**1. Introduction**

The normal central nervous system (CNS) is completely devoid of resident plasma cells, so finding infiltrating B‐cells in the CNS is rather exceptional. Infiltrating activated CD20+CD23+ B‐cells in brain parenchyma number to less than 0.1–1.5 cell/cm2 [1–3]. Therefore, no intrathe‐

© 2016 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

calimmunoglobulin(Ig) synthesis (ITS)occurs inthebasal state, andlocallyproducedIgs testify to former or active CNS infection or inflammatory processes. Therefore, while it is easy to demonstrate high levels of ITS, quantifying low levels poses technical problems.

Although ITS is one of the most sensitive biological clues for MS, it was long considered to be nonspecific or a bystander of CNS inflammation since the latter mainly involves T‐cells and no specific antibody target has yet emerged. The recent success of anti‐CD20 therapy in relapsing and progressive MS sheds light on a more central role of non‐CNS B‐cells in the pathophysiology of MS. However, progressive MS pathophysiology is still poorly understood and CNS resident B‐cells may play an even greater role. Since locally synthesized Igs are the main B‐cell by‐products, ITS offers a good opportunity for studying resident B‐cells. The role of cerebrospinal fluid (CSF) B‐cells has not yet been elucidated and is probably diverse as follows: (1) cytotoxic effect of antibodies; (2) local antigen presentation to T‐cells; and (3) secretion of cytokines playing a complex regulating role, including a possible Ig‐unrelated CNS cytotoxicity. Therefore, CNS‐trapped B‐cells not only synthesize oligoclonal bands (OCBs) but might also play a pivotal role in the slow‐burning CNS inflammation.

### **2. Quantitative measures of CSF IgG**

#### **2.1. Pitfalls in assessing ITS level**

Synthesis of Igs does not occur in the normal CNS and the tiny Ig concentration measured in normal CSF reflects a low‐rate passive diffusion through the blood‐brain barrier (BBB) to CSF. It was long thought that although virtually all the molecules may diffuse from serum to CSF, BBB permeability positively correlated with the molecular weight [4]. For example, the ratio decreases from 1:205 with albumin (65 kDa) to 1:440 with IgG (150 kDa) and 1:900 with IgM (970 kDa) [5]. Moreover, the permeability of the BBB commonly increases during CNS pathologies, leading to an increase in CSF concentrations of blood‐borne proteins and Igs. As

**Figure 1. CSF IgG passively diffused from blood to CSF in normal population. Left panel (A). Plot of CSF/serum quotients with hyperbolic function of quotient ratios ("'Reibergram")**. Reference range defined by *Q*Lim : *Q*mean±2SD or ±3SD involving, respectively, 96 and 99% of the normal population. Dotted line is the upper normal limit (>0.7) of IgG index, intersecting the *Q*Lim curve at two points. **Right panel (B). Probability curve of basal** *Q***IgG for a given** *Q***Alb**. The maximum probability is obtained for *Q*mean. Points are obtained from a simulated healthy population.

a consequence, the level of intrathecally synthesized Igs related to CNS inflammation is obscured by the passively diffused CSF Igs concentration and requires a mathematical approach that takes into account the permeability and blood concentration of the targeted molecules. The albumin quotient (or ratio), *Q*Alb = [AlbCSF]/[Albserum], is indicative of BBB dysfunction and increases with its permeability. It is also influenced by the patient's age and underlying CNS pathologies. In the basal state, which is devoid of intrathecal IgG synthesis, the ratio [IgGCSF]/[IgGserum] is proportional to *Q*Alb.

calimmunoglobulin(Ig) synthesis (ITS)occurs inthebasal state, andlocallyproducedIgs testify to former or active CNS infection or inflammatory processes. Therefore, while it is easy to

