Sustained cell-mediated but not humoral responses in rituximab-treated rheumatic patients after vaccination against SARS-CoV-2

Abstract

Objectives

B-cell depleting monoclonal antibodies are associated with increased COVID-19 severity and impaired immune response to vaccination. We aimed to assess the humoral and cell mediated (CMI) immune response after SARS-CoV-2 vaccination in rituximab (RTX)-treated rheumatic patients.

Methods

Serum and whole blood samples were collected from RTX-treated rheumatic patients 3–6 months after last vaccination against SARS-CoV-2. Serum was tested by ELISA for quantitative detection of anti-spike SARS-CoV-2 IgG. Cell-mediated variant-specific SARS-CoV-2 immunity (CMI) was assessed by interferon-γ release assay Covi-FERON FIA. Patients were interviewed for breakthrough COVID-19 infection (BTI) 3 months post sampling.

Results

Sixty patients were studied after a median (IQR) of 179 (117–221.5) days from last vaccine to sampling. Forty (66.7%) patients had positive Covi-FERON and 23 (38.3%) had detectable anti-spike IgG. Covi-FERON positive patients had lower median RTX cumulative dose [6 (4–10.75) vs 11 (6.75–14.75) grams, (P = 0.019)]. Patients with positive anti-spike IgG had received fewer RTX cycles [2 (2–4) vs 6 (4–8), P = 0.002] and cumulative dose [4 (3–7) vs 10 (6.25–13) grams, P = 0.002] and had shorter time from last vaccination to sampling [140 (76–199) vs 192 (128–230) days, P = 0.047]. Thirty-seven percent were positive only for Covi-FERON and 7% only for anti-spike IgG. Twenty (33.3%) BTI occurred post sampling, exclusively during Omicron variant predominance. The proportion of patients with CMI response against Delta variant was lower in patients who experienced BTI (25% vs 55%, P = 0.03).

Conclusions

Four out of ten RTX-treated vaccinated patients show lasting cell-mediated immune response despite undetectable anti-spike antibodies. Cumulative RTX dose affects both humoral and cell-mediated responses to SARS-CoV-2 vaccines. Cell-mediated immune responses call for attention as a vaccine efficacy marker against SARS-CoV-2.

Introduction

Since the beginning of the COVID-19 pandemic it has been well established that SARS-CoV-2 infection poses an increased risk for severe Coronavirus Disease 2019 (COVID-19) in immunosuppressed individuals with autoimmune rheumatic diseases (ARDs) [1]. Several patients with ARDs are treated with rituximab (RTX), a chimeric anti-CD20 monoclonal antibody that induces lasting B-cell depletion, is associated with increased COVID-19 severity [2, 3] and risk of persistent viral infection, while it also impairs humoral response after vaccination leading to low antibody titers [4]. However, existing evidence suggests that vaccination induces a T-cell response, despite impaired humoral immunity [4, 5].

Cell-mediated immunity (CMI) response is crucial for protection against infection [6] as well as severe outcomes of COVID-19 [7]. Indeed, robust interferon (IFN) and adaptive immunity responses are associated with favorable clinical outcomes in acute infection as SARS-CoV-2 specific T-cell responses are necessary for effective viral clearance [8]. On the other hand, humoral immunity wanes significantly six months post vaccination [9], whereas T-cell immune responses are retained up to ten months after vaccination [10]. These responses are maintained for the Omicron variant, as opposed to very low to absent antibody cross-neutralization [11], which is probably attributed to the fact that SARS-CoV-2 specific T cells are characteristic of a wide breadth of targeted viral epitopes that are invariant between Omicron and previous variants [12]. The duration, however, of CMI against SARS-CoV-2 has not been studied in RTX-treated patients.

In this study we aim to study the long-term CMI and humoral response against SARS-CoV-2 in RTX-treated vaccinated patients with ARDs.

