Seddon JM, Gensler G, Milton RC, Klein ML, Rifai N. Association Between C-Reactive Protein and Age-Related Macular Degeneration. JAMA. 2004;291(6):704–710. doi:10.1001/jama.291.6.704
Author Affiliations: Epidemiology Unit, Massachusetts Eye and Ear Infirmary, Department of Ophthalmology, Harvard Medical School, and Department of Epidemiology, Harvard School of Public Health, Boston, Mass (Dr Seddon); The EMMES Corporation, Rockville, Md (Mr Gensler and Dr Milton); Casey Eye Institute, Oregon Health and Science University, Portland (Dr Klein); and Department of Laboratory Medicine, Children's Hospital, Harvard Medical School, Boston, Mass (Dr Rifai).
Context C-reactive protein (CRP) is a systemic inflammatory marker associated
with risk for cardiovascular disease (CVD). Some risk factors for CVD are
associated with age-related macular degeneration (AMD), but the association
between CRP and AMD is unknown.
Objective To test the hypothesis that elevated CRP levels are associated with
an increased risk for AMD.
Design, Setting, and Participants A total of 930 (91%) of 1026 participants at 2 centers in the Age-Related
Eye Disease Study (AREDS), a multicenter randomized trial of antioxidant vitamins
and minerals, were enrolled in this case-control study. There were 183 individuals
without any maculopathy, 200 with mild maculopathy, 325 with intermediate
disease, and 222 with advanced AMD (geographic atrophy or neovascular AMD).
The AMD status was assessed by standardized grading of fundus photographs,
and stored fasting blood specimens drawn between January 1996 and April 1997
were analyzed for high-sensitivity CRP levels.
Main Outcome Measure Association between CRP and AMD.
Results The CRP levels were significantly higher among participants with advanced
AMD (case patients) than among those with no AMD (controls; median values,
3.4 vs 2.7 mg/L; P = .02). After adjustment for age,
sex, and other variables, including smoking and body mass index, CRP levels
were significantly associated with the presence of intermediate and advanced
stages of AMD. The odds ratio (OR) for the highest vs the lowest quartile
of CRP was 1.65 (95% confidence interval [CI], 1.07-2.55; P for trend = .02). The OR for CRP values at or above the 90th percentile
(10.6 mg/L) was 1.92 (95% CI, 1.20-3.06), and the OR for CRP values at or
above the mean plus 2 SDs (16.8 mg/L) was 2.03 (95% CI, 1.03-4.00). A trend
for an increased risk for intermediate and advanced AMD with higher levels
of CRP was seen for smokers (OR, 2.16; 95% CI, 1.33-3.49) and those who never
smoked (OR, 2.03; 95% CI, 1.19-3.46) with the highest level of CRP.
Conclusion Our results suggest that elevated CRP level is an independent risk factor
for AMD and may implicate the role of inflammation in the pathogenesis of
Age-related macular degeneration (AMD) is a burden to the elderly population,
and its consequences are increasing because treatment options are limited.
Prevention remains the best approach for decreasing the impact of this leading
cause of blindness. Knowledge about modifiable factors related to AMD has
increased considerably during the past decade, including most notably cigarette
smoking,1- 3 nutritional
factors,4- 7 obesity,8,9 and lipid levels.10
Many factors associated with AMD are also related to cardiovascular
disease (CVD). We have hypothesized that cardiovascular disorders and AMD
share common antecedents and proposed that novel biomarkers associated with
CVD be evaluated for their potential relationship with AMD.11 One
of these factors is C-reactive protein (CRP), a marker of systemic inflammation,
which has been shown to be an independent indicator of risk for cardiovascular
and peripheral arterial disease.12,13 Given
the similarity of the risk profile for the 2 diseases, we designed a study
to explore the relationships between AMD and CVD biomarkers, including CRP.
