Introduction

Multiple sclerosis (MS) is a chronic inflammatory immune-mediated disease of the central nervous system (CNS) that mainly affects young adults and bears the risk of physical and cognitive disability [1].

Diagnosis of MS requires the combination of clinical signs and symptoms with paraclinical findings obtained by magnetic resonance imaging (MRI) and cerebrospinal fluid (CSF) analysis [2]. Evidence of intrathecal immunoglobulin G (IgG) synthesis in the CSF, although not specific for MS, increases diagnostic certainty in the appropriate clinical setting [3] and substitutes for dissemination in time according to current diagnostic criteria [2].

Besides establishing MS diagnosis, one of the main challenges for neurologists counselling patients with MS is weighing benefits versus risks of certain disease-modifying therapies (DMTs) [4]. An ever-increasing number of DMTs have been proven to reduce the number of relapses, accumulation of disability and brain MRI activity [5] and current treatment concepts recognize the importance of early treatment towards suppressing disease activity below the level of detectability [6]. However, the interindividual courses of MS are extremely variable [7] and there is also a certain risk for treatment-associated adverse events. Since criteria guiding decisions when to start treatment in early MS and, in case, whether to choose a moderately or a highly efficacious DMT are still controversially debated, there is an urgent need of biomarkers to predict disease activity [4, 8]. So far, the number of brain MRI lesions and the presence of intrathecal IgG synthesis in the CSF imply some prognostic value [9].

As depicted above, the value of CSF analysis for diagnosis of MS and for prediction of disease activity after the first demyelinating CNS event is unquestioned. However, for the detection of intrathecal IgG synthesis as a marker for intrathecal B cell activity, several different laboratory methods have been developed in the last half century. Quantitative methods that require the measurement of IgG concentrations in CSF and serum followed by calculation of certain formulae such as IgG index [10], Reiber [11] or Auer & Hegen formulae [12] referring patient’s individual values to a predefined upper normal limit are mainly hampered by their low sensitivity. The detection of oligoclonal IgG bands (OCB) by isoelectric focusing (IEF) followed by immunoblotting is nowadays the gold standard. This technique compares paired CSF and blood samples of each individual patient. Intrathecal IgG synthesis is present if OCB are present in CSF without corresponding bands in serum [13]. It ensures a high diagnostic sensitivity and specificity both of approximately 90% [14]. However, this method enables only a qualitative determination of intrathecal IgG synthesis (i.e., returns either a positive or a negative result), is technically demanding, time-consuming, costly and rater-dependent [13].

Kappa free light chains in the CSF as an emerging biomarker

Besides intact immunoglobulins that consist of light chains and heavy chains bound together via disulfide bonds and noncovalent interactions [15], B cells also produce light chains in 10–40% excess over heavy chains and secrete them as free forms into the blood circulation [16]. These free light chains (FLC) have a molecular weight of approximately 24 kD and consists of two immunoglobulin domains, a constant region that specifies the isotype of free light chain (either κ or λ) and a variable domain (Fig. 1; [15]). If bound, the variable light chain domain is part of the immunoglobulin antigen binding site; the function in the free forms is not fully elucidated [16]. κ‑FLC exist mainly in the form of monomers, whereas λ‑FLC are present as covalent dimers [16]. In the last few years, a multitude of studies have highlighted the value of κ‑FLC in CSF as another biomarker—instead of immunoglobulins—for intrathecal B cell activity in patients with MS, not least due to significant methodological advantages.

