Abstract
Background/Aim: The systemic administration of anticancer drugs may cause ocular adverse reactions (OARs). However, such adverse events are generally rare and occur with an unknown frequency. This study aimed to investigate the tendency of occurrence of OARs induced by systemic anticancer drugs using a large spontaneous pharmacovigilance database in Japan. Patients and Methods: The safety signals for eight OARs (periorbital and eyelid, conjunctival, corneal, scleral, lacrimal, lens, retinal, and optic nerve disorders) and their associations with anticancer drugs were evaluated by analyzing reporting odds ratios (RORs) and information components (ICs) based on data from the Japanese Adverse Drug Event Report (JADER). Results: Safety signals associated with anticancer drugs were detected for periorbital and eyelid disorders (imatinib), conjunctival disorders (imatinib and lapatinib), corneal disorders (S-1, erlotinib, capecitabine, cetuximab, gefitinib, vandetanib, trastuzumab emtansine, lapatinib), lacrimal disorders (S-1, pembrolizumab), lens disorders (lenalidomide, pomalidomide, elotuzumab, tamoxifen, bexarotene, venetoclax), retinal disorders (encorafenib, binimetinib, tamoxifen, nab-paclitaxel, trametinib, dabrafenib), and optic nerve disorders (tamoxifen and blinatumomab). Some anticancer drugs showed differences in safety signals based on sex and age. Conclusion: Safety signals indicative of the risk of occurrence of OARs were observed for several anticancer drugs, and several hitherto unreported ocular adverse events requiring caution were also detected. Our results will help predict the occurrence of OARs by oncologists, ophthalmologists, pharmacists, and other healthcare professionals.
- Anticancer drug
- ocular adverse reactions
- Japanese Adverse Drug Event Report
- JADER
- safety signal
- pharmacovigilance database
Certain anticancer agents induce ocular adverse reactions (OARs), such as periorbital and eyelid disorders, conjunctival disorders, corneal disorders, scleral disorders, lens disorders, and retinal disorders, optic nerve disorders, lacrimal disorders (1, 2). Some well know OARs are corneal and lacrimal disorder caused by S-1 (2-6), and retinal disorder caused by tamoxifen (7, 8). Novel molecular target anticancer drugs such as B-RAF and MEK inhibitors have also been reported to cause OARs such as uveitis, dry eyes, conjunctival disorders, and retinopathy (9-12). Most OARs may be without discontinuation of anticancer therapy. However, it is important to be aware of irreversible, life-threatening, and vision-threatening OARs (1). Thus, to predict and manage OARs, it is important to know the risk for OARs, the involved eye region, and patient characteristics. However, as drug-induced OARs are generally rare or occur with an unknown frequency in adults and children, there is inadequate evidence to guide patient screening for the potential risks of anticancer drugs (13). Furthermore, information on OARs for recently approved anticancer drugs is limited to that identified in the early stages of development, and in clinical trial reviews/case reports (1).
In recent years, a large pharmacovigilance database of real-world spontaneous adverse drug reactions (ADRs) has been used to evaluate drug-associated safety signals (14). The Japanese Adverse Drug Event Report (JADER) is a nationwide database of spontaneous reports of ADR published by the Pharmaceuticals and Medical Devices Agency (PMDA), a pharmaceutical regulatory authority in Japan. The calculation of reporting odds ratio (RORs) and information components (ICs) (15-18) based on safety signal data from JADER allows for the investigation of anticancer drug-induced OARs.
Herein, we investigated the safety signals for OARs in patients systemically administered anticancer drugs to examine the tendency of OARs occurrence, the region of the eye, and patient characteristics, using the JADER database.
