The Association of Cataract With Leukocyte Telomere Length in Older Adults: Defining a New Marker of Aging

Abstract

Lens transparency, or the magnitude of cataract severity, is a potential in vivo marker of aging distinguishable from diagnosed cataract. To explore lens transparency as a marker of aging, we determined its association with leukocyte telomere length (LTL) measured with quantitative polymerase chain reaction. Cataract severity was directly measured in 259 participants, and prevalent cataract and incident cataract surgery were ascertained in 2,750 participants of the Health, Aging, and Body Composition Study. LTL was unassociated with clinical cataract outcomes. Six of 259 had successfully aged lenses and a mean LTL of 5,700 bp, whereas 253/259 with poorly aged lenses had a mean LTL of 4,770 bp. Participants with a 1,000 bp greater mean LTL had nearly half the odds of any cataract (odds ratio = 0.47, 95% confidence interval 0.22–1.02) after adjustment. Lens transparency might be associated with longer LTL in community-dwelling older adults and should be investigated further as a possible biomarker of aging.

INDIVIDUAL health is highly variable. During their lifetimes, some individuals experience disease, accidents, and poor health; others seemingly escape these burdens. Underlying this variability is a single process that all living things experience, aging. It is critical to determine how we age if we hope to optimize the well-being of individuals. Specifically, markers of primary aging are needed as intermediate outcomes to understand the aging process and potential early benefits of preventive interventions.

The human lens is a marker of interest for several reasons. Lens changes, such as greater thickness and reduced transparency, can be detected in vivo in young adulthood, prior to the onset of disease and disability, and can be distinguished from known diseases of the lens, such as cataract (1–4). Lens proteins (crystallins) are set down in fetal development and are the only proteins in the body that do not turn over during life. Thus, changes in lens proteins due to aging are not obscured by removal or repair processes and may reflect lifelong exposure. Furthermore, the same tissue can be repeatedly measured. Crystallins are natively clear proteins that belong to the family of chaperone proteins, which critically maintain protein conformation, particularly under stress, throughout the body (5–7). Disruption of lens crystallins, which leads to decreased transparency, may therefore reflect widespread breakdown of maintenance machinery, a process that underlies aging across tissues (6,7). Indeed, the absence of crystallins that are normally found in the lens, brain, heart, skin, and skeletal muscle is associated with aging phenotypes (8–10). Because crystallins are natively clear, lens transparency can also be measured quickly, accurately, noninvasively, and nonhazardously (3,11,12). These qualities make lens transparency an attractive candidate marker of aging, though it has been nearly uninvestigated in population aging research.

To explore lens transparency as a marker of aging requires determining its association with other potential aging markers. Leukocyte telomere length (LTL), which shortens with age, inflammation, and oxidation, has become an increasingly prominent marker of aging (13). It has been associated with longevity and some age-related outcomes independent of chronic disease in some, but not all, community-dwelling cohorts examined (14–17). Subsequently, LTL could be a useful marker against which to correlate lens transparency. The hypothesis that lens transparency may be associated with LTL is also supported by the longstanding observation that the vast majority of individuals with the premature aging disorder Werner syndrome, which is characterized by accelerated telomere shortening, have early onset of bilateral cataracts (18,19).

In this analysis, we sought to explore the human lens as a possible marker of aging by studying the association of cataract with LTL in a large cohort of community-dwelling older adults, the Health, Aging, and Body Composition Study (Health ABC), and an ophthalmic substudy of Health ABC, the Age-Related Maculopathy Ancillary Study (ARMA). In doing so, we distinguish lens transparency, a possible primary aging phenotype, from cataract, a disease phenotype. We specifically hypothesized that (a) longer LTL is associated with a lower prevalence of cataract, (b) longer LTL is associated with a lower incidence of cataract surgery, and (c) longer LTL is associated with a lower odds of lens opacity.

