Derisking CD20-therapies for long-term use

Highlights

• The balance between risks and benefit changes with long-term use of anti-CD20, and we need evidence-based approaches to reduce anti-CD20 dosing intervals and guide retreatment.

• Vaccinations and adequate infectious screening before starting anti-CD20 are paramount.

• Family planning needs to be discussed upfront with every woman of childbearing age.

• Infusion reactions should be adequately managed to avoid treatment interruption.

• Pharmacovigilance involves measuring immunoglobulin levels annually, counseling adequate fever management, reviewing vaccination status and recommending female patients to actively engage with the local breast cancer screening programs.

Abstract

Anti-CD20 have quickly become the mainstay in the treatment of multiple sclerosis (MS) and other neuroinflammatory conditions. However, when they are used as a maintenance therapy the balance between risks and benefits changes. In this review, we suggested six steps to derisk anti-CD20. Firstly and secondly, adequate infectious screening followed by vaccinations before starting anti-CD20 are paramount. Third, family planning needs to be discussed upfront with every woman of childbearing age. Fourth, infusion reactions should be adequately managed to avoid treatment interruption. After repeated infusions, it becomes important to detect and prevent anti-CD20-related adverse events. Fifth, we recommended measuring immunoglobulin levels and reviewing vaccinations annually as well as counselling adequate fever management. For female patients, we emphasised the importance to engage with the local breast cancer screening programs. Sixth, to fundamentally derisk anti-CD20 therapies, we need evidence-based approaches to reduce dosing intervals and guide retreatment.

Introduction

Anti-CD20-monoclonal antibodies (mAb) have become the cornerstone in the management of several neuroinflammatory conditions, and in particular multiple sclerosis (MS). Currently, rituximab, ocrelizumab and ofatumumab are used in clinical practice depending on national regulations (Hauser et al., 2017; Montalban et al., 2017; Hauser et al., 2020, 2008; Sellebjerg et al., 2020). These mAb differ in many respects: immunogenicity, epitope binding site, the extent of antibody- versus complement-mediated cytotoxicity and mode of administration (Table 1) (Hauser et al., 2017; Montalban et al., 2017; Hauser et al., 2020, 2008; EMA, 2021a, 2021b, 2021c). However, these differences are of unknown importance when it comes to their efficacy to suppress central nervous system (CNS) inflammation (Sellebjerg et al., 2020). Anti-CD20-mAb are currently prescribed as maintenance treatments and suppress B cells from the pre-B cell to early plasmablast stage (Hauser et al., 2017; Sellebjerg et al., 2020). Repeated infusions or injections with anti-CD20 therapy also perturb the development of germinal centres in lymph nodes and spleen and other secondary lymphoid tissue (Ramwadhdoebe et al., 2019). This implies that several B cell functions such as antigen presentation, antibody formation including class-switching, affinity maturation, memory B-cell and plasmablast maturation and antibody production are impaired. Consequently, long standing use of anti-CD20-mAb increases the frequency of decreased IgA, IgM as well as IgG levels which are key players in humoral immunity (Derfuss et al., 2020; Marcinnò et al., 2018; Avouac et al., 2021; Tallantyre et al., 2018). This has implications when being exposed to new infectious agents or vaccines, and results in blunting of vaccine responses (Smets et al., 2021; Kado et al., 2016) and possibly peripheral tumor immunosurveillance (Degnim et al., 2017). Therefore, prolonged and sustained B-cell depletion changes the risk/benefit profile of the treatment and requires rethinking of the existing risk mitigation strategies. In this review, we would like to outline the steps that need to be taken for derisking anti-CD20 therapies for long-term use.

Before starting with anti-CD20-mAb it is necessary to organize the laboratory tests which are summarized in Table 2. This panel is aimed at preparing patients for treatment with anti-CD20-mAb and, in case treatment plans change, to start with any other immunosuppressive therapy. Screening for human immunodeficiency virus 1 and 2, syphilis, hepatitis B (HBV) and hepatitis C virus (HCV) is necessary to exclude a chronic active infection (EMA, 2021b). Especially HCV screening is important as the prevalence of viraemic HCV is 0.64% in the EU and the disease often remains undiagnosed (Razavi et al., 2017). Anti-CD20-mAb pose a uniquely high risk of HBV-associated hepatitis and liver failure, sometimes associated with death (Pei et al., 2010; Reddy et al., 2015). Importantly, patients with serologic evidence of HBV infection were excluded from ocrelizumab trials (Hauser et al., 2017; Montalban et al., 2017). Given the known risk for HBV flares or reactivation, all patients should be screened for the presence of infection with HBsAg and anti-HBc. Patients with any marker of infection (HBsAg⁺ or HBsAg⁻/Anti-HBc⁺) have more than 10% risk of HBV reactivation and should undergo antiviral prophylaxis during and for 12 months after cessation of the therapy (Reddy et al., 2015; Epstein et al., 2018).