Patients and Methods
Data source. Data from the JADER database (open access database) for the period between April 2004 and April 2021 were obtained from the PMDA website (https://www.pmda.go.jp/english/index.html). The JADER dataset consisted of three data tables, as follows: demographic information “demo” table, drug information “drug” table, and adverse reaction information “reac” table, which included 699,701 patients, 3,897,477 cases, and 1,109,327 cases, respectively. The “demo” table included patient demographic data such as sex and age. Patients with blanked/unknown sex or age data in the “demo” table and those with duplicated data in the “drug” and “reac” tables were excluded. The demo table was linked to the “drug” and “reac” tables using the patient identification number of each case. The contribution of the medications to the ADRs was classified into three categories: suspected drug, concomitant drug, and interaction. The “suspected drug” category was extracted in the present study. To evaluate the safety signals for OARs in patients who received anticancer drugs systemically, the anticancer drugs with “oral”, “intravenous”, and “arterial” routes of administration were selected. The data of 622,488 patients were used in the present study (Figure 1). For subgroup analysis, younger patients were defined as those in the “under 10s”, “10s”, “20s”, “30s”, “40s”, “50s”, and “60s” age groups, and elderly patients were defined as those in the “70s”, “80s”, “90s”, and “100s” age groups, as previously reported (19).
Flow diagram of the study.
As per a search for prescription drugs, the PMDA website (https://www.pmda.go.jp/PmdaSearch/iyakuSearch accessed in August 2021) listed 189 anticancer drugs, and all drugs were selected for analysis (Table I).
Anticancer drugs included in the analysis.
Definition of OARs. Eight OARs were extracted from the “reac” table according to the Standardized MedDRA Queries (SMQ), which are groupings of preferred terms (PTs) related to a medical condition defined in the Medical Dictionary for Regulatory Activities (MedDRA 24.1J). The eight OARs were as follows: periorbital and eyelid disorders (SMQ 2000179), 95 PTs; conjunctival disorders (SMQ 20000175), 78 PTs; corneal disorders (SMQ 20000156), 101 PTs; scleral disorders (SMQ 20000182), 26 PTs; lacrimal disorders (SMQ 20000176), 50 PTs; lens disorders (SMQ 20000155), 34 PTs; retinal disorders (SMQ 20000158), 185 PTs; and optic nerve disorders (SMQ 20000148), 39 PTs.
Statistical analysis. In this study, RORs and ICs were used for evaluating safety signals for OARs. The ROR is an ADR signal index and is the OR of reporting a particular ADR versus all other ADRs associated with the target drugs compared to the reporting odds for all other drugs in the database (19, 20). However, the results obtained using RORs might be unreliable when the sample size is small (21). The IC is an ADR signal index of the Bayesian Confidence Propagation Neural Network (BCPNN) analysis calculated based on the Bayesian statistic approach. It can be used to detect ADR signals even in small samples (18, 22). RORs, ICs, and their 95% confidence intervals (CIs) were calculated using a two-by-two contingency table (Table II) and equations as per previous studies as below (18, 19, 22).
Two-by-two contingency table.
ROR equations:
IC equations:
The calculations were performed using Excel for Microsoft 365 (Microsoft Corporation, Redmond, WA, USA). The safety signals for OARs were positive when the lower limit of the 95% CI of the ROR exceeded 1, and that of the IC exceeded 0 (17).
