Association between chronic kidney disease and open-angle glaucoma in South Korea: a 12-year nationwide retrospective cohort study

Various non-intraocular pressure factors have been identified as possible risk factors for open-angle glaucoma (OAG). However, there is still controversy around the association between OAG and chronic kidney disease (CKD). In this study, we used a nationwide cohort to investigate the risk of OAG in the 12 years following a diagnosis of CKD. This retrospective cohort study included 1,103,302 subjects from the Korean National Health Insurance Service National Sample Cohort database. The CKD group (n = 1318) included patients who were initially diagnosed with CKD between 2003 and 2008. The subjects in the comparison group were matched at a 1:5 ratio using propensity scores. In multivariate Cox regression analysis, a diagnosis of CKD was significantly associated with an increased incidence of OAG (hazard ratio [HR] = 1.546, 95% confidence interval [CI] 1.363–1.754, p < 0.001). Further analysis revealed that the risk of OAG increased with the severity of CKD (mild to moderate CKD [CKD stage 1–3]: HR = 1.280, 95% CI 1.077–1.521, p = 0.005; advanced CKD [CKD stage 4–5]: HR = 1.861, 95% CI 1.589–2.180, p < 0.001). In subgroup analysis, female CKD patients had a greater risk of developing OAG than males, and subjects with CKD aged ≥ 40 years were more likely to develop OAG compared with those aged < 40 years. Our study demonstrates that CKD is a significant risk factor for OAG and that severe CKD is associated with an increased risk of developing OAG.

www.nature.com/scientificreports/ In light of above, in this study, we investigated the risk of developing OAG after CKD diagnosis using a representative sample of approximately 1.1 million South Koreans from the National Health Insurance Service-National Sample Cohort 2002-2015 (NHIS-NSC 2002-2015. In addition, we analyzed the risk of developing OAG according to CKD severity. Table 1 shows the baseline characteristics of the study population and the differences between the OAG group and the control group. There was a significant difference in the prevalence of CKD in the OAG group and the control group (p < 0.001). Subjects in the OAG group were older (p < 0.001), had lower income (p < 0.001), were more likely to live in rural areas (p < 0.001), and had lower CCI scores (p < 0.001). Hypertension (p < 0.001), diabetes mellitus (p < 0.001), and hyperlipidemia (p < 0.001) were more prevalent in the OAG group than in the control group. There were no significant differences in sex or ischemic stroke.

Results
The risk of a CKD patient developing OAG during the 12-year follow-up period was analyzed using multivariate Cox hazard regression analysis. The risk of developing OAG during the 12-year follow-up period was significantly higher in the CKD group than the control group (CKD group: adjusted HR = 1.546, 95% CI 1.363-1.754, p < 0.001) ( Table 2). Higher CKD stages were associated with a higher risk of OAG (CKD group 1: adjusted HR = 1.280, 95% CI 1.077-1.521, p = 0.005; CKD group 2: adjusted HR = 1.861, 95% CI 1.589-2.180, p < 0.001) ( Table 3). Older age, low income, rural residency, hypertension, diabetes mellitus, and hyperlipidemia were also associated with an increased risk of OAG.
In subgroup analyses based on multivariate Cox regression, the adjusted HR of OAG in the female CKD patients was higher (adjusted HR = 2.989) than that in the male CKD patients (adjusted HR = 2.041, Fig. 1). www.nature.com/scientificreports/ Moreover, the risk of developing AD in patients with OAG was higher in the subgroup of age ≥ 40 years (HR = 2.861) than that in the subgroup of age < 40 years (HR = 0.279). As shown in the Kaplan-Meier survival curves in Fig. 1, the risk of OAG was higher for patients at CKD stages 1-3 (mild to moderate CKD) and CKD stages 4-5 (advanced CKD) than in the non-CKD comparison group (log-rank test, < 0.001) (Fig. 2). In addition, the cumulative incidence of OAG during the follow-up period was significantly higher in advanced CKD than mild to moderate CKD (log-rank test, p < 0.001). Furthermore, the risk of developing OAG was higher for CKD patients in both male and female subgroups, as well as in the subgroup of age ≥ 40 years (log-rank test, < 0.001, Fig. 3A-C). However, no significant difference in the cumulative OAG incidence was observed between the CKD and comparison group in the subgroup of age < 40 years (log-rank test, p = 0.3022, Fig. 3D).

