Predictors of an Immunogenic Response to the BNT162b2 mRNA COVID-19 Vaccination in Patients with Autoimmune Inammatory Rheumatic Diseases Treated with Rituximab: A Multicenter Study

Treatment with rituximab (RTX) blunts SARS-CoV-2 vaccination-induced humoral response. We sought to identify predictors of a positive immunogenic response to the BNT162b2 mRNA vaccine in patients with autoimmune inammatory rheumatic diseases (AIIRD) treated with RTX (AIIRD-RTX). We analyzed 108 AIIRD-RTX patients and 122 immunocompetent controls immunized with BNT162b2 mRNA vaccine participating in a multicenter vaccination study. Immunogenicity was dened by positive anti-SARS-CoV-2 S1/S2 IgG measured at 2 to 6 weeks after the second vaccine dose. We used a stepwise backward multiple logistic regression to identify predicting factors for a positive immunogenic response to vaccination and develop a predicting calculator, further validated in an independent cohort of AIIRD-RTX patients (n=48) immunized with the BNT162b2 mRNA vaccine. identify clinical predictors for a seropositive immunogenic vaccination response to BNT162b2 vaccination. A diagnosis of rheumatoid arthritis (as opposed to AAV and IIM), low number of total RTX courses prior to vaccination, high serum total IgG level prior to last RTX course, and extended interval between RTX treatment and vaccination confer a high probability to achieve a seropositive immunogenic response to vaccination. The study provides the clinicians with a practical prediction calculator for assessing the probability of seropositive response to mRNA BNT162b2 vaccination easily applied in daily practice. predicting RTX response and vaccine ecacy proposed prediction prospective cohorts

vaccines trials, uncertainty regarding the response to vaccination in AIIRD patients was raised by the medical community. Reassuringly, several prospective controlled studies proved that most patients with rheumatic diseases could mount an adequate immunogenic response to SARS-CoV-2 vaccination, acknowledging a lower post-vaccination level of anti-SARS-CoV-2 immunoglobulin G (IgG) antibodies in AIIRD patients compared to immunocompetent controls. (6)(7)(8)(9)(10)(11) Among other factors, B-cell depleting therapy signi cantly contributes to a reduced immunogenic response to SARS-CoV-2 vaccination in patients with AIIRD. (8-10,12-18) Indeed, B-cell depleting therapy has been consistently associated with reduced immunogenicity induced by in uenza and pneumococcal vaccines in patients with AIIRD. (19) Rituximab (RTX), a widely used B-cell depleting therapy for rheumatoid arthritis (RA), anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV), and other rheumatic diseases, was identi ed as a risk factor for COVID-19 complications and a severe disease course (20)(21)(22)(23)(24), emphasizing the importance of SARS-CoV-2 vaccination in vulnerable RTX-treated patients. On the other hand, RTX was linked to a reduced immunogenic response to SARS-CoV-2 vaccination, corresponding to a low seroconversion rate ranging between 24% and 49%.(8, 10,11,[14][15][16][17][18]25) Furthermore, B-cell depleting therapy was reported to impair not only humoral but also a cellmediated immune response to SARS-CoV-2 mRNA vaccination (11), although two recent studies detected T-cell-mediated immune response in the majority of RTX-treated patients irrespective of the humoral response. (12,16) However, the extent of anti-COVID-19 protection conferred by T-cell-mediated immune response in RTX treated AIIRD (AIIRD-RTX) patients remains unknown.
In the setting of the COVID-19 pandemic, clinicians commonly face a challenge concerning the optimal timing of vaccination in relation to RTX treatment, despite a general recommendation to delay the B-cell depleting therapy in relation to SARS-CoV-2 vaccination.(5) Identifying routine available predictors of a seropositive immunogenic response to SARS-CoV-2 vaccination in AIIRD-RTX patients may assist in a patient-tailored vaccination approach. Indeed, several studies found a direct correlation between detectable CD19 peripheral B cell counts and an immunogenic response to mRNA SARS-CoV-2 vaccination. (11,15,16) However, routine measurement of CD19 B cells prior to RTX treatment has not been recommended for most AIIRD and, therefore, might be unavailable in daily practice. In a prospective study conducted in a large cohort of AIIRD patients and immunocompetent controls reported by our group, we found that time interval between RTX administration and vaccination had a critical role in predicting an immunogenic response to the BNT162b2 mRNA vaccine .(8) Therefore, we sought to investigate additional predictors associated with an immunogenic response to mRNA BNT162b2 vaccination in AIIRD-RTX patients participating in our ongoing prospective vaccination study. We further developed a calculator based on clinical and laboratory data available in daily clinical practice to predict a seropositive immunogenic response conferred by SARS-CoV-2 vaccination in AIIRD-RTX patients.

