Lens Aging in Relation to Nutritional Determinants and Possible Risk Factors for Age-Related Cataract FREE

AGE-RELATED CATARACT is the leading cause of visual disability.¹ Several nutritional determinants possibly influence age-related cataract. Vitamins C and E and the carotenoids lutein and zeaxanthin are present in human lenses, with vitamin C at concentrations severalfold that of plasma.^{2 - 4} They all can reduce oxidative stress, which influences the development of cataract.^{5 - 7} Low vitamin C content in the lens seemed to be a good indicator of cataract severity.⁸

Plasma levels of vitamin C were not associated with risk of cataract.^{9 - 11} Studies^{9 - 12} on a possible association between plasma levels of vitamin E and risk of cataract showed conflicting results. Self-reported consumption of vitamin C and E supplements was associated with a reduced risk of cataract.^{13 - 16} However, studies^{17 - 21} on consumption of vitamin C supplements showed controversial results. Conflicting results were also reported for multivitamin or vitamin E use.^{22 - 24}

Intake of foods rich in lutein and zeaxanthin, like spinach and broccoli, was associated with a lower relative risk of cataract.^{12 ,17 ,25 - 29} Lutein and zeaxanthin are also components of macular pigment in the eye.^{30 - 31} A significant association has been found between macular pigment optical density (MPOD) and concentrations of serum and adipose tissue lutein,^{32 - 33} be it that there are significant differences between men and women.^{34 - 35} Furthermore, intake of lutein modifies the MPOD.^{32 ,36 - 37} The MPOD might be a better measure of the lutein and zeaxanthin content in the eye than serum and adipose tissue values of carotenoids, although there is no independent evidence to support this hypothesis.

Besides nutritional determinants, other factors might influence cataract. Smoking has been identified to increase cataract risk.^{16 ,22 ,38 - 43} However, smoking is also associated with diminished carotenoid⁴⁴ and vitamin C⁴⁵ status, and there seemed to be a diminished risk for cataract in smokers who used multivitamins compared with smokers not using multivitamins.⁴⁶ Height and abdominal adiposity are independent risk factors for cataract.^{47 - 48} The association between body mass index (BMI) and cataract is controversial.^{22 ,49 - 54} Subjects with darker-colored irises tend to have a higher risk of cataract.^{55 - 57} Finally, women have a somewhat higher incidence of cataract than men.^{58 - 61}

There is a strong relation between lens aging and cataract, the latter being the end stage of deterioration of the aging lens.^{62 - 64} Lens senescence can be quantified by the increase in lens optical density (LOD), which has been studied extensively.^{62 ,65 - 70} Although these studies showed strong positive associations between LOD and age, there are substantial individual differences at each age. This study investigates whether nutrition and the possible risk factors for cataract previously mentioned could account for part of these individual differences. Fundus reflectance yields reliable estimates of the LOD.^{67 ,71} It was used in this study, having the advantage that the MPOD was obtained simultaneously.

The subjects in our study were recruited from the pool of volunteers of TNO Nutrition and Food Research and through advertising in local and regional newspapers and on television. The cross-sectional data analyzed herein were gathered during the baseline measurements of an intervention study designed to test an unrelated hypothesis. Advertising was aimed at recruiting people who have a general interest in participating in nutrition studies. Respondents who expressed potential interest received a questionnaire eliciting data on daily fruit and vegetable consumption and lifestyle factors. This questionnaire was designed for ranking subjects according to their intake and not for quantitative measurements. Only respondents in the highest quintile of fruit and vegetable consumption and those in the lowest tertile (calculated in the total group) were invited for an oral briefing and screening. In the lowest stratum, vitamin supplement users were excluded. The major inclusion criteria were age between 18 and 75 years and consumption of an average Western diet. The main exclusion criteria were pregnancy and/or lactation, wanting to become pregnant, serum cholesterol level greater than 290 mg/dL (>7.5 mmol/L) and/or triglyceride level greater than 204 mg/dL (>2.3 mmol/L) if not undergoing stabilized hypercholesterolemia or hyperlipidemia treatment, and anticoagulant therapy. A total of 380 volunteers were enrolled in the study. Two volunteers squinted too much during the reflectance measurement. Furthermore, one measurement failed for other reasons and one subject withdrew from the study before the reflectance measurement. Data from these 4 subjects were not included in the statistical analyses. Eight subjects were measured in the left eye instead of the right eye. Their measurements were included in the analysis, because there is a good correlation between both eyes.⁷² The final study group comprised 177 men and 199 women. The study was performed according to the International Conference on Harmonization of Technical Requirements of Registration of Pharmaceuticals for Human Use guidelines for good clinical practice, and was approved by a medical ethical committee. Informed consent was obtained from all subjects.

