to confirm the prognostic capability of baseline sNfL for on-study disease activity and disease worsening in individuals with RMS who were randomly assigned to receive either oral teriflunomide 14 mg once daily or subcutaneous ofatumumab 20 mg every 4 weeks for up to 30 months.1 A panel of neurology experts from the University of Texas Southwestern Medical Center in Dallas, led by Kyle Blackburn, MD, reviewed the results of the study and evaluated the implications on clinical practice.


The study cohort included 1882 patients with RMS from the phase 3 ASCLEPIOS 1/2 trials ( Identifiers: NCT02792218; NCT02792231), of whom 1746 had available baseline sNfL samples. The median baseline sNfL level was 9.3 pg/mL. The researchers stratified patients into ”high” (sNfL >9.3 pg/mL) and “low” (sNfL ≤9.3 pg/mL) classifications to prognosticate the annualized rate of new and enlarging T2 (neT2) lesions in years 1 and 2, the annualized relapse rate, the annual percentage change in whole brain and regional brain volume, and disability outcomes. The investigators performed similar analyses with patients who had been diagnosed within the last 3 years and were treatment naive at baseline.


A total of 870 patients (49.8%) were classified into the high baseline sNfL group, while 876 patients (50.2%) were classified into the low baseline sNfL group. The baseline high sNfL vs low sNfL was deemed prognostic of increased on-study T2 lesion formation. At year 1, the relative increase was 158% in the ofatumumab treatment group (P <.001) and 69% in the teriflunomide treatment group (P <.001). This finding persisted into year 2, with a 65% increase in the ofatumumab treatment group (P =.124) and a 46% increase in the teriflunomide treatment group (P =.003).

Among patients with high sNfL who received ofatumumab, the rate of brain volume loss between years 1 and 2 was significantly lower than that observed in patients who received teriflunomide, yielding a relative reduction of 24.5% (-0.32 vs -0.43; P =.019). The effect of treatment with ofatumumab by baseline sNfL category was consistent for reduction in thalamic volume loss (relative reduction, 40.6%; -0.56 vs -0.94; P =.003). A single assessment of sNfL at baseline was not deemed prognostic for on-study relapses or worsening of disability. The researchers reported that these findings were similar in the subgroup of patients who were recently diagnosed and treatment-naive.


Kyle Blackburn, MD

Natalia Gonzalez Caldito, MD

Victor Salinas, MD, PhD

Kyle Blackburn, MD, assistant professor; Natalia Gonzales Caldito, MD, fourth-year resident; and Victor Salinas, MD, PhD, fourth-year resident, Department of Neurology, University of Texas Southwestern Medical Center, Dallas participated in this journal club discussion.

Kyle Blackburn, MD: We selected this article for discussion because recent research has shown that NfL may represent a prognostic and monitoring biomarker in MS that could help assess disease activity.

Natalia, will you speak briefly about NfL?

Natalia Gonzales Caldito, MD: Yes, NfL is emerging as a key biomarker for MS. In this article, the researchers studied NfL as a biomarker for disease activity and also in estimating a patient’s response to treatment. High NfL levels have been found to correlate with brain volume loss so measuring sNfL levels may help identify patients at higher risk for brain damage.

KB: Right, NfL is a measure of neuro-axonal damage. Several studies in MS have found that high NfL levels seems to correlate with certain measures of disease activity. Typically, higher levels of NfL occur around the time of a relapse or development of new gadolinium-enhancing (Gd+) lesions. A decrease in NfL levels is observed in patients who are undergoing treatment for MS. I think people are really excited about this marker as a potential adjunct to the traditional ways in which we monitor disease activity in MS.

ASCLEPIOS 1/2 were 2 randomized trials assessing the efficacy of ofatumumab, which is one of the newer therapies for MS, vs teriflunomide. In these trials, they found that ofatumumab resulted in a significant reduction in the relapse rate and in the development of neT2 lesions and Gd+ lesions. This ultimately led the US Food and Drug Administration (FDA) to approve ofatumumab for the treatment of RMS. Dr Salinas, can you review the analysis of the trials?

Victor Salinas, MD, PhD: Yes, patients pooled from these 2 trials were stratified on the basis of their sNfL levels using a cutoff that was actually derived from the distribution of all sNfL levels across the patients, taking the median value at baseline to designate patients into a high sNfL group and low sNfL group. The study sought to determine what kind of prognostic information is contained in these baseline sNfL measurement groups. The researchers went on to compare aspects of these 2 cohorts, namely the appearance of neT2 lesions at various time points, relapse rates, disability progression, and other radiologic findings, including brain atrophy and localized brain atrophy.

