Vaccine response following anti-CD20 therapy: a systematic review and meta-analysis of 905 patients

Immunocompromised patients including those with hematologic malignancy are at increased risk of infections, including those that are vaccine preventable.^1-3 COVID-19 is a viral infection with high mortality in immunocompromised patients.^4,5 In healthy vaccine recipients, severe COVID-19 can be prevented; however, patients with immunocompromising conditions and/or those on immunosuppression were excluded from initial vaccine trials.^6-9 With the advent of the COVID-19 vaccination rollout, clinicians are faced with questions regarding the effectiveness of vaccination in immunocompromised patients, particularly in those receiving anti-CD20 therapy.

Although anti-CD20 therapy is known to efficiently deplete peripheral B cells, these constitute only 2% of the total B-cell population.¹⁰ The effect of anti-CD20 treatment on B cells located in peripheral lymphoid tissues is less clear.^11-16 Additionally, anti-CD20 therapy has less of an effect on memory B cells¹⁷ and long-lived plasma cells,¹⁸ which do not express the CD20 antigen. Furthermore, the peripheral blood B-cell compartment begins to recover ?6 to 9 months after therapy.¹⁰ There is limited and conflicting evidence on the ability of anti-CD20–treated patients to mount a vaccine response to both primary and recall vaccinations. Previous reports reviewing vaccine response following anti-CD20 therapy have been limited by their scope.^19-21 

As the COVID-19 pandemic continues and vaccinations begin, to help inform counseling, public health policies, and future research on vaccination in this vulnerable patient population, reviewing the literature regarding vaccine responsiveness of patients receiving anti-CD20 therapy is critical. The objective of this study was to perform a systematic review of the literature on vaccine responsiveness in patients receiving anti-CD20 therapy.

This study is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Definitions used in this article are in accordance with the World Health Organization (WHO) Report on Biological Standardization.²² Immunological correlates of protection (ICP) have been established for a number of different vaccines. When an ICP has been established, the seroprotection (SP) rate is the percentage of subjects who achieve an antibody level at or above the ICP. The seroconversion (SC) rate is a predefined increase in serum antibody concentration or titer. In subjects without a detectable or quantifiable antibody (below the lower limit of detection or quantification) prior to vaccination, SC is usually defined as achieving a quantifiable antibody level postvaccination. In subjects with a quantifiable antibody prior to vaccination (due to previous exposure or vaccination), SC is commonly defined by a predefined fold increase from pre- to postvaccination.

SP and SC have been standardized for influenza vaccines.²³ For other vaccines (eg, tetanus, diphtheria, pertussis, pneumovax conjugate and pneumovax polysaccharide, Haemophilus influenzae B, hepatitis B, hepatitis A, polio, varicella zoster), SP and SC definitions are not uniformly standardized. In our review, if an article used response criteria that did not meet the WHO definition for either SP or SC as indicated in this section, then the vaccination response rate was described as seroresponse (SR).

All studies published on response to clinically available vaccinations in patients treated with anti-CD20 therapy were considered for inclusion, including noncancer and cancer populations. Only studies reporting SP, SC, and SR rates, and/or measurement of cellular (T-cell) immune response, were included. Case reports, conference proceedings, and studies with <10 patients who had received anti-CD20 therapy were excluded. Only English language reports were included. Full inclusion criteria are available in supplemental Table 1.

PubMed and EMBASE databases were searched up to the week of 4 January 2021.

The full search strategy is available in supplemental Table 2. Two authors (A.V. and I.G.) independently conducted the search strategy, and results were compared with ensure concordance. Differences in classification were discussed and resolved, with a third author (L.K.H.) available for resolution of disagreements. Titles and abstracts of articles were reviewed and any that were clearly irrelevant were excluded. Full texts of remaining articles were reviewed to find studies that met the inclusion criteria. Additionally, systematic reviews discovered in the search that focused on vaccine response in patients receiving immunosuppressive therapy were screened to identify additional references.

A data extraction form was used to extract relevant information from the articles. Information was extracted on study design; geographic location; type of immunogenicity studies conducted (humoral vs cellular); the anti-CD20 drug used by patients; patient population; sample size; median/mean age; distribution by sex; type of vaccine administered; how response was defined; SP, SC, and SR rates; cellular response data if available; and safety data for patients with hematologic malignancies. Patients were considered to be on active treatment if the average interval between anti-CD20 therapy and vaccination was <12 weeks; otherwise, treatment durations were divided into average interval 3 to 6 months, 6 to 12 months, and >12 months. Data on controls (healthy and disease controls not receiving anti-CD20 therapy) in the study and their outcomes were also extracted.

