Introduction
B-cell-depleting anti-CD20 monoclonal antibody therapies have demonstrated significant clinical efficacy in individuals with neuroinflammatory disorders and are increasingly gaining traction as a therapeutic approach. In particular, open label data consistently demonstrate favourable efficacy for rituximab (a chimeric anti-CD20 monoclonal antibody) on clinical outcomes in patients with neuromyelitis optica (NMO) [
1] as well as a reduction in MRI and clinical measures of disease activity in phase-II studies of relapsing multiple sclerosis (MS) [
2‐
4]. More recently, ocrelizumab (a fully humanised anti-CD20 monoclonal) has also been shown to be effective in relapsing MS and is the first therapeutic agent to have exhibited a reduction in disability progression in a phase-III study of primary progressive MS [
5,
6]. It has gained FDA approval and is expected to become a key drug for MS globally in the next few years.
Anti-CD20 monoclonal antibody therapy is typically used as maintenance therapy when it is prescribed for NMO or MS. This is in contrast to its widespread use in many non-neurological diseases including rheumatoid arthritis and anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis, in which anti-CD20 therapy is usually employed as a short-term remission-inducing agent before the subsequent introduction of an alternative maintenance therapy [
7]. While the use of maintenance anti-CD20 therapy has increased significantly in neurology, long-term safety data on its use in neuroinflammatory disease remain scarce.
A profound and durable depletion of circulating B cells occurs within days of rituximab infusion [
8,
9]. The fact that plasma cells do not express CD20 and are, therefore, resistant to the immediate depletory effects of anti-CD20 therapy is expected to preserve humoral immunity [
10], but secondary antibody deficiency does occur [
9,
11‐
14]. However, its frequency or consequences in neurological diseases have not been fully explored.
Here, we present a UK cohort of 50 individuals with NMO treated over extended periods with rituximab, five of whom developed serious infection in the context of secondary antibody deficiency. We consider the implications of this case series and outline potential considerations in the long-term use of anti-CD20 monoclonal antibodies in NMO and other neuroinflammatory diseases.
Patient identification and acquisition of laboratory data
Management of individuals with NMO in England is coordinated by the Walton Centre, Liverpool, and John Radcliffe Hospital, Oxford. Patients in Wales are referred to regional specialist services, including Cardiff. Disease-specific databases and detailed longitudinal clinical data are available in all centres. Interrogation of the Liverpool and Cardiff datasets was used to identify any individual with a diagnosis of NMO or NMO spectrum disorder (NMO-SD) who had been treated with rituximab since 2007 (Liverpool n = 50, Cardiff n = 6). Results of laboratory investigations including serum immunoglobulins were known for 50 out of 56 patients. Review of case records was undertaken in a subset of patients known to have experienced severe infection (defined as requiring hospital admission and IV antibiotics) in the context of hypogammaglobulinaemia (reduced blood concentration of IgG, IgM and/or IgA).
All patients provided consent for their data to be used as part of ethically approved research studies (reference numbers 15/L0/1433 and 05/WSE03/111). Immunoglobulin levels (IgG, IgA and IgM) were assayed by nephelometry (Siemens BN2 Nephelometer; Siemens) and specific antibody titres against
Haemophilus influenzae (Hib),
Clostridium tetani (tetanus) and Pneumococcal capsular polysaccharide were determined by ELISA (The Binding Site, Birmingham, UK). Protective cut-off levels of circulating specific antibodies were tetanus > 0.1 IU/ml,
Haemophilus influenzae > 1mcg/ml and Pneumococcal antibodies > 50 mg/L [
15].
A total of 33 archived serum samples were available, from disease-specific biobanks, from the five cases with severe infection. Samples were analysed for immunoglobulin levels over 14.6 patient-years (mean 2.7 samples per patient per year). A total of 19 samples were analysed for disease-specific Igs over 15.3 patient-years (mean 1.3 samples per patient per year).
Discussion
B-cell-depleting anti-CD20 monoclonal antibody therapies demonstrate favourable clinical efficacy for individuals with CNS inflammatory disease but long-term safety data are scarce. Hypogammaglobinaemia was recorded in 64% of patients (32 out of 50) for whom data were available in this UK cohort. Severe infection associated with secondary antibody deficiency occurred in 5 out of 50 (10% of the entire cohort). This value may underestimate the true prevalence, as infection and immunoglobulin data were not systematically surveyed in all patients.
This case series illustrates several additional insights. First, rituximab had sustained clinical efficacy in patients that had not been achieved with prior disease-modifying therapies (mean pre-treatment annualised relapse rate of 1.8 versus 0.4 post-rituximab). This was a key factor influencing the decision to continue using rituximab in all cases, despite the episodes of infection. Second, laboratory and clinical surveillance is required to detect the earliest signs of these treatment-associated infections. Third, a combination of serum immunoglobulin measurement plus disease-specific antibody titres appears optimal to determine risk, maximise the opportunity for preventative measures and avoid the development of bronchiectasis.
