Kappa Free Light Chains Reflect Treatment Response and Progression Independent of Focal Inflammation in Multiple Sclerosis

ABSTRACT

Background

Cerebrospinal fluid (CSF) kappa free light chains (KFLC) is a sensitive marker of intrathecal immunoglobulin synthesis and is incorporated into the 2024 McDonald criteria for multiple sclerosis (MS). How disease‐modifying therapies (DMTs) influence KFLC dynamics and their association with treatment response remains unclear.

Objective

To determine whether intrathecal KFLC synthesis changes during DMT and whether these changes are associated with clinical outcomes.

Methods

Patients with treatment‐naïve relapsing–remitting MS were prospectively enrolled at the Sahlgrenska MS Center (Gothenburg, Sweden). Paired CSF and serum samples were collected at baseline and after 12 months of treatment with dimethyl fumarate (DMF) or natalizumab (NTZ). KFLC index was calculated as [(CSF KFLC/serum KFLC)/(CSF albumin/serum albumin)]. Treatment response was assessed using no evidence of disease activity‐3 (NEDA‐3) and progression independent of relapse and MRI activity (PIRMA⁺).

Results

Forty‐eight patients were included (DMF n = 26, NTZ n = 22). NTZ reduced KFLC index from a median 117.4 (63.6–171.9) at baseline to 78.8 (39.4–104.0) at 12 months (adjusted p = 0.003), corresponding to a median decline of −51.0 (IQR −82.1 to −24.7), whereas DMF produced no meaningful change. In NTZ‐treated patients maintaining NEDA‐3 or without PIRMA⁺, KFLC index declined markedly (both p < 0.01). Change in KFLC index predicted outcomes, yielding an AUC of 1.0 for NEDA‐3 and 0.79 for non‐PIRMA⁺.

Conclusions

Natalizumab appears to reduce intrathecal KFLC synthesis, and a decline in KFLC index may reflect effective suppression of CNS humoral immunity. ΔKFLC index could serve as a sensitive biomarker of treatment response and disease stability in MS.

In a prospective cohort of treatment‐naïve patients with relapsing–remitting multiple sclerosis, paired CSF and serum samples were obtained at baseline and after 12 months of disease‐modifying therapy. Natalizumab (NTZ), but not dimethyl fumarate (DMF), reduced the KFLC index. A decline in KFLC index predicted favorable long‐term outcomes, including NEDA‐3 and absence of PIRMA⁺. Figure created with Biorender.

1. Introduction

Cerebrospinal fluid (CSF) kappa free light chains (KFLC), a marker of intrathecal immunoglobulin synthesis, have emerged as a sensitive and specific biomarker in multiple sclerosis (MS) diagnostics [1]. A recent systematic review and meta‐analysis comparing the diagnostic accuracy of the KFLC index with that of IgG oligoclonal bands (OCB) included 32 studies comprising approximately 3300 patients with clinically isolated syndrome (CIS) or MS and 5800 controls [2]. The results of this work, together with the supporting body of evidence, led to the inclusion of CSF KFLC synthesis in the revised 2024 McDonald criteria for MS diagnosis [3].

Beyond its diagnostic utility, the KFLC index also carries important prognostic information. Elevated baseline KFLC index values have been associated with increased risk of early disease activity in radiologically and clinically isolated syndromes (RIS and CIS) as well as in relapsing–remitting MS (RRMS) [4, 5, 6, 7], development of MS symptoms in people with RIS [7], progression independent of relapse activity [6], and cognitive decline [8].

Despite this growing evidence, relatively few studies have explored the influence of disease‐modifying therapies (DMTs) on intrathecal KFLC synthesis [9, 10, 11], and none have examined the relationship between the KFLC index and treatment response in MS. The present study was designed to address this knowledge gap. We hypothesized that suppression of intrathecal KFLC synthesis would be associated with favorable treatment response on disease activity and progression, particularly in patients receiving high‐efficacy therapy.

2. Materials and Methods

2.1. Study Design

Patients with suspected onset of MS were prospectively enrolled in a cohort study at the Multiple Sclerosis Center, Sahlgrenska University Hospital, Gothenburg, Sweden, between April 2014 and June 2016. The original cohort has been described previously [12]. Participants were followed prospectively to evaluate biomarkers of inflammation and neurodegeneration.

