Индекс Свободных цепей света Каппы и знак центральной веин: два новых биомаркера для многоразового диагностирования склероза.

Key Summary Points

The kappa free light chain (kFLC) index is highly correlated with oligoclonal bands status. Its positivity achieves high sensitivity and specificity for MS diagnosis, with an increased specificity for very high kFLC index if used as a quantitative variable. People with clinically isolated syndrome (pwCIS) and radiologically isolated syndrome (pwRIS) with a positive kFLC index have a higher risk of developing MS (prediction of relapse or MRI dissemination in space and time). The different technics used for kFLC measurement (i.e. turbidimetry, nephelometry, ELISA-based methods) show similar accuracy in detecting intrathecal Ig synthesis, with high reproductibility. The central vein sign (CVS) assessment should be performed on merged 1mm Fluid-attenuated inversion recovery (FLAIR) and T2*-weighted sequences or 'susceptibility-weighted imaging (SWI), on typical MS-suggestive lesions (ovoid, not confluent and > 3 mm in diameter). A CVS-positive MR scan, considering at least 6 CVS-positive lesions (Select-6 rule), or at least 40% of CVS-positive lesions (40% threshold rule), in patients presenting with low lesion load (Select-6 rule not applicable), reaches high sensitivity and specificity for MS diagnosis.

Introduction

The diagnostic criteria for multiple sclerosis (MS) have evolved over the years. The 2017 revision [ 1 ] combines clinical, imaging, and biological evidence and is widely used in clinical practice. They allow MS to be diagnosed early, after the first clinical event suggestive of the disease in most patients [ 2 – 4 ]. Their high sensitivity was justified by the expert panel to limit undiagnosed cases, to limit the "wait and see" attitude, and ultimately to administer disease-modifying therapies (DMT) to the majority of patients to prevent neurological disability. The consequences of the 2017 McDonald criteria were (1) a higher risk of misdiagnosis [ 5 ] associated with lower specificity compared to the 2010 McDonald criteria [ 2 – 4 ], (2) the virtual disappearance of the clinically isolated syndrome (CIS), a term used to identify patients at risk of clinical and/or MRI inflammatory activity after their first clinical event, and (3) the removal of elevated IgG Index in the cerebrospinal fluid (CSF) as a helpful diagnosis biomarker. The spectrum of MS has thus been shifted to the left.

Since 2009, the first description of the radiologically isolated syndrome (RIS) [ 6 ], defined as individuals with incidentally discovered demyelinating-appearing white matter MRI lesions in the central nervous system (CNS), the vision of MS has evolved. However, in 2017, there was not enough evidence to suggest that people with RIS (pwRIS) could be defined as having preclinical MS. Since then, our knowledge of pwRIS has increased dramatically, thanks to the work of the RIS Consortium (RISC). In 2014 and 2020, the RISC found in their international prospective cohort that 34% and 51.2% of pwRIS presented with a clinical event suggestive of MS after 5 and 10 years of follow-up, respectively [ 7 , 8 ]. Young age, spinal cord, and gadolinium-enhancing MRI lesions at diagnosis of RIS were cumulative predictors of MS within the first 2 years of follow-up [ 9 ]. In two international randomised, placebo-controlled clinical trials, teriflunomide and dimethyl fumarate, two DMTs used to treat relapsing–remitting (RR) MS, reduced the occurrence of a clinical event at week 96 in pwRIS by a factor of 4–5 [ 10 , 11 ]. Finally, the RISC shows that 29% of asymptomatic people with a low number of demyelinating-appearing MRI lesions (not meeting the 2009 RIS criteria) experienced a transition to symptomatic MS within 5 years of follow-up [ 12 ]. The number of people with symptomatic disease increased to 38% if they had a spinal cord MRI lesion with the presence of oligoclonal bands (OCB) on cerebrospinal fluid (CSF) analysis [ 12 ]. Therefore, in 2023, the RIS diagnostic criteria were revised, identifying people at risk of symptomatic MS within the first 5 years of follow-up with an accuracy of 60% [ 12 ].

Given all these data, there is strong evidence that most pwRIS have preclinical MS, shifting the MS spectrum even further to the left. In 2024, international experts from 16 countries, all with expertise in clinical management, radiology, methodology, epidemiology, and patient perspectives in MS, co-sponsored by ECTRIMS and the National MS Society, met to review and update the 2017 McDonald criteria. According to the literature update, one of the primary outcomes was the ability to diagnose MS at a preclinical stage and include new biomarkers. This manuscript has reviewed the rationale for retaining the kappa free light chain index (kFLC index) and the central vein sign (CVS) as specific biomarkers for diagnosing preclinical MS, as presented by Montalban et al. [ 13 ].

Proposed 2024 McDonald Criteria for Preclinical MS

Of the proposed changes to the McDonald criteria, including the optic nerve as a fifth topography to fulfill dissemination in space (DIS) and the need for additional biomarkers to diagnose MS in patients presenting with vascular or headaches comorbidities should be considered [ 13 ]. Preclinical MS diagnosis could be made in individuals with no symptoms suggestive of MS if they have (1) typical demyelinating white matter lesions on T2-weighted MRI images in two of the five DIS locations and (2) a specific MRI or biological diagnostic biomarker. The presence of at least one gadolinium-enhancing lesion or new T2-weighted lesion (dissemination in time, DIT), the detection of intrathecal immunoglobulin synthesis based on oligoclonal IgG bands (OCB) or kFLC index positivity, and the identification of at least six lesions harboring a CVS are the three diagnostic biomarkers that, when added to DIS, support the diagnosis of preclinical MS. In patients with comorbidities (vascular risk factors, known vascular or headache disorder) or age > 50 years old, intrathecal immunoglobulin synthesis, the presence of at least 6 CVS-positive lesions, and the presence of at least one spinal cord lesion should be considered as additional biomarkers to support the diagnosis. This definition of MS at the preclinical stage slightly differs from the RIS criteria. Table 1 resumes the 2024 preclinical MS diagnostic and 2023 RIS criteria. Compared to the 2017 McDonald criteria, two new biomarkers will be used in clinical practice to diagnose preclinical MS: the kFLC index and the CVS.