Results
Safety signals for OARs with anticancer drug use. Table III shows the results of the disproportionality analysis with anticancer drugs which showed RORs with the lower limit of the 95% CI >1 and that of IC >0 for each of the OARs. For periorbital and eyelid disorders, safety signals were detected only with imatinib mesilate [ROR=4.61 (2.85-7.46) and IC=1.92 (1.23 to 2.61)]. Safety signals for the following disorders were detected: conjunctival disorders, imatinib mesilate [ROR=3.14 (1.85-5.32 and IC=1.44 (0.69 to 2.19)] and lapatinib tosilate hydrate [ROR=6.13 (2.91-12.94) and IC=1.89 (0.86 to 2.92)]; corneal disorders, eight anticancer drugs as below: S-1 [ROR=12.66 (10.34-15.50) and IC=3.34 (3.04 to 3.63)], erlotinib hydrochloride [ROR=7.32 (5.08-10.55) and IC=2.56 (2.03 to 3.09)], capecitabine [ROR=2.55 (1.50-4.33) and IC=1.19 (0.44 to 1.95)], cetuximab [ROR=2.53 (1.36-4.72) and IC=1.14 (0.26 to 2.02)], gefitinib [ROR=2.42 (1.25-4.66) and IC=1.07 (0.15 to 1.99)], vendetanib [ROR=105.11 (49.88-221.51) and IC=3.05 (2.01 to 4.08)], trastuzumab emtansine [ROR=10.39 (4.29-25.19) and IC=2.01 (0.82 to 3.20)], and lapatinib tosilate hydrate [ROR=4.04 (1.51-10.81) and IC=1.32 (0.02 to 2.62)]. None of the anticancer drugs were associated with safety signals for scleral disorders. For lacrimal disorders, safety signals were detected with S-1 [ROR=41.41 (34.62-49.53) and IC=4.61 (4.36 to 4.86)] and pembrolizumab [ROR=2.45 (1.44-4.16) and IC=1.14 (0.38 to 1.89)]. For lens disorders, safety signals were detected with six anticancer drugs, as below; lenalidomide hydrate [ROR=11.93 (9.59-14.86) and IC=3.27 (2.95 to 3.59)], pomalidomide [ROR=9.59 (6.14-14.99) and IC=2.73 (2.09 to 3.38)], elotuzumab [ROR=11.79 (6.64-20.94) and IC=2.67 (1.85 to 3.48)], tamoxifen citrate [ROR=5.08 (2.27-11.36) and IC=1.67 (0.57 to 2.77)], bexarotene [ROR=31.89 (11.68-87.11) and IC=2.14 (0.81 to 3.47)], and venetoclax [ROR=39.06 (12.18-125.29) and IC=1.89 (0.39 to 3.38)]. For retinal disorders, safety signals were detected with six anticancer drugs, as below: encorafenib [ROR=159.01 (117.11-215.90) and IC=5.35 (4.95 to 5.74)], binimetinib [ROR=170.51 (124.94-232.68) and IC=5.37 (4.97 to 5.77)], tamoxifen citrate [ROR=7.62 (5.37-10.81) and IC=2.62 (2.11 to 3.13)], nab-paclitaxel [ROR=2.67 (1.82-3.91) and IC=1.31 (0.76 to 1.86)], trametinib dimethyl sulfoxide [ROR=8.51 (5.22-13.86) and IC=2.54 (1.84 to 3.24)], and dabrafenib mesilate [ROR=6.03 (3.39-10.73) and IC=2.09 (1.27 to 2.90)]. For optic nerve disorders, safety signals were detected with tamoxifen citrate [ROR=5.62 (2.10-15.06) and IC=1.54 (0.24 to 2.84)] and blinatumomab [ROR=9.32 (2.98-29.13) and IC=1.59 (0.14 to 3.05)].
Reporting odds ratios and information components of anticancer drugs for ocular adverse reactions.
Subgroup analysis of safety signals for OARs with anticancer drug use. To evaluate differences in safety signals based on sex or age (younger/elderly groups), subgroup analyses were performed for anticancer drugs which were reported to be associated with three or more cases of OARs in Table III.