Inclusion criteria were a diagnosis of RRMS according to the revised 2017 McDonald criteria [13] (applied retrospectively) and a minimum follow‐up duration of 3 years. Exclusion criteria included concomitant neurological, ophthalmological, or systemic inflammatory diseases. All patients were newly diagnosed, treatment‐naïve at the time of inclusion, and initiated DMT according to the treating physician's recommendation.

For the purpose of the present study, we included a subset of patients from this cohort who were treated with either dimethyl fumarate (DMF) or natalizumab (NTZ) and for whom paired CSF and serum samples were available at baseline (prior to DMT initiation) and after 12 months of treatment.

The dates of first symptom onset, diagnosis, relapses, and concomitant conditions were recorded at baseline and verified at the end of the study. Neurological and functional assessments, including the Expanded disability status scale (EDSS), 9‐hole peg test (9‐HPT), timed 25‐foot walk test (T25FWT), symbol digit modalities test (SDMT), and magnetic resonance imaging (MRI), were performed at baseline, 6 months, and 12 months. Brain MRI was performed on a 3.0 Tesla MRI scanner (Philips Achieva dStream, head coil type with 16 coil channels). All sequences were obtained after a standard dose of iv gadolinium. We acquired conventional T1‐weighted, T2‐weighted, and fluid‐attenuated inversion recovery (FLAIR). A trained neuroradiologist blinded to clinical data assessed the T2 lesion and contrast‐enhancing lesion (CEL) count.

2.2. Evaluation of Treatment Response: Study Endpoints

Two clinical endpoints were evaluated: no evidence of disease activity‐3 (NEDA‐3) and progression independent of relapse and MRI activity (PIRMA⁺). ΔKFLC index was calculated from baseline to 12 months, and NEDA‐3 and PIRMA⁺ events were assessed only during the subsequent 12–60 months of follow‐up.

NEDA‐3 was defined as the absence of any of the following events during the follow‐up period: (i) clinical relapses, (ii) confirmed disability worsening (CDW) sustained for at least 6 months, or (iii) new T1 gadolinium‐enhancing lesions and/or new or enlarging T2‐weighted lesions on MRI [14]. A clinical relapse was defined as new or worsening neurological symptoms lasting at least 24 h and not attributable to another cause [13]. Patients were thus dichotomized into those who maintained NEDA‐3 status throughout follow‐up and those who exhibited evidence of disease activity (EDA‐3).

CDW was defined as an increase in the EDSS score sustained for at least 3 months, by ≥ 1.5 points if baseline EDSS was 0, ≥ 1.0 point if baseline EDSS was between 1.0 and 5.0, and ≥ 0.5 points if baseline EDSS was ≥ 5.5.

PIRMA⁺ was defined as the occurrence of any of the following sustained outcomes in the absence of clinical relapses or MRI evidence of inflammatory activity (i.e., new or enlarging T2 lesions and/or CELs):

• CDW as defined above,

• ≥ 20% increase in T25FWT or 9‐HPT times, or

• ≥ 10% decrease in SDMT score [15].

A PIRMA⁺ event was recorded only if the worsening was confirmed at a follow‐up visit more than 3 months after the initial event, with no relapse or new MRI activity within the 3‐month interval between baseline and confirmation assessments. A PIRMA⁺ event was considered sustained if confirmed at the final study visit.

To ensure high sensitivity, we used a roving baseline EDSS approach [16] together with a composite EDSS/T25FWT/9‐HPT/SDMT evaluation to identify PIRMA⁺ events [15].

2.3. Analyses of Intrathecal Immunoglobulin Synthesis

Paired CSF and serum samples were collected at baseline as part of the routine diagnostic work‐up. Patients who consented to longitudinal follow‐up and a second lumbar puncture contributed repeat paired CSF and serum samples at 12 months. All material was processed and stored according to the BioMS‐EU consensus protocol for CSF biomarker research in MS [17]. Blood (EDTA tubes) and CSF (polypropylene tubes) were obtained concurrently, centrifuged on‐site to separate plasma, aliquoted into polypropylene cryovials, and stored at −80°C. No more than one freeze–thaw cycle occurred prior to analysis. Concentrations of KFLC in serum and CSF were measured post hoc using the N Latex FLC kappa kit on an Atellica NEPH 630 instrument (Siemens), according to the manufacturer's instructions. The KFLC index was calculated as [(CSF KFLC/serum KFLC)/(CSF albumin/serum albumin)].