Table 1 Proposed 2024 McDonald diagnostic criteria for preclinical MS and 2023 RIS criteria Clinical symptoms 2024 MS criteria 2023 RIS criteria No typical symptom for MS T2-weighted MRI lesions DIS 1 (/5) a DIS ≥ 2 (/5) a DIS ≥ 1 (/4) b ≥ 3 2005 DIS criteria [ 132 ] Diagnostic biomarker Diagnosis cannot be made without typical symptom for MS 1 from the 3 below required: DIT c  ≥ 6 lesions with CVS  OCB or kFLC index positivity 2 from the 3 below required:  DIT c  Spinal cord lesion consistent with CNS demyelination  OCB positivity Not required Additional biomarker required if comorbidities d 1 from the 3 below required:  ≥ 6 lesions with CVS  OCB or kappa index positivity  ≥ 1 spinal cord lesion consistent with CNS demyelination Not required Not required RIS Radiologically isolated syndrome; DIS dissemination in space; CVS central vein sign; CNS central nervous system; MS Multiple sclerosis; OCB oligoclonal bands; kFLC kappa-free light chain index a According to the proposed 2024 DIS criteria, including optic nerve as a new topography b According to the 2017 DIS criteria (optic nerve topography not included) c At least one gadolinium-enhancing lesion at any time or at least one new T2-weighted lesion on follow-up MRI scan d The additional biomarker used to consider the diagnosis of MS in patients with comorbidities should differ from the one used to complete the “diagnostic biomarker”

Proposed 2024 McDonald diagnostic criteria for preclinical MS and 2023 RIS criteria

1 from the 3 below required:

DIT c

≥ 6 lesions with CVS

OCB or kFLC index positivity

2 from the 3 below required:

DIT c

Spinal cord lesion consistent with CNS demyelination

OCB positivity

1 from the 3 below required:

≥ 6 lesions with CVS

OCB or kappa index positivity

≥ 1 spinal cord lesion consistent with CNS demyelination

RIS Radiologically isolated syndrome; DIS dissemination in space; CVS central vein sign; CNS central nervous system; MS Multiple sclerosis; OCB oligoclonal bands; kFLC kappa-free light chain index

a According to the proposed 2024 DIS criteria, including optic nerve as a new topography

b According to the 2017 DIS criteria (optic nerve topography not included)

c At least one gadolinium-enhancing lesion at any time or at least one new T2-weighted lesion on follow-up MRI scan

d The additional biomarker used to consider the diagnosis of MS in patients with comorbidities should differ from the one used to complete the “diagnostic biomarker”

Ethical Approval

This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

The Rationale for Integrating the Kappa Free Light Chain Index for the Diagnosis of Preclinical MS

Cerebrospinal Fluid B-Cell Antibody Production and the Emergence of kFLC Index

It is widely reported that most people with MS (pwMS) who undergo lumbar puncture have cerebrospinal fluid (CSF) inflammation. Meningeal follicular plasmocytes and activated B cells strongly promote this inflammation [ 14 , 15 ]. In clinical practice, CSF-restricted B-cell immunoglobulin (Ig) production, also known as intrathecal Ig synthesis, is determined by the detection of surnumerous IgG production in the CSF compared to the serum by (1) measuring the IgG concentration in both compartments (IgG index calculation) or (2) the presence of CSF-restricted IgG bands (OCB) using the qualitative isoelectro focusing method, the latter being the gold standard.

The presence of OCB (type 2 or 3 profiles [ 16 ]) supports the diagnosis of MS with an overall sensitivity ranging from 80 to 100% [ 17 – 20 ] and an overall specificity ranging from 54 to 94%, depending on the control population to be compared [ 21 ]. Such discrepancy of results comes from the existence of different detection methods with different diagnostic sensitivities [ 22 ], but also from the fact that the result interpretation can be rater-dependent [ 23 , 24 ]. The use of isoelectric focusing on polyacrylamide gels as opposed to agarose and the use of silver staining as opposed to immunoblotting or immunofixation tend to achieve higher diagnostic sensitivity, but are not used in all laboratories [ 22 ].

OCB positivity is also associated with follow-up MRI and clinical activity in patients presenting with a clinically isolated syndrome (CIS), adding information to baseline MRI in predicting MS [ 20 , 25 – 27 ]. Together, these data led to the introduction of OCB as a diagnostic biomarker in the 2017 revision of the McDonald criteria [ 1 ].

In pwRIS, OCB is detected in almost 70% of the cases [ 12 ]. As described in CIS, OCB positivity is associated with a higher risk of clinical attack during follow-up and conversion to symptomatic disease, fulfilling MS diagnostic criteria [ 12 , 28 – 30 ].

However, intrathecal Ig synthesis in MS is not limited to IgG production [ 31 – 33 ]. Under physiological conditions, plasmocytes produce and secrete active Ig containing two homogeneous heavy chains (IgG, IgA, IgM, IgE, or IgD) linked by their Fab domain to 2 homogeneous light chains (kappa and lambda). Plasmocytes overproduce light chains relative to heavy chains and secrete them as inactive Ig, named free light chains (FLC). In clinical practice, the concentration of FLC can be measured in body fluids, particularly serum and CSF, to detect overproduction of one or both FLC, reflecting B cell Ig production. Since the 1990s, oligoclonal overproduction of kappa, but not lambda, FLC has been described in MS [ 34 ].

