Fourth dose of BNT162b2 vaccine for patients with autoimmune rheumatic diseases in a nationwide setting

Abstract

Objective

The effectiveness of COVID-19 vaccinations wanes due to immune evasion by the B.1.1.529 (Omicron) variant and diminished antibody titres over time. We aimed to evaluate the benefit of a fourth vaccination dose in patients with autoimmune rheumatic diseases (ARDs).

Methods

This retrospective analysis included ARD patients aged 18 years or older and members of Clalit Health Services in Israel (which at the time of the study insured 52% of the entire population), and covered the period from 16 January 2022 to 31 March 2022, when the predominant SARS-CoV-2 variant was Omicron. We compared patients without previous COVID-19 infection who had received three doses of the BNT162b2 vaccine (the control group) with those who had received the fourth dose. The primary outcome was COVID-19 infection, which was analysed using multivariate Cox regression in the entire cohort and within ARD subgroups. Secondary outcomes were COVID-19–related hospitalizations and COVID-19–related death.

Results

We included 43 748 ARD patients, of whom 27 766 and 15 982 were in the control and fourth vaccination groups, respectively. COVID-19 infection occurred in 6942 (25.0%) of the control group and 1754 (11.0%) of the fourth dose group (P < 0.001). Patients vaccinated with the fourth dose had a lower risk of COVID-19 infection than the entire cohort [Hazard Ratio (HR) 0.54, 95% CI 0.52, 0.58] and throughout every subgroup regardless of the baseline characteristic or medical treatment, except for rituximab. A similar association was observed for risk of COVID-19–related hospitalization (HR 0.36, 95% CI 0.22, 0.61) and of COVID-19–related death (HR 0.41, 95% CI 0.24, 0.71).

Conclusion

A fourth BNT162b2 vaccination of ARD patients was associated with favourable outcomes compared with three doses among patients with no history of COVID-19 infection.

Introduction

The ongoing COVID-19 pandemic has had a devastating effect on almost every aspect of life. Therefore, developing a vaccine against the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is considered one of the outstanding achievements of modern science [1]. Currently, there are six SARS-CoV-2 vaccines commercially approved globally, and all are highly effective in preventing symptomatic COVID-19 infection, complications, and mortality in the general population [2, 3]. However, the emergence of new SARS-CoV-2 variant strains and the waning immunogenicity of second and third doses has raised concerns regarding the long-term effectiveness of the SARS-CoV-2 vaccine [4, 5].

COVID-19 patients with autoimmune rheumatic diseases (ARDs) showed a higher risk of having poorer outcomes (COVID-19–related hospitalization and COVID-19–related death) than the general population [6, 7]. The high incidence of comorbidities, the intrinsic alterations of the immune system function, and the concomitant use of ARD patients’ immunosuppressant treatments might influence their prognosis when coping with COVID-19 infection [8–10].

Immune evasion of the B.1.617.2 (Delta) and the Omicron variants decreased the effectiveness of the SARS-CoV-2 vaccines [11]. However, a recent study found that a fourth dose of the BNT162b2 (Pfizer–BioNTech) vaccine effectively reduced new SARS-CoV-2 infections, related hospitalizations, and death among persons 60 years or older [12]. Therefore, the current study aims to evaluate the effect of a fourth dose of the BNT162b2 vaccine among patients with ARDs.

Methods

Study population

We extracted data from Clalit Health Services (CHS), the largest Health Provider Organization in Israel, operating mandatory health insurance and insuring 52% of the entire population of Israel [12]. The database comprises real-time input from outpatient clinics, hospitals, laboratories, pharmacies, and vaccination systems. In addition, electronic health-care records are accessible at the patient level. The CHS database features high validity and reliability, as described elsewhere [13, 14].

We retrospectively included all CHS members diagnosed with ARDs who were vaccinated with a fourth COVID-19 vaccine during the Omicron outbreak in Israel and who had no previous diagnosis of COVID-19. We identified ARD patients based on the International Classification of Diseases 9 (ICD-9) criteria for RA, SpA (including PsA, and AS), SLE, and SSc (Supplementary Table S1, available at Rheumatology online). The Institutional Review Board (IRB) of the Emek Medical Center and the CHS Research Room Committee approved the study and exempted the requirement of collecting an informed consent form at enrolment (IRB approval number EMC-0220–20). Public representatives of patients and public services were part of those committees.

