Drug-related macular edema: a real-world FDA Adverse Event Reporting System database study

Abstract

Purpose

This study aims to assess the risks associated with drug-induced macular edema and to examine the epidemiological characteristics of this condition.

Methods

This study analyzed data from the U.S. Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) database from January 2004 to June 2024 to conduct a disproportionality analysis identifying drugs with positive signals of drug-induced ME. Additionally, the onset time of ME associated with these drugs was examined.

Results

In the FAERS database, a total of 490 drugs were reported to pose a risk of drug-induced ME. Disproportional analysis and screening further identified 8 drugs that significantly increased this risk. Among these, one is ophthalmic drugs, including Latanoprost (ROR = 5.51), and ten are non-ophthalmic drugs, including Cefuroxime (ROR = 75.93), Fingolimod (ROR = 30.69), and Siponimod (ROR = 20.51).

Conclusions

This study utilizes the FAERS database to investigate potential associations between drug use and the occurrence of ME, rapidly identify drugs that may induce the condition, and propose research strategies. These findings hold significant value for guiding clinical medication practices.

Introduction

Drug-induced macular edema (ME) is a significant, though often underestimated, adverse effect in pharmacotherapy, particularly relevant to ophthalmology [1]. Damage to the macular region of the retina can lead to profound, irreversible impairment in visual function. It involves pathological changes in the macula, the retina’s central area, characterized by degeneration or functional impairment of its structures [1]. Since macular is crucial for fine vision and color discrimination, its damage severely diminishes life quality of patients [2]. The mechanisms underlying drug-induced ME are complex, including direct drug toxicity, metabolite accumulation, and changes in local blood flow. Additionally, some drugs may contribute to macular damage through oxidative stress, inflammatory pathways, or mitochondrial dysfunction, further complicating the condition’s prevention and management [3].

The increased use of chronic disease management and various systemic treatments has heightened awareness of medications potentially linked to maculopathy. For instance, anticoagulants prescribed for cardiovascular conditions may cause ischemic macular damage by disrupting retinal microcirculation. Similarly, some immunomodulators can induce toxicity in retinal pigment epithelial cells, impairing macular function [4]. Moreover, certain anticancer agents, particularly targeted therapies, may directly or indirectly harm the macula due to their actions on rapidly dividing cells. Despite some case reports and small-scale studies addressing macular toxicity, the incidence, mechanisms, and risk factors of drug-induced ME remain underrecognized and inadequately studied. For clinicians, understanding drug-induced ME mechanisms and evaluating drug safety are vital in treatment selection to minimize vision loss risks. For patients, early risk recognition and timely visual monitoring are essential to enable interventions before irreversible damage ensues.

Existing data on ME largely stems from clinical trials and observational studies, which limit the scope of populations, diseases, and medications evaluated. In real-world settings, large-scale, comprehensive studies and robust data on adverse reactions, such as ME, are lacking. Notably, no extensive analysis of drugs linked to an elevated risk of ME, derived from the FDA’s Adverse Event Reporting System (FAERS), is currently available. Therefore, this study aims to identify potential risk signals of drugs that may increase the risk of ME, assess associated risk levels, and systematically examine the relationship between these drugs and maculopathy [5]. This analysis will support physicians in evaluating the risk of vision impairment when selecting treatment options and guide future foundational research on underlying mechanisms. Ultimately, it will enhance patient visual function protection and improve medication safety. The graphical abstract of this article is shown in Fig. 1.

Fig. 1

Graphical abstract

Materials and methods

Data source

The FDA electronically processes and collects data on adverse reactions, which are published quarterly in ASCII and XML formats on its official website, where it is freely available for download. Adverse events symptoms are encoded using the internationally standardized and clinically validated Medical Dictionary for Regulatory Activities (MedDRA). Relevant data retrieval and extraction are facilitated by the Open Vigil 2.1-MedDRA tool, accessible via the provided link. To improve data accuracy and reliability, only cases reported by healthcare professionals (doctors and pharmacists, code = MD) were included [6]. From January 2004 to June 2024, 21,035,995 original entries were recorded; after removing duplicate reports, 17,785,793 entries remained. Due to the redundancy of drug names under commercial brands, deduplication yielded 1,117 unique drugs. The data cleaning process is depicted in Fig. 2.

