Efficacy and safety of tocilizumab in patients with refractory generalized myasthenia gravis

Abstract Background We aimed to compare the efficacy of tocilizumab with conventional immunotherapy in refractory patients with acetylcholine receptor antibody‐positive (AChR‐Ab+) generalized myasthenia gravis (gMG). Methods This single‐center prospective cohort study was based on patients from an MG registry study in China and conducted from February 10, 2021 to March 31, 2022. Adult refractory patients with AChR‐Ab+ gMG were assigned to tocilizumab or conventional immunotherapy groups. The primary efficacy outcome was the mean difference of MG activities of daily living (MG‐ADL) change at weeks 4, 8, 12, 16, 20, 24 corresponding to that at the baseline between the two groups. A generalized estimating equation model was used for the primary outcome analysis. Safety was assessed based on adverse events. Results Of 34 eligible patients, 20 (mean [standard deviation] age, 53.8 [21.9] years; 12 [60.0%] female) received tocilizumab and 14 received conventional immunotherapy (45.8 [18.0] years; 8 [57.1%] female). The tocilizumab group had greater reduction in MG‐ADL score at week 4 (adjusted mean difference, −3.4; 95% CI, −4.7 to −2.0; p < 0.001) than the conventional immunotherapy group, with significant differences sustained through week 24 (adjusted mean difference, −4.5; 95% CI, −6.4 to −2.6; p < 0.001). At week 24, the proportion of patients achieving higher levels of MG‐ADL (up to 7‐point reduction) and QMG (up to 11‐point reduction) scores improvement was significantly greater with tocilizumab. Tocilizumab had acceptable safety profiles without severe or unexpected safety issues. Conclusion Tocilizumab is safe and effective in improving the MG‐ADL score and reducing prednisone dose in refractory AChR‐Ab+ gMG, suggesting tocilizumab has the potential to be a valuable therapeutic option for such patients.


| INTRODUC TI ON
Myasthenia gravis (MG) is an acquired autoimmune disease caused by acetylcholine receptor (AChR) antibodies in approximately 80% of cases. 1 Most patients can be successfully managed with nonspecific immunotherapies, such as corticosteroid and non-steroidal immunosuppressive treatments (NSIST).However, the adverse events (AEs) of long-term nonspecific immunotherapies remain a hurdle for MG treatment.A case-control study reported that around 48.2% of patients with MG expressed concerns regarding short-and long-term treatment AEs and that 55.8% discontinued treatment due to AEs associated with current therapies. 2Despite these therapies, MG may not be well controlled or refractory due to an incomplete response or poor tolerance.
Recently, there has been a surge in the development of new therapies targeting B and T cells, component 5 (C5), neonatal Fc receptor (FcRn), and cytokines.[5][6][7][8] Interleukin 6 (IL-6), a keystone cytokine, is involved in the survival, expansion, and maturation of B cells and plasmablasts, as well as the production of autoantibodies, making it a potential candidate for MG treatment.Studies on experimental autoimmune MG (EAMG) have shown that clinical symptoms can be reversed using anti-IL-6 antibodies.Suppression of EAMG by anti-IL-6 antibodies was accompanied by a decrease in the overall rat anti-AChR antibody titer and a reduced number of B cells compared with the control treatment. 9Additionally, the disruption of IL-6 signaling may switch immune responses in vivo from the induction of regulatory T cells to pathogenic IL-17-producing T helper cells. 10Currently, a study evaluating the efficacy and safety of satralizumab (an IL-6 receptor blocking agent) as an add-on therapy in patients with gMG is ongoing.Nevertheless, the efficacy of targeted IL-6 treatment for MG remains unsupported by empirical evidence.
Tocilizumab, the first humanized anti-IL-6 receptor monoclonal antibody (mAb), has been approved for the treatment of rheumatoid arthritis (RA) and juvenile idiopathic arthritis and is now being investigated in neuromyelitis optica spectrum disorder (NMOSD), with promising results in preventing disease relapse. 11Data from long-term cumulative safety analysis demonstrated that tocilizumab had a safety profile in the treatment of patients with RA. 12 Although no studies have been performed yet, case reports of refractory MG responding favorably to tocilizumab suggest its possible efficacy in MG. 13 Thus, this study aimed to compare the efficacy and safety of tocilizumab with those of conventional immunotherapy in refractory patients with AChR antibody-positive generalized MG (AChR-Ab+ gMG).