KB: It is an interesting way to look at things. They were trying to assess the prognostic value of sNfL levels. They split the patients into 2 groups: approximately 50% above a certain threshold and 50% below it based upon that median sNfL level in the study. They then looked at the outcomes of the trial, eventually comparing the treatment outcomes between ofatumumab and teriflunomide. In clinical practice, we often will have a patient on an early disease course and we have to try to apply some logic or some measure like T2 lesion formation to decide whether the patient needs an aggressive treatment plan or just monitoring. It is really intriguing to think about having another data point like NfL to help us make the decision about how aggressive we should be in treating patients. Natalia, do you want to dive into the findings from the trial?

NGC: In addition to the 2 groups, there also was a subgroup of newly diagnosed patients who had no Gd+ T1 lesions at baseline. Based on the sNfL levels, the researchers investigated if there were any difference in disease progression outcomes. But, in the subgroup, the researchers also found that the higher the baseline sNfL, the greater the risk for neT2 lesions. I thought that was very interesting. I know with newly diagnosed patients, there is always a question as to which patient is going to do worse and have a more severe disease course down the line. With this new biomarker, we could potentially have more information to answer this question.

This trial also analyzed sNfL levels post treatment. It was expected that post treatment, when there is higher sNfL, there is a higher percentage of patients having more lesions within the next month. However, the reason behind the statistically significant difference in duration could not be explained. The investigators thought that perhaps it could be because of the shorter time and the sensitivity. It is also possible the Expanded Disability Status Scale (EDSS) is not sensitive enough for short-term changes and smaller changes that are harder to pick up with the EDSS.

Considering the change in brain volume post-treatment, we know that the most pronounced was thalamic atrophy. Atrophy rates in thalamic volumes were the most pronounced in the higher sNfL group, regardless of treatment assignment. There is a lot of information in this article but these are the findings that stood out to me the most. All of them align to the same idea: the higher the sNfL levels at baseline, the greater the lesions, atrophy, and perhaps tendency towards more significant disability outcomes.

KB: I think that is a really good way to summarize it. They found that if you had a high sNfL level, there was a higher rate of neT2 lesions and a higher rate of atrophy. Also, you specifically mentioned thalamic atrophy, which has always been an area of particular interest with MS. What were the clinical correlates to this, Victor?

VS: One detail that they could not find was a statistically significant difference between the 2 cohorts in the annualized relapse rate and that is something that I tried to delve into a little bit more myself. If you look at the methods, they looked at neT2 lesions and annualized relapse rate separately using these regression models. Curiously, for the neT2 lesions, their regression model did not take into account the presence of Gd+ T1 lesions, but the annualized relapse rate model did. You can see in Table 1 that the presence of Gd+ T1 lesions at baseline was higher for patients who were in the high baseline sNfL category.

It is possible that they may not have found a difference between the 2 cohorts, because it was already conveyed or controlled for by the presence of Gd+ T1 lesions at baseline. Of course, they later go on to show that when they looked at patients in the subgroup who lacked Gd+ T1 lesions at baseline, there was still a significantly higher rate of neT2 lesion formation in those with a high vs low baseline sNfL. But I thought that could be a potential reason why they did not see a difference in the annualized relapse rate.

I think that a key limitation in a lot of the clinical studies in MS, especially now with the availability of highly efficacious treatment, is that you have to accrue large patient samples and follow them for long periods of time in order to have relapses that can be compared between 2 cohorts. Of course, they also go on to talk about the confirmed disease worsening or progression, which is effectively a threshold change in the EDSS that is sustained after a certain period of time. They did not find a statistically significant difference, but you can see from the data that there is a trend towards increased disability progression between the 2 cohorts that changes with treatment.

KB: I think that you bring up really good points there. You are right, whenever study authors compared high vs low sNfL groups, they did not see a statistically significant difference in terms of relapse rates. Of course, they say that in both groups, the relapse rates were relatively low. It may be that we have not followed these patients for a long enough duration to be able to detect that difference. This is probably the same for understanding disability worsening. Certainly, the authors also mentioned that this trial was not powered to detect a difference in disability worsening with sNfL levels. In the results section, researchers looked at the sNfL levels and how the 2 different treatments in the high vs low sNfL groups affected the different outcomes that they were measuring. Natalia, will you comment on the effect of ofatumumab vs teriflunomide in these patients?