Blood samples from subjects were obtained between 8 AM and 9:30 AM, after an overnight fast. For the analysis of carotenoids and α-tocopherol, blood was collected in tubes containing clot activator and gel (Vacutainer Systems; Becton Dickinson, Franklin Lakes, NJ). These tubes were immediately stored in a closed box to avoid breakdown of carotenoids by UV light. Tubes were centrifuged within 15 to 30 minutes after collection at approximately 2000g for 10 minutes at approximately 4°C to obtain serum. After centrifugation, serum was removed and stored at approximately−80°C. Before storage, all serum was handled in subdued light.

For the analysis of vitamin C, blood was collected in tubes containing lithium heparin (Vacutainer Systems). A 0.5-mL aliquot of blood was added to 2 mL of metaphosphoric acid (50 g/L) (Mallinckrodt Baker, Deventer, the Netherlands) before freezing to preserve the vitamin C concentration during storage. This mixture was stored at approximately −80°C.

For the analysis of total cholesterol, blood was collected in tubes containing clot activator and gel (Vacutainer Systems). Tubes were centrifuged within 15 to 30 minutes after collection at approximately 2000g for 10 minutes at approximately 4°C to obtain serum. After centrifugation, serum was removed and stored at approximately −20°C.

Subcutaneous adipose tissue was obtained by needle biopsy from the lateral buttock,⁷³ using a 16-gauge needle attached to a plastic container in which the tissue was collected by connecting a vacuum tube. Samples were kept in the plastic container, immediately placed on dry ice, and stored at approximately −80°C.

All carotenoids, α-tocopherol, and vitamin C were analyzed by high-performance liquid chromatography. The coefficients of variation for the analysis of lutein, zeaxanthin, and α-tocopherol were 3.7%, 11.5%, and 2.8%, respectively (at mean concentrations of 0.16, 0.03, and 20.90 µmol/L in the quality control samples, respectively). The coefficient of variation for the analysis of vitamin C was 10%. Total cholesterol level was analyzed by enzymatic conversion to a stable chromogen, which could easily be detected by colorimetry with a commercially available kit (Boehringer Ingelheim, Mannheim, Germany). Broekmans et al⁷⁴ provide more details.

Spectral fundus reflectance was measured with a densitometer (Utrecht Retinal Densitometer).⁷⁵ Briefly, a rotating wheel (14 revolutions per second) offered a sequence of 14 interference filters in the range of 430 to 740 nm to enable quasi-simultaneous measurement of the reflectance across the visual spectrum. Light reflected from the fundus was measured in a detection field of 1.5°, concentric within the illumination field of 1.8°. To obtain an estimate of the mean LOD and MPOD in this area, spectral fundus reflectance was measured under 2 conditions: (1) with the instrument's entry and exit pupils aligned to the peak of the Stiles-Crawford function and (2) 2 mm temporal to the Stiles-Crawford peak.⁷¹ A detailed optical model of foveal reflection was used to arrive at individual estimates of variables such as the equivalent thickness of the blood layer, MPOD, LOD, and optical density of melanin.⁷¹ In short, the incoming light was assumed to reflect at the inner limiting membrane, the discs in the outer segments of photoreceptors, and the sclera. By using known spectral characteristics of the different absorbers within the eye (lens, macular pigment, melanin, and blood), the optical densities of the pigments and the percentage reflectance at the interfaces were optimized to fit the measured data at all wavelengths. The spectral characteristic of the LOD was composed of a nonaging and an aging part.⁶⁵ Only the aging part was optimized in the fit procedure. The subjects' pupils were dilated with 0.5% tropicamide.

Smoking status was obtained by a questionnaire. Iris color was graded by a set of 4 standard photographs, providing a 5-grade classification system.⁷⁶ Weighing the subject wearing indoor clothing, without shoes, wallet, and keys, assessed body weight. Height was measured without shoes. The BMI was calculated as weight in kilograms divided by the square of height in meters.