| Study design
This single-center prospective cohort study, which was based on patients from an MG registry study (https:// neuro reg.se/ ) in China that covered more than 2000 patients with MG, was approved by the Ethics Committee of Tangdu Hospital, Fourth Military Medical University (202102-06).All the patients provided written informed consent for registration in the Tangdu MG registry.Patients receiving tocilizumab provided independent written informed consent before treatment and were not incentivized to participate in this study.This study followed the Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines for cohort studies and was registered with the Chinese Clinical Trial Registry (ChiCTR2100043273).

| Participants
We identified patients in the MG registry study who received tocilizumab or conventional immunotherapy (i.e., prednisone, azathioprine, tacrolimus, cyclosporin A, and mycophenolate mofetil) between February 10, 2021 and September 30, 2021.Adult refractory patients with AChR-Ab+ gMG were eligible for inclusion.Patients with refractory MG were defined as the patients had an inadequate response (MG Post-Intervention Status is unchanged or worse) to ≥1 immunosuppressants used in adequate doses for an adequate duration (including daily prednisone at doses ≥30 mg over a minimum of 3 months and 3 or more months treatment with azathioprine ≥150 mg/day, tacrolimus ≥3 mg/day, cyclosporin A ≥ 100 mg/day, and mycophenolate mofetil ≥1500 mg/day since disease onset), with persistent symptoms or side effects (including low blood cells, severe osteoporosis, femoral head necrosis, alopecia, obesity, gastrointestinal ulcers, tremor, psychiatric symptoms, etc.) that limit functioning defined by the patient and physician. 14Eligible patients should be categorized to have Myasthenia Gravis Foundation of America (MGFA) class II to IVa; and had a Quantitative Myasthenia Gravis (QMG) score of ≥11 or a Myasthenia Gravis Activities of Daily Living (MG-ADL) score of ≥5 (with <50% of the score due to ocular symptoms).
The exclusion criteria were as follows: active hepatitis B, seropositivity for hepatitis C, seropositivity for human immunodeficiency virus with a low CD4 count, intravenous immunoglobulin (IVIg) or plasma exchange within 1 month of screening, receiving biologics such as rituximab and eculizumab within 6 months of screening, undergoing thymectomy within 6 months or planning to undergo thymectomy in the next 3 months of screening, and concurrent invasive thymoma.

| Data collection
Demographic and clinical data were collected from the Tangdu MG registry database and validated against medical records.Anti-AChR antibody was tested in all patients using a cell-based assay (CBA) method, which was reported have a higher sensitivity and specificity than ELISA and radioimmunoassay methods. 15The following baseline data were collected: sex; age; age at onset; disease duration; body mass index; comorbid autoimmune diseases (e.g., the presence of systemic lupus erythematosus, multiple sclerosis, RA, Hashimoto's thyroiditis, and NMOSD); seropositive for antititin antibody; and thymic status classified as normal, hyperplasia, thymoma, or other types based on radiology and pathological evaluations (patients who underwent thymectomy were evaluated based on pathology findings and those who did not undergo surgery were evaluated based on radiology findings); previous thymectomy; time from disease onset to surgery; MG crisis history; QMG, MG-ADL, and Myasthenia Gravis Composite (MGC) scores; dose and duration of prednisone and NSIST previously and at enrollment; and previous treatment with biologics such as rituximab and eculizumab.

