Ocular Complications Following COVID-19 Vaccination – Coincidence, Correlation, or Causation?

As of September 2022, the World Health Organization (WHO) had collected over 600 million confirmed cases of coronavirus disease 2019 (COVID-19) that resulted collectively in roughly 6.5 million deaths worldwide.¹ More than two-thirds of the world’s population, or about 5.4 billion people, have received at least one COVID-19 immunization since the introduction of the first mRNA-based vaccines in December 2020. Remarkably, the first effective COVID-19 vaccines were available for widespread use less than 9 months after severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection was declared to be pandemic in March of that same year.² Such unprecedentedly rapid development and delivery of what are now more than two dozen SARS-CoV-2-targeted vaccines has reduced COVID-19-related morbidity and mortality dramatically³ – an amazing achievement for biomedical research, the pharmaceutical industry, and international regulatory authorities.^4–6

While the benefits of commercially available vaccines are clear, targeted activation of the immune system to help prevent severe infectious diseases and their complications is not without risks.⁷ Mild, yet manageable, side-effects of vaccination are well-recognized, including localized injection site reactions and constitutional signs and symptoms such as headache, fatigue, fever, chills, arthralgias, myalgias, and rashes. Ocular complications, including uveitis,^8–16 can occur within days to weeks of immunization as well, and such events have been reported following COVID-19 vaccination with varying levels of evidence.^16–23 Whether such descriptions represent coincidence or chance, correlation—which is best thought of as a mathematical or statistical association yet to be demonstrated to be causal, or true causation remains unclear in the vast majority of cases.²⁴ Many have proposed, and we support, the use of the Naranjo approach for assessing suspected associations between the administration of a given therapeutic agent or vaccine and suspected adverse reactions, wherein the greatest weight is given to elimination of alternate causes or risk factors for each complication or adverse event, close temporal association, and, most notably, recurrence on rechallenge (Table 1).^8,25 This issue of Ocular Immunology and Inflammation (OII) contains 20 additional reports of ocular complications following administration of one or more COVID-19 vaccine doses.^26–45

Table 1. Naranjo scoresheet for assessing suspected associations between the administration of therapeutics and adverse reactions.

QUESTION	YES	NO	UNKNOWN	SCORE
1. Are there previous DEFINITIVE reports of this association?	+1	0	0
2. Did the complication or adverse event occur in close temporal association to administration of the agent?	+2	−1	0
3. Did the complication or adverse event resolve or improve when the agent was discontinued or a reversal agent was given?	+1	0	0
4. Did the complication or adverse event recur when the agent was re-administered?	+2	−1	0
5. Were alternative causes for the complication or adverse event effectively eliminated?	−1	+2	0
6. Did the reaction reappear when a placebo was administered?	−1	+1	0
7. Was the agent known to be present at the site of the complication or adverse event at pharmacologically relevant concentrations?	+1	0	0
8. Did the severity of the complication or adverse event increase or decrease in parallel with changes in the dose of the agent?	+1	0	0
9. Did administration of agents of a similar class or mechanism or action produce similar complications or adverse events?	+1	0	0
10. Was the complication or adverse event objectively confirmed and/or measured?	+1	0	0

Scores range from 0 to 13. A score of 0 doubtful; 1 to 4 possible; 5 to 8 probable; 9 to 13 definite. Examples of drugs definitely associated with uveitis include systemic cidofovir, rifabutin, bisphosphonates, and sulfonamides; topical metpranolol, brimonidine, and prostaglandin analogues; intraocular cidovofir and anti-VEGF agents; and the Bacille Calmette-Guerin (BCG) vaccine. Modified after London et al.⁸ and Naranjo et al.²³

Ferrand et al.²⁶ retrospectively reviewed 25 subjects who developed uveitis following COVID-19 vaccination at one of four tertiary referral centers in Germany (n = 14), Italy (n = 8), or India (n = 3) between February and July 2021. Most patients (n = 19; 76.0%) received two immunizations, and among the total 44 delivered doses 30 were Pfizer/BioNTech (BNT162b2 mRNA), five were Moderna (mRNA-1273), seven were AstraZeneca (ChAdOx1 nCoV-19), and two were Bharat Biotech (Covaxin; BBV-152). Patient age ranged from 20 to 72 years (mean 43.2 years), 19 of the 25 subjects (76.0%) were female, and the time from vaccination to onset of uveitis ranged from 6 hours to 27 days (mean 10.7 days; median 12 days). Nineteen patients (76.0%) had a prior history of uveitis. Presentations included anterior uveitis in 14 (56.0%; four of which were bilateral, three of which were due to recurrent, unilateral herpes simplex-associated keratouveitis, and two of which had unilateral granulomatous features), intermediate uveitis in three (12%; one bilateral), posterior uveitis in four (16.0%; all bilateral), and panuveitis in one (4%; recurrence of prior Vogt-Koyanagi-Harada disease). While the likelihood of causality was graded using the aforementioned Naranjo algorithm,²³ wherein a grade of 1 to 4 reflects a possible association, 5–8 a probable, and 9 or greater is definite (Table 1), the authors did not indicate specifically which components of the algorithm lead them to categorize 16 cases (64.0%) as possible and 9 (36.0%) as probable. Uveitis occurred after the first immunization in 21 patients (84.0%), after the second in two (8.0%) and after both in two (8%). These last two patients, each of whom received serial AstraZeneca vaccines, provide perhaps the strongest evidence for causality. Both had a history of anterior uveitis and developed similarly localized inflammation following immunization. Of note, however, none of the 17 other subjects who received two immunizations and who developed uveitis following the first dose were noted to develop uveitis after a second immunization. All patients responded well to traditional therapy, leading the authors to conclude that the benefits of COVID-19 vaccination far outweigh the risks of either recurrent or de novo uveitis.

