Drug-Related Keratitis: A Real-World FDA Adverse Event Reporting System Database Study

Abstract

Purpose: This study aimed to assess the drug risk of drug-related keratitis and track the epidemiological characteristics of drug-related keratitis.

Methods: This study analyzed data from the U.S. Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) database from January 2004 to December 2023. A disproportionality analysis was conducted to assess drug-related keratitis with positive signals, and drugs were classified and assessed with regard to their drug-induced timing and risk of drug-related keratitis.

Results: A total of 1606 drugs were reported to pose a risk of drug-related keratitis in the FAERS database, and, after disproportionality analysis and screening, 17 drugs were found to significantly increase the risk of drug-related keratitis. Among them, seven were ophthalmic medications, including dorzolamide (reporting odds ratio [ROR] = 3695.82), travoprost (ROR = 2287.27), and brimonidine (ROR = 2118.52), and 10 were non-ophthalmic medications, including tralokinumab (ROR = 2609.12), trazodone (ROR = 2377.07), and belantamab mafodotin (ROR = 680.28). The top three drugs having the highest risk of drug-related keratitis were dorzolamide (Bayesian confidence propagation neural network [BCPNN] = 11.71), trazodone (BCPNN = 11.11), and tralokinumab (BCPNN = 11.08). The drug-induced times for non-ophthalmic medications were significantly shorter than those for ophthalmic medications (mean days, 141.02 vs. 321.96, respectively; P < 0.001). The incidence of drug-related keratitis reached its peak in 2023.

Conclusions: Prevention of drug-related keratitis is more important than treatment. Identifying the specific risks and timing of drug-induced keratitis can support the development of preventive measures.

Translational Relevance: Identifying the specific drugs related to medication-related keratitis is of significant importance for drug vigilance in the occurrence of drug-related keratitis.

Introduction

Keratitis is an inflammation of the corneal tissue caused by the weakening of the defense mechanisms of the cornea and the presence of exogenous or endogenous pathogenic factors.¹ Drug-related keratitis, as an important subset of these exogenous factors, can be triggered either by the direct damaging effects of the drug on the cornea or by the drug reducing the immune barrier function of the cornea, making it more susceptible to infections and foreign bodies and subsequently inducing keratitis. It is often observed in cases where the primary ocular diagnosis is unclear and there is prolonged, frequent, or combined use of ophthalmic formulations or systemic medications. It may also occur during the treatment of primary ocular diseases or after ocular surgery. However, systematic research on drug-related keratitis is still lacking. Some studies focus on drug-related corneal lesions due to common toxic drugs found in ophthalmic formulations including antibiotics, antiviral drugs, surface anesthetics, non-steroidal anti-inflammatory drugs, antiglaucoma drugs, and preservatives.^2–5 

Common risk factors for keratitis include wearing contact lenses or having ocular surface diseases, previous ocular trauma, or ocular surgery. Some common pathogens associated with keratitis include Pseudomonas aeruginosa, Staphylococcus aureus, coagulase-negative staphylococci, and Streptococcus pneumoniae.⁶ However, it is also essential to recognize the continuous occurrence of drug-related keratitis in clinical practice. Some topical or systemic medications significantly increase the risk of drug-related keratitis.⁷ It is imperative to continually raise public awareness of the potential adverse reactions of topical ophthalmic or systemic medications, thereby enabling the adoption of more personalized medication treatment plans for different individuals. 

Our study, based on the U.S. Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) database, provides, to the best of our knowledge, the first data support for drug-related keratitis. It evaluates the iatrogenic risk of drug-related keratitis associated with topical ophthalmic or systemic medications using large-scale real-world data. This study aimed to offer supportive data for enhancing drug vigilance and providing guidance for personalized medication in clinical practice. Additionally, it may serve as supplementary information regarding potential adverse reactions of drug-related keratitis not yet mentioned in some drug labels. 