Methods
This study is part of the ongoing prospective observational multicenter study investigating immunogenicity and safety of the BNT162b2 mRNA COVID-19 vaccine in adult AIIRD patients, focusing on RTX-treated patients compared to immunocompetent controls. The study was conducted at the Rheumatology Departments of Tel Aviv Sourasky, Carmel, and Hadassah Medical Centers, Israel, between December 2020 and June 2021. The study protocol was described in detail elsewhere.(8) For the prediction model validation, data from an independent cohort of RTX-treated BNT162b2 mRNA vaccinated AIIRD patients from the Rambam Medical Health Care Campus were used.
The study was performed in accordance with the principles of the Declaration of Helsinki and approved by the research ethics committees of the four medical centers: TLV-1055-20, CMC-0238-20, HMO-0025-21, RMB-417-20, respectively. All study participants gave written informed consent on recruitment into the study.

Study aims
The primary end point was to identify independent predictors associated with seropositive immunogenic response to the BNT162b2 (P zer-BioNTech) vaccine in adult AIIRD-RTX patients.
Secondary end points included 1. Immunogenicity of the BNT162b2 mRNA vaccine in adult AIIRD-RTX patients compared with immunocompetent controls.
2. Development of a calculator to predict the probability of a seropositive immunogenic response to the BNT162b2 mRNA vaccination in AIIRD-RTX patients and its validation in an independent cohort of vaccinated AIIRD-RTX patients.

Safety of vaccination.
4. Effect of vaccination on disease activity in AIIRD-RTX patients strati ed by positive and negative immunogenic response to vaccination.

Study population
This study included consecutive AIIRD patients treated with RTX up to 8 years prior to the BNT162b2 vaccination. The AIIRD-RTX group included 86 patients who participated in the vaccination study reported by our group (8) (29), and the EULAR/ACR classi cation criteria for idiopathic in ammatory myopathy (IIM) (30) patients were recruited.
The control group included a sample of the immunocompetent population, with one additional participant added to the original study control group (n=122).
An independent prospective cohort of BNT162b2 mRNA vaccinated AIIRD-RTX patients (n=48) from the Rambam Medical Health Care Campus was used for the validation of the prediction model.
Exclusion criteria for all study groups included pregnancy, history of past vaccination allergy, and previous COVID-19 infection and for controls -history of AIIRD, immunosuppressive treatment, and previous COVID-19 infection.

Data collection
Demographic and clinical characteristics, including AIIRD diagnosis and anti-rheumatic medications, were reported by the participants and con rmed by reviewing the electronic medical records (EMR) by the study investigators (VF, TE, DZ, HP). The dates of the seasonal 2020 in uenza vaccination and BNT162b2 mRNA vaccination were recorded. Medications included conventional synthetic disease modifying antirheumatic drugs (csDMARDs), glucocorticoids (GC), other immunosuppressive medications (mycophenolate mofetil), and intravenous immunoglobulin (IVIg). Doses of methotrexate (MTX) and prednisone were recorded. Speci c details regarding RTX treatment were collected from the EMR, including immunoglobulin G levels (IgG, mg/dL) up to 3 months prior to last RTX course; total number of RTX courses, regardless of indication; the dose of each RTX course and the cumulative RTX dose (mg); the date of last RTX course. Hypogammaglobulinemia was de ned as a total IgG level (prior to last RTX course) of less than 500 mg/dL. The time interval between last RTX course and BNT162b2 vaccine was calculated in days. Data on CD19-positive B cell counts at the time of RTX administration prior the vaccination were unavailable.

Vaccination procedure
All participants were administered the two-dose regimen of the BNT162b2 P zer BioNTech mRNA vaccine according to national guidelines. Each 30-µg dose was given as an intramuscular injection in the deltoid muscle, the second dose given 3 weeks after the rst dose.

Vaccine Immunogenicity
All study participants had a serological test performed 2 to 6 weeks after the second vaccine dose. SARS-CoV-2 S1/S2 IgG antibodies were measured by the food and drug administration (FDA) authorized LIAISON (Diasorin) quantitative assay, with 98% sensitivity and speci city.(31) A cut-off of 15 binding antibody units (BAU) was considered as a positive immunogenic response, according to the manufacturer's instructions. In the validation group, anti-spike-receptor-binding-domain (RBD) antibodies were measured by the SARS-Cov-2 IgG II Quant (Abbott) assay (a chemiluminescent microparticle immunoassay) on the ARCHITECT ci8200system from Abbott. This test was considered positive when titers were above 50 AU/ml, according to the manufacturer's instructions. Whereas an inter-assay validation test between the two assays was out of this study scope, published data suggest a good diagnostic performance and strong correlations with neutralizing antibodies for both. (32,33) Vaccine safety Study participants were questioned (by phone or in-person) regarding adverse events 2 weeks after the 1st vaccine dose and 2-6 weeks after the 2nd vaccine dose.

Assessment of AIIRD activity
Pre-vaccination disease activity was retrieved from the medical records within 3 months before vaccination. Post-vaccination disease activity was clinically assessed 2-6 weeks after the 2nd dose. Disease activity indices used were Clinical Disease Activity Index (CDAI), Simpli ed DAI (SDAI), and DAS-28-CRP for RA, Systemic Lupus Disease Activity Index (SLEDAI) for SLE, and patient and physician global assessment (PGA and PhGA, respectively), using a visual analogue scale (VAS) of 0-10 mm, for vasculitis and IIM.

