Association Between Osteoporosis and Age-Related Macular Degeneration: The Korea National Health and Nutrition Examination Survey

Abstract

Purpose: Previous studies have reported a possible link between low bone mineral density and AMD. The aim of the present study was to investigate the association between osteoporosis and AMD in a South Korean cohort.

Methods: This cross-sectional, nationwide study included 3496 women and 2789 men who had participated in the Korean National Health and Nutrition Examination Survey from 2008 to 2011. All retinal photographs were graded using an international classification and grading system. Osteoporosis was assessed using dual-energy x-ray absorptiometry. Multivariate logistic regression analysis was performed to examine the relationship between osteoporosis and AMD after adjustment for potential confounders, including age, the body mass index, dietary calcium intake, and the serum vitamin D level. The odds ratios (OR) for other aging-related eye diseases, including cataract, open-angle glaucoma, and diabetic retinopathy, were analyzed in accordance with the presence of osteoporosis.

Results: Multivariate regression analysis revealed that osteoporosis was significantly associated with all types of AMD (early and late: OR, 1.31; P = 0.017) and early AMD (OR, 1.36; P = 0.007) in women. Late AMD was not associated with osteoporosis (OR, 0.84; P = 0.670). In men, osteoporosis was not associated with any type of AMD. In women, the status of osteoporosis in the femoral neck showed a linear relationship with AMD (P = 0.004). Although osteoporosis was associated with AMD in women, it showed no association with other age-related eye diseases; this suggested a disease-specific association.

Conclusions: Our findings suggest that osteoporosis plays a role in AMD development in postmenopausal women.

Osteoporotic fractures are an increasing socioeconomic burden.¹ Osteoporosis is common in elderly postmenopausal women, although it is asymptomatic until a fracture occurs. Risk factors, including decreased exposure to estrogen, systemic inflammation, and a decreased vitamin D level, can induce changes in osteoblast–osteoclast interactions to result in destruction of the bone-remodeling process and progressive bone mineral loss.² Generally, decreased visual acuity is also recognized as an important risk factor for fractures³ because of the high risk of frequent falls.^4,5 Several researchers have suggested that an improvement in the visual acuity may prevent falls and hip fracture.^6,7 Aging-related visual impairment combined with osteoporosis can elevate the fracture risk. However, the relationship between osteoporosis and most aging-related eye diseases remains unclear. Earlier studies have revealed that systemic inflammation and vitamin D could affect both osteoporosis and aging-related eye diseases.^8,9 Metabolic syndrome has also been considered as a link between osteoporosis and several eye diseases.¹⁰ 

AMD is the most common progressive retinal disorder resulting in poor vision in developed countries.¹¹ Previous results from the Study of Osteoporotic Fracture, which included a Caucasian cohort, demonstrated that low bone mineral density (BMD) was linked with AMD.¹² A recent immunohistologic study showed that bone mineral (hydroxyapatite) spherules in Bruch's membrane in human eyes could initiate the formation of several deposits external to the retinal pigment epithelium (RPE).¹³ These sub-RPE deposits are closely associated with the formation of drusen and progression of AMD. Another study based on the National Health and Nutrition Examination Survey (NHANES) reported that calcium supplementation for the prevention of osteoporosis was associated with an increased prevalence of AMD.¹⁴ A cohort study implemented in Australia revealed that AMD was significantly associated with total hip replacement surgery for hip fracture.¹⁵ On the basis of these previous findings, we hypothesized that bone mineral loss is closely associated with the prevalence of AMD in the Korean population and conducted the present study to investigate the relationship between osteoporosis and aging-related eye diseases, including AMD, cataract, open-angle glaucoma (OAG), and diabetic retinopathy, through multivariate regression analyses and comparisons of odds ratios (OR) in order to identify a disease-specific association in a South Korean cohort. 

Methods

This study was based on the Korean NHANES (KNHANES; available online in the public domain at http://knhanes.cdc.go.kr) conducted between 2008 and 2011.¹⁶ The institutional review board of the Korean Center for Disease Control and Prevention approved the study protocol. All participants in the study provided written informed consent for the use of their health information for research purposes. KNHANES is a nationwide, population-based, cross-sectional survey conducted by the Division of Chronic Disease Surveillance of the Korea Centers for Disease Control and Prevention. For the current study, we randomly selected all participants from 200 (2008–2009, KNHANES IV) plus 192 (2010–2011, KNHANES V) enumeration districts using stratified sampling accounting for the following factors: population, sex, age, region, and type of residential area. KNHANES itself comprised health records based on a health interview, health examination, and nutrition survey. The dataset contained BMD measurements for the entire hip, femoral neck, and lumbar spine. BMD was measured by dual-energy x-ray bone densitometry (DEXA) using Hologic Discovery (Hologic, Inc., Bedford, MA, USA). The World Health Organization defines osteoporosis as a BMD of 2.5 SDs below the peak bone mass of a young, healthy, sex- and race-matched reference population. The normal bone mass group included participants with a T score of −1.0 and above. The osteopenia group included participants with a T score between −1.0 and −2.5. The osteoporosis group included participants with a T score of −2.5 and below. Osteoporosis was defined as the presence of a low BMD (T-score, ≤−2.5) in any of the following sites: entire hip, femoral neck, and lumbar spine. 