Ocrelizumab mildly increased the risk of herpes virus infections in its phase-III trial in MS (2.1% ocrelizumab; 1.0% interferon-β) (Hauser et al., 2017). However, nearly all infections were mild to moderate and there was no need for routine antiviral prophylaxis (Hauser et al., 2017; Epstein et al., 2018). Nonetheless, patients without a history of chickenpox or uncertain VZV vaccination status should be tested for anti-VZV-antibodies and get vaccinated when absent (Farez et al., 2019). Anti-CD20-mAb are not associated with a significant risk of tuberculosis (TB) reactivation given their specificity for CD20 without affecting T cell-mediated immunity (Epstein et al., 2018). Screening for latent TB was not necessary in three phase-III ocrelizumab trials (Hauser et al., 2017; Montalban et al., 2017) and in rituximab trials the risk of reactivation of TB was similar to placebo (Liao et al., 2016). Screening for measles, rubella, mumps is necessary when vaccination status in uncertain (Reyes et al., 2020). Routine cervical/human papillomavirus (HPV) screening according to national guidelines is critical for all women with MS, especially as the use of immunosuppressive drugs is associated with an increased risk of HPV-associated malignancies (Garland et al., 2017; WHO, 2017).

We recommend determining IgM, IgA and IgG levels at baseline to allow differentiation between a primary versus secondary hypogammaglobulinemia later on (Kaplan et al., 2014; Compagno et al., 2014). In a French registry, patients with low IgG levels at baseline (< 6 g/L) were 4.8 times more likely to develop severe infections (Gottenberg et al., 2010).

Immunoprophylaxis is preferable to treatment, reducing both infectious morbidity and mortality as well as interruptions to MS therapy (Epstein et al., 2018). Against this background, one vaccine is mandatory and three are highly recommended before going ahead with anti-CD20 and are directed against Hepatitis B, SARS-CoV-2, influenza and pneumococcus (Table 3). In patients previously vaccinated against HBV, an anti-HBsAg⁺ titre of > 100 IE/L is considered protective (Saco et al., 2018). Repeat vaccination with the standard HBV vaccine dose, increased the rate of seroconversion to 83–87.5% in healthy non-responders (Jafarzadeh et al., 2008). People who are not adequately vaccinated should be advised against starting treatment. In selected unvaccinated individuals with no alternative disease-modifying treatment (DMT) options and an aggressive disease course, an accelerated vaccination schedule could be considered on a case-by-case basis (Nothdurft et al., 2002; Saltoğlu et al., 2003). Regarding COVID-19, several observational studies have shown a small though consistently present increased risk for hospitalisation and/or intensive care unit (ICU) admission in pwMS treated with anti-CD20-mAb, irrespective of other demographic risk factors (Sormani et al., 2021; Simpson-Yap et al., 2021). In a nation-wide Swedish cohort, pwMS appeared at higher risk of influenza-related hospitalisations, with an increased relative risk of 3.57 (CI 95% 3.06–4.15) (Montgomery et al., 2013). The most common serious infections were urinary tract infections, pneumonia, and cellulitis (Wolinsky et al., 2020; Hauser et al., 2020). Ocrelizumab may especially increase the risk of severe pneumonia in patients with swallowing difficulties (Montalban et al., 2017; EMA, 2021b). Importantly, immunoprophylaxis before starting with anti-CD20-mAb needs to be counselled sufficiently as vaccine hesitancy is widespread amongst pwMS (Diem et al., 2021). A minimal delay of two weeks between vaccination and initiation anti-CD20-mAb is recommended (Reyes et al., 2020).