Fig. 1
figure 1

Schematic illustration of the molecular structure of immunoglobulins and free light chains. B cells produce (a) intact immunoglobulins and (b) in excess free light chains. Both immunoglobulins and FLC serve as biomarker for B cell activity. a Immunoglobulins consist of two identical heavy chains (blue) and two identical light chains (green). Each heavy chain consists of four immunoglobulin domains linked by a hinge region. Differences in the structure of the constant regions (CH1, CH2 and CH3) determine the isotype (IgG, M, A, D, E) and subclass of the immunoglobulin (e.g., IgG1–4), while the variable domain (VH) contributes to the antigen binding site. Each light chain consists of two immunoglobulin domains. Differences in the structure of the constant region (CL) determine the isotype of free light chain (either κ or λ), while the variable domain (VL) contributes to the antigen binding site. Accordingly, both the heavy chains and light chains form the amino-terminal variable (V) regions responsible for antigen recognition; the carboxyl-terminal constant (C) regions mediate effector functions. b Free light chains show the same structure as light chains bound within the intact immunoglobulin. FLC have a molecular weight of approximately 24 kD and consist of the two immunoglobulin domains CL and VL. Differences in the structure of the constant region (CL) determine the isotype of the free light chain (either κ or λ). Whereas κ‑FLC mainly exist in the form of monomers, λ‑FLC are present as covalent dimers. CH constant heavy chain domain, CL constant light chain domain, Fab fragment antibody binding, Fc fragment crystallisable, FLC free light chain, VH variable heavy chain domain, VL variable light chain domain

Full size image

History of FLC detection

FLCs were discovered more than 150 years ago, when in 1847 Henry Bence Jones described a protein in the urine of a patient with severe bone pain and fractures that precipitated upon addition of nitric acid [17]. The so-called Bence Jones proteins evolved to an important diagnostic marker for patients with multiple myeloma. More than 100 years after its discovery, the Bence Jones protein was identified as monoclonal FLC [18]. Developments in laboratory methods brought up protein electrophoresis and immunofixation electrophoresis; however, these methods still had limited sensitivity so that low level FLC under physiological or oligosecretory conditions, e.g., in immune-mediated diseases, were not detectable, and allowed only qualitative determination [18]. Attempts to quantify FLC were initially hindered by difficulties of producing antibodies specific to FLC that do not cross-react with light chains bound in intact immunoglobulins. The breakthrough was achieved in 2001 by Bradwell and coworkers who dissociated light chains from heavy chains and then raised antibodies directed against unique epitopes on FLC that are normally “hidden” in the conformational structure of an intact immunoglobulin [19]. These anti-human FLC-specific antibodies could then be used to develop assays that exclusively detect FLC at least a hundred times more sensitive than previous methods with detection limits down to approximately 1 mg/L. Nowadays, FLC can be measured in serum as well as in CSF by use of two types of detection antibodies: either polyclonal [19] or monoclonal [20] detection antibodies.

Elevated κ-FLC in the CSF of patients with MS

κ‑FLC in the CSF—similar to immunoglobulins or other proteins—originate either from blood by diffusion across the blood–CSF barrier or are produced within the intrathecal compartment under pathological conditions [21]. Conceptually, it seems necessary to determine the locally synthesized κ‑FLC fraction separate from the blood-derived fraction (as it is also done for IgG). Most studies calculated the κ‑FLC index that considers the CSF/serum albumin quotient (Qalb) which is an established marker for the blood–CSF barrier function [22] and corrects for the absolute serum κ‑FLC level. The κ‑FLC index is determined by following formula [23, 24]:

$$\kappa -FLC\,\textit{index}=\frac{\kappa -\mathrm{FLC}_{\mathrm{CSF}}/\kappa -\mathrm{FLC}_{\text{Serum}}}{Q_{\mathrm{alb}}}$$

It has been consistently shown that the κ‑FLC index reaches a high diagnostic accuracy to identify patients with MS. An overview of current evidence—retrieved in a systematic literature search [23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40]—is provided in Table 1. For the κ‑FLC index, diagnostic sensitivity ranges from 52 to 98% (weighted average: 87%) and specificity ranges from 68 to 100% (weighted average: 89%). For OCB, sensitivity ranges from 37 to 100% (weighted average: 84%) and specificity from 74 to 100% (weighted average: 90%). The reported sensitivity of OCB is in accordance with a previous meta-analysis [41]. Applying a difference-in-differences model showed that the mean difference of diagnostic sensitivity between κ‑FLC index and OCB was +2% and of specificity was −2%, i.e., overall the diagnostic accuracy of κ‑FLC index and OCB was equal.

