Abstract
Background
In addition to cardiotoxicity, ocular toxicity induced by chemotherapeutic agents is not uncommon.
Objective
This study aimed to explore the association between ocular adverse events and major adverse cardiovascular events (composite endpoint) caused by chemotherapy, and whether specific ocular events could be potential predictors of some specific components of the composite endpoint.
Methods
A total of 5378 newly diagnosed patients (age > 18 y/o) with any malignancy or metastatic solid tumors who received chemotherapy from January 1997 to December 2010 were enrolled from the Taiwan National Health Insurance Research Database. Patients who developed new incident ocular diseases were classified as the study group, and those who did not develop incident ocular diseases as the control group.
Results
After propensity score matching, there was a significant increase in the incidence of stroke in the ocular diseases group compared to the no ocular diseases group (13.4% vs. 4.5%, p < 0.0001). Tear film insufficiency, keratopathy, glaucoma, and lens disorders were associated with a significantly higher risk of stroke. A longer duration of methotrexate and a longer duration with higher total amount of tamoxifen were associated with both incident ocular diseases and incident stroke. Cox proportional hazards regression showed that the only independent risk factor for stroke was incident ocular diseases [Adjusted relative risk (95% confidence interval): 2.96 (1.66-5.26), p = 0.0002]. In addition, incident ocular disease was the most significant risk factor compared with other traditional cardiovascular risk factors.
Conclusions
Incident ocular diseases related to chemotherapy were associated with a significantly higher risk of stroke.
Keywords: Chemotherapy, Methotrexate, Stroke, Tamoxifen, Tear film insufficiency
Abbreviations
ARR, Adjusted relative risk
cDDD, Cumulative defined daily dose
CI, Confidence intervals
CV, Cardiovascular
DDD, Defined daily dose
Gb-3, Globotriaosylceramide
HR, Hazard ratio
IQR, Interquartile range
MDR, Multidrug resistance
MTX, Methotrexate
NHI, National Health Insurance
NHIRD, National Health Insurance Research Database
RR, Relative risk
95% CI, 95% confidence interval
INTRODUCTION
Despite great progress in the systemic treatment of solid tumors in past years, these treatments offer only limited benefits in the majority of patients while exposing them to the very real possibility of toxicity from such management.1 In addition to cardiotoxicity,2,3 other systemic organ damage driven by the administration of chemotherapeutic agents while treating cancer is well known.4-6 Among them, ocular toxicity induced by chemotherapeutic agents is not uncommon, a common complaint of patients, easily detected by ophthalmologists, and generally not preventable.7,8 Chemotherapy regimens may produce a broad spectrum of ocular disorders such as dry eye, keratitis, lens disorders, retinopathy, and optic neuropathy.7,9-12 Some ocular symptoms can be warning signs of specific cardiovascular diseases.13,14 An interesting feature of cardiovascular disease, taking atherosclerosis as an example, is that disease progress is not only present in the body, but can also occur in the eyes.15 Individuals with changes in ocular blood vessels have been reported to be at a higher risk of heart attack, heart failure or stroke.16 In addition, examining the eyes is an easy and non-invasive method for doctors to check the condition of the vascular system. However, to date, little research has investigated whether ocular changes may be a potential predictor of specific treatment-related composite endpoints including major adverse cardiovascular events. To the best of our knowledge, few studies have explored the time sequence and the associations of incident ocular diseases with incident cardiovascular (CV) events caused by chemotherapy. Furthermore, if certain ocular diseases occur earlier than the CV events and their associations are significant, it is unclear whether these incident ocular diseases could potentially predict certain CV events in the future. In addition, it is also important to investigate which chemotherapy drugs (including cumulative dose and treatment duration) are associated with these eye and CV associations. Hence, this study aimed to explore these issues.
METHODS
Data resource
The National Health Insurance (NHI) program in Taiwan was initiated in March 1995, and it currently covers more than 95% of the population, making it one of the biggest resources for health research in the world. Information available from the National Health Insurance Research Database (NHIRD) includes patient demographics, disease diagnoses, contracted medical care institutions, medical expenditure, and prescription claims data. To ensure anonymity, the patients’ identification is encrypted and investigators are only permitted to perform data linkage, processing and statistical analyses on an assigned computer in a strictly monitored room. Using an encrypted personal identifier for each patient, researchers can connect files to acquire socio-demographic data, longitudinal medical history and other information. Researchers are only allowed to carry out statistical analysis. This study was approved by the Institutional Review Board (Approval No. KMUHIRB-E(1)-20220212) at Kaohsiung Medical University Hospital.
Study design
A sample of 5378 patients who had any malignancy or metastatic solid tumor and who received chemotherapy (cisplatin, carboplatin, methotrexate, etoposide, tamoxifen, carmustine, pentostatin, leuprolide, vemurafenib, gefitinib) were selected from the NHIRD. The exclusion criteria were: 1) age at cohort entry < 18 years, 2) first chemotherapy drug use after Jan. 1, 2009, 3) undergoing radiotherapy prior to chemotherapy, 4) diagnosis of an ocular disease before chemotherapy, 5) fewer than three clinical visits, and 6) < 2 years of follow-up. Patients with new incident ocular diseases identified according to the following ICD-9-CM codes were defined as the study group: retinopathy: 362.x, 363.x, 379.23 and 379.24; uveitis 364.00, 364.01, 364.10, and 364.24; glaucoma 365.x; tear film insufficiency 375.15; keratopathy 370.x and 371.x; lens disorders 366.x and 379.x; optic neuropathy 377.x. The patients without incident ocular diseases were defined as the control group. The composite endpoint was defined as including myocardial infarction (ICD-9-CM 410.x, 412.x), congestive heart failure (ICD-9-CM 428.x), stroke or cerebrovascular disease (ICD-9-CM 430.x-438.x), dementia (ICD-9-CM 290.x), hemiplegia or paraplegia (ICD-9-CM 342.x, 344.1), renal disease (ICD-9-CM 582.x, 583.0-583.7, 585, 586, 588.x), end-stage renal disease (ICD-9-CM 585 plus ICD-9-CM procedure codes 58001C, 58002C, and 580011A), peripheral vascular disease (ICD-9-CM 443.9, 441.x, 785.4, V43.4, plus ICD-9-CM procedure code 38.48).
Statistical analysis
Data of continuous and categorical variables were analyzed using the t-test or Wilcoxon rank sum test and chi-square test to compare the data of chemotherapy patients with and without ocular diseases. Continuous data were presented as mean (standard deviation) or median [interquartile range (IQR)], and categorical data were presented as percentages. Adjusted relative risk (ARR) with 95% confidence interval (95% CI) for ocular diseases with the risk of incident composite endpoint and its individual components (see Table 2) were calculated after adjusting for age on chemotherapy, sex, hypertension, diabetes mellitus and radiotherapy using a Cox proportional hazards model. A propensity analysis for stroke or dementia was performed through logistic regression to obtain a 2-digit match of the propensity score for each patient with the covariates, such as chemotherapy age group, sex, hypertension, diabetes mellitus, and radiotherapy region (1:1 matching for chemotherapy patients with ocular diseases and without ocular diseases). We estimated the cumulative incidence of stroke in the patients with and without ocular diseases separately by fitting Kaplan-Meier curves and comparing them using the log-rank test. A Cox proportional hazards regression model was used to calculate hazard ratio (HR). The matched data set of age group, sex, hypertension, diabetes mellitus and radiotherapy were used in the STRATA statement such that each unique value for age group, sex, hypertension, diabetes mellitus and radiotherapy was defined as a stratum. A Cox proportional hazards regression model, including chemotherapy by sex interaction term, was used. Average chemotherapy drug dose (DDD, defined daily dose per day) was calculated as cumulative defined daily dose (cDDD) divided by total drug prescription days. The average chemotherapy drug dose was classified using two approaches: stratifying the chemotherapy exposure by DDD category (0 < DDD ≤ 52.0, 52.0 < DDD ≤ 363.5, 363.5 < DDD ≤ 920.0, DDD > 920.0 per day) according to a quartile method. In addition, the quartile method was used to classify the L02BA01 tamoxifen of average chemotherapy drug dose (0 < DDD ≤ 312.0, 312.0 < DDD ≤ 660.0, 660.0 < DDD ≤ 1274.0, DDD > 1274.0 per day). A Cox proportional hazards regression model, including DDD or tamoxifen DDD by chemotherapy (without/with ocular diseases) interaction term was used. All statistical analyses were performed using SAS software, version 9.4 (SAS Institute, Cary, NC, USA). All statistical tests were two-sided, and a two-tailed p value of < 0.05 was considered significant.
