Abstract
Background
In relapsing-remitting multiple sclerosis (RRMS), the assessment of clinical disease activity can be challenging.
Objectives
To determine the diagnostic potential of serum neurofilament light (sNfL) and glial fibrillary acidic protein (sGFAP) to distinguish a relapse from other causes of deterioration.
Methods
In this multicenter, prospective study, RRMS patients with new neurological symptoms in the last 14 days were followed for 12 weeks. A diagnosis was established by the treating physician or, when in doubt, a panel of experienced neurologists. Blood samples were taken at baseline and week 12.
Results
A total of 65 patients were included. At baseline, patients with a clear relapse had a significantly higher median sNfL (14.6 pg/mL) than those with a clear other cause (9.5 pg/mL, p = 0.004). Although not significant after correction for multiple testing, median sGFAP was also higher in relapse patients (73.0 vs 64.6 pg/mL, p = 0.036). An sNfL value below 6.0 pg/mL had a high sensitivity (67% at baseline (CI 22.3–95.7%) and 100% at follow-up (CI 54.1–100%)) to rule out a relapse.
Conclusions
Analysis of sNfL level can be useful as an add-on investigation to determine whether disease activity is present in patients with RRMS presenting with new symptoms.
Keywords: Relapsing-remitting multiple sclerosis, relapse, pseudo relapse, neurofilament light, glial fibrillary acidic protein
Introduction
In patients with relapsing-remitting multiple sclerosis (RRMS), the prevention of disease activity is an important treatment goal. However, the assessment of clinical disease activity can be challenging for the treating physician. A relapse is defined as a monophasic clinical episode with patient-reported symptoms and objective findings typical of MS, reflecting a focal or multifocal inflammatory demyelinating event in the central nervous system, developing acutely or sub acutely, with a duration of at least 24 h, with or without recovery, and in the absence of fever or infection. 1 Despite this clear definition, in daily practice, the distinction between a relapse, a pseudo relapse or another mimic is often difficult. The physician can use a combination of history, neurological examination, laboratory studies and MRI, but there is no gold standard and no specific laboratory tests to diagnose a relapse. 2
A biomarker assessing disease activity could be useful to distinguish a relapse from other causes of deterioration. The most extensively studied biomarker for MS is neurofilament light (NfL). Previous studies showed that elevated serum NfL (sNfL) levels are associated with recent relapses, and future disease activity and progression.3–7 Several pro- and retrospective studies described a higher serum or cerebrospinal fluid (CSF) NfL in patients with a recent relapse (<2 weeks–3 months) compared to patients in remission.8–10 Some other studies report on the natural course of NfL during a relapse and found an incline of NfL in the first weeks after a relapse and then a decline in several weeks or months.11,12 Another important fluid biomarker is glial fibrillary acidic protein (GFAP), which appears mainly associated with disease severity and disease progression instead of inflammatory activity.9,13–20 However, previous findings also showed a relationship between (CSF or serum) GFAP and recent relapses.21,22
Although extensively studied, data on the usability of both biomarkers to support or rule out a relapse diagnosis are limited. Therefore, we aimed to determine the diagnostic potential of sNfL and GFAP levels to support the physician to distinguish a relapse from other causes of deterioration.
Methods
Population
We used data from a multicenter, prospective, observational study. Between February 2020 and December 2022, this study recruited patients with RRMS or clinically isolated syndrome (CIS) from six outpatient clinics in the Netherlands. Patients were eligible if they experienced (sub)acute neurological symptoms, possibly caused by a relapse, within 14 days before the presumed baseline visit. Additional inclusion criteria were age between 18 and 65 years and a score of ≤5.5 on the Expanded Disability Status Scale (EDSS) prior to the baseline visit. Patients were excluded if they experienced a relapse or received treatment with methylprednisolone (MP) within one month prior to the inclusion in the study. Patients could only be included in the study once.
At four study sites, a blood sample was collected from each patient at baseline and W12 to assess NfL and GFAP levels. This subgroup of patients was used for the current analysis.
Study procedures
Study visits were planned at baseline (D0) and after 12 weeks (W12). At baseline, we collected demographic characteristics, medical history, the clinical course of MS, current immunomodulatory treatment (IMT) and an estimate of the EDSS score before onset of the new symptoms. The treating physician determined the considered cause of the new symptoms (diagnosis) and the degree of certainty about the diagnosis at both visits. During both visits, information on MP treatment and MRI results was documented. A blood sample was taken at D0 and W12 to assess NfL and GFAP levels.
Outcome measures
Diagnosis
The considered cause of the new symptoms was specified during each visit. The treating physician could choose one of the options as described in Table 1. We only used the W12 diagnosis for the current analysis, as this was considered the most reliable.
Table 1.
