Peripheral CD19+ B-cell counts and infusion intervals as a surrogate for long-term B-cell depleting therapy in multiple sclerosis and neuromyelitis optica/neuromyelitis optica spectrum disorders

Background With ocrelizumab another drug is available for the treatment of multiple sclerosis (MS). Little is known on the long-term use of ocrelizumab on immune cell subsets, and no surrogate markers are available. Rituximab (RTX) has been in off-label use for the treatment of MS, neuromyelitis optica (NMO) and neuromyelitis optica spectrum disorder (NMOSD) for > 10 years. Objective We evaluated the long-term depletion and repopulation rate of peripheral CD19+ B-cells as a potential surrogate for the clinical outcome, and whether it may serve for dosage and time-to-infusion decision making. Methods We evaluated the CD19+ and CD4+/8+ T-cell counts in n = 153 patients treated with RTX (132 MS, 21 NMO/NMOSD). The dosages ranged from 250 to 2000 mg RTX. Depletion/repopulation rates of CD19+ B-cells as well as long-term total lymphocyte cell counts, were assessed and corroborated with EDSS, ARR (annualized relapse rate), MRI, and time to reinfusion. Results CD19+ B-cells’ repopulation rate significantly varied depending on the dosage applied leading to individualized application intervals (mean 9.73 ± 0.528 months). Low/absent CD19+ B-cell counts were associated with reduced ARR, EDSS, and GD+-MRI-lesions. Long-term B-cell-depleting therapy led to a transiently skewed CD4+/8+ T-cell ratio due to reduced CD4+ T-cells and absolute lymphocyte counts, which recovered after the second cycle. Conclusion Our data suggest that CD19+ B-cell repopulation latency may serve as surrogate marker for individualized treatment strategies in MS and NMO/NMOSD, which proved clinically equally effective in our cohort as evaluated by previous studies.


Introduction
Although the T-cell (Th1 and Th17) mediated pathogenesis of MS is well established, B-cells and the humoral immune involvement are also increasingly recognized as drivers of the autoimmune disease and the concomitant neurodegeneration [18]. Hence, therapies either exclusively targeting B-cells, such as rituximab, ocrelizumab, and ofatumumab, or B-cells along with T-cells, such as alemtuzumab have proven to be effective in clinical trials [6].
The recent approval of the humanized antibody ocrelizumab depleting CD20 + B-cells, based on its efficacy shown in two phase III clinical trials in relapsing-remitting MS (RRMS; OPERA I and II studies) and one in Gisa Ellrichmann and Jan Bolz contributed equally as first authors.
Ralf Gold and Aiden Haghikia contributed equally as senior authors.
primary progressive MS (PPMS; ORATORIO study), led to more than ten compounds (nine substance classes) for the treatment of MS [7,8,11]. However, the principle of B-cell depletion with its predecessor rituximab (RTX)-a chimeric monoclonal anti-CD20 antibody that binds to cell surface CD20 and induces antibody dependent cell-mediated cytotoxicity-has long been off-label use in MS and other autoimmune disorders. Several phase II studies and case series have previously shown its efficacy in MS [8,9]. More recently, retrospective studies analyzing large Swedish MS cohorts receiving RTX treatment for a mean time of ~ 22 months provided further evidence for its beneficial therapeutic effect in RRMS and progressive MS [16,17]. The rapid onset of its therapeutic effect, as shown in MRI measures after RTX initiation suggests that the mechanism by which RTX exerts its therapeutic effects is based not only on depleting B-cells as potential antibody producing cells, but also their involvement in T-cell activation [2]. In addition to MS, RTX has lately proven effective in NMOSD, another autoimmune demyelinating disease of the central nervous system that in contrast to MS is mediated by antibodies directed to the astrocytic aquaporin-4 [12].
Using ocrelizumab, the re-dosing is carried out at a fixed interval and dosage (600 mg every 6 months). Unlike ocrelizumab, there is no standard protocol for RTX infusions. At our centers, it is a current mode of clinical practice to follow B-cell counts as a measure of RTX reinfusion.
The aim of our study was to evaluate repopulation rate of peripheral CD19 + B-cells, as well as other lymphocyte subtypes as a potential surrogate marker for individual application intervals in patients with MS and NMO/NMOSD treated with RTX.