RESULTS
Patients
From January 1997 to December 2010, 5378 patients with any newly diagnosed malignancy or metastatic solid tumor who received chemotherapy (ICD-9 codes 140.x-172.x, 174.x-195.8, 200.x-208.x, 196.x-199.1) were identified from the NHIRD as the study population. After excluding patients with the first chemotherapy drug use after Jan. 1, 2009 or with < 2 years of follow-up (n = 1520), who received radiotherapy before chemotherapy (n = 476), diagnosed with incident ocular diseases before chemotherapy (n = 719), and with < 3 clinic visits during follow-up (n = 431), 483 cancer patients who had received chemotherapy and developed incident ocular diseases and 1651 cancer patients who had received chemotherapy but without developing incident ocular diseases were identified (Figure 1). No subjects were lost to follow-up in this cohort during the 14-year study period.
Figure 1.
Study flow.
Baseline characteristics
The patients with ocular diseases were significantly older than those without ocular diseases, and they had a significantly higher prevalence of hypertension and diabetes mellitus, but significantly fewer received radiotherapy. Significantly less cisplatin and epirubicin but more etoposide and leuprorelin were found in the chemotherapy regimens in the patients with ocular diseases (Table 1).
Table 1. Characteristics of chemotherapy patients with ocular diseases and without ocular diseases.
| Chemotherapy with ocular diseases (n = 483) | Chemotherapy without ocular diseases (n = 1651) | p value | |
| Age, mean (SD) years | 58.05 (10.9) | 48.22 (11.7) | < 0.0001* |
| Sex (males), n (%) | 97 (20.1) | 329 (19.9) | 0.9401 |
| Hypertension, n (%) | 300 (62.1) | 505 (30.6) | < 0.0001* |
| Diabetes mellitus, n (%) | 180 (37.3) | 232 (14.1) | < 0.0001* |
| Radiotherapy, n (%) | 95 (19.7) | 458 (27.7) | 0.0004* |
| Chemotherapy drug | |||
| Cisplatin, n (%) | 25 (5.2) | 171 (10.4) | 0.0005* |
| Carboplatin, n (%) | 5 (1.0) | 18 (1.1) | 0.9179 |
| Methotrexate, n (%)* | 117 (24.2) | 355 (21.5) | 0.2050 |
| Etoposide, n (%) | 1 (0.2) | 33 (2.0) | 0.0057* |
| Tamoxifen, n (%) | 301 (62.3) | 983 (59.5) | 0.2725 |
| Carmustine, n (%) | 1 (0.2) | 1 (0.1) | 0.3548 |
| Leuprorelin, n (%) | 16 (3.3) | 27 (1.6) | 0.0210* |
| Gefitinib, n (%) | 4 (0.8) | 14 (0.9) | 0.9666 |
| Doxorubicin, n (%) | 62 (12.8) | 247 (15.0) | 0.2433 |
| Epirubicin, n (%) | 111 (23.0) | 504 (30.5) | 0.0013* |
| Any malignancy, n (%) | 483 (100.0) | 1651 (100.0) | 1.0000 |
| ICD-9 codes: 140.x-173.x | 141 (29.2) | 442 (26.8) | 0.2936 |
| ICD-9 codes: 174.x-195.x | 412 (85.3) | 1352 (81.9) | 0.0816 |
| ICD-9 codes: 200.x-208.x | 29 (6.0) | 128 (7.8)0 | 0.1954 |
| Metastatic solid tumor, n (%) | |||
| ICD-9 codes: 196.x-199.x | 145 (30.0) | 575 (34.8) | 0.0494* |
* p < 0.05. Hypertension and diabetes mellitus were defined as ≥ 3 clinic visits. Data of continuous and categorical variables were analyzed using t-test and chi-square test to compare the data of chemotherapy patients with ocular diseases and without ocular diseases.
SD, standard deviation.
Endpoint during the 14-year follow-up period
The mean follow-up periods were 4.19 ± 3.11 and 7.00 ± 3.59 years in the patients with and without ocular diseases, respectively. During the follow-up period, there was no significant difference in the composite endpoint between the two groups, with 483 in the ocular diseases group and 1651 in the no ocular diseases group and the adjusted relative risk (RR) with 95% CI after adjusting for age on chemotherapy, sex, hypertension, diabetes mellitus and radiotherapy was 1.18 (0.88-1.58) (p = 0.2682) (Table 2). However, there were significant increases in the incidence rates of stroke and dementia (two individual components of the composite endpoint) in the ocular diseases group compared to the no ocular diseases group, with ARRs (95% CIs) of 1.81 (1.21-2.71) (p = 0.0036) and 2.20 (1.07-4.54) (p = 0.0329), respectively (Table 2). In order to clarify the relationships of incident ocular diseases with stroke and dementia, each propensity analysis of the two was performed through logistic regression to obtain a 2-digit match of the propensity score for each patient with the covariates, including chemotherapy age group, sex, hypertension, diabetes mellitus, and radiotherapy region (Table 3 and Supplemental Table 1). After propensity score matching, the incident stroke rate was still significantly increased in the ocular diseases group compared to the no ocular diseases group (13.4% vs. 4.5%, p < 0.0001) (Table 3), but dementia was not (2.8% vs. 1.8%, p = 0.3655) (Supplemental Table 1). The median (IQR) stroke event was 4.6 years (IQR 2.7-7.5) in the ocular diseases group and 5.8 years (IQR 3.3-9.5) in the no ocular diseases group (p < 0.0001). Since incident ocular diseases occurred significantly ahead of incident stroke, we further investigated whether the incident ocular diseases could assess the risk of incident stroke. To clarify the relationship between incident ocular diseases with incident stroke, we used time-to-event analysis with a Cox proportional hazards regression model (Figure 2). The results showed that the ocular diseases group had a significantly higher risk of developing stroke [adjusted HR: 4.23 (95% CI: 2.29-7.81), p < 0.0001]. Since there was a trend of more males in the no ocular diseases group for stroke, another Cox proportional hazards regression model regarding sex was analyzed, which showed significantly higher risks of developing incident stroke in both male and female groups [adjusted HR: 4.54 (95% CI: 1.69-12.16), p = 0.0026 in males and 4.04 (1.85-8.82), p = 0.0005 in females] (Table 4).
Table 2. The association of the chemotherapy related incident ocular diseases with the risk of incident composite endpoint (major adverse cardiovascular events, including all individuals), and its individual components.