Methods

We conducted a cross-sectional study to examine the humoral and cell-mediated responses to SARS-CoV-2 vaccination in patients with ARDs treated with RTX. The study was conducted from 24 May to 20 July 2022, in two tertiary academic hospitals in Athens, Greece. The recruited cohort was prospectively followed up until October 2022 for post-sampling breakthrough SARS-CoV-2 infections (BTI).

Eligible patients were those with ARD treated with RTX who had received at least two doses of SARS-CoV-2 vaccine. We recruited patients whose last RTX infusion was administered from 1 December 2021 to 28 February 2022. Patients were initially informed via telephone call and, if they agreed to participate, an on-site appointment was scheduled before the upcoming RTX infusion. During this evaluation, data on demographics, comorbidities and known previous SARS-CoV-2 infection were collected. Previously known COVID-19 infection was identified via the National COVID-19 Registry, which is depository for all PCR and laboratory-performed antigen tests, or by a self-reported positive antigen test performed at home. Vaccination data (e.g. dates and type of vaccine administered) were retrieved from the national COVID-19 registry. Subsequently, dates of underlying ARD diagnosis, initiation of RTX and previous RTX infusions, RTX dose per cycle, previous and concurrent antirheumatic medications were retrieved from individual medical records.

Serum samples were tested by commercial ELISA for qualitative detection of IgG against the nucleoprotein (NP) of SARS-CoV-2 (Aeskulisa^® SARS-CoV-2 NP IgG, Aesku Diagnostics, Wendelsheim, Germany) and quantitative detection of IgG against the spike protein (S) of SARS-CoV-2 (Aeskulisa^® SARS-CoV-2 S1 IgG) according to the manufacturer’s instructions. Antibody activities of the samples were considered positive if >12, negative if <8 or equivocal if 8–12 U/ml. As reported by the manufacturer, the specific U/ml units of the Aeskulisa^® SARS-CoV-2 S1 IgG test can be considered equivalent, when multiplied by 2, to the international standard units IU/ml of the first World Health Organization (WHO) International Standard for anti‐SARS‐CoV‐2 Immunoglobulins. We also investigated CMI response by measuring IFN-γ secreted by T cells in response to SARS-CoV-2 antigens by using a SARS-CoV-2 specific interferon gamma release assay (IGRA) kit with enzyme-linked immunosorbent assay (ELISA) (Covi-FERON ELISA, SD Biosensor, Suwon, Republic of Korea) according to the manufacturer’s instructions. In brief, one-mL of whole blood collected in lithium-heparin tubes was added in each one of the six Covi-FERON tubes: Nil tube (negative control), Original SP tube [contained specific Spike Protein (SP) antigen derived from SARS-CoV-2 and 20I/501Y.V3 variants, (Wuhan/UK) – from here on referred as original tube], Variant SP tube [contained specific SP antigen derived from 20H/501.V2 and 20J/501Y.V3 variants (Brazilian/South Africa) – from here on Beta tube], Delta tube (contained specific SP antigen derived from B.1.617.2.), NP Antigen tube [contained specific SARS- CoV-2 Nucleocapsid Protein (NP) antigen], Mitogen tube (positive control). After incubating for 16–24 h at 37°C, plasma was separated by centrifuging the tubes for 15 min at 2200$×$g. Levels of IFN-γ were assessed by ELISA and the results were interpreted as follows: (i) reactive Covi-FERON assay at ≥0.25 IU/ml (Antigen-nil) and/or ≥0.25 and ≥25% of Nil value and Mitogen-Nil value is ≥0.5 IU/ml; (ii) non-reactive Covi-FERON assay at <0.25 IU/ml (Antigen-nil) and/or ≥0.25 IU/ml and <25% of Nil value and Mitogen- Nil value is ≥0.5 IU/ml; and (iii) intermediate Covi-FERON assay at <0.25 IU/ml (Antigen-nil) and/or ≥0.25 IU/ml and <25% of Nil value and Mitogen-Nil value is <0.5 IU/ml.