Study population
The study was approved by the local ethics committees (regno 4493-12), and was intended to identify biological surrogate markers for the efficacy of drugs in MS and NMO/ NMOSD. For analyses of CD56 + NK-cells and Th1-cells in healthy controls and RRMS patients, ethical permit was approved by the local ethics committees (reg-no 15-5351).
Patients who had received RTX and fulfilling either the McDonald criteria for MS or the criteria proposed by Wingerchuk et al. for NMO/NMOSD were included into the study [14,19]. RTX was administrated when other options where exhausted or MRI indicated a B-cell pathology. The dosage and application interval was generally based on B-cell repopulation (2-5%), but did not follow a standardized protocol and differed on timing of patients' clinic visits, underlying disease activity, side effects, patients' conditions and physicians' decision. That is the reason, why patients' clinical visit intervals and applied RTX dose might vary in a single patient. In general, earlier (~ 10 years ago) the RTX dosages were higher and resembled protocols used by oncologists for the treatment of lymphoma patients. All RTX doses up to 1000 mg were applied in a single infusion, higher doses were given equally distributed within a 2-week interval. Beside this protocol, no double-dosing was administrated. For inclusion treatment should at least be once between April 2006 and September 2016. For baseline data, the EDSS and MRI before the first RTX infusion were used. Patients were treated as in-or out-patients in two German MS centers (Department of Neurology of the St. Josef-Hospital, Ruhr-University Bochum and the Department of Neurology, Friedrich-Alexander-University Erlangen). Anonymized patients' data were used for stratification analyses, i.e. treatment response.

Stratification
The respective absolute long-term lymphocyte cell counts (cells/µl) were corroborated with disability progression (EDSS), MRI and ARR (annualized relapse rate). A relapse was defined as a worsening of clinical symptoms that lasted longer than 24 h and was not related to any infectious disease. We recorded the change in number of T1and T2-weighed lesions as well as numbers of gadolinium enhancing (GD + ) T1-lesions. EDSS, cerebral and spinal MRI compared to baseline were assessed before RTX therapy and after 12, 24 and 36 months.

Statistics
Statistical analyses were performed using GraphPad Prism 6.0. Data are presented as a mean ± standard deviation (SD). For statistical analysis, group differences were evaluated using one-way analysis of variance (ANOVA) followed by Bonferroni's post-hoc test (GraphPad Prism 6.0 software, San Diego, CA, USA). The probability levels of p* ≤ 0.05, p** ≤ 0.01, and p*** ≤ 0.001 were considered statistically significant for all statistical tests.

Patients' characteristics and treatment
A total of 153 patients treated with at least one course of B-cell depleting RTX therapy were included in our study ( Table 1). Data of 21 NMO/NMOSD patients and 132 MS patients were analyzed. 72 of the MS patients were diagnosed as RRMS and 60 of them as secondary progressive MS (SPMS). Mean age of patients at disease onset was 33.58 ± 11.97 years and mean age at first RTX infusion was 41.69 ± 12.35 years. Mean follow-up was 26.28 ± 21.08 months. Mean RTX dose per treatment course was 717 mg ± 456 taking every infusion into account. If RTX infusion within a 2-week interval is regarded as one cycle, mean RTX dose is higher (828 mg ± 589) ( Table 2). We did not observe differences in treatment response or dosing depending on previous therapy. As there would be too many subgroups resulting in small patient numbers per group, we decided to forego those subanalyses concerning pre-treatment.
In a sub-population (n = 25), CD56 + NK-, Treg-and Th1-cells were evaluated (Fig. 4a-c). NK-cells were normalized under RTX treatment and were lower in RRMS as compared with controls. Blood samples were obtained at remission. RRMS group was treated with dimethyl fumarate, detailed characteristics of the different groups are given in Table 4.

Clinical course and MRI
Less than 40% (28/72) of patients with relapsing forms of MS and one-third (7/21) of the NMO/NMOSD patients had a relapse during the observation period. ARR in our MS cohort significantly decreased from 1.55 ± 1.36 2 years before RTX treatment to 0.26 ± 0.52 during follow-up (83% reduction, p*** < 0.0001) ( Table 5). ARR in NMO/ NMOSD patients decreased from 2.22 ± 1.89 to 0.42 ± 0.56 (81% reduction, p*** < 0.0001). In patients with relapse under RTX average time until first relapse after last infusion was 5.35 ± 3.04 months (Table 5). There was no correlation between B-cell counts at the time-point of reinfusion and clinical course in patients with relapses.
One year after first application of RTX, 130 EDSS of the 132 MS patients were available for analyses. 32 MS patients improved, 75 remained stable, and 23 worsened ( Table 6). 24/36 months after first treatment 101/77 follow-ups were available. 21/21 MS patients had a better score compared to baseline, 45/30 patients did not change and in 20/16 patients a progression of EDSS was documented ( Fig. 5a; Table 6). In patients diagnosed with NMO/NMOSD 6 improved, 12 remained stable and in 3 patients a progress in EDSS was documented. At 24/36 months, 15/10 EDSS were available 4/2 improved, 6/4 remained stable and 5/4 had a higher EDSS compared to baseline.
Cerebral MRI scans at baseline were available for 150 patients (98%). The number of patients with Gd + lesions significantly decreased during therapy (p*** < 0.001) from 31 at baseline to 5 at 12 months after treatment. Analyzing spinal MRI, number of patients with Gd + lesions decreased from 35 at baseline to 2 at 12 months (p*** < 0.001). In  Table 7). Due to varying infusion intervals and doses, over time clinical course was analyzed regarding mean annual RTX doses and CD19 + B-cell at reinfusion/relapse. We did not observe difference in mean RTX dose or CD19 + B-cell counts in regard of EDSS, MRI or clinical relapse (Table 8).

