**4. Intrathecal Ig synthesis is a robust and predictive marker in MS**

#### **4.1. Factors influencing ITS level**

ITS in MS is characterized by numerous unambiguous OCBs (≥10 OCB) that are highly specific of MS [24]. Monoclonal bands are exceptional and should be considered carefully. Repeated LP in these atypical cases demonstrates a broadening of ITS or a different diagnosis [25]. Owing to their lack of specificity [25], ≥2 OCB are required in most studies to define an oligoclonal

Absence of OCB (type 1 pattern) is observed in less than 10% in recent studies, whereas most MS patients display the type 2 pattern and only some have type 3 (**Figure 3**). Among OCB+ patients, patterns 2 and 3 are observed in about 90 and 10% in relapsing‐remitting (RR) and secondary progressive (SP) patients, respectively, whereas the distribution is 40 and 60% in primary progressive (PP) patients [27], suggesting that pattern responses may be linked to

**Figure 3. Types of isoelectric focusing patterns on agarose gels** [26]: Type 1, no OCB in CSF or serum (normal pat‐ tern). Type 2, OCB restricted to CSF, absent in serum. Indicates low‐level ITS. Typical pattern in MS. Type 3, Identical OCB in both serum and CSF with extra bands in CSF. Pattern seen during systemic synthesis associated with intrathe‐ cal synthesis. Seen in MS. Type 4, OCBs in CSF mirror those in serum. Indicates a systemic IgG synthesis and passive transfer of OCB from blood to CSF, without any local synthesis. From a purely theoretical point of view, a low local synthesis with similar B‐cell clones cannot be completely deciphered from this situation. Type 5, Ladder‐type identical OCB in both serum and CSF typically associated with monoclonal IgG proteins. Peripheral IgG synthesis without local

OCB are almost always present when *Q*IgG> *Q*Lim, but are also commonly present when *Q*IgG *Q*Lim [26, 28, 29]. Nonetheless, about 5–10% of MS patients fail to demonstrate any OCB or an elevated IgG index. Nevertheless, the question remains whether ITS is really absent from such patients' CSF. Our short review argues for a probable faint ITS in the rare 'CSF‐negative'

A high CSF IgA synthesis has been demonstrated in a few patients lacking the classical IgG synthesis [30, 31]. Free light chain (FLC) sensitivity seems to be near 100% but the (expected) specificity lower than OCB makes FLC less useful for routine clinical purposes [32, 33]. Oligoclonal κ‐FLCs are detected in about 50% of MS patients without OCB [34, 35]. Isoelectric focusing with affinity blotting against known antigens overcomes the limitation induced by

**3.2. Absence of OCB in MS patients is a technical limitation**

patients, making ITS the most valuable marker of MS to date.

pattern with optimal specificity.

56 Trending Topics in Multiple Sclerosis

pathophysiology.

synthesis.

About 95% of patients display OCB but a quantifiable IgG synthesis only occurs in about 70% of them. Several studies have shown that the level of ITS is highly variable from one patient to another, which suggests a genetic determinism. Females are more prone to OCB positivity (92% vs 84%) and to a higher IgG index (1.32 ± 0.92 vs 1.03 ± 0.54) [46, 47]. An eth‐ nic influence on IgG index level and OCB status has also been confirmed by several studies. Black‐ascending patients are more prone to a higher IgG index, a higher rate of IgG synthe‐ sis, and OCB positivity [48, 49]. The IgG index level is still higher in Afro‐Americans after adjustment for OCB status [49]. The association of female sex to higher IgG index was also confirmed in this group [49].

Prevalence of abnormal OCB status and IgGLoc is lower in Asian patients than in Caucasian patients [50–52]. In Asian patients having the strongest genetic risk factors (DRB1\*1501), the prevalence of OCB positivity remains lower than in Caucasians (<72% vs 97%) [53, 54]. Moreover, irrespective of genetic background, extreme latitudes may exert a positive influence on OCB status whereas tropical latitudes exert a negative influence [53, 55].