| Interventions
Investigators grouped patients into the tocilizumab group or the control group based on whether they received tocilizumab treatment.The treatment schedule was determined by the patient and physician together.Investigators were responsible to follow up prospectively after achieving informed consent forms.Patients in the tocilizumab group received intravenous tocilizumab (8 mg/ kg every 4 weeks) for 24 weeks and were allowed to keep their original treatment regimen including prednisone and NSIST.The dose and administration frequency of tocilizumab was based on the recommended treatment regimen for patients with RA.
Patients in the control group received prednisone and/or NSIST.
During the study period, the dose of corticosteroids or immunosuppressants was allowed to be adjusted based on the symptoms change and AEs at each visit.If the symptoms worsened, as defined by an MG-ADL score of ≥2, the prednisone dose increased to 0.5 mg/kg/day until minimal manifestation status (MMS) was attained, and the maximum dose did not exceed 0.75-1 mg/kg/ day.If the symptoms improved, the initial dose was maintained for 4-6 weeks; subsequently, the prednisone dose was reduced by 10 mg every 2 weeks until a dose of 20 mg was reached, with subsequent slowing of the tapering to 5 mg monthly.The prednisone dose was reduced in the same manner in cases of major steroid-related AEs.Treatment information and drug-related AEs were recorded at each visit.

| Outcomes
The primary efficacy outcome was the mean difference in the MG-ADL score change at each study visit (weeks 4, 8, 12, 16, 20, and   24), corresponding to that at the baseline between the two groups.
Secondary outcomes included the following: (1) the mean difference in QMG and MGC score changes compared with that at the baseline; 7][18] To minimize subjective differences among assessors, all patients' assessment was conducted by an independent assessor who was not involved in the patients' treatment.Additionally, we also established unified inquiry standards for each item of QMG to reduce the bias.

Safety was assessed based on AEs reported by investigators or participants. All AEs were recorded according to Common
Terminology Criteria for Adverse Events (version 5.0).

| Follow-up
Assessments were performed every 4 weeks for up to 24 weeks.All participants in the tocilizumab group underwent face-to-face visits for 24 weeks.Six participants in the control group failed to perform face-to-face visits at 4 or 8 weeks owing to the impact of COVID-19 pandemic.We performed visits by video using a uniform case report form and collected MG-ADL scores and treatment feedback.To avoid the effect of video follow-up on the collected data, the same neurologist performed the assessment with a unified questionnaire that was used as face-to-face visit.

| Statistical analyses
We calculated the sample size using a repeated measures design by comparing the means of the two groups. 19,20As a preliminary study, there were no relevant data available for reference, so we calculated the sample size according to the results of other clinical trials in MG as well as conservative intraclass correlations.The planned sample size of 20 participants can provide 84% power to detect a difference between the two groups on the primary efficacy outcome (mean change in MG-ADL score from the baseline to week 24).A design with 7-times repeated measures and a compound symmetry covariance structure were employed.Assuming a 4 mean difference of MG-ADL and a standard deviation of 4 between the two groups, and the intraclass correlation coefficient was 0.5 and an alpha level was 0.05.
We conducted normality tests on all continuous variables using the Shapiro-Wilk test.Continuous variables were presented as the mean (standard deviation) for normally distributed data and median (interquartile range) for non-normally distributed variables.Quantitative data are presented as numbers (percentages).
Baseline characteristics between the two groups were compared using the Student's t-test or Mann-Whitney U-test for continuous variables and Fisher's exact test for categorical variables.A generalized estimating equation (GEE) model was used for the primary outcome analysis.Baseline MG-ADL, QMG, and MGC scores and daily prednisone dose were separately included in the GEE model as covariates.A pairwise comparison was performed to compare the mean differences in the MG-ADL, QMG, and MGC scores and daily prednisone dose at each study visit between the two groups.
The proportions of MG-ADL and QMG responders were analyzed using Fisher's exact test.In addition, MG-ADL and QMG responders were redefined based on different improvement threshold points (a ≥3-, 4-, 5-, 6-, or 7-point reduction in the MG-ADL score and ≥6-, 7-, 8-, 9-, 10-, or 11-point reduction in the QMG score) and compared between the two groups at week 24.There were no missing data for the MG-ADL.The missing data mainly associated with QMG and MGC, and a recorded QMG and MGC was available for 200 of 206 total visits (97.1%).The missing data, accounting for 2.9%, was not imputed by multiple imputation approach, because GEE model can address the issue of randomly missing data. 21nsitivity analyses for the primary and secondary outcomes  S1.