Nanji and Fraunfelder²⁷ described two patients reported to the National Registry for Drug-Induced Ocular Side Effects who were found to have evidence of anterior uveitis 3 weeks and 3 days, respectively, after their first dose of the Moderna vaccine. The first patient, who had a history of bilateral anterior uveitis following hepatitis B vaccination, was found to have new posterior synechiae formation in her left eye at the three-week visit, but showed no signs of keratic precipitates or anterior chamber cell or flare and so received no treatment. The second patient, who was found to have mild unilateral anterior uveitis with fibrin and posterior synechiae formation 3 days after immunization, responded promptly to local therapy. The authors identified an additional 59 cases of anterior uveitis in the WHO global database collected from over 140 member countries, including the United States. This cohort ranged in age from 24 to 88 years, 80% were female, and the time from vaccination to onset of symptoms ranged from 0 to 31 days, with a median of 3 days. Data was not available to accurately grade the likelihood of each association using the Naranjo approach.

Rallis et al.²⁸ retrospectively reviewed 10 patients who experienced reactivation of herpetic eye disease within 28 days of COVID-19 vaccination. Five patients each received the AstraZeneca and Pfizer/BioNTech vaccines. All were seen at tertiary referral centers in the United Kingdom with active herpetic eye disease (HED) following the first of two COVID-19 immunizations. Two patients who had discontinued their anti-viral agents showed recurrence of HED following their second dose. Of the eight patients who showed no recurrence following their second dose, six (75%) were still on an anti-viral agent. The mean interval from vaccination to herpetic reactivation/activation was 12.3 ± 10.3 days. Subjects who received the AstraZeneca vaccine tended to present earlier than those who received the Pfizer/BioNTech vaccine (4.6 ± 1.5 days vs 20.0 ± 9.4 days; nominal p-value = 0.007). The series included four subjects who had herpes simplex-associated keratitis—one of which was bilateral, and six with varicella zoster-associated keratitis – five of whom had concurrent herpes zoster ophthalmicus (HZO). Five of the 11 affected eyes (63.6%) had anterior uveitis. All affected eyes were treated with topical and systemic anti-viral agents and topical corticosteroids, with a mean time to resolution of 3.9 ± 1.6 weeks. All five patients who developed VZV-associated keratitis in the setting of HZO received the Pfizer/BioNTech vaccine, but also tended to be older (median age 68 years, range 59–95 years vs median age of 47.5 years, range 44–59 years). The authors cited previous studies describing the occurrence of HED following vaccination, including immunization against VZV, smallpox, rubella, and influenza. Walter et al.⁴⁶ described recurrent varicella zoster activation following each of two serial hepatitis A vaccines. The authors suggested use of prophylactic anti-viral therapy in patients with a history of HED – particularly if reactivation/activation followed the first of two or more planned immunizations. Separate reports by Sangoram et al.²⁹ and Singh et al.³⁰ in this issue of OII described the occurrence of PCR-positive HSV-1-associated uveitis and retinitis, respectively, following immunization with ChAdOx1 nCoV-19 from the Serum Institute of India (CoviShield^TM). Others have reported reactivation of toxoplasmic retinochoroiditis following COVID-19 vaccination.^47–49

Patel et al.³¹ described four patients seen at a retina referral center in Dallas, Texas, who developed posterior segment manifestations, including uveitis, following immunization with the Moderna (n = 1), Johnson & Johnson (JNJ-78436735; n = 1) and Pfizer/BioNTech (n = 2) vaccines between March and August of 2021. Onset occurred at 1 day, 2 days, 2 weeks, and 1 month, respectively. In addition to active vitreous inflammation in each patient, two patients had macular retinal pigment epithelial changes, and one each had peripheral yellow choroidal lesions, optic disc edema, and severe anterior chamber inflammation with hypopyon formation. All findings resolved following treatment with systemic corticosteroids, supporting their inflammatory etiology. The authors stated that their patients each had a Naranjo grade of “probable,” although the actual grade and which factors contributed were not provided. Other posterior segment findings following COVID-19 vaccination described in this issue of OII included VKH disease,^32–35 Multiple Evanescent White Dot Syndrome,^36,37 acute retinal necrosis,³⁸ acute macular neuroretinopathy (AMN),^39,40 acute foveolitis,⁴¹ central retinal vein occlusion,⁴² and non-arteritic anterior ishemic optic neuropathy with macular star formation.⁴³ Additional reports of transient visual field loss⁴⁴ and non-necrotizing anterior scleritis⁴⁵ are also included.

Together, these findings support and expand the growing number of ocular complications reported following COVID-19 vaccination. The extent to which such findings were truly related to immunization remains unknown, however, since the simple, and most common, assumption of causation based upon the occurrence of ocular inflammation following a given vaccine dose is inadequate to establish a causal relationship between immunization and the complication or adverse event. We would encourge those who present or report suspected drug or vaccine-associated findings to rigorously apply and report the Naranjo criteria on a case by case basis, and for those who undertake population-based studies to systematically explore the most important of the Naranjo criteria - namely whether alternate explanations were effectively eliminated, time from vaccination to onset of the complications or adverse events, and whether or not the same or a similar finding occurred following planned, subsequent vaccine doses. The latter criterion of demonstrated recurrence with rechallenge is particularly important in establishing causation. Of note, most reported COVID-19-associated ocular complications have been both mild to moderate and readily reversible following treatment with regional and/or systemic corticosteroids. We would, therefore, further encourage clinicians to consider completing planned COVID-19 dosing schedules whenever prudent and possible – perhaps with temporary anti-microbial coverage for those who experienced re-activation of HED or toxoplasmic retinochoroiditis after prior immunization.