CSF and serum albumin concentrations were determined using the ALBT2 reagent cassette on a cobas c module instrument (Roche), and the CSF/serum albumin ratio was calculated as [CSF albumin (mg/L)/serum albumin (g/L)].

CSF‐specific IgG OCBs were determined using an in‐house isoelectric focusing (IEF) method on 7.7% polyacrylamide gels followed by silver staining. Matched CSF and serum samples from each patient were run on adjacent lanes, and CSF‐specific OCBs were defined as additional bands in the gamma region that were absent in the corresponding serum sample. For quality control, a positive CSF sample with known CSF‐specific OCBs was included on each gel. A cut‐off of ≥ 2 CSF‐restricted bands was considered OCB‐positive.

All analyses were performed by board‐certified laboratory technicians blinded to clinical data, using standardized procedures for quality control and run approval. Analyses were conducted at the Sahlgrenska Neurochemistry Laboratory in Mölndal, Sweden.

2.4. Statistics

Demographic and clinical data are presented as counts and percentages or median and interquartile range (IQR), as appropriate. Data distribution was assessed with the Shapiro‐Wilks test. The Mann–Whitney U test, χ2 test, and Fisher's exact test were used for group comparisons, as appropriate.

Wilcoxon matched‐pairs signed rank test was used to compare treatment groups at baseline and 12 months follow‐up and to compare DMF‐ and NTZ‐treated patients according to NEDA‐3/EDA‐3 and non‐PIRMA⁺/PIRMA⁺ status at both time points. The Holm‐Šídák method was used to set the p‐value threshold for multiple comparisons (alpha = 0.05). Paired change in KFLC index was calculated for each patient as the difference between 12‐month and baseline values, and median and for ΔKFLC index were derived directly from the distribution of individual paired differences.

ΔKFLC index was calculated as 12 M—baseline. To investigate the association of ΔKFLC index with disease activity in NTZ‐treated patients, we first compared NEDA‐3 versus EDA‐3 and non‐PIRMA⁺ vs. PIRMA⁺ with the Mann–Whitney U test. Then, we performed ROC curve analysis with both outcomes (NEDA‐3 and non‐PIRMA⁺) and calculated area under the curve with 95% CI (Wilson‐Brown method), optimal ΔKFLC index with the Youden index, and sensitivity and specificity. Box plots and ROC curves are shown for visual illustration. To determine whether KFLC index changes reflected true intrathecal synthesis rather than blood–CSF barrier effects, we performed partial Spearman correlations between ΔKFLC index and ΔCSF KFLC while adjusting for ΔQAlb (ratio of CSF‐to‐serum albumin). Partial correlations were chosen given the non‐normal distribution of biomarker changes. All statistical analyses were performed with IBM SPSS 30 and GraphPad Prism 10.6.1.

2.5. Ethical Standards

All patients participated voluntarily in the study and provided written informed consent. The study conformed to the Code of Ethics of the World Medical Association (Declaration of Helsinki). The Regional Ethics Review Board in Gothenburg, Sweden, approved the study (Reference number 895‐13).

3. Results

3.1. Demographic and Clinical Data at Baseline and Follow‐Up

The original prospective cohort [12] included 43 RRMS patients initiating DMF and 31 initiating NTZ. Of these, 26 DMF‐treated and 22 NTZ‐treated patients underwent a second lumbar puncture at 12 months and were therefore eligible for inclusion in the present analysis. Patients who contributed 12‐month CSF samples while receiving other DMTs (n = 16) were excluded because fewer than five individuals were treated with any single agent, precluding reliable subgroup analyses. Demographic, clinical, and CSF data at baseline and follow‐up are presented in Table 1. At baseline, 26 patients were assigned to DMF and 22 to NTZ. The DMF and NTZ groups did not differ significantly in age, sex, or disease duration. Patients starting NTZ had slightly higher EDSS at baseline (p = 0.03) compared with those on DMF. More patients starting NTZ had CELs at baseline compared with those starting DMF (54.2% vs. 38.5%, p = 0.023). As expected, more patients initiating DMF had an EDA‐3 event during follow‐up compared with NTZ (61.5% vs. 22.7%, p = 0.009).

TABLE 1.

Demographic, clinical, and CSF data at baseline and follow‐up in RRMS patients treated with either DMF or NTZ.