Possible diagnoses and their definition.
| Diagnosis | Definition / examples |
|---|---|
| Relapse | The development of new or worsening neurological symptoms attributable to MS. Symptoms must persist for more than 24 h, should be in the absence of fever and must be preceded by a stable or improving neurological state for at least 30 days (because of the study representing daily practice, without always an exact EDSS before onset of new symptoms, no obligatory objective neurological worsening was added to this definition) |
| Pseudo relapse | Worsening of preexisting or previously experienced neurological symptoms during an episode of fever, infection or heat (because this study was conducted during the COVID-19 pandemic, we also included worsening of symptoms shortly after a vaccination as a pseudo relapse) |
| Deterioration of MS due to other causes | Worsening of preexisting or previously experienced neurological symptoms during an episode of, for example, stress, fatigue or metabolic disturbances |
| Other | Consisted of peripheral neurological diseases, other central neurological diseases, functional neurological symptoms, MS-related symptoms and miscellaneous |
Certainty on diagnosis
At each visit, the treating physician was asked how certain he or she was about the cause of the new symptoms. This was scored on a five-point Likert scale, with the answer options ‘very sure’, ‘sure’, ‘neutral’, ‘unsure’ and ‘very unsure’. Here, we only used the certainty on the diagnosis at W12. In case of the answers ‘neutral’, ‘unsure’ or ‘very unsure’, a panel, consisting of three experienced neurologists, judged the available information and determined the cause of the new symptoms.
Blood samples and biomarker analysis
Blood samples were taken at D0 (before possible start of MP treatment) and W12. Blood was collected in Vacutainer serum separator tubes, stored for at least one hour at room temperature and then centrifuged at 1800 g for 10 min. One mL of the supernatant was aliquoted into polypropylene cryovials and stored at −80°C. At the end of the study, samples from all study sites were shipped on dry ice to the central laboratory. The NfL and GFAP concentrations were measured using single-molecule array immunoassay on an HD-X according to the manufacturer's instructions (Quanterix, Billerica, USA). The biomarker measurements were not known to the treating physicians.
Statistical analysis
We performed a sample size calculation for the assessment of differences in NfL between two groups (relapse and no relapse). Input was based on several studies investigating NfL levels in MS patients with and without clinical or radiological disease activity.3–5,7 Assuming a prevalence of relapses of 50% in each group and expected group means of 34 pg/mL (relapse) and 25 pg/mL (no relapse) and a standard deviation within a group of 20, a total of 158 patients (79 in each group) would be needed to achieve 80% power to detect differences among the means with a 0.05 significance level.
Results are presented as absolute and relative frequencies for categorical variables and medians with interquartile ranges (IQRs) for continuous data. We displayed results for the total group and split the group into two, based on the certainty of the diagnosis; a ‘no clear diagnosis’ group for the answers ‘neutral’, ‘unsure’ or ‘very unsure’, and a ‘clear diagnosis’ group for the answers ‘sure’ or ‘very sure’. Because of a small number of patients in the different diagnosis groups other than ‘relapse’, we categorized all patients into two groups for further analysis; a ‘relapse’ and ‘no relapse’ group. Differences between groups were measured using a Mann–Whitney U test, Pearson chi-square or Fisher's exact test where appropriate.
The results of the blood tests are displayed as median pg/mL (IQR). We calculated the relative and absolute difference between baseline and W12 samples. Also, we calculated a standardized score (Z score) for NfL levels by comparing the absolute NfL concentrations to a reference database. 6 The Z scores were adjusted for age and a default value of 25 for body mass index. We separately displayed the results for patients with and without a relapse in the ‘clear diagnosis’ group, as the diagnosis in this group was most reliable. Differences in biomarker concentrations between baseline and W12 samples were calculated using a Wilcoxon signed rank test, and the differences between direction of change of NfL and GFAP (decline or incline) in the ‘relapse’ and ‘no relapse’ group were measured using Pearson Chi-square test. Differences in changes over time based on diagnosis, IMT status, MP treatment status and MRI results were measured using a Kruskal–Wallis test or a Mann–Whitney U test.
The diagnostic value of NfL and GFAP was determined using a receiver operating characteristic (ROC) curve in the ‘clear diagnosis’ group, where the diagnosis of the treating physician served as gold standard. A cut-off value of an area under the ROC curve (AUC) of 0.65 was used to determine which measures should be further explored. We determined three cut-off values for each measure. The first was the one with the highest Youden's Index, which is a measure for the performance of diagnostic test, ranging from minus one to one, with the highest value indicating the cut-off value with the highest combined sensitivity and specificity. 23 The other two cut-off values were chosen based on a high specificity, with conservation of a sensitivity of at least 20%, and on a high sensitivity, with conservation of a specificity of at least 20%. Lastly, we used the three cut-off values in the ‘no clear diagnosis’ group, to find out sensitivity and specificity for each measure in this group, with the panel diagnosis as gold standard. The 95% confidence intervals were calculated using the Clopper-Pearson method.
Statistical analyses were carried out with SPSS version 29. p-values were adjusted for multiple testing using the false discovery rate of Benjamini and Hochberg. 24
Ethics
This study was conducted in accordance with the Declaration of Helsinki. The study was approved by the medical ethics committee of Brabant, Tilburg, the Netherlands (NL68078.028.19). All patients provided written informed consent before entering the study.
Data availability
Anonymized data will be available from the corresponding author upon reasonable request.
Results
Baseline characteristics
A total of 119 patients were included in the RELAMS study. This was a lower number than expected, as the inclusion rate was limited for several reasons, such as the COVID-19 pandemic, late presentation of patients and a lower rate of relapses due to highly efficient IMT. The inclusion period was extended for more than a year; however, the inclusion rate did not improve. Sixty-five patients had blood samples taken at both baseline and W12. Baseline characteristics are summarized in Table 2. The median age was 42 years (IQR 32–49 years), and 83.1% were female. The median disease duration was 6.5 years (IQR 2–14 years), and the median EDSS before enrollment was estimated at 2.0 (IQR 1.0–3.0). A total of 36 patients (55%) used IMT at baseline, from which 70% on platform therapies (injectables, dimethylfumarate and teriflunomide) and 30% on highly active therapies (fingolimod, oral cladribine, monoclonal antibodies). There were no significant differences in baseline characteristics between the patients with (n = 65) and without blood samples (n = 54).