METHODS

Health ABC Study Population

We used the Health ABC study population to examine prevalence of cataract and incident cataract surgery. All Medicare-eligible people in Memphis, TN, and Pittsburgh, PA, were identified to volunteer in the Health ABC study. The study also sought an enriched sample of African Americans. Because the Health ABC study was originally designed to examine incident mobility disability, eligibility criteria were no reported difficulty in walking for 1/4 mile, walking up 10 steps, getting in and out of bed or chairs, bathing or showering, dressing, or eating; no need of using a cane, walker, crutches, or other special equipment to get around; not enrolled in a lifestyle intervention trial; free of life-threatening illness; and had plans to stay in the geographic area for ≥3 years. Due to these criteria, the study population is slightly healthier than the age-matched general population. The University of Pittsburgh and the University of Tennessee Institutional Review Boards approved all procedures related to Health ABC, and all research adhered to the Declaration of Helsinki. For the cross-sectional analysis of the association of baseline cataract diagnosis with LTL, the cohort included 2,750 Health ABC participants (41% black, 48% male, aged 70-79 years) who had available LTL measurements (89.4% of the original Health ABC cohort). The prospective analysis of incident cataract surgery included all those who did not report a diagnosis of cataract in either eye at baseline (n = 1,505).

ARMA Study Population

We used the ARMA study, an ophthalmic substudy within Health ABC, to examine lens transparency. The ARMA study included sensitive measures of lens transparency and recruited participants from the Health ABC site in Memphis, TN, to take part in a study of the role of carotenoids and inflammation in macular aging and the pathogenesis of age-related maculopathy. ARMA participants consisted of an enriched sample of participants originally enrolled in the Health ABC study examined between Year 6 and Year 7 of the Health ABC study. Recruitment was based on stratification by a cumulative vision score from the Health ABC Year 3 tests (ie, based on the outcome of binocular best corrected visual acuity to Bailey–Lovie charts, binocular contrast sensitivity measured with Pelli–Robson charts, and stereoacuity characterized with the Frisby test) (20). Inclusion criteria were non-random and included self-reported diagnosis of age-related macular degeneration (n = 55); best vision function score (n = 100); or worst vision function score (n = 200). Exclusion criteria were diagnosis of type 1 or type 2 diabetes of ≥6 years duration or presence of diabetic retinopathy at the time of the eye exam; known history of glaucoma, or documented presence thereof on examination; diagnosis of autoimmune disease or monoclonal gammopathy; and illiteracy based on the inability to perform visual acuity tests, which negated reliable calculation of vision score, unless the patient was referred for diagnosis of age-related macular degeneration. Our cross-sectional analysis of lens transparency included all 259 Health ABC participants who had their mean LTL determined at baseline and who also had their lens transparency measured as part of the ARMA study. Though ARMA participants were non-randomly selected from the Health ABC study Memphis cohort using vision scores, selection was unassociated with baseline mean LTL (p = .76), age, gender, smoking history, body mass index (BMI), total cholesterol, and history of hypertension, heart disease, or cancer (all p > .05). However, ARMA participants were less likely to be black, more likely to use statins, and had lower mean high-density lipoprotein, oxidized low-density lipoprotein (oxLDL), C-reactive protein, interleukin-6, and fasting glucose (all p < .05), which was expected given the ARMA selection criteria.

Cataract Ascertainment and Lens Transparency Measures

At the baseline Health ABC clinic visit, trained interviewers asked participants if a physician ever told them that they had a cataract in one eye (unilateral cataract) or both eyes (bilateral cataract). We analyzed unilateral and bilateral cataract independently as measures of prevalent baseline cataract. Because retrospective determination of cataract diagnosis could introduce recall bias, we also analyzed incident cataract surgery documented prospectively by trained interviewers at annual clinic visits and at 6-month intervals with a home, phone, or proxy questionnaire.

Screening for ARMA included a dilated anterior segment exam by one observer (A.I.) to verify the presence of the crystalline lens and to ascertain the presence of cataract (nuclear, cortical, or both). Lens transparency was graded according to the validated Age-Related Eye Disease Study (AREDS) criteria (21). AREDS grading was accomplished by comparing an individual’s lens transparency measured using a slit lamp at the time of examination to standard graded photographs. The AREDS grading system for nuclear cataract is a semi-continuous, 8-step–based scale, with one indicating complete transparency and eight indicating complete cataract, and where opacity of intermediate severity between steps can be classified by 0.5 intervals. The AREDS scale for anterior cortical and posterior cortical cataract is measured as the percent of the viewing field occupied by the opacity (0%–100%).