Fig. 2

The flow diagram of screening reports containing macular edema elicited by diverse agents from the FAERS database

Data extraction

The data collection for this study covered the period from the first quarter of 2004 through the second quarter of 2024. We identified the number of ME reports and their corresponding PRIMARYID codes from the preferred term “macular oedema” in the REAC file. Using these PRIMARYIDs, we then identified and eliminated duplicate reports within the DEMO file to accurately determine the number of ME cases. Additionally, we extracted information on case age and the reporters’ countries (REPORTER_COUNTRY). Our analysis was restricted to drugs primarily suspected of causing these events, excluding those categorized as “secondary suspect drugs,” “concomitant drugs,” and “interaction drugs” due to the substantial uncertainty of their association with adverse events.

Exclusion criteria

Patients receiving glaucoma treatment, including those using ophthalmic medications such as Latanoprost and Bimatoprost, were excluded from the study. This decision was made to avoid confounding due to the potential pre-existing macular edema in these patients, which might already be part of their glaucoma management. As glaucoma patients typically undergo regular ophthalmic follow-ups, the detection of macular edema in these individuals is more likely, which could introduce bias into the results.

Additionally, patients with diabetes as the primary indication for treatment, particularly those receiving insulin or other anti-diabetic medications, were excluded. This was done to avoid confounding, as insulin therapy is frequently used to manage diabetic macular edema (DME). The elevated signals for macular edema in this cohort may reflect the treatment of DME rather than a direct relationship between insulin or other anti-diabetic medications and macular edema.

Statistical analysis

In our study, we applied multiple disproportionality analysis methods to identify potential adverse event signals including the Reporting Odds Ratio (ROR) as detailed by Rothman et al. [7], the Proportional Reporting Ratio (PRR) as described by Evans et al. [8], the Bayesian Confidence Propagation Neural Network (BCPNN) introduced by Bate [9], and the Multi-Item Gamma Poisson Shrinker (MGPS) method from Heo et al. [10]. Each method was statistically analyzed using signal detection parameters derived from a 2 × 2 contingency table, as provided in Supplementary Table S1. Detailed formulas and criteria for signal generation are provided in Supplementary Table S2. To robustly identify potential associations between drugs and ME, a signal was considered valid only if it met the criteria across all four algorithms, indicating a potential link. Additionally, we analyzed the duration of drug use leading to drug-induced ME, comparing onset times across different drugs. Using the BCPNN algorithm, we categorized the ME risks associated with various drugs, divided drug-related ME adverse reactions into quartiles, and assessed the durations of these reactions.

Results

Subject descriptive analysis

Based on data from the FAERS database between 2004 and 2024, a total of 1,885 patients reported experiencing drug-related ME adverse events. The average age of these patients was approximately 59.78 ± 16.23 years, with 48.2% being female. The age of female patients primarily ranged between 55 and 70 years, while male patients were mainly between 60 and 75 years (Fig. 3B). Additionally, we observed an upward trend in reported cases of drug-related ME, peaking in 2015 (Fig. 3A). In terms of reporting sources (Fig. 3C and D), the top five countries were the United States (n = 674, 35.76%), Japan (n = 288, 15.28%), France (n = 153, 8.12%), Germany (n = 150, 7.96%), and the United Kingdom (n = 57, 3.02%). The primary outcomes for these cases were categorized as “Other Serious (Important Medical Event)” (77.08%) and “Hospitalization (Initial or Prolonged)” (10.98%) (Fig. 3E). For further details, please refer to Table 1.

Fig. 3

Characteristics of reports involved in drug-induced macular edema from the FAERS database. Notes Fig. 3A displays a timeline chart showing the distribution of reported adverse events of ME over time. Figure 3B depicts a pyramid chart of age distribution among patients reporting adverse events of ME, categorized by gender. Figure 3C presents a histogram of the distribution of reported adverse events of ME by country. Figure 3D showcases a heatmap of the distribution of reported adverse events of ME by country. Figure 3E illustrates a pie chart representing the distribution of outcomes among patients experiencing adverse events of ME

Table 1 Baseline data of macular oedema patients reported in the FAERS database

Classification of drugs associated with macular edema

We selected the top 30 drugs from a total of 490 drugs based on the number of reported cases and conducted signal calculations. Additionally, we verified the brand and generic names of these drugs. Considering that some drugs are intended to treat macular-related diseases but may still show a positive signal due to insufficient efficacy, we excluded these drugs from the analysis. Ultimately, we identified the top 14 drugs with the highest number of reports, with specific details available in Fig. 4. We further calculated the signals for these 14 drugs, applying four positive signal screening criteria based on disproportionality analysis. As a result, we identified 8 drugs associated with drug-induced ME, including 1 drug primarily used in ophthalmology and 7 in non-ophthalmology. The ophthalmic drugs were Latanoprost (n = 5). In non-ophthalmic studies, the leading drugs were Fingolimod (n = 303), Cefuroxime (n = 66) and Paclitaxel (n = 40). For more details, please refer to Fig. 3; Table 2.

