Vaccine-associated uveitis following SARS-CoV-2 vaccination: A CDC-VAERS database analysis

Purpose To assess the risk of vaccine-associated uveitis (VAU) following SARS-CoV-2 vaccination and evaluate uveitis onset interval and clinical presentations in the patients. Design A retrospective study from December 11, 2020, to May 9, 2022, using the Centers for Disease Control and Prevention (CDC) Vaccine Adverse Events Reporting System (VAERS). Participants Patients diagnosed with VAU following administration of BNT162b2 (Pfizer-BioNTech), mRNA-1273 (Moderna), and Ad26.COV2.S (Janssen) vaccine, worldwide. Methods A descriptive analysis of the demographics, clinical history and presentation was performed. We evaluated the correlation between the three vaccines and continuous and categorical variables. A post-hoc analysis was performed between uveitis onset-interval after vaccination, and age, dose, and vaccine type. Finally, a 30-day risk analysis for VAU onset post-vaccination was performed. Main Outcome Measures The estimated global crude reporting rate, observed to expected ratio of VAU in the United States, associated ocular and systemic presentations, and onset duration. Results A total of 1094 cases of VAU were reported from 40 countries with an estimated crude reporting rate (per million doses) of 0.57, 0.44, and 0.35 for BNT162b2, mRNA-1273, and Ad26.COV2.S, respectively. The observed to expected ratio of VAU for the cases reports from the United States was comparable for BNT162b2 (0.023), mRNA-1273 (0.025), and Ad26.COV2.S (0.027). Most cases of VAU were reported in patients who received BNT162b2 (n=853, 77.97%). The mean age of patients with VAU was 46.24±16.93 years, and 68.65% (n=751) were women. Most cases were reported after the first dose (n=452, 41.32%) and within the first week (n=591, 54.02%) of the vaccination. The onset interval for VAU was significantly longer in patients who received mRNA-1273 (21.22± 42.74 days) compared to BNT162b2 (11.42 ± 23.16 days) and rAd26.COV2.S (12.69 ± 16.02 days) vaccines (p<0.0001). The post-hoc analysis revealed a significantly shorter interval of onset for the BNT162b2 compared to the mRNA 1273 vaccine (p<0.0001) and the first dose compared to the second dose (p=0.0021). The 30-day risk analysis showed a significant difference between the three vaccines (p<0.0001). Conclusions The low crude reporting rate and observed-expected ratio suggests a low safety concern for VAU. This study provides insights into possible temporal association between reported VAU events and SARS-CoV-2 vaccines, however further investigations are required to delineate the associated immunological mechanisms.


INTRODUCTION
The global COVID-19 pandemic caused by severe acute respiratory syndrome coronavirus (SARS-CoV-2) led to swift vaccine development and approval. Since the beginning of the pandemic, 336 vaccine candidates have been developed and 32 vaccines are currently authorized for use globally. 1 On December 11, 2020, the first vaccine received emergency use authorization from the United States Food and Drug Administration (FDA) for a large-scale, vaccination program to prevent the spread of SARS-CoV-2 and reduce its severity in infected patients. 2 Amongst the authorized vaccines, BNT162b2 (Pfizer Inc./BioNTech SE, Mainz, Germany) and mRNA-1273 (Moderna Therapeutics Inc., Cambridge, MA) are based on messenger RNA whereas Ad26.COV2.S (Janssen Pharmaceuticals, Beerse, Belgium) uses a recombinant replication-incompetent adenovirus type 26 vector to stimulate an immune response in the recipients. [3][4][5] As all the SARS-CoV-2 vaccines were approved for emergency use authorization, the Centers for Disease Control and Prevention (CDC) expanded the purview of its Vaccine Adverse Event Reporting System (VAERS), a passive surveillance platform that functions as an early warning system for potential vaccine adverse events. 6 Several ophthalmic disorders, including uveitis, were added as the adverse events of interest to the system. Several reports in the literature have highlighted the temporal association between uveitis and universally administered vaccines such as hepatitis B, human papilloma virus (HPV), influenza, Bacille Calmette-Guérin (BCG), measles, mumps, and rubella (MMR), and varicella vaccines. [7][8][9][10][11][12][13] Benage and Fraunfelder identified 289 cases of vaccine-associated uveitis (VAU) published in the literature and reported by the surveillance systems (including VAERS) over 26 years. 14 Although the precise immunopathological mechanisms that cause VAU are yet to be delineated, several hypotheses attribute it to the immune response to vaccine adjuvants, molecular mimicry between vaccine peptide fragments and uveal self-peptides, and delayed hypersensitivity and subsequent immune complex deposition as the potential causes. 15-J o u r n a l P r e -p r o o f 18 As of June 2022, uveitis is one of the most commonly reported ophthalmic adverse events following SARS-CoV2 vaccination, with over 70 published reports and case series. 19 Since the initiation of the most extensive vaccination program, several studies have evaluated the safety concern of inflammatory disorders (such as Guillain-Barré Syndrome, myocarditis, etc.) following SARS-CoV2 vaccination using the VAERS database. [20][21][22][23][24][25][26] For a comprehensive insight into the potential association between VAU and the three FDA emergency use authorized COVID-19 vaccines [BNT162b2, mRNA-1273, and Ad26.COV2.S], we analyzed the largest cohort of VAU cases using the VAERS database.
Herein, we determine the global crude reporting rate and the observed-expected ratio of uveitis since the initiation of the vaccination program. We also report the clinical characteristics in patients diagnosed with VAU and assessed the association between demographics and duration of uveitis onset after vaccination.