The Statistical Product and Service Solutions software package, version 8.0.2 (SPSS Inc, Chicago, Ill), was used for data analysis. Pearson product moment correlation coefficients were calculated between continuous variables. To adjust for age effects, we also calculated partial correlation coefficients, controlling for age. t Tests and an analysis of variance were used to evaluate possible differences in continuous variables between men and women, between smokers and nonsmokers, between subjects with high and low fruit and vegetable intake, and between subjects with different iris colors. χ² Tests were used to evaluate possible differences between categorical baseline characteristics (sex, iris color, smoking status, and fruit and vegetable intake). If variables were not normally distributed, they were log transformed. Regression analysis was used to evaluate the association between categorical variables (sex, iris color, and smoking), covariates (age, MPOD, vitamins, and lutein, zeaxanthin, and cholesterol levels), and LOD.

The mean ± SD LOD at 420 nm was 0.52 ± 0.17. Study group characteristics are given in Table 1 and Table 2. For ranking prestudy, the participants were classified into those with a diet low in fruit and vegetables, fruit juices included (57% of the study sample), and those with a diet rich in these foods (43% of the study sample). The latter group consisted of significantly more women (62%) than the former (46%) (P = .001). The percentage of male smokers (32%) did not differ from that of female smokers(27%) (P = .33). The LOD did not differ significantly between men and women, between the groups with high and low fruit and vegetable intake, or between smokers and nonsmokers (Table 1).

Iris color was classified in 331 participants. The participants predominantly(65%) had blue or gray irises. The distribution of subjects in the 5 iris categories did not differ between men and women (P =.72). The LOD did not differ among the iris categories (Table 1).

Serum concentrations were significantly lower in men than in women for lutein, zeaxanthin, vitamin C, and total cholesterol. The lutein concentration in adipose tissue was also significantly lower in men than in women. The MPOD was significantly higher for men than for women (Table 2).

Age correlated most significantly with LOD (r =0.71, P<.001), explaining 50% of the variance. Furthermore, serum and adipose lutein levels, serum α-tocopherol level, height, BMI, and total cholesterol level also correlated significantly with LOD (Table 2). Other previously mentioned characteristics (MPOD and serum zeaxanthin and vitamin C levels) showed no association with LOD. We also found a significant correlation between age and serum lutein, serum α-tocopherol, total cholesterol levels, BMI, and height (Table 2). Therefore, we calculated the partial correlation coefficient between all variables and LOD, controlling for age (Table 2). All previously mentioned correlations disappeared. Only MPOD showed a small significant negative correlation with LOD (r = −0.12, P<.05). This could imply that MPOD provided more detailed information on lutein and zeaxanthin content in the lens than on serum lutein, serum zeaxanthin, and adipose tissue lutein content. To test this hypothesis, we used linear regression analysis to identify factors that are associated most significantly with LOD. We used the forward conditional method with default selection criteria (P<.05 to enter and P>.10 to remove). Age, MPOD, and serum lutein, serum zeaxanthin, adipose tissue lutein, serum vitamin C, and α-tocopherol levels were used as independent variables and LOD as the dependent variable. Total cholesterol level was included because of its correlation with serum carotenoids, α-tocopherol, height, and BMI. Height, BMI, sex, smoking status, and iris color were also included. We only found significant β coefficients for age (mean ± SD, .008 ± .0004 [P<.001]; 95% confidence interval [CI], .008-.009) (16% per 10 years) and for MPOD (mean ± SD,−.096 ± .041 [P = .02]; 95% CI, −.177 to −.016) (−2.7% per SD in MPOD).

We stratified our data (≤50 and >50 years) and applied the regression analysis on both strata. The mean ± SD LOD was 0.68 ± 0.16 for the older subjects and 0.45 ± 0.11 for the younger subjects. The mean± SD MPOD was 0.34 ± 0.16 and 0.32 ± 0.14 for the older and younger subjects, respectively. For the older subjects (n = 114), we found significant β coefficients for age (mean ± SD, .011 ± .002[P<.001]; 95% CI, .007-.015) and MPOD (mean ± SD, −.240 ± .086 [P = .005]; 95% CI,−.410 to −.071) (−5.6% per SD in MPOD). For the younger subjects (n = 262), we only found a significant β coefficient for age(mean ± SD, .006 ± .0007 [P<.001]; 95% CI, .004-.008), and there was no association between LOD and MPOD.