Fig. 4

The identification and classification of drugs associated with macular edema

Table 2 Statistical values and distribution of drug-related macular edema

Drug-associated risk of macular edema and drug-induced onset time

We assessed the potential risk of drug-induced ME for each drug based on ROR values and further evaluated their drug-induced onset times. Among the drugs associated with ME, the top five drugs with the highest risk were Cefuroxime (ROR = 75.93), Fingolimod (ROR = 30.69), Encorafenib (ROR = 19.77),Siponimod (ROR = 20.51), Latanoprost (ROR = 5.51). Regarding drug-induced onset time, the top five drugs with the shortest median onset times were Siponimod (41.00 days), Paclitaxel (79.00 days), Encorafenib (122.33 days), and Fingolimod (535.94 days). For more details, please refer to Table 3; Figs. 5 and 6.

Table 3 Drug-Induced Time distribution of drug-related macular oedema caused by different drugs

Fig. 5

Forest plots of different agents inducing macular edema under ROR classification. Notes Drugs associated with drug-induced ME are predominantly distributed among ophthalmic medications, oncology-related medications, respiratory system/infection medications, nervous system/pain management medications, endocrine system, cardiovascular system medications, immune system drugs

Fig. 6

Time to event onset of macular edema elicited by various drugs. Notes Fig. 6A shows a cumulative risk timeline of drug induction for eye system drugs and non-eye system drugs. Figure 6B exhibits a violin plot for time disparities in two group induction

Discussion

In this study, we analyzed 490 maculopathy-related drugs reported in the FAERS database since its establishment in 2004. We then selected the top 14 drugs based on the number of reports and conducted a disproportionality analysis. Ultimately, we identified 8 drugs associated with drug-induced ME, including ophthalmic drugs, oncology drugs, antibiotics, immunosuppressants, immunomodulators, cardiovascular medications, and non-steroidal anti-inflammatory drugs (NSAIDs), shown in Fig. 3. We further analyzed the risk of these drug-induced maculopathies and the differences in drug-induced onset times. To our knowledge, this is the first large-scale real-world study on drug-induced maculopathy based on the FAERS database.

Although most studies focus on ophthalmic drugs related to ME, it is equally important to recognize that non-ophthalmic drugs can also increase the risk of ME. Our study revealed that drugs associated with a high risk of ME are not limited to a specific category but include ophthalmic drugs (Latanoprost), antibiotics (Cefuroxime 75.93), immunomodulators (Fingolimod, Siponimod), oncology drugs (Protein-Bound Paclitaxel, Encorafenib, Paclitaxel), and cardiovascular drugs (Sildenafil).

Considering that ophthalmic medications have been implicated in increasing the risk of ME, further understanding of the underlying mechanisms is crucial. Cystoid macular edema (CME) is defined by the thickening of the retina at the macula, accompanied by cystic alterations in the outer plexiform and inner nuclear layers [11]. Although glaucoma in isolation is not regarded as a risk factor for the onset of CME following cataract surgery, the presence of untreated elevated intraocular pressure (IOP) accompanied by glaucomatous damage to the optic nerve or retinal nerve fiber layer, as well as associated visual field defects, have been demonstrated capable of elevating the risk of developing postoperative pseudophakic CME [12]. Current research presents inconclusive findings concerning the risk of pseudophakic CME associated with preoperative use of prostaglandin analogs (PGAs). In a retrospective analysis of 12 patients treated with latanoprost, 8 patients intentionally ceased treatment one week before surgery, while 4 patients maintained the regimen. Notably, none of the patients who discontinued the treatment developed CME, in contrast to all patients who continued the treatment, who did develop CME (P = 0.003) [13]. Other studies have indicated that the preoperative administration of PGAs does not elevate the risk of developing CME [14, 15]. In our study, Latanoprost (ROR, 5.51) commonly used PGAs for treating glaucoma and elevated intraocular pressure, which demonstrated a high risk of inducing ME based on data from the FAERS database. In a clinical trial involving cataract patients administered Latanoprost, it was reported that the use of Latanoprost results in the disruption of the blood-aqueous barrier and significantly elevates the incidence of ME [16]. However, most case reports on latanoprost-associated ME involved patients with cataract surgery complications, and only a few highlighted the patients’ diabetes status, which increases their risk of developing ME [15]. In conclusion, Latanoprost is highly effective in treating glaucoma; however, after excluding patients with glaucoma, our analysis shows a strong association between Latanoprost and the increased risk of ME in certain individuals. Therefore, clinicians should consider individual patient risk factors when prescribing these medications, especially in those with pre-existing conditions or those with other risk factors for the development of ME.

