Cataract risk had a positive association with metabolic syndrome and negative with a traditional balanced diet in middle-aged adults

The prevalence of cataracts is steadily increasing among middle-aged and elderly Asians. We hypothesized that adults aged 40–77 years with cataracts would show an association with metabolic syndrome and its components and have different dietary patterns and lifestyles. Methods The study was performed using the Korean genome and epidemiology study (KoGES; a large-scale hospital-based cohort study) data collected during 2004–2013. Subjects were classied as cases(945 cataract patients) and controls(27,454 healthy controls) based on responses to the question "have you ever been diagnosed to have a cataract?”.

hyperglycemia, smoking, long-term exposure to sunlight, and heavy alcohol consumption [9][10][11], though reports on relations between alcohol consumption and smoking and cataract have been somewhat inconsistent. Lipid pro les have also been reported to affect cataract incidence but the relationship remains controversial. Since cataract formation is known to be associated with oxidative stress, the in uence of vitamin C on cataract development has attracted attention [12], but its effects are not obvious. Furthermore, these risk factors have been largely studied in Western countries, and studies on Asians are relatively recent. However, relations between dietary intakes (e.g., nutrient intakes and dietary patterns) and cataract risk have not been studied in Asians. Koreans have 5 distinct dietary patterns, that is, a traditional balanced diet, a bread-based diet, a rice-based diet, a noodle-based diet, and an alcoholbased diet. We hypothesized that middle-aged adults with cataracts display associations with metabolic syndrome and its components and dietary patterns and lifestyles. The present study was performed using data obtained during the Korean Genome and Epidemiology Study (KoGES; a large, annual, hospital-based cohort study) for the period 2004-2013.

Subject characteristics and biochemical results
A total of 28,342 Korean adults aged 40-77 years participated from 2004-2013 in the KoGES, which was organized by the Korean Center for Disease and Control [13]. This study was approved by the Institutional Review Board of the Korean National Institute of Health for the Korean Genome and Epidemiology Study (KoGES; KBP-2015-055) and Hoseo University (1041231-150811-HR-034-01). All that participated in KoGES provided written informed consent.
General characteristics, including socioeconomic factors, were acquired using survey responses [14].
Anthropometric, biochemical, and dietary parameters were assessed. Weight, height, and waist and hip circumferences were measured and body mass indices were calculated, as previously described [15]. Serum and plasma were separated from blood collected after an overnight-fast [16]. Serum glucose and plasma total and HDL cholesterol and triglyceride concentrations in a fasting state were measured using a Hitachi 7600 Automatic Analyzer (Hitachi, Tokyo, Japan), and plasma LDL concentrations were calculated using the Friedewald Equation (plasma LDL concentration = plasma total cholesterol concentration -plasma HDL concentration -plasma triglyceride concentration/5). Blood HbA1c contents were determined using a ZEUS 9.9 automatic analyzer (Takeda, Tokyo, Japan).
Education status was classi ed as less than high school, high school, and college or higher and smoking status as current-smokers, past-smoker, and non-smoker. Alcohol beverage types, intake amounts, and frequencies during the previous year were obtained from KoGES data and daily alcohol intakes were calculated by multiplying frequencies of alcohol drinking by the average amount of alcohol consumed daily. Participants were classi ed as light drinkers (< 0.3 g/day), moderate drinkers (0.3-2 g/day), or heavy drinkers (> 2 g/day). Daily coffee intake was assessed in the same manner and participants were categorized into three groups using daily coffee intake tertiles. Daily physical activity was estimated by summing physical activities and daily physical activity levels were classi ed as light, moderate, or heavy using tertiles.

Cataract Diagnosis
In the KoGES survey, subjects were asked the question "Have you ever been diagnosed with a cataract?" Those that answered "yes" were assigned to the case group and the others were included in the control group.

De nition Of Metabolic Syndrome
Metabolic syndrome was de ned as the presence of a cluster of ≥ 3 of the following: 1) abdominal obesity (waist circumference ≥ 90 cm for men and ≥ 85 cm for women); 2) elevated blood pressure (average systolic blood pressure ≥ 130 mmHg or diastolic blood pressure ≥ 85 mmHg); 3) low HDLcholesterol level (< 40 mg/dl for men and < 50 mg/dl for women); 4) elevated serum triglyceride level (≥ 150 mg/dl); and 5) elevated fasting blood glucose level (≥ 110 mg/dl). [17] Participants taking medication for dyslipidemia, hyperglycemia, or hypertension were included in the metabolic syndrome group.