A profound and durable depletion of circulating B cells occurs within days of rituximab infusion [
8,
9]. As plasma cells do not express CD20, they are unaffected and have hitherto provided some assurance that humoral immunity would be sufficiently preserved [
10]. However, it remains unclear whether long-term humoral immunity results entirely from a self-sustaining long-lived plasma cell (LLPC) population that survive for decades, or whether regular replenishment of plasma cells by memory B cells is required every few months [
16]. Despite the apparent survival of LLPCs after B-cell depletion [
17‐
19], several clinical studies have observed a dose-dependent reduction in serum immunoglobulins following anti-CD20 therapy [
11,
12,
14,
20]. This implies that long-term humoral immunity may be more reliant on replenishment of plasma cells from the B-cell progenitor pool than previously thought.
The finding that B-cell depletion lasts an average of 6 months [
21], led to regular 6-monthly dosing of rituximab during many early regimens. However, the considerable inter-individual variation in repopulation time has led several authors to recommend retreatment guided by B-cell repopulation, mainly to avoid risk of neurological relapse in those who repopulate early. The use of CD19 + (a pan B-cell marker) monitoring in NMO was initially suggested but the incidence of NMO relapses occurring even at low levels of CD19 + B-cell repopulation suggests this approach may compromise disease control [
9,
11]. The observation that the return of CD27 + (memory B cells) coincides with the return of disease activity in rheumatoid arthritis prompted use of CD27 + B-cell monitoring to guide retreatment of NMO [
11,
22]. Retreatment guided by CD27 + repopulation has been reported to lower the cumulative dose of rituximab while maintaining remission in the majority of patients (91 out of 100 cases) but is not universally available in clinical laboratories [
11,
23].
These data may have implications for use of similar B-cell depleting drugs, notably the newly approved ocrelizumab for MS. Despite considerable inter-individual variation observed in time-to-repopulation of B cells following ocrelizumab [
24], dosing is recommended to occur at regular 6-monthly intervals [
25]. No data were reported on immunoglobulin levels in phase-III ocrelizumab studies in MS but there was no excess of serious infection in ocrelizumab-treated patients during the 96 weeks of follow-up [
6,
25]. It is possible that the risk of secondary antibody deficiency may be higher in an NMO population where other immunosuppressive agents often precede anti-CD20 therapy. The subnormal IgG and pneumococcal antibodies observed pre-rituximab in case 4 (who had already received mitoxantrone, azathioprine and mycophenolate over 10 years) highlight the possible contribution of prior immunotherapy to secondary antibody deficiency. A review of risk factors predisposing to the development of hypogammaglobulinaemia and infections post-rituximab identified low immunoglobulins prior to treatment, maintenance therapy with rituximab, prior immunosuppression, concomitant purine analogues such as mycophenolate and chronic lung or heart disease as well as older age [
26]. In four out of five of our cases, the combined duration of previous immunosuppressive therapy and rituximab treatment exceeded 5 years, suggesting that secondary antibody deficiency may be a dose- and duration-dependent phenomenon. Therefore, it is possible that other CD-20-depleting drugs including ocrelizumab may with long-term treatment have similar effects. However, case 3 had received only 4 months of immunosuppression and 11 months of rituximab therapy before developing severe infection associated with secondary antibody deficiency. Possible explanations for the early development of severe infection in this case include the co-administration of maintenance corticosteroid.
The occurrence of recurrent or complicated infections of the upper and/or lower respiratory tract in our case series is typical of antibody deficiency and illustrates the need for surveillance that extends beyond blood monitoring [
27]. Monitoring of patients on anti-CD20 therapy should include regular questioning on infective burden with a particular focus on sino-pulmonary symptoms.
In retrospect, using archived sera, we were able to demonstrate downward trends in serum immunoglobulin levels that occurred after the commencement of rituximab. Due to low awareness of this complication of B-cell-depleting therapies, serum immunoglobulin surveillance was not routinely undertaken in this series until recently, leading to delays in detection and treatment in some cases. However, the hypogammaglobulinaemia that coincided with severe infection was often modest and the presence of subnormal baseline protective levels of disease-specific antibodies to HiB and/or pneumococcus suggests that they may add additional functional information to help stratify for the risk of infection during anti-CD20 therapy. Test vaccination as used in the assessment of humoral immunodeficiency may also play a role.
Screening of patients before rituximab therapy for serum immunoglobulins and disease-specific antibodies opens up the possibility for therapeutic vaccination, which is more challenging once B cells are depleted. Detection of on-therapy secondary antibody deficiency ought to prompt close infection surveillance and consideration of back-up or prophylactic antibiotics and/or IGRT to prevent infection and end organ damage. If further pooled long-term data confirm secondary antibody deficiency as a prevalent complication, management algorithms allied to those used for primary antibody deficiency may be required [
27].
In summary, this case series highlights a serious complication of anti-CD20 depletion therapy when used as remission maintenance treatment for neuroinflammatory disease. While NMOSD is a relatively rare disease, this observation is potentially generalisable to other diseases treated with B-cell depletion. It is particularly important in the ocrelizumab era of MS treatment. Progressive MS patients who may benefit from ocrelizumab may have additional comorbidities that predispose them to infections. Heightened awareness of this potentially preventable and treatable complication is crucial to avoid added morbidity and mortality and to allow patients to continue benefiting from a highly effective mode of therapy.