Statistical Analysis
Differences between categorical variables were tested with the Fisher's exact test. Differences between numeric variables were tested with t-test. A stepwise backward multiple logistic regression for predicting a seropositive response to vaccination was applied to AIIRD patients with all data available (n=104). The AIIRD diagnosis was a dummy variable, meaning that each participant could have only one diagnosis. The model included all individual variables that showed p<0.2 signi cance level between seropositive and seronegative result. The rule for leaving the variable in the model was p < 0.2. Multicollinearity between signi cant variables was assessed by Pearson correlations. The prediction calculator was tested on an independent cohort of 48 BNT162b2 vaccinated AIIRD-RTX patients (validation group), including 21 patients with a positive immunogenic response (responders) to vaccination and 27 patients without a detectable immunogenic response (non-responders) to vaccination. The calculated xed values for the tested population were plotted on a receiver operator characteristic (ROC) curve to select the optimal discriminative cut-off to predict a positive response to vaccination. Sensitivity, speci city, positive predictive, and negative predictive values (PPV and NPV, respectively) were calculated based on this optimal cut-off of the ROC curve.

Study participants
The study included 108 AIIRD-RTX patients and 122 controls, all vaccinated with the two-dose regimen of the BNT162b2 mRNA vaccine. Demographics and clinical data of the AIIRD-RTX population are presented in table 1. AIIRD-RTX patients were signi cantly older than controls, mean ± standard deviation (SD) 61.45±14.96 vs 50.83±14.64 years, p<0.0001. In both patient and control groups, the majority were females, 76.85% (n=83) and 64.75% (n=79), p=0.06, with a high uptake of the seasonal 2020 in uenza vaccination prior to the BNT162b2 mRNA vaccination, 84.11% (n=90) and 81.65% (n=89), p=0.72, respectively. The most common AIIRD diagnosis was RA in 45.37% (n=49), followed by AAV (21.3%, n=23), IIM (16.67%, n=18), SLE (10.19%, n=11), and other vasculitides (5.66%, n=6).  correlation between the IgG levels and a total number of RTX courses, but not strong enough to assume multicollinearity, permitting to consider both variables as independent predictors. Next, we validated the model using data from an independent AIIRD-RTX cohort (n=48) vaccinated with the BNT162b2 mRNA vaccine, including 21 responders and 27 non-responders.  (supplementary table S4 and gure S1) The pattern of disease activity changes was similar in RA patients who mounted a seropositive immunogenic response to vaccination and those who did not. SLE disease activity measured by SLEDAI remained stable for 8 of 9 patients who had available pre-and postvaccination SLEDAI. PGA and PhGA-VAS scores before and after vaccination were overall stable in AAV, IIM, and patients with other vasculitides.

Discussion
The optimal timing of COVID-19 vaccination in AIIRD patients treated with RTX remains debatable.
Herein, we report the analysis of a large group of RTX treated AIIRD patients (n=108) vaccinated with the two-dose BNT162b2 mRNA vaccine regimen, representing a subset of the ongoing multicenter controlled vaccination trial conducted within a nationwide vaccination campaign.
An immunogenic serologic response against the BNT162b2 mRNA vaccine was observed in all immunocompetent controls, yet in only 41.7% of the RTX-treated patients, with signi cantly lower postvaccination anti-spike S1/S2 IgG antibody levels in the latter group. This nding aligns with the previous studies con rming the negative impact of RTX on vaccine-induced immunogenicity. (8-18) The safety pro le of vaccination was similar among patients with positive and negative immunogenic responses to vaccination and consistent with the report of the main study.(8) We further identi ed predictors for a positive immunogenic response to vaccination de ned in a binary mode by the positive versus negative anti-spike S1/S2 antibodies, without considering the antibodies' titer. The predictors included the diagnosis of rheumatoid arthritis as opposed to AAV and IIM, a low number of total RTX courses prior to vaccination, high serum total IgG levels prior to last RTX course, and a longer interval between RTX treatment and vaccination.
To date, a number of studies addressed this topic, reaching a consensus that CD19 reconstitution at the time of vaccination plays a critical role in mounting the immunogenic response to vaccination. (11,12,14-

Conclusion
This is the rst multicenter study based on a diverse AIIRD population treated with RTX to identify clinical predictors for a seropositive immunogenic vaccination response to BNT162b2 vaccination. A diagnosis of rheumatoid arthritis (as opposed to AAV and IIM), low number of total RTX courses prior to vaccination, high serum total IgG level prior to last RTX course, and extended interval between RTX treatment and vaccination confer a high probability to achieve a seropositive immunogenic response to vaccination. The study provides the clinicians with a practical prediction calculator for assessing the probability of seropositive response to mRNA BNT162b2 vaccination easily applied in daily practice.   ROC curve for determining the predicting calculator optimal cut-off for a seropositive immunogenic response to the BNT162b2 mRNA vaccination in rituximab treated AIIRD patients.