A comprehensive ophthalmologic examination was conducted by ophthalmologists dispatched by the Korean Ophthalmological Society (KOS) in a vehicle equipped with ophthalmic devices. The KOS National Epidemiologic Survey Committee periodically trained the ophthalmologists. Fundus photographs were obtained using a digital fundus camera (TRC-NW6S; Topcon, Tokyo, Japan).¹⁷ For each eye in each participant, a 45° digital retinal fundus image centered on the macula and fovea was obtained. All fundus photographs were subjected to preliminary and detailed grading. Preliminary grades were assigned to the retinal images by trained ophthalmologists using the International Age-Related Maculopathy Epidemiological Study Group grading system.¹⁸ Nine retinal specialists who were approved by KOS performed detailed grading. Final grading was based on the detailed grading, and one retinal specialist resolved any discrepancy between the preliminary and detailed grades. Early AMD was defined by the presence of one of the following on fundus photographs: soft indistinct drusen or reticular drusen and hard or soft distinct drusen with pigmentary abnormalities (increased pigmentation or hypopigmentation of RPE) in the absence of signs of late AMD. If the fundus photographs showed signs of wet AMD or geographic atrophy, the patient was diagnosed with wet AMD, which was defined as RPE detachment or serous detachment of the neurosensory retina, sub-RPE hemorrhage, and subretinal fibrous scarring. Geographic atrophy was defined as a circular discrete area (175 μm in diameter) of retinal depigmentation with visible choroidal vessels in the absence of signs of wet AMD. For participants with AMD in only one eye, AMD was defined according to the affected eye. For those with asymmetrical AMD in both eyes, the worst affected eye was considered. 

We also investigated the association of aging-related eye diseases with the status of osteoporosis for comparisons to confirm the disease-specific association between AMD and osteoporosis. Participants with OAG were defined as those who fulfilled the diagnostic criteria specified by the International Society for Geographical and Epidemiological Ophthalmology.⁹ We adopted the standard Lens Opacities Classification System III (LOCS III) for diagnosing cataract.¹⁹ Participants with diabetic retinopathy were identified through fundus examinations for patients with diabetes who met the glucose level and HbA1c criteria defined by the American Diabetes Association. 

Information on risk factors associated with AMD and osteoporosis was obtained through face-to-face interviews. Each participant completed a questionnaire that required information pertaining to age, alcohol consumption, the smoking status, hypertension, diabetes, age at menopause, and hormone use. Body weight and height were measured with the participants in light indoor clothing without shoes. Body mass index (BMI) was calculated using the following formula: weight (kg)/[height(m)]². Dietary calcium intake was assessed using a 24-hour dietary recall questionnaire administered by a trained dietitian. Fasting blood samples were collected from individual participants in order to measure the serum 25-hydroxyvitamin D level using a gamma counter (1470 Wizard; Perkin-Elmer, Turku, Finland) with an RIA (Dia-Sorin, Stillwater, MN, USA). 

Using χ² tests for categorized variables and one-way ANOVA for continuous variables, we compared the characteristics of participants according to the category of AMD. To confirm an independent association between AMD and bone mineral loss, we built adjusted logistic regression models. For evaluation of the effects of BMD on the AMD prevalence, BMD was grouped into quartiles. The analysis for postmenopausal women was adjusted for potential confounders associated with osteoporosis and AMD. Age, alcohol consumption,²⁰ current smoking,²¹ hypertension,²² and diabetes mellitus²³ are well-known risk factors for AMD, while BMI,²⁴ the residential area,²⁵ atopic dermatitis,²⁶ asthma,²⁷ rheumatoid arthritis,²⁸ dietary calcium intake,²⁹ serum vitamin D level,³⁰ duration of menopause, and past or current hormone³¹ use are well-established factors associated with osteoporosis. We evaluated an effect modification by including interaction terms between BMD and sex, BMD and age, and BMD and BMI. In addition, we fitted generalized additive models with cubic spline regression in order to visualize the dose-response relationship between BMD and AMD. The resulting spline models represented adjusted log ORs for AMD. To test for the linear trend relationship with the status of osteoporosis, we regarded each category as a simple and continuous variable in a multivariate-adjusted model. We also compared demographic characteristics between included and excluded KNHANES participants (Supplementary Table S1). A P value of less than 0.05 was considered statistically significant. All statistical analyses were performed using the statistical software packages IBM SPSS Statistics 18.0 (IBM SPSS, Inc., Armonk, NY, USA) and R software version 3.4.3 (The Comprehensive R Archive Network; available in the public domain at http://cran.r-project.org). 