The study population (n=622,448) was classified into the male (n=319,982) and female (n=302,466) groups. Table IV shows the RORs and ICs of the anticancer drugs by sex. In the male group, safety signals were associated with imatinib mesilate for periorbital and eyelid disorders [ROR=7.91 (4.43-14.11) and IC=2.33 (1.51 to 3.15)], pembrolizumab for lacrimal disorders [ROR=3.20 (1.64-6.25) and IC=1.35 (0.41 to 2.29)], bexarotene for lends disorders [ROR=53.25 (16.41-172.76) and IC=1.91 (0.40 to 3.43)], and blinatumomab for optic nerve disorders [ROR=23.39 (7.42-73.75) and IC=1.82 (0.35 to 3.29)]. On the other hand, the following safety signals were detected only in the female group: conjunctival disorders, lapatinib tosilate hydrate [ROR=5.94 (2.81-12.56) and IC=1.86 (0.82 to 2.89)]; corneal disorders, capecitabine [ROR=4.16 (2.39-7.23) and IC=1.73 (0.95 to 2.52)], cetuximab [ROR=4.25 (1.75-10.28) and IC=1.45 (0.26 to 2.64)], and trastuzumab emtansine [ROR=9.26 (3.81-22.49) and IC=1.95 (0.75 to 3.14)]; and tamoxifen citrate was identified for lens disorders [ROR=5.06 (2.25-11.36) and IC=1.66 (0.56 to 2.77)], retinal disorders [ROR=8.76 (6.16-12.47) and IC=2.77 (2.26 to 3.28)], and optic nerve disorders [ROR=4.73 (1.76-12.70) and IC=1.43 (0.12 to 2.73)].
Reporting odds ratios and information components of anticancer drugs for ocular adverse reactions in sex-based subgroups.
The RORs and ICs of the anticancer drugs were then analyzed according to younger (n=368,129) and elderly (n=254,319) groups (Table V). Safety signals pertaining to imatinib mesilate for conjunctival disorders [ROR=4.01 (2.26-7.11) and IC=1.68 (0.87 to 2.49)], vandetanib [ROR=212.73 (97.24-465.35) and IC=3.10 (2.03 to 4.17)], trastuzumab emtansine [ROR=13.12 (4.86-35.38) and IC=1.93 (0.62 to 3.24)], and lapatinib tosilate hydrate [ROR=5.82 (2.17-15.65) and IC=1.56 (0.25 to 2.86)] for corneal disorders, pembrolizumab for lacrimal disorders [ROR=3.86 (1.99-7.50) and IC=1.56 (0.63 to 2.49)], and blinatumomab for optic nerve disorders [ROR=9.35 (2.99-29.28) and IC=1.59 (0.13 to 3.05)] were detected only in younger group. Safety signals associated with tamoxifen citrate for lens disorders [ROR=6.98 (3.11-15.67) and IC=1.90 (0.79 to 3.00)], retinal disorders [ROR=8.38 (5.80-12.12) and IC=2.70 (2.16 to 3.23)], and optic nerve disorders [ROR=5.65 (2.10-15.16) and IC=1.54 (0.24 to 2.84)] were also identified only in younger group. Safety signals associated with capecitabine [ROR=3.29 (1.63-6.64) and IC=1.37 (0.40 to 2.35)] and cetuximab [ROR=3.56 (1.68-7.52) and IC=1.41 (0.38 to 2.44)] were detected only in elderly group.
Reporting odds ratios and information components of anticancer drugs for ocular adverse reactions in younger/elderly subgroups.
Discussion
Our study investigated the risk of occurrence of OARs induced by anticancer drugs based on safety signal data from the JADER pharmacovigilance database in Japan. Safety signals for periorbital, eyelid, and conjunctival disorders were detected with imatinib. Periorbital edema and conjunctival hemorrhage are common OARs with imatinib (23), which is consistent with our results. Lapatinib was also associated with safety signals for conjunctival disorders. This drug is used for breast cancer treatment with capecitabine or aromatase inhibitors, but safety signals were not detected with capecitabine or aromatase inhibitors such as letrozole. Lapatinib inhibits HER2 as well as EGFR; this dual inhibition might contribute to conjunctival disorders, though further investigation is needed. In corneal disorders, tyrosine kinase inhibitors such as the EGFR inhibitors erlotinib, cetuximab, and gefitinib, HER2 inhibitors trastuzumab emtansine and lapatinib, and the multikinase inhibitor vendetanib were associated with safety signals. EGFR has an important role in maintaining and restoring the outermost layer of the cornea, and EGFR inhibitors including vandetanib induce corneal disorders (24, 25). Our results also support that EGFR inhibitors may contribute to corneal disorders. In our study, safety signals for corneal disorders were observed with trastuzumab emtansine (trastuzumab-DM1, a HER2 antibody-cytotoxic conjugate), but not with trastuzumab or trastuzumab deruxtecan. Two published case reports revealed that trastuzumab emtansine-treated patients developed corneal disorders (26, 27). Thus, trastuzumab emtansine might confer a higher risk for corneal disorders compared to trastuzumab or trastuzumab deruxtecan. S-1 and capecitabine also showed safety signals for corneal disorders, in line with previous reports of corneal complications with these drugs (3-6, 28). 