Several studies have confirmed a strong association between OCB status and human leukocyte antigen (HLA). The DRB1\*15 allele confers a higher risk of OCB positivity whereas the genotypes DRB1\*04:04/\*04:05 or DRB1\*03:01/\*04:01 are associated with OCB negativity [56, 57]. Interestingly, DRB1\*1501 also increases the prevalence of abnormal IgG and IgA indexes but not of the IgM index [54]. In Asian patients, even though these determinants are conserved, OCB status is also driven by a strong interaction with latitude [53].

Genetic markers of γ chains (GMs) display 18 serologic specificities and all of the variants but two are expressed in the Fc region of γ chains. Some of these variations have been shown to strongly influence antibody‐dependent cell‐mediated cytotoxicity (ADCC) and complement‐ dependent cytotoxicity (CDC) [58]. In an animal model of rheumatoid arthritis, the recognition of collagen‐II epitopes by VH chains strongly depended on GM haplotypes [59]. Interestingly, GM haplotypes are in almost total linkage disequilibrium among world populations and are therefore typical of geographic origin unless there is a genetic admixture [60]. Gm21\* haplo‐ types were associated with high ITS in a Caucasian population although Gm5\*;3 haplotypes are associated with a low level of ITS96, 97. Studies are required concerning non‐Caucasian (African and Asian descent) haplotypes. Gm haplotypes are neither correlated with serum IgG level nor with the risk of MS [61, 62]. The causes of interaction between GM allotypes and ITS level remain elusive and the relationship between allotypes and CNS FcRn function (which enhances IgG clearance from CSF via Fc interaction) is unknown. Other non‐HLA unknown genes also influence OCB status [56, 63].

Healthy siblings of MS patients display a hyperimmune condition termed "MS immunopathic trait" that is characterized by (1) one or more CSF OCB, (2) an exaggerated response to a variety of viral antigens, and (3) an increased BBB permeability [64, 65]. CSF OCB were found in up to 19% of MS siblings in contrast to 4% of an unrelated control population [66, 67], some of them also displaying an elevated IgG index [66] and up to half of these siblings having both OCB‐ and measles‐specific IgG local synthesis [64, 67]. It is not clear whether ITS observed in these cases reflects a presymptomatic MS in high‐risk patients, elite non‐progressors, or an unrelated predisposition to ITS.

#### **4.2. ITS is robust over time**

ITS occurs as a very early disease event. In pediatric cohorts with very early onset MS (before 6 years), only 8% showed IgG OCB, a figure increasing to 90% with 69% of intrathecal production in two early onset cohorts (<15 y) and even more after a further relapse [68, 69]. In adults, the proportion of OCB positivity tends to appear at re‐examination [70] or increases as the disease progresses (RR = 325/360, SP: 25/25, and PP: 39/39) [27]. In a longitudinal study of CSF from 19 patients taken at multiple time points up to 12 years (mean 8 years), OCB never abated with time and demonstrated a robust pattern [71]. Most of the changes concerned band intensity and the acquisition of a new band [25, 72], although band loss was sometimes described in earlier reports [25, 71]. The same stability of OCB over time was observed with IgG, IgM, and IgA [13, 71, 73].

adjustment for OCB status [49]. The association of female sex to higher IgG index was also

Prevalence of abnormal OCB status and IgGLoc is lower in Asian patients than in Caucasian patients [50–52]. In Asian patients having the strongest genetic risk factors (DRB1\*1501), the prevalence of OCB positivity remains lower than in Caucasians (<72% vs 97%) [53, 54]. Moreover, irrespective of genetic background, extreme latitudes may exert a positive influence