| Ocular diseases (n = 483) | No ocular diseases (n = 1651) | p value | |
| Composite endpoints | |||
| No. of events (%)/total patients | 72 (36.36)/198 | 224 (20.70)/1082 | |
| Crude RR (95% CI) | 1.76 (1.35-2.29) | 1.00 | < 0.0001* |
| Adjusted RR (95% CI) | 1.18 (0.88-1.58) | 1.00 | 0.2682 |
| Prior event or clinic visits < 3 | 285 | 569 | |
| Myocardial infarction | |||
| No. of events (%)/total patients | 2 (0.42)/474 | 4 (0.24)/1639 | |
| Crude RR (95% CI) | 1.73 (0.32-9.44) | 1.00 | 0.5273 |
| Adjusted RR (95% CI) | 0.62 (0.11-3.53) | 1.00 | 0.5854 |
| Prior event or clinic visits < 3 | 9 | 12 | |
| Congestive heart failure | |||
| No. of events (%)/total patients | 16 (3.78)/423 | 26 (1.63)/1592 | |
| Crude RR (95% CI) | 2.32 (1.24-4.32) | 1.00 | 0.0082* |
| Adjusted RR (95% CI) | 0.92 (0.48-1.76) | 1.00 | 0.8013 |
| Prior event or clinic visits < 3 | 60 | 59 | |
| Stroke (cerebrovascular disease) | |||
| No. of events (%)/total patients | 50 (13.44)/372 | 61 (3.99)/1528 | |
| Crude RR (95% CI) | 3.37 (2.32-4.89) | 1.00 | < 0.0001* |
| Adjusted RR (95% CI) | 1.81 (1.21-2.71) | 1.00 | 0.0036* |
| Prior event or clinic visits < 3 | 111 | 123 | |
| Dementia | |||
| No. of events (%)/total patients | 18 (3.89)/463 | 13 (0.80)/1632 | |
| Crude RR (95% CI) | 4.88 (2.39-9.96) | 1.00 | < 0.0001* |
| Adjusted RR (95% CI) | 2.20 (1.07-4.54) | 1.00 | 0.0329* |
| Prior event or clinic visits < 3 | 20 | 19 | |
| Hemiplegia or paraplegia | |||
| No. of events (%)/total patients | 3 (0.63)/479 | 3 (0.18)/1642 | |
| Crude RR (95% CI) | 3.43 (0.69-16.98) | 1.00 | 0.1313 |
| Adjusted RR (95% CI) | 1.64 (0.3-9.12) | 1.00 | 0.5717 |
| Prior event or clinic visits < 3 | 4 | 9 | |
| Renal Disease | |||
| No. of events (%)/total patients | 24 (5.87)/409 | 46 (2.96)/1552 | |
| Crude RR (95% CI) | 1.98 (1.21-3.24) | 1.00 | 0.0067* |
| Adjusted RR (95% CI) | 1.11 (0.66-1.88) | 1.00 | 0.6851 |
| Prior event or clinic visits < 3 | 74 | 99 | |
| End-stage renal disease | |||
| No. of events (%)/total patients | 4 (0.83)/481 | 7 (0.43)/1644 | |
| Crude RR (95% CI) | 1.95 (0.57-6.67) | 1.00 | 0.2855 |
| Adjusted RR (95% CI) | 1.14 (0.31-4.23) | 1.00 | 0.8464 |
| Prior event or clinic visits < 3 | 2 | 7 | |
| Peripheral vascular disease | |||
| No. of events (%)/total patients | 5 (1.10)/456 | 12 (0.75)/1604 | |
| Crude RR (95% CI) | 1.47 (0.52-4.16) | 1.00 | 0.4722 |
| Adjusted RR (95% CI) | 0.69 (0.23-2.05) | 1.00 | 0.4995 |
| Prior event or clinic visits < 3 | 27 | 47 |
* p < 0.05. Adjusted relative risk (RR) with 95% confidence intervals (CI) and their p values were calculated after adjustment for age on chemotherapy, sex, hypertension, diabetes mellitus and radiotherapy by using Cox proportional hazards regression model.
Note: In order not to sacrifice some temporal related specific event regarding to chemotherapy with ocular diseases (e.g., stroke), the exclusions for each endpoint of prior events or clinic visits < 3 were performed in this step.
Table 3. Characteristics of chemotherapy patients with ocular diseases and matched without ocular diseases.
| Ocular diseases | No ocular diseases | p value | |
| N | 358 | 358 | |
| Stroke events, n (%) | 48 (13.4) | 16 (4.5) | < 0.0001* |
| Chemotherapy age, mean (SD) years | 56.52 (10.9) | 55.19 (13.3) | 0.1425 |
| Chemotherapy age ≥ 55 years, n (%) | 202 (56.4) | 178 (49.7) | 0.0723 |
| Sex, males, n (%) | 63 (17.6) | 84 (23.5) | 0.0520 |
| Hypertension, n (%) | 197 (55.0) | 184 (51.4) | 0.3302 |
| Diabetes mellitus, n (%) | 116 (32.4) | 103 (28.8) | 0.2917 |
| Radiotherapy, n (%) | 73 (20.4) | 82 (22.9) | 0.4141 |
| Metastatic solid tumor, n (%) | 109 (30.4) | 120 (33.5) | 0.3781 |
| Comorbidities, n (%) | |||
| Myocardial Infarction | 2 (0.6) | 1 (0.3) | 0.5629 |
| Congestive heart failure | 9 (2.5) | 16 (4.5) | 0.1541 |
| Dementia | 11 (3.1) | 5 (1.4) | 0.1293 |
| Hemiplegia or paraplegia | 1 (0.3) | 1 (0.3) | 1.0000 |
| Renal disease | 16 (4.5) | 13 (3.6) | 0.5695 |
| End-stage renal disease | 3 (0.8) | 2 (0.6) | 0.6536 |
| Peripheral vascular disease | 3 (0.8) | 4 (1.1) | 0.7041 |
| Peptic ulcer disease | 40 (11.2) | 37 (10.3) | 0.7174 |
* p < 0.05. Comorbidities were defined as ≥ 3 clinic visits. Data of continuous and categorical variables were analyzed using t-test and chi-square test to compare the data of chemotherapy patients with ocular diseases and without ocular diseases. A propensity analysis was performed through logistic regression to obtain a 2-digit match of the propensity score for each patient with the covariates, such as chemotherapy age group, sex, hypertension, diabetes mellitus, and radiotherapy region (1:1 for chemotherapy patients with ocular diseases and without ocular diseases). SD, standard deviation.
Figure 2.
Cumulative incidences of stroke between chemotherapy patients with and without incident ocular diseases.
Table 4. Association of chemotherapy with ocular diseases with risk of incident stroke in terms of gender.
| Chemotherapy | Stroke events/total subjects, % | Person-years | Events per 1000 person-years (95% CI) | HR (95% CI) | p value | p for interaction* |
| Men | ||||||
| Without ocular diseases | 7/84, 8.33 | 551.64 | 12.69 (11.67-13.79) | 1.00 | ||
| With ocular diseases | 14/63, 22.22 | 299.58 | 46.73 (41.73-52.34) | 4.54 (1.69-12.16) | 0.0026* | |
| Women | ||||||
| Without ocular diseases | 9/274, 3.28 | 1828.86 | 4.92 (4.70-5.15) | 1.00 | ||
| With ocular diseases | 34/295, 11.53 | 1546.82 | 21.98 (20.91-23.10) | 4.04 (1.85-8.82) | 0.0005* | |
| Combined group | ||||||
| Without ocular diseases | 16/358, 4.47 | 2380.5 | 6.72 (6.46-7.00) | 1.00 | ||
| With ocular diseases | 48/358, 13.41 | 1846.4 | 26.00 (24.84-27.21) | 4.23 (2.29-7.81) | < 0.0001* | 0.9396 |
* p < 0.05. Hazard ratios (HRs) with 95% confidence intervals (CI) and their p values were calculated by using Cox proportional hazards regression model. Age group, sex, hypertension, diabetes mellitus and radiotherapy were used in the STRATA statement such that each unique value for age group, sex, hypertension, diabetes mellitus and radiotherapy defines a stratum. Cox proportional hazards regression model, including chemotherapy × sex interaction term, was applied.
The sub-phenotypes of incident ocular diseases associated with an increased risk of stroke
Adjusted RR and p values were calculated according to the ocular disease phenotypes compared to those without ocular diseases and adjusted for chemotherapy, age, sex, hypertension, diabetes mellitus and radiotherapy using a Cox proportional hazards regression model. Tear film insufficiency, followed by keratopathy, glaucoma, and lens disorders were associated with significantly higher risks of stroke, with RRs and 95% CIs of 5.08 (2.47-10.44), 4.27 (1.53-11.93), 3.72 (1.23-11.19) and 2.98 (1.66-5.35), respectively (all p < 0.05) (Supplemental Table 2).