Our knowledge of FLC in neuroimmunology has evolved with both nephelometric and turbidimetric FLC quantification methods commonly used in hematological malignancies. Cerebrospinal fluid overproduction of FLC can be estimated using various formulae, the most commonly used being the FLC index, calculated from CSF and serum concentrations of FLC and albumin, using:

Some authors have suggested that the kFLC intrathecal synthesis should be evaluated using Reiber’s diagram, using the hyperbolic function as a reference, based on the fact that the variation of the FLL quotient (Q FLC ) to the albumin quotient (Q A ) is not linear, as suggested by the calculation of the FLC index [ 35 ]. Some studies have found very good results using this interpretation of kFLC results, with increased sensitivity compared to the kFLC index, but with lower specificity, while it has been found to be positive in non-MS inflammatory and non-inflammatory conditions [ 36 ]. Independent studies directly comparing the two techniques in large MS populations have found similar diagnostic accuracy with Reiber’s diagram (kFLC intrathecal fraction calculation) and the kFLC index calculation [ 37 , 38 ]. As the 2024 MS diagnostic criteria only include the kFLC index as an intrathecal kFLC synthesis biomarker, we will not elaborate on Reiber’s diagram in this review.

Diagnostic Accuracy of the kFLC Index in MS and RIS

Many multicentre cohort studies have evaluated the diagnostic performance of the kFLC index by comparing its median values between pwMS and controls [ 18 , 37 – 51 ]. These studies found a high diagnostic accuracy of the kFLC index, used as a positive or negative variable according to site-specific threshold values to discriminate pwMS. Table 2 summarizes the main findings of these studies. The overall sensitivity and specificity of a positive kFLC index for diagnosing pwMS is estimated to be 88% and 89%, respectively, similar to OCB [ 52 ]. The positivity of the kFLC index has a slightly higher sensitivity than OCB [ 38 ], probably because the kFLC concentration reflects not only IgG overproduction but also IgM and others not tested in routine CSF analysis [ 53 – 55 ]. The kFLC index has also been evaluated in primary progressive MS and pediatric-onset MS (POMS), with results similar to those in relapsing MS (RMS) [ 56 , 57 ]. Most of the kFLC index cutoffs reported to define its positivity ranged from 2.4 to 10.6, with an increased cutoff when patients with other inflammatory neurological disorders were used as controls [ 38 ]. A major challenge to the implementation of the kFLC index in clinical practice has been the lack of a consensus cut-off to define intrathecal kFLC synthesis. The discrepancy of all reported cut-offs was mainly due to the populations studied but also to how the authors dealt with undetectable CSF kFLC values. An important point is that all available technologies for measuring CSF kFLC have different low detection limits (i.e. 0.28 mg/L for Optilite and 0.09 mg/L for certain Siemens), which leads to different precision for the quantification of very low CSF kFLC levels, affecting the cut-offs obtained in these studies. In addition, the management of undetectable values was performed differently. It is clear that CSF kFLC values below the low detection limit of the analyser should be considered as negative [ 52 ]. According to a recent meta-analysis, an optimised kFLC index threshold for clinical practice could be 6.1 [ 52 ]. Of importance, most pwMS have a very high kFLC index, making the difference between the proposed thresholds negligible. A recent multicenter study shows a 98% concordance when comparing site-specific kFLC index thresholds (i.e. 4.6–7.5) to the 6.1 proposed [ 56 ], meaning that most reported cutoffs could be used to assess intrathecal Ig synthesis, with better specificity but lower sensitivity when using higher cutoffs.

Table 2 The ability of the kFLC index to separate pwMS from controls N (MS/control) Method (analyser/assay) KFLC index cutoff Accuracy Presslaueur et al. 2016 a [ 39 ] 120/60 Behring/Freelight 5.9 0.94 Christiansen et al. 2018 a [ 40 ] 133/97 SPAplus/Freelight 5.2 0.88 Vasilij et al. 2018 a [ 19 ] 101/50 Siemens/Freelight 8.8 0.85 Senel et al. 2019 [ 41 ] 75/1149 Behring/N-latex 9.6 0.94 Crespi et al. 2019 [ 43 ] 127/258 Behring/N-latex 5.0 0.93 Emersic et al. 2018 [ 133 ] 80/50 Behring/N-latex 5.3 0.96 Reiber et al. 2020 [ 35 ] 95/256 Siemens/N-latex 5.0 0.93 Gaetani et al. 2020 a [ 44 ] 64/106 Behring/Freelight 7.8 0.85 Cavalla et al. 2020 [ 45 ] 140/233 Behring/N-latex 6.2 0.85 Leurs et al. 2020 a [ 46 ] 526/219 Optilite/Freelight 6.6 0.86 Ferraro et al. 2020 [ 47 ] 146/299 Optilite/Freelight 5.8 0.73 Ferraro et al. 2020 [ 18 ] 84/458 Optilite/Freelight 6.2 0.86 Sanz Diaz et al. 2021 a [ 68 ] 36/197 Optilite/Freelight 6.6 0.90 Rosenstein et al. 2021 a [ 37 ] 223/101 Siemens/N-latex 4.6 0.90 Bernardi et al. 2022 [ 49 ] 171/235 Optilite/Freelight 6.4 0.86 Levraut et al. 2022 a [ 38 ] 765/856 Optilite/Freelight 8.9 0.89 Zondra Revendova et al. 2024 a [ 51 ] 439/1312 SPAplus/Freelight 8.9 0.93 kFLC kappa-free light chain index, pwMS people with multiple sclerosis, MS Multiple sclerosis; CIS clinically isolated syndrome a Studies that put MS and CIS in the same group

The ability of the kFLC index to separate pwMS from controls

kFLC kappa-free light chain index, pwMS people with multiple sclerosis, MS Multiple sclerosis; CIS clinically isolated syndrome

a Studies that put MS and CIS in the same group

Only one study evaluated the kFLC index in pwRIS [ 58 ], and a few others included pwRIS as MS controls [ 37 , 59 ]. In the study by Levraut et al. [ 58 ], which included 36 pwRIS, the median kFLC index was 19, with 61% of them having a positive kFLC index. The authors found that pwRIS with a positive kFLC index had a 3.5-fold increased risk of MRI activity during follow-up, supporting the diagnosis of preclinical MS. The studies of Vecchio et al. [ 59 ] and Rosenstein et al. [ 37 ] pooled patients with CIS and RIS. Both studies found a median kFLC index higher in this group than in other non-MS controls [ 37 , 59 ].