Patients were once again contacted via telephone between 1 October and 14 October 2022 and were asked if they acquired SARS-CoV-2 infection after the sampling date. For those with BTI, we collected data on outcomes, such as need for hospitalization, duration of hospitalization or death.

The study was approved by the Institutional Ethics Committee of Attikon University Hospital (ID: 487/3–9-2020) and written informed consent was provided by all participants. There was no funding for this study.

Outcomes

Primary outcome was the positivity rate in Covi-FERON, indicating SARS-CoV-2 specific CMI and anti-spike IgG positivity, as well as the titer of the latter. We then tried to identify factors associated with Covi-FERON and anti-spike antibodies positivity. Lastly, patients with BTI in the prospective phase of the study were compared with those who did not get infected in regard to IGRA and antibody positivity.

Statistical analysis

Continuous characteristics are presented as mean [± standard deviation (SD)] or median values [interquartile range (IQR)], based on normality of distribution by the Kolmogorov–Smirnov method. Categorical characteristics are presented as absolute (n) and relative frequencies (%). There were no missing values in our data. Comparisons were based either on the Pearson χ² test in case of the categorical measures, or the independent samples t test (or the K-sample equality-of-median test, if needed) in case of the continuous measures.

Results

Patient characteristics

Sixty patients were included, 43 (71.7%) were women, with a mean age of 61.1 ± 13.8 years and mean disease duration of 6.5 ± 5.1 years. Most frequent indications for RTX treatment were rheumatoid arthritis (RA, 31.7%), systemic lupus erythematosus (SLE, 25%), ANCA-associated vasculitis (AAV, 18.3%) and inflammatory myositis (10%). Arterial hypertension (26.7%), dyslipidaemia (26.7%), diabetes mellitus (18.3%) and cardiovascular disease (15%) were the most common comorbidities, while the mean BMI was 28.3 ± 5.7 kg/m².

All patients have been treated with RTX, with a median of 4 (2.25–6.75) cycles and 7 (4.25–12) grams of cumulative dose prior to sampling. The median time from last RTX administration to sampling was 139 (113–176.5) days. Concurrent antirheumatic medications mostly included hydroxychloroquine (HCQ, 28.3%), methotrexate (MTX, 26.7%) and mycophenolate mofetil (MMF, 11.7%). Thirty-four (56.7%) out of 60 patients were receiving glucocorticoids in a median dose of 5 (3.25–5) mg of prednisone.

All patients had received at least two vaccines against SARS-CoV-2 before sampling, with 13.3% having been vaccinated with two, 66.7% with three and 20% with four vaccines. Regarding the timing of vaccination between the RTX cycles, 56.7% received their last vaccine 1–4 months after and 43.3% 1–2 months before RTX. The median time from the most recent vaccine administration and sampling was 179 (117–221.5) days. Twelve (20%) patients reported a history of COVID-19 before sampling, with three of them (25%) needing hospitalization and all three of them being discharged (Table 1).

Cell mediated (CMI) and humoral immune response

Regarding CMI, 40 (66.7%) patients had a positive, 18 (30%) a negative and 2 (3.3%) an indeterminate Covi-FERON test. The proportion of patients with a positive reaction in the original, Beta and Delta Covi-FERON tube were 63.3%, 50% and 45%, respectively. Twenty-four (40%) patients had a positive reaction in all three tubes, eight (13.3%) in two and seven (11.7%) in only one tube (Table 2). The median IFN-γ concentration was 0.41 (0.18–1.73), 0.25 (0.02–0.83) and 0.16 (0.04–0.85) IU/μL for the original, Beta and Delta tube, respectively (P <0.001). In pairwise comparison, IFN-γ concentrations in the original tube were higher compared with the Beta (P <0.001) and Delta tube (P <0.001), whereas no difference was noted between Beta and Delta tubes (P = 0.21) (Supplementary Fig. S1, available at Rheumatology online).

