The impact of disease-modifying therapies on immunoglobulin blood levels in patients with multiple sclerosis: a retrospective cross-sectional study

Background: Although disease-modifying therapies (DMTs) in multiple sclerosis (MS) are known to target the immune system, mechanisms of action, efficacy, safety, and tolerability profiles differ. The long-term impact of DMTs on the immune system and its relation to infectious complications is still poorly understood. Objectives: To analyze the effect of DMTs on serum immunoglobulin (Ig) levels under consideration of patient demographics and therapy duration. Design: We included 483 patients on DMTs, 69 patients without DMTs, and 51 controls in this retrospective cross-sectional study. Methods: IgG, IgM, and IgG subclass 1–4 levels of patients with MS under treatment with DMTs were compared with treatment naive MS patients and controls by multivariate linear regression. Further, Ig levels stratified by DMTs were analyzed regarding therapy duration. Results: MS patients treated with fingolimod (FG), natalizumab, and B-cell depleting therapies (BCDT) demonstrated significantly lower IgG and IgM levels than healthy controls after a median treatment of 37, 31, and 23 months, respectively (p < 0.05). Treatment with dimethyl fumarate (DMF) and teriflunomide was associated with lower IgG, but not IgM levels. DMF and BCDT were also associated with lower IgG1 levels, while FG led to a reduction of IgG2. Treatment with interferon-beta (IFN) and glatiramer acetate (GA) had no impact on Ig levels. Analysis of subgroups by linear regression also showed a time-dependent decrease of Igs levels in patients treated with BCDT with a median annual reduction of IgG of 3.2% and IgM of 6.2%. Conclusion: Treatment with DMTs, except GA and IFN, was associated with a decrease in Ig levels. DMTs differed in the extent of decreasing Ig levels but also in their differential effects on Ig subclasses. Monitoring of Ig levels should be considered in patients on long-term treatment with DMTs, particularly those on BCDT, to identify patients at risk of low immunoglobulin levels.


Introduction
During the past few decades, several diseasemodifying therapies (DMTs) for multiple sclerosis (MS) have been approved worldwide.
Although these therapies are all known to target the immune response in MS, mechanisms of action, and efficacy, safety and tolerability profiles differ.

The impact of disease-modifying therapies on immunoglobulin blood levels in patients with multiple sclerosis: a retrospective cross-sectional study
Many of these DMTs lead to an increase of infectious complications. However, little is known on the effect of DMTs on immunoglobulin (Ig) serum levels. Of all DMTs administered in MS most is known in regard to rituximab and its effect on immunoglobulin serum levels. Rituximab, a B-cell depleting therapy (BCDT), is used for treating many autoimmune-mediated and hematological diseases. Various studies have shown an association between low levels of both IgG and IgM and rituximab therapy. [1][2][3][4][5][6][7] Two recent studies evaluating the long-term effect of rituximab on Ig levels in patients with neuromyelitis optica spectrum disorders discovered reductions in both IgG and IgM over time. 8,9 In some studies, a prolonged hypogammaglobulinemia even years after the last treatment with rituximab was demonstrated. 5,10 In 2017, ocrelizumab, a slightly modified BCDT, was approved for use in patients with MS. Data on its impact on Ig levels are still sparse. However, an initial report on the pivotal Phase III trials of ocrelizumab and their open-label extensions showed a reduction of IgG of approximately 17%, of IgM of between 56% and 58%, and IgA of 20% and 21% over the course of 5.5 years under therapy with ocrelizumab. 11 Furthermore, a comparative study on treatment outcomes over the first year with rituximab and ocrelizumab showed a continuous reduction of IgG levels in patients under treatment with ocrelizumab, while they remained stable under rituximab. IgM levels were shown to drop similarly under both drugs. 12 Regarding further DMTs used in the treatment of MS, one recent study showed reduced serum levels of IgG and IgM in patients with MS in comparison to healthy controls. Within this cohort, rituximab, natalizumab (NZ), and fingolimod (FG) were associated with significantly reduced immunoglobulin levels. A multivariable linear regression, however, showed no time dependency of hypoimmunoglobulinemia. 13 Similar results were found in a further study in regard to NZ, displaying significantly decreased levels of both immunoglobulins in serum. This study furthermore demonstrated a significant time dependency. 14 So far, only sparse data have been published in regard to the effect of DMTs on the individual IgG subclasses 1-4 and most of these data refer to rituximab. Several studies showed that rituximab leads to a reduction of IgG4 explaining the effective use in combatting IgG4-related systemic diseases. [15][16][17] The aim of this study was to identify the effect of the most commonly used DMTs, namely interferon-beta (IFN), glatiramer acetate (GA), dimethyl fumarate (DMF), teriflunomide (TFM), FG, NZ, and BCDT (ocrelizumab and rituximab) on IgM, IgG and its subclasses 1-4 in serum of patients with MS in comparison to healthy controls and untreated MS patients.

Study design and inclusion criteria
In this retrospective cross-sectional study, patient data collected at the Department of Neurology of the Technical University of Munich were analyzed regarding their serum Ig levels. We compared IgG, IgM, and IgG subclass 1-4 serum levels between patients with MS with or without immunotherapy (DMT+ versus DMT-patients) with normal controls under consideration of patient demographics and therapy duration. Patients treated with rituximab and ocrelizumab were pooled (BCDT) for most analyses because their mode of action and treatment effects are highly similar. Ocrelizumab was only approved in 2017, median therapy duration was 17 months for patients treated with ocrelizumab in comparison to 30.5 in patients treated with rituximab. As our results show that therapy duration plays a vital role on the reduction of Ig levels, a combined analysis of the two reduces the risk of introducing bias.
Inclusion criteria for the MS groups were defined as follows: To increase statistical power, we decided to collectively analyze data from patients treated with rituximab and ocrelizumab and define these as patients under BCDT.