Several studies have confirmed a strong association between OCB status and human leukocyte antigen (HLA). The DRB1\*15 allele confers a higher risk of OCB positivity whereas the genotypes DRB1\*04:04/\*04:05 or DRB1\*03:01/\*04:01 are associated with OCB negativity [56, 57]. Interestingly, DRB1\*1501 also increases the prevalence of abnormal IgG and IgA indexes but not of the IgM index [54]. In Asian patients, even though these determinants are conserved,

Genetic markers of γ chains (GMs) display 18 serologic specificities and all of the variants but two are expressed in the Fc region of γ chains. Some of these variations have been shown to strongly influence antibody‐dependent cell‐mediated cytotoxicity (ADCC) and complement‐ dependent cytotoxicity (CDC) [58]. In an animal model of rheumatoid arthritis, the recognition of collagen‐II epitopes by VH chains strongly depended on GM haplotypes [59]. Interestingly, GM haplotypes are in almost total linkage disequilibrium among world populations and are therefore typical of geographic origin unless there is a genetic admixture [60]. Gm21\* haplo‐ types were associated with high ITS in a Caucasian population although Gm5\*;3 haplotypes are associated with a low level of ITS96, 97. Studies are required concerning non‐Caucasian (African and Asian descent) haplotypes. Gm haplotypes are neither correlated with serum IgG level nor with the risk of MS [61, 62]. The causes of interaction between GM allotypes and ITS level remain elusive and the relationship between allotypes and CNS FcRn function (which enhances IgG clearance from CSF via Fc interaction) is unknown. Other non‐HLA unknown

Healthy siblings of MS patients display a hyperimmune condition termed "MS immunopathic trait" that is characterized by (1) one or more CSF OCB, (2) an exaggerated response to a variety of viral antigens, and (3) an increased BBB permeability [64, 65]. CSF OCB were found in up to 19% of MS siblings in contrast to 4% of an unrelated control population [66, 67], some of them also displaying an elevated IgG index [66] and up to half of these siblings having both OCB‐ and measles‐specific IgG local synthesis [64, 67]. It is not clear whether ITS observed in these cases reflects a presymptomatic MS in high‐risk patients, elite non‐progressors, or an

ITS occurs as a very early disease event. In pediatric cohorts with very early onset MS (before 6 years), only 8% showed IgG OCB, a figure increasing to 90% with 69% of intrathecal production in two early onset cohorts (<15 y) and even more after a further relapse [68, 69]. In adults, the proportion of OCB positivity tends to appear at re‐examination [70] or increases as

on OCB status whereas tropical latitudes exert a negative influence [53, 55].

OCB status is also driven by a strong interaction with latitude [53].

confirmed in this group [49].

58 Trending Topics in Multiple Sclerosis

genes also influence OCB status [56, 63].

unrelated predisposition to ITS.

**4.2. ITS is robust over time**

When focusing on the clonal repertoire of CSF Ig, the highest number of clones was found in patients with the longest disease duration, suggesting a continuous clonal expansion over time [74]. However, the low number of patients included cannot totally rule out disease heteroge‐ neity [74]. In a longitudinal study of CSF B‐cells in two MS patients, clonal rearrangements on VH (VH3 and VH4), D segment, JH, and HCDR3 sequences were conserved at two time points 1 and 4 years apart [75]. The antibody index of phagotopes from the entire panel of clones extracted from a single patient was stable at 2 years [76]. The peptidic targets of the OCB IgG are constant over time [72] and the activity index (AI) against infectious agents also remained stable over several years [77, 78]. These findings are in line with a slightly delayed oligoclonal immune response immediately after the onset of MS. However, once initiated, CSF IgG secretion persists over time indefinitely and is little altered qualitatively at the clonal level [71]. Other CSF parameters such as albumin index, IgG, and IgM index remained essentially the same in the clinically isolated syndrome (CIS), RR‐MS, and progressive MS groups [27, 79].