Discussion
In this population-based cohort study, we evaluated the risk of OAG in CKD patients using propensity score matching. We used multivariate Cox proportional hazard regression analysis and found that CKD was a significant risk factor for OAG. In a subgroup analysis in which CKD patients were divided into two groups based on CKD severity, there was an increased risk of OAG in CKD patients with more advanced disease. To our knowledge, this is the first study using a large longitudinal cohort from a nationwide database to evaluate the risk of OAG in CKD patients by disease stage and to show a significant increase in OAG risk with worsening kidney function.
Several population-based studies have explored the relationship between CKD and OAG, with conflicting results. A study using National Health and Nutrition Examination Survey data found no significant association  www.nature.com/scientificreports/ between CKD and glaucoma 33 . Another report on a large pool of Asian population-based studies showed that neither lower estimated glomerular filtration rate (eGFR) nor CKD were associated with primary OAG (POAG) prevalence 31 . However, subgroup analysis revealed a significant association between lower eGFR and POAG prevalence in East Asians, including Korean and Chinese individuals 31 . This association was supported by a recent population-based cross-sectional study from South Korea demonstrating a positive association between lower eGFR and POAG 34 , suggesting that kidney function decline is a risk factor for POAG. The discrepancies in study results may be due to ethnic differences and the higher prevalence of normal-tension glaucoma (NTG)  www.nature.com/scientificreports/ in East Asian countries. Vascular risk factors, including hypertension, diabetes mellitus, and hyperlipidemia, are considered more important in NTG than POAG 6 . Our findings, which are based on a longitudinal cohort study, are consistent with previous results demonstrating a causal relationship between CKD and OAG. It has been suggested that CKD and OAG share several pathophysiological mechanisms. The most frequently mentioned mechanism is the RAS. The RAS is a systemic mechanism that is essential for maintaining blood pressure and electrolyte homeostasis, and has been reported to be upregulated in CKD patients 35 . In the eye, the RAS has a role in the production of aqueous humor from the ciliary body and its outflow through the trabecular meshwork 36,37 . Especially, Angiotensin II (AngII), a major active component of RAS system affected by the release of renin, has been reported to increase production and decrease outflow of AH by acting on non-pigmented epithelium of ciliary body as well as trabecular meshwork [38][39][40] . Another in vivo study using rabbits showed that intracamerally intected AngII diminished uveoscleral outflow 41 , one of main AH outflow pathway other than trabecular meshwork. Such changes in the eye may result in increased IOP, which is a well-known risk factor for the pathogenesis of glaucoma. Moreover, AngII also has been reported to play a significant role in the regulation of ophthalmic microcirculation by controlling contraction and relaxation of vascular endothelium 42,43 . Disruption in blood-brain barrier of the optic nerve head in glaucoma, as previously reported 44 , may result in local passage of systemic RAS component to the optic nerve and retina, leading to dysregulation of the retinal microcirculation. In addition, an animal study using a rodent model of glaucoma found that angiotensin-converting enzyme inhibitors and angiotensin receptor antagonists may have a beneficial effect on RGC survival 45 , suggesting that RAS may modulate RGC survival through exerting a neuroprotective effect.
Other possible common pathologic mechanisms between CKD and OAG may include atherosclerosis and oxidative stress. It has been previously reported that decreased renal function may induce atherosclerosis 21,46 , and another study by Shim et al 22 . showed that POAG was significantly associated with increased systemic arterial stiffness, suggesting possible connection between the two conditions. Moreover, oxidative stress has been considered as one of well-known pathogenic mechanisms of OAG 23 , which also has been suggested as important underlying mechanism in CKD pathogenesis 24 . Also, low eGFR may reflect accumulation of reactive oxygen species 47 , and this may lead to increased susceptibility of glaucoma development by stimulating RGC apoptosis and glial activation in posterior segment of eye 48 . Such hypothesis may be supported by our findings that subjects suffering from more severe CKD condition showed increased HRs of developing OAG. Furthermore, previous studies showed significant association between elevated serum uric acid level and open-angle glaucoma [49][50][51] . Interestingly, serum uric acid level was reported to be commonly elevated in patients with CKD 52 , suggesting possible connection between the two diseases. However, further large scale prospective studies are needed to exactly evaluate these relationships.
Several previous cross-sectional population-based studies have demonstrated an association between lower eGFR and POAG 31,34 , but whether a causal relationship was present is uncertain. Our study, using a longitudinal population-based cohort design, showed an increased risk of OAG in CKD patients with advanced disease. We found that patients in advanced CKD group (CKD stages 3-5) had a significantly increased risk of OAG compared to mild to moderate CKD group (CKD stages 1-2). Although eGFR is the standard for staging CKD, the results from this study suggests that lower eGFR is an independent risk factor for OAG in CKD patients. Further studies should identify the pathogenetic mechanism underlying this association.
Our study has several limitations that should be considered when interpreting the results. First, the diagnoses of OAG, CKD, and other comorbidities were based entirely on KCD codes, which may be less accurate than the results of standard diagnostic tests performed in a real clinic. In this study, however, we included only OAG patients who met all three inclusion criteria: having the KCD code for an OAG diagnosis; having the code for visual field tests; and having a prescription for anti-glaucoma eye drops. As a result, the selection of OAG subjects used in the statistical analyses may have had high specificity. Second, the data used in our study were based only on patients who had visited a hospital or clinic during the study period. Therefore, the data excluded asymptomatic patients and those who had not visited hospitals for personal or economic reasons. This may have caused selection bias, leading to an underestimation of the true prevalence or incidence of OAG and CKD. However, the effects of any such bias or underestimation is thought to be limited; our government-run healthcare system includes almost the entire population of the country and is reasonably low-cost, making healthcare highly accessible.
In conclusion, our 12-year nationwide cohort study demonstrated that CKD patients have a higher risk of OAG than those without CKD. Moreover, more severe disease is associated with a significant increase in the risk of OAG. This indicates that lower eGFR is associated with OAG. The results of this study further support eye screening in CKD patients for early detection and treatment of OAG.