To assess the quality of cohort studies and substudies of trials with a comparison group, the Newcastle-Ottawa assessment tool was used.^24,25 For 1-arm cohort studies, the Joanna Briggs Institute Critical Appraisal Checklist for Cohort Studies was used.²⁶ For randomized studies, the Cochrane Risk of Bias Scale was used.²⁷ See supplemental Table 3 for details on each tool. Assessment was conducted by 2 authors (A.V. and I.G.), while a third author (L.K.H.) was available for differences of opinion.

The primary outcomes were the SP, SC, and/or SR rates to each vaccine. As the pandemic influenza vaccination has standardized definitions for SP and SC,²³ this enabled meta-analysis of point estimates of SC to vaccination in patients treated with anti-CD20 therapy, as well as relative benefit ratios (RBs) comparing SC in anti-CD20–treated patients to healthy or disease controls. Secondary outcomes were cellular response to vaccinations, and safety data on vaccinations in patients with hematologic malignancies. Due to limitations in data, subgroup and sensitivity analyses were not performed.

The majority of the data synthesis was descriptive. For meta-analysis, the principal summary measures used were pooled estimates and RBs with 95% confidence intervals (CIs). Heterogeneity between estimates was assessed using the I² statistic, and interpreted per the Cochrane Handbook recommendations: I² 0% to 40%, heterogeneity likely not substantial; 30% to 60%, moderate heterogeneity; 50% to 90%, substantial heterogeneity; and 75% to 100%, considerable heterogeneity.²⁸ Pooled estimates were transformed using the Freeman-Tukey double arcsine method,²⁹ and the final pooled results were back-transformed with a 95% CI for ease of interpretation. Meta-analysis was performed using a random-effects model (DerSimonian and Laird) using the MetaXL (www.epigear.com) add-in for Microsoft Excel, as well as Review Manager 5.4 (Cochrane Collaboration, 2020).

Figure 1 shows the flow diagram for study selection. A total of 38 peer-reviewed studies (31 cohort, 3 phase 3, 2 phase 3 substudies, 1 phase 1, 1 phase 1/2 substudy) comprising 905 anti-CD20–treated patients from North America, Europe, and Asia were included. Nineteen studies assessed vaccine responses in patients receiving anti-CD20 for hematologic malignancy; 19 studies assessed vaccine responses in patients receiving anti-CD20 for autoimmune diseases such as rheumatoid arthritis or multiple sclerosis.

Tables 1 and 2 list the summary characteristics of the included studies. Supplemental Table 4 lists study definitions of vaccine response assessment.

Supplemental Table 5 lists the results of the risk-of-bias assessment for individual studies. Twenty-eight studies had a low risk of bias.

Table 3 provides a high-level summary of all of the reports of humoral immune responses in patients receiving rituximab. Details of individual studies are available in Tables 1 and 2. As outlined in both tables, SP, SC, and/or SR rates were lower in anti-CD20–treated patients than in healthy controls or disease controls, regardless of the time that had elapsed from treatment. In patients on active anti-CD20 therapy, response to vaccination was poor regardless of the vaccine studied, with SC rates of 0% to 25%.^30-37 Response rates appeared to improve as more time elapsed after anti-CD20 therapy, with the best responses in patients >12 months from therapy, though this did not always translate to an equivalent response to controls.

Based on the descriptive nature of the data, ascertaining whether patients had better responses to T-cell–dependent vaccinations than T-cell–independent vaccinations is difficult.

Thirteen studies (222 anti-CD20 treated patients and 993 control patients) reported on response rates to the pandemic influenza vaccine. Ten of the 13 studies included patients with hematologic malignancies (rather than autoimmune disease) treated with anti-CD20 therapy. The pooled estimates of SC for patients treated with anti-CD20 therapy are listed in Table 4. Patients on active (<3 months since last dose) anti-CD20 therapy had poor response to vaccination, with a pooled estimate of SC after 1 vaccine dose of 3% (95% CI, 0% to 9%; 6 studies, 75 patients). This improved incrementally over time with pooled estimates of 21% (95% CI, 0% to 65%), 50% (95% CI, 4% to 96%), and 41% (95% CI, 19% to 65%) within 3 to 6 months, 6 to 12 months, or >12 months from last dose of anti-CD20 therapy, respectively, though the sample size was limited (see Table 4).