Side effects
Only major side effects were recorded. One patient was hospitalized for severe pneumonia. In one patient, a reactivation of hepatitis B was observed. In general, there were rare side effects and RTX was well tolerated.

Discussion
There is growing evidence for the efficacy of B-cell depleting therapies in various autoimmune diseases [4]. This has previously resulted in the FDA approval of RTX for the treatment of rheumatoid arthritis in 2006. Two phase II trials in MS, and several open label trials in MS and NMO/NMOSD underlined the efficacy also in MS and NMO/NMOSD with a favorable safety profile of the anti-CD20 monoclonal antibody RTX. Finally, its successor, the humanized antibody ocrelizumab (Ocrevus ® , Roche, Switzerland) was recently approved for the treatment of RRMS and PPMS by the FDA and the EMA.
Yet, the RTX dosages used significantly vary in different cohorts, and ocrelizumab is approved only at a defined dosage. Hence, in our study, we evaluated the long-term depletion and repopulation rate of peripheral CD19 + B-cells under RTX treatment as a potential surrogate marker for the clinical outcome.
We showed that CD19 + B-cell repopulation was significantly faster when 250 mg RTX was applied. Higher doses of RTX (500-2000 mg) did not lead to sustained B-cell depletion, which might indicate a ceiling effect. Second, long-term RTX treatment did not induce a substantial change in total T-cell populations over time. In a long-term manner, the number of CD4 + -and CD8 + -cells, as well as CD4/CD8 ratio did not change significantly compared to pre-RTX levels. This is well in line with previous studies [13]. We did not observe any severe side effects such as  secondary autoimmunity (SAI). This is in contrast to T-and B-cell-depleting therapies, i.e. alemtuzumab, where SAI are seen in more than 30%. These side effects are attributed to an excessive repopulation of B cells accompanied by the depletion of (regulatory) T-cells [1,5]. Moreover, reduced numbers of CD19 + B-cells were associated with reduced ARR, EDSS and reduced numbers of Gd + enhancing lesions. Besides the small number and different treatment regimens, NMO/NMOSD and MS patients did neither show significant differences in intensity, nor duration of cell depletion.
Continuous monitoring of CD19 + B-cells may be a sufficient tool for an individualized decision making on dosage and reinfusion intervals of B-cell depleting therapies. Fixed dosage and infusion intervals, as in therapy regime of ocrelizumab, may explain the increased risk for infections and malignancies. In our cohort, dosage of RTX ranging 500-1000 mg led to longer reinfusion intervals. Higher dosages of RTX neither lead to extended reapplication intervals, nor any additional treatment effects, and may be avoided [16].
Despite the therapeutic effect being closely associated with the absence of CD19 + B-cells, we did not observe a correlation between B-cell counts at the time-point of reinfusion and clinical course or MRI outcome in patients in whom relapses did occur. This might be due to different proportion of progenitor CD19 + B-cells and mature B-cells after the first cycle of RTX treatment. After replenishment of the B-cell compartment there are mainly naïve B-cells following repletion of circulating B-cells. CD27 + memory phenotype cells stay at significantly lower levels in peripheral blood [13]. It has been recently shown that memory B cells drive proliferation of self-reactive brain-homing CD4 + T-cells in patients with multiple sclerosis and that RTX strongly reduces autoproliferation and proinflammatory cytokine responses [10]. This may be one of the reasons for the longlasting beneficial clinical effects of anti-CD20 therapy [13,15]. However, B-cell responses in MS patients have shown  to be per se altered in MS [3]. We observed an increase of B cells over time that is due to statistical effects and patients' behavior. Number of patients being analyzed decreased over time and simultaneously variations in CD19 + B-cell counts increased as there were single statistical outliers. Our study has several limitations, the main being its retrospective character. Another limitation arises from the different dosages used over time. The varying infusion protocols reflect the evolution of RTX titration toward lower dosage over the past 10 years in off-label use, and allow the differentiated analysis of B-cell repopulation rates.
Nevertheless, we could build up a robust database over a long-term period of 100 months with a sufficiently large number of patients. Theoretically, other immune cell subsets bearing CD20 antigen may contribute to the efficacy of RTX and may be overlooked in FACS analysis-guided infusion cycles. At least, in our cohort, the clinical effect was very robust.
In summary, in contrast to treatment of rheumatoid arthritis or lymphoma, 500 mg or even 250 mg RTX in intervals of approximately 6-9 months are an effective and at the same time safe therapy option in MS. In our cohort, higher dosage in elongated intervals proved equally effective. However, dosages beyond 1000 mg did not reveal additional effects in regard to B-cell depletion, administration interval and clinical outcome. Given the large inter-individual range of B-cell recovery time, we suggest that the CD19 + B-cell repopulation rate may serve as surrogate marker to appraise individually adapted therapy intervals. Although the next generation of anti-CD20 monoclonal antibodies is approved for the treatment of MS, many mechanistic aspects and longterm immunological consequences remain unclear. Analysis of more than 10 years of RTX use in MS may shed further light into these questions.