Methods and materials
Database and study sample. Our  To analyze the effect of CKD on the risk of OAG, a control group was selected using propensity scores with nearest neighbor matching at a 1:5 ratio of CKD to non-CKD subjects. Ultimately, 25,560 subjects were included: 4260 in the CKD group and 21,300 in the non-CKD comparison group. These groups included 1318 OAG cases. Propensity scores were calculated using socioeconomic parameters, including age group (≤ 49, 50-59, 60-69, 70-79, ≥ 80), sex, household income percentile (≤ 30th, 31st-70th, ≥ 71st percentile according to insurance premium), residential area (urban included cities including metropolitan cities and rural included all other areas), and Charlson comorbidity index (CCI) scores (< 3, ≥ 3).
Outcomes and comorbidities. We compared the incidence of OAG in the matched populations. CKD was the main outcome in the analysis. Subjects with CKD were divided into two groups based on CKD severity (mild to moderate CKD: CKD stages 1-3; advanced CKD: CKD stages 4-5). Socioeconomic variables and comorbid conditions were selected as confounding factors. Socioeconomic variables were age, sex, household income, residential area, and CCI. Comorbid conditions included hypertension (KCD codes I10-I15), diabetes mellitus (KCD codes E10-E14), hyperlipidemia (KCD codes E78.0-E78.5), and ischemic stroke (KCD code I61).
Statistical analysis. The chi-square test was used to compare the groups. Multivariate Cox proportional hazard regression analysis was performed to calculate HRs with 95% confidence intervals (CI) for estimating the incidence of OAG in CKD patients. Survival curves for OAG were generated. All statistical analyses were conducted using SAS 7.3 and STATA 16.