Study cohort
Baseline patient demographics and therapy duration stratified by drug can be found in Table 1.
Of the 603 patients included in this study, 552 suffered from MS, while 51 served as normal controls. Within the group of MS patients, 69 patients did not receive immunotherapy and therefore constituted the DMT-group, while the other 483 patients were defined as DMT+ (the exact distribution of patients among groups can be found in Table 1). Sera for sub-analysis of IgG subgroups was available for 286 (47.4%) patients (the exact distribution of patients among groups can be found in Table 1).  Figure 1).

Comparison of IgG and IgM levels between groups
The highest proportion of patients with IgG values below the lower limit of normal was found in patients treated with FG (n = 23; 21.9% of patients), shortly followed by patients treated with NZ (n = 9; 19.6% of patients). Patients treated with BCDT showed the highest proportion of IgM values below the lower limit of normal (according to our laboratory standards), with a total of 32 IgM values below 40mg/dl (26.2% of patients; Table 2).
We performed a multivariate linear regression, comparing Ig levels between therapy groups, in which we included age and gender as covariates.
The control group was set as a reference to compare Ig levels between groups.
This resulted in lower IgG and IgM levels in patients treated with FG, NZ, and BCDT than healthy controls (p < 0.05) after a median treatment of 37, 30.5, and 23 months respectively (Table 2, Figure 1). Treatment with DMF and TFM lead to relatively low IgG, but not IgM levels (Table 2, Figure 1). The most pronounced effect on IgG was seen in patients treated with FG, with 13.2% lower IgG levels than normal controls, while for IgM the most pronounced effect was seen in patients medicated with NZ (27.5% lower than controls).
DMT-naive patients and patients treated with GA or IFN did not show lower Ig levels than healthy controls (p > 0.05, Table 2, Figure 1). Age and male gender were also associated with lower IgG and IgM levels, in accordance with current literature (IgG: p = 0.002 and p = 0.018; IgM: p = 0.014 and p < 0.001, respectively).

Multivariate linear regression of timedependent Ig reduction within therapy subgroups
After stratification of our patients into subgroups based on the medication applied, Ig levels in patients under treatment with FG, NZ, and BCDT were analyzed in regard to therapy duration. Individual multivariate linear regression  Figure  2). We did not observe major differences between patients treated with rituximab or ocrelizumab with respect to their impact on IgG and IgM levels when the therapy duration was set at 39 months (the highest therapy duration in the ocrelizumab group, Supplementary figure).

Comparison of IgG subclasses 1-4 between groups
We performed a multivariate linear regression comparing IgG subclasses 1-4 between therapy groups, using the control group as the reference group and including age, gender, and total IgG as further covariates.
The highest levels of IgG1 were found in the control group (median: 794 mg/dl; range: 254-1482 mg/dl), while the lowest levels were in the group of patients under therapy with NZ (median: 526 mg/dl; range: 265-1075 mg/dl). Multivariate linear regression resulted in lower IgG1 levels in patients treated with DMF (p < 0.05), with 7.6% lower values than normal controls. Patients treated with BCDT also showed 6% lower IgG1 levels than controls; however, the p-value was exactly 0.05, thus closely missing statistical significance.
IgG2 levels were found to be 10.6% lower in patients under therapy with FG (p < 0.05) than in controls.
IgG3 and IgG 4 levels did not differ between groups (p > 0.05). Comparison between controls and DMT-naive MS patients did not result in differences of any IgG subclasses. Age and gender did not seem to be associated with lower IgG subclasses 1-3. Interestingly, female gender was associated with lower IgG4 levels in this subset of patients. For further details, please see Table 3. Furthermore, we visualized IgG subclass levels of each therapy group as percent from controls in Figure 3.

Discussion
In this explorative study, we investigated total IgG and IgM as well as IgG subclasses in patients with MS. Treatment naive MS patients did not differ in their Ig levels from controls, challenging the hypothesis of a previous study that MS in itself may cause hypoimmunoglobulinemia. 13 We observed lower levels of IgM and IgG under therapy with FG, NZ, and BCDT compared with *Statistical significance compared to controls (see Table 2 19 and Barmettler et al. 7 demonstrating that low IgG levels are associated with a higher risk of infection in patients under rituximab, regularly monitoring of Ig levels in patients receiving BCDT can be useful to allow an earlier identification of patients at risk for infectious complication. 7,19 As stated within the methods section, we decided to collectively analyze the effects of rituximab (n = 74) and ocrelizumab (n = 48) on the Ig levels of our cohort to increase statistical power. As both DMTs are monoclonal CD20 antibodies, no difference in their effects on Ig levels should be expected. A linear multivariate regression including individual analysis of these two DMTs on IgG and IgM levels supported this hypothesis. Subanalysis of their effect on IgG subclasses was not performed due to insufficient sample size.
The largest strength of our analysis lies in our large cohort of patients with balanced participant numbers between groups, which increases the power of our findings. However, in regard to IgG subclasses, only a subset of patients with conserved sera underwent ELISA testing, leading to a smaller cohort (Table 1). For some DMTs only, a small number of sera were available for testing, limiting the statistical power of analysis for these groups.