Although ITS is one of the most sensitive biological clues for MS, it was long considered to be nonspecific or a bystander of CNS inflammation since the latter mainly involves T‐cells and no specific antibody target has yet emerged. The recent success of anti‐CD20 therapy in relapsing and progressive MS sheds light on a more central role of non‐CNS B‐cells in the pathophysiology of MS. However, progressive MS pathophysiology is still poorly understood and CNS resident B‐cells may play an even greater role. Since locally synthesized Igs are the main B‐cell by‐products, ITS offers a good opportunity for studying resident B‐cells. The role of cerebrospinal fluid (CSF) B‐cells has not yet been elucidated and is probably diverse as follows: (1) cytotoxic effect of antibodies; (2) local antigen presentation to T‐cells; and (3) secretion of cytokines playing a complex regulating role, including a possible Ig‐unrelated CNS cytotoxicity. Therefore, CNS‐trapped B‐cells not only synthesize oligoclonal bands

demonstrate high levels of ITS, quantifying low levels poses technical problems.

(OCBs) but might also play a pivotal role in the slow‐burning CNS inflammation.

Synthesis of Igs does not occur in the normal CNS and the tiny Ig concentration measured in normal CSF reflects a low‐rate passive diffusion through the blood‐brain barrier (BBB) to CSF. It was long thought that although virtually all the molecules may diffuse from serum to CSF, BBB permeability positively correlated with the molecular weight [4]. For example, the ratio decreases from 1:205 with albumin (65 kDa) to 1:440 with IgG (150 kDa) and 1:900 with IgM (970 kDa) [5]. Moreover, the permeability of the BBB commonly increases during CNS pathologies, leading to an increase in CSF concentrations of blood‐borne proteins and Igs. As

**Figure 1. CSF IgG passively diffused from blood to CSF in normal population. Left panel (A). Plot of CSF/serum quotients with hyperbolic function of quotient ratios ("'Reibergram")**. Reference range defined by *Q*Lim : *Q*mean±2SD or ±3SD involving, respectively, 96 and 99% of the normal population. Dotted line is the upper normal limit (>0.7) of IgG index, intersecting the *Q*Lim curve at two points. **Right panel (B). Probability curve of basal** *Q***IgG for a given** *Q***Alb**.

The maximum probability is obtained for *Q*mean. Points are obtained from a simulated healthy population.

**2. Quantitative measures of CSF IgG**

**2.1. Pitfalls in assessing ITS level**

52 Trending Topics in Multiple Sclerosis

The daily IgG synthesis rate can be assessed by Tourtellotte's formula (median value 29 mg/ day), but inconsistent values are obtained at the individual level owing to assumptions [6]. The linear approach where the IgG index = *Q*IgG/QAlb (normal values <0.7) does not take into account either normal *Q*IgG variance nor nonlinear correlations with *Q*Alb, leading to an approximation around the limit (**Figure 1**, left panel). Moreover, since QAlb increases with age, the IgG index is thought to decrease mechanically without any change in ITS [7]. Data are best fitted by an empiric hyperbolic function (the "Reibergram") [8] whose constant parameters were fitted with a large dataset of 4154 control patients [9]. From a theoretical point of view, the hyperbolic function is the application of Fick's laws of diffusion applied to albumin and Igs [9]. For a given *Q*Alb in a population of normal patients, the distribution of *Q*IgG follows a normal law around the mean curve as *Q*mean=*f*(*Q*Alb) (**Figure 1**, right panel) [9]. With intrathecal synthesis (ITS), the IgG concentration in the CSF is the sum of IgG passively diffused from the blood and synthesized intrathecally

$$\underline{\mathbf{Q}}\_{\mathrm{lgG}} = \left( \left[ \underline{\mathbf{I}} \, \mathbf{g} \, \mathbf{G}\_{\mathrm{CSF\\_passive}} \right] + \left[ \underline{\mathbf{I}} \, \mathbf{g} \, \mathbf{G}\_{\mathrm{CSF\\_Loc}} \right] \right) \\ / \left[ \underline{\mathbf{I}} \, \mathbf{g} \, \mathbf{G}\_{\mathrm{serum}} \right] = \underline{\mathbf{Q}}\_{\mathrm{gbG\\_pass}} + \underline{\mathbf{Q}}\_{\mathrm{lgb\\_Low}} $$