The pooled RBs for SC of patients treated with anti-CD20 therapy compared with healthy controls and compared with disease controls are listed in Table 5 and visualized in Figures 2 and 3. Patients on active anti-CD20 therapy had significantly lower SC rates after 1 dose of the pandemic influenza vaccine compared with healthy controls (RB, 0.05 [95% CI, 0-0.73]) and compared with disease controls (RB, 0.22 [95% CI, 0.09-0.56]). The results did not change significantly even when limiting only to studies of patients with hematologic malignancy (Table 5).

Six studies reporting cellular response data on 171 patients treated with anti-CD20 therapy were included. Descriptive data are available in Table 2. Notably, all but 1 study³⁸ included patients who were receiving concomitant immunosuppressive therapy that may have impaired T-cell function (eg, mycophenolate, cyclosporine). Three of the included studies found no significant difference in T-cell vaccine response between anti-CD20–treated patients vs healthy or disease controls.^39–41 In these studies, the average interval between anti-CD20 and vaccination ranged from being on active treatment to 19 months. The other 3 studies found that patients treated with anti-CD20 had lower T-cell responses compared with healthy or disease controls, or to patients with normal B-cell subsets.^38,42,43 The average interval between anti-CD20 therapy and vaccination in these studies was 6 to 9 months.

Of the 19 studies reporting data on patients with hematologic malignancies, 9 reported data on adverse events related to vaccination.^{30,31,33,37,40,41,44–47} The majority of studies found that adverse reactions were mild, and 1 study noted that there were no differences in side effect profiles in patients compared with healthy controls.⁴⁰ Only 1 study reported serious adverse events (SAEs): a phase 1 study of a novel inactivated varicella zoster vaccine schedule reported 10 SAEs related to vaccination among the 80 patients studied,⁴¹ with the most common SAEs being febrile neutropenia and pneumonia.

There are over 900 000 people living with lymphoma in the United States alone,⁴⁸ and anti-CD20 therapies have been a component of first-line therapy for many lymphoproliferative disorders for over 20 years. Furthermore, anti-CD20 therapies are an important treatment modality for numerous autoimmune conditions, including multiple sclerosis and rheumatoid arthritis, where their use is widespread. Understanding vaccine responsiveness in the context of B-cell–depleting therapies, especially given the number of patients that will need vaccination against COVID-19, is imperative.

To interpret vaccine responses (SP, SC, and/or SR rates), it is important to note that there are 2 types of vaccine antigens. A primary antigen is an antigen that the patient has not been previously exposed to, which elicits an extrafollicular response that causes B-cell proliferation and differentiation into plasma cells, and generally reaches a peak value 4 weeks after immunization.⁴⁹ A recall antigen is one in which an anamnestic immune response is anticipated (eg, tetanus booster vaccination), in as little as 7 days.⁴⁹ For recall antigens, SP and SC rates postvaccination are expected to be high. From that perspective, we chose to focus our meta-analysis on the pandemic influenza vaccine from the 2009 H1N1 pandemic, as this vaccine was a primary antigen, analogous to the COVID-19 vaccine. As patients would have no preexisting memory B cells to this antigen, this allows the best assessment of B-cell–depleted patients’ immune response to vaccination. Additionally, the pandemic influenza vaccine has standardized definitions of SP and SC. The SC rate was chosen as the primary outcome to meta-analyze as SC demonstrates the ability to mount a response to vaccination (by either a several-fold increase in antibody titers, or conversion from negative titers to protective titers). SP rates can be adequate in patients who have had previous exposure to an antigen, even if they do not respond to the vaccination given.

