Preliminary data indicate that B cell depletion may be effective in autoimmune disease in the areas of rheumatology, nephrology, neurology and dermatology [7]. A greater amount of evidence for the effectiveness of B cell depletion has been gathered in rheumatoid arthritis (RA), with recent emerging data indicating that B cell depletion may also be effective in the treatment of antineutrophil cytoplasmic antibody (ANCA)-associated granulomatous vasculitis [8-10]. Rituximab, a monoclonal antibody that selectively targets CD20+ B cells and leads to their depletion, has demonstrated significant efficacy and a good safety profile in clinical trials conducted in patients with active RA [11-17]. The long-term efficacy and safety of rituximab in RA is particularly relevant as many of the autoimmune diseases are relatively rare, and as such, clinical development of drugs for these conditions will be less likely.

In the current analysis across autoimmune diseases, the overall rate of infections and serious infections compares well with analyses of rituximab in RA [17] and SLE [63]. There are few studies that provide controlled trial data in other autoimmune diseases; however, in the HERMES study where relapsing-remitting multiple sclerosis was treated with rituximab or a placebo, a greater proportion of patients in the placebo group had serious infectious events than those receiving rituximab (5.7% versus 2.9%) [56]. The strict exclusion criteria employed in this study restricted entry of high-risk patients, and so this study does not truly reflect what might be expected in a real-life clinical setting. Overall, the current evidence would suggest that serious infections occur in a small proportion of patients with autoimmune diseases during rituximab therapy. In contrast, two controlled studies in SLE (EXPLORER) and lupus nephritis (LUNAR) have shown slightly elevated serious infection rates of 9.5% [59] and 16.4% [58], respectively, during rituximab therapy in combination with SOC (including mycophenolate mofetil (in EXPLORER and LUNAR), and methotrexate and azathioprine (in EXPLORER)) and high-dose corticosteroids, which might indicate that the regimens studied may not have been appropriate for these populations. An elevated serious infection rate was also reported in the rituximab group of the RITUXIVAS study, although, these patients with ANCA-associated vasculitis also received cyclophosphamide and high-dose corticosteroids as part of the regimen, and no differences in the rates were shown as compared with the control group, which received a cyclophosphamide-based regimen (18% for each regimen) [20]. In these three studies, serious infection rates were similar or elevated in the control groups compared with the rituximab regimens. Therefore, and in addition to, a potential relationship with underlying disease, concomitant immunomodulating therapies cannot be ruled out as a cause [20,58,59]. In particular, corticosteroids have an important additional effect on T cells and have been shown to be an important predisposing factor for infections [64]. Moreover, reduced IgG levels at baseline have been identified as risk factor for infections in RA [17] for which insufficient data were available in the registry to search for their role in modulating infectious risks and so further studies are needed. Data on Ig levels and B cell depletion were lacking from this current analysis, and therefore, these data are needed in subsequent registry studies.

Of interest is that the infection (serious and non-serious) rates over time remained stable with multiple courses of rituximab in RA [17]. In addition, it has been reported that retreatment with rituximab based on treatment to target DAS28 remission in patients with RA compared with an 'on demand' approach resulted in better disease control with comparable safety profiles [65]. Furthermore, retreatment with rituximab at a fixed interval of six months provided greater improvements in disease control at one year with a serious infection rate similar to that in patients with RA who received a single course [15,66]. These data might suggest that control over disease activity may be closely related to safety and the risk of infections. With the suggestion from the current clinical outcomes data that responses may be improved with greater doses of rituximab in patients with various autoimmune diseases, a stable safety profile over time will provide physicians with the confidence to prescribe multiple courses of rituximab for patients with autoimmune conditions. Although most of the complex systemic diseases reported here may require higher doses per se, dose-findings studies for rituximab within the individual diagnoses are still lacking and preclude any far-reaching conclusions.

A significant relationship between higher doses and a reduced rate of infections (Poisson regression model) was also demonstrated in this analysis. However, it cannot be excluded that infections led to discontinuation of therapy and resulted in lower treatment doses. Indeed, the majority of clinically relevant infections occurred within seven months of the first rituximab infusion, likely reflecting that severe infections led to discontinuation of rituximab. Therefore, a selection bias of patients initially developing infections remains possible. Alternatively, control of disease activity by successful rituximab therapy may have led to reduced disease activity and reduced subsequent infections as suggested by fixed-dosing intervals of rituximab in RA [65]. The overall safety data of this registry do not indicate that certain diseases may have a higher risk for any adverse event, including a higher propensity of infections. This conclusion can be further substantiated by the observation that during rituximab therapy the lowest frequencies of certain co-therapies were registered (Table ​(Table3),3), which reflects the overall clinical effect by this B cell directed therapy as well as that these co-therapies likely do not substantially confound the safety database. The increase of various therapeutic modalities after rituximab therapy is likely related to increased disease activity or when insufficient response to rituximab required new treatments.