Dietary Assessment And Dietary Patterns
Food intakes were calculated using responses to a semi-quantitative food frequency questionnaire (SQFFQ) composed of 106 food items that Koreans mainly consume. Frequencies and average serving sizes were checked for the 106 food items. Usual serving sizes of individual food items were rated as 0.5, 1, or 2-fold those of displayed reference serving sizes. The midpoint of each food frequency category was used for the frequencies of each food item. The average intake of each food item was calculated by multiplying average daily intake frequency by average selected serving size. The quality of the SQFFQ was validated using 3-day food records conducted by subjects [18]. SQFFQ data on dietary patterns of subjects were classi ed using principal component analysis (PCA). Daily intakes of energy and nutrients (e.g., protein, carbohydrate, and fat) were calculated using Can-Pro 2.0 nutrient assessment software developed by the Korean Nutrition Society [17].

Statistical analysis
Statistical analysis was performed using SAS version 9.4 (SAS Institute, Cary, NC, USA). The 106 food items in SQFFQ were categorized into 30 prede ned food groups (Supplemental Table 1) and were used as variables for factor analysis, which was performed using the FACTOR procedure. The number of factors in PCA was selected to retain eigenvalues of > 1.5 [19]. The orthogonal rotation procedure (varimax) and post-rotated factor loadings were used with PCA. The selected factors described distinct dietary patterns comprised of speci c foods and major contributory foods were indicated by factorloading values of ≥ 0.40 [20]. Factor scores for each pattern were determined by summing the intakes of all food groups weighted by factor loading. Factor scores of each dietary pattern were categorized into tertiles to compare lifestyle factors and nutrient intakes.
Frequency distributions of classi cation variables were calculated using the chi-squared test, and the two-sample t-test was used to determine the means and standard deviations of continuous variables, which included body mass index (BMI) and serum lipid pro les, in the case and control groups. The signi cance of intergroup differences was determined by one-way analysis of variance (ANOVA) with adjustments for covariates. Intergroup multiple comparisons were performed using the Tukey test.

Results
General characteristics of the study population according to the presence of cataracts The demographic characteristics of subjects with and without cataracts are summarized in Table 1. For subjects aged ≥ 45 years, 945 (3.3%) had a cataract and 27,454 (96.7%) did not. The mean age in the cataract group (54.7 years) was higher than in the control group (53.3 years). Women (57.6%) had much more cataract than men (42.4%). The prevalence of cataract was four times as high in those with a lower income (<$4,000) than in those with a higher-income (≥$4,000), was greater in subjects educated to middle school than in those educated to a higher level, and was greater in subjects with high activity levels than in those with low activity levels. Cataract prevalence was also higher in subjects aged > 60 years, in those that smoked, and in heavy alcohol consumers. Intakes of coffee, energy, and macronutrients were not signi cantly different in the cataract and non-cataract groups. Results are means ± standard deviation or number (%).

Adjusted Means Of Metabolic Parameters Related To Metabolic Syndrome
The frequencies of metabolic syndrome and obesity were signi cantly higher in the cataract group (32.2 and 37.3%, respectively) than in the non-cataract group (21.9 and 32.7%, respectively) (P < 0.0001; Table 2). However, mean BMIs and waist and hip circumferences were not signi cantly different in these two groups. Fasting plasma glucose concentrations and blood HbA1c levels were much higher in the cataract group (97.5 vs 94.3 mg/dL and 5.86 vs 5.71%, respectively; P < 0.0001; Table 2), but serum total cholesterol and LDL cholesterol levels were lower in the cataract group (Table 2). Serum HDL and triglyceride concentrations and systolic blood pressure and diastolic blood pressure were similar in the two groups, but group serum C-reactive peptide levels (an index of in ammation) were not signi cantly different ( Table 2).