	DMF (n = 26)	NTZ (n = 22)	p
Clinical and demographic data
Age, y, m (IQR)	34.5 (29.5–39.0)	32.5 (25.8–40.0)	0.58 a
Sex, F, n (%)	17 (65.4)	19 (86.4)	0.18 b
Disease duration, y, m (IQR)	0.0 (0.0–1.25)	0.0 (0.0–1.0)	0.46 a
Baseline EDSS, m (IQR)	2.0 (1.0–2.5)	2.5 (2.0–3.0)	0.03 a
Baseline T2 lesions, n (%)			0.37 b
1–9	16 (61.5)	9 (40.9)
≥ 10	10 (38.5)	13 (59.1)
Baseline CEL (yes), n (%)	10 (38.5)	26 (54.2)	0.023 b
Methylprednisolone treatment at baseline (yes), n (%)	5 (19.2)	7 (31.8)	0.34 b
EDA‐3 (yes), n (%)	16 (61.5)	5 (22.7)	0.009 b
Clinical relapse, n (%)	4 (25)	0 (0)
MRI activity, n (%)	13 (81.3)	2 (40)
CDW, n (%)	3 (18.8)	5 (100)
PIRMA+ (yes), n (%)	8 (30.8)	11 (50)	0.24 b
CDW, n (%)	3 (37.5)	5 (45.5)
SDMT worsening, n (%)	7 (87.5)	7 (63.6)
9‐HPT/T25FWT worsening, n (%)	2 (25)	5 (45.5)

	Baseline	12 months	p	Baseline	12 months	p
CSF data
Serum KFLC ng/L, m (IQR)	12.5 (10.2–13.5)	11.5 (9.5–13.3)	0.28 c	12.8 (10.5–16.7)	12.7 (9.4–14.5)	0.01 c
CSF KFLC ng/L, m (IQR)	5.4 (2.2–10.3)	3.2 (1.1–7.6)	0.007 c	6.5 (3.8–15.6)	3.4 (1.6–6.3)	< 0.001 c
KFLC index, m (IQR)	89.4 (34.1–192)	55.5 (26.2–182.3)	0.27 c	117.4 (63.6–171.9)	78.8 (39.4–104)	< 0.001 c
Qalb, m (IQR)	4.3 (2.9–6.4)	4.2 (2.8–5.5)	0.02 c	5.0 (3.9–5.7)	4.1 (3.4–4.8)	0.003 c
IgG OCBs ≥ 2 (yes), n (%)	26 (100)	26 (100)	1.0 b	22 (100)	22 (100)	1.0 b

Note: Bold text indicates p‐values < 0.05.

Abbreviations: 9‐HPT, 9 hole peg test; T25FWT, Timed 25‐ft walk test; CDW, confirmed disability worsening; CEL, contrast‐enhancing lesions; CSF, cerebrospinal fluid; DMF, dimethyl fumarate; EDA‐3, evidence of disease activity‐3; EDSS, expanded disability status scale; IgG, immunoglobulin G; IQR, interquartile range; KFLC, kappa free light chains; OCB, oligoclonal bands; NTZ, natalizumab; PIRMA, progression independent of relapse and magnetic resonance imaging activity; Qalb, albumin quotient; RRMS, relapsing–remitting multiple sclerosis; SDMT, symbol digit modalities test.

^a Mann–Whitney U test.

^b Fisher's exact test or Pearson Chi square test.

^c Wilcoxon matched‐pairs signed rank test.

3.2. Changes in KFLC Index by Treatment

Only NTZ‐treated patients demonstrated a statistically significant reduction in KFLC index at 12 months (median 78.8, IQR [39.4–104]) compared with baseline (117.4 [63.6–171.9], adjusted p = 0.003; Figure 1A). The corresponding median change (ΔKFLC index) in NTZ‐treated patients was −51.0 (IQR −82.1 to −24.7). In DMF‐treated patients, KFLC index decreased from a median 89.4 (34.1–192.0) at baseline to 55.5 (26.2–182.3) at 12 months, corresponding to a median reduction of −6.3 (IQR −28.7 to +19.1), indicating no consistent reduction in intrathecal KFLC synthesis. When analyzing CSF KFLC concentrations, both DMF‐ and NTZ‐treated patients exhibited reductions over time (adjusted p = 0.03 and < 0.001, respectively; Table 1 [CSF data], Figure 1B).

FIGURE 1.