Telomere Measurement

DNA was extracted from isolated peripheral blood mononuclear leukocytes. Average LTL was measured using a validated quantitative polymerase chain reaction method, as described previously (22). This method measures the relative average LTL in genomic DNA by determining the ratio of telomere repeat copy number to single-copy gene copy number (T/S ratio) in experimental samples relative to a reference sample. The coefficient of variation for the relative T/S ratios was 5.8%. Assuming a normal distribution for relative T/S ratios in repeated measurements of the same sample, samples differing in average LTL by as little as 11.4% should be distinguishable by this method at the 95% confidence level. A T/S “ratio unit” of 1.0 measured by the quantitative polymerase chain reaction method is equivalent to a mean LTL of 4,270 bp, the known LTL of the reference DNA used in this study. Thus, the LTL unit presented was converted to base pairs by using this conversion factor.

Covariates

Trained interviewers administered the baseline Health ABC questionnaire between April 1997 and May 1998 to assess demographic and socioeconomic characteristics, health behaviors, health status, and medical history. Within 2 weeks of the interview participants visited the University of Pittsburgh or University of Tennessee clinics for baseline biologic, anthropometric, and functional measures and a blood draw. We selected possible confounders measured at baseline based on those documented in the literature. Demographic characteristics included age, gender, race (white or black), and study site. Smoking was categorized as never, former, or current. BMI (kilograms per square meter) was calculated using baseline height and weight. The prevalence of physician-diagnosed diabetes, ischemic heart disease, congestive heart failure, cerebrovascular disease, and cancer (excluding non-melanoma skin cancer) was determined using algorithms based on self-reporting and medication use. Diabetes was defined as an elevated fasting glucose (≥126 mg/dL) or abnormal 2-h oral glucose tolerance test (≥200 mg/dL). Blood pressure was measured with the participant seated for 5 minutes and then 1 minute after standing. Participants were asked to bring all prescription and over the counter medications used in the previous 2 weeks. Medications were coded using the Iowa Drug Information System (23). Assay reproducibility was checked using a blind duplicate system whereby 5% of aliquots had a duplicate aliquot prepared. Blind duplicate aliquots were analyzed simultaneously, and results were matched to determine the Pearson correlation coefficient and coefficient of variation.

Statistical Analysis

We assessed whether covariates could be potential confounders by quantifying their association with LTL, cataract, cataract surgery, or lens transparency using Pearson correlation coefficients, the Student's t-test, χ² statistic, and analysis of variance. We chose the significance level p < .10 as a cutoff for inclusion of possible confounders in multivariate models. Age, BMI, cholesterol, high-density lipoprotein, LDL, oxLDL, C-reactive protein, interleukin-6, fasting insulin, and fasting glucose were treated as continuous covariates, whereas gender, race, study site, smoking (ever/never), comorbidities, and medication use were treated categorically. oxLDL, C-reactive protein, interleukin-6, and fasting insulin levels were log transformed in models to account for skewness, though nontransformed descriptive statistics are reported.

We added LTL to all models in kilobasepairs such that ratio measures reflect the odds of outcome associated with a 1,000 bp increase in mean LTL. LTL was measured on 45 quantitative polymerase chain reaction plates so we included a random effect variable in all models to account for slight LTL measurement variation between plates. For analyses of cataract in Health ABC, we used logistic regression to generate odds ratios and 95% confidence intervals to determine the association between LTL (predictor) and baseline cataract (outcome). We employed multivariate logistic regression for adjustment. We used a frailty model for Cox proportional hazards regression to calculate the hazard ratio for incident cataract surgery (outcome) using LTL as the predictor. Multivariate proportional hazards regression was used to adjust for confounding.

Because we were interested in exploring the lens as a marker of successful aging, in the ARMA study, we decided a priori upon a stringent classification of a truly transparent lens to clearly distinguish successful lens aging from cataract. Using the AREDS grading criteria, we defined successful lens aging as ≤3.0 nuclear grade and 0% cortical grade. An aged lens was defined as having nuclear opacity >3.0 grade or cortical opacity >0% grade or previous placement of an intraocular lens (n = 102). Specific types of cortical opacity were also classified using the AREDS criteria. The reference group for all logistic models of lens transparency was participants with a transparent lens (≤3.0 nuclear grade and 0% cortical grade, n = 6). We used logistic regression and multivariate logistic regression for all ARMA analyses.