Several other vaccines are optional and need to be considered on a case-by-case basis. As the measles-mumps-rubella (MMR) vaccine is a live-attenuated vaccine, it is highly recommended in antibody-negative patients who did not have the vaccine in childhood (Reyes et al., 2020). If vaccination history is unclear, the meningococcal vaccine against common serotypes also needs to be considered, particularly in at-risk groups i.e. military recruits, university/boarding school students or patients living in crowded communal accommodation. Similar to the pneumococcal vaccine, the MenACWY is a glycoconjugate vaccine in which vaccination response is known to be more dependent on the presence of naïve B cells (Mitchell et al., 2014). The recombinant VZV vaccine has recently been licensed for use in immunocompromised patients after high efficacy against zoster reactivation was shown in adults with solid tumor malignancies receiving chemotherapy (Mullane et al., 2019). It could similarly be of use in pwMS as approximately 2.1 and 1.0% of individuals treated with ocrelizumab or interferon-β, respectively, are affected with zoster in the first 96 weeks of their treatment (Hauser et al., 2017; Smets et al., 2021). Although national funding is unlikely to be available, there are potential benefits of primary HPV vaccination or upgrading immune status from the bivalent (Cervarix) and quadrivalent (Gardasil) to the newer polyvalent vaccine (Gardasil-9). Immunosuppression is associated with an increased risk of HPV-associated malignancies (Garland et al., 2017). In addition, the 9-valent HPV protects against 90% of cervical cancers and 95% of anal cancers vs. 70 and 90% with the quadrivalent vaccine, respectively (Garland et al., 2017). Of note, Gardasil-9^TM has not been used in routine vaccination programs. As delaying treatment to complete the full three-dose HPV vaccine schedule before starting treatment is not justified, we recommend to give the initial two doses (0, 1–2 months) before starting anti-CD20-mAb and the third dose one month before the second infusion (month 12) (Reyes et al., 2020). Alternatively, a recent study showed that just one dose of HPV vaccine is also associated with a significant reduction in cervical disease (Rodriguez et al., 2020).

All women of childbearing potential should have the possibility of pregnancy discussed before starting on anti-CD20-mAb (Dobson and Hellwig, 2021; Dobson et al., 2019). First, anti-CD20 therapies are not known to interact with hormonal contraception (EMA, 2021a; Dobson et al., 2019). Second, the European Medicine Agency (EMA) advises the use of contraception until 12 months after the last anti-CD20 infusion (EMA, 2021a, 2021b, 2021c). However, the risks of congenital malformations or other adverse pregnancy outcomes with exposure to anti-CD20-mAb within this window are likely to be low based on data from rituximab registries (Smith et al., 2020; Mao-Draayer et al., 2020). As both rituximab and ocrelizumab take about 4 months to be eliminated (EMA, 2021a, EMA, 2021b) and as IgG does not cross the placenta to any significant degree until late in the second trimester, a more pragmatic approach would be to advise women to attempt pregnancy around 1–3 months after the last infusion (Dobson and Hellwig, 2021). This minimizes foetal exposure while offering immunomodulatory benefit for several additional months (Dobson and Hellwig, 2021). The ocrelizumab phase-II extension trial monitored patients after receiving three or four treatment cycles during an 18 month treatment-free period (Baker et al., 2020). As there were no reported relapses nor signs of MRI activity during this interval, we suggest that redosing of anti-CD20-mAb in patients trying to fall pregnant only needs to be considered after 12 months (Baker et al., 2020) Third, the EMA recommends to discontinue anti-CD20-mAb when breastfeeding (EMA. 2021a, EMA, 2021b). However, anti-CD20-mAb are only detected at very low levels in breastmilk: < 10 and < 1% of the maternal dose for ocrelizumab and rituximab, respectively. Concentrations below 10% are generally considered safe for an infant to breastfeed (LaHue et al., 2020). Similar to other proteins most immunoglobulins have low oral bioavailability. Reassuringly, no infectious or developmental adverse events were seen in 368 infants followed for ≥ 6 months after exposure to breast milk of mothers treated with multifarious mAb (LaHue et al., 2020).