A few exceptions should be mentioned as follows: (1) IgM OCBs behave differently, if absent at onset, no further IgM OCBs appear [73]; when present, they tend to disappear after a few years (RR: 109/360, SP: 15/25, PP: 0/39) [27]. When oligoclonal IgMs are present, the disease lasts for less than 5 years in 90% of patients, but when IgMs are absent, it lasts for more than 5 years in 60% of patients. (2) Anti‐MBPs in CSF are high during relapses in about 90% of patients but vanish during the 6 weeks following onset, relapse treatment appearing to precipitate this event [80, 81].

Each patient has a private and unique CSF OCB pattern, the so‐called OCB fingerprint [82, 83]. The pattern persists even though the total CSF IgG is significantly reduced, for example, by steroids [83]. The CNS IgG synthesis rate remains roughly stable over several months [6]. When steroids are used to reduce the IgG index, pretreatment values are recovered in a few weeks or months [83]. Intrathecal steroids rarely obliterate OCB for more than 1 month [71], but this modification can be corrected when equal amounts of CSF Ig are applied in IEF.

Unfortunately, OCBs provide basic information and the future development of microarrays may provide more precise insight into the intimate evolution of clonality. For example, the involution of the antibody response against heat shock protein (HSP) in serum is a central modification in the immune response during the transition from RR to the progressive form [84], but the outcome of this immune signature in CSF is unknown.

The persistence of OCB suggests a local IgG production by long‐lived plasma cells residing inside the CNS, suggesting that the B‐cell fostering properties of the CNS deserve consider‐ ation [85]. On the other hand, minor changes in OCB pattern are in accordance with competition between plasma cells for a limited number of survival niches, as observed in the bone marrow. It seems unlikely that such a discrepancy between brain volume and the small number of CNS plasma cells can be explained only by the B‐cell fostering properties of the CNS. Moreover, the intrathecal IgG synthesis rate is roughly stable over time, even though persistent local inflammation would be expected to induce a gradual increase in synthesis rate. These data suggest that the number of CNS niches is limited and that they are quickly saturated. Therefore, the putative role of CNS tertiary lymphoid organs (TLOs) deserves further consideration.

#### **4.3. None of the available MS drugs deplete intrathecal Ig synthesis**

In a complete review of the literature, we demonstrated that none of the available (approved or experimental) treatment proved to abate intrathecal Ig synthesis (review in reference [86]). A null or minimal (<20%) impact on intrathecal IgG synthesis was observed after the admin‐ istration of azathioprine, β‐interferons, cytarabine (intravenous or intrathecal), lomustine, 5‐ fluorouracil, CNS irradiation, mitoxantrone, methotrexate, cyclophosphamide, cyclosporin A, cladribine, alemtuzumab, daclizumab, fingolimod, or stem cell transplantation following myeloablation.

Steroid infusion decreases blood Ig levels in a few days [87]. However, high‐dose steroid infusions have little effect on IgG index and OCB number [88–90], irrespective of the dosage or the mode of administration (intramuscular, intravenous, and intrathecal) [91]. Even if steroids can transiently lower the IgG index in most but not all patients, the decrease in the range of CSF IgG synthesis is low [90, 92]. The mean antibody reactivity and the mean number of targeted antigens in CSF are decreased in the 2 months following steroids [93].

Rituximab depletes CSF B‐cells but fails to significantly decrease intrathecal IgG synthesis (<-20%) [45, 94], possibly because ITS may partly depend on plasma cells (CD20-), which are constitutively resistant. This failure to lower intrathecal IgG secretion was predictable from the absence of the effect of blood‐infused rituximab upon serum IgG and IgA levels, contrary to a minor effect upon IgM levels [45, 95]. No data are available on intrathecal IgM synthesis after rituximab treatment.

Given the low diffusion of rituximab to CSF (<0.2%) and the growing body of evidence demonstrating the safe use of intrathecally infused rituximab, a rationale to infuse intrathecal rituximab in progressive MS recently emerged (review in reference [96]). Data obtained from a single patient receiving intrathecal rituximab (10 mg per month for 2 months) showed a major effect upon CSF cytokine levels although intrathecal IgG synthesis was unchanged [97, 98] (unpublished).