| Role of the funding source
The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

| Demographic and clinical characteristics
Overall, 275 patients with MG were registered during the study period, among whom 37 fulfilled the inclusion and exclusion criteria.
Three patients in the control group withdrew from the study due to receiving rituximab; thus, 34 patients were included in the final analysis (Figure 1).The individual immunotherapy courses of the 34 patients are shown in Table S2.S2).There were no significant differences observed in any other baseline characteristics between the two groups (Table 1).p < 0.001), which was sustained until week 24 (adjusted mean difference, −8.0; 95% CI, −12.5 to −3.5; p = 0.002; Table 2; Figure 2).The individual change in the MG-ADL, QMG, and MGC scores as well as daily prednisone dose from the baseline to 24 weeks are shown in Figure S1.

F I G U R E 1
It was noted that individual scores were significantly higher at two follow-ups, especially at week 24.The reason for this phenomenon was two young women in the tocilizumab group had transient worsening of symptoms during menstruation.At week 24, these two patients had menstruating, resulting in higher QMG and MGC scores.
The daily prednisone dose was reduced more significantly in the tocilizumab group (adjusted mean difference, −6.8 mg; 95% CI, −11.0 to −2.7; p = 0.007) at 4 week and sustained until 24 weeks (adjusted mean difference, −13.1 mg; 95% CI, −19.4 to −6.7; p < 0.001; Table 2; Figure 2).The adjusted means of all score measures and daily prednisone dose calculated using the GEE are shown in Table S3 and

| Sensitivity analyses
In the sensitivity analyses, the proportions of MG-ADL and QMG responders from week 4 to 24 were larger in the tocilizumab group than in the control group (95.0% vs. 35.7% in the MG-ADL score, 90.0% vs. 42.9% in the QMG score at week 24, p < 0.001; Table S4; Figure 3A,B).Additionally, the proportion of patients with higher levels of improvement in the MGADL, QMG, and MGC scores was significantly greater in the tocilizumab group at week 24 (up to a 7point reduction in the MG-ADL, 45% vs. 7.1%, p = 0.02; up to an 11point reduction in the QMG, 40% vs. 7.1%, p = 0.05; up to a 12-point reduction in the MGC, 50% vs. 7.1%, p = 0.01; Figure 3C-E).
The results of GEE analyses showed that group, time, and groupby-time interaction effects were observed in the full, simplified, and final models, and the adjusted mean differences of the MG-ADL, QMG, and MGC scores and daily prednisone dose changes were consistent with the main outcomes (Tables S5-S8).

| Safety
Tocilizumab showed acceptable safety profiles, with a total of 10 tocilizumab-related AEs.All were graded 1/2 and manageable.
Details of the AEs are shown in Table S9.refractory cases. 22Recently, innovative drugs have emerged as important therapeutic tools that promise to change the treatment algorithm for MG.4][25][26][27][28] The RINOMAX trial found that a single low rituximab dose was clinically effective for newly diagnosed gMG. 29With two reports of new therapies for MG targeting FcRn and C5, 4,5 a new era for the treatment of this autoimmune disease is emerging.

| DISCUSS ION
Reportedly, IL-6 drives the pathogenesis of MG, which is supported by the link between IL-6 and B cells and follicular helper T (TFH) cells. 30Previous studies have identified IL-6 as a promoter of the differentiation and proliferation of B cells and an inducer of the maturation of B cells into antibody-producing plasma cells, as well as the generation of TFH cells.Theoretically, the blockage of IL-6 is supposed to suppress various ongoing autoimmune diseases. 31cilizumab, a humanized IL6-receptor mAb, can reduce the risk of NMOSD relapse as compared to azathioprine.In patients with RA when rituximab failed despite adequate B cell depletion, IL-6-directed TA B L E 2 Efficacy for the primary and secondary outcomes by generalized estimated equation analysis.therapy may be an effective treatment choice. 32Satralizumab, a pharmacologically optimized IL6-receptor mAb, has been approved for the treatment of NMOSD. 33A phase III clinical trial aiming to evaluate the effectiveness and safety of satralizumab in MG is still ongoing (NCT04963270).NMOSD and MG are mediated by pathogenic autoantibodies of AQP4 and NMJ postsynaptic proteins (AChR, MuSK, and LRP4), respectively.These two diseases share some similar pathogenic mechanisms including the breakdown of tolerance, the collaboration of T cells and B cells, imbalances in T helper 1 (Th1)/ Th2/Th17/regulatory T cells, and complement system activation.
A prospective, open-label, single-arm study has evaluated the efficacy and safety in patients with AChR+ gMG in a 48-week follow-up study. 34Tocilizumab was associated with a good clinical response and safety, as evidenced by significant improvements in QMG score (from 15. if started early after diagnosis. 29It is warranted to further explore the efficacy of tocilizumab monotherapy and of initiation in early-stage.
The reported AEs associated with tocilizumab in our study were consistent with its known safety profile in RA. [36][37][38] The long-term safety profile of tocilizumab (mean treatment duration,