Comparison of chemotherapy drug administration patterns and stroke rate in the patients with and without incident ocular diseases
Supplemental Table 3 shows the chemotherapy drug administration patterns in the patients with and without incident ocular diseases. The total dosage of methotrexate was the same between the two groups. However, the duration of usage was significantly longer in the ocular diseases group than in the no ocular diseases group [median (IQR), 90.0 (42.0-284.0) days vs. 63.0 (28.0-140.0) days, p = 0.0446]. As to tamoxifen, the average daily dosage (total dosage/total dosage use days) was the same in both groups. However, the duration and, therefore, the total dosage were significantly longer and higher in the ocular diseases group than in the no ocular diseases group [median (IQR), 779.0 (448.0-1370.0) days vs. 532.0 (252.0-904.0) days, and 17000.0 (8960.0-28120.0) mg vs. 10700.0 (5380.0-18880.0), both p < 0.0001]. Furthermore, analysis of the average dosage effect of tamoxifen to stroke by DDD showed no significant dose effect (Supplemental Table 4). Therefore, tamoxifen toxicity was found to be duration dependent but not dosage dependent.
Incident ocular diseases but not chemotherapy were associated with the occurrence of stroke
To investigate whether incident ocular diseases, chemotherapy with methotrexate and/or tamoxifen or both were independent risk factors associated with stroke, further Cox proportional hazards regression analysis was performed adjusting for chemotherapy, age, sex, hypertension, diabetes mellitus and radiotherapy (Supplemental Table 5-1). In Supplemental Table 5-2, neither methotrexate nor tamoxifen was a risk for stroke. Only incident ocular diseases were associated with the risk of stroke, with an ARR and 95% CI of 2.96 (1.66-5.26) (p = 0.0002). In addition, incident ocular diseases with or without methotrexate/tamoxifen both significantly increased the risk of stroke [adjusted HR (95% CI) 5.91 (2.08-16.77), p = 0.0009 in those without methotrexate/tamoxifen and 4.03 (1.43-11.38), p = 0.0084 in those with methotrexate/tamoxifen] (Table 5) [ARR (95% CI) 4.19 (1.49-11.80, p = 0.0067 in those without methotrexate/tamoxifen and 3.75 (1.37-10.26), p = 0.0099 in with those methotrexate/tamoxifen] (Supplemental Table 5-1). After adjusting for the baseline risk factors, chemotherapy and radiotherapy, incident ocular diseases for stroke, age (chemotherapy at an age of ≥ 55 y/o), hypertension and diabetes mellitus were still risk factors for stroke with ARRs and 95% CIs of 2.96 (1.66-5.26, p = 0.0002), 2.47 (1.29-4.72, p = 0.0061), 2.45 (1.27-4.72, p = 0.0073), and 1.73 (1.04-2.87, p = 0.0339), respectively. Among them, incident ocular diseases were the highest risk (Supplemental Table 5-2).
Table 5. Ocular diseases predicted risk of incident stroke.
| Stroke events (%) | Total subjects | Crude HR (95% CI) | p value | Adjusted HR (95% CI) | p value | |
| Methotrexate | ||||||
| No | 48 (8.6) | 559 | 1.00 | 1.00 | ||
| Yes | 16 (10.2) | 157 | 1.15 (0.65-2.02) | 0.6377 | 1.13 (0.64-2.00) | 0.6784 |
| Tamoxifen | ||||||
| No | 30 (10.3) | 290 | 1.00 | 1.00 | ||
| Yes | 34 (8.0) | 426 | 0.71 (0.43-1.15) | 0.1654 | 0.93 (0.47-1.82) | 0.8247 |
| Methotrexate/Tamoxifen | ||||||
| No | 18 (9.4) | 191 | 1.00 | 1.00 | ||
| Yes | 46 (8.8) | 525 | 0.79 (0.46-1.36) | 0.3945 | 1.03 (0.52-2.03) | 0.9297 |
| Ocular diseases | ||||||
| No | 16 (4.5) | 358 | 1.00 | 1.00 | ||
| Yes | 48 (13.4) | 358 | 4.30 (2.37-7.78) | < 0.0001 | 3.99 (2.20-7.23) | < 0.0001 |
| Methotrexate/Tamoxifen and Ocular diseases | ||||||
| No and No | 5 (4.3) | 115 | 1.00 | 1.00 | ||
| Yes and No | 11 (4.5) | 243 | 0.84 (0.29-2.45) | 0.7549 | 1.20 (0.39-3.72) | 0.7559 |
| No and Yes | 13 (17.1) | 76 | 6.46 (2.29-18.27) | 0.0004 | 5.91 (2.08-16.77) | 0.0009 |
| Yes and Yes | 35 (12.4) | 282 | 3.35 (1.30-8.58) | 0.012 | 4.03 (1.43-11.38) | 0.0084 |
Adjusted hazard ratio (HR) and p value were calculated after adjustment for chemotherapy age, sex, hypertension, diabetes mellitus and radiotherapy by using Cox proportional hazards regression model. CI, confidence interval.
DISCUSSION
To the best of our knowledge, this is the first and biggest observational study to explore the associations between incident ocular diseases caused by chemotherapy and the risk of CV events. Among those CV events, the ocular diseases group was significantly associated with a higher risk of developing stroke. As to the phenotypes of ocular diseases, tear film insufficiency, followed by keratopathy, glaucoma, and lens disorders were associated with significantly higher risks of stroke. Comparing the chemotherapy drugs and their patterns of usage, only a longer duration of methotrexate was associated with both incident ocular diseases and incident stroke after propensity score matching for baseline characteristics. However, tamoxifen was significantly associated with both incident ocular diseases and incident stroke in terms of both duration and total amount of usage. We further analyzed chemotherapy with methotrexate and tamoxifen and incident ocular diseases with regards to the risk of stroke, and the results showed that the phenotype of incident ocular diseases was the only risk for stroke but not methotrexate and tamoxifen. In addition, the phenotype of incident ocular diseases had the highest risk compared to traditional risks including age, hypertension and diabetes mellitus.
Chemotherapy-related incident ocular diseases
Among the phenotypes of ocular diseases, tear film insufficiency, keratopathy, glaucoma, and lens disorders were most strongly associated with the risk of stroke. These phenotypes have also been reported after chemotherapy.9,17 In addition, a longer duration of methotrexate and both a longer duration and subsequent increased total amount of tamoxifen were associated with both incident ocular diseases and incident stroke. Methotrexate has been reported to have the side effect of decreased tears.9 Tamoxifen has been associated with an increased rate of keratopathy18,19 and lens disorders.17,20,21 Mitomycin has been associated with an increased rate of glaucoma,22 but few studies have reported on glaucoma with methotrexate and tamoxifen. In addition, steroids are included in many chemotherapy regimens, and steroid-induced glaucoma should also be taken into consideration.23
Chemotherapy-related incident stroke
Stroke can occur in a variety of tumor-related conditions, including direct invasion, coagulopathy, chemotherapy side effects and nonbacterial thrombotic endocarditis.24 Among the chemotherapy drugs, tamoxifen is well known to increase the risk of both stroke and venous thrombosis, especially in women.25,26 In addition, tamoxifen has been associated with an increased risk of atrial fibrillation.27 Methotrexate has also been associated with an increased risk of stroke.28 In addition, methotrexate has also been reported to provoke new-onset atrial fibrillation.29 In patients with solid tumors, coagulation is mildly activated, and cancer treatment can further enhance the formation of systemic and cerebral arterial or venous thrombosis.30 In addition, cancer and atrial fibrillation are both independent risk factors for ischemic stroke, and systemic inflammation and autonomic dysregulation are thought to play critical roles.31
The association between phenotypes of ocular diseases and stroke
The cancer patients receiving chemotherapy with incident ocular diseases had a significantly higher risk of developing stroke with an adjusted HR and 95% CI of 4.23 (2.29-7.81) (p < 0.0001), and the curves separated very early within 1 year. In addition to the cumulative incidence rate of 8.76% within 3 years, the individual cumulative incidence rates were 1.74% in first year, 3.93% in second year, and 8.76% in third year, which showed the novel findings in the study – that the phenotype of ocular diseases is a strong potential predictor of stroke (Figure 2), possibly through inflammation-related atherothrombotic effects due to the stroke endpoint. For the possible mechanisms or connections of specific sub-phenotypes of ocular diseases with stroke, tear film insufficiency, or dry eye disease, has been more frequently associated with comorbidities of not only stroke, per se, but also of stroke-related risk factors including ischemic heart disease, diabetes mellitus, hyperlipidemia, cardiac arrhythmias, and peripheral vascular disorders.32 In addition, tear film insufficiency has also been significantly associated with solid tumors without metastasis.32 As to the relationship of keratopathy and lens disorders with stroke, Anderson-Fabry disease, a disease of vascular endothelial accumulation of globotriaosylceramide (Gb-3), may suggest a possible link.33 Gb3 is overexpressed in many human tumors and tumor cell lines with inherent or acquired multidrug resistance (MDR). Gb3 is co-expressed and interplays with the membrane efflux transporter P-gp encoded by the MDR1 gene. P-gp can act as a lipid flippase and stimulate Gb3 induction when tumor cells are exposed to cancer chemotherapy.34,35 In addition, patients with stroke have been reported to have a high prevalence of lens disorders.36 Glaucoma has been reported to be a significant risk factor for subsequent stroke.37-39 Abnormal ocular blood flow has also been implicated as a risk factor for both glaucoma and stroke.40
Limitations
The present study had some limitations. First, personal variables, such as smoking habits, alcohol intake, body mass index, and physical activity were not available in the NHIRD. Second, the presence of comorbidities relied on the claims data based on ICD-9-CM diagnosis codes, which could have potentially led to disease misclassification. Therefore, we used hospital admission follow-up records of cardiovascular events with the main ICD-9-CM diagnosis codes as incident events to minimize possible misclassification. Third, blood pressure profiles were not available in the NHIRD, and details of blood pressure control were unknown. Finally, the project has closed, the deadline to access the NHIRD has passed, and no further analysis is allowed. Owing to these limitations, a large-scale, prospective trial is necessary to confirm our findings. Fourth, two issues should be considered when evaluating an observational cohort study. First, loss of follow-up. However, the NHIRD represents > 99% of Taiwan’s population and no one was lost during follow-up in our study. Second, it is often difficult to blind the investigators who are assessing the study outcomes. However, both the exposure status and outcomes of the study participants were assessed by their own doctors who did not participate in or knew about this study, which may have attenuated the influence of bias on the assessment of the outcomes. Finally, there was a female predominance, and therefore it was possible to have skewed cancer types in our study. The advantage of an observational cohort study design is the possibility of assessing incidence rates, relative risks, and causality.