We simultaneously tested for interaction between LTL and all included covariates using a significance level of 0.10. Nonsignificant interactions were excluded from the final multivariate models. We used a significance level of 0.05 except as noted earlier. SAS 9.1 (SAS Institute, Cary, NC) was used for all analyses except proportional hazards regression (Stata 10.0, StataCorp, College Station, TX).

RESULTS

Prevalent Cataract

Mean LTL at baseline for all participants was 4,860 bp (SD, 1,360 bp). Though correlations were weak (all r < .103), LTL was shorter with greater age, BMI, oxLDL, interleukin-6, fasting glucose, and fasting insulin and longer with greater high-density lipoprotein and total cholesterol (all p < .10). At baseline, mean LTL was 377 bp greater in women (p < .0001), 107 bp greater in blacks (p = .043), and 131 bp greater in statin users (p = .016). Smoking history was associated with shorter LTL (p = .0008).

At baseline, 36.4% and 25.6% of participants reported a history of unilateral and bilateral cataract, respectively. Both unilateral and bilateral cataract was significantly associated with age and gender (Table 1, bilateral cataract data not shown). Participants reporting a history of unilateral cataract had greater BMI, were more likely to be from Pittsburgh, and had a greater burden of comorbidities. Crude analyses illustrated no association between LTL and unilateral or bilateral cataract (Table 2). After adjustment for relevant confounders, the magnitude of effect increased slightly but remained nonsignificant.

Table 1.

Baseline Characteristics of the Health ABC Study Population by Prevalent Cataract at Baseline or Incident Cataract Surgery

	Unilateral Cataract		p*	9-y Cataract Surgery		p*
	Yes (n = 984)	No (n = 1,719)		Yes (n = 353)	No (n = 1,152)
	M (SD) or n (%)	M (SD) or n (%)		M (SD) or n (%)	M (SD) or n (%)
Age (y)	74.1 (2.9)	73.4 (2.8)	<.0001	73.5 (2.8)	73.1 (2.7)	.05
Gender
Male	428 (43.5)	880 (51.2)	.0001	157 (44.5)	652 (56.6)	<.0001
Female	556 (56.5)	839 (48.8)		196 (55.5)	500 (43.4)
Race
White	584 (59.4)	1014 (59.0)	.85	215 (60.9)	652 (56.6)	.15
Black	400 (40.6)	705 (41.0)		138 (39.1)	500 (43.4)
Study site
Memphis	466 (47.4)	905 (52.6)	.008	153 (43.3)	599 (52.0)	.004
Pittsburgh	518 (52.6)	814 (47.4)		200 (56.7)	553 (48.0)
Smoking status
Never	418 (42.5)	752 (43.9)	.022	161 (45.6)	490 (42.7)	.14
Current	83 (8.4)	194 (11.3)		30 (8.5)	141 (12.3)
Former	482 (49.0)	769 (44.8)		166 (45.9)	517 (45.0)
Comorbidities
Diabetes mellitus	173 (17.6)	229 (13.4)	.003	58 (16.5)	144 (12.6)	.06
Hypertension	397 (40.4)	655 (38.1)	.46	133 (37.7)	450 (39.1)	.87
Cerebrovascular disease	88 (9.1)	99 (5.9)	.002	17 (4.9)	74 (6.5)	.27
History of MI	24 (2.5)	51 (3.0)	.42	11 (3.2)	34 (3.0)	.88
History of cancer	190 (19.4)	272 (15.9)	.023	60 (17.1)	176 (15.4)	.46
Current statin use	123 (12.5)	231 (13.5)	.46	60 (17.1)	147 (12.8)	.044

Note: *p Value from t-test, analysis of variance, or χ²-test.

Table 2.

Models of Lens Disease: Likelihood of Cataract and Cataract Surgery With Each 1,000 bp Increase in Leukocyte Telomere Length

Cataract Outcome	Prevalence or Rate	Crude	Crude	Adjusted	Adjusted
		OR/HR	95% CI	OR/HR*	95% CI*
Baseline unilateral cataract	36.4%	0.95	0.89–1.01	0.95	0.89–1.01
Baseline bilateral cataract	25.6%	0.99	0.93–1.06	0.99	0.92–1.06
9-y incident cataract surgery	17.2/1000py	1.03	0.96–1.12	1.02	0.94–1.10

Notes: CI = confidence interval; HR = hazards ratio; OR = odds ratio.