Results

A total of 37,753 participants were initially enrolled in KNHANES 2008 to 2011. We restricted our study population to middle-aged and elderly participants, who are generally considered to be at high risk of osteoporosis. Both female and male participants were included in this study. We excluded participants younger than 40 years and those who did not attend the interviews and health examinations (N = 19,503), those who did not undergo the DEXA test (N = 9849), those who did not undergo fundus examinations (N = 1135), and those who were not tested for the serum 25-hydroxyvitamin D level (N = 981). Finally, 3496 women and 2789 men were considered eligible for the current analysis (Fig. 1). 

Figure 1

 

A flow diagram showing the selection of study participants for the present study based on the KNHANES 2008 to 2011.

The characteristics of the study participants are shown in Table 1. The prevalence of all types of AMD (early and late AMD), early AMD, and late AMD were 9.70% (N = 339), 9.01% (N = 315), and 0.69% (N = 24) among women and 10.00% (N = 279), 8.96% (N = 250), and 1.04% (N = 29) among men, respectively. Compared with the control group participants (no AMD), women with AMD were older and exhibited a lower BMI, higher rate of hypertension, and lower dietary calcium intake. They also tended to exhibit a low BMD in the hip, lumbar spine, and femoral neck, which resulted in a higher rate of osteoporosis (P < 0.001). Compared with the control group participants, men with AMD were older and exhibited a lower BMI and BMD in the hip and femoral neck. For females, the prevalence of osteoporosis was 33.1% for the control group, 54.9% the early AMD group, and 50.0% for the late AMD group. The prevalence of osteoporosis in male was 7.6% for the control group, 7.2% for the early AMD group, and 10.3% for the late AMD group. 

Table 1

 

Demographic and Clinical Characteristics According to the Status of AMD in Participants in the KNHANES Conducted Between 2008 and 2011

 

Using a likelihood ratio test for interaction terms to evaluate the effect modification in the multivariate model for ORs of all AMD types and early AMD, we found a significant interaction between BMD and sex (P for interaction < 0.05; Table 2). Therefore, ORs are presented separately for men and women. The interactions terms between BMD and age and BMD and BMI were not significant in all multivariate models for AMD (Supplementary Table S2). 

Table 2

 

Adjusted ORS for AMD According to the BMD in Participants in the KNHANES Conducted Between 2008 and 2011

 

According to univariate analyses, all AMD types were significantly associated with increased age, a lower BMI, hypertension, a lower dietary calcium intake, duration of menopause, past or current hormone use, and osteoporosis (Supplementary Table S3). Multivariate analysis adjusted for all the possible factors listed in Table 3 indicated that the prevalence of AMD significantly increased in accordance with the prevalence of osteoporosis in women. After adjustment for all factors, increased age (OR, 1.08; P < 0.001), and osteoporosis (OR, 1.31; P = 0.017) were significantly associated with AMD. This association mainly resulted from the strong association of early AMD with osteoporosis (OR, 1.36; P = 0.007). Late AMD showed no association with osteoporosis (OR, 0.84; P = 0.670). For men, multivariate analysis showed no association between osteoporosis and all AMD types (OR, 0.95; P = 0.664), early AMD (OR, 0.96; P = 0.749), and late AMD (OR, 0.75; P = 0.406). 

Table 3

 

Adjusted ORs for AMD According to Demographic Factors for Participants in the KNHANES Conducted Between 2008 and 2011

 

Table 4 presented the ORs for AMD according to the status of osteoporosis (quartiles of BMD in the entire hip, lumbar spine, and femoral neck). The numeric data for each group are shown in Supplementary Table S4. According to multivariate analyses, the severity of femoral neck osteoporosis showed a linear association with all AMD types (P = 0.004) and early AMD (P = 0.006) in women. Moreover, trend analysis showed no relationship between the severity of osteoporosis and OR for late AMD in women. In men, there was no association between BMD and all AMD types. 