Adverse advents within 24 h of infusion and infusion reactions (IRR) are common when infusing anti-CD20-mAb, and a potential source of treatment discontinuation and/or absenteeism. Hence, it is important to manage them proactively and provide reassurance for patients that the frequency of these IRR decreases with ongoing infusions. In the pooled OPERA population, 27.5% of ocrelizumab-treated patients experienced one or more IRR with the first infusion versus 7.8% after the fourth infusion (Mayer et al., 2019; Hauser et al., 2017). IRR can be reduced by appropriate prophylactic treatment. The standard pre-treatment medications are methylprednisolone 100 mg, chlorphenamine 10 mg and paracetamol 1 g (EMA, 2021b). Premedication with corticosteroids, particularly in combination with the use of antihistamines, resulted in the highest reduction in IRR frequency (19.2% in OPERA; 16.7% in ORATORIO) (Mayer et al., 2019). Further reductions in IRR frequency can be obtained instructing patients to increase hydration from one day before the infusion and to take cetirizine 10 mg and ranitidine 75 mg on the night before and morning of the infusion (Conte et al., 2019). IRR need to be managed by adjusting the infusion rate and using medications to alleviate symptoms (EMA, 2021b). Symptomatic treatment of IRR include oral paracetamol (1 g) and IM/slow IV antihistamines, such as diphenhydramine (25–100 mg) (Mayer et al., 2019). If a patient experiences a severe IRR (e.g. dyspnea), the infusion should be interrupted immediately and symptoms appropriately treated (EMA, 2021b). However, as the frequency of IRR decreases with ongoing infusions (Hauser et al., 2017) and the time point for reinfusion precedes the median repopulation of CD20⁺ cells (Table 1, Baker et al., 2020) it is probably not necessary to continue the pre-treatment regimens after the first set of infusions. This is especially relevant in the context of COVID-19 as recent use of methylprednisolone in pwMS was associated with worse outcome (Sormani et al., 2021). For 20 mg subcutaneous ofatumumab, injection side reactions (ISRs) are less frequent (20.2% ofatumumab; 15.0% teriflunomide). In addition, there was only limited benefit of premedication on the frequency of ISRs: 18.2% no medication vs. 12.7% any premedication (Hauser et al., 2020).

Treatment with anti-CD20-mAb results in a decrease in total immunoglobulins, mainly driven by reduction in IgM. The 6-year follow-up data of the OPERA and ORATORIO trials showed that 30.7% of patients had developed an IgM deficiency (lower limit of normal (LLN) 0.4 g/L), 6.0% an IgA deficiency (LLN 0.7 g/L) and 7.1% an IgG deficiency (LLN 5.65 g/L) (Derfuss et al., 2020). No long-term data are available for ofatumumab, but the frequencies of low IgM and IgG at 96-weeks suggest a similar pattern (Hauser et al., 2020). Real-world data in other neuroinflammatory conditions tend to report higher numbers. In a group of 15 neuromyelitis optica spectrum disorder (NMOSD) patients receiving a mean of six rituximab infusions during a median of 70 months, 73% showed reduced IgG level (<7 g/L) and 20% severe reduced IgG level (<4 g/L) (Marcinnò et al., 2018). Amongst 50 NMOSD patients receiving rituximab for several years, the proportion with reduced IgG level (< 4.5 g/L) was 38% (Tallantyre et al., 2018). In a cohort of mostly treatment-naïve individuals, the frequency of persistent reduced IgG level < 7, 6, and 4 g/L after rituximab was 31%, 23%, and 6%, respectively (Avouac et al., 2021). In RA, the most important risk factors to develop hypogammaglobulinemia were the number of rituximab infusions and age at first infusion (Christou et al., 2017). While no serious consequences have so far been reported for IgM or IgA hypogammaglobulinemia, a link between low levels of IgG and microbial infections is clearly documented (Derfuss et al., 2020; Avouac et al., 2021; Kridin and Ahmed, 2020; Md Yusof et al., 2019; Vollmer et al., 2020).

In OPERA and ORATORIO, there was an association between levels of IgG below the LLN and serious infections (OR per 100 patient years: 5.48 95%CI 3.0–9.2), although the overall number of infections remained low (n = 14 in 10/152 patients) (Derfuss et al., 2020). The most common severe infections were urinary tract infection, cellulitis, and pneumonia (Derfuss et al., 2020; Avouac et al., 2021). Amongst individuals treated with ocrelizumab, observational cohorts showed that especially higher levels of disability carry an increased infection risk, in combination with longer treatment duration and severe comorbidities (Oksbjerg et al., 2021). This comes on top of the fact that older age is one of the most consistent risk factor for vulnerability to any kind of infection (Salter et al., 2021; El Chakhtoura et al., 2017). In an NMOSD cohort, the occurrence of severe infections was linked to the presence of urinary tract dysfunction, the dosing intervals and the interaction between IgG level and urinary tract dysfunction (Avouac et al., 2021). In CNS demyelinating diseases, there was a more than 8-fold increased odds for infection in wheelchair-bound patients treated with rituximab (Vollmer et al., 2020). However, it is currently unclear how to predict which low IgG patients will become symptomatic in terms of severe/recurrent infections.