years) from several phase III trials and open-label extensions
did not demonstrate an increase in serious or treatment-related AEs, and the data were overall consistent with those of postmarketing trials and clinical experience. 39Since NSIST is allowed in the tocilizumab group, the risk of infection should be a concern.
However, no AEs of infection were reported in our study, because the steroid tapering was allowed for patients when symptoms improved.In addition, 40% (8/20) patients in the tocilizumab group did not use NSIST due to prior treatment failure or intolerance to side effects.Notably, three of four young or child-bearing period female patients had irregular menstruation, and we found they took tacrolimus that is known to cause menstrual irregularities.
Their menstrual cycles returned to normal when they ceased tacrolimus.It seems that menstrual irregularities might be attributed to the use of tacrolimus.Due to the short observation period and small sample size, the exact cause of the menstrual irregularities cannot be confirmed and the risk of tocilizumab in MG may be underestimated.A study with larger sample size will be needed to study the safe profile of tocilizumab.
In the era of innovative drugs, physicians should know the cost/ effectiveness ratio for disease treatment and their availability.
Economic aspects are also important sources of high disease burden.For eculizumab, with an estimated annual cost of $653,100, the incremental cost-effectiveness ratio was $5,210,000 per QALY/evLYG.
Additionally, its high availability promises the sustainability of tocilizumab in clinical practice, particularly in the developing countries.
This study had several limitations.First, the sample size was relatively small, which may introduce selection and measurement bias, potentially undermine the validity and reliability of the overall representation.The measurement and random errors may compromise the statistical accuracy of the results, thus limit the external validity of the outcomes to a larger population.Although the repeated-measures design indicated that the sample size was adequate to provide statistical power to confirm group differences, larger sample sizes was still warranted to provide more accurate information.Second, the observational design might introduce confounding bias and selection bias despite the relatively balanced baseline characteristics.Thus, we performed several sensitivity analyses to further evaluate the impact of potential confounders on the stability of the primary outcome, which were consistent with the main analyses.However, it is still impossible to eliminate unincorporated confounding factors that may result in bias.This issue will be addressed in an ongoing multicenter randomized controlled trial (NCT05067348).In addition, we further expand the cohort size and conduct long-term observation of the existing cohort.

| CON CLUS ION
In conclusion, tocilizumab is safe and effective in improving the MG-ADL score and reducing prednisone dose in refractory patients with AChR-Ab+ gMG.Refractory MG is still a challenge for both patients and neurologists.Based on the above results, tocilizumab has the potential to be a valuable therapeutic option for such patients.The ongoing controlled randomized trial and the extension study of tocilizumab will provide additional insight on the effectiveness and long-term safety of tocilizumab in refractory AChR-Ab+ gMG.
were performed by introducing different covariates and factors into the GEE model.Three models were constructed.The full model included group-time-interaction effects, baseline measures of outcomes, and other baseline variables; the simplified model included group-time-interaction effects, baseline measures of outcomes, and variables with significant fixed effects of the full model; and the final model included group-time-interaction effects and baseline measures of outcomes.p Values of <0.05 were considered statistically significant.All analyses were performed using R 4.13 (R Foundation for Statistical Computing), and the packages used in the analyses are shown in Table Of the 34 patients, 20 (58.8%) were female individuals.The age was 50.5(20.5)years, while the age at onset was 46.7 (20.8) years.The disease duration was 36.5(18.3,61.8) months.Twenty (including 12 female) patients received tocilizumab, while 14 (including 8 female) patients received conventional immunotherapy.Patients in the tocilizumab group had slightly worse symptoms than those in the control group, which was reflected by the MG-ADL score (mean 9.0 vs. 7.9), QMG score (mean 16.7 vs. 15.0) and MGC score (mean 16.8 vs. 15.6)but did not reach statistical significance.No significant differences were found in daily prednisone dose between tocilizumab and control group (mean 25.0 ± 14.1 vs. 26.4± 12.6 mg, p = 0.76).In each of the two groups, two patients had previously undergone rituximab administration at a dosage of 500 mg, with a frequency of once every 6-8 months, and both individuals received more than two treatment cycles.Despite these interventions, the desired treatment outcome was not achieved (Table