Detectable humoral immune response, as assessed by the measurement of anti-S IgG, was noted in 23 (38.3%) patients with a median anti-S IgG value of 5.87 (0–82.7) U/ml. Only five (8.3%) patients had detectable anti-N IgG levels. The median anti-S IgG concentrations were higher in patients with positive Covi-FERON compared with those with negative test [9.9 (2.4–89) vs 0 (0–17), P = 0.019]. Eighteen (31.6%) patients were found positive for both Covi-FERON and anti-S IgG, 14 (24.6%) were negative for both assays, whereas 21 (36.8%) were positive only for Covi-FERON and four (7%) only for anti-S IgG (Table 2, Fig. 1).

After excluding the two patients with the indeterminate Covi-FERON results, we compared those with positive results with those with negative results. Patients with positive Covi-FERON had significantly lower median RTX cumulative dose [6 (4–10.75) vs 11 (6.75–14.75) grams, P = 0.019]. We did not find any differences in the disease and patient characteristics, in the mean lymphocyte count at the time of sampling, in the number of administered vaccines, in the mean interval between the last vaccination and sampling or in the timing of last vaccination between RTX cycles. A difference was noted between the last RTX administration and the time of sampling, with Covi-FERON positive patients having a shorter interval compared with Covi-FERON negative patients (Table 3).

Similarly, patients with positive humoral responses had received a lower number of RTX cycles [2 (2–4) vs 6 (4–8), P = 0.002] and cumulative dose [4 (3–7) vs 10 (6.25–13) grams, P = 0.002]. Additionally, the median time from last vaccination to sampling was shorter in those with positive results [140 (76–199) vs 192 (128–230) days, P = 0.047] (Table 4).

Nine (15%) patients also had a positive reaction to NP tube, with a median IFN-γ concentration of 1.5 (0.77–3.06) IU/ml. Among them, all nine had positive Covi-FERON results, whereas only three (33%) had positive anti-S IgG. We performed a subgroup analysis, excluding these patients that had evidence of CMI after natural infection. After excluding the indeterminate results, we found that 31 (63%) and 20 (40%) had positive Covi-FERON and anti-S IgG results, respectively (Supplementary Table S1, available at Rheumatology online). Similar to the analysis of the total cohort, Covi-FERON (+) patients had higher anti-S IgG titers and had received lower cumulative RTX dose (Supplementary Table S2, available at Rheumatology online).

Finally, regarding comorbidities, despite a higher number of diabetics in the non-responders group, no other statistical difference was found between responders and non-responders both for humoral and cellular immune responses (data not shown).

Breakthrough infections (BTI) post sampling

The patients were followed for a mean duration of 145 ± 13 days post sampling. During this follow-up period, 20 (33.3%) BTI occurred, with four (20%) needing hospitalization and all of them surviving. After comparing the cell-mediated and humoral responses between patients that experienced a BTI post sampling vs those that did not, we did not find any difference in the proportion of patients with positive Covi-FERON between the two groups (60% vs 73.7%, P = 0.28), nor in the median anti-S IgG levels [7.8 (0.6–51) vs 5.1 (0–104), P = 0.78]. IFN-γ levels, although higher in the patients that did not experience BTI post-sampling, did not differ significantly in any of the tubes. Interestingly, when comparing the results of the separate Covi-FERON tubes, we did find a statistically significant difference in the proportion of patients with positive delta tube results [BTI (+) vs BTI (–): 25% vs 55%, P = 0.03] (Table 5) and this finding was retained in the sub-analysis of 51 patients with a negative NP tube (17.6% vs 47.1%, P = 0.041).

Discussion

Our study focused on the humoral and CMI response to vaccination in ARD patients treated with the B-cell depleting agent RTX. We observed a reduced humoral post-vaccine response (38.6%) in contrast to a retained long-term CMI response (68.4%). Through the prospective observation of our cohort post sampling and by documenting 20 BTI, our findings could have clinical implications regarding the protective effect of CMI in this high-risk population for severe COVID-19.