There are no consensus guidelines regarding immunoglobulin replacement therapy (IGRT) for patients with severely depressed IgG levels in the absence of infection (Kado et al., 2016; Srivastava and Wood, 2016; Kado et al., 2017). There is also no consensus on the cut-offs to define hypogammaglobulinemia ranging from 4 to 7 g/L (Kado et al., 2016). Nonetheless, it would be good practice to measure immunoglobulin levels annually and to increase awareness about the possible risk of severe infections (Table 4). Rheumatology experts build the decision to start IGRT on three elements: IgG levels below the LLN, the presence of recurrent/severe infections and failure to respond to the pneumococcal polysaccharide or tetanus toxoid vaccine (Kado et al., 2016; Srivastava and Wood, 2016; Kado et al., 2017). However, the latter seroconversion antibody tests are not routinely available. Only if IgG levels are less than 2.5 g/L in a patient with recurrent or severe infections, it is recommended to start IGRT immediately (Srivastava and Wood, 2016; Kado et al., 2017). In all other cases, there is a tendency to, at least initially, monitor patients with appropriate antibiotic therapy for infectious episodes. If fever develops in an individual on anti-CD20-mAb with low IgG, immediate antibiotic initiation is indicated. Moreover, prescribing prolonged antibiotic courses (e.g., 14 days co-amoxiclav 625 mg TID for an uncomplicated respiratory tract infection) is recommended as prevention of infection reduces end organ damage and mortality (Srivastava and Wood, 2016).

It is important not to assume proper understanding of fever management. In hypo- or asplenic patients, a medic alert bracelet/card incorporating vaccination history and information booklet significantly raised awareness. Pre-intervention, none of the patients reported having a prescribed antibiotic if more than 2 h away from medical care at fever onset while this number increased to 18/28 (64%) post-intervention (O'Neill et al., 2020). Prophylactic antibiotics, such as macrolides (e.g. azithromycin 500 mg 3x/week), might be useful but their use is not evidence-based (Srivastava and Wood, 2016), and is likely to select for resistant organisms. The focus on adequate management of infectious episodes rather than IGRT is in line with the management of patients with a functional splenectomy in whom the immunological deficits are similar to those associated with longstanding rituximab use (Rubin and Schaffner, 2014). The burden of starting IGRT should not be underestimated and the decision cannot be taken lightly. In the setting of secondary hypogammaglobulinemia, IGRT needs to be given intravenously every 3–4 weeks or subcutaneously on a weekly or biweekly basis, depending on insurance coverage and patient preference (Kado et al., 2017). Based on guidelines for IGRT in people with common variable immune deficiency, it is recommended to start with a dose of 400–600 mg/kg body weight every month and adjust dosing based on infections and IgG trough level, or steady state level in subcutaneous administration (Kado et al., 2017; Cunningham-Rundles, 2010). Target IgG levels range from 6 to 9 g/L, although a higher level may be required in some patients to protect them from recurrent infections (Kado et al., 2017; Cunningham-Rundles, 2010).