Although the literature was heterogenous in terms of patient population, vaccine types, and interval since anti-CD20 therapy, we documented 4 main findings: (1) vaccination appears safe in patients on anti-CD20 therapy; (2) the response to vaccination in patients on active anti-CD20 therapy is low and approaches 0%; (3) anti-CD20 therapy lowers patients’ vaccine response beyond the impact of their disease or other treatments, given the higher response to vaccination in disease controls; and (4) response to vaccination improves incrementally over time but may not reach the level of healthy controls even 12 months after therapy. Although not the primary aim of our study, we also noted that patients with lymphoproliferative disorders not exposed to anti-CD20 therapy (on active cancer therapy, previous non-anti-CD20 treatment, or never treated) had reasonable responses to vaccination (eg, SC rates 27% to 80% to pandemic influenza vaccine, see Tables 1 and 2).

It is notable that SC rates in studies of patients on active anti-CD20 therapy were 0% to 25%,^30–37 regardless of the vaccine type assessed. A small number of studies reported that vaccine responses did not vary over time.^39,50,51 However, these studies may have been underpowered to detect a time trend, as our meta-analysis for pandemic influenza vaccine found that SC rates did improve as time between last anti-CD20 treatment and vaccination increased.

Whether 2 vaccine doses enhance immunogenicity in patients receiving anti-CD20 therapy remains unclear. When assessing the response to pandemic influenza vaccination, the point estimate of SC in patients on active therapy was just 3% (95% CI, 0% to 9%) after 1 dose, and 12% (95% CI, 2% to 27%) after 2 doses. Five studies assessed SC after 1 and 2 doses of pandemic influenza vaccine in patients receiving anti-CD20 vs disease controls. As illustrated in Tables 4 and 5, the point estimate of SC does not appear appreciably different in those receiving 1 or 2 doses of vaccine, nor do patients on anti-CD20 seem to improve in their SC rates compared with disease controls. This finding is unchanged when limiting the analysis only to patients with hematologic malignancies. However, due to the small sample sizes of included studies, the relative impact or lack of impact of booster vaccinations remains essentially unknown.

Our third finding was that the reduction in response to vaccination seen in patients on anti-CD20 therapy is likely above and beyond the immunocompromising effects of the disease alone and the other treatments used in the disease. The SC rates to the pandemic influenza vaccination in patients on active anti-CD20 therapy or even 3 to 6 months from therapy were significantly lower than in disease controls (RBs of 0.22 [95% CI, 0.09-0.56] and 0.44 [95% CI, 0.23-0.84], respectively). However, by an average of 6 to 12 months from anti-CD20 therapy, the difference between groups was less pronounced, and by >12 months the difference seemed trivial, though we are limited by the number of studies available for analysis.

Given the finding that anti-CD20 therapy abrogates vaccine responses, one strategy may be to delay anti-CD20 therapy (when possible) until after vaccination. However, currently we do not know how long of a gap is required between vaccination and anti-CD20 treatment of vaccination to be effective. One small study of patients with relapsed low-grade lymphoma³⁶ assessed this strategy: 5 patients were given a primary antigen vaccine (hepatitis A) 2 weeks prior to starting on rituximab treatment. None of patients developed a humoral response. On the other hand, 5 of 10 patients immunized 2 weeks before rituximab had a response to a recall antigen (polio virus or tetanus toxoid). These results highlight the need for further research to determine the optimal gap between vaccination, particularly of a novel primary vaccine, and anti-CD20 therapy. In the absence of data, clinicians and patients need to weigh the expected benefits of initiating anti-CD20 therapy against its anticipated impact on vaccine response. In rare instances, it may be appropriate to delay anti-CD20 therapy to allow for COVID-19 vaccination.

Finally, the response to vaccination does seem to improve incrementally over time as the B-cell compartment recovers following rituximab treatment. Generally, the effect of anti-CD20 therapy was least pronounced >1 year after treatment, although even these patients had impaired responses compared with healthy controls, with SC rates of 33% to 100% for pandemic infuenza,^40,44 3% to 43% for seasonal influenza,^44,52 9% to 75% for tetanus, diphtheria, and pertussis,^46,53 85% for haemophilus influenzae type b,⁴⁶ and 56% for hepatitis B.⁵⁴ The finding was also noted in correlative analyses in a study by de Lavallade,⁴⁰ which found that the interval between therapy and vaccination was significantly longer in patients who were seroprotected compared with those who were not (4.7 vs 17.5 months; P = .001). Whether the degree of B-cell recovery is associated with response remains unclear, as this is not generally measured in clinical practice, and thus was not a focus of this review.