Table 4

 

Adjusted ORs for AMD Stratified According to the Quartile of BMD for Participants in the KNHANES Conducted Between 2008 and 2011

 

To explore the overall association between BMD and AMD, we used generalized additive models with cubic spline regression (Fig. 2). The smooth curves generated from the cubic spline model, which was fully adjusted for the variables listed in Table 3, exhibited an increase in OR for all AMD types with a decrease in BMD in the overall pattern for women. Early AMD showed a similar pattern. When we focused on the sparse area representing a femoral neck BMD of less than 0.4 g/cm² and more than 0.8 g/cm², a linear pattern was identified in an analysis of late AMD. However, obvious trends were not observed for men. We also found that decreased BMDs in the entire hip and lumbar spine were grossly associated with an increase in OR for AMD in women, although the linear trends were weaker than those for the BMD in the femoral neck (Supplementary Fig. S1). 

Figure 2

 

Adjusted log odds ratios with 95% CIs for the risk of AMD according to the BMD in the femoral neck, obtained using generalized additive models with cubic spline regression for postmenopausal female participants in the KNHANES 2008 to 2011. The regression model has been adjusted for age, sex, BMI, residential area, alcohol consumption, current smoking, hypertension, diabetes mellitus, atopic dermatitis, asthma, rheumatoid arthritis, dietary calcium intake, the serum 25-hydroxyvitamin D level, the duration of menopause, and past or current hormone use.

Sex- and disease-associated differences in the association between aging-related eye diseases and osteoporosis are presented in Table 5. In women, AMD showed a stronger association with osteoporosis than did other eye diseases. Osteoporosis was not significantly associated with cataract (OR, 1.08; P = 0.480), OAG (OR, 1.02; P = 0.940), and diabetic retinopathy (OR, 1.41; P = 0.211). In addition, multivariate analyses showed that cataract (OR, 0.95; P = 0.812), OAG (OR, 1.27; P = 0.613), and diabetic retinopathy (OR, 0.52; P = 0.305) were not associated with osteoporosis in men. In both men and women, osteopenia showed no relationship with aging-related ocular diseases. 

Table 5

 

Comparison of ORs for Eye Diseases, Including AMD, Cataract, OAG, and Diabetic Retinopathy, According to the Status of Osteoporosis in Participants in the KNHANES 2008 to 2011

 

Discussion

To the best of our knowledge, this is the first study to identify the association between BMD and AMD in a large Asian population. We attempted to investigate the association between osteoporosis and multiple eye diseases, including AMD, cataract, OAG, and diabetic retinopathy. In the study population, which was a part of KNHANES 2008 to 2011, cross-sectional findings indicated that postmenopausal women with a lower BMD in the femoral neck had a significantly higher OR for early AMD. This suggests that osteoporosis may be associated with drusen formation. Other aging-related eye diseases did not show a significant association with osteoporosis. Therefore, this comparison implies an underlying mechanism, not a simple aging process that accounts for a disease-specific association between AMD and osteoporosis in women. 

The relationship between osteoporosis and progression of AMD is not fully understood. Bone mineral loss may stimulate circulating calcium and phosphate.³² These components may increase hydroxyapatite spherules in RPE and initiate drusen formation.^13,33 According to Tompson's study, hydroxyapatite may precipitate on extracellular lipid droplets at the RPE–choroid interface.¹³ The hydroxyapatite spherule may bind to different proteins, which promotes further deposition of lipids and leads to the formation of macroscopic sub-RPE deposits. Decreased exposure to estrogen, increased systemic inflammation, a lower vitamin D level, and metabolic diseases may interrupt bone remodeling and accelerate the deposition of minerals in the sub-RPE layer.^34,35 Local inflammation may boost oxidative stress, resulting in high lipid and mineral deposition in Bruch's membrane.³⁶ In bone, inflammation may increase the activity of bone-resorbing osteoclasts, leading to a decreased BMD. Therefore, it is possible to explain that osteoporosis is associated with a significantly increased prevalence of AMD (Fig. 3). 

Figure 3

 

Diagram showing the possible relationship between AMD and osteoporosis in a South Korean cohort selected from the KNHANES 2008 to 2011.

Age-related changes in early and late AMD have been reported to occur in Bruch's membrane, including hydroxyapatite deposition and drusen calcification.³⁷ Osteoporotic changes are typically related to ectopic calcification, including abnormal mineralization of the peripheral vessels, heart valves, and kidneys.³⁸ This association between tissue calcification and osteoporosis may be slightly connected with our study results. It is possible that secondary inflammation due to elevated calcium levels may be associated with the pathophysiology of cataract, OAG, and diabetic retinopathy. However, unlike AMD, these diseases are not directly associated with the calcification and deposition of lipid or protein products. Because calcium metabolism affects both oxidative stress and calcification during the pathogenesis of AMD, osteoporosis may be associated with AMD but not with other aging-related eye diseases. 