5-Fluorouracil, a metabolite of capecitabine and S-1, interferes with nucleic acid synthesis in keratocytes, which causes corneal damage (dependent on the 5-fluorouracil concentration) (29, 30). Among the various pyrimidine analogue drugs, S-1 and capecitabine may particularly affect corneal keratocytes and corneal epithelial cells, possibly due to formulation characteristics that increase the blood concentration of 5-fluorouracil. Safety signals for lacrimal disorders were also detected with S-1 and pembrolizumab. In line with our results, several studies report that S-1 treatment may lead to lacrimal disorders (3-5). Ocular damage is a known immune-related AE (irAE) associated with PD-1/PD-L1 inhibitors (31, 32). Our results suggest that lacrimal disorders are the most likely ocular disorders associated with pembrolizumab use. Safety signals for lens disorders were detected with the immunomodulatory drugs lenalidomide and pomalidomide. Interestingly, several studies reported on the development of cataracts with lenalidomide and pomalidomide. In a phase 2 study for primary vitreoretinal lymphoma, grade 3 cataract events occurred in 6/13 (54.45%) patients treated with a lenalidomide/rituximab and methotrexate regimen; however, the cataracts may have been related to the vitrectomy (33). In Japan, lenalidomide and pomalidomide have been approved for multiple myeloma, and lenalidomide is approved for adult T-cell leukemia lymphoma, myelodysplastic syndrome, follicular lymphoma, and marginal zone lymphoma. Lenalidomide and pomalidomide are used in combination with dexamethasone for treating multiple myeloma. The anti-SLAMF7 monoclonal antibody elotuzumab, which was associated with safety signals for lens disorders, is also used with lenalidomide/dexamethasone for multiple myeloma (34). These drugs may cause cataract and other lens disorders. Tamoxifen, an anti-estrogen, was associated with safety signals for lens, retinal, and optic nerve disorders. The ocular toxicity of tamoxifen has been reported to cause keratopathy (35), crystalline retinopathy and macular edema (7, 36), and optic neuropathy (37, 38). In agreement with the above, our study suggests that tamoxifen may increase the risk of a broad spectrum of ocular disorders. It is reported that bexarotene, a third-generation retinoid X receptor-selective retinoid, induces cataract formation in dogs and rats. Several phase 2/3 trials for refractory/persistent early-stage cutaneous T-cell lymphoma reported that new lens opacity occurred in 2/18 patients (39). In this study, we found a safety signal for lens disorders with bexarotene, suggesting that patients treated with bexarotene should be monitored for lens disorders. The B-cell lymphoma 2 (BCL-2) inhibitor venetoclax is approved for relapsed/refractory chronic lymphocytic leukemia and acute myeloid leukemia, and OARs have not been associated with this drug in previous clinical studies (40, 41). However, we reveal important data linking safety signals for lens disorders with venetoclax use. Combination therapy with BRAF/MEK inhibitors such as encorafenib/bimetinib and dabrafenib/trametinib dimethyl sulfoxide were associated with safety signals for retinal disorders, and this has been previously reported as well (9-12). Nab-paclitaxel was associated with safety signals in retinal disorders, and paclitaxel, but not docetaxel (no cases for retinal disorders were detected) showed a trend associated with safety signals for retinal disorders [ROR=1.41 (1.01-1.96) and IC=0.47 (−0.02 to 0.95)]. It is reported that taxanes such as paclitaxel, nab-paclitaxel, and docetaxel induce macular edema at a low frequency (42, 43). Our results suggest that nab-paclitaxel and paclitaxel may increase the risk of retinal damage. In our study, blinatumomab (anti-CD3/CD19 antibody) was also associated with safety signals for optic nerve disorders. Blinatumomab, which has been approved for relapsed/refractory B-cell acute lymphoblastic leukemia, caused neurological events such as tremors, dizziness, mental confusion, and encephalopathy in 98/189 patients (52%) (44). Although there have been few reports of optic nerve disorders associated with blinatumomab, our results revealed that safety signals of optic neuropathy were associated with blinatumomab; this may be related to neurotoxicity caused by blinatumomab.