The investigation of inflammatory biomarkers in AMD is rendered even
more biologically plausible by the observation that inflammation is associated
with angiogenesis and that neovascularization can occur in inflammatory eye
diseases,14 similar to the most advanced and
debilitating neovascular form of AMD. Furthermore, in addition to CVD, stroke
and Alzheimer disease may also have an inflammatory component.13,15 It
is possible that AMD represents another chronic, age-related inflammatory
disease that is manifested in the eye and other organs, including the heart
and brain. Therefore, we examined the relationship between CRP levels and
AMD in the multicenter Age-Related Eye Disease Study (AREDS).
This study is ancillary to the AREDS. Details of the AREDS design have
been described elsewhere.16 AREDS is a prospective
cohort study designed to assess the incidence, clinical course, prognosis,
and risk factors for AMD and cataract. AREDS also includes a double-masked
randomized clinical trial to assess the effects of high-dose antioxidants
(vitamins C and E and beta carotene) on AMD and cataract and the effect of
high-dose zinc on AMD.
Eleven AREDS clinical centers nationwide enrolled 4757 participants
between February 1993 and January 1998. Participants include 1117 controls,
1063 with mild AMD, 1621 with intermediate AMD, and 956 with advanced AMD.
All participants were randomized into 1 of 4 treatment groups. Persons with
AMD were randomized to receive high-dose antioxidants, zinc, antioxidants
and zinc, or a placebo. Persons without AMD were randomized to receive either
high-dose antioxidant vitamins or a placebo.
Participants were aged 55 to 80 years at enrollment and were required
to be in overall good health. Potential participants were excluded if they
had diseases with a poor 7-year survival prognosis (eg, end-stage cancer,
advanced heart disease), hemochromatosis or Wilson disease, or oxalate kidney
stone, alcoholism, or drug abuse or were unwilling or unable to discontinue
the use of nonstudy antioxidant vitamin or zinc supplementation. Persons were
also excluded if they had visual acuity less than 20/32 in both eyes, advanced
AMD or laser photocoagulation for AMD in both eyes, bilateral cataract extraction
without signs of AMD, or other eye diseases that potentially compromised evaluation
of study outcomes or if they used medications known to be toxic to the lens
Participants were followed up at 6-month intervals, when information
was collected on changes in visual acuity, disease incidence and progression,
and risk factors from a visual acuity test, dilated lens and fundus examination,
and a clinical interview. In addition, at the annual visit (which occurred
at 12-month intervals beginning 12 months after randomization), blood samples
were drawn for specified AREDS tests, fundus and lens photographs were taken
(except at the first annual visit), and a refraction was completed. Since
the end of the clinical trial in April 2001, participants have been examined
The AREDS Executive Committee, the AREDS Data and Safety Monitoring
Committee, and the National Eye Institute approved this ancillary study in
1995. The 2 sites with the most participants were chosen to participate. The
Massachusetts Eye and Ear Infirmary (Boston, Mass) and the Devers Eye Institute
(Portland, Ore) enrolled 1026 participants (517 and 509, respectively) into
the AREDS clinical trial. Between January 1996 and April 1997, 930 of the
1026 participants (91%) had blood specimens drawn after randomization for
this ancillary study, 465 at each clinic. All but 1.3% of the specimens were
from participants who had been fasting for at least 8 hours. Blood samples
were processed immediately and then frozen in liquid nitrogen freezers at
−140°C. The study was approved by the human subjects committees
of the 2 clinical centers, and all participants signed an informed consent
Case-control definitions are adopted from a previous AREDS publication.9 According to reading center grading of fundus photographs
at the visit most closely associated with the specimen draw, participants
in this ancillary study were divided into 4 maculopathy groups by size and
extent of drusen in each eye, presence of geographic atrophy, and neovascular
disease. These groups, numbered serially and defined by increasing severity
of drusen or type of AMD, were as follows.
Group 1 (No Drusen). In this group (n = 183),
each eye had no drusen or nonextensive small drusen, no pigment abnormalities,
no advanced AMD, and no disqualifying ocular conditions. Most participants
had visual acuity of 20/32 or better in both eyes.