| 5 of 11 RUAN
Study flowchart showing the selection process for patients with MG.Three patients whose treatment was switched to rituximab after conventional immunotherapy failed withdrew from the study in the control group.MG, myasthenia gravis.et al.TA B L E 1 Baseline demographic and clinical characteristics of the patients.
5 at the baseline to 4 at week 12) and MG-ADL score (from 14.5 at the baseline to 4 at week 12), decreased AChR antibody titers (from 15 at the baseline to 6.8 at week 12), and discontinued prednisone without serious AEs.These findings suggest that IL-6 receptor could be a therapeutic target for MG, although future studies involving larger sample size and a control group are still required.Current rapid-acting treatments in MG include IVIg administration, plasma exchange, and steroid use.Tocilizumab has a rapid onset of action, with efficacy observed as early as week 4 in our study, making it a potential first-line treatment option that may be able to partially address the limitations of plasma supply and AEs because of steroid use.Additionally, tocilizumab led to a remarkable symptom improvement compared with conventional immunotherapy.Many patients had improvement beyond the clinically meaningful threshold, achieving up to 7-point reductions in MGADL and 11-point reductions in QMG.As an add-on treatment to baseline immunosuppressants, tocilizumab was shown to make the process of prednisone reduction faster.It is noteworthy two patients in the cohort received tocilizumab monotherapy because of treatment-resistance or intolerance to the side effects of conventional drugs.Their symptoms improved greatly with MG-ADL reducing from 12 at the baseline to 3 at week 24 and from 13 at the baseline to 5 at week 24, respectively, suggesting the efficacy of tocilizumab monotherapy.In multiple sclerosis, an earlier start of effective treatment is associated with reduced damage to neuro-axonal connections. 35In MG, low-dose rituximab showed efficacy in some patients with generalized AChR-gMG, especially F I G U R E 2 Mean changes in MG-ADL (A), QMG (B), MGC (C) scores, and the daily prednisone dose (D) from the baseline to 24 weeks in the tocilizumab and control groups.The line charts show mean change in all scores measure and the daily prednisolone dose from the baseline to 24 weeks in the tocilizumab and control groups with data points showing mean scores and error bars showing 95% confidence interval of mean scores.MG-ADL, myasthenia gravis activities of daily living; MGC, myasthenia gravis composite; QMG, quantitative myasthenia gravis.

F I G U R E 3
Sensitivity analyses.Proportion of the MG-ADL responders (A) and the QMG responders (B) form week 4 at week 24 in the tocilizumab and control groups.Minimum point improvement in MG-ADL (C), QMG (D), and MGC (E) scores at 24 weeks from the start of therapy.MG-ADL, myasthenia gravis activities of daily living; MGC, myasthenia gravis composite; QMG, quantitative myasthenia gravis.
Eight patients previously used two NSIST and one patient used three NSIST.
lizumab with those of conventional immunotherapy for the treatment of refractory patients with AChR-Ab+ gMG, providing new insights into the possibility of treating MG with IL-6 receptor blockers.Currently, advances in MG treatment have led to a marked reduction in disease-associated mortality, however, an unmet need still exists for the management of the disease especially for the b GEE model separately included the baseline MG-ADL, QMG, MGC scores and daily prednisone dose as covariates. a