The kFLC Index Predicts Early Disease Activity in MS and RIS

OCB has diagnostic and prognostic values in MS, CIS, and RIS, predicting MRI activity and the risk of clinical relapse [ 12 , 20 , 25 – 29 ]. All studies found that the median kFLC index was higher in patients with CIS (pwCIS) who experienced a clinical relapse during follow-up than in stable patients [ 44 , 58 , 60 – 62 ]. According to the studies of Arrambide et al. [ 62 ] ( n  = 214), Gaetani et al. [ 44 ] ( n  = 23), and Levraut et al. [ 58 ] ( n  = 146), pwCIS with a positive kFLC index are more likely to develop a second clinical attack during follow-up. This increased risk was estimated to be twofold higher than in pwCIS with a negative kFLC index [ 62 ]. Some authors also suggested that using the kFLC index as a continuous variable may quantify this risk as being twofold higher every year when the kFLC index value is > 100 [ 61 ], or 40% higher every year for each increase of 100 in the kFLC index [ 58 ]. Finally, one study found that the risk of a second clinical relapse in pwCIS ( n  = 86) was higher in subjects with elevated kFLC index than in subjects with elevated neurofilament light chains (NfL) [ 63 ].

In pwMS, it has been shown that patients with an elevated kFLC index at baseline may present with more active disease (i.e. a more elevated EDSS scoring or more T1-weighted gadolinium-enhancing lesion on MRI scans), an increased risk of reaching clinical disability during follow-up, confirmed disability worsening independent of relapse activity, and a shorter time to high-efficacy DMT use [ 54 , 58 , 64 ].

In RIS, only one study has evaluated the risk of achieving DIT or clinical attack according to the kFLC index [ 58 ], and it found a 3.6-fold increased risk of DIT in pwRIS having a positive kFLC index, compared to pwRIS with a negative kFLC index [ 58 ]. However, due to a low power ( n  = 36 pwRIS, median follow-up duration of 2 years), the ability of the kFLC index to predict clinical attack in pwRIS could not be evaluated in this study. An ongoing international study ( NCT06395662 ) aims to stratify the risk of clinical conversion in pwRIS according to various serum and CSF biomarkers, including the kFLC index. The results of this study will provide further information on the ability of the kFLC index to predict clinical disease in pwRIS.

kFLC Index and OCB: Two Correlated Biomarkers Reflecting Intrathecal Ig Synthesis

Our knowledge of the kFLC index in MS is mainly based on studies that examined symptomatic patients (e.g. CIS or clinically definite MS). Very few studies have focused on pwRIS. Therefore, adding the kFLC index as a biomarker for preclinical MS could be discussed.

There is strong evidence that the kFLC index and OCB are correlated and reflect the same pathophysiological CSF process [ 43 , 47 , 65 – 67 ], with an agreement between both biomarkers of between 85 and 90%. It has even been shown that the level of the kFLC index correlates with the number of OCB [ 35 ]. Discordant cases are more likely to favour a positive kFLC index with negative OCB (higher sensitivity). Considering the high diagnostic and prognostic performance of the kFLC index in symptomatic MS, the ability of OCB to predict symptomatic MS in pwRIS, the correlation between both biomarkers, and the growing knowledge about the diagnostic and prognostic performance of the kFLC index in pwRIS, it was justified to propose both as equivalent biomarkers to detect intrathecal Ig synthesis in preclinical MS.

The remaining question could be which biomarker biologists should focus on first to detect CSF Ig synthesis. As the kFLC index is cost-effective, time-saving, and a more sensitive biomarker than OCB [ 43 , 67 , 68 ], many authors have suggested using the kFLC index as a screening tool and using OCB only in cases with high suspicion of CNS inflammation and a negative kFLC index to definitively rule out intrathecal Ig synthesis [ 50 , 52 ]. Such algorithms have been reported to be very accurate for MS diagnosis [ 50 ]. However, using the kFLC index as a first-line biomarker may lead to misdiagnosis, while kFLC index positivity with negative OCB testing is reported in patients with CNS infection [ 55 ], AQP4 IgG-positive neuromyelitis optica spectrum disorder (AQP4+ NMOSD) [ 53 , 69 ] or people living with HIV and having neurological symptoms [ 70 ]. In addition, the use of the kFLC index may add value to the OCB test in detecting intrathecal Ig synthesis in patients with equivocal isoelectric focusing results (i.e. pattern V) [ 71 ]. As recently suggested by the German CSF Society, both biomarkers could be performed at the same time, as lumbar puncture is invasive and usually done once during the diagnostic work-up [ 72 ], even though both biomarkers are highly correlated. Therefore, laboratories should use the one (or both) they are familiar with, taking into account cost, procedure time, invasiveness of the lumbar puncture and their own experience with these biomarkers.