Table 1 Diagnostic value of κ‑free light chain index in patients with multiple sclerosis
Full size table

The wide range of diagnostic sensitivity and specificity for both the κ‑FLC index and OCB arises from a certain heterogeneity between studies. It is evident that specificity of κ‑FLC index is lowered when patients with inflammatory neurological disease (IND) were included into the control group. κ‑FLC in CSF are—similar to CSF-restricted OCB—a sign of intrathecal inflammation and thus can support the diagnosis of MS, but they are not specific for MS. The spectrum of diseases which show intrathecal κ‑FLC synthesis is probably identical to that with CSF-restricted OCB, even though studies on the frequency of intrathecal κ‑FLC synthesis in other neurological disease are still rare. Apart from a mixture of different IND as part of control populations (Table 1) that had κ‑FLC synthesis in up to 32%, dedicated disease-specific studies exist only for a few entities, e.g., neuroborreliosis [43, 44].

κ-FLC index associated with early MS disease activity

There are only a few studies on the predictive value of the κ‑FLC index in MS. An overview is given in Table 2. These studies reported that the presence of intrathecal κ‑FLC synthesis is associated with conversion from CIS to MS [45,46,47,48,49] and that the κ‑FLC index predicted the time to conversion to MS as well as disability progression [49, 50]. However, these studies had some methodological limitations. A multivariate approach that considers other already known risk factors especially MRI activity is critical to identify the independent prognostic effect of the κ‑FLC index and to weigh its impact on the outcome.

Table 2 Prognostic value of κ‑free light chain index in patients with multiple sclerosis
Full size table

There is one recent study that fulfills these requirements providing class II evidence that in patients with early MS, high κ‑FLC index is an independent risk factor for early second clinical attack. A cohort of 88 patients with a first CNS demyelinating event (mostly monofocal, 45% myelitis, 30% optic neuritis, 24% affection of brainstem/cerebellum), at a mean age of 33 years and with a female predominance of 68% were followed over 4 years. In all, 38 (43%) patients converted to clinically definite MS (CDMS) within the observation period. In multivariate Cox regression analysis adjusting for age, sex, MRI lesion load and activity at baseline, administration of corticosteroids at baseline and DMT during follow-up revealed that κ‑FLC index predicts time to second clinical attack. This study showed that patients with κ‑FLC index > 100 at baseline had a twice as high probability for a second clinical attack within 12 months than patients with low κ‑FLC index; within 24 months, the chance in patients with high κ‑FLC index was 4 times as high as in patients with low κ‑FLC index. The median time to second attack was 11 months in patients with high κ‑FLC index, whereas 36 months in those with low κ‑FLC index [52].

Advantages of κ-FLC index compared to OCB

Current evidence suggests that determination of κ‑FLC index has some advantages over OCB detection. Even though it seems that there is no relevant difference with regard to diagnostic accuracy (Table 1), κ‑FLC can be easily measured by nephelometry which is—in contrast to the detection of OCB—a reliable, labor-saving and cost-efficient method [20]. Moreover, κ‑FLC index returns a metric result covering a range from approximately 1 up to 500 [23], i.e., it is a quantitative parameter, while OCB status is dichotomous returning either a positive or negative result as assessed by visual inspection [13]. The advantage of a metric result seems important especially for predicting disease activity. In the most recent study on the predictive value of κ‑FLC index—as previously mentioned [52]—which included patients with a first CNS demyelinating event, OCB were detected in 95% of patients who converted to CDMS during the 4‑year follow-up (CDMS converters), whereas non-converters were OCB positive also in 86% of cases. As a continuous variable, κ‑FLC index overcame the weak performance of OCB by further stratification. κ‑FLC index also significantly differed between OCB-positive CDMS converters and OCB-positive nonconverters and predicted CDMS conversion also within the cohort of OCB-positive patients [52]. Despite these promising results and clear methodological advantages of κ‑FLC index over OCB, the latter is still considered the gold standard. Before κ‑FLC index might be introduced into clinical routine, a few issues still need to be clarified, e.g., whether calculation of intrathecal κ‑FLC synthesis is superior to determination of absolute κ‑FLC concentrations in CSF or which cut-off should be applied. These two open issues are discussed in the following.