Since the advent of the mRNA vaccines, accumulating data have suggested an attenuated humoral immune response in ARD patients treated with rituximab [13]. Even after a third or a booster fourth vaccine dose, RTX-treated patients have significantly lower anti-S IgG and neutralizing antibody levels compared with patients not receiving rituximab [14, 15]. In accordance with these findings, we observed that a post-vaccination humoral immune response was noted in 23/60 (38.3%) patients. Compared with non-responders, patients with a detectable humoral response had received a lower number or rituximab cycles, highlighting the cumulative adverse effect of B-cell depletion. The negative correlation of time elapsed from last vaccination to sampling with the presence of a humoral response is probably consistent with the antibodies’ half-life kinetics [16].

The majority of the published studies have examined the effect of RTX on immunogenicity in the short term of 2–4 weeks post vaccination [17, 18]. In contrast, we aimed at evaluating the long-lasting immune responses, sampling our patients at a median time of 6 months from the most recent vaccine. The timing of last vaccination between rituximab cycles, dichotomized as 0–4 months after the last or 1–2 months before the next infusion, did not have a statistically significant difference in the mounting of a humoral response, although existing data suggest otherwise [19]. According to these, ACR suggests SARS-CoV-2 vaccination to be scheduled at least 5 months after last dose [20], while EULAR recommends postponing next infusion in order to optimize vaccine efficacy [21].

Cellular immunity was assessed via an IGRA with the proportion of patients with a positive reaction in the original, Beta and Delta Covi-FERON tube being 63.3%, 50% and 45%, respectively. This peptide-stimulation assay uses IFN signalling as a surrogate marker for effective T-cell immune responses and its use is well validated in other infections, such as latent tuberculosis [22]. The timing and magnitude of IFN response has been associated with the severity of COVID-19 [23], whereas other pivotal studies have shown that genetic defects in the IFN pathway or autoantibodies against them predispose to severe infection [24, 25].

In contrast to the impaired antibody response, mRNA vaccines can potentially induce CMI in B-cell depleted patients, despite the absence of humoral immunity [4, 26]. In our cohort, CMI was noted in two-thirds of RTX-treated rheumatic patients, with the responders characterized by a lower median RTX cumulative dose. Of interest, we noticed a low percentage of Covi-FERON indeterminate results, similar to the rate observed with the use of similar IGRA assays for latent tuberculosis diagnosis in rheumatic patients [27]. Lymphopenia has been considered a risk factor for indeterminate or negative IGRA results in some studies [28]; however, this finding was not confirmed in others [27, 29]. In our cohort, we did not observe any difference in lymphocyte count between Covi-FERON positive and negative patients.

Reports from multiple sclerosis and rheumatologic patients on anti-CD20 treatment suggested a robust T-cell response post vaccine [3, 30–32]. However, all the previous studies examined the immune responses 2–4 weeks after the vaccine series completion. Our study, by sampling the patients 6 months post vaccination, is the only one, to our knowledge, that shows a sustained cellular response to vaccination.

The presence of neutralizing antibodies has traditionally been considered the best predictor of protection from future infections: however, the concept of ‘cellular sensitization without seroconversion’ has emerged during the COVID-19 pandemic [8]. Accumulating evidence from non-human primates as well as in oncologic patients receiving anti-CD20 agents, suggests that T-cell immunity has a potential protective clinical effect in circumstances of attenuated humoral immunity [33, 34].

In our cohort, 20 BTI were noted in a median 145 days post sampling. Four patients required hospitalization and no deaths occurred. The follow-up period was characterized by the predominance of the highly infectious Omicron variant. Although no differences were noted in the percentage of patients with positive Covi-FERON between the BTI and non-BTI groups, it is intriguing to consider that the overall favorable outcomes could be attributed to the mounting of long-lasting post-vaccination CMI to the majority of our cohort. Comparing cell-mediated responses between patients that experienced a BTI vs those that did not, a statistically significant difference was noted in the percentage of patients with a positive Delta tube result [BTI (+) vs BTI (–): 25% vs 55%, P = 0.03], despite the fact that Delta tube responses in terms of quantitative IFN-γ production were lower than the respective one of the original tube. This finding could lead to the hypothesis that previous cellular sensitization to the delta variant through vaccination or infection could have a protective role against a subsequent Omicron variant infection.