In cancer treatment, the potential side effects of drugs are a significant concern, especially the complication of ME, which is a key focus of this study. Chemotherapy remains the primary treatment for most cancers, including breast, lung, genitourinary, and head and neck cancers. The taxane class of drugs, including docetaxel, paclitaxel, and nanoparticle albumin-bound paclitaxel, are microtubule-stabilizing agents used to treat various malignancies. However, information on the frequency and incidence of ocular adverse events associated with these agents, including CME, remains limited. Most reports of taxane-induced CME come from case studies [17, 18]. In our study, we found that paclitaxel (ROR, 7.42) and protein-bound paclitaxel (ROR, 14.82) were associated with a higher risk of drug-induced CME. A retrospective cohort study using a US private health database with over 150 million enrollees compared women receiving taxanes with those receiving tamoxifen. The results indicate an increased risk of CME with taxane-based chemotherapy, although the findings were imprecise due to the wide confidence interval (HR 1.33; 95%CI 0.70–2.53) [19]. A case report described two metastatic breast cancer patients who developed CME following treatment with nab-paclitaxel and bevacizumab [20]. One patient developed CME after a single dose of bevacizumab, suggesting that nab-paclitaxel alone was the primary cause, rather than the combination of the two drugs. The second patient experienced worsening vision despite discontinuing bevacizumab, but after stopping nab-paclitaxel, her vision improved within two months, and a three month follow-up showed significant reduction in CME. This further supports nab-paclitaxel as the primary causative agent. It has been postulated that taxane-induced CME results from disruption of the blood-retinal barrier, leading to fluid accumulation in intracellular spaces [21]. Additionally, taxanes may cause retinal cell toxicity due to their mechanism of inhibiting microtubular reorganization [21]. The only definitive treatment for taxane-induced CME is cessation of the causative agent [22], with visual improvement observed as early as three weeks after discontinuation. Although taxane-induced CME is reversible, careful consideration should be given to its impact on visual function and the clinical efficacy of taxane-based chemotherapy. The incidence of taxane-related maculopathy may be underestimated, requiring clinicians to exercise caution when prescribing taxanes. Patients receiving taxane-based therapy should monitor their vision regularly and undergo ocular examinations if any visual symptoms develop.

Encorafenib, a BRAF inhibitor, is typically used in combination with MEK inhibitors (MEKi) for the treatment of BRAF-mutated solid tumors [23,24,25,26]. However, studies on ocular adverse events related to BRAFi + MEKi combination therapy are limited, with most findings originating from clinical trials. Reported ocular adverse events included uveitis [27, 28], chorioretinopathy [27, 29], retinal detachment [27, 29], blurred vision [27, 29, 30], and visual impairment [27]. In the literature, the most frequently studied ocular adverse events are MEKi-associated retinopathy (MEKAR) and uveitis [31,32,33]. Our disproportionality analysis identified a significant association between Encorafenib and ME (ROR, 19.77). However, the number of reports in our FAERS study is relatively small (n = 13), and the discrepancy between clinical trial data and FAERS data may stem from variations in the definition of terms. For example, in the coBRIM and COLUMBUS clinical trials, “serous retinopathy” encompassed a wide range of preferred terms, including ME, retinal detachment, chorioretinitis, chorioretinopathy, CME, macular retinal pigment epithelium detachment, and several others [32]. Thus, the reporting rate in our FAERS data apperars to provide a reasonable assessment. However, larger epidemiological studies are needed to further evaluate the risk of ME associated with Encorafenib.