Open issues

Determining κ-FLC index or absolute CSF κ-FLC values

As mentioned above, one might argue that determining the locally synthesized fraction of κ‑FLC separate from the blood-derived fraction is necessary to capture an intrathecal inflammatory process. And indeed, the majority of studies used the κ‑FLC index (Table 1) or calculated an intrathecal κ‑FLC fraction by empirically determined Qalb-dependent reference limits [25, 43, 51, 53], whereas some studies included the CSF/serum κ‑FLC ratio (Qκ‑FLC) [30, 38, 54,55,56,57,58]. Other authors determined the absolute CSF κ‑FLC concentrations [31] arguing that the contribution of blood-derived FLC to the total CSF FLC concentration is low in cases with intrathecal synthesis. In fact, the intrathecal fraction of κ‑FLC is greater than 80% in most MS patients [23], and around 15% of CIS/MS patients showed even higher absolute κ‑FLC concentrations in CSF than in serum that proves an intrathecal synthesis per se [59]. To further elaborate this research question, a recent study separated patients into low and high CSF κ‑FLC categories (based on median values) and observed that CSF κ‑FLC concentration, Qκ‑FLC and κ‑FLC index showed similar diagnostic performance in the high category, but not in the low category with inferiority of CSF κ‑FLC and to some extent also of Qκ‑FLC. This is in line with a previous study reporting that QFLC depends almost exclusively on the amount of intrathecally synthesized FLC in cases of intrathecal B cell activity (defined by presence of oligoclonal FLC bands), whereas a correlation of Qalb and QFLC was observed in cases of absent intrathecal B cell activity (defined by negative oligoclonal FLC bands) [56]. Thus, there is evidence that the impact of serum κ‑FLC levels and Qalb is negligible in patients with strong intrathecal κ‑FLC synthesis, but probably not in patients with only low or modest intrathecal κ‑FLC production. Further studies applying multivariate statistics are required to compare these different approaches.

Establishing cut-off values

Before κ‑FLC index can be introduced into clinical routine, cut-offs have to be established. Different cut-off values might apply depending on the clinical question, e.g., to provide an upper reference limit as determined in a control population (either healthy or e.g. a symptomatic control [42]), to differentiate MS from other IND or to classify patients according to their risk for future disease activity. The so far published cut-off values differentiating MS from other neurological diseases ranged from 3.09 to 12.45 (Table 1). As κ‑FLC index values indeed vary between diseases with high values in MS, followed by other IND and then by non-IND [32, 35, 60], different cut-off values might be useful. For example, one study showed that patients with MS had κ‑FLC index of approximately 90, whereas patients with neuromyelitis optica spectrum disease that is relevant differential diagnosis of MS had values of roughly 20 and control patients values of 4 [60].

Studies that address reproducibility of κ‑FLC index using different assays, platforms and cut-offs between centers are needed, too. Although some work has already been performed in terms of absolute serum κ‑FLC concentrations, this is still lacking for κ‑FLC index. κ‑FLC index might show different robustness, as a ratio (of the CSF/serum κ‑FLC concentration, used for calculation of the κ‑FLC index) is usually less prone to laboratory variations.

Conclusions

κ‑FLC are a promising biomarker that might replace OCB detection. With regard to its diagnostic value, κ‑FLC index shows a high accuracy similar to that of OCB, but has also significant methodological advantages as an easy, reliable, fast, labor- and cost-saving method. With regard to its prognostic value, the benefit could evolve—either stand alone or in combination with others—to identify early MS patients with a higher risk for further disease activity, e.g., shorter time to a second attack. These patients could be advised to start DMT early or use highly effective DMT, as there is evidence that the time to the second attack has a prognostic impact on long-term disability [61, 62] and that early treatment significantly delays conversion to CDMS as well as disability progression [63,64,65]. Conversely, there is a certain proportion of patients showing a mild disease course who may not need a potentially harmful, psychologically distressing and, last but not least, costly DMT.

Whereas the high diagnostic value is already supported by a multitude of studies, further studies are still required to replicate the independent prognostic value of κ‑FLC index in early MS. Apart from harmonization efforts as depicted above to establish a widely applicable cut-off to definite positivity, potential influential factors such as corticosteroid treatment [52, 66], DMT or different disease phases (relapse versus stable remission) on κ‑FLC index also have to be explored.

Thus, there is convincing evidence that κ‑FLC index reliably indicates intrathecal inflammation in MS, might replace OCB determination and probably takes us one step closer to tailored medicine in MS.