CONCLUSION
Chemotherapy-related incident ocular diseases were associated with a significantly increased risk of stroke in this 14-year follow-up study. Tear film insufficiency, keratopathy, glaucoma, and lens disorders were the leading phenotypes for stroke. A longer duration of methotrexate and a longer duration and higher total amount of tamoxifen were associated with incident ocular diseases. Finally, incident ocular diseases occurring after chemotherapy were associated with the risk of future stroke.
DECLARATION OF CONFLICT OF INTEREST
All authors declare no conflict of interest.
SUPPLEMENTARY MATERIALS
Supplemental Table 1. Chemotherapy with ocular diseases revealed no significant association with incident dementia.
| Ocular diseases | No ocular diseases | p value | |
| N | 433 | 433 | |
| Dementia events, n (%) | 12 (2.8) | 8 (1.8) | 0.3655 |
| Chemotherapy age, mean (SD) years | 57.0 (10.7) | 55.6 (13.8) | 0.0997 |
| Chemotherapy age ≥ 55 years, n (%) | 251 (58.0) | 233 (53.8) | 0.2180 |
| Sex, males, n (%) | 82 (18.9) | 95 (21.9) | 0.2733 |
| Hypertension, n (%) | 255 (58.9) | 231 (53.3) | 0.1003 |
| Diabetes mellitus, n (%) | 146 (33.7) | 129 (29.8) | 0.2146 |
| Radiotherapy, n (%) | 92 (21.2) | 93 (21.5) | 0.9339 |
| Metastatic solid tumor, n (%) | 129 (29.8) | 154 (35.6) | 0.0701 |
| Comorbidities, n (%) | |||
| Myocardial infarction | 2 (0.5) | 1 (0.2) | 0.563 |
| Congestive heart failure | 12 (2.8) | 13 (3.0) | 0.8392 |
| Stroke | 36 (8.3) | 26 (6.0) | 0.1875 |
| Hemiplegia or paraplegia | 2 (0.5) | 2 (0.5) | 1.0000 |
| Renal disease | 19 (4.4) | 20 (4.6) | 0.8698 |
| End-stage renal disease | 3 (0.7) | 3 (0.7) | 1.0000 |
| Peripheral vascular disease | 4 (0.9) | 6 (1.4) | 0.5247 |
| Peptic ulcer disease | 52 (12.0) | 46 (10.6) | 0.5198 |
Comorbidities were defined as ≥ 3 clinic visits. Data of continuous and categorical variables were analyzed using t-test and chi-square test to compare the data of chemotherapy patients with ocular diseases and without ocular diseases. A propensity analysis was performed through logistic regression to obtain a 2-digit match of the propensity score for each patient with the covariates, such as chemotherapy age group, sex, hypertension, diabetes mellitus, and radiotherapy region (1:1 for chemotherapy patients with ocular diseases and without ocular diseases). SD, standard deviation.
Supplemental Table 2. Phenotypes of chemotherapy related incident ocular diseases with their risk of following incident stroke.
| Stroke events/total subjects, % | Crude RR (95% CI) | p | Adjusted RR (95% CI) | p | |
| Chemotherapy without ocular diseases | 16/358, 4.47 | 1.00 | 1.00 | ||
| Chemotherapy with ocular diseases | 48/358, 13.41 | 3.00 (1.70-5.28) | 0.0001 | 3.11 (1.76-5.49) | < 0.0001 |
| Retinopathy | |||||
| No | 41/271, 15.13 | 3.39 (1.90-6.03) | < 0.0001 | 3.20 (1.79-5.72) | < 0.0001 |
| Yes | 3/41, 7.32 | 1.64 (0.48-5.62) | 0.4333 | 2.28 (0.65-7.97) | 0.1966 |
| Other phenotypes | 4/46, 8.70 | 1.95 (0.65-5.82) | 0.2337 | 2.89 (0.95-8.85) | 0.0627 |
| Uveitis | |||||
| No | 48/342, 14.04 | 3.14 (1.78-5.53) | < 0.0001 | 3.19 (1.80-5.64) | < 0.0001 |
| Yes | 0/8, 0.00 | - | - | - | - |
| Other phenotypes | 0/8, 0.00 | - | - | - | - |
| Glaucoma | |||||
| No | 40/306, 13.07 | 2.92 (1.64-5.22) | 0.0003 | 2.98 (1.66-5.35) | 0.0002 |
| Yes | 4/30, 13.33 | 2.98 (1.00-8.92) | 0.0505 | 3.72 (1.23-11.19) | 0.0196 |
| Other phenotypes | 4/22, 18.18 | 4.07 (1.36-12.17) | 0.0121 | 4.22 (1.39-12.8) | 0.0111 |
| Tear film insufficiency | |||||
| No | 20/209, 9.57 | 2.14 (1.11-4.13) | 0.0232 | 2.07 (1.07-4.00) | 0.0307 |
| Yes | 15/89, 16.85 | 3.77 (1.86-7.63) | 0.0002 | 5.08 (2.47-10.44) | < 0.0001 |
| Other phenotypes | 13/60, 21.67 | 4.85 (2.33-10.08) | < 0.0001 | 4.97 (2.38-10.41) | < 0.0001 |
| Keratopathy | |||||
| No | 39/284, 13.73 | 3.07 (1.72-5.50) | 0.0002 | 2.98 (1.66-5.35) | 0.0003 |
| Yes | 5/37, 13.51 | 3.02 (1.11-8.25) | 0.0308 | 4.27 (1.53-11.93) | 0.0056 |
| Other phenotypes | 4/37, 10.81 | 2.42 (0.81-7.24) | 0.1141 | 3.48 (1.15-10.57) | 0.0276 |
| Lens disorders | |||||
| No | 4/79, 5.06 | 1.13 (0.38-3.39) | 0.8234 | 2.79 (0.89-8.79) | 0.0796 |
| Yes | 39/250, 15.60 | 3.49 (1.95-6.25) | < 0.0001 | 2.98 (1.66-5.35) | 0.0003 |
| Other phenotypes | 5/29, 17.24 | 3.86 (1.41-10.53) | 0.0084 | 5.72 (2.05-15.94) | 0.0009 |
| Optic nerve disorders | |||||
| No | 47/345, 13.62 | 3.05 (1.73-5.38) | 0.0001 | 3.15 (1.78-5.57) | < 0.0001 |
| Yes | 0/8, 0.00 | - | - | - | - |
| Other* | 1/5, 20.00 | 4.48 (0.59-33.74) | 0.146 | 3.49 (0.45-26.85) | 0.2292 |
* Stroke events before retinopathy, uveitis, glaucoma, tear film insufficiency, corneal opacity, cataract, or optic nerve and clinic visits < 3.