This study also analyzed sex- and age-related differences in the safety signals for OARs. The safety signals associated with HER2 inhibitors/antibodies used for breast cancer such as lapatinib, trastuzumab emtansine, and tamoxifen were detected only in the female group because of the disproportionality of the total ARs. The pyrimidine analogue capecitabine and the EGFR inhibitor cetuximab showed safety signals for corneal disorders among females, while the safety signals associated with the pyrimidine analogue S-1 and EGFR tyrosine kinase inhibitors erlotinib and gefitinib did not differ by sex. Although further investigation is needed, female patients treated with capecitabine and cetuximab should be carefully monitored for corneal disorders. On the other hand, among male patients, imatinib, pembrolizumab, bexarotene, and blinatumomab were associated with safety signals. Sex was not reported to be a risk factor for imatinib-induced OARs in gastrointestinal stromal tumors (GIST) (23), for irAE development within six months of pembrolizumab initiation (45), and for time to first neurological events induced by blinatumomab (44). Although these studies did not assess the risk for the OARs identified in our study, our results suggest that there are sex differences in the applicable OARs for these drugs. Imatinib, vandetanib, trastuzumab emtansine, lapatinib, pembrolizumab, tamoxifen, and blinatumomab were associated with safety signals in younger patients, while capecitabine and cetuximab were associated with safety signals in elderly patients. For trastuzumab emtansine, lapatinib, tamoxifen, and blinatumomab, the number of cases with total adverse reactions in elderly patients was much smaller than in younger patients, understanding the safety signals of these drugs should be done carefully. As with sex differences, capecitabine but not S-1, and cetuximab but not erlotinib and gefitinib were associated with safety signals for corneal disorders in elderly patients. On the other hand, multikinase inhibitors including vandetanib showed safety signals in younger patients. This may be due to the inhibition of the RET proto-oncogene and VEGFR (related to cell proliferation), but further investigation is needed.
Our study has several limitations. First, anticancer therapy may be administered as monotherapy or as a combination regimen with multiple anticancer drugs. Although the safety signals were evaluated by focusing on the suspect drugs, the influence of concomitantly administered drugs cannot be ruled out. Second, biases including under- or over-reporting and confounding caused by comorbidities cannot be ruled out in this large spontaneous reporting system (16, 19). Third, the number of cases with OARs was small, especially for subgroup analysis. To avoid detection of false-positives, we defined safety signals as those with a significant difference in both RORs and ICs. Future studies with larger sample sizes are needed for a more accurate evaluation of safety signals.
Conclusion
Safety signals indicative of the risk of OARs were observed for several anticancer drugs; several new OARs requiring caution were also detected. Our results may help oncologists, ophthalmologists, and pharmacists to predict the occurrence of OARs.
Acknowledgements
The Authors would like to thank Editage (https://www.editage.com/) for editing and reviewing this manuscript for English language.
Footnotes
Authors’ Contributions
JT, TK and SY designed this study. JT carried out the survey of the JADER database. JT and TK performed the statistical analyses. JT, TK, MK, YI and SY drafted the manuscript. All Authors approved the final manuscript.
Conflicts of Interest
The Authors declare that they have no conflicts of interest.
- Received May 31, 2022.
- Revision received July 8, 2022.
- Accepted July 11, 2022.
- Copyright © 2022 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).