Early-onset neutropenia after rituximab has been described in individual case reports suggesting a close temporal relationship (Arroyo-Ávila et al., 2015; Adler et al., 2019; Shah et al., 2019). Late-onset neutropenia (LON) following rituximab is also well-recognised. LON is usually defined as absolute neutrophil count below 1.5 × 10⁹/L which occurs more than one and less than 12 months after anti-CD20-mAb (Kado et al., 2016; Salmon et al., 2015). In ORATORIO, the proportion of ocrelizumab patients presenting with early or late decreased neutrophils was higher (12.9%) than in placebo (10.0%). However, only 1% of the patients in the ocrelizumab group had less than 0.5 × 10⁹/L neutrophils (0% placebo) and two were treated with granulocyte colony-stimulating factor (e.g. filgrastim) (EMA, 2021b). In the ofatumumab trial, one neutropenic sepsis was reported (Hauser et al., 2020). Amongst 2624 individuals diagnosed with an autoimmune disease receiving rituximab, LON was observed in 40/2624 (1.5%) when performing 3 or 6 monthly follow-up bloods (Salmon et al., 2015). Neutropenia occurred after a mean period of 5 months after the last infusion, was usually mild (85%> 0.5 × 10⁹/L) and recovered fully in all with or without filgrastim (Salmon et al., 2015). A non-opportunistic serious infection was diagnosed in 5/40 of neutropenic individuals, all with favourable outcome after treatment (Salmon et al., 2015). In other smaller cohorts, the incidence varied between 4.6 and 6.5% which was partially dependent on the frequency of blood monitoring (Tesfa et al., 2011; Abdulkader et al., 2014; Besada et al., 2012; Monaco et al., 2016). This time frame suggests that a considerable number of cases go unnoticed. Therefore, we do not recommend to implement blood monitoring to detect asymptomatic neutropenia after each cycle (Table 4) (Kado et al., 2016; Salmon et al., 2015). When retreatment is offered after LON, a recurrent episode of neutropenia was observed in about 16% and was always mild (Salmon et al., 2015; Monaco et al., 2016). Authors differ on the need for filgrastim which seems to speed up but not alter the resolution of the neutropenia and has been linked with MS flares (Monaco et al., 2016; Openshaw et al., 2000).

In a large cohort of rheumatoid arthritis (RA) patients treated with rituximab, there was no evidence of an increased risk of malignancy of any type over time (Van Vollenhoven et al., 2015). In the two phase-III ocrelizumab trials, 15 patients with malignancies were identified (1.1%) versus 4 cases in the comparator groups (0.38%). From these, 6/15 were breast cancers while no such cases were observed in the placebo or interferon-β groups (Hauser et al., 2017; Montalban et al., 2017). In the ofatumumab trial, there were 5 neoplasms in the active arm (1/5 invasive breast cancer) while there were 4 in the teriflunomide arm (Hauser et al., 2020). In July 2019, a total of 95 cases of breast cancer were reported amongst 76,554 women exposed to ocrelizumab based on post-marketing data (Ng et al., 2020). Although the different breast cancers did not fall into a single pathological category and the incidence was still within the background rate expected for an MS population (Greenfield and Hauser, 2018), an imbalance towards anti-CD20 cannot be excluded. Therefore, it is paramount that female patients follow standard breast cancer screening per local guidelines, especially focusing on self-breast examination. In addition, we would recommend discontinuing anti-CD20 therapy in people presenting with a breast neoplasm (Table 4). In a cross-sectional study, the absence of CD20⁺ cells in benign breast lobules was associated with a higher risk of subsequent breast cancer (OR 5.7 95% CI 1.4–23.1), implying that B cells may play a role in preventing progression from benign to malignant disease (Degnim et al., 2017).

At week 48 of the MS rituximab trial, 24.6% of participants were positive for human anti-drug antibodies (ADAs) (Hauser et al., 2008). In a cross-sectional cohort of 336 patients, there was a significant association between both presence and titres of ADAs and incomplete B-cell depletion, but not with infusion/adverse reactions or clinical outcomes at the group level (Dunn et al., 2018). In the ocrelizumab trials, ADAs developed in 3/825 (0.4%) across three phase-III trials, with neutralizing antibodies developing only in 1 patient (Hauser et al., 2017). Ofatumumab is a fully humanized antibody which does not generate ADAs (Hauser et al., 2020). Therefore, screening for ADAs is not necessary while being treated with ocrelizumab or ofatumumab, unless patients fail treatment in the presence of poor depletion or rapid repopulation of peripheral B cells (Table 4).