Methods

Data Source

This study collected adverse event data from the FAERS database spanning from January 1, 2004, to December 31, 2023 (https://fis.fda.gov/extensions/FPD-QDE-FAERS/FPD-QDE-FAERS.html). Further analysis evaluated all reported cases of keratitis adverse reactions among subjects and subsequently screened the drugs used by these subjects to provide data support for the epidemiological characteristics and potential drug distribution of drug-related keratitis. The database is primarily comprised of seven datasets: patient demographic and administrative information (DEMO); drug and biologic information (DRUG); adverse events (REAC); patient outcomes (OUTC); report sources (RPSR); drug therapy start and end dates (THER); and drug use and diagnosis indications (INDI). Over the specified time frame, the database contained a total of 20,629,811 raw entries. After removing duplicate data based on the primary ID number, 17,379,609 entries remained. Among these reported entries, there were 8716 reports of keratitis-related adverse events involving 7176 subjects who experienced keratitis adverse reactions during drug use associated with 1606 drugs. Additionally, we cross-referenced and replaced drug names with both generic and brand names using the DrugBank (https://go.drugbank.com/) database,⁸ excluding drugs with fewer than three reported cases and those with identical generic names but different brand names. Ultimately, 497 drugs were retained. The data cleaning process is illustrated in Figure 1. The datasets presented in this study can be found in online repositories. The names of the repositories and accession numbers can be found at the FAERS Public Dashboard (https://www.fda.gov/drugs/questions-and-answers-fdas-adverse-event-reporting-system-faers/fda-adverse-event-reporting-system-faers-public-dashboard). 

Figure 1.

 

Data cleaning process for drug-related keratitis in the FAERS database.

Identification of ADRs

The definition of adverse drug reaction events analyzed in this study is derived from the Medical Dictionary for Regulatory Activities (MedDRA, version 20.0; http://www.meddra.org/).⁹ Adverse events were encoded using MedDRA preferred terms (PTs), and standardized MedDRA queries were employed in this study to identify PTs related to keratitis. In this research, we utilized PTs with a narrow scope.¹⁰ The PTs for keratitis included in our study are “keratitis,” “ulcerative keratitis,” “punctate keratitis,” “corneal ulcer,” “keratitis interstitial,” “vernal keratoconjunctivitis,” “atopic keratoconjunctivitis,” “photokeratitis,” “exposure keratitis,” “allergic keratitis,” “superior limbic keratoconjunctivitis,” and “keratitis sclerosing.” 

Statistical Analysis

Signal detection utilized the reporting odds ratio (ROR),¹¹ proportional reported ratio (PRR),¹² Bayesian confidence propagation neural network (BCPNN),¹³ and Multi-item Gamma Poisson Shrinker (MGPS) of the disproportionality method.¹⁴ These four methods are based on mining potential positive signals through the comparison of target events and target drugs with all other events and drugs using a fourfold table calculation method (Tables 1 and 2). The criteria for positive signals are as follows: (1) for ROR, the standard is a ≥ 3 and 95% confidence interval (CI) of the ROR (ROR_{95% CI lower limit}) > 1; (2) for PRR, the standard is a ≥ 3 and 95% CI of the PRR (PRR_{95% CI lower limit}) > 1; (3) for BCPNN, the standard is a ≥ 3 and the lower limit of the 95% CI of the information component (IC₀₂₅)>0; and (4) for MGPS, the standard is a ≥ 3 and empirical Bayesian geometric mean lower 95% CI for the posterior distribution (EBGM₀₅) ≥ 2.^15,16 The “a” represents the number of target adverse events that occur with target drugs; other details can be found in Table 1. In our study, the drugs selected as positive signals had to meet the criteria of the above four methods, indicating a potential correlation between drugs and events. Further assessment of the drug-related keratitis risk was conducted using the BCPNN value for each drug. Additionally, we categorized drugs into ophthalmic and non-ophthalmic medications and evaluated the drug-induced time using cumulative risk curves.¹⁷ Statistical analysis was performed using SPSS Statistics 26.0 (IBM, Chicago, IL), Prism 10.1.2 (GraphPad, Boston, MA), Excel 2019 (Microsoft, Redmond, WA), and R 4.2.2 (R Foundation for Statistical Computing), with a significance level set at P < 0.05. In the R data analysis process, we utilized major packages including ggplot2 3.4.4, ggrepel 0.9.4, dplyr 1.1.4, and DescTools 0.99.52. 