Changes in CSF KFLC index and concentrations during treatment with dimethyl fumarate (DMF) and natalizumab (NTZ). (A) KFLC index and (B) CSF KFLC concentrations at baseline and after 12 months of treatment with DMF or NTZ. (C, D) KFLC index changes in DMF‐ (C) and NTZ‐treated (D) patients stratified by no evidence of disease activity (NEDA‐3) or evidence of disease activity (EDA‐3). (E, F) KFLC index changes in DMF‐ (E) and NTZ‐treated (F) patients stratified by progression independent of relapse and MRI activity (PIRMA⁺) or absence thereof (non‐PIRMA⁺). Horizontal bars represent medians. Statistical comparisons were performed using the Wilcoxon matched‐pairs signed‐rank test. ns = not significant; p < 0.01 (**), p < 0.001 (***).

3.3. Association of KFLC Index With NEDA‐3

We next examined whether KFLC index was associated with treatment response as defined by NEDA‐3 at 12‐month follow‐up. Table 1 demonstrates the main drivers of EDA‐3 events in the present cohort. Only one EDA‐3 event coincided with sampling at 12 months. All other EDA‐3 events occurred at later time points. In NTZ‐treated patients who maintained NEDA‐3 status, KFLC index at 12 months (median 67.8, IQR [32.1–97.0]) was significantly lower than at baseline (119.7 [63.4–172.2]; adjusted p < 0.001; Figure 1D). In contrast, no significant change was observed in NTZ‐treated patients with evidence of disease activity (EDA‐3) (104.7 [74.5–313.4] vs. 113.3 [68.6–270.4]; adjusted p = 0.31).

Among DMF‐treated patients, KFLC index remained unchanged regardless of disease activity: NEDA‐3 group (47.3 [24.2–188.3] vs. 73.7 [29.2–159.9]; adjusted p = 0.68) and EDA‐3 group (75.9 [30.8–166.8] vs. 89.4 [41.1–214.6]; adjusted p = 0.68) (Figure 1C).

3.4. Association of KFLC Index With PIRMA ⁺

We then assessed KFLC index in relation to progression independent of relapse and MRI activity (PIRMA⁺). Table 1 shows the main drivers of PIRMA⁺ in each group. Only one PIRMA⁺ event coincided with KFLC index measurement at 12 months. All other events occurred at later time points. In NTZ‐treated patients who remained non‐PIRMA⁺, KFLC index at 12 months (48.3 [35.6–92.9]) was significantly lower than at baseline (119.7 [62.9–182.2]; adjusted p = 0.002; Figure 1F).

In NTZ‐treated patients who experienced a sustained PIRMA⁺ event, KFLC index showed a numerical but nonsignificant reduction (86.4 [63.1–104.7] vs. 113.3 [70.3–161.0]; adjusted p = 0.28).

In DMF‐treated patients, no statistically significant change was observed in either non‐PIRMA⁺ (55.5 [26.2–141.5] vs. 95.4 [44.1–192.0]; adjusted p = 0.13) or PIRMA⁺ groups (69.8 [29.5–434.2] vs. 65.4 [31.6–314.0]; adjusted p = 0.55) (Figure 1E).

3.5. Predictive Value of ΔKFLC Index

Finally, we examined whether the change in KFLC index (ΔKFLC index) predicted treatment response in NTZ‐treated patients.

Median ΔKFLC index in patients who achieved and had sustained NEDA‐3 was −51.0 (IQR −82.1 to −24.7), compared with 9.0 (−6.2 to 45.0) in those with EDA‐3 (p < 0.001; Figure 2A). ROC curve analysis showed excellent discrimination for NEDA‐3, with an AUC of 1.0 (95% CI 1.0–1.0, p < 0.001). A ΔKFLC index cutoff of −9.0 yielded 100% (95% CI 81.6%–100%) sensitivity and 100% (95% CI 56.5%–100%) specificity (Figure 2B).

FIGURE 2.

Change in KFLC index (ΔKFLC) predicts treatment response and disease stability in natalizumab‐treated patients. (A) ΔKFLC index in patients maintaining no evidence of disease activity (NEDA‐3) compared with those exhibiting evidence of disease activity (EDA‐3). (B) Receiver operating characteristic (ROC) curve of ΔKFLC index for predicting NEDA‐3 status (AUC = 1.0, 95% CI 1.0–1.0, p < 0.001). (C) ΔKFLC index in patients without progression independent of relapse and MRI activity (non‐PIRMA⁺) compared with those with sustained PIRMA⁺. (D) ROC curve of ΔKFLC index for predicting non‐PIRMA⁺ status (AUC = 0.79, 95% CI 0.60–0.98, p = 0.02). Horizontal bars represent medians. Statistical comparisons were performed using the Mann–Whitney U test. p < 0.05 (*), p < 0.001 (***).