Data extraction

We included all CHS ARD patients 18 years or older who were eligible to receive the fourth Pfizer BioNTech (BNT162b2) COVID-19 vaccine in Israel, i.e. those vaccinated with a third dose of BNT162b2 at least 4 months earlier. The BNT162b2 is the only vaccine available for the mass vaccination campaign in Israel. The study period was from 16 January 2022 to 31 March 2022, when the predominant SARS-CoV-2 variant in Israel was Omicron [13]. On 26 December 2021, the fourth dose of the BNT162b2 mRNA COVID-19 vaccine became available for the population 60 years and older and/or immune-compromised patients (including all ARD conditions) above 18 years and/or health-care workers. Patients with a COVID-19 diagnosis before the study period (confirmed either by a PCR test or an antigen test or who had a documented diagnosis recorded in the electronic medical record) were excluded from the analysis.

We extracted information on baseline demographic, medical and pharmaceutical data, and vaccination status. Vaccination status was defined as (1) patients vaccinated with only two doses and a booster (later referred to as ‘control patients’), or (2) patients vaccinated with the fourth vaccine a week after receiving the fourth dose (later referred to as “fourth dose”). In addition, we extracted information on purchases of glucocorticoids, conventional synthetic, biologic, and targeted synthetic DMARDs made during the 3 months prior to the study period (except for 6 months of rituximab), based on the CHS prescription system.

The primary outcome was a diagnosis of COVID-19 during the study period. The diagnosis of COVID-19 infection was based on at least one of the following: a documented SARS-CoV-2 positive PCR test, a documented SARS-CoV-2 positive antigen test, or a documented record of COVID-19 infection from the CHS medical diagnosis database. Other study outcomes were COVID-19–related hospitalization and COVID-19–related death. COVID-19–related hospitalization was defined as any hospitalization at the COVID-19 unit or any hospitalization 14 days after the diagnosis of COVID-19. Similarly, COVID-19–related death was defined as death at the COVID-19 unit or any death within 30 days following the COVID-19 diagnosis.

All data were extracted from CHS using the Clalit Research Data-sharing platform powered by MDClone (https://www.mdclone.com).

Statistical analysis

Descriptive statistics are expressed as number and percentage, median and interquartile range (IQR), or mean and standard deviation [mean (s.d.)]. The study outcomes were analysed using multivariable Cox regression, with the hazard ratio (HR) being a point estimate of any association between an exposure and an outcome. During the Omicron outbreak, we identified 1693 new COVID-19 cases among ARD patients during the first 3 weeks after the fourth vaccination became available in Israel. Since the recently infected subjects would not receive the fourth vaccine, the subjects receiving this vaccination were likelier to stay longer without the virus. To minimize this version of immortal time bias, we allowed a window of 3 weeks after the fourth vaccination became available. The index date for the beginning of the analysis was 16 January 2022. Patients infected with COVID-19 during the first 3 weeks were excluded from the analysis (Supplementary Table S2, available at Rheumatology online). The final analysis included only patients infected with COVID-19 for the first time after 16 January 2022. Exposure to vaccination was defined as receiving the vaccine at any time during the entire study period. We also evaluated factors associated with the probability of receiving the fourth dose by applying a multivariate logistic model (Supplementary Table S3, available at Rheumatology online).

The final model was selected mainly based on the clinical relevance of the covariates, their statistical significance, and the discriminatory ability of the model. The final model was adjusted for age, sex, BMI, smoking status, the Charlson comorbidity index, hypertension, and diabetes in the background. In addition, the model was adjusted for propensity score (PS) to address the imbalance between the fourth dose and the control group. We calculated the probability of receiving the fourth dose using a multivariable logistic regression model. The PS model was adjusted for sex, age, the length of ARD diagnosis, smoking, the Charlson comorbidity index, dementia, malignancy, and current treatment with prednisone (Supplementary Fig. S1, available at Rheumatology online). In addition, we plotted predicted survival curves based on a multivariable Cox regression model for each outcome stratified by the vaccination status. We analysed the primary outcome using the same method in selected subgroups, stratified by patients’ baseline demographic and medical and therapeutic characteristics. In every iteration, we selected only patients within the tested subgroup (e.g. only RA patients), and then we applied the Cox multivariate models as detailed above. Finally, the third dose was given in Israel between July and November 2021. Therefore, the exposure time between those who received the booster early and those who were recently vaccinated may have differed (Supplementary Fig. S2, available at Rheumatology online). To address this issue, we conducted a sensitivity analysis, including 1:1 matching for age, sex, and time interval from the third vaccine to the study index (16 January 2022), using the same multivariate models mentioned above. We analysed the data using SPSS version 26.0, IBM Corp. We considered P < 0.05 as statistically significant in our analysis.