Anti-CD20-mAb influence the immune response against vaccines and the safety of non-live vaccines is not established (EMA, 2021b). In rituximab, the most informative results originated from a controlled trial including 103 people with RA comparing antibody responses in a methotrexate and methotrexate-rituximab arm (Bingham et al., 2010). Vaccine responses against tetanus vaccine were preserved amongst both immunosuppressed groups 24 weeks after treatment whereas responses to a neoantigen and pneumococcal vaccine were decreased. These findings were in line with the reduced seroprotection rate at 3–5 weeks following vaccination against influenza (H1N1) virus in rituximab (18.8%) vs. azathioprine (83.3%), IFN-β (87.5%) and healthy controls (100%) in 26 people with NMOSD (Kim et al., 2013). Other studies also showed blunted post-rituximab responses following vaccination with haemophilus influenzae (Nazi et al., 2013) and HBV (Richi et al., 2020) component vaccines. A randomized, open-label study showed important reductions in humoral response to tetanus, pneumococcus and seasonal influenza vaccination in ocrelizumab compared to interferon-β or no therapy (Bar-Or et al., 2020). An Israeli observational cohort measuring humoral response 1 month after the second dose of the BNT162b2-COVID-19 showed 22.7% of patients treated with ocrelizumab developed humoral IgG response irrespective of normal absolute lymphocyte count (Achiron et al., 2021). Although a case has been described in which VZV-antibodies were lost after ocrelizumab (Rapisarda et al., 2021), the majority of the evidence suggests that anti-CD20-mAb do not significantly affect pre-existing humoral immunity at 96 weeks (Ziemssen et al., 2017). Insufficient data are available to understand what happens to antibody titres beyond this interval. However, vaccine efficacy is not exclusively antibody-mediated and also relies on cellular (predominantly T-cell mediated) mechanisms. After influenza vaccination in rituximab cohorts, there were similar CD4⁺ T-cell recall responses (Arad et al., 2011), potent plasmablast responses and normal expansion of influenza-specific memory B cells (Cho et al., 2017). A phase-I/II study of an mRNA vaccine candidate, BNT162b1, produced by Moderna and closely related to their COVID-19 vaccine, demonstrated a robust cellular immune response, as evidenced by skewed T-helper type 1 response, interferon-γ production by CD8⁺ and CD4⁺ T cells, and expansion of these memory T cells (Vishnevetsky et al., 2021). Therefore it is highly recommended to not withhold people from vaccines while being on anti-CD20 and emphasize the benefits of the annual flu vaccination and potentially also repeat SARS-CoV-2 vaccinations (Reyes et al., 2020). To optimize humoral response, we recommend to administer single-jab vaccines two to four weeks before the next anti-CD20-mAb infusion. The first dose of a double-jab vaccine can be given as early as 12 weeks post-infusion/injection and the booster doses 2–4 weeks before the next infusion or 12 weeks after (Table 4). Of note, when the 6-month infusion cycles are maintained, there is most likely no optimal timing to guarantee an antibody response and thus maximal protection against SARS-CoV-2 (Achiron et al., 2021; Lucas et al., 2021; Sormani et al., 2021).

The infection signal associated with IgG hypogammaglobulinemia and reduced humoral vaccine responses while being on treatment with anti-CD20-mAb cast a shadow on their continued use as a maintenance therapy. On the contrary, it is still unclear whether anti-CD20-mAb could also be used as an induction therapy followed by a wait-and-see approach or a maintenance therapy. From the ocrelizumab phase-II extension trial, we learned that the annualized relapse rate remained low during the drug-free 18 months follow-up period (Baker et al., 2020). In this period, there were very few new T1-gadolinium-enhancing or T2-lesions detected (Baker et al., 2020). The phase-I rituximab extension study (2000 mg/cycle) in MS also reported maintained benefit 12 months after the last infusion (Bar-Or et al., 2008). These results suggest that the 6-monthly infusion intervals may not be necessary.

There are three different strategies that could be explored to derisk anti-CD20-mAb for long-term use. First, monitoring the repopulation of specific proinflammatory B cell subsets such as memory B cells could be of value in determining the timing for reinfusion. In a prospective, open-label Italian cohort, reassuring results were obtained when reinfusion was guided by the level of memory B cell repletion (Novi et al., 2020). Moreover, the mean infusion interval was extended to 367 days (Novi et al., 2020). Second, MRI activity and clinical relapses could guide the need for retreatment which is in line with the approach in other immune reconstitution therapies. Third, an induction therapy with anti-CD20-mAb could be followed by a maintenance therapy with a different treatment. Ideally, this DMT would have no impact on immunoglobulin levels and reassuring data on brain volume loss (e.g. dimethyl fumarates or teriflunomide) (Radue et al., 2017; Gold et al., 2020). However, this treatment strategy should be substantiated by clinical trials in which maintenance treatment with anti-CD20 is compared to induction anti-CD20 followed by maintenance of another DMT. However, large number of participants and lengthy follow-up would be needed to show a difference on inflammatory outcomes measures such as annualized relapse rate and lesion burden on MRI. Safety as a primary outcome measure could be more promising and reduce the numbers of patients needed in such trials.