Associations Between Cataract Risk And Socioeconomic And Metabolic Parameters
The adjusted ORs and 95% CIs of cataract risk were exhibited in models 1 and 2 according to the binary groups of socioeconomic and metabolic parameters after adjusting covariates ( Table 3). The covariates used in model 1 were age, sex, residence area, and body mass index and model 2 included model 1 covariates plus family income, education, smoking, physical activity, menopause, and energy, fat, protein, carbohydrate, alcohol, coffee, and cholesterol intakes. Subjects ≥ 55 years old had a 6.7-and 5.6-fold higher risk of cataract than subjects < 55 years old in models 1 and 2, respectively (Table 3). Sex was not signi cantly associated with cataract risk. Subjects educated to high school or more and with a monthly income of ≥ $1500 dollars had a lower cataract risk than those that had received less education and were earning less (Table 3).  3 The criteria for the high group when the values were dichotomized into two groups with the cutoff point. 4 The no-cataract group was the reference.
Subjects with metabolic syndrome were found to have 1.23-and 1.15-fold higher risks of cataract than those without, by models 1 and 2, respectively (Table 3). BMI and waist circumference were not associated with cataract risk, but fasting plasma glucose and HbA1c levels were signi cantly higher in those with a cataract than in those without by model 2 (OR = 1.920, 95% CI = 1.586-2.325 and OR = 2.071, 95% CI = 1.718-2.497; P < 0.0001). Interestingly, systolic blood pressure (≥ 140 mmHg), but not diastolic blood pressure (≥ 90 mmHg), was signi cantly associated with a 1.16-fold increase in cataract risk by model 1 (P = 0.0495) ( Table 3). Serum concentrations of total cholesterol, LDL, HDL, and triglyceride and serum C-reactive peptide were not associated with cataract risk by model 1 or 2 (Table 3).
Cataract incidence, nutrient intake, and lifestyles according to dietary pattern tertiles Based on food frequency results, dietary intakes of subjects were divided into four dietary patterns using PCA ( Table 4). The ve dietary patterns were; a traditional balanced diet (TB), a meat and sh diet (MF), a bread and cookie diet (BC), a grain-based diet (GB), and a coffee and alcohol diet (CA). Adjusted means of energy and nutrient intake are presented according to dietary pattern tertiles (T1, T2, and T3) in Table 3, after adjusting for age, sex, residence area, and BMI (Table 4). Cataract incidences were not signi cantly different between TB tertiles. Daily energy intakes were slightly but signi cantly higher in tertile 3 (T3) than in T1. Carbohydrate intake was lower and protein, fat, Ca, and coffee intakes were higher in T3 than T1 (Table 4). For TB, the prevalence of non-smokers and those with high physical activity levels were higher in T3 than in T1. Alcohol intake was not signi cantly different among the tertiles. For the MF diet, cataract incidence was lower in T3 than in T1. CHO, fat, and Ca intakes were lower but protein and alcohol intakes were higher in T3 than in T1, but daily energy intakes were not signi cantly different among tertiles ( Table 4). The prevalence of smoking was greater in T3 than in T1.
For the BC diet, the prevalence of cataracts was lower in T3 than in T1. Carbohydrate and fat intakes were higher, but protein, Ca, coffee, and alcohol intakes were lower in T3 than in T1. No signi cant difference between smoking prevalence was observed among the tertiles. For the GB diet, the prevalence of cataract was higher in T3 than in T1 (Table 4). A high intake GB diet was lower in CHO, protein, fat, and coffee and higher in Ca and alcohol, and CHO, fat, protein, and coffee intakes were lower but Na and alcohol intake were higher in T3 than in T1. No signi cant difference was observed between the GB tertile groups in terms of V-C, V-A, or β-carotene intakes (Data not shown). Adjusted with family income, education, smoke, physical activity, menopause, and intake of energy, alcohol, coffee, fat, protein, carbohydrate, and cholesterol.

Association Between Cataract Risk And Dietary Patterns
Those who frequently consumed a TB diet (a high-intake group) had signi cantly lower ORs (0.839; 95% CI = 0.700-0.999; P < 0.05) for cataracts than those that consumed a TB diet infrequently (reference group; a low-intake group) in model 1, but not in model 2 (Table 5). On the other hand, those with a high intake GB diet had higher ORs for cataracts than those with a low intake GB diet by model 1 (OR = 1.289) and model 2 (OR = 1.291). MF and BC diet patterns did not exhibit an association with cataract risk (Table 5). 2 Model 2: model 1 plus income, education, smoke, physical activity, menopause, and intake of alcohol, coffee, fat, protein, carbohydrate, and cholesterol. 3 The cut-off points of the low-, medium-, and high-intake groups were assigned by the tertiles.
Daily energy intake was not associated with cataract risk by model 1 or 2. However, the percentage of macronutrient intake in uenced cataract risk. A high carbohydrate intake (energy %) had a 1.42 higher risk of cataract by model 1 but was not found to present a higher risk by model 2 (Table 5). High intakes of fat, protein, and Ca were associated with lower cataract risks by 0.74-, 0.74-, and 0.73-fold by model 1 but no association was indicated by model 2. However, no signi cant association was observed between cataract risk and physical activity or Na or coffee intake by model 1 or 2 (Table 5). Moderate alcohol intake, but not heavy alcohol intake, lowered cataract risk by 0.80 fold as compared with low alcohol intake by model 1 (P < 0.05) and 2 (P < 0.001). Interestingly, former smokers had a higher risk of cataract than non-smokers by models 1 and 2 (Table 5).