Table 1.

 

Four-Grid Table of Disproportionality Analysis Method: A Contingency Table for the Proportion Imbalance Analysis

 

Table 2.

 

Principle of Disproportionality Analysis and Standard of Signal Detection

 

Table 3.

 

Baseline Data for Drug-Related Keratitis Patients Reported in the FAERS Database

 

Results

Baseline Subject Information

A total of 7176 subjects were reported to have experienced drug-related keratitis adverse reactions in the FAERS database from 2004 to 2023. The age of the subjects was primarily centered around 56.2 ± 20.3 years, with females comprising the majority at 52.0%. The age distribution of females experiencing drug-related keratitis was concentrated in the 55- to 70-year age range, whereas males were mainly concentrated in the 60- to 75-year age range (Fig. 2A). Furthermore, we observed a gradual increase in the number of reported cases of drug-related keratitis over the years, peaking in 2023, with the incidence being significantly higher in females than males each year (Fig. 2B). The primary outcomes for these subjects were concentrated in “other serious (important medical event)” (67.7%) and “hospitalization (initial or prolonged)” (19%) (Fig. 2C). The predominant routes of drug administration were ophthalmic (22.7%) and intraocular (20.5%) (Fig. 2D). The majority of reports originated from the United States (40.9%) and France (8.8%) (Figs. 2E, 2F). More details can be seen in Table 3. 

Figure 2.

 

Distribution of demographic data for drug-related keratitis. (A) Population pyramid of subjects with drug-related keratitis categorized by gender and age. (B) Bar chart showing the annual reporting counts of drug-related keratitis by gender. (C) Distribution of outcomes among subjects. (D) Distribution of modes of drug administration among subjects. (E, F) Bar chart of reporting countries and a heatmap of reporting countries, respectively.

Distribution of Drugs Causing Drug-Related Keratitis

We screened out 76 drugs from a pool of 497 drugs that simultaneously met four positive signal screening criteria based on disproportionality analysis. We further verified the brand and generic names of these drugs. Considering that some drugs are intended to treat keratitis-related diseases but may still exhibit positive signals due to their inadequate efficacy, we excluded such drugs from our analysis. Ultimately, we identified 17 drugs associated with drug-related keratitis, of which seven are predominantly used in ophthalmology and 10 in non-ophthalmology. The top three drugs in ophthalmology were dorzolamide (ROR = 3695.82), travoprost (ROR = 2287.27), and brimonidine (ROR = 2118.52); in non-ophthalmology, they were tralokinumab (ROR = 2609.12), trazodone (ROR = 2377.07), and belantamab mafodotin (ROR = 680.28). For more detailed information, please refer to Figure 3 and Table 4. Subgroup analyses based on age, gender, and underlying diseases can be found in Supplementary Figures S1 to S3. 

Figure 3.

 

Distribution of drugs causing drug-related keratitis, with a total of 76 drugs identified as positive signals through disproportionality analysis. After excluding drugs intended for the treatment of keratitis and drugs with identical generic names but different brand names, 17 drugs were retained, including seven ophthalmic drugs and 10 non-ophthalmic drugs. The heatmap color intensity indicates the relative risk of drug-related keratitis, with darker colors representing higher risks.

Table 4.

 

Statistical Values and Distribution of Drug-Related Keratitis

 

Drug Risk and Drug-Induced Time of Drug-Related Keratitis

We assessed the potential risk of drug-related keratitis for each drug based on BCPNN value and further evaluated their drug-induced time. In the risk assessment of drug-related keratitis, the top five drugs with the highest risk were dorzolamide (BCPNN value = 11.71), trazodone (BCPNN value = 11.11), tralokinumab (BCPNN value = 11.08), travoprost (BCPNN value = 10.81), and brimonidine (BCPNN value = 10.77). Regarding drug-induced time, the top five drugs with the shortest median drug-induced time were zoledronic acid, imiquimod, allopurinol, docetaxel, and lamotrigine, arranged in descending order. For more detailed information, please refer to Table 5 and Figure 4. The therapeutic effects of the above-mentioned drugs can be found in Table 6. 