Table 2.
Baseline characteristics.
| Baseline characteristic | Patients with blood samples (n = 65) |
|---|---|
| Age, years, median (IQR) | 42 (32–49) |
| Female, n (%) | 54 (83.1) |
| Diagnosis RRMS, n (%) | 63 (96.9) |
| Disease duration, years, median (IQR) | 6.5 (2–14) |
| Documented relapses in past 2 years, median (IQR) | 0 (0–1) |
| IMT at baseline, n (%) | 38 (58.5) |
| (Peg)Interferon, n (% of IMT) | 4 (10.5) |
| Glatiramer acetate, n (% of IMT) | 8 (21.1) |
| Dimethylfumarate, n (% of IMT) | 7 (18.4) |
| Teriflunomide, n (% of IMT) | 6 (15.8) |
| Fingolimod, n (% of IMT) | 2 (5.2) |
| Natalizumab, n (% of IMT) | 3 (7.9) |
| Ocrelizumab, n (% of IMT) | 6 (15.8) |
| >1 year after last Cladribine gift, n (% of IMT) | 2 (5.2) |
| EDSS before symptom onset, median (IQR) | 2.0 (1.0–3.0) |
IQR: interquartile range; RRMS: relapsing-remitting multiple sclerosis; IMT: immunomodulatory treatment; EDSS: expanded disability status scale.
Diagnosis at 12-week follow-up visit
The diagnosis by the treating physician at W12 was relapse in 37/65 patients (57%). Pseudo relapse was diagnosed in 13% and deterioration of MS due to other causes in 15% (Figure 1A). The treating physician was sure or very sure about the diagnosis in 47 patients (72%, ‘clear diagnosis’; Figure 1B). The expert panel assessed the diagnosis of the other 18 patients (‘no clear diagnosis’; Figure 1C and D). The panel was also not sure in three cases; they diagnosed these cases as a relapse, but had also arguments to diagnose them as a pseudo relapse. There were no differences in baseline characteristics between the ‘no clear diagnosis’ and ‘clear diagnosis’ groups.
Figure 1.
Distribution of week 12 diagnosis. (a) Distribution in total group, diagnosis by treating physician (n = 65). (b) Distribution in ‘clear diagnosis’ group, diagnosis by treating physician (n = 47). (c) Distribution in ‘no clear diagnosis’ group, diagnosis by treating physician (n = 18). (d) Distribution in ‘no clear diagnosis’ group, diagnosis by the expert panel (n = 18).
Blood samples
In the total group, median NfL was 11.4 pg/mL at baseline and 11.1 pg/mL at W12, and median GFAP was 73.0 pg/mL at baseline and 64.6 pg/mL at W12 (Table 3). No significant differences between the measurements at D0 and W12 were found. A significantly higher median NfL at baseline was found in patients in the ‘clear diagnosis’ group with a relapse (14.6 pg/mL) compared to patients without a relapse (9.5 pg/mL, p = 0.004; Table 3). Glial fibrillary acidic protein values at baseline were 73.0 pg/mL in the total group, 76.0 pg/mL in the ‘clear diagnosis’ relapse group and 64.6 pg/mL in the ‘clear diagnosis’ no relapse group (p = 0.036, significant before correction for multiple testing only; Table 3). Between patients with and without a relapse, no significant difference in change over time for both NfL and GFAP was found.
Table 3.
Results of blood samples.
| Measure | Total group (n = 65) | ‘Clear diagnosis’ group - Relapse (n = 32) | ‘Clear diagnosis’ group - No relapse (n = 15) | p value |
|---|---|---|---|---|
| Median pg/mL (IQR) | Median pg/mL (IQR) | Median pg/mL (IQR) | ||
| NfL D0 | 11.4 (7.7–15.3) | 14.6 (10.5–23.1) | 9.5 (6.2–12.6) | 0.004 |
| NfL W12 | 11.1 (7.8–16.4) | 14.4 (10.0–24.4) | 10.1 (6.0–13.8) | 0.025 |
| GFAP D0 | 73.0 (51.6–90.0) | 76.0 (63.0–98.5) | 64.6 (49.9–72.4) | 0.036 |
| GFAP W12 | 64.6 (49.3–87.8) | 75.3 (54.4–89.6) | 62.1 (53.8–79.5) | 0.263 |
| Z score NfL D0 | 1.3 (0.3–2.0) | 1.6 (1.1–2.7) | 0.6 (0.2–2.0) | 0.020 |
| Z score NfL W12 | 1.2 (0.6–2.0) | 1.7 (0.9–2.5) | 1.0 (0.3–1.9) | 0.070 |
| Absolute Δ NfL | 0.4 (−1.5–2.7) | 0.9 (−2.2–6.5) | −0.1 (−1.3–4.3) | 0.819 |
| Absolute Δ GFAP | −1.5 (−11.7–6.3) | −2.2 (−13.7–6.4) | 2.4 (−8.7–10.1) | 0.361 |
| Relative Δ NfL | 4.8 (−13.4–29.0) | 5.4 (−18.5–30.9) | −2.0 (−13.5–44.6) | 0.648 |
| Relative Δ GFAP | −3.6 (−15.7–9.7) | −3.8 (−15.3–12.8) | 3.1 (−11.3–16.8) | 0.373 |
Δ: difference between D0 and W12. Significant p values after correction for multiple testing are in bold. p values which are only significant before correction for multiple testing are in italics.