^*All adjusted for age, gender, site, smoking, body mass index, high-density lipoprotein, cholesterol, and statin use.

Incident Cataract Surgery

Over 9 years of follow-up, 353 of 1,505 participants underwent cataract surgery for an average 9-year risk of 23.5% and a rate of 17.2/1,000 person-years. The average length of follow-up was 8.1 years. Incident cataract surgery was associated with age, gender, cholesterol, study site, and statin use (Table 1). LTL was not associated with 9-year risk of cataract surgery before or after adjustment (adjusted hazards ratio = 1.02, 95% confidence interval 0.94–1.10; Table 2).

Lens Transparency

Mean LTL at baseline for all ARMA participants was 4,790 bp (SD, 1,050 bp). At the time of the ARMA study eye exam, 97.7% of participants had evidence of any cataract, including 95.0% with nuclear opacity (AREDS grade > 3.0), 79.2% with anterior cortical opacity, and 52.5% with posterior cortical opacity. The six individuals with successfully aged lenses (nuclear grade ≤ 3.0 and 0% cortical grade) had a mean LTL of 5,700 bp and no history of smoking, cerebrovascular disease, or cardiovascular disease. For each 1,000 bp increase in mean LTL, participants were nearly 50% less likely to have any cataract or an intraocular lens (ie, an aged lens; Table 3). There was a similar protective effect for nuclear opacity, any cortical opacity, and cortical opacity subtypes. Adjustment for confounders included in the previous cataract models did not alter estimates appreciably.

Table 3.

Models of Successful Lens Aging: Odds of Lens Opacity Versus no Opacity With Each 1,000 bp Increase in Leukocyte Telomere Length

Opacity Outcome	N	M (SD) LTL (bp)	OR (95% CI)	OR (95% CI)*
No opacity	6	5,700 (1,550)	Ref	Ref
Nuclear opacity	246	4,780 (1,040)	0.51 (0.27–1.00)	0.48 (0.22–1.03)
Anterior cortical opacity	205	4,790 (1,040)	0.52 (0.27–1.01)	0.50 (0.23–1.05)
Posterior cortical opacity	136	4,760 (1,050)	0.50 (0.24–1.02)	0.47 (0.21–1.05)
Any cortical opacity	214	4,800 (1,040)	0.52 (0.27–1.01)	0.49 (0.23–1.05)
Any opacity or IOL	253	4,770 (1,040)	0.51 (0.26–0.99)	0.47 (0.22–1.02)

Notes: CI = confidence interval; OR = odds ratio.

^*Adjusted for age and gender.

DISCUSSION

This is the first report of the association between cataract and LTL in human participants. Using a stringent a priori-determined cutoff to define successful lens aging, it is striking that we identified individuals with a mean LTL nearly 1,000 bp longer than other individuals with less transparent lenses. Our analysis suggests that higher lens transparency was likely associated with longer LTL. Both transparency and LTL may indicate cumulative oxidation (13,24). Alternatively, senescence of the lens epithelium has been associated with non-enzymatic alterations in lens crystallins, and this may correlate with senescence-related immune aging. In contrast, diagnosed cataract, a clinical disease phenotype, was not associated with LTL. This could be because self-reported physician diagnosis of cataract is a biased measure because cataract surgery is more dependent on access to care or impact on quality of life than the severity of the cataract or because diagnosed cataract is a marker of disease and not necessarily aging. Cataractous lenses are different from aged human lenses in gene expression patterns (25), and the clinical classification of “cataract” or “no cataract” does not reflect the subtle morphological, histological, and biochemical changes that occur with lens aging (3,26–28). These data provide preliminary evidence that lens transparency might be an in vivo marker of primary aging, whereas diagnosed cataract is likely not. With further exploration, lens transparency might emerge as a valuable intermediate marker for use in intervention studies attempting to modify the aging process.

Using our stringent definition of transparency, we identified only six individuals free of opacity in the ARMA substudy. Subsequently, this definition illustrates a truly rare phenotype of successful aging. This rarity is akin to using the oldest old, such as centenarians and super-centenarians, to search for longevity-associated factors, yet may be more feasible because the phenotype can be identified at younger ages. Interestingly, at the Health ABC baseline clinic visit, these six participants had no history of smoking, cerebrovascular disease, or cardiovascular disease in addition to markedly longer LTL, yet their mean age was 72.3 years (range 70-75 years). This is further proof that sensitive lens measurements may provide a window on whole organism aging.