Results

Patients’ characteristics

We identified 45 441 ARD patients during the study period and excluded 1693 patients who had been infected with COVID-19 during the first 3 weeks. The final cohort included 43 748 ARD patients, and 15 982 (36.5%) received the fourth dose of the vaccine (Table 1). Patients in the fourth dose group had different characteristics. They were older (71.5 vs 56.5 years, P < 0.001), had a higher rate of RA (58.1% vs 53.4%, P < 0.001) and scleroderma (8.1% vs 6.5%, P < 0.001), and had a longer median duration of ARD (9.0 vs 7.0 years, P < 0.001). They also had a higher median Charlson comorbidity index (5.0 vs 3.0, P < 0.001). In addition, the fourth dose group showed a higher usage of low-to-medium dosage of prednisone, and lower use of csDMARDs, AZA, TNF alpha-blockers, and IL-17 inhibitors (P < 0.01 for all). The confirmed SARS-CoV-2 infection rate was higher in the control group (25.0% vs 11.0%, P < 0.001), yet COVID-19–related hospitalizations (0.2% vs 0.2%, P = 0.1) and deaths (0.2% vs 0.1%, P = 0.51) were similar.

Characteristics of the cohort stratified by SARS-CoV-2 (at the end of the follow-up) are shown in Supplementary Table S4, available at Rheumatology online. The SARS-CoV-2–positive group was younger (56.0 vs 63.5 years, P < 0.001), with higher rates of non-smokers (68.8% vs 64.6%, P < 0.001), SpA (28.4% vs 24.9%, P < 0.001) and SLE (14.7% vs 11.6%, P < 0.001). They also had a lower median Charlson comorbidity index (3.0 vs 4.0, P < 0.001).

Study outcomes

Predicted survival curves of the study outcomes are shown in Fig. 1. Patients who were vaccinated with the fourth dose had a lower risk of COVID-19 infection (HR 0.54, 95% CI 0.52, 0.58), COVID-19–related hospitalization (HR 0.36, 95% CI 0.22, 0.61) and COVID-19–related death (HR 0.41, 95% CI 0.24, 0.71). The multivariate Cox regression COVID-19 infection risk stratified by subgroups is shown in Fig. 2. Patients vaccinated with the fourth dose had a lower risk of COVID-19 infection than the entire cohort (HR 0.54, 95% CI 0.52, 0.58) and throughout every subgroup, regardless of the baseline characteristic or medical treatment, except for rituximab treatment. In addition, our sensitivity analysis (Supplementary Fig. S3, available at Rheumatology online) showed that receiving the fourth dose was associated with a lower risk of COVID-19 infection (HR 0.76, 95% CI 0.71, 0.81), COVID-19–related hospitalization (HR 0.42, 95% CI 0.24, 0.74) and COVID-19–related death (HR 0.43, 95% CI 0.25, 0.76).

Discussion

We found that the fourth dose of the BNT162b2 vaccine was associated with a lower COVID-19 infection rate among ARD patients. In addition, patients who were vaccinated with the fourth dose had a lower risk of COVID-19–related hospitalization and COVID-19–related mortality. Furthermore, our results remained consistent regardless of the type of background ARD diagnosis, medical background, or treatment.

Various studies have found that COVID-19 vaccinations are effective and safe for most ARD patients [14, 15]. However, concerns have been raised regarding the immunogenicity of COVID-19 vaccination in the ARD population. Nevertheless, the majority of ARD patients will have a seroconversion response, though in lower titres compared with healthy individuals [9, 16]. On the other hand, the protection of COVID-19 vaccines waned over several months after the second and even the third dose, leading to a higher risk of COVID-19–related hospitalization over time [5, 17]. The waning efficacy is explained by the lagging of the approved vaccines behind the relevant SARS-CoV-2 predominant variant and the expected decrease in antibody and neutralization titres over time [18]. Moreover, the intrinsic changes in immune function and immunosuppressant utilization expose the ARD population to a further decline in COVID-19 vaccination effectiveness [19, 20].