IQR: interquartile range; GFAP: glial fibrillary acidic protein.
The D0 and W12 measurements of NfL and Z NfL were higher in patients with new T2 lesions or gadolinium-enhancing lesions on MRI, which occurred more often in patients with a clear relapse (Supplementary Tables 1 and 2). The biomarker measurements did not differ significantly based on the use of IMT (platform; n = 25, highly active; n = 13, or no IMT; n = 27) or MP treatment status (yes; n = 23, or no; n = 42). Although not significant, we found lower values of both baseline NfL and GFAP for patients on highly active therapies compared to platform therapies, and for patients on platform therapies compared to no IMT (NfL 9.6 vs 11.9 vs 12.5 pg/mL and GFAP 65.6 vs 72.4 vs 75.1 pg/mL for, respectively, highly active therapies, platform therapies and no IMT; Supplementary Tables 3–6). No significant differences in baseline EDSS, EDSS before symptom onset, change between these EDSS values or relapses in the previous two years were found to explain this difference. Also, we found non-significant higher NfL and GFAP values and a stronger incline of NfL in the MP-treated patients compared to the not MP-treated patients (Supplementary Tables 7–9). These patients had a significantly higher baseline EDSS than patients without MP treatment (EDSS 3.5 vs EDSS 2.5, p = 0.013; Supplementary Tables 10 and 11). This was partially based on a (not significantly) higher baseline EDSS in the relapse group compared to the no relapse group (Supplementary Table 12). However, in the relapse group, we still found a significant higher baseline EDSS in the MP-treated patients compared to not MP-treated patients (Supplementary Table 13).
Diagnostic value of Nfl and GFAP to distinguish relapse from no relapse
Five of the blood sample measures (NfL D0, NfL W12, GFAP D0, Z score NfL D0 and Z score NfL W12) had an AUC higher than 0.65 (range 0.417–0.765), with the highest value for NfL D0 (Supplementary Table 14). A combination of any two of these parameters did not lead to a higher AUC than the highest value of the individual parameters. We further explored the values higher than 0.65 (Table 4). Based on the cut-off at the value of the highest Youden's Index, a baseline NfL of 12.8 pg/mL had the highest diagnostic value, reaching a sensitivity of 62.5% (CI 43.7–78.9%) and a specificity of 86.7% (CI 59.5–98.3%, Youden's Index 0.492). The highest sensitivity was found for both D0 and W12 NfL (6.0 pg/mL, 100% sensitivity, CI 89.1–100%), and the highest specificity was found for the Z score of NfL at D0 (Z score 2.7, 100% specificity, CI 78.2–100%).
Table 4.
Cut-off values of the blood sample measures and their diagnostic value in the ‘clear diagnosis’ group.
| Measure (AUC) | Chosen cut-off | Highest Youden's index | High sensitivity | High specificity |
|---|---|---|---|---|
| NfL D0 (0.77) | Cut-off | 12.8 (0.492) | 6.0 (0.235) | 14.4 (0.465) |
| Sensitivity (CI) | 62.5 (43.7–78.9) | 100 (89.1–100) | 53.1 (34.7–70.9) | |
| Specificity (CI) | 86.7 (59.5–98.3) | 20 (4.3–48.1) | 93.3 (68.1–99.8) | |
| NfL W12 (0.70) | Cut-off | 12.3 (0.390) | 6.0 (0.235) | 14.8 (0.335) |
| Sensitivity (CI) | 65.6 (46.8–81.4) | 100 (89.1–100) | 46.9 (29.1–65.3) | |
| Specificity (CI) | 73.3 (44.9–92.2) | 20 (4.3–48.1) | 86.7 (59.5–98.3) | |
| GFAP D0 (0.69) | Cut-off | 72.7 (0.456) | 47.0 (0.106) | 90.0 (0.246) |
| Sensitivity (CI) | 65.6 (46.4–81.4) | 90.6 (75.0–98.0) | 31.3 (16.1–50) | |
| Specificity (CI) | 80 (51.9–95.7) | 20 (4.3–48.1) | 93.3 (68.1–99.8) | |
| Z score NfL D0 (0.71) | Cut-off | 0.6 (0.408) | 0.3 (0.271) | 2.7 (0.219) |
| Sensitivity (CI) | 87.5 (71.0–96.5) | 93.8 (79.2–99.2) | 21.9 (9.3–40) | |
| Specificity (CI) | 53.3 (26.6–78.7) | 33.3 (11.8–61.6) | 100 (78.2–100) | |
| Z score NfL W12 (0.67) | Cut-off | 0.6 (0.338) | 0.4 (0.302) | 2.4 (0.148) |
| Sensitivity (CI) | 93.8 (79.2–99.2) | 96.9 (83.8–99.9) | 28.1 (13.7–46.7) | |
| Specificity (CI) | 40 (16.3–67.7) | 33.3 (11.8–61.6) | 86.7 (59.5–98.3) |
NfL and GFAP cut-offs displayed in pg/mL. Cut-off values based on the highest Youden's index, a high sensitivity and a high specificity, with corresponding sensitivity, specificity and confidence intervals. Best performing cut-off displayed in bold.