Table 5.

 

Drug-Induced Time Distribution of Drug-Related Keratitis Caused by Different Drugs

 

Figure 4.

 

Distribution of risks and drug induction times for drug-related keratitis, arranged in descending order based on drug risk and drug induction time.

Table 6.

 

Drugs Associated With Drug-Related Keratitis and Their Specific Therapeutic Purposes

 

Comparison of Drug-Induced Times Between Different Types of Drugs

We divided drugs into ophthalmic medications (seven types) and non-ophthalmic medications (10 types) based on whether or not they were used in ophthalmology. We evaluated the differences in drug-induced times using cumulative risk curves. The results showed that, at the same level of risk, the drug-induced time for non-ophthalmic medications was significantly shorter than that for ophthalmic medications. In the comparison of drug-induced time between the two groups, the drug-induced time for non-ophthalmic medications was significantly shorter than that for ophthalmic medications (mean days, 141.02 vs. 321.96, respectively; P < 0.001). For more detailed information, please refer to Figure 5. 

Figure 5.

 

Drug induction time for ophthalmic and non-ophthalmic medications. (A) Cumulative risk curve distribution for ophthalmic and non-ophthalmic medications showing a significant difference between the two groups (P < 0.001). (B) Violin plot illustrating the drug induction times between the two groups, with non-ophthalmic medications having significantly shorter drug induction times compared to ophthalmic medications (P < 0.001).

Discussion

In this study, based on the FAERS database, we have confirmed the epidemiological characteristics of drug-related keratitis, the occurrence of which has been increasing annually since 2004 and reached a peak in 2023. Using disproportionality analysis, we identified 17 drugs that significantly increase the risk of drug-related keratitis; seven are found in ophthalmic medications and 10 in non-ophthalmic medications. We also assessed the risk values and drug-induced times for this series of drugs. We demonstrated that non-ophthalmic medications have a shorter drug-induced time for drug-related keratitis compared to ophthalmic medications. Our study provides data support for guiding personalized medication for some patients with drug-related keratitis and adds supplementary information to the labeling of some drugs regarding potential ocular adverse reactions. 

The potential ocular toxicity of medications used to treat glaucoma has been confirmed in several studies. There have been reports that local administration of dorzolamide can induce marginal keratitis. This phenomenon is not due to the preservative benzalkonium chloride in dorzolamide, as one patient continued to use timolol maleate, which contains the same preservative as dorzolamide but did not induce marginal keratitis. Therefore, it is evident that the occurrence of marginal keratitis in this patient was a result of dorzolamide itself and not the preservative benzalkonium chloride.¹⁸ Long-term use of anti-glaucoma medications can lead to decreased tear secretion, decreased conjunctival goblet cells, and increased macrophages and lymphocytes, thereby affecting the normal function of the corneal epithelium.¹⁹ Furthermore, the topical use of prostaglandin derivatives may increase the risk of recurrence of herpetic keratitis.^20,21 Additionally, signs of keratoconjunctivitis have been observed in patients with glaucoma using brimonidine, with the allergen primarily concentrated in the active ingredient brimonidine itself rather than the preservative.²² Moreover, cases of drug-related keratitis have been reported with the use of carbonic anhydrase inhibitor brinzolamide for glaucoma treatment.²³ In our study, we found that several medications used to treat glaucoma carry a high risk of causing drug-related keratitis, including prostaglandin analogs and carbonic anhydrase inhibitors. These drugs also tend to have a longer drug-induced time, typically exceeding 100 days. This longer time further emphasizes the risk of ocular adverse reactions, including drug-related keratitis, associated with these medications when used long-term. 