Patients without a sustained PIRMA⁺ event had a median ΔKFLC index of −51.0 (−104 to −27.3), compared with −9.7 (−60 to 11.7) in those with PIRMA⁺ (p = 0.02; Figure 2C). ROC curve analysis for non‐PIRMA⁺ yielded an AUC of 0.79 (95% CI 0.60–0.98, p = 0.02). A cutoff of −20 provided 90.9% (95% CI 62.3%–99.5%) sensitivity and 63.6% (95% CI 35.4%–84.4%) specificity (Figure 2D).

ΔKFLC index was strongly correlated with ΔCSF KFLC (Spearman r = 0.82, p < 0.001), and this association persisted after adjusting for ΔQAlb (partial r = 0.80, p < 0.001), indicating that changes in KFLC index were not solely attributable to alterations in blood–CSF barrier function.

In DMF‐treated patients, ΔKFLC index did not differ significantly between NEDA‐3 and EDA‐3 groups (median −8.77 [IQR −39.1 to 2.34] vs. 7.27 [−12.7 to 74.9], p = 0.94) (Figure S1A). ROC analysis demonstrated no discriminatory ability (AUC = 0.51, 95% CI 0.29–0.74, p = 0.92) (Figure S1B). Similarly, ΔKFLC index was not significantly associated with non‐PIRMA⁺ status (median −18.8 [−33.5 to 11.7] vs. −3.56 [−37.9 to 14.7], p = 0.10) (Figure S1C). ROC analysis yielded an AUC of 0.71 (95% CI 0.48–0.94, p = 0.10), indicating imprecise and nonsignificant discrimination (Figure S1D).

4. Discussion

In this prospective study of newly diagnosed, treatment‐naïve patients with RRMS, we demonstrate that the KFLC index decreases significantly during NTZ therapy but not during treatment with DMF. Moreover, a reduction in KFLC index was associated with absence of disease activity and NEDA‐3 and lack of progression independent of relapse or MRI activity (non‐PIRMA⁺). Importantly, the change in KFLC index (ΔKFLC) demonstrated high predictive accuracy for both endpoints, with an AUC of 1.0 for NEDA‐3 and 0.79 for non‐PIRMA⁺. To our knowledge, this is the first study to link dynamic changes in intrathecal KFLC synthesis with treatment response in MS.

NTZ acts by blocking α4‐integrin–mediated lymphocyte trafficking across the blood–brain barrier, thereby reducing intrathecal B‐ and plasma‐cell activity [18, 19]. Moreover, NTZ has been previously shown to reduce or eliminate IgG OCBs [20, 21, 22]. The observed decline in KFLC index under NTZ thus likely reflects effective suppression of CNS immunoglobulin synthesis secondary to diminished immune‐cell migration. In contrast, DMF exerts primarily systemic immunomodulatory effects through Nrf2‐mediated antioxidant and anti‐inflammatory pathways, which may not sufficiently inhibit intrathecal immunoglobulin production during the first treatment year [23, 24, 25]. These mechanistic differences provide a plausible explanation for the treatment‐specific KFLC responses observed.

The predictive performance of ΔKFLC index in NTZ‐treated patients is particularly noteworthy. A decrease greater than −9 perfectly distinguished patients maintaining NEDA‐3 status from those with evidence of disease activity, while a cut‐off of −20 predicted the absence of PIRMA⁺ events with high sensitivity. This finding implies that longitudinal KFLC monitoring may serve as a sensitive pharmacodynamic marker, providing early indication of treatment efficacy or suboptimal response, complementary to MRI and clinical endpoints.

However, although ΔKFLC index perfectly discriminated NEDA‐3 from EDA‐3 in this small NTZ‐treated subgroup (AUC 1.0), the 95% confidence intervals for sensitivity (81.6%–100%) and specificity (56.5%–100%) indicate considerable statistical uncertainty and suggest that such performance is unlikely to generalize unchanged to larger cohorts. Similarly, for non‐PIRMA⁺, the wide confidence intervals around sensitivity (62.3%–99.5%) and specificity (35.4%–84.4%) underline the exploratory nature of these findings and the need for validation in larger, independent datasets.

