Abstract
Background/Aim: Immune checkpoint inhibitors (ICIs) are widely used anticancer drugs, but they can trigger immune-related adverse events (irAEs), including rare but potentially blinding uveitis. In this study, the associations between ICIs and uveitis, and the time to uveitis onset, were evaluated using data from the Japanese Adverse Drug Event Report (JADER) database.
Patients and Methods: We performed disproportionality analysis using data from the JADER database (April 2004–June 2024). Positive signals were identified using reporting odds ratios (RORs) with 95% confidence intervals (CIs). Time-to-onset analyses were performed with the Weibull distribution model, and sex-based differences were assessed with Kaplan–Meier curves and the log-rank test.
Results: Among 914,713 cases in the JADER database, 346 were suspected ICI-induced uveitis. Positive signals were detected for nivolumab [ROR, 9.05 (95%CI=7.81-10.50)], ipilimumab [9.82 (8.20-11.76)], and pembrolizumab [4.94 (4.05-6.01)] but not for other ICIs. The median onset time was approximately 2 months (65 days for nivolumab, 61 days for ipilimumab, and 70 days for pembrolizumab). Pembrolizumab-induced uveitis occurred significantly earlier in female patients than in male patients (29.5 vs. 91 days, p=0.015). Onset patterns were classified as early failure for nivolumab and random failure for ipilimumab and pembrolizumab. The outcomes were mostly good, and severe cases of uveitis were rare.
Conclusion: ICI-induced uveitis typically occurs within two months of treatment initiation, with pembrolizumab-induced uveitis occurring earlier in females than in males. Because causation cannot be established on the basis of this analysis, large-scale prospective clinical studies are needed.
- Adverse reactions
- immune checkpoint inhibitors
- Japanese Adverse Drug Event Report database
- time-to-onset
- uveitis
Introduction
Uveitis is an inflammatory condition that affects various parts of the eye, including the uvea, retina, retinal vessels, and vitreous (1, 2). Uveitis is a serious condition that can cause blindness (3), accounting for 10-15% of reported blindness cases (4). In developing countries, approximately 25% of irreversible blindness is attributable to uveitis and its complications (5). Drug-induced uveitis has been reported as an adverse event for several medications (6).
Immune checkpoint inhibitors (ICIs) are widely used anticancer drugs with a different mechanism of action than previous cytotoxic anticancer drugs. ICIs are classified as programmed cell death protein 1 (PD-1) inhibitors, programmed death ligand 1 (PD-L1) inhibitors, and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) inhibitors. In Japan, commonly used ICIs include PD-1 inhibitors (nivolumab, pembrolizumab, and cemiplimab), PDL-1 inhibitors (atezolizumab, avelumab, and durvalumab), and CTLA-4 inhibitors (ipilimumab and tremelimumab). Because ICIs activate the immune system, they are associated with immune-related adverse events (irAEs) that can affect almost all organs and systems, including the liver, gastrointestinal tract, skin, lungs, musculoskeletal system, nervous system, and endocrine glands (7-13). Although most irAEs are reversible and of low severity (13), early identification is critical to ensure uninterrupted ICI treatment.
The pathogenesis of uveitis is thought to involve an abnormal immune response (14), which suggests that ICIs may be associated with uveitis. Ocular adverse events occur in approximately 1% of patients treated with ICIs (15). However, because of the low incidence of ICI-associated uveitis, existing information is largely limited to case reports and case series (16-18).
To investigate such rare adverse events, large-scale datasets, such as spontaneous adverse event reporting systems, are valuable tools. Such databases have been used in several studies to investigate ICI-related irAEs (19-24). Although the time-to-onset of ICI-associated uveitis has been evaluated in some of these studies (20, 22, 23), investigations of the onset timing specific to each ICI are lacking.
Although rare, ICI-associated uveitis can cause blindness, and predicting the time-to-onset of ICI-associated uveitis is critical for early detection and intervention. In this study, we investigated the association between each ICI and the time to uveitis onset by analyzing data in the Japanese Adverse Drug Event Report (JADER) database.
Patient and Methods
Database analysis. We obtained data from the JADER database, which is managed by the Pharmaceuticals and Medical Devices Agency (PMDA). The JADER database contains fully anonymized data submitted by healthcare professionals and pharmaceutical companies. The database is structured into four tables: “Demo”, “Drug”, “Reac”, and “Hist”. The “Demo” table includes basic patient information, such as age, sex, and reporting year. The “Drug” table contains details on drug administration, including the drug name, route of administration, start and end dates, dosage, and classification of involvement in adverse reactions (e.g., suspected, concomitant, or interaction). The “Reac” table contains adverse events, outcomes, and onset dates, and the “Hist” table contains information on patient medical history. Adverse events in the JADER database are categorized using the preferred terms (PTs) from the Japanese version of the Medical Dictionary for Regulatory Activities (MedDRA/J). The JADER database is available on the PMDA website (25).