As expected, the LOD increased significantly with age. Furthermore, the MPOD showed an inverse relationship with LOD after adjustment for age. The inverse association between MPOD and LOD could reflect the greater likelihood of subjects with a high MPOD to have other healthy behaviors that retard lens aging. We found a significant, but small, correlation between MPOD and serum vitamin C level (r = 0.12, P =.02). However, serum vitamin C level and other markers for a healthy lifestyle, which are possible risk factors for age-related cataract, like BMI and smoking, have been included in the regression analysis. Only MPOD showed a significant contribution.

Similar to Hammond et al,⁷⁷ we stratified our data (≤50 and >50 years). Only for the older subjects, we found a significant association between LOD and MPOD. To study this in more detail, we added an interaction term (age × MPOD) in the regression analysis. However, the interaction term did not contribute significantly. It could be due to a nonlinear dependence of the LOD on age, not accounted for in the linear model analysis. Pokorny et al⁶⁵ determined, for subjects older than 60 years, a 3-fold increase in the rate at which the LOD changes as a function of age. In this study, we found the mean ± SD β coefficient for age to be 1.76 ± 0.39 times larger for subjects older than 50 years than for the younger subjects. For a cut point at age 60 years, we found the mean ± SD β coefficient for age to be 2.09 ± 0.81 times larger for the older subjects than for the younger subjects. We added an extra quadratic age term in the general linear model analysis, which indeed turned out to contribute significantly. However, the age × MPOD interaction term remained nonsignificant. Thus, in the total study group, we could not find a significant change in the regression coefficient for the MPOD with age, although in the stratified data the regression coefficient was only significant for the older subjects.

Like others,^{34 ,78} we found a significantly higher MPOD in men than in women, despite lower serum and adipose tissue levels. Furthermore, like others,^{34 - 35} we also found a higher correlation between MPOD and concentrations of serum and adipose tissue lutein in men than in women. Thus, the ability to transport lutein and zeaxanthin from the blood into the eye might be less in women than in men. It may be the reason that we found the LOD to show an inverse association with MPOD, rather than with serum and adipose tissue values of lutein and zeaxanthin. It may also be an explanation for the higher incidence of cataract in women than in men, apart from the impact of estrogen.⁷⁹

We found no difference in LOD between smokers and nonsmokers. Hammond et al⁸⁰ studied LOD in relation to smoking behavior in younger subjects (n = 41) without cataract. They found a significant dose-response relationship between smoking frequency and LOD. Their data indicate that smoking is directly related to age-related increases in LOD throughout life, and that these increases persist even after smoking cessation. We had no information on duration of smoking and no information of ex-smokers. This may have lessened the strength of our analysis.

Hammond et al⁸¹ also found significantly higher LODs in subjects with dark irises compared with subjects with light irises. We similarly divided our study group and found a mean ± SD LOD of 0.52 ± 0.17 for subjects with light irises (n = 268 [the first 2 categories in Table 1]) and of 0.54 ± 0.17 for subjects with dark irises (n = 63 [the last 3 categories in Table 1]). Although the LOD was higher in subjects with dark irises, this difference was not significant(P = .42). Hammond et al found only differences in LOD between light and dark irises for older subjects (>45 years). Younger subjects showed no differences in LOD. We could not find any age effect in our study group. Model analysis of the reflectance spectra yielded a mean± SD retinal melanin optical density of 0.87 ± 0.18 for subjects with light irises and of 1.02 ± 0.27 for subjects with dark irises(P<.001). The spectral absorbance of melanin and the aging component of the LOD show a slow decrease with increasing wavelength, without any characteristic spectral features. However, because of choroidal blood absorption, melanin shows its spectral fingerprint only for wavelengths greater than 600 nm, whereas the LOD is determined for wavelengths of less than 600 nm. As a consequence, their absolute values can be determined independently and reliably, and there is no indirect covariance between iris color and LOD from our variable fitting.

In summary, the LOD showed a significant association with age and MPOD. Macular pigment is composed of lutein and zeaxanthin, the only carotenoids found in human lenses. The MPOD is associated with serum and adipose lutein levels and can be modified by lutein intake. Inverse relationships between MPOD and LOD suggest that lutein and zeaxanthin in the eye may retard aging of the lens.

Submitted for publication February 5, 2002; final revision received July 1, 2002; accepted August 6, 2002.

Corresponding author and reprints: Tos T. J. M. Berendschot, PhD, Department of Ophthalmology, Universitair Medisch Centrum Utrecht, AZU E03.136, Heidelberglaan 100, PO Box 85500, NL-3508 GA Utrecht, the Netherlands (e-mail: tosb@isi.uu.nl).