Adjusted relative risk (RR) and p value were calculated by ocular diseases phenotypes, when compared with no ocular diseases, after adjustment for chemotherapy age, sex, hypertension, diabetes mellitus and radiotherapy by using Cox proportional hazards regression model. CI, confidence interval.
Supplemental Table 3. Comparison of chemotherapy drug administration patterns and stroke rate in patients with and without incident ocular diseases.
| Ocular diseases | No ocular diseases | p value | |
| N | 358 | 358 | |
| Stroke events, n (%) | 48 (13.4) | 16 (4.5) | < 0.0001* |
| L01XA01 cisplatin, n | 22 | 39 | |
| Total clinic visits (frequencies), median (IQR) | 5.0 (3.0-6.0) | 3.0 (2.0-4.0) | 0.0620 |
| Total drug use (days), median (IQR) | 31.0 (15.0-56.0) | 18.0 (11.0-35.0) | 0.0580 |
| Total dosage, median (IQR), mg | 6.0 (4.0-13.0) | 4.5 (3.0-9.0) | 0.6373 |
| Total dosage/total dosage use days | 0.2 (0.1-0.5) | 0.3 (0.1-0.8) | 0.2576 |
| L01XA02 carboplatin, n | 5 | 5 | |
| Total clinic visits (frequencies), median (IQR) | 4.0 (4.0-6.0) | 5.0 (4.0-6.0) | 0.8362 |
| Total drug use (days), median (IQR) | 20.0 (7.0-23.0) | 42.0 (19.0-43.0) | 0.2788 |
| Total dosage, median (IQR), mg | 230.0 (140.0-400.0) | 2700.0 (1800.0-4050.0) | 0.1753 |
| Total dosage/total dosage use days | 11.5 (9.5-20.0) | 64.3 (30.3-180.0) | 0.3235 |
| L01BA01 methotrexate, n | 92 | 65 | |
| Total clinic visits (frequencies), median (IQR) | 9.0 (6.0-12.5) | 9.0 (3.0-12.0) | 0.1638 |
| Total drug use (days), median (IQR) | 90.0 (42.0-284.0) | 63.0 (28.0-140.0) | 0.0446* |
| Total dosage, median (IQR), mg | 300.0 (145.0-610.0) | 300.0 (85.0-450.0) | 0.2909 |
| Total dosage/total dosage use days | 2.3 (1.3-4.3) | 3.6 (1.7-7.1) | 0.0426* |
| L01CB01 etoposide, n | 0 | 6 | |
| Total clinic visits (frequencies), median (IQR) | - | 3.0 (2.0-6.0) | - |
| Total drug use (days), median (IQR) | - | 41.5 (14.0-58.0) | - |
| Total dosage, median (IQR), mg | - | 635.0 (320.0-1500.0) | - |
| Total dosage/total dosage use days | - | 23.9 (15.9-30.0) | - |
| L02BA01 tamoxifen, n | 229 | 197 | |
| Total clinic visits (frequencies), median (IQR) | 27.0 (17.0-50.0) | 20.0 (9.0-33.0) | < 0.0001* |
| Total drug use (days), median (IQR) | 779.0 (448.0-1370.0) | 532.0 (252.0-904.0) | < 0.0001* |
| Total dosage, median (IQR), mg | 17000.0 (8960.0-28120.0) | 10700.0 (5380.0-18880.0) | < 0.0001* |
| Total dosage/total dosage use days | 20.0 (20.0-20.0) | 20.0 (20.0-20.0) | 0.9862 |
| L01AD01 carmustine, n | 1 | 0 | |
| Total clinic visits (frequencies), median (IQR) | 2.0 (2.0-2.0) | - | - |
| Total drug use (days), median (IQR) | 28.0 (28.0-28.0) | - | - |
| Total dosage, median (IQR), mg | 600.0 (600.0-600.0) | - | - |
| Total dosage/total dosage use days | 21.4 (21.4-21.4) | - | - |
| L02AE02 leuprorelin, n | 7 | 9 | |
| Total clinic visits (frequencies), median (IQR) | 31.0 (12.0-66.0) | 12.0 (11.0-34.0) | 0.1859 |
| Total drug use (days), median (IQR) | 790.0 (342.0-1841.0) | 336.0 (84.0-950.0) | 0.1887 |
| Total dosage, median (IQR), mg | 116.3 (45.0-247.5) | 45.0 (41.3-127.5) | 0.1859 |
| Total dosage/total dosage use days | 0.1 (0.1-0.1) | 0.1 (0.1-0.1) | 0.3489 |
| L01XE02 gefitinib, n | 1 | 3 | |
| Total clinic visits (frequencies), median (IQR) | 39.0 (39.0-39.0) | 3.0 (1.0-18.0) | 0.4370 |
| Total drug use (days), median (IQR) | 510.0 (510.0-510.0) | 79.0 (14.0-250.0) | 0.4370 |
| Total dosage, median (IQR), mg | 127250.0 (127250.0-127250.0) | 19750.0 (3500.0-62500.0) | 0.4370 |
| Total dosage/total dosage use days | 249.5 (249.5-249.5) | 250.0 (250.0-250.0) | 0.3318 |
| L01DB01 doxorubicin, n | 50 | 48 | |
| Total clinic visits (frequencies), median (IQR) | 4.0 (2.0-7.0) | 4.0 (2.0-7.0) | 0.6504 |
| Total drug use (days), median (IQR) | 35.0 (12.0-87.0) | 23.0 (8.5-58.0) | 0.1377 |
| Total dosage, median (IQR), mg | 290.0 (180.0-400.0) | 270.0 (145.0-400.0) | 0.3937 |
| Total dosage/total dosage use days | 6.9 (4.0-13.3) | 9.7 (4.3-17.0) | 0.3578 |
| L01DB03 epirubicin, n | 79 | 99 | |
| Total clinic visits (frequencies), median (IQR) | 6.0 (4.0-7.0) | 6.0 (3.0-7.0) | 0.9134 |
| Total drug use (days), median (IQR) | 42.0 (16.0-72.0) | 42.0 (21.0-80.0) | 0.5269 |
| Total dosage, median (IQR), mg | 360.0 (200.0-520.0) | 400.0 (200.0-600.0) | 0.3450 |
| Total dosage/total dosage use days | 7.5 (4.8-14.3) | 7.1 (4.3-15.0) | 0.5288 |
* p < 0.05. p values were calculated by using Wilcoxon rank sum test to compare the data of with ocular diseases and without ocular diseases. IQR, interquartile range.
Supplemental Table 4. Average chemotherapy dosage effect in ocular diseases group on the risk of incident stroke.
| Stroke events | Total subjects | Crude RR (95% CI) | p value | p for interaction | |
| Chemotherapy use | |||||
| 0 < DDD ≤ 52.0 | 19 (10.5) | 181 | 1.00 | ||
| 52.0 < DDD ≤ 363.5 | 14 (7.91) | 177 | 0.75 (0.38-1.50) | 0.4216 | |
| 363.5 < DDD ≤ 920.0 | 15 (8.38) | 179 | 0.80 (0.41-1.57) | 0.5142 | |
| DDD > 920.0 | 16 (8.94) | 179 | 0.85 (0.44-1.66) | 0.6357 | 0.2927 |
| L02BA01 tamoxifen | |||||
| 0 < DDD ≤ 312.0 | 11 (10.28) | 107 | 1.00 | ||
| 312.0 < DDD ≤ 660.0 | 5 (4.67) | 107 | 0.45 (0.16-1.31) | 0.1438 | |
| 660.0 < DDD ≤ 1274.0 | 9 (8.49) | 106 | 0.83 (0.34-1.99) | 0.6704 | |
| DDD > 1274.0 | 9 (8.49) | 106 | 0.83 (0.34-1.99) | 0.6704 | |
| Other chemotherapy drugs | 30 (10.34) | 290 | 1.01 (0.5-2.01) | 0.9859 | 0.2691 |
Average chemotherapy drug dose [defined daily dose (DDD) per day] was calculated as cDDD divided by total drug prescription days. The average chemotherapy drug dose was classified by using two approaches: stratifying the chemotherapy exposure into categorizing the DDD (0 < DDD ≤ 52.0, 52.0 < DDD ≤ 363.5, 363.5 < DDD ≤ 920.0, DDD > 920.0 per day) according to a quartile method. In addition, the quartile method was adopted to classify the L02BA01 tamoxifen of average chemotherapy drug dose (0 < DDD ≤ 312.0, 312.0 < DDD ≤ 660.0, 660.0 < DDD ≤ 1274.0, DDD > 1274.0 per day).