kFLC Index Laboratory Considerations

To calculate the kFLC index, kFLC and albumin concentrations are required in both serum and CSF. The recommended tests for FLC measurement are based on turbidimetric and/or nephelometric principles. Both techniques are automated, cost-effective, and can provide reproducible FLC concentrations [ 73 ]. The two main platforms used in MS studies to measure FLC are the Optilite (Biding Site: turbidimetry) using the Freelight assay and Siemens platforms such as the BN Prospec (Siemens Healthcare Diagnostics Products: nephelometry) using the N-Latex FLC kit. An ELISA-based method has also been reported to be effective in measuring kFLC in serum and CSF [ 39 ]. Independent studies have reported that all technics have equivalent diagnostic performance in separating pwMS from controls, with excellent inter-laboratory and inter-assay concordance, especially for CSF KFLC concentrations and KFLC index [ 74 , 75 ]. The stability of samples over time has not been directly evaluated. However, some studies measured kFLC and albumin concentrations on samples frozen at – 80 °C [ 18 , 65 , 68 ], others at – 20 °C [ 76 – 78 ], and yet others on fresh samples [ 38 , 58 ]. All reported very high accuracy of the kFLC index for MS diagnosis. It has been reported that kFLC quantification is not affected by patient- or treatment-related factors such as blood contamination of the CSF [ 53 ] or moderately effective DMT [ 79 , 80 ]. However, impaired renal function and monoclonal gammopathy must be considered as influencing factors on blood kFLC concentration (i.e. increased) that could lead to false negative kFLC index calculation [ 81 ]. Also, exposure to highly effective DMT prior to lumbar puncture should be considered when interpreting the kFLC index, as these treatments have been reported to decrease CSF kFLC concentrations [ 80 ].

Conclusion About the kFLC Index in Preclinical MS

In conclusion, the kFLC index is a new biomarker with good stability and excellent inter-laboratory and inter-assay concordance, allowing the estimation of intrathecal Ig synthesis with similar performance to OCB (both biomarkers are highly correlated). It is elevated in pwMS, but also in most of pwRIS, and can predict MS (whether based on clinical or MRI criteria) in both pwRIS and CIS. For all these reasons, kFLC index positivity is included as a specific diagnostic biomarker, as OCB, in the proposed 2024 McDonald criteria [ 13 ]. Table 3 summarizes why the kFLC index has been incorporated into the proposed revised criteria.

Table 3 Key points explaining why the KFLC index positivity has been integrated into the proposed 2024 McDonald criteria for preclinical MS diagnosis kFLC index positivity achieves high sensitivity (88%) and specificity (89%) for MS diagnosis, similar to OCB The higher the kFLC index, the higher the specificity for MS diagnosis Patients with CIS with a positive kFLC index have a higher risk of developing MS (prediction of clinical relapse or MRI DIS and DIT), similar to OCB pwRIS with a positive kFLC index have a higher risk of having MRI DIT, more than OCB The higher the kFLC index, the higher the risk of MS diagnosis in CIS and RIS High agreement between OCB and kFLC index kFLC index measurement is time and cost effective compared to OCB detection kFLC measurement is fully automated, with high inter-laboratory and inter-assay concordance kFLC concentrations are stable in frozen samples kFLC kappa-free light chain index, MS Multiple sclerosis, OCB oligoclonal bands, DIS dissemination in space, DIT dissemination in time, CIS clinically isolated syndrome, RIS Radiologically isolated syndrome

Key points explaining why the KFLC index positivity has been integrated into the proposed 2024 McDonald criteria for preclinical MS diagnosis

kFLC kappa-free light chain index, MS Multiple sclerosis, OCB oligoclonal bands, DIS dissemination in space, DIT dissemination in time, CIS clinically isolated syndrome, RIS Radiologically isolated syndrome

Rationale for Integrating the Central Vein Sign for the Diagnosis of Preclinical MS

An MRI Biomarker Based on MS Plaque Morphology

Central vessels (mainly veins and venules) in MS plaques have been reported in pathological studies [ 82 , 83 ]. The perivascular space surrounding these veins is thought to be a privileged site for immune cells to interact with antigen-presenting cells, which can trigger inflammation, leading to lesion formation around the veins [ 83 – 85 ]. With the development of susceptibility-based magnetic resonance imaging, observing these central veins in MS plaques became possible using ultrahigh-field MRI [ 86 – 88 ] and, later on, 3T and even 1.5T MRI [ 89 – 92 ]. To our knowledge, there is only one report of the CVS in MS lesions located in the spinal cord [ 93 ], although there is pathological evidence of a central vein in spinal cord lesions.

MRI Sequences and Interpretation for the Assessment of the Central Vein Sign

Structural imaging of cerebral veins is best achieved using the T2*-based contrast mechanism, which exploits the magnetic properties of blood, which appears hypointense [ 94 ]. These T2*-weighted images can be post-processed using susceptibility-weighted imaging (SWI) to further enhance venous conspicuity [ 95 ]. When added to Fluid-attenuated inversion recovery (FLAIR)-weighted images, which are sensitive for detecting MS plaques, the vein can be visualised within the lesion, appearing as a hypointense point or line (depending on how the vessel crosses the lesion). Many MR sequences have been reported to be able to detect the CVS in MS lesions, some without the need for gadolinium injection, such as the FLAIR*, which combines 1 mm 3D FLAIR (for lesion detection) and 3D EPI (for vein detection) [ 96 ], and others such as SWAN-venule, which combines 1 mm 3D FLAIR and T2*-weighted angiography (SWAN) acquired after intravenous gadolinium injection [ 97 ]. To our knowledge, only one study has compared MR sequences in 3T MRI to determine which one might be more valuable in clinical practice for the detection of the CVS and found a better accuracy with optimized T2*-weighted imaging compared to SWI [ 98 ]. Similar results were found using a 7T MRI [ 99 ].