In contrast, ΔKFLC index showed no significant predictive value in DMF‐treated patients. Although the ROC analysis for non‐PIRMA⁺ yielded an AUC above 0.7, the wide confidence interval and nonsignificant p‐value indicate statistical instability, likely reflecting limited sample size rather than a robust treatment effect.

Because QAlb decreased over time, particularly in NTZ‐treated patients, we examined whether KFLC index dynamics could be driven primarily by changes in blood–CSF barrier permeability. However, ΔKFLC index remained strongly associated with ΔCSF KFLC after adjustment for ΔQAlb (partial r = 0.80, p < 0.001), supporting the interpretation that the KFLC reduction reflects suppression of intrathecal immunoglobulin synthesis rather than barrier effects alone.

Our data extend previous observations that highly effective DMTs can reduce CSF KFLC concentrations [10, 11]. Duell et al. recently showed a gradual decline in KFLC index following B‐cell‐depleting and trafficking‐blocking therapies. However, their study included only a very small number of NTZ‐treated patients and did not evaluate treatment response outcomes. Interestingly, a decrease in KFLC index was also observed in their DMF–treated cohort, but sampling intervals varied considerably between patients, which may have influenced the results. Süße et al. reported lower intrathecal KFLC levels in patients receiving “very high‐efficacy” DMTs, but this study was cross‐sectional, and did neither assess concentrations before and after therapy intraindividually nor treatment response outcomes. We could previously demonstrate an association between high baseline KFLC index and EDA‐3 and PIRA at follow‐up [6]. The present findings complement these studies by demonstrating a clear association between KFLC reduction and clinical response, operationalized by NEDA‐3 and PIRMA⁺ endpoints.

The results suggest that serial KFLC measurement could serve as an accessible biomarker of treatment efficacy and disease stability, paralleling the use of serum or CSF neurofilament light chain (NfL) as a marker of neuroaxonal injury. A decreasing KFLC index under high‐efficacy therapy might indicate suppression of compartmentalized inflammation, whereas persistently elevated or increasing values could signal ongoing B‐cell activity and risk of disease progression.

This study has several limitations. The sample size was modest, and subgroup analyses, particularly those stratified by treatment and clinical outcome, were limited by statistical power. This was especially relevant in the DMF‐treated subgroup, where the absence of statistically significant associations between ΔKFLC index and clinical outcomes should not be interpreted as definitive evidence of no treatment effect. Rather, moderate effects may have remained undetected due to limited sample size. The wide confidence intervals observed in ROC analyses further reflect statistical imprecision and underscore the exploratory nature of these findings.

Nevertheless, prospective CSF studies with harmonized sampling at baseline and at a fixed interval after treatment initiation are challenging to design and execute. Within these constraints, we believe the present data provide informative preliminary insight into treatment‐related KFLC dynamics. The cohort was single‐center and restricted to the first treatment year, precluding assessment of long‐term prognostic value. In addition, treatment allocation was not randomized and may have introduced selection bias. It was not possible to perform multivariable models adjusting for baseline EDSS, contrast‐enhancing lesions, or steroid exposure due to the limited number of NEDA‐3 and PIRMA⁺ events, which would violate recommended events‐per‐variable thresholds. Accordingly, the associations observed between ΔKFLC and clinical outcomes should be interpreted as exploratory and unadjusted. Finally, while KFLC quantification is analytically robust, inter‐laboratory standardization remains limited, emphasizing the need for harmonized cut‐off values and assay calibration.

Larger multicenter studies with extended follow‐up are warranted to confirm these findings and to determine whether changes in KFLC index predict long‐term disability outcomes or treatment failure. Mechanistic studies examining CSF B‐cell subsets and clonality during therapy may also help elucidate the biological basis of KFLC dynamics.

In summary, we show that NTZ, but not DMF, reduces intrathecal KFLC synthesis, and that a decrease in KFLC index is associated with favorable treatment response and reduced progression independent of inflammatory activity. These findings identify KFLC index as a promising candidate biomarker for monitoring intrathecal immune activity and therapeutic efficacy in MS.

Author Contributions

Igal Rosenstein: conceptualization, investigation, funding acquisition, methodology, validation, visualization, software, formal analysis, project administration, data curation, supervision, resources. Clas Malmeström: writing – review and editing, data curation. Hlin Kvartsberg: writing – review and editing, methodology, formal analysis, data curation. Markus Axelsson: conceptualization, writing – review and editing, resources, data curation. Jan Lycke: writing – review and editing, resources, data curation, supervision, project administration. Magnus Johnsson: writing – review and editing, data curation. Lenka Novakova: conceptualization, funding acquisition, writing – review and editing, project administration, data curation, supervision, resources. Henrik Zetterberg: funding acquisition, methodology, writing – review and editing, data curation, supervision, resources.