AUC: area under the curve; GFAP: glial fibrillary acidic protein; Sens: sensitivity; Spec: specificity; CI: confidence interval.
When we implemented the found cut-off values in the ‘no clear diagnosis’ group, the diagnostic value of the different blood sample measures was lower than in the ‘clear diagnosis’ group (Supplementary Table 15). Because the panel was also not sure in three cases, we excluded them from this analysis, which resulted in better diagnostic values. In that analysis, the highest sensitivity was found for NfL at W12 (6.0 pg/mL, 100% sensitivity), and the highest specificity was found for the Z score of NfL at D0 (Z score 2.7, 100% specificity; Table 5).
Table 5.
Diagnostic value of the established cut-off values in the ‘no clear diagnosis’ group.
| Measure (AUC) | Chosen cut-off | Highest Youden's index | High sensitivity | High specificity |
|---|---|---|---|---|
| NfL D0 (0.77) | Cut-off (pg/mL) | 12.8 | 6.0 | 14.4 |
| Sensitivity (CI) | 16.7 (0.4–64.1) | 66.7 (22.3–95.7) | 16.7 (0.4–64.1) | |
| Specificity (CI) | 88.9 (51.8–99.7) | 22.2 (2.8–60) | 88.9 (51.8–99.7) | |
| NfL W12 (0.70) | Cut-off (pg/mL) | 12.3 | 6.0 | 14.8 |
| Sensitivity (CI) | 33.3 (4.3–.77) | 100 (54.1–100) | 0 (0–45.9) | |
| Specificity (CI) | 88.9 (51.8–.99) | 22.2 (2.8–60) | 88.9 (51.8–99.7) | |
| GFAP D0 (0.69) | Cut-off (pg/mL) | 72.7 | 47.0 | 90.0 |
| Sensitivity (CI) | 50.0 (11.8–88.2) | 66.7 (22.3–95.7) | 16.7 (0.4–64.1) | |
| Specificity (CI) | 44.4 (13.7–78.8) | 33.3 (7.5–70.1) | 66.7 (29.9–92.5) | |
| Z score NfL D0 (0.71) | Cut-off | 0.3 | 0.6 | 2.7 |
| Sensitivity (CI) | 66.7 (22.3–95.7) | 66.7 (22.3–95.7) | 0 (0–45.9) | |
| Specificity (CI) | 55.6 (21.2–86.3) | 55.6 (21.2–86.3) | 100 (66.4–100) | |
| Z score NfL W12 (0.67) | Cut-off | 0.6 | 0.4 | 2.4 |
| Sensitivity (CI) | 66.7 (22.3–95.7) | 83.3 (35.9–99.6) | 0 (0–45.9) | |
| Specificity (CI) | 44.4 (13.7–78.8) | 44.4 (13.7–78.8) | 88.9 (51.8–99.7) |
N = 15 (excluding three cases where panel also was not sure). Cut-off values based on Youden's index, a high sensitivity and a high specificity, with corresponding sensitivity, specificity and confidence intervals. Best performing cut-off displayed in bold.
AUC: area under the curve; GFAP: glial fibrillary acidic protein; Sens: sensitivity; Spec: specificity; CI: Clopper Pearson confidence interval.
Discussion
In this prospective study, we found a significantly higher median sNfL at baseline in RRMS or CIS patients with a clear relapse compared to patients with a clear other cause of deterioration. Although not significant, sGFAP also showed a trend towards higher baseline values in relapse patients. We did not find a consistent decline or incline over time in both NfL and GFAP in both the relapse and non-relapse patients. While the diagnostic value of sGFAP for a relapse diagnosis was limited, we found a high diagnostic value for NfL and the Z score of NfL at baseline and W12 follow-up (100% sensitivity for a cut-off value of W12 NfL of 6.0 pg/mL and 100% specificity for a cut-off value of the Z score of NfL at D0 of 2.7).
After correction for multiple testing, we found different significant levels of the raw values of sNfL versus the Z score of NfL in the comparison between relapse and no relapse. This might be caused by our small sample size. Although theoretically the Z score has a higher sensitivity to detect pathological values, we found almost comparable results for sNfL and the Z score of NfL. This suggests that for the use of NfL to distinguish relapse from no relapse, it is not necessary to use the Z score. We did not include the Z score of GFAP, as opposed to NfL, there are no well-validated Z scores yet. The AUC did not change when we corrected the analysis for age or sex.
Our findings on the difference in biomarkers between relapse versus no relapse are in line with other studies investigating the differences between biomarkers in active MS versus MS in remission.8–10 We did not find a difference between median values short after onset and 12 weeks after onset, which corresponds to the results of colleagues investigating the natural course of NfL after a relapse.11,12 However, due to the timing of our samples, we were not able to replicate the finding that sNfL first inclines in the first weeks after a relapse and then steadily declines in several weeks or months.11,12 The diagnostic values we found for the baseline measures approach the values found by Johnsson et al. 12 We were not able to compare the follow-up findings, as the time difference with this study is too large (5.5 weeks after the relapse in Johnsson et al. vs 12 weeks in our study).