The North American Imaging in Multiple Sclerosis Cooperative (NAIMS) has produced a consensus statement to assist radiologists in the clinical assessment of the CVS [ 96 ]. Firstly, not all white matter hyperintensities on T2-FLAIR images should be retained for the CVS assessment. The presence of a central vein needs to be assessed in highly suspicious inflammatory demyelinating lesions represented by non-confluent lesions with a diameter > 3 mm. Small (< 3 mm) lesions could be interpreted as false negatives on 1.5 and 3T MRI due to lower resolution compared to 7T MRI [ 99 , 100 ]. If the lesion is eligible, the CVS should be evaluated on dedicated merged FLAIR and T2*- or SWI-weighted sequences in at least two perpendicular MRI planes, looking for a thin hypointense dot in one plane and a line in the perpendicular plane. The hypointense line should cross the lesion centrally, regardless of the lesion’s shape. Figure  1 shows examples of CVS-positive” and CVS-negative” lesions.

Central vein sign assessment on 3T MRI scan. A Axial 3D FLAIR ( on the left ) and SWAN-venule ( on the right ) MR images from a 46-year-old woman with MS (two distinct clinical events: sensitive myelitis and diplopia). The 3D FLAIR images highlights typical MS lesions (i.e. Dawson’s fingers lesion in the top left image and U-fiber lesion in the bottom left image ). This patient had also brainstem, and cervical spinal cord lesion, with 2 enhanced-lesions on T1-weighted gadolinium sequences. SWAN-venule sequences showed a positive CVS MR scan (more than 6 CVS positive lesions). She also had positive OCB and kFLC index testing. However, according to the NAIMS criteria, both selected images should not be considered for CVS assessment (confluent or non-ovoid lesion on 3D FLAIR sequences). B Axial 3D FLAIR ( on the left ) and SWAN-venule ( all 3 images on the right ) MR images from a 48-year-old asymptomatic woman, diagnosed with RIS. The SWAN-venule sequences shows typical ovoid lesions, most of them harboring a CVS ( last image on the right ). 3D FLAIR images also show typical periventricular, brainstem and cervical spinal cord lesions. The SWAN-venule sequence showed a positive CVS MR scan (more than 6 CVS positive lesions). She also had positive OCB and kLFC index. This asymptomatic patient presenting with typical lesions in 3 of the 5 DIS topography, and having intrathecal Ig synthesis and a CVS positive MR scan, fulfill 2024 McDonald criteria of preclinical MS. C Axial 3D FLAIR ( on the left ) and SWAN-venule ( on the right ) MR images from a 46-year-old woman presenting with fatigue and dizziness. She suffers from rheumatoid arthritis and high blood pressure. 3D FLAIR sequences showed no periventricular, brainstem, cerebellar or spinal cord lesions. No ovoid lesions harboured a CVS on SWAN-venule sequences. She was diagnosed as having small cerebral vessel disease

Diagnostic Accuracy of the CVS for MS and RIS Diagnosis

Many independent multicenter cohort studies have evaluated the CVS as a diagnostic biomarker in pwMS [ 90 , 98 , 101 – 107 ], but very few in pwRIS [ 89 , 108 , 109 ]. As this review focuses on the clinical application of the CVS, all studies performed on 7T MRI will not be discussed in this section. Table 4 summarises the studies evaluating the CVS as a diagnostic biomarker by comparing pwMS or pwRIS and controls using 3T MRI scans. Overall, pwMS had a higher number of lesions with a CVS than controls [ 89 , 98 , 106 , 110 – 112 ], even in POMS [ 107 ]. Attention should be paid to patients with Myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD), who may present with a challenging number of CVS-positive lesions (almost 33%, according to the study of Cagol et al.) [ 106 ]. In patients with suspected MS, adding at least one CVS-positive lesion to the MRI DIS criteria increases the diagnostic accuracy for MS from 78 to 86% [ 113 ], as reported with OCB [ 27 ]. Maggi et al. [ 114 ] showed in their study of 51 challenging cases that using the CVS could improve MS diagnosis in more than 90% of the cases. To our knowledge, no longitudinal studies have evaluated whether the proportion of CVS-positive lesions at CIS diagnosis can predict an earlier time to CDMS. Still, Cagol et al. [ 106 ] reported that the CVS was the imaging biomarker that could predict MS diagnosis with better accuracy than cortical lesions or other typical lesion locations.

Table 4 Diagnostic performances of the central vein sign in pwMS and pwRIS on 3T MRI scans MS RIS n (MS/control) 40% rule accuracy Select-6 rule accuracy n (RIS/control) 40% rule accuracy Select-6 rule accuracy Georges et al. 2016 [ 134 ] 68/19 0.86 – – – – Clarke et al. 2020 [ 135 ] 12/23 0.87 – – – – Cortese et al. 2018 [ 110 ] 18/16 NMOSD 0.83 – – – – Sinnecker et al. 2019 [ 98 ] 323/164 0.73 – – – – Suthiphosuwan et al. 2019 [ 108 ] – – – 20/0 90% sensitivity 95% sensitivity Maggi et al. 2020 [ 114 ] 51 undiagnosed 27/12 0.97 0.90 – – – Sacco et al. 2023 [ 111 ] 26/14 MOGAD 0.97 – – – – Landes-Chateau et al. 2024 [ 89 ] 45/90 0.90 0.88 45/90 0.90 0.87 Borrelli et al. 2024 [ 112 ] 185/100 0.96 0.92 – – – Cagol et al. 2024 [ 106 ] 599/452 0.82 0.84 – – – Daboul et al. 2024 [ 116 ] 37/41 – 0.82 – – – pwMS people with multiple sclerosis, pwRIS people with radiologically isolated syndrome, MS multiple sclerosis, NMOSD neuromyelitis optica spectrum disorder, MOGAD Myelin oligodendrocyte glycoprotein antibody-associated disease, RIS Radiologically isolated syndrome