Although a direct comparison cannot be made between the safety analysis of rituximab in patients with RA [17] and the current study for several reasons (heterogeneous patient populations, different dosing regimens and different study periods), it appears from the long-term analysis in RA that IRRs diminish over time; therefore, it would not be unreasonable to expect that the rate of IRRs over time in various autoimmune diseases might decrease with each course of rituximab. However, caution should be taken in interpreting these data as patients with more severe IRRs may discontinue rituximab and, therefore, provide a distorted impression. In addition, IRRs were defined in this analysis as events that occurred during or within 24 hours of an infusion, which does not capture those that may have occurred after one week, for example; therefore, the data recorded in this registry may represent an under-reporting of the level of IRRs. Fortunately, the vast experience with rituximab in RA and NHL has led to effective techniques in managing these infusion-related complications.

In the RA clinical trial programme for rituximab, there were few deaths that occurred in patients in the rituximab groups [12,17]. The number of deaths across autoimmune diseases appears to be relatively high considering the short observation period of the current study, even though the proportion of deaths is no greater than that reported in the Spanish registry; however, the follow-up period in the Spanish registry was considerably longer than that of the current study [60]. In placebo-controlled trials investigating the efficacy and safety of rituximab in autoimmune diseases for one year or more, the proportion of deaths appears low (< 3%) [56-59]. The discrepancy in safety profiles between the real-life, observational studies and randomised, controlled trials may be explained by consideration of the study entry criteria. In real-life studies, the entry criteria are more 'relaxed' than in randomised, controlled trials, which often result in higher-risk patients being included; consequently, outcomes such as mortality rates can be elevated in real-life studies. Overall, no specific pattern emerged from this analysis of GRAID in relation to the type of infection leading to death, prior therapies, age, gender and dosing of rituximab. The deaths predominantly occurred in a short amount of time following rituximab therapy; however, patients had previously received intensive immunosuppressive therapy and had uncontrolled severe disease, both of which would enhance the risks of severe infection and life-threatening complications [62,67-71]. In a study of patients with refractory RA [12], an infection was complicated by an existing heart condition during rituximab therapy, which led to the death of a patient, thus illustrating the difficulty in clearly relating cause of death in the context of severe confounding risk factors. In the recent RITUXIVAS study, a relatively high number of deaths occurred early in the course of therapy (median time to death: 81 days) [8]; however, overall 18% of patients died in the control group (cyclophosphamide), which compared well with 18% in the treatment arm (cyclophosphamide + rituximab). Together with the data of the current study with the highest death rates among patients with ANCA-associated granulmatous vasculitides, the recent trial results of RITUXIVAS indicate the challenges of this disease and suggest that previous and co-therapies as well as comorbidities play a substantial role in this complication. Patients in this and the RAVE study also had an elevated number of malignancies in a short time period [8,9], which was not observed in the current GRAID analysis. In these challenging and often difficult to treat conditions, therapeutic options are limited and physicians are often required to weigh up the benefit:risk ratio of newer therapies off-label.

Response to rituximab in patients with refractory RA has been proven in a number of well-designed clinical studies [11-16]. The current analysis and previous analyses of registry data [60,61] provide an indication that the effectiveness of rituximab may not be limited solely to the treatment of RA. In addition, our data are very much consistent with reported open-label experiences [21,22,25,26,30-33,36-38,40,43,44,72]. However, the interpretation of these data is limited without the inclusion of a control group, and some placebo-controlled studies of rituximab in specific autoimmune diseases failed to provide conclusive evidence as to its benefit [9,56-59]. The current study differs from the lupus studies (LUNAR [58] and EXPLORER [59]) in that data were collected from otherwise refractory patients while both controlled studies included patients who were stable on SOC at study entry and received rituximab plus SOC in the active group compared with SOC in the control group. Further endorsement of the benefit of rituximab in a refractory population has been provided by the RAVE study [9]. In a subgroup of patients with ANCA-associated granulomatous vasculitis who had relapsed with SOC (in particular, cyclophosphamide), rituximab was shown to be superior to SOC (P = 0.01), whereas in the overall population that included newly diagnosed disease, there was no significant difference shown between rituximab and the control. Therefore, the bulk of evidence would suggest that treatment of autoimmune diseases, other than RA, with rituximab represents a therapeutic option to consider.

An additional interesting analysis of the GRAID data revealed that there were no significant differences in the distribution patterns for complete, partial and no response across the different autoimmune diseases. However, some patient groups were very small and given the lack of prospective use of validated outcome measures, it cannot be excluded that certain diagnoses have distinct response profiles.