An important note to our findings is that the follow-up samples are potentially confounded by MP treatment. However, we did not find a significant difference in change over time of the biomarkers between MP-treated and not MP-treated patients. The finding that MP-treated patients have slightly higher baseline and follow-up values of both sNfL and sGFAP might reflect high disease activity, as both baseline EDSS and EDSS before symptom onset are higher in MP-treated relapse patients compared to not MP-treated relapse patients. Another clue for this reflection of disease severity is the finding that sNfL shows a stronger incline in MP-treated patients compared to the overall population.
The strengths of this prospective study are the study design, which reflects daily practice well, and the set-up of an expert panel for a diagnosis which is as reliable as possible and can act as a gold standard. We were able to show that baseline sNfL clearly differs between patients with and without a relapse, which could support a physician in his certainty on the diagnosis and subsequent management options. However, some precautions must be taken, as having a high sNfL value at a certain time point does not mean there is a certain relapse. First, sNfL is not specific for MS and inflammatory disease activity, and second, a patient with active MS could also experience a pseudo relapse or other causes of deterioration. This can explain the limited diagnostic values found in this study. On the contrary, a low sNfL value seems safe to use as an argument to rule out a relapse as potential cause of deterioration. We found in both the ‘clear diagnosis’ and the ‘no clear diagnosis’ group that a value of NfL at baseline or W12 follow-up below 6.0 pg/mL has a high sensitivity to rule out a relapse. Therefore, sNfL could possibly be used both at onset of new symptoms and up until 12 weeks after onset for this purpose. However, this study has several limitations. The sample size was below the required number of patients to achieve 80% power, and therefore, there was a low number of patients without a clear diagnosis. However, we still were able to show significant results, although with consequent large confidence intervals corresponding to the diagnostic values. Also, there was a lack of pre-deterioration values of the investigated biomarkers. Therefore, our results need to be confirmed in a larger cohort, with blood samples before, during and after clinical disease activity, and reliable diagnosis judgement (panel based).
In conclusion, we found a significant difference in median baseline sNfL between patients with and without a relapse. The diagnostic value of sNfL is promising and can possibly be useful as an add-on investigation to determine whether disease activity is present in patients with RRMS presenting with new symptoms.
Supplemental Material
Supplemental material, sj-docx-1-mso-10.1177_20552173251370830 for Relapse or no relapse in multiple sclerosis: Can disease activity biomarkers support the clinician? by Jeske van Pamelen, Marleen J A Koel-Simmelink, Birgit I Lissenberg, Edo P J Arnoldus, Janet de Beukelaar, Judith van Vliet, Joep Killestein, Charlotte E Teunissen and Leo H Visser in Multiple Sclerosis Journal – Experimental, Translational and Clinical
Acknowledgements
The authors would like to thank our patients for study participation and the MS and research nurses in the participating centers for their involvement with data acquisition. The RELAMS study group includes the following investigators: EPJ Arnoldus, MD, PhD, Department of Neurology, Elisabeth-TweeSteden Hospital, Tilburg, the Netherlands; J de Beukelaar, MD, PhD, Neurologist, Department of Neurology, Albert Schweitzer Hospital, Dordrecht, the Netherlands; J Killestein, MD, PhD, Neurologist, MS Center Amsterdam, Department of Neurology, Amsterdam Neuroscience, Amsterdam UMC Location Vrije Universiteit Amsterdam, Amsterdam, the Netherlands; JP Mostert, MD, PhD, Department of Neurology, Rijnstate Hospital, Arnhem, the Netherlands; J van Pamelen, MD, Department of Neurology, Elisabeth-TweeSteden Hospital, Tilburg, the Netherlands; LC van Rooij, MD, PhD, Department of Neurology, Maasstad Hospital, Rotterdam, the Netherlands; LH Visser, MD, PhD, Department of Neurology, Elisabeth-TweeSteden Hospital, Tilburg, the Netherlands; and J van Vliet, MD, PhD, Department of Neurology, Jeroen Bosch Hospital, ‘s-Hertogenbosch, the Netherlands.
Footnotes
Consent to participate: All patients provided written informed consent before entering the study.
Data availability: Anonymized data will be available from the corresponding author upon reasonable request.