An unexpected partial repression of intrathecal IgG synthesis was obtained with natalizumab treatment in some series [99–101] but results were null or lower in two others [102, 103]. These findings need to be replicated in another large cohort. Three non‐mutually exclusive explan‐ ations may be put forward. First, α4β1‐integrin is expressed by B‐cells (CD19+ and CD138+) at higher levels than CD3+ T‐cells, and natalizumab impedes B‐cell trafficking to the brain and renewal of the CNS plasmablast pool. Second, apart from cytokines, T‐cells also play a supportive survival role in the survival niches of plasma cells [104]. Third, natalizumab inhibits the CNS migration of dendritic cells, which in turn may affect the maintenance of CNS lymphoid tissue [105], and dendritic cell counts in MS brains were lower than expected [106]. Of note, a significant decrease in IgM (≈-45%) (and less significantly in IgG1≈-18 to 36%) plasma levels also occurs during natalizumab treatment but is not correlated with treatment duration, suggesting that it effectively perturbs the IgG synthesis pathway [103, 107]. Future experiments should examine the kinetics of intrathecal IgG secretion after withdrawal of natalizumab. Data obtained from a single patient devoid of OCB under natalizumab and discontinuing treatment for progressive multifocal leukoencephalopathy (PML) showed that OCB returned shortly with a slightly modified pattern [101], whereas a sustained negativation was observed at 6 months in two other patients [99]. However, in a series of 23 patients, no increase of ITS or OCB was observed at 14 months after natalizumab discontinuation [108], suggesting that the impact of natalizumab on ITS is probably low.

A single patient treated by EBV‐specific adoptive immunotherapy (autologous CD8+ T‐cells activated against EBV) regained a normal level of IgG index at month 4 [109]. This single case needs to be replicated.

#### **4.4. Prognostic value of ITS**

bone marrow. It seems unlikely that such a discrepancy between brain volume and the small number of CNS plasma cells can be explained only by the B‐cell fostering properties of the CNS. Moreover, the intrathecal IgG synthesis rate is roughly stable over time, even though persistent local inflammation would be expected to induce a gradual increase in synthesis rate. These data suggest that the number of CNS niches is limited and that they are quickly saturated. Therefore, the putative role of CNS tertiary lymphoid organs (TLOs) deserves

In a complete review of the literature, we demonstrated that none of the available (approved or experimental) treatment proved to abate intrathecal Ig synthesis (review in reference [86]). A null or minimal (<20%) impact on intrathecal IgG synthesis was observed after the admin‐ istration of azathioprine, β‐interferons, cytarabine (intravenous or intrathecal), lomustine, 5‐ fluorouracil, CNS irradiation, mitoxantrone, methotrexate, cyclophosphamide, cyclosporin A, cladribine, alemtuzumab, daclizumab, fingolimod, or stem cell transplantation following

Steroid infusion decreases blood Ig levels in a few days [87]. However, high‐dose steroid infusions have little effect on IgG index and OCB number [88–90], irrespective of the dosage or the mode of administration (intramuscular, intravenous, and intrathecal) [91]. Even if steroids can transiently lower the IgG index in most but not all patients, the decrease in the range of CSF IgG synthesis is low [90, 92]. The mean antibody reactivity and the mean number

Rituximab depletes CSF B‐cells but fails to significantly decrease intrathecal IgG synthesis (<-20%) [45, 94], possibly because ITS may partly depend on plasma cells (CD20-), which are constitutively resistant. This failure to lower intrathecal IgG secretion was predictable from the absence of the effect of blood‐infused rituximab upon serum IgG and IgA levels, contrary to a minor effect upon IgM levels [45, 95]. No data are available on intrathecal IgM synthesis