Diagnostic performances of the central vein sign in pwMS and pwRIS on 3T MRI scans

51 undiagnosed

27/12

pwMS people with multiple sclerosis, pwRIS people with radiologically isolated syndrome, MS multiple sclerosis, NMOSD neuromyelitis optica spectrum disorder, MOGAD Myelin oligodendrocyte glycoprotein antibody-associated disease, RIS Radiologically isolated syndrome

Only two studies focused on pwRIS. Both found a high prevalence of CVS-positive lesions in pwRIS [ 89 , 108 ]. The study by Landes-Chateau et al. [ 89 ] reported the same proportion of CVS-positive lesions in pwRIS and age- and sex-matched pwMS. Whether the proportion of CVS-positive lesions at RIS diagnosis can predict clinical disease remains unanswered.

Application of the CVS in Clinical Practice: The Select-6 Rule

The main clinical limitations of CVS studies to date are (1) the time required for radiologists to search for CVS in each lesion on the scan and (2) the retrospective comparison of well-established MS and non-MS diagnoses, which amplifies differences and does not reflect clinical practice.

To avoid the first limitation, some authors have attempted to identify simplified algorithms for CVS detection without loss of diagnostic performance. In the Select-3 [ 104 ] and Select-6 [ 115 ] rules, i.e. the presence of at least 3 and 6 CVS-positive lesions, respectively, on one scan drastically improves MRI reading time, especially in individuals with many lesions. While Select-3 had a high sensitivity for MS diagnosis, its specificity remained low, with a significant loss of accuracy compared to the proportion of CVS-positive lesions [ 89 , 106 , 116 , 117 ]. In contrast, the Select-6 rule performs similarly predicting MS and RIS, with a slightly lower sensitivity and higher specificity than the 40% threshold used in many studies [ 89 , 106 , 116 , 117 ]. Therefore, the Select-6 rule is preferred to the 40% threshold for evaluating a CVS-positive MRI scan in patients with moderate to high lesion burden, and the 40% CVS-positive lesion is used in patients with very few lesions because the Select-6 is not applicable.

To our knowledge, only one study has evaluated the predictive value of the CVS in patients suspected of having early MS [ 117 ]. The authors found that patients with a Select-6 positive MRI scan at baseline were diagnosed with MS in 81% of the cases at diagnosis and in 95% of the cases at 1-year follow-up [ 117 ]. In addition, an ongoing multicenter study (NCT04024269) evaluates the CVS's ability to predict MS in typical pwCIS who do not meet the 2017 McDonald criteria. The results of this study will provide further information on the diagnostic performance of the CVS in CIS.

Although CVS is an interesting MRI biomarker for diagnosis, it has been shown that half of pwMS who develop new T2-weighted lesions during follow-up have only CVS-positive lesions [ 118 ]. Another study found that age and vascular risk factors (i.e. hypertension, diabetes, obesity, and smoking) were cumulative factors reducing the number of CVS-positive lesions in pwMS [ 119 ]. One interpretation could be that the CVS may differentiate asymptomatic MRI activity from non-specific white matter hyperintensities due to small cerebral vessel disease and may be helpful in the choice of DMT switch for suspected asymptomatic MS evolution. To date, no study has compared DMT switching according to the CVS status of new MRI lesions.

Conclusion about CVS in Preclinical MS

In conclusion, both the Select-6 and the 40% threshold rules performed equally to detect specific MS lesions. The sensitivity and specificity of the CVS are reported to be elevated, as is the case for the kFLC index in pwMS and pwRIS. T2*- and SWI-weighted MR sequences are available to detect the CVS, with slightly better accuracy favouring T2*-weighted images. The CVS is reported to be the more specific MRI biomarker for MS, even when compared to other lesion locations included in the DIS criteria. Finally, the pwMS and pwRIS have the same proportion of CVS-positive lesions. For all these reasons, the Select-6 rule has been incorporated into the proposed 2024 McDonald criteria for preclinical MS [ 13 ]. Table 5 summarizes key points as to why the CVS has been included in the proposed revised criteria.

Table 5 Key points explaining why the central vein sign has been integrated into the proposed 2024 McDonald criteria for preclinical MS diagnosis The CVS assessment should be performed on MS-suggestive lesions (not confluent and > 3 mm in diameter) If eligible, a lesion can be retained as vein-centred (CVS-positive) if it presents with a thin hypointense dot in one plane, and a hypointense line in the perpendicular plane. The line should crosses the lesion centrally, regardless of lesion’s shape. Lesions crossed by numerous veins should be considered as CVS-negative A CVS-positive MR scan is considered when at least 6 lesions are CVS-positive (Select-6 rule) In patients presenting with low lesion load (Select-6 rule not applicable), CVS-positive MR scan is considered when at least 40% of the lesions are CVS-positive (40% threshold rule) The CVS assessment should be performed on merged 1mm FLAIR and T2*-weighted sequences or SWI T2*-weighted sequences are slightly more accurate than SWI to visualise the CVS The Select-6 or 40% threshold rules are sensitive and very specific for MS diagnosis, and performed equally Diagnostic performances of the CVS (whether the use of the Select-6 or the 40% threshold rule) is similar in pwMS and pwRIS The CVS assessment could serve to define DIT in newly appearing white matter hyperintensities in pwMS MS Multiple sclerosis, CVS central vein sign, FLAIR Fluid-attenuated inversion recovery, SWI susceptibility-weighted imaging, pwMS people with Multiple sclerosis, pwRIS people with Radiologically isolated syndrome, DIT dissemination in time