J. van Pamelen received a research grant for the BIA study by Merck, and a travel grant for a scientific meeting by Merck, outside the submitted work; E.P.J. Arnoldus received research grants for research projects (Merck, Novartis, Sanofi-Genzyme), honoraria for lectures (Merck, Novartis, Sanofi-Genzyme, Biogen, Teva) and consulting fees (Merck, Novartis, Sanofi-Genzyme, Biogen, Teva); J. de Beukelaar received research grants from the Dutch National MS Foundation (OZ2019-19, OZ2021-19 and P2023-003), Stichting BeterKeten (OPZ 2019-019), Vaillant Fonds (20072020), Albert Schweitzer Hospital: various Stipendia (ASz2018-07, ASz2019-01, ASz2021-011, ASz2021-012, Asz 2024-06), mProve (131224) and Asz Vriendenfonds (191124); J. Killestein received research grants for multicentre investigator-initiated trials DOT-MS (NCT04260711, ZonMW), Supernext (NCT04225312, Treatmeds) and BLOOMS (NCT05296161, ZonMW and Treatmeds), received consulting fees (F Hoffmann-La Roche, Biogen, Teva, Merck, Novartis, and Sanofi/Genzyme (all payments to institution)) and honoraria for lectures (F Hoffmann-La Roche, Biogen, Immunic, Teva, Merck, Novartis, and Sanofi/Genzyme (all payments to institution)), and is member of the adjudication committee of MS clinical trials of Immunic (payments to institution only); C.E. Teunissen has research contracts with Acumen, ADx Neurosciences, AC-Immune, Alamar, Aribio, Axon Neurosciences, Beckman-Coulter, BioConnect, Bioorchestra, Brainstorm Therapeutics, C2N diagnostics, Celgene, Cognition Therapeutics, EIP Pharma, Eisai, Eli Lilly, Fujirebio, Instant Nano Biosensors, Novo Nordisk, Olink, PeopleBio, Quanterix, Roche, Toyama, Vivoryon. Her research is supported by the European Commission (Marie Curie International Training Network, grant agreement No 860197 (MIRIADE) and No 101119596 (TAME), Innovative Medicines Initiatives 3TR (Horizon 2020, grant no 831434) EPND ( IMI 2 Joint Undertaking (JU), grant No. 101034344) and JPND (bPRIDE, CCAD), European Partnership on Metrology, co-financed from the European Union's Horizon Europe Research and Innovation Programme and by the Participating States ((22HLT07 NEuroBioStand), Horizon Europe (PREDICTFTD, 101156175), CANTATE project funded by the Alzheimer Drug Discovery Foundation, Alzheimer Association, Michael J Fox Foundation, Health Holland, the Dutch Research Council (ZonMW), Alzheimer Drug Discovery Foundation, The Selfridges Group Foundation, Alzheimer Netherlands. CT is recipient of ABOARD, which is a public-private partnership receiving funding from ZonMW (#73305095007) and Health~Holland, Topsector Life Sciences & Health (PPP-allowance; #LSHM20106). CT is recipient of TAP-dementia, a ZonMw funded project (#10510032120003) in the context of the Dutch National Dementia Strategy. She is editor in chief of Alzheimer Research and Therapy, and serves on editorial boards of Molecular Neurodegeneration, Alzheimer's & Dementia, Neurology: Neuroimmunology & Neuroinflammation, Medidact Neurologie/Springer, and serves on committee to define guidelines for Cognitive disturbances, and one for acute Neurology in the Netherlands. She had consultancy/speaker contracts for Aribio, Biogen, Beckman-Coulter, Cognition Therapeutics, Eli Lilly, Merck, Novo Nordisk, Olink, Roche and Veravas; L.H. Visser received research grants for research projects (Merck), honoraria for lectures (Merck) and consulting fees (Merck, Novartis, Janssen-Cilag and Roche); The remaining authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Ethical considerations: This study was conducted in accordance with the Declaration of Helsinki. The study was approved by the medical ethics committee of Brabant, Tilburg, the Netherlands (NL68078.028.19).
Funding: The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The RELAMS study was funded by the Dutch National MS Foundation (OZ2018–007).
Supplemental material: Supplemental material for this article is available online.
ORCID iDs: Jeske van Pamelen https://orcid.org/0000-0003-2181-0195
Marleen J A Koel-Simmelink https://orcid.org/0000-0003-3939-7394
Birgit I Lissenberg https://orcid.org/0000-0001-9448-1826
Edo P J Arnoldus https://orcid.org/0000-0002-5944-5931
Janet de Beukelaar https://orcid.org/0009-0000-6089-9896
Judith van Vliet https://orcid.org/0009-0006-4043-4729
Joep Killestein https://orcid.org/0000-0002-6347-5957
Charlotte E Teunissen https://orcid.org/0000-0002-4061-0837
Leo H Visser https://orcid.org/0000-0002-0704-4538
Contributor Information
Jeske van Pamelen, Department of Neurology, Elisabeth-TweeSteden Hospital, Tilburg, The Netherlands; MS Center Amsterdam, Amsterdam Neuroscience, Amsterdam UMC Location Vrije Universiteit Amsterdam, Amsterdam, The Netherlands;.
Marleen J A Koel-Simmelink, Neurochemistry Laboratory, Department of Laboratory Medicine, Amsterdam Neuroscience, Amsterdam UMC Location Vrije Universiteit Amsterdam, Amsterdam, The Netherlands.
Birgit I Lissenberg, Department of Epidemiology and Data Sciences, Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands.
Edo P J Arnoldus, Department of Neurology, Elisabeth-TweeSteden Hospital, Tilburg, The Netherlands.
Janet de Beukelaar, Department of Neurology, Albert Schweitzer Hospital, Dordrecht, The Netherlands.
Judith van Vliet, Department of Neurology, Jeroen Bosch Hospital, 's Hertogenbosch, The Netherlands.
Joep Killestein, MS Center Amsterdam, Department of Neurology, Amsterdam Neuroscience, Amsterdam UMC Location Vrije Universiteit, Amsterdam, The Netherlands.
Charlotte E Teunissen, Neurochemistry Laboratory, Department of Laboratory Medicine, Amsterdam Neuroscience, Amsterdam UMC Location Vrije Universiteit Amsterdam, Amsterdam, The Netherlands.
Leo H Visser, Department of Neurology, Elisabeth-TweeSteden Hospital, Tilburg, The Netherlands.