Several pathogenetic factors are shared by osteoporosis and AMD, including vitamin D levels, calcium intake, metabolic diseases, and factors associated with oxidative stress. Most previous studies have focused on the association between the vitamin D level and AMD.³⁹ One previous research showed that calcium supplements for the prevention of osteoporosis was associated with the occurrence of AMD in an elderly population.¹⁴ Vitamin D supplementation was also associated with protection against AMD.⁴⁰ A recent study reported that women with vitamin D deficiency and a genetic risk (CFH and CFI genotypes) were at a higher risk of developing AMD.⁴¹ Our results showed that osteoporosis was significantly associated with AMD after adjustment for dietary calcium intake and the vitamin D level. Hypertension and dyslipidemia are well-known risk factors for AMD and may also influence the development of osteoporosis.^42,43 Several signaling pathways associated with oxidative stress may affect both bone mineral loss and AMD. For example, the Wnt signaling pathway, which is important for osteoblast differentiation and bone formation, can affect the progression of AMD.^44,45 

In the present study, no significant results were derived for men, possibly because of the low prevalence of osteoporosis and AMD. An increased prevalence of drusen and neovascular AMD in women has been documented in previous reports.^46,47 Estrogen may contribute to sex-associated differences in the development of AMD associated with osteoporosis. In women, estrogen plays a critical role in bone remodeling and has anti-inflammatory effects. Moreover, it regulates a couple of signaling pathways that are connected with the pathogenesis of AMD.³⁴ An earlier study stated that exogenous estrogen use has a protective effect against AMD progression, and it supported the presence of a sex-associated difference in AMD development.⁴⁸ Men generally have a heavier bone mass than women, and they do not experience menopause, which results in dramatically low estrogen levels and rapid bone mineral loss. This hormonal and menopausal etiology associated with osteoporosis should be further explored in order to identify a more detailed pathophyisiology for AMD. 

We also observed that osteoporosis in the femoral neck was more strongly associated with AMD than was osteoporosis in the entire hip and lumbar spine. The reason for this differential association remains unclear. Generally, trabecular bone (typical of the lumbar spine) is enormously affected by secondary osteoporosis that is caused by factors, such as malabsorption, liver disease, rheumatoid arthritis, and medications, when compared with cortical bone (typical of the proximal femur, including the femoral neck).⁴⁹ Another explanation is that most bone mineral loss after 65 years of age primarily occurs in the intracortical bone, not in the endocortical or trabecular bone.⁵⁰ One study reported that mineral loss in the cortical bone is more strongly associated with ischemic heart disease than is mineral loss in the trabecular bone; this finding is consistent with our study findings.⁵¹ 

The present study has some limitations. First, the results could not confirm a causal relationship because of the cross-sectional study design. Our health-related data was based on a health interview survey conducted on one occasion. Several factors, such as BMD, calcium intake, the serum vitamin D level, and BMI, may differ depending on the time of measurement. Second, this study was based on a single Asian population. Generally speaking, the genetic background influences the incidence and type of osteoporosis and AMD. Therefore, it remains unclear whether our results are relevant to other ethnic groups. Third, there is a possibility of a selection bias because both ophthalmic examinations and AMD measurements were not performed for all participants. Moreover, younger participants were excluded, and this may represent a bias. However, a stratified, multistage, clustered probability design was used to select representative samples of noninstitutionalized Koreans for KNHANES, which reduces the possibility of selection bias. Finally, we did not consider concomitant corticosteroids, because information on inhaled and oral steroid use was not fully collected as part of KNHANES. Long-term corticosteroid use could serve as a critical confounding factor in relation to osteoporosis.⁵² 

In conclusion, it is noteworthy that this study investigated the disease-specific relationship between AMD and low BMD in a female population. The results imply a possible sex-associated difference in the uptake and metabolism of calcium, and that osteoporosis may be involved in AMD development in women. In order to confirm the effect of bone mineral loss on AMD progression, more in-depth studies on bone remodeling and sub-RPE mineral deposition should be conducted. 

Acknowledgments

Supported by a faculty research grant from Yonsei University College of Medicine for 2017 (6-2017-0089; Yonsei University, Seoul, South Korea). 

Disclosure: T.K. Yoo, None; S.H. Kim, None; J. Kwak, None; H.K. Kim, None; T.H. Rim, None