Key points explaining why the central vein sign has been integrated into the proposed 2024 McDonald criteria for preclinical MS diagnosis

MS Multiple sclerosis, CVS central vein sign, FLAIR Fluid-attenuated inversion recovery, SWI susceptibility-weighted imaging, pwMS people with Multiple sclerosis, pwRIS people with Radiologically isolated syndrome, DIT dissemination in time

Discussion

The kFLC index and CVS are specific biomarkers for MS. Their addition to the 2017 McDonald criteria has been shown to improve MS diagnosis accuracy [ 62 , 113 ], while the 2017 criteria are reported to be highly sensitive but not specific [ 2 – 4 ]. The use of such specific biomarkers is essential in the 2024 revision of the McDonald criteria because neurologists will be able to diagnose MS earlier due to (1) the addition of optic nerve as the fifth DIS topography, (2) the ability to diagnose MS in pwCIS with only one typical DIS involvement, and (3) in asymptomatic patients (Montalban 2024). The use of additional specific biomarkers in comorbid patients is likely to improve misdiagnosis, which is reported to be common in MS [ 5 , 120 , 121 ]. This approach limits doubtful preclinical MS diagnoses in patients with vascular or headache comorbidities, while 2 of the 4 following biomarkers (positive kFLC index or OCB, 6 CVS-positive lesions, the presence of one typical spinal cord lesion, or DIT) need to be added to the DIS criteria to confirm the diagnosis [ 13 ]. By adding intrathecal Ig synthesis and 6 CVS-positive lesions, one as a specific biomarker and the other as an additional biomarker, the specificity for MS will be increased, as it is reported to be close to 100% [ 89 , 116 , 117 ]. An interesting point could be using the CVS to confirm DIT in new T2-weighted lesions, which could increase DIT’s specificity [ 118 ].

Most of our knowledge about the kFLC index and CVS comes from patients with symptomatic disease. Evidence for the use of such biomarkers in pwRIS is scarce. Nevertheless, the inclusion of both biomarkers in the preclinical MS diagnostic work-up is supported by (1) the similarity of the results obtained in pwMS and pwRIS, (2) the high specificity reported in clinical disease, suggesting that their presence in asymptomatic patients reinforces MS diagnosis, and (3) the correlation of both biomarkers with others (i.e. OCB) already known to predict the course of the disease in its early stages including RIS.

Neurologists should be aware of the mimicking diseases that may present with an elevated kFLC index or a Select-6 positive MR scan. In a small cohort, it has been reported that half of AQP4-positive NMOSD patients have a positive kFLC index [ 69 ]. This finding is confusing, whereas the prevalence of OCB positivity has been reported to be lower in NMOSD [ 122 , 123 ]. This result needs to be confirmed in larger cohorts. One explanation could be that NMOSD patients are currently associated with other autoimmune disorders (i.e. systemic lupus or primary Sjögren syndrome) known to have oligoclonal bands in one-third of the cases when neurological symptoms are present [ 124 – 126 ]. The kFLC index being more sensitive than OCBand, reflecting not only IgG intrathecal synthesis, its prevalence in central nervous system demyelination associated with connective tissue disorder should be studied. Some authors have suggested that patients with MOGAD may present with a high proportion of CVS-positive lesions [ 106 , 127 ] and that the CVS may not be able to accurately separate MOGAD from MS using the proportion of CVS-positive lesions [ 127 ]. However, given the very small number of T2-weighted lesions eligible for the CVS assessment per patient in MOGAD ( n  = 1 in the Cortese et al. study) [ 127 ], this finding should not be a clue in clinical practice. Moreover, MOGAD patients are very unlikely to present with Ig intrathecal synthesis [ 69 , 128 , 129 ]. Applying intrathecal Ig synthesis and at least 6% or 40% CVS-positive lesions in clinically challenging cases will provide high MS specificity [ 89 , 116 , 117 ] and avoid misdiagnoses. However, the sensitivity will be lower, and this combination of biomarkers needs to be confirmed in dedicated prospective studies.

In preclinical MS, neither NMOSD and MOGAD will be part of the challenging mimicking disorders, while both are symptomatic and require immunotherapy for clinical improvement [ 130 , 131 ]. MRI will be, as it actually is in pwRIS, the cornerstone of preclinical MS diagnosis, while clinical symptoms will not be specific. Differential diagnoses in pwRIS are commonly represented by vascular white matter hyperintensity and headache disorders that are unlikely to present with a positive kFLC index [ 18 , 38 , 41 ] or CVS-positive lesions [ 89 , 98 , 106 ]. Therefore, both biomarkers will undoubtedly be beneficial in separating inflammatory versus vascular MRI lesions in asymptomatic patients and be very specific for preclinical MS diagnosis.

Acknowledgments

Author Contribution

Michael Levraut and Christine Lebrun-Frenay design and draft the manuscript, Cassandre Landes-Chateau, Mikael Cohen and Lydiane Mondot review the manuscript for intellectual content.

Funding

No funding or sponsorship was received for this study or publication of this article.

Data Availability

Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.

Declarations

Conflict of Interest

Michael Levraut, Cassandre Landes-Chateau, Mikael Cohen, and Lydiane Mondot have nothing to disclose related to this study. Christine Lebrun-Frenay was part of the expert panel that reviewed and proposed the 2024 update of the McDonald criteria.

Ethical Approval

This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Patient Consent

Patients' consent was obtained for their MR images to be published anonymously.

Footnotes

Priori publication: This manuscript focuses on the 2024 McDonald criteria that were presented by Montalban X. and co-workers at the last ECTRIMS Committee 17–20 September 2024, Copenhagen, Denmark. The 2024 McDonald criteria have not yet been published and this review is based on what was presented at the 2024 ECTRIMS meeting.

References

Associated Data

Data Availability Statement

Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.