The emergence of many viral variants of concern (VOCs) and the immune escape that characterizes them has limited the ability of post-vaccine humoral immunity to protect from an initial infection. However, COVID-19 vaccines characteristically retain their ability to prevent severe disease irrespective of the circulating VOC at any given time period. The breadth of cellular recognition of viral proteins impedes the ability of VOCs to evade cellular immune control, potentially mirroring the sustained vaccine efficacy for severe disease through the retained T-cell immunity [12, 35].

We believe that our study adds some valuable information to the existing body of knowledge. First, we show that CMI against SARS-CoV-2 can be sustained for a significant duration in the majority of RTX-treated rheumatic patients, even in the absence of humoral response. Moreover, we provide some evidence that adequately vaccinated patients, such as our cohort where almost 90% of them had received at least three vaccine doses, have favorable outcomes after COVID-19. We also suggest that mounting CMI after vaccination could be protective against emerging VOC in high-risk immunocompromised patients. The major limitations of our study include the relatively small number of patients examined and a simplified approach to CMI analysis. In-depth analysis and characterization of CD4+ and CD8+ T-cell memory cells in larger populations could delineate the cellular correlates of protection and further elucidate the potential protective role of SARS-CoV-2 specific T-cell immunity. The small sample size could have precluded the recognition of other factors associated with impaired immune response, such as comorbidities, despite that we did not find significant differences between responders and non-responders with regard to their prevalence [36]. Furthermore, the absence of a control group with unvaccinated or inadequately vaccinated patients makes the estimation of vaccine efficacy and the correlation with humoral and cell-mediated immune response difficult. Moreover, given the milder disease course of Omicron variant, we cannot predict the exact effect of vaccination on the favorable outcomes of our patients who developed COVID-19, whereas the study of CMI at only one timepoint does not provide information on the kinetics of the immune response. We also recognize that using only one assay for assessing CMI could be among the limitations of our study. However, Covi-FERON test has demonstrated high sensitivity and specificity (91.7% and 100%, respectively) in immunocompetent adults infected by SARS-CoV-2 using a cutoff value of 0.24 IU/ml [37], as in this study. In a study of vaccinated healthy individuals, CMI was measured both by Covi-FERON ELISA and T-SPOT Discovery SARS-CoV-2 assay (Oxford Immunotec, Oxfordshire, UK), a standardized enzyme-linked immuno-spot (ELISPOT) assay [38]. Positive results were concordant between the two tests, but T-SPOT assay yielded more invalid results (0–2.1% and 0–16.7% for Covi-FERON and T-SPOT assays, respectively). In contrast, Sim et al. compared CMI measured by two interferon-γ release assays (IGRAs) in vaccinated healthy individuals, Covi-FERON ELISA and QuantiFERON SARS-CoV-2 (QIAGEN, Germantown, MD, USA) and observed weak correlation between them [39]. FluoroSpot-based IGRA, another commercially available assay for CMI measurement, has demonstrated varying sensitivity (66.8–96.1%) and specificity (80.4–85.3%) depending on the study population and the commercially available peptide pool used [40]. Finally, the immunosuppressive state of the patients could have affected the sensitivity of the assay used in our study.

In conclusion, SARS-CoV-2 vaccination of rituximab-treated rheumatic patients led to long-lasting cellular sensitization with impaired seroconversion. The favorable clinical outcomes of this at-risk population underpin the significance of COVID-19 vaccination and call for the attention of vaccine efficacy through the prism of CMI.

Supplementary material

Supplementary material is available at Rheumatology online.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Funding

No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

Disclosure statement: The authors have declared no conflicts of interest.

References