Crude relative risk (RR) and p value were calculated by using Cox proportional hazards regression model. p for interaction: Cox proportional hazards regression model, including DDD or tamoxifen DDD × chemotherapy (without/with ocular diseases) interaction term, was applied. CI, confidence interval.
Supplemental Table 5-1. Ocular diseases predicted risk of incident stroke.
| Stroke events (%) | Total subjects | Crude RR (95% CI) | p value | Adjusted RR (95% CI) | p value | |
| Methotrexate | ||||||
| No | 48 (8.6) | 559 | 1.00 | 1.00 | ||
| Yes | 16 (10.2) | 157 | 1.19 (0.67-2.09) | 0.5529 | 1.08 (0.61-1.92) | 0.7849 |
| Tamoxifen | ||||||
| No | 30 (10.3) | 290 | 1.00 | 1.00 | ||
| Yes | 34 (8.0) | 426 | 0.77 (0.47-1.26) | 0.3004 | 1.10 (0.56-2.14) | 0.7857 |
| Methotrexate/tamoxifen | ||||||
| No | 18 (9.4) | 191 | 1.00 | 1.00 | ||
| Yes | 46 (8.8) | 525 | 0.93 (0.54-1.60) | 0.7933 | 1.24 (0.65-2.36) | 0.5189 |
| Ocular diseases | ||||||
| No | 16 (4.5) | 358 | 1.00 | 1.00 | ||
| Yes | 48 (13.4) | 358 | 3.00 (1.70-5.28) | 0.0001 | 2.93 (1.66-5.18) | 0.0002 |
| Methotrexate/tamoxifen and ocular diseases | ||||||
| No and No | 5 (4.3) | 115 | 1.00 | 1.00 | ||
| Yes and No | 11 (4.5) | 243 | 1.04 (0.36-3.00) | 0.9404 | 1.54 (0.51-4.66) | 0.4450 |
| No and Yes | 13 (17.1) | 76 | 3.93 (1.4-11.04) | 0.0092 | 4.19 (1.49-11.80) | 0.0067 |
| Yes and Yes | 35 (12.4) | 282 | 2.85 (1.12-7.29) | 0.0282 | 3.75 (1.37-10.26) | 0.0099 |
Adjusted relative risk (RR) and p value were calculated after adjustment for chemotherapy age, sex, hypertension, diabetes mellitus and radiotherapy by using Cox proportional hazards regression model. CI, confidence interval.
Supplemental Table 5-2. Ocular diseases predicted risk of incident stroke.
| Crude RR (95% CI) | p | Adjusted RR (95% CI) | p | |
| L01BA01 methotrexate | 1.19 (0.67-2.09) | 0.5529 | 0.92 (0.50-1.69) | 0.7916 |
| L02BA01 tamoxifen | 0.77 (0.47-1.26) | 0.3004 | 1.01 (0.50-2.07) | 0.9712 |
| Ocular diseases | 3.00 (1.70-5.28) | 0.0001 | 2.96 (1.66-5.26) | 0.0002 |
| Chemotherapy at age ≥ 55 years old, n (%) | 3.83 (2.05-7.18) | < 0.0001 | 2.47 (1.29-4.72) | 0.0061 |
| Sex, males, n (%) | 1.89 (1.12-3.19) | 0.0168 | 1.86 (0.90-3.87) | 0.0962 |
| Hypertension, n (%) | 3.81 (2.03-7.14) | < 0.0001 | 2.45 (1.27-4.72) | 0.0073 |
| Diabetes mellitus, n (%) | 2.27 (1.39-3.70) | 0.0010 | 1.73 (1.04-2.87) | 0.0339 |
| Radiotherapy, n (%) | 0.37 (0.16-0.87) | 0.0220 | 0.54 (0.23-1.26) | 0.1548 |
Adjusted relative risk (RR) and p value were calculated by using Cox proportional hazards regression model. CI, confidence interval.
Acknowledgments
Financial support
We thank the National Health Insurance Administration, Ministry of Health and Welfare, Taiwan for its support. This work was supported by a grant from the Kaohsiung Medical University Research Foundation (KMU-M111007) and a grant from Kaohsiung Medical University Hospital (KMUH111-1M38).
Author contributions
The first author, Dr. Kai-Chun Cheng, and corresponding author, Dr. Kai-Hung Cheng, participated on the generation of the original ideas, in the study design and on the analysis of data, in drafting of the manuscript, in revising it critically for important intellectual content and in final approval of the manuscript submitted. Other authors participated in 1) conception and design or analysis and interpretation of data, or both: Hung-Pin Tu, 2) drafting of the manuscript or revising it critically for important intellectual content: Hung-Pin Tu and Tsung-Hsien Lin.
DECLARATION OF FUNDING
There is no sponsorship/funding.
DECLARATION OF FINANCIAL/OTHER RELATIONSHIPS
There are no relationships to be declared.