References
- 1.Thompson AJ, Banwell BL, Barkhof F, et al. Diagnosis of multiple sclerosis: 2017 revisions of the McDonald criteria. Lancet Neurol 2018; 17: 162–173. [DOI] [PubMed] [Google Scholar]
- 2.Wang C, Ruiz A, Mao-Draayer Y. Assessment and treatment strategies for a multiple sclerosis relapse. J Immunol Clin Res 2018; 5: 20161207. [PMC free article] [PubMed] [Google Scholar]
- 3.Disanto G, Barro C, Benkert P, et al. Serum neurofilament light: a biomarker of neuronal damage in multiple sclerosis. Ann Neurol 2017; 81: 857–870. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Novakova L, Zetterberg H, Sundström P, et al. Monitoring disease activity in multiple sclerosis using serum neurofilament light protein. Neurology 2017; 89: 2230–2237. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Varhaug KN, Barro C, Bjørnevik K, et al. Neurofilament light chain predicts disease activity in relapsing-remitting MS. Neurol Neuroimmunol Neuroinflamm 2018; 5: e422. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Benkert P, Meier S, Schaedelin S, et al. Serum neurofilament light chain for individual prognostication of disease activity in people with multiple sclerosis: a retrospective modelling and validation study. Lancet Neurol 2022; 21: 246–257. [DOI] [PubMed] [Google Scholar]
- 7.Kuhle J, Kropshofer H, Haering DA, et al. Blood neurofilament light chain as a biomarker of MS disease activity and treatment response. Neurology 2019; 92: e1007–e1015. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Teunissen CE, Iacobaeus E, Khademi M, et al. Combination of CSF N-acetylaspartate and neurofilaments in multiple sclerosis. Neurology 2009; 72: 1322–1329. [DOI] [PubMed] [Google Scholar]
- 9.Meier S, Willemse EAJ, Schaedelin S, et al. Serum glial fibrillary acidic protein compared with neurofilament light chain as a biomarker for disease progression in multiple sclerosis. JAMA Neurol 2023; 80: 287–297. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Solís-Tarazona L, Raket LL, Cabello-Murgui J, et al. Predictive value of individual serum neurofilament light chain levels in short-term disease activity in relapsing multiple sclerosis. Front Neurol 2024; 15: 1354431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Malmeström C, Haghighi S, Rosengren L, et al. Neurofilament light protein and glial fibrillary acidic protein as biological markers in MS. Neurology 2003; 61: 1720–1725. [DOI] [PubMed] [Google Scholar]
- 12.Johnsson M, Stenberg YT, Farman HH, et al. Serum neurofilament light for detecting disease activity in individual patients in multiple sclerosis: a 48-week prospective single-center study. Mult Scler 2024; 30: 664–673. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Norgren N, Sundström P, Svenningsson A, et al. Neurofilament and glial fibrillary acidic protein in multiple sclerosis. Neurology 2004; 63: 1586–1590. [DOI] [PubMed] [Google Scholar]
- 14.Axelsson M, Malmeström C, Nilsson S, et al. Glial fibrillary acidic protein: a potential biomarker for progression in multiple sclerosis. J Neurol 2011; 258: 882–888. [DOI] [PubMed] [Google Scholar]
- 15.Martínez MAM, Olsson B, Bau L, et al. Glial and neuronal markers in cerebrospinal fluid predict progression in multiple sclerosis. Mult Scler 2015; 21: 550–561. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Abdelhak A, Huss A, Kassubek J, et al. Serum GFAP as a biomarker for disease severity in multiple sclerosis. Sci Rep 2018; 8: 14798. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Ayrignac X, Le Bars E, Duflos C, et al. Serum GFAP in multiple sclerosis: correlation with disease type and MRI markers of disease severity. Sci Rep 2020; 10: 10923. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Högel H, Rissanen E, Barro C, et al. Serum glial fibrillary acidic protein correlates with multiple sclerosis disease severity. Mult Scler 2020; 26: 210–219. [DOI] [PubMed] [Google Scholar]
- 19.Madill E, Healy BC, Molazadeh N, et al. Serum glial fibrillary acidic protein predicts disease progression in multiple sclerosis. Ann Clin Transl Neurol 2024; 11: 2719–2730. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Rosenstein I, Nordin A, Sabir H, et al. Association of serum glial fibrillary acidic protein with progression independent of relapse activity in multiple sclerosis. J Neurol 2024; 271: 4412–4422. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Burman J, Zetterberg H, Fransson M, et al. Assessing tissue damage in multiple sclerosis: a biomarker approach. Acta Neurol Scand 2014; 130: 81–89. [DOI] [PubMed] [Google Scholar]
- 22.Kuhle J. New Promising Biomarkers: GFAP, IgM, Complement. Milan: ECTRIMS. 2023. [Google Scholar]
- 23.Youden WJ. Index for rating diagnostic tests. Cancer 1950; 3: 32–35. [DOI] [PubMed] [Google Scholar]
- 24.Hochberg YBY. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Ser B (Methodol) 1995; 57: 289–300. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplemental material, sj-docx-1-mso-10.1177_20552173251370830 for Relapse or no relapse in multiple sclerosis: Can disease activity biomarkers support the clinician? by Jeske van Pamelen, Marleen J A Koel-Simmelink, Birgit I Lissenberg, Edo P J Arnoldus, Janet de Beukelaar, Judith van Vliet, Joep Killestein, Charlotte E Teunissen and Leo H Visser in Multiple Sclerosis Journal – Experimental, Translational and Clinical
Data Availability Statement
Anonymized data will be available from the corresponding author upon reasonable request.