Fingolimod and Siponimod, selective S1P1 receptor agonists, are commonly used in the treatment of Multiple sclerosis (MS). MS is a chronic autoimmune demyelinating disease characterized by diverse clinical, radiological, and pathological features. A systematic review [34] identified 21 studies reporting fingolimod-associated ME, with symptoms completely resolving upon drug discontinuation. However, one study noted that symptoms reappeared two months after the reintroduction of fingolimod [35]. In a Phase 3 clinical trial [36], 2% of patients (n = 18) in the Siponimod group developed ME, compared to less than 1% (n = 1) in the placebo group. The development of ME is likely linked to the activation of the S1P pathway, which impacts vascular permeability [37]. S1P is a bioactive lipid that plays a crucial role in processes such as cell proliferation, migration, and survival. It has been shown to influence vascular permeability, a key factor in the pathogenesis of ME. In diabetic retinopathy, elevated S1P levels have been associated with disruption of the blood-retinal barrier (BRB), leading to increased permeability of retinal endothelial cells and fluid accumulation in the macula.

Cefuroxime, as a broad-spectrum antibiotic, has been widely used in ophthalmic surgery, especially in the prevention of intraocular inflammation after surgery. Some previous cases found that transient ME following intracameral cefuroxime injection during phacoemulsification, where varying doses (9 mg/0.1 mL [38], 2 mg/0.1 mL [39], and 1 mg/0.1 mL [40,41,42]) were associated with ME. The issue was resolved after switching from corneal incision injection to direct anterior chamber injection, suggesting that cefuroxime may disrupt retinal barriers and induce retinal edema [42]. Another population-based cohort study assessed the effectiveness and retinal safety of intracameral cefuroxime injection at the conclusion of cataract surgery for preventing postoperative endophthalmitis [43]. The study shows that cefuroxime injections significantly reduce the risk of postoperative endophthalmitis, particularly in high-risk cases such as those with capsular rupture, while no increased risk of CME was observed with the use of cefuroxime. These results further indicate that the elevated incidence of CME could be highly linked to the administration of high-dose cefuroxime, potentially due to dilution errors during the preparation of the cefuroxime solution.

Sildenafil citrate, a phosphodiesterase-5 (PDE5) inhibitor initially approved for erectile dysfunction, has since been approved for pulmonary arterial hypertension, benign prostatic hyperplasia, and lower urinary tract symptoms, and is being researched for potential use in treating diabetes mellitus and chronic kidney disease [44]. A case report presents one patient who developed acute anterior uveitis and ME after using sildenafil citrate, suggesting a potential link between sildenafil use and intraocular inflammation in predisposed individuals [44]. Studies have shown that PDE5 inhibitors like sildenafil can increase choroidal blood flow (CBF) with a comparatively minimal impact on retinal vasculature [45]. Given these effects on ocular blood flow, the rise in CBF following sildenafil administration might contribute to the onset of ME.

Additionally, we analyzed the time-to-onset (TTO) for drug-induced CME. Our results showed that Siponimod had the shortest median TTO at 41 days, while Fingolimod exhibited the longest median TTO at 535.94 days among all drugs studied. The median TTO for Paclitaxel was 79 days, Encorafenib 122.33 days. This study utilized the comprehensive FAERS database, offering a broader perspective on TTO variability for different drugs associated with CME. These findings highlight the need for clinicians to be vigilant for signs of CME, particularly during early treatment stages with drugs that show shorter TTOs.

This study has several limitations. First, the FAERS database is a spontaneous reporting system, relying on voluntarily submitted adverse event reports, which may result in underreporting. Second, many reports lack sufficient evidence to establish a causal relationship between the reported adverse events and drug exposure. This challenge in confirming causality is a common limitation in pharmacovigilance and observational cohort studies. Furthermore, FAERS cases frequently contain incomplete information, including dosage, comorbidities, onset time, and other crucial details. Consequently, our data mining results should be interpreted cautiously, with conclusions drawn through comprehensive, evidence-based assessments.

Conclusion

This study utilizes the FAERS database to explore potential associations between drug exposure and the occurrence of ME, enabling the rapid identification of medications that may contribute to ME and proposing strategies for further investigation. Management of drug-induced ME requires a patient-centered approach, particularly for cancer patients, where treatment adjustments may be limited, while other patient groups may have more flexibility in managing medications. Future research should integrate real-world data with experimental validation to develop a comprehensive safety profile for drugs potentially associated with ME, thereby supporting safer therapeutic approaches.