Trazodone is a medication clinically used to treat depression, and there have been reports of its adverse effects causing dry eye keratitis.²⁴ On the other hand, a history of medication for inflammatory bowel disease and anxiety and depression has also been linked to the occurrence of keratitis.²⁵ Among patients treated with belantamab mafodotin for multiple myeloma, two-thirds have reported drug-related punctate keratitis, further suggesting dysfunction of corneal epithelial stem cells.^26–28 The potential reasons for the significant ocular adverse reactions caused by belantamab mafodotin, an antibody–drug conjugate (ADC), include the presence of abundant blood flow in the eyes, a large population of rapidly growing cell subgroups, and a variety of cell surface receptors.²⁹ Drug-related keratitis caused by a series of ADCs is not uncommon and often occurs concomitantly with other visual function disorders such as blurred vision and dry eye.³⁰ In our study, we found that psychiatric drugs such as trazodone and lamotrigine, as well as the ADC drug belantamab mafodotin, carry a higher risk of causing drug-related keratitis. These findings are consistent with previous studies and case reports mentioned above. 

The risk of ocular adverse reactions associated with oncology-related medications has been reported in several studies, but the assessment of drug-related keratitis for these medications is still lacking. Moreover, the adverse reactions of some drugs are significantly correlated with the dose and duration of use.³¹ Taxane drugs are commonly used in cancer chemotherapy and have a high ocular surface toxicity. In a cross-sectional study, it was found that subjects treated with paclitaxel had higher Ocular Surface Disease Index scores, further confirming the ocular toxicity of this class of drugs.³² In 2020, nearly 200 million women worldwide were diagnosed with breast cancer, with nearly 60% of them having invasive breast cancer, often requiring chemotherapy. Paclitaxel and docetaxel are the preferred drugs for these subjects. In populations undergoing long-term use of taxane chemotherapy drugs, there is a need for increased awareness of the risk of keratitis.³³ 

The risk of ocular adverse reactions associated with some medications used to treat skin-related diseases has also been increasingly recognized in recent years. For example, the most common ocular adverse reactions of imiquimod, used to treat periorbital skin lesions, include conjunctivitis and eye discomfort.³⁴ Additionally, randomized controlled trials investigating the use of imiquimod in the treatment of periorbital nodular basal cell carcinoma have shown its ability to inhibit tumor angiogenesis, stimulate innate and adaptive immunity, and induce tumor cell apoptosis. However, it is important to acknowledge that local use of this medication can lead to adverse ocular reactions, including conjunctivitis (95%), keratitis (84%), and foreign body sensation (79%).³⁵ Furthermore, the medication dupilumab, used to treat atopic dermatitis, has been reported to cause adverse reactions such as corneal ulceration. However, more common adverse reactions include superficial keratitis, conjunctivitis, blepharitis, and dry eye.³⁶ As dupilumab is a systemic monoclonal antibody introduced in recent years for the treatment of moderate to severe atopic dermatitis, the potential mechanisms underlying its ocular adverse reactions remain unclear. Some hypotheses suggest that dupilumab may disrupt the production of mucin and goblet cells in the conjunctiva through the inhibition of interleukin (IL)-4 and IL-13, leading to tear film instability and subsequent corneal erosion.³⁷ Therefore, it is important to remain vigilant regarding the risk of ocular adverse reactions associated with some skin medications, and adjustments in medication may be necessary for patients with skin diseases who also suffer from keratitis. 

Due to the lack of a gold standard for diagnosing drug-related keratitis, it is challenging to differentiate it from exacerbation of primary ocular diseases or complications following ocular surface or intraocular surgeries. Consequently, clinicians may mistakenly interpret corneal lesions during medication as worsening of the underlying eye condition, leading to increased dosage or prolonged medication duration, thereby exacerbating the condition. Therefore, when encountering patients with abnormal corneal changes or keratitis, clinicians should consider the possibility of drug-related keratitis. Having awareness of this diagnosis prevents indiscriminate escalation of eye drop dosage or additional use of other eye drops, thereby reducing further medication-induced damage. Clinicians should elevate their diagnostic awareness of drug-related keratitis to improve early detection rates and minimize lesion damage. In summary, when prescribing medications to high-risk patients susceptible to drug-related keratitis, it is crucial to prevent the toxic effects of eye drops and the ocular disease risks associated with systemic medications. The most important aspect is to accurately diagnose and treat the primary disease comprehensively, avoiding the indiscriminate mixing and abuse of multiple medications and minimizing the frequency and duration of medication use. Additionally, to our knowledge, our study is the first to discover that the drug-induced time for drug-related keratitis from systemic medications is significantly shorter than that from ophthalmic medications. We analyzed the potential reasons for this finding, which may include the following two aspects: First, adverse reactions of systemic medications prior to market release primarily focus on the heart, liver, kidneys, and other organs,³⁸ and the risks of adverse effects on the eyes, especially the cornea, are relatively underestimated and potentially overlooked. As a result, the drug-induced time for systemic medications could be shorter than that for ophthalmic medications. Second, in our study, the main ophthalmic medications causing drug-related keratitis were found to be glaucoma-related drugs, which have a longer treatment cycle.³⁹ This might be one of the reasons why the drug-induced time for ophthalmic medications is significantly longer than that for systemic medications. Finally, our study also found that reports of drug-related keratitis have shown an increasing trend over the years. The potential reasons for this may include population growth and a significant increase in the use of certain drug groups, such as biologics.⁴⁰ 