Discussion
The prevalence of cataracts among the elderly has increased worldwide, and cataract surgery is the main therapeutic method used to prevent vision loss. In the present study, we found that metabolic syndrome was positively associated with cataract prevalence after adjusting for confounding factors. Furthermore, cataract prevalence was positively associated with grain-based dietary patterns and negatively associated with a traditional balanced diet. High carbohydrate intake was associated with a 1.4-fold increase in cataract risk and high fat, protein, and Ca intake was in association with a 0.74-fold decrease in model 1. These results suggest that a balanced diet containing su cient protein and Ca protect against cataract development in middle-aged adults.
However, further analysis of results showed that relations between cataract prevalence and the components of metabolic syndrome differed from previous ndings. For example, in the present study, BMI, waist circumferences, systolic and diastolic blood pressures, hypertension, lipid pro les (LDL, HDL, and triglyceride) were not signi cantly associated with cataract prevalence. Total cholesterol and LDL levels were signi cantly lower in the cataract group than in the non-cataract group, although this association disappeared after adjusting for confounding factors in logistic regression analysis. Although the exact reason is unclear, one possible explanation for this observation is that patients with cataract might have taken cholesterol-lowering drugs (e.g., statins), given that patients with a cataract may also have dyslipidemia. However, the pathogenic and protective effects of statins on cataracts are poorly documented and widely debated [28,29]. It was recently reported that the prolonged use of statins may predispose cataract development [30], but others failed to identify an association [31]. Unfortunately, the administration of cholesterol-lowering drugs was not included in the analysis in the present study. Further studies are needed to determine whether cholesterol-lowering drugs might be associated with cataractogenesis.
BMI and abdominal obesity (as determined by waist circumference) have been suggested to be risk factors of cataract development. Cheung and Wond have reported that obesity was positively associated with the presence of cataract [32], but other studies disagreed. In a study conducted in Singapore obesity was found to be positively related to the prevalence of cortical cataracts but negatively related to that of nuclear cataracts [21]. Another study based on the Korean National Health and Nutritional Examination Survey data showed that overweight was signi cantly associated with a lower risk of cataract development [22]. However, we found no association between obesity and cataract prevalence.
The present study revealed signi cant associations between cataract prevalence serum glucose levels and hemoglobin A1c levels, which is consistent with previous ndings [33,34]. In particular, we observed linear associations between hemoglobin A1c and fasting serum glucose levels and the presence of cataracts using generalized linear models, which suggests high fasting glucose levels and diabetes are risk factors of cataract development. It has also been suggested that advanced glycation of lens ber proteins [35], hyperosmotic pressure, and the sorbitol pathway are possible mechanisms of cataract formation [36].
Tryptophan/selenium imbalance and calcium, vitamin C, vitamin A, and excessive energy intakes are generally considered to be cataract risk factors [37]. However, we found vitamin C, vitamin A, and carotene intakes were not associated and that calcium intake was negatively associated with cataract risk. The average intake of vitamins A and C and carotene was higher than DRI in both case and control groups. However, the average intake of calcium intake (~ 449 mg/day) in our subjects was ~ 50% of DRI and calcium intakes were too low to enhance cataract risk [38]. Furthermore, in the present study, macronutrient intake was observed to be associated with cataract risk that is consistent with metabolic syndrome risk [39,40]. A high carbohydrate diet (> 70 energy %) and a low-fat diet (< 15 energy %) were found to be associated with metabolic syndrome in a previous using KNHANES data. [40,41] The present study showed a consistent result in cataract risk. Metabolic syndrome is known to be associated with cataract risk [42] and macronutrient nutrient intake showed similar patterns for the risk of both diseases. The present study shows that high CHO, low fat, and low protein diets increased the risk of cataract in middle-aged adults.
The present study has several limitations. First, the presence of cataracts was evaluated based on selfreports and not on clinical examinations, which may have caused bias. Second, KoGES data did not include details of current cataract status or information on the cataract severity or subtype. Third, the data obtained could not be used to determine causalities because of the cross-sectional design of KoGES. Forth, nutrient intakes and life-styles were determined using self-reported questionnaires, and thus, were prone to subjective errors. However, it should be added that the semi-quantitative food questionnaire used contained 106 commonly consumed 106 food items and was validated using 3 dayfood records over 4 seasons.

Conclusion
Summarizing, the present study shows an association exists between metabolic syndrome and the presence of cataract in a large hospital-based cohort, and that fasting serum glucose was a strong and independent risk factor of cataract, whereas other components of metabolic syndrome (i.e., obesity, lipid pro les, and hypertension) were not. Furthermore, the study shows strict glucose control is more important terms for preventing cataract formation than blood pressure or dyslipidemia control.