Group 2 (Intermediate Drusen). In this group
(n = 200), at least 1 eye had 1 or more intermediate-sized drusen, extensive
small drusen, or pigment abnormalities associated with AMD. Neither eye had
large drusen, advanced AMD, or a disqualifying ocular condition. Most participants
had visual acuity of 20/32 or better in both eyes.
Group 3 (Large Drusen or Intermediate AMD). In
this group (n = 325), at least 1 eye had either 1 or more large drusen, approximately
20 intermediate-sized soft drusen, or approximately 65 intermediate-sized
hard drusen. Neither eye had advanced AMD, a disqualifying ocular condition,
or presence of geographic atrophy with diameter at least one eighth that of
the average disc, and most participants had visual acuity of 20/32 or better
in both eyes. Also included were persons in whom one eye met these criteria
and the other eye had either a disqualifying ocular condition or visual acuity
of 20/32 or less not caused by AMD.
Group 4 (Geographic Atrophy or Neovascular AMD–Advanced
AMD). In this group (n = 222), at least 1 eye had geographic atrophy
definitely present (with diameter at least one eighth that of the average
disc; n = 58) or neovascular AMD (further defined below; n = 164). In most
cases, the other eye had visual acuity of 20/32 or better, with no evidence
of advanced AMD or a disqualifying ocular condition.
Neovascular AMD included choroidal neovascularization or retinal pigment
epithelial (RPE) detachment in 1 eye (nondrusenoid RPE detachment, serous
sensory, or hemorrhagic retinal detachment), subretinal hemorrhage, subretinal
pigment epithelial hemorrhage, subretinal fibrosis, or evidence of confluent
photocoagulation for neovascular AMD. The term neovascular is used as a summary term for this group of participants because most
persons in this group have direct evidence of choroidal neovascularization,
according to the assessment of fundus photographs. A few participants in this
group had serous RPE detachments.
The AREDS clinical trial7 showed that
rates of progression to advanced AMD in groups 1 and 2 were low (5-year rates
of 0.5% and 1.3%, respectively), and they were therefore combined here into
one larger control group. For regression analyses, to enhance statistical
power, group 3 (5-year rate of progression of approximately 18%) was combined
with group 4 (5-year rate of progression of approximately 43%) to form the
Serum samples were thawed and assayed for CRP, which was measured with
a high-sensitivity assay as in previous studies of CVD.12,13 The
concentration of CRP was determined by using an immunoturbidimetric assay
on the Hitachi 911 analyzer (Roche Diagnostics, Indianapolis, Ind), with reagents
and calibrators from Denka Seiken (Niigata, Japan). In this assay, an antigen-antibody
reaction occurs between CRP in the sample and an anti-CRP antibody that has
been sensitized to latex particles and agglutination results. This antigen-antibody
complex causes an increase in light scattering, which is detected spectrophotometrically,
with the magnitude of the change proportional to the concentration of CRP
in the sample. The coefficients of variation of the assay at concentrations
of 0.91, 3.07, and 13.38 mg/L are 2.81%, 1.61%, and 1.1%, respectively.
General risk-factor and dietary interviews were conducted at baseline,
and slit-lamp biomicroscopy and ophthalmoscopy were performed at the blood
drawing. The baseline risk factor variables that were considered in the analyses
can be divided into 5 classes: demographic, medical, dietary or supplementation,
use of medication, and ocular factors. For analysis, continuous variables
(body mass index, weight change from the age of 20 years, and sunlight exposure)
were categorized by quartiles or tertiles, according to the group without
drusen (group 1).