DATA SOURCE
This retrospective cohort study was conducted using the CDC-VAERS database (Centers for Disease Control and Prevention, Atlanta, GA). The VAERS is the national early warning system that monitors the safety of vaccines after they are authorized or licensed for use by the FDA. The database is publicly available, de-identified, anonymous data of vaccine related adverse events reported by patients, parents (for minor patients), clinicians, vaccine manufacturers, and regulatory bodies worldwide. The VAERS data are available through the Wide-ranging Online Data for Epidemiologic Research (WONDER) platform, developed and operated by the CDC. 27 The database includes demographic information, date of vaccination and adverse event onset, brief medical and surgical history, current co-morbidities and medications, history of adverse events, and a detailed report of the clinical signs and symptoms and the diagnoses of the adverse events post-vaccination. All the reports submitted to VAERS J o u r n a l P r e -p r o o f that appear to be false or fabricated to mislead CDC and FDA are reviewed before being added to the VAERS database. A false VAERS report violates Federal law (18 U.S. Code § 1001) and is punishable by a fine and imprisonment. The reports are then evaluated by third-party professional coders, who assign appropriate medical terminology using Medical Dictionary for Regulatory Activities (MedDRA) preferred terms based on the unstructured data in the submitted reports. 28 On requesting explicit permission to analyze and publish these data, we were informed that CDC WONDER allows access to the information freely and use, copy, distribute, or publish this information without additional or explicit permission. 29

STUDY POPULATION
The patients diagnosed with VAU who received BNT162b2, mRNA-1273, and Ad26.COV2.S vaccines between December 11, 2020, and May 9, 2022, were included in the study. The VAU cases were reported from 40 countries, and the data from the United States was reported from

STATISTICAL ANALYSIS
The statistical analysis was performed using R Studio (R Foundation for Statistical Computing, Vienna, Austria). The crude reporting rates were estimated using the number of VAU reports (by vaccine type) per million COVID-19 vaccine doses. The 30-day observed to expected ratios for the cases in the United States were calculated using the formula -(personyears x background rate)/100,000, where background rates were measured per 100,000 personyears. The person-years at risk for uveitis within 30 days of vaccination were calculated as the number of persons who received at least one vaccine dose x 30/365.25. The assessment of the observed-expected ratio analysis was limited to reports from the United States due to the lack of accurate global vaccination data and background rates of uveitis which are highly variable in different populations. The background rate for the United States population was referenced from the study reporting the incidence rate of uveitis in the United States by Acharya and colleagues. 30 The total number of vaccinated individuals and the doses administered in the United States during the study period were obtained from the publicly accessible CDC data. 31 A descriptive analysis of the social demographic characteristics and vaccination data was performed. We assessed the association between the onset interval of uveitis and vaccine type, age, sex, and dosage using the one-way ANOVA test. Since the history of COVID-19, uveitis and other inflammatory disorders, and ocular and systemic presentations were categorical variables, a Pearson's Chi-square test of association to evaluate the risk associated with the three vaccines. A post-hoc analysis was performed to evaluate the variability in VAU onset duration in the age groups, dose, and vaccine type. A reverse Kaplan-Meier risk analysis was also performed for the three vaccines. The missing values in the data were indicated, and J o u r n a l P r e -p r o o f the Na.rm code accounted for them during the analysis. The value of p<0.05 was considered statistically significant.  Additionally, the mean onset interval was longest in patients diagnosed with VAU after the second dose (18.89  33.82 days). The analyses evaluating the association of onset interval with vaccine type, sex, age, and dosage are summarized in Table 2.  Table 3.