where *Q*IgG\_basal is the *Q*IgG of the same patient before the onset of ITS. In clinical practice, only *Q*IgG is directly available but not *Q*IgG\_basal, so the exact IgGCSF\_Loc concentration can only be approximated by using Reiber's discrimination curve. The upper limit of the reference range, *Q*Lim, is usually set arbitrarily as *Q*mean+3SD and involves >99% of the normal population. By applying this definition, intrathecal IgG synthesis is considered to be present when *Q*IgG> *Q*Lim. The major drawback of this reference range is the loss of sensitivity for cases displaying a low level of ITS (*Q*Lim> *Q*IgG> *Q*IgG\_basal). In common practice, demonstrating ITS in such cases requires a CSF‐restricted OCBs positivity. As expected, restricting the reference range to *Q*Lim+2SD instead of +3SD increases the percentage of abnormal *Q*IgG in MS cohorts by 6–10% for IgG and up to 20% for IgM, which increases the risk of false positivity to 4% [7].

The true amount of intrathecally (or locally) synthesized IgG should be calculated as follows:

$$\operatorname{IgG}\_{\text{Loc}}\left(\mathsf{mg}\,\,\middle|\,\,L\right) = \left(\mathcal{Q}\_{\text{lg}\,\text{G}} - \mathcal{Q}\_{\text{lg}\,\text{G}\_{-}\,\text{basal}}\right) \ge \left[\operatorname{Ig}\,\mathcal{G}\_{\text{surn}}\right].$$

However, since *Q*IgG\_basal is unavailable, it can be replaced by either *Q*Lim or *Q*mean. As previously demonstrated, replacing *Q*IgG\_basal by *Q*Lim confers maximal specificity in single patient studies but with the drawback of an unavoidable underestimation of ITS. The range of IgGLoc underestimation may vary widely from 1 to 50 mg/L according to *Q*Alb (unpublished results). In cohort studies, *Q*IgG\_basal is advantageously replaced by *Q*mean, which provides a closer estimation of the exact IgGLoc [7].

Lastly, inter‐assay variability may directly impact *Q*IgG or the IgG index. For example, a 10% decrease in serum IgG directly means a 10% increase in IgG index, and 10% variations of the IgG index from day to day are commonly reported [7, 10]. As a consequence, minor fluctuations in IgG levels may be translated into normal or abnormal *Q*IgG results, although the intrathecal IgG synthesis rate is not really impacted. A final pitfall of ITS assessment relates to the properties of CNS‐targeting antibodies themselves, which are capable of brain adsorption that can potentially abolish low levels of specific antibodies that are synthesized locally or spill over to the CSF [11].

#### **2.2. Predicted changes in ITS measures in response to treatments**

Using formulas and normal values obtained from the literature, we simulated results of a cohort with tunable ITS level (unpublished results). This model provides the advantage of being able to compare the calculated (approximated) IgGLoc with the fixed IgGLoc.

IgGLoc estimation based on *Q*mean fitted well with the exact IgGLoc, even in small cohorts and for small ITS (<1 mg/L). On the other hand, individual or cohort estimations of IgGLoc based on *Q*Lim were strongly biased in a range dependent on *Q*Alb.

**Figure 2. Effect of plasma exchange on** *Q***IgG**. Simulated population with ITS +5 mg/L. *Q*IgG values increase to abnormal range whereas ITS level remains unchanged.

Plasma exchange depletes both serum IgG and IgG passively transferred through the BBB. Therefore, assuming that ITS remains constant during the procedure, the contrast in locally synthesized and passively diffused IgG in CSF is dramatically tuned by plasma exchange (**Figure 2**). For example, after a 90% decrease in [IgGserum], IgG in CSF originates almost entirely from local synthesis. The IgG index increases whereas the precision of the IgGLoc calculation (based on Qmean) remains unchanged.

For a given level of ITS, a decrease in BBB permeability decreases QIgG\_basal in a nonlinear response whereas *Q*Loc remains constant. Therefore, *Q*IgG and the IgG index may shift to abnormal values and IgGLoc(*Q*Lim) increases incorrectly, whereas IgGLoc(*Q*mean) remains con‐ stant.

As a consequence, monitoring of the ITS level should be based on the unbiased IgGLoc(*Q*mean).