Demographic. The demographic variables included
age, sex, race, education, and sunlight exposure (adult lifetime average annual
ocular UV-B exposure, adapted from McCarty et al).17
Medical. Medical variables included history
of smoking, body mass index, weight change (increase or decrease) since the
age of 20 years, hypertension (systolic blood pressure >160 mm Hg, diastolic
blood pressure >90 mm Hg, or current use of antihypertensive medication),
history of CVD (at least 1 of the following: newly developed heart disease
after enrollment but before blood draw, occurrence of a stroke or myocardial
infarction after enrollment but before blood draw, history of angina and taking
an angina medication [dipyridamole, propranolol, β-blocker, calcium channel
blocker, nitroglycerin, or isobide dinitrate], taking a CVD medication [furosemide,
angiotensin-converting enzyme inhibitor, digoxin, blood-thinning medication,
cholesterol-lowering medication]), diabetes (under treatment for diabetes),
Dietary or Supplementation. The dietary or
supplement variables included an antioxidant index and use of study treatment
containing antioxidants. The antioxidant index was based on dietary results
from a modified Block Food Frequency questionnaire (AREDS
Manual of Operations, 1992) completed at the participant's baseline
visit. Three measures were assessed: carotenoid intake (alpha carotene, beta
carotene, lutein, lycopene, and beta-cryptoxanthin), vitamin C intake, and
vitamin E intake. Participants were grouped as having high antioxidant intake
(above the highest quartile of intake for 2 of the 3 measurements), low antioxidant
intake (below the lowest quartile of intake for 2 of the 3 measurements),
or mixed antioxidant intake.
Participants randomized to receive the study supplements containing
high-dose antioxidants or high-dose antioxidants and zinc comprised the antioxidant
treatment group. Participants randomized to receive the study supplements
containing zinc or placebo comprised those not in the antioxidant treatment
Use of Medication. Use of medication was defined
as current use with 5 or more lifetime years of regular use. These medications
included hydrochlorothiazide, diuretics (other than hydrochlorothiazide),
aspirin, antacids, nonsteroidal anti-inflammatory drugs, thyroid hormones, β-blockers,
and, for women, estrogen and progesterone.
Ocular. Ocular variables included iris color
and refractive error. Iris color was graded at the reading center by comparing
photographs of each eye with standards on a scale from 1 (light or blue) to
4 (dark or brown); a person's eyes were considered light if both eyes were
code 1, dark if both eyes were code 4, and mixed if at least 1 eye was code
2 or code 3 or eyes were not of the same code. A person was considered myopic
if both eyes were myopic by −1.0 diopters (D) spherical equivalent refractive
error or more, hyperopic if both eyes had +1.0 D spherical equivalent refractive
error or more, or other, which includes emmetropes and mixed cases.
The median values and interquartile ranges of CRP were calculated for
each maculopathy group, and the most advanced AMD grade was compared with
group 1 by using a nonparametric test of all P values.
Conditional logistic regression analysis (SAS procedure LOGISTIC; version
8.02, SAS Institute Inc, Cary, NC) was used to estimate odds ratios (ORs)
and 95% confidence intervals (CIs) after the population was divided into quartile
groups according to the quartile cut points of CRP values for the disease-free
group 1, as done in previous studies of CRP and CVD.12,13 Prevalence
ORs, which describe the association between disease and CRP (comparing cases
in groups 3 and 4 with controls in groups 1 and 2), were computed for each
CRP quartile group relative to the lowest quartile group. A test for linear
trend was calculated according to the median levels of CRP within each quartile
Multivariate ORs were estimated from conditional logistic regression
models and were adjusted for age (57-65, 66-70, and 71-83 years), sex, race
(white vs other), smoking (ever smoked vs never smoked), education (never
completed high school, high school graduate, some college, or college graduate),
body mass index (measured as weight in kilograms divided by the height in
meters squared; <23.9, 23.9-29.9, >29.9), antioxidant index (low, mixed,
high), diabetes, history of CVD, hypertension, and antioxidant treatment (taking
study supplement containing antioxidants vs taking study supplement containing
no antioxidants). History of CVD and hypertension were not correlated. Other
risk factors described previously were evaluated as potential covariates but
did not reach statistical significance in this ancillary study population
(including weight change from the age of 20 years, sunlight exposure, arthritis,
anti-inflammatory drugs, thyroid hormones, β-blocker use, hormone use
[women], other medications, iris color, and refractive error).
To evaluate a possible threshold effect, additional analyses were performed
comparing cases with controls according to levels