The lens has several real and hypothesized advantages as a marker of aging. First, the lens is unique in that nuclear lens cells are present at birth and do not turn over throughout life. Therefore, the lens has the potential to reflect risk factors for aging that are encountered throughout life, and the same tissue can be measured repeatedly. Second, though a somewhat immune privileged site, the lens is exposed to the internal and external environments. Subsequently, transparency might reflect both in vivo and ex vivo risk factors for aging. Third, using modern imaging techniques, transparency can be measured on a continuous scale, allowing finer risk stratification. Fourth, lens imaging is noninvasive, nonhazardous, and quick, so it may be easily adopted by researchers and clinicians. The utilization of a grading system, such as the AREDS one applied here (21) or the LOCS-III or the OCCCGS systems (29,30), offers a simple, fast, inexpensive, reproducible, and accurate means to ascertain lens transparency. Although AREDS grading is designed to be applied by trained graders after the eye exam on standard photos obtained with a photographic slit lamp, adding to the complexity and duration of the eye exam and significantly increasing cost, it surely offers an accurate and minimally biased assessment of lens transparency. Other imaging techniques, such as Scheimpflug photography, can provide a more accurate assessment of the nucleus, can be acquired quickly, and allow finer quantification of transparency, though it can be expensive and requires training.

Our results are strengthened by several factors. This study represents one of the largest samples of LTL measurements available in prospective longitudinal studies, a necessity given the inherent heterogeneity in human LTL (31). Coupled with measurement of lens transparency and LTL using validated methods (21,22), low loss-to-follow-up and long average length of follow-up, we believe that our estimates are minimally affected by random error and bias. Similar to previous studies, we also found that LTL was greater in women, blacks, nonsmokers (32), and those without cardiovascular disease (33–35). The adjusted odds ratios for covariates included in our multivariate models also agree with those previously published (36–39), suggesting that our models are accurate because they are consistent with those from other populations. Though the ARMA sample was small, which limited power, adjustment for other confounders did not alter the magnitude of estimates appreciably.

Our study does have some limitations. Although we determined that the ARMA population was relatively similar to the overall Health ABC cohort at baseline, selection of the ARMA study population was non-random with respect to outcome, which could introduce selection bias. It is possible that residual confounding affects our final estimates due to incomplete adjustment, as well as possible inaccuracies in lens transparency estimation introduced by the AREDS grading system. Although we found no significant interactions, unidentified modification might remain. Finally, we did not examine associations with posterior subcapsular cataract, though this is a rare lesion (36–38).

In conclusion, we found that greater lens transparency is likely associated with longer LTL in a sample of community-dwelling older adults. This suggests that lens transparency might serve as an in vivo marker of primary aging. We recommend that future studies of this association employ a prospective approach to more accurately portray the biological dynamic that occurs, namely studying telomere attrition and declining transparency using repeated measures. This would also mirror how these markers might be employed in the future to monitor biologic age across the lifespan. Because lens transparency decreases beginning in adolescence and LTL is heterogeneous at birth, it would be most sensible to investigate whether the rate of telomere shortening is associated with rate of change in lens transparency in a study of participants with a wide age range. Finally, testing whether lens transparency is a predictor of mortality or longevity may provide the strongest evidence of its usefulness as a biomarker of aging.

FUNDING

This research was supported in part by the Intramural Research Program of the National Institutes of Health (NIH), National Institute on Aging (NIA). The Health ABC study is funded by contracts N01-AG-6-2101, N01-AG-6-2103, and N01-AG-6-2106 from the NIH, Bethesda, MD. The ARMA study is supported by National Eye Institute Grant K23 EY000409; Contracts N01 AG62101, N01 AG62103, and N01 AG62106 and the Intramural NIA Research Program; the International Retinal Research Foundation, Birmingham, AL; Macular Degeneration Research, American Health Assistance Foundation, Clarksburg, MD; and a Career Development Award (A.I.) and an unrestricted grant to the UTHSC Department of Ophthalmology from Research to Prevent Blindness. Telomere length measurement was supported by grant 5U19AG023122-05 (S.R.C.) from the Longevity Consortium.