During
The post-hoc analysis between the different doses and VAU onset showed a significant difference between the onset intervals of the first and second doses (p=0.021). We also found the VAU onset interval was significantly shorter in patients who received the BNT162b2 vaccine compared to mRNA-1273 vaccines. The post-hoc analyses between onset interval and age groups, vaccine type, and dose are detailed in Tables 4a, b, and c. The 30-day reverse Kaplan-Meier risk analysis showed a higher risk of VAU with BNT16b2 compared to mRNA-1273 and Ad26.COV2.S vaccines (p<0.0001). (Figure 1)

DISCUSSION
The initiation of the vaccination program to immunize people against SARS-CoV-2 was a critical step in managing the COVID-19 pandemic, which has impacted every nation worldwide. The three FDA emergency use authorized vaccines -BNT162b2, mRNA-1273, and Ad26.COV2.S, have shown high efficacy against SARS-CoV-2 and significantly reduced incidence of severe disease, hospitalizations, and long-term effects of this respiratory virus. 32-potential systemic adverse effects, including ocular disorders. The population-based studies have reported several adverse events possibly associated with these vaccines, including pericarditis, arrhythmia, deep-vein thrombosis, pulmonary embolism, myocardial infarction, intracranial hemorrhage, and thrombocytopenia; however, the evidence of VAU following SARS-CoV2 vaccination is limited to a few case reports and series. 19,[33][34][35][36] Only one large-scale database study from Israel, including 188 patients with non-infectious uveitis following SARS-CoV2 vaccination, was recently published. 37 Several years ago, Aguirre and colleagues reported the generation of a uveitic reaction in a canine model on injecting adenovirus 1, which was attributed to the type III hypersensitivity response due to generation of antigen-antibody complexes in the aqueous humor. 38 Recent studies have reported the detection of SARS-CoV-2 RNA in the aqueous humor and other ocular tissues of patients infected with the virus, leading to a similar inflammatory response involving immune complex deposition. 39,40 Since BNT162b2 and mRNA-1273 are mRNA delivery vaccine platforms, it can be speculated that viral mRNAinduced immune response may be causing VUA in some patients post-vaccination. On the contrary, Rabinovitch and colleagues attributed VAU caused by mRNA vaccines to type I immune response leading to elevated levels of interferons. 41,42 They suggested that the mRNA delivery through the vaccines leads to the activation of RNA-sensing molecules (TLR3, TLR7, MDA5, and RIG-I), leading to activation of autoimmune processes in these patients. However, it has been reported that modified nucleobase (N1-methylpseudouridine) added to the SARS-CoV-2 vaccines suppresses the vaccine-induced immunostimulatory response. 43 The Ad26.COV2.S vaccine is a replication-incompetent recombinant adenovirus type 26 viral vector which expresses SARS-CoV-2 spike protein. 5 In the past, Cunningham and colleagues have attributed delayed-type hypersensitivity and immune responses observed in VAU to the molecular similarities between uveal self-peptides and vaccine peptides. 18  and were diagnosed with anterior uveitis (90.96%) following vaccination. 37 Interestingly, in the cases reported to VAERS, very few patients with VAU had been previously diagnosed with uveitis (9.7%) or systemic autoimmune diseases (1.2%) and only 44.9% of the cases were diagnosed with anterior uveitis following vaccination.

LIMITATIONS
This study reporting the VAU cases following SARS-CoV-2 vaccination has several limitations. VAERS is a passive surveillance system that records adverse event reports from pharmaceutical companies, physicians, drug regulators, and patients globally. Despite the mandatory requirement to report vaccine-associated adverse events, underreporting and delayed reporting are common. In some cases, the submitted reports are incomplete, and lack uniformity in data reporting, and several reports have missing data points such as ethnicity, which are considered important risk factors associated with uveitis. [44][45][46][47] The VAERS data is