This study also has certain limitations that should be acknowledged. First, although disproportionality analysis served to identify a statistical method used to determine correlations between targeted drugs and adverse events, it did not establish a definitive causal relationship between targeted drugs and adverse reactions, nor did it exclude other confounding factors such as age, gender, country, race, or underlying diseases. Second, the data utilized in this study originated from the FAERS database, which relies on voluntary and spontaneous reporting so it may be influenced by recent research or media reports, potentially introducing bias.⁴¹ Third, the modes of drug administration in the FAERS database are not specified in detail. For example, it is unclear whether “ophthalmic” includes topical eye drops or also encompasses periocular injections. Similarly, the term “intraocular” is not explicitly defined to indicate whether it includes intravitreal implants or intravitreal injections. Additionally, the severity of adverse reactions caused by these drugs is not thoroughly reported; only outcome information related to the subjects is provided. Therefore, further validation using data from alternative adverse drug reaction databases is still warranted to enhance the robustness and reliability of these findings. 

Conclusions

This study utilized a large dataset from real-world adverse drug reaction databases to screen a series of drugs that may induce drug-related keratitis, assessing the risk values and drug-induced time for different medications. By comparing and analyzing the differences in drug-induced time between ophthalmic and non-ophthalmic medications, it revealed important epidemiological characteristics of drug-related keratitis. This provides valuable data-driven guidance for reducing the incidence of drug-related keratitis and guiding clinical medication. By describing the risk profile and occurrence patterns of drug-related keratitis, this study lays the foundation for strengthening clinical decision-making and patient care in the context of medication-related ocular complications. 

Acknowledgments

Supported by grants from the National Natural Science Foundation of China (82271054 to ZL) and Natural Science Foundation of Fujian (2023J01012 to CH). 

Authors' Contributions: Shi-Nan Wu conceived the research idea. Shi-Nan Wu, Xiao-Dong Chen, Yu-Qian Wang, Dan Yan and Shu-Jia Guo conducted data cleaning and literature review. Shi-Nan Wu, Dan-Yi Qin, Chang-Sheng Xu, Shao-Pan Wang, Qing-He Zhang, Lin Chen, Yu-Wen Liu, Caihong Huang and Lingfangzi Zhu contributed to drafting and critically revising the work for intellectual content. Shi-Nan Wu conducted the analysis and created the figures and tables. Jiaoyue Hu, and Zuguo Liu provided a critical review of the manuscript. All authors have read and approved the manuscript. 

Ethical Statements: The data source for this study is the public database FAERS database, which does not contain ethics. Previous studies of the database did not require ethical approval. Therefore, this study does not require the approval of the Ethics Committee. 

Data Availability: The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: FAERS Publish Dashboard (https://www.fda.gov/drugs/questions-and-answers-fdas-adverse-event-reporting-system-faers/fda-adverse-event-reporting-system-faers-public-dashboard). 

Disclosure: S.-N. Wu, None; X.-D. Chen, None; Q.-H. Zhang, None; Y.-Q. Wang, None; D. Yan, None; C.-S. Xu, None; S.-P. Wang, None; L. Zhu, None; D.-Y. Qin, None; S.-J. Guo, None; L. Chen, None; Y.-W. Liu, None; C. Huang, None; J. Hu, None; Z. Liu, None