NGC: In patients treated with ofatumumab or teriflunomide, patients with a higher sNfL levels at baseline had higher atrophy rates. At 3 months and 6 months, they did not see the same results in patients in the low sNfL group but there was a significant difference in treatments outcomes. Meaning that, even when there was a high sNfL, ofatumumab was more efficacious in those markers than teriflunomide.

KB: Part of this goes back to the original ASCLEPIOS 1/2 trials. I think it is important to highlight that they did not just check sNfL levels at the beginning of the study. In the actual ASCLEPIOS 1/2 trials, they actually reported the sNfL levels. We did see a higher reduction in sNfL levels. The sNfL levels were lower in the ofatumumab-treated group vs the teriflunomide-treated group. In the results, they discuss the effect of ofatumumab on relapse rates, disability progression, and brain volume loss.

Another interesting analysis that I think is worth talking about, although not explicitly mentioned in the discussion, is that they looked at patients who did not have Gd+ lesions. I think that is really intriguing, and both of you have alluded to this. NfL is a marker that is often very elevated after a relapse or after a new Gd+ lesion. Dr Salinas, would you comment on the findings regarding patients who did not have Gd+ lesions?

VS: I think the results of those are in the supplementary figures, but they also discussed this in the text. It was the case that these patients had decreased neT2 lesion frequency in the cohort with high sNfL, which I thought was interesting. There is data showing that sNfL is a prognostic marker for progressive MS, which is characterized by the accrual of disability in the absence of T1 Gd+ lesions, or so-called active lesions. I see sNfL possibly integrating all of this. I think one of the next challenges would be to differentiate the value of sNfL as telling us something about active information and relapses vs the insidious accrual of disability or degeneration that happens gradually in patients. But going back to the original question, even in patients who were free of Gd+ lesions, there was indeed a higher frequency of T2 lesion formation in patients with elevated sNfL levels.

KB: Just to piggyback on that question, Victor, what do you think is the clinical implication for that? So again, we see patients in clinic, and we are having to advise them and make a decision on the appropriate disease-modifying therapy (DMT) based upon a lot of factors. But do you think this study makes an argument that sNfL should be part of that conversation?

VS: I think so. It can form part of our arsenal of data that we use to guide our patients to make the most informed decisions. Obviously, this paper and the ASCLEPIOS trials already provide evidence that ofatumumab is a more efficacious medication that has the ability to reduce disease activity in patients with high or low sNfL levels. I think in cases in which we have patients who might be initially more apprehensive about taking an infusion or injectable, this is data that we can use to convince them of the benefits.

We will still rely on the many clinical and demographic variables that we use to determine if a patient is at higher risk and if more efficacious treatments should be pursued. What I like about sNfL is that it seems to be able to integrate both clinical and radiographic information into a single measurement. Even with continued reliance on measures like EDSS for disability, we know that a lot of our patients suffer from other things that the EDSS doesn’t capture, including issues with sleep and mood. In this regard, the use of sNfL could help us make more informed, data-driven decisions for our patients.

KB: Natalia, what are your takeaways from this study?

NGC: I agree with everything that Victor said and I have 2 main takeaways. I think that sNfL can be used as a biomarker to assess for response to treatment in future trials. Also, I think in the clinic setting, it could be most helpful in cases in which the patient is not having obvious signs of aggressive MS but they are at that halfway point when it’s unclear if they will benefit from an infusion, injectable, or oral disease-modifying therapy. This also is true in those cases in which the patient is doing well, the EDSS is 0, but they have some lesions in the brain. The sNfL would give us that extra piece of information to help guide our decision making. However, given the novelty of this biomarker, I think that a question that has not been completely answered is the cutoff. How do we decide if the sNfL is high or low? I think that is something that still has to be further categorized.

KB: I think that is a very good point, Natalia. This study definitely divided patients into high and low groups based upon the median sNfL across the trial, but defining the meaningful cutoff is probably the biggest barrier to using sNfL in the clinical setting at the moment. Of course, there still are barriers in terms of age adjustments, because we know that NfL increases with age, so there needs to be some adjustment for age whenever an individual’s sNfL is calculated. So, yes, I agree, there definitely needs to be more standardization in terms of determining what is considered to be a high vs low sNfL level.

This article was edited for clarity and length.


Ziemssen T, Arnold DL, Alvarez E, et al. Prognostic value of serum neurofilament light chain for disease activity and worsening in patients with relapsing multiple sclerosis: results from the phase 3 ASCLEPIOS I and II trialsFront Immunol. 2022;13:852563. doi:10.3389/fimmu.2022.852563

Posted by Haymarket’s Clinical Content Hub. The editorial staff of Neurology Advisor had no role in this content’s preparation.

Reviewed July 2022