Notably, we should be cautious applying immunological correlates of protection (which have only been validated in young healthy adults⁵⁵) to immunocompromised patients.⁵⁶ Whether an incremental rise in titers not meeting criteria for SP or SC might still be valuable in immunocompromised patients to guard against clinical infection remains unclear. Additionally, cellular responses are also important in vaccination-mediated protection,⁵⁷ and although these are also affected by anti-CD20 therapy, the degree to which cellular immunity is impacted is less clear.³⁵ Studies were small and contradictory in their findings with respect to whether T-cell immunity was preserved or diminished following anti-CD20 therapy. Some of the studies were also confounded by the inclusion of patients who were on concomitant immunosuppressive therapies that impair T-cell function (eg, mycophenolate).

There are several limitations to our work. First, the pool of studies included in this systematic review was heterogenous with respect to patient population, vaccine studied, definitions of SP and SC, and interval between anti-CD20 treatment and vaccination. Although precise estimates of interval between anti-CD20 and vaccination were available for some studies, others only provided median/mean intervals and ranges. Definitions of SP and SC varied between studies, which was expected as there are no widely used definitions for SP and SC other than for pandemic influenza vaccinations. We sought to limit the heterogeneity encountered as much as possible by categorizing studies both by vaccine studied and by average interval between anti-CD20 and vaccination. We also only meta-analyzed data related to the pandemic influenza vaccine, which has standard definitions for SP and SC. Nonetheless, heterogeneity in interval from anti-CD20 therapy does remain, and the most accurate interpretation of amalgamated data only applies to patients on active treatment as only these studies had homogenous populations. Additionally, the sample sizes of each study were small, so direct extrapolation of SP, SC, and/or SR from any 1 study should be avoided. Finally, we did not have data on baseline hypogammaglobulinemia for all studies, and whether concomitant immunoglobulin replacement therapy was used for patients, as this can provide an exogenous source of SP detected in laboratory studies. We also did not collect data on other concomitant immunomodulatory medications used for patients and the timing of their use with respect to vaccination, which may also have impacted their immune responses.

This systematic review highlights the available literature on both humoral and cellular vaccination response in patients treated with anti-CD20 therapy. We found limited and heterogenous data, suggesting the need for a robust prospective study in a large sample of patients. The COVID-19 vaccination rollout represents an opportunity to study primary vaccine responses in immunocompromised patients, including those receiving anti-CD20 therapy.

We report that SC after vaccination in patients on active anti-CD20 therapy is low and approaches 0%. The low response rate of patients on anti-CD20 therapy appears to be largely attributable to the anti-CD20 therapy rather than the disease or other treatments, and response to vaccination seems to improve incrementally over time but may not reach the level of healthy controls even 12 months after therapy. During the current pandemic, this argues for the need for rapid, real-world studies to guide clinical practice. We need to quickly determine the SC rate following COVID-19 vaccination, whether it can be improved upon with booster shots, and/or influenced via the timing of vaccination relative to anti-CD20 therapy. Additionally, these data serve as a reminder to continue stringent evidence-based infection-prevention strategies such as infection control measures, physical distancing, and appropriate shielding for patients on anti-CD20 therapy.⁵⁸ Additionally, in an effort to decrease the household risk of exposure, advocating for vaccination of the caregivers and household contacts of patients receiving anti-CD20 therapy may be appropriate. Finally, while we await future prospective data addressing this issue, we do recommend COVID-19 vaccination for patients receiving anti-CD20 therapy as even a muted response to vaccination is likely better than no response.

The authors thank the authors and patients of all studies included in this review.

Contribution: A.V. performed literature search, article selection, analysis, and manuscript writing; I.G. performed article selection and manuscript review and revision; S.D.B. and M.C. reviewed and revised the manuscript; and L.K.H. conceived of the study, assisted with analysis, and reviewed and revised the manuscript.

Conflict-of-interest disclosure: L.K.H. was one of the principal investigators on a study partially funded by Gilead Sciences. The remaining authors declare no competing financial interests.

ORCID profiles: A.V., 0000-0001-5369-1501; I.G., 0000-0001-9306-2945.

Correspondence: Lisa K. Hicks, St. Michael’s Hospital, Room 2-084, Donnelly Wing, 30 Bond St, Toronto, ON M5B 1W8, Canada: e-mail lisak.hicks@unityhealth.to.