In light of these concerns, the Israeli Ministry of Health began to offer a fourth dose of the COVID-19 vaccine in early January 2022 [21]. Magen et al. evaluated the efficacy of the fourth dose nationwide using the CHS database in Israel [12]. Their results indicated that, compared with a standard three doses, the fourth vaccine lowers the risk of COVID-19 infection, hospitalization and death by 45%, 62% and 74%, respectively. In another Israeli study, fourth vaccine efficacy was evaluated among the population aged 60 years or older using the Israeli Ministry of Health database [22]. The risk of severe COVID-19 infection after 1 month was lower than in the three-dose group by a factor of 3.5. Yet, another study carried out among health-care workers found that a fourth vaccination restored the antibody titres (after the third dose) but was unable to prevent COVID-19 infection [23].

Previous studies examining the efficacy of the fourth vaccine for immunocompromised patients in general and within the ARD population are scarce. A study conducted among 92 kidney transplant recipients who showed a weak serologic response after three doses found that the fourth vaccine produced a satisfactory serologic response in 50% of the subjects [24]. A similar study included 164 ARD patients with a poor serologic response. The main finding was that seroconversion occurred in 95.1% of the patients after the fourth dose, compared with only 66.4% after the third dose [25].

This study extends recent findings on possible association between a fourth COVID-19 vaccination and a favourable outcome for ARD patients. In this study, we describe for the first time the benefits of the fourth dose in lowering the risk of clinical outcomes, e.g. COVID-19 infection, COVID-19-related hospitalization, and COVID-19-related mortality among ARD patients. A strength of our study was that we used a nationwide continuous real-time database, reflecting the real-life setting. For instance, similar to findings of previous studies, approximately one-third of the patients in our cohort used glucocorticoids, and another third used csDMARDs [26, 27]. The finding that the fourth vaccine was associated with better COVID-19 outcomes, regardless of the immunosuppressant therapy, is reassuring. These results are supported by the studies mentioned above, which have shown a high seroconversion rate. Moreover, the BNT162b2 was approved to use a year before the Omicron outbreak. Yet, we assume that some of the efficiency of the fourth dose in our cohort may be due to high antibody titres, but these were not evaluated in this study.

This study has several limitations, which provide opportunities for further research. First, causality between the fourth dose and beneficial COVID-19 outcomes cannot be assumed due to the retrospective nature of this study. In addition, including ARD patients with no history of COVID-19 infection in this analysis may limit the external validity of our results. Second, although our analysis was adjusted for baseline epidemiological, clinical, and therapeutical characteristics, additional factors that drive patients’ behaviour (e.g. risk aversion, health system trust, and compliance with preventive measures) were unavailable. To reduce the bias of these variables, we applied a PS-adjusted analysis for the primary outcome, but we acknowledge that other factors than the receiving of the fourth vaccine may have influenced the outcomes. For example, previous studies reported a positive association between seroconversion after vaccination and protection against COVID-19 infection. However, antibody titre information was not collected in the database used for this analysis. Third, the COVID-19–related hospitalization rate in our cohort was relatively low (0.2%). Nevertheless, our results are aligned with those in previous reports from Europe and North America. Fourth, our cohort of ARD patients was assembled using ICD-9 codes. Unfortunately, CHS policy does not allow researchers to validate the data extracted by ICD-9 codes by comparing it with data extracted from identified records. Hence, we could not assess activity level or remission in our cohort. On the other hand, >80% of our patients utilized at least one glucocorticoid or DMARD regimen, validating the active ARD status of the study population. Additionally, we expect this potential bias to be non-differential with respect to the study groups. It is important to note that the Pfizer BNT162b2 vaccine is the only COVID-19 vaccine available in Israel. Hence, our findings are best generalized in health systems using the same type of vaccine. Finally, we extracted data from prescription regimens, which are only a surrogate for actual treatment use. However, our results for our cohort’s immunosuppressant utilization were similar to those of past studies (see above), and we believe they approximately reflect actual utilization.

Despite those limitations, this is the first study that has evaluated the efficacy of a fourth COVID-19 vaccine for ARD patients in a wide-scale setting. We found that the fourth dose may have an added benefit among this vulnerable population by reducing COVID-19 infections, hospitalizations and deaths. Therefore, we hope ARD patients will consider adopting this measure to reduce the COVID-19 burden.

Supplementary material

Supplementary material is available at Rheumatology online.

Data availability

Owing to the Clalit Health Services data privacy regulations, the raw data for this study cannot be shared.

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.

Acknowledgements

We would like to acknowledge the assistance of Mrs Snait Ayalon, MBA, from Ha’Emek Medical Center Research Center, and the technical support of the Clalit Health Services Research Room Team.

References