REFERENCES
- 1.Cheng KH, Wu YW, Hou CJ, et al. An overview of cardio-oncology, a new frontier to be explored. Acta Cardiol Sin. 2021;37:457–463. doi: 10.6515/ACS.202109_37(5).20210706A. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Zamorano JL, Lancellotti P, Rodriguez Munoz D, et al. 2016 ESC Position Paper on cancer treatments and cardiovascular toxicity developed under the auspices of the ESC Committee for Practice Guidelines: The Task Force for cancer treatments and cardiovascular toxicity of the European Society of Cardiology (ESC). Eur Heart J. 2016;37:2768–2801. doi: 10.1093/eurheartj/ehw211. [DOI] [PubMed] [Google Scholar]
- 3.Lenneman CG, Sawyer DB. Cardio-oncology: an update on cardiotoxicity of cancer-related treatment. Circ Res. 2016;118:1008–1020. doi: 10.1161/CIRCRESAHA.115.303633. [DOI] [PubMed] [Google Scholar]
- 4.Plenderleith IH. Treating the treatment: toxicity of cancer chemotherapy. Can Fam Physician. 1990;36:1827–1830. [PMC free article] [PubMed] [Google Scholar]
- 5.Mitchell EP. Gastrointestinal toxicity of chemotherapeutic agents. Semin Oncol. 1992;19:566–579. [PubMed] [Google Scholar]
- 6.Kreisman H, Wolkove N. Pulmonary toxicity of antineoplastic therapy. Semin Oncol. 1992;19:508–520. [PubMed] [Google Scholar]
- 7.Vizel M, Oster MW. Ocular side effects of cancer chemotherapy. Cancer. 1982;49:1999–2002. doi: 10.1002/1097-0142(19820515)49:10<1999::aid-cncr2820491009>3.0.co;2-b. [DOI] [PubMed] [Google Scholar]
- 8.Imperia PS, Lazarus HM, Lass JH. Ocular complications of systemic cancer chemotherapy. Surv Ophthalmol. 1989;34:209–230. doi: 10.1016/0039-6257(89)90105-7. [DOI] [PubMed] [Google Scholar]
- 9.Al-Tweigeri T, Nabholtz JM, Mackey JR. Ocular toxicity and cancer chemotherapy. A review. Cancer. 1996;78:1359–1373. doi: 10.1002/(SICI)1097-0142(19961001)78:7<1359::AID-CNCR1>3.0.CO;2-G. [DOI] [PubMed] [Google Scholar]
- 10.Jack MK, Hicks JD. Ocular complications in high-dose chemoradiotherapy and marrow transplantation. Ann Ophthalmol. 1981;13:709–711. [PubMed] [Google Scholar]
- 11.Kaida T, Ogawa T, Amemiya T. Cataract induced by short-term administration of large doses of busulfan: a case report. Ophthalmologica. 1999;213:397–399. doi: 10.1159/000027462. [DOI] [PubMed] [Google Scholar]
- 12.Hilliard LM, Berkow RL, Watterson J, et al. Retinal toxicity associated with cisplatin and etoposide in pediatric patients. Med Pediatr Oncol. 1997;28:310–313. doi: 10.1002/(sici)1096-911x(199704)28:4<310::aid-mpo12>3.0.co;2-g. [DOI] [PubMed] [Google Scholar]
- 13.Greenberg PB, Chen AJ, Wu WC. Sudden vision loss and mortality: The Jackson Heart Study. Ophthalmic Epidemiol. 2016;23:285–291. doi: 10.1080/09286586.2016.1215476. [DOI] [PubMed] [Google Scholar]
- 14.Gudmundsson LS, Scher AI, Aspelund T, et al. Migraine with aura and risk of cardiovascular and all cause mortality in men and women: prospective cohort study. BMJ. 2010;341:c3966. doi: 10.1136/bmj.c3966. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Varma DD, Cugati S, Lee AW, et al. A review of central retinal artery occlusion: clinical presentation and management. Eye (Lond) 2013;27:688–697. doi: 10.1038/eye.2013.25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Andrikopoulos GK, Alexopoulos DK, Gartaganis SP. Pseudoexfoliation syndrome and cardiovascular diseases. World J Cardiol. 2014;6:847–854. doi: 10.4330/wjc.v6.i8.847. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Omoti AE, Omoti CE. Ocular toxicity of systemic anticancer chemotherapy. Pharm Pract (Granada) 2006;4:55–59. [PMC free article] [PubMed] [Google Scholar]
- 18.Yoshizawa K, Sasaki T, Kuro M, et al. Corneal damage induced in adult mice by a single intraperitoneal injection of N-ethyl-N-nitrosourea. In Vivo. 2011;25:609–615. [PubMed] [Google Scholar]
- 19.Kashiwagi H. Ocular disorders of anticancer drugs--ocular side effects. Gan To Kagaku Ryoho. 2010;37:1639–1644. [PubMed] [Google Scholar]
- 20.Parkkari M, Paakkala AM, Salminen L, et al. Ocular side-effects in breast cancer patients treated with tamoxifen and toremifene: a randomized follow-up study. Acta Ophthalmol Scand. 2003;81:495–499. doi: 10.1034/j.1600-0420.2003.00116.x. [DOI] [PubMed] [Google Scholar]
- 21.Paganini-Hill A, Clark LJ. Eye problems in breast cancer patients treated with tamoxifen. Breast Cancer Res Treat. 2000;60:167–172. doi: 10.1023/a:1006342300291. [DOI] [PubMed] [Google Scholar]
- 22.Susanna R, Jr., Oltrogge EW, Carani JC, et al. Mitomycin as adjunct chemotherapy with trabeculectomy in congenital and developmental glaucomas. J Glaucoma. 1995;4:151–157. [PubMed] [Google Scholar]
- 23.Yamashita T, Kodama Y, Tanaka M, et al. Steroid-induced glaucoma in children with acute lymphoblastic leukemia: a possible complication. J Glaucoma. 2010;19:188–190. doi: 10.1097/IJG.0b013e3181af321d. [DOI] [PubMed] [Google Scholar]
- 24.Newton HB. Neurologic complications of systemic cancer. Am Fam Physician. 1999;59:878–886. [PubMed] [Google Scholar]
- 25.Bushnell C. The cerebrovascular risks associated with tamoxifen use. Expert Opin Drug Saf. 2005;4:501–507. doi: 10.1517/14740338.4.3.501. [DOI] [PubMed] [Google Scholar]
- 26.Dignam JJ, Fisher B. Occurrence of stroke with tamoxifen in NSABP B-24. Lancet. 2000;355:848–849. doi: 10.1016/S0140-6736(05)72466-1. [DOI] [PubMed] [Google Scholar]
- 27.Veronesi U, Maisonneuve P, Rotmensz N, et al. Tamoxifen for the prevention of breast cancer: late results of the Italian Randomized Tamoxifen Prevention Trial among women with hysterectomy. J Natl Cancer Inst. 2007;99:727–737. doi: 10.1093/jnci/djk154. [DOI] [PubMed] [Google Scholar]
- 28.Saynak M, Cosar-Alas R, Yurut-Caloglu V, et al. Chemotherapy and cerebrovascular disease. J BUON. 2008;13:31–36. [PubMed] [Google Scholar]
- 29.Wooten MD, Reddy GV, Johnson RD. Atrial fibrillation occurring in a patient taking etanercept plus methotrexate for rheumatoid arthritis. Del Med J. 2000;72:517–519. [PubMed] [Google Scholar]
- 30.Rogers LR. Cerebrovascular complications in cancer patients. Neurol Clin. 2003;21:167–192. doi: 10.1016/s0733-8619(02)00066-x. [DOI] [PubMed] [Google Scholar]
- 31.Fitzpatrick T, Carrier M, Le Gal G. Cancer, atrial fibrillation, and stroke. Thromb Res. 2017;155:101–105. doi: 10.1016/j.thromres.2017.05.006. [DOI] [PubMed] [Google Scholar]
- 32.Wang TJ, Wang IJ, Hu CC, et al. Comorbidities of dry eye disease: a nationwide population-based study. Acta Ophthalmol. 2012;90:663–668. doi: 10.1111/j.1755-3768.2010.01993.x. [DOI] [PubMed] [Google Scholar]
- 33.Tuttolomondo A, Pecoraro R, Simonetta I, et al. Neurological complications of Anderson-Fabry disease. Curr Pharm Des. 2013;19:6014–6030. doi: 10.2174/13816128113199990387. [DOI] [PubMed] [Google Scholar]
- 34.Behnam-Motlagh P, Tyler A, Grankvist K, et al. Verotoxin-1 treatment or manipulation of its receptor globotriaosylceramide (gb3) for reversal of multidrug resistance to cancer chemotherapy. Toxins (Basel) 2010;2:2467–2477. doi: 10.3390/toxins2102467. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.De Rosa MF, Ackerley C, Wang B, et al. Inhibition of multidrug resistance by adamantylgb3, a globotriaosylceramide analog. J Biol Chem. 2008;283:4501–4511. doi: 10.1074/jbc.M705473200. [DOI] [PubMed] [Google Scholar]
- 36.Olubor OJ, Uhumwangho OM, Omoti AE. Ocular disorders in stroke patients in a tertiary hospital in Nigeria. Niger J Clin Pract. 2016;19:397–400. doi: 10.4103/1119-3077.179290. [DOI] [PubMed] [Google Scholar]
- 37.Lee MS, Kuo LL, Tan EC, et al. Is normal-tension glaucoma a risk factor for stroke? -A 10-year follow-up study. PLoS One. 2017;12:e0179307. doi: 10.1371/journal.pone.0179307. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.French DD, Margo CE. Open angle glaucoma and stroke. Ophthalmology. 2010;117:1653. doi: 10.1016/j.ophtha.2010.03.016. [DOI] [PubMed] [Google Scholar]
- 39.Ho JD, Hu CC, Lin HC. Open-angle glaucoma and the risk of stroke development: a 5-year population-based follow-up study. Stroke. 2009;40:2685–2690. doi: 10.1161/STROKEAHA.109.554642. [DOI] [PubMed] [Google Scholar]
- 40.Huck A, Harris A, Siesky B, et al. Vascular considerations in glaucoma patients of African and European descent. Acta Ophthalmol. 2014;92:e336–e340. doi: 10.1111/aos.12354. [DOI] [PMC free article] [PubMed] [Google Scholar]