Funding

This work was supported by grants from the Swedish Society of the Neurologically Disabled; the Research Foundation of the Multiple Sclerosis Society of Gothenburg; the Edit Jacobson Foundation; Foundation of Swedish MS research. Gothenburg Medical Society (GLS‐1021566). HZ is a Wallenberg Scholar and a Distinguished Professor at the Swedish Research Council supported by grants from the Swedish Research Council (#2023‐00356, #2022‐01018, and #2019‐02397), the European Union's Horizon Europe research and innovation programme under grant agreement No 101053962, Swedish State Support for Clinical Research (#ALFGBG‐71320), the Alzheimer Drug Discovery Foundation (ADDF), USA (#201809‐2016862), the AD Strategic Fund and the Alzheimer's Association (#ADSF‐21‐831376‐C, #ADSF‐21‐831381‐C, #ADSF‐21‐831377‐C, and #ADSF‐24‐1284328‐C), the European Partnership on Metrology, co‐financed from the European Union's Horizon Europe Research and Innovation Programme and by the Participating States (NEuroBioStand, #22HLT07), the Bluefield Project, Cure Alzheimer's Fund, the Olav Thon Foundation, the Erling‐Persson Family Foundation, Familjen Rönströms Stiftelse, Familjen Beiglers Stiftelse, Stiftelsen för Gamla Tjänarinnor, Hjärnfonden, Sweden (#FO2022‐0270), the European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska‐Curie grant agreement No 860197 (MIRIADE), the European Union Joint Programme—Neurodegenerative Disease Research (JPND2021‐00694), the National Institute for Health and Care Research University College London Hospitals Biomedical Research Centre, the UK Dementia Research Institute at UCL (UKDRI‐1003), and an anonymous donor.

Conflicts of Interest

I.R. has received compensation for lectures from Biogen, Novartis, Merck, and Sanofi, and has served on advisory boards for Sanofi. M.A. has received compensation for lectures and/or advisory boards from Biogen. Genzyme. M.J. has received compensation for lectures from Merck, and has served on advisory boards for Merck. C.M. has received honoraria for lectures and advisory board memberships from Biogen, Merck, Novartis, and SanofiAventis. H.Z. has served at scientific advisory boards and/or as a consultant for Abbvie, Acumen, Alector, Alzinova, ALZpath, Amylyx, Annexon, Apellis, Artery Therapeutics, AZTherapies, Cognito Therapeutics, CogRx, Denali, Eisai, Enigma, LabCorp, Merck Sharp & Dohme, Merry Life, Nervgen, Novo Nordisk, Optoceutics, Passage Bio, Pinteon Therapeutics, Prothena, Quanterix, Red Abbey Labs, reMYND, Roche, Samumed, ScandiBio Therapeutics AB, Siemens Healthineers, Triplet Therapeutics, and Wave, has given lectures sponsored by Alzecure, BioArctic, Biogen, Cellectricon, Fujirebio, LabCorp, Lilly, Novo Nordisk, Oy Medix Biochemica AB, Roche, and WebMD, is a co‐founder of Brain Biomarker Solutions in Gothenburg AB (BBS), which is a part of the GU Ventures Incubator Program, and is a shareholder of CERimmune Therapeutics (outside submitted work). J.L. has received travel support and/or lecture honoraria and has served on scientific advisory boards for Alexion, Almirall, Biogen, Bristol Myers Squibb, Janssen, Merck, Novartis, Roche and Sanofi; and has received unconditional research grants from Biogen and Novartis, and financial research support from Sanofi. L.N. has received lecture honoraria from Biogen, Novartis, Teva, Sanofi and Merck, has served on advisory boards for Merck, Janssen and Sanofi, has received unconditional research grant from Novartis and Sanofi. The other authors declare no conflicts of interest.

Supporting information

Figure S1: Change in KFLC index (ΔKFLC) and its association with treatment response and disease stability in dimethyl fumarate‐treated patients.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request. Restrictions may apply due to privacy, ethical regulations, or participant confidentiality, and therefore the dataset is not publicly available. Any shared data will be de‐identified in accordance with applicable guidelines.