Data extraction. We downloaded data from the JADER database covering the period from April 2004 to June 2024. We focused on 8 ICIs used in Japan: atezolizumab, avelumab, cemiplimab, durvalumab, ipilimumab, nivolumab, pembrolizumab, and tremelimumab. Only cases of ICI-related uveitis classified as “suspected” were included in the analysis. Uveitis cases were identified on the basis of 11 PTs defined in MedDRA/J version 27.0 (Table I). We analyzed patient age and sex, reporting year, drug name, administration start date, clinical outcome, adverse event onset date, and patient medical history.
List of preferred terms of uveitis-related adverse events.
Disproportionality analysis. The reporting odds ratio (ROR) is used by the PMDA and the Netherlands Pharmacovigilance Center to detect signals of adverse events in spontaneous reporting systems (26). The ROR, along with its 95% confidence interval (95%CI), is a quantification of the disproportionality in adverse event reporting between groups (24, 27). Therefore, the ROR is conceptually analogous to the odds ratio in case–control studies, representing the odds of exposure in reported cases compared with non-cases. For this study, we calculated the ROR and 95%CI for uveitis adverse events in patients receiving ICIs. A signal for adverse reactions was considered positive if the lower limit of the 95%CI was >1.
Time to uveitis onset. When a positive signal for ICI-related uveitis was detected in the disproportionality analysis, we analyzed the time-to-onset. Cases were excluded if either the start date of ICI administration or the onset date of the adverse event was unavailable. Cases were also excluded if they were identified as duplicates on the basis of matching treatment initiation dates, adverse event onset dates, patient age and sex, and primary disease. The time-to-onset was defined as the difference between the start date of ICI administration and the onset date of uveitis, plus 1. To evaluate the pattern of onset, we used the Weibull distribution model and examined the Weibull shape parameters (19, 28). Weibull shape parameters characterize the distribution of failure rates over time, where failure rates correspond to the onset of adverse events. The shape parameter α indicates the distribution breadth, with larger values representing a wider spread. The parameter β reflects the hazard ratio as follows:
i) If β>1 and the 95%CI does not include 1, the frequency of adverse events is interpreted as increasing over time.
ii) If β<1 and the 95%CI does not include 1, the frequency of adverse events is interpreted as decreasing over time.
iii) If β=1 or the 95%CI includes 1, the frequency of adverse events is considered constant over time. Additionally, we evaluated the influence of patient sex on the time to uveitis onset using the Kaplan–Meier method and the log-rank test.
Statistical analysis. All the statistical analyses were performed with JMP Pro 17.2 (SAS Institute Inc., Cary, NC, USA). A p-value of <0.05 was considered statistically significant.
Results
Database analysis. The JADER database contains a total of 914,713 drug adverse events dated from April 2004 to June 2024. Of these drug adverse events, 346 were cases of suspected ICI-induced uveitis. The characteristics and clinical outcomes of patients with suspected ICI-induced uveitis are shown in Table II. More patients were male than female, and the most commonly affected age group was patients in their 60s and 70s. The most common cancer type was malignant melanoma, followed by lung and kidney cancers. Among ICI therapies, PD1 monotherapy accounted for approximately 60% of cases, and combination therapy with PD1 and CTLA-4 accounted for approximately 40% of cases. The clinical outcome was favorable in most cases. The number of reported cases gradually increased from 2016 to 2019, with no major changes thereafter (Figure 1).
Characteristics and clinical outcomes of patients with uveitis as suspected adverse reactions to immune checkpoint inhibitors included in the Japanese Adverse Drug Event Report database.
Yearly number of cases of suspected immune checkpoint inhibitor-induced uveitis.
Disproportionality analysis. We calculated the ROR and 95%CI for suspected ICI-induced uveitis (Table III). Owing to the limited number of cases (≤1 each), RORs could not be calculated for cemiplimab, tremelimumab, or avelumab. Positive signals for uveitis were detected for nivolumab, ipilimumab, and pembrolizumab, whereas no signals were detected for atezolizumab or durvalumab.
Reporting odds ratios of suspected uveitis adverse reactions to immune checkpoint inhibitors.
Time-to-onset analysis. We analyzed the time to uveitis onset associated with nivolumab, ipilimumab, and pembrolizumab, for which positive signals were detected in the disproportionality analysis. The analysis included 85 cases for nivolumab, 40 cases for ipilimumab, and 45 cases for pembrolizumab that satisfied the eligibility criteria. The median [interquartile range (IQR)] time-to-onset of suspected uveitis was 65 (range=29-128) days for nivolumab, 61 (range=22-72) days for ipilimumab, and 70 (range=23-150) days for pembrolizumab, which is approximately two months (Table IV). The onset pattern for nivolumab was early failure, and the onset pattern for ipilimumab and pembrolizumab was random failure. In sex-based comparisons, the time to uveitis onset for pembrolizumab was significantly earlier in female patients [median, 29.5 days (IQR=20-87)] than in male patients [91 days (44-220), p=0.015; Figure 2]. The time to uveitis onset did not differ significantly by sex for nivolumab or ipilimumab.
Time-to-onset and Weibull parameters of suspected immune checkpoint inhibitor-related uveitis.
Time-to-onset and cumulative incidence of uveitis among affected patients after treatment with (A) nivolumab, (B), ipilimumab, and (C) pembrolizumab. Statistical analysis was performed using the log-rank test, and a p-value of less than 0.05 was considered statistically significant. IQR: Interquartile range.
Discussion
In this study, we analyzed the associations between ICIs and uveitis, as well as the time to ICI-associated uveitis onset, using data from the Japanese pharmacovigilance database. Positive signals for uveitis were detected for nivolumab, ipilimumab, and pembrolizumab but not for atezolizumab or durvalumab. Given that a positive signal for uveitis induced by atezolizumab was detected in a previous study of data from the U.S. Food and Drug Administration Adverse Event Reporting System database (22), the lack of a positive signal in this study may be related to the limited number of reports of adverse events for atezolizumab and durvalumab in the JADER database.
We found that the number of suspected cases of ICI-induced uveitis in the JADER database was greater in male patients than in female patients. However, no significant difference in the number of cases between male and female patients was observed in previous studies of data from other pharmacovigilance databases (22, 23). There are numerous possible explanations for this discrepancy. First, the discrepancy may be due to differences in the patients’ cancer types. A greater proportion of patients with lung cancer was included in this study than in previous studies, and lung cancer patients are more likely to be male than female (29). Second, the discrepancy may reflect underlying racial or ethnic differences in the occurrence of adverse events. Race or ethnicity may influence pharmacokinetic factors, including drug efficacy and the frequency of adverse events (30). Third, the discrepancy may reflect a limitation of spontaneous reporting systems, which cannot account for all patients who receive ICIs in actual clinical practice.
Sex differences in adverse reactions to cancer treatment have been reported previously (31). For example, the incidence of irAEs associated with ICIs is greater in premenopausal females than in postmenopausal females and males (32). Sex differences in specific irAEs are also observed, with pneumonia occurring more frequently in females and skin toxicity occurring more frequently in males. Thus, the frequency of uveitis as an adverse event may also differ by sex.
The clinical outcomes in this study were generally favorable, with serious outcomes reported in approximately 10% of cases. This percentage is similar to that found in previous studies of data from other pharmacovigilance databases (22, 23).
We found that the median time to ICI-induced uveitis onset was approximately two months after initiation of the drug, which is consistent with the results of previous pharmacovigilance and case series studies, in which onset times ranging from 6 to 12 weeks were reported (17, 18, 23). A study of Japanese pharmacovigilance data revealed that the onset times of ICI-induced autoimmune disease adverse events varied, with type 1 diabetes and pemphigoid occurring as late as 4 or 5 months after drug initiation and myasthenia gravis and rheumatoid arthritis occurring as early as approximately one month after drug initiation (24).
To our knowledge, this is the first study in which the time to ICI-induced uveitis onset by sex was analyzed. We found that the time to pembrolizumab-induced uveitis onset was significantly shorter for female patients than for male patients. This finding may be attributable to the higher incidence of systemic autoimmune disease in female patients than in male patients (33). Differences in the expression of genes related to immune function may contribute to the increased incidence of irAEs in females.
Additionally, pembrolizumab clearance is lower and the area under the curve (AUC) is higher in female patients than in male patients (34). Thus, sex hormone and autoimmune influences, as well as sex-based pharmacokinetic differences in response to pembrolizumab, may explain this finding.
Study limitations. First, data from a spontaneous adverse event reporting database were analyzed, but this database does not contain data for all patients treated with ICIs. Therefore, we cannot conclude that females generally have a greater incidence of ICI-induced uveitis than males do.
Second, the JADER database contains missing and duplicate data. Missing sex data may have influenced the sex differences that we observed, although sex information was lacking for only 2% of the cases. Duplicate cases could not be excluded from the disproportionality analysis but were excluded from the time-to-onset analysis to minimize their influence. Third, we did not include concomitant medications or the history of autoimmune disease in our analyses, but these factors may cause uveitis. For example, the incidence of ocular immune disease is increased in patients with uveitis or other preexisting ocular inflammatory conditions (35). Therefore, large-scale prospective clinical studies are needed to account for these factors.
Conclusion
Drug-induced uveitis is a rare but sight-threatening adverse event. Given the widespread use of ICIs in cancer treatment, the occurrence of ICI-induced uveitis is an important issue. The findings from this study suggested that ICI-induced uveitis typically occurs approximately two months after treatment initiation. Notably, pembrolizumab-induced uveitis onset may occur earlier in females than in males, suggesting that patients should be monitored closely during the initial stages of treatment.
Footnotes
Authors’ Contributions
Naohito Ide and Ken-ichi Sako conceived and designed the study. Naohito Ide wrote the manuscript. Naohito Ide and Ken-ichi Sako analyzed the data. Ken-ichi Sako and Tomoji Maeda interpreted the results and contributed to the discussion. All the Authors read and approved the final manuscript.
Conflicts of Interest
The Authors have no conflicts of interest to disclose in relation to this study.
Funding
This work was supported by the Nihon Pharmaceutical University Research Grant received in 2024.
- Received March 27, 2025.
- Revision received April 10, 2025.
- Accepted April 11, 2025.
- Copyright © 2025 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