Given the low diffusion of rituximab to CSF (<0.2%) and the growing body of evidence demonstrating the safe use of intrathecally infused rituximab, a rationale to infuse intrathecal rituximab in progressive MS recently emerged (review in reference [96]). Data obtained from a single patient receiving intrathecal rituximab (10 mg per month for 2 months) showed a major effect upon CSF cytokine levels although intrathecal IgG synthesis was unchanged [97, 98]

An unexpected partial repression of intrathecal IgG synthesis was obtained with natalizumab treatment in some series [99–101] but results were null or lower in two others [102, 103]. These findings need to be replicated in another large cohort. Three non‐mutually exclusive explan‐ ations may be put forward. First, α4β1‐integrin is expressed by B‐cells (CD19+ and CD138+) at higher levels than CD3+ T‐cells, and natalizumab impedes B‐cell trafficking to the brain and renewal of the CNS plasmablast pool. Second, apart from cytokines, T‐cells also play a supportive survival role in the survival niches of plasma cells [104]. Third, natalizumab inhibits

of targeted antigens in CSF are decreased in the 2 months following steroids [93].

**4.3. None of the available MS drugs deplete intrathecal Ig synthesis**

further consideration.

60 Trending Topics in Multiple Sclerosis

myeloablation.

after rituximab treatment.

(unpublished).

Although the prognostic value of OCB positivity is not consensual, ITS is consistently associ‐ ated with a worse outcome. In CIS patients, none of the IgG, IgA, or IgM indexes are predictive of clinical conversion to MS [110]. On the other hand, OCB positivity in CIS increases the risk of having a second attack irrespective of magnetic resonance imaging (MRI) results [111] and the number of OCB is an independent risk factor for relapses [112], whereas their absence is associated with a lower clinical conversion at 5 years and with a better clinical outcome [113]. Interestingly, double OCB positivity (IgG and IgM) is associated with an odds ratio (OR) of second relapses higher than fulfilling of Barkhof's criteria (OR 55.7 vs 3.5) [114]. The MRZ pattern is also an independent risk factor of clinical conversion [115] and all CIS patients displaying OCB targeting lipid antigens relapse in the first year [113].

In clinically defined MS patients, the IgG synthesis rate correlates with MRI burden and the IgG1 index correlates with expanded disability status scale (EDSS). Free light chains in CSF are predictive of impairment [116]. CSF IgM and IgM OCB are associated with a poorer clinical outcome [73, 117] and a lower brain volume [118, 119]. IgM OCBs reacting against myelin lipids are associated with a higher relapse rate, lower efficiency of IFNβ, and a faster EDSS increase [120, 121]. Time and probability to reach EDSS 3 and 4 are strongly correlated with the presence of IgM in CSF [122, 123]. The IgM index in CSF at onset is strongly correlated with a higher last follow‐up EDSS [117, 124]. The time to conversion to the secondary progressive phase is shorter in the presence of IgM BOC or an elevated IgM index [73, 123, 124]. These attractive findings concerning IgM were established in a single group of patients. Stauch et al. [13] did not find any influence of IgM OCB on the evolution of MS in a pediatric group and raised some technical concerns about the IgM tests used previously, which may invalidate the results. Another group claimed negative results but they also obtained a shorter, albeit nonsignificant, time to relapse after CIS in OCB‐IgM+ patients [125]. In a large retrospective study, disease severity appeared to be identical between OCB‐positive and ‐negative patients [126], whereas in a study including 1800 MS patients, the number of OCB was considered to correlate with the course of the disease [127]. The mean number of OCB in the group of benign patients was 2.9 ± 3.6 compared to 5.7 ± 4.9 in the severe group (*p* < 0.06) [127]. Therefore, the evidence points to a poorer clinical prognosis in MS patients in association with ITS. On the other hand, OCB negativity correlates with a better outcome in many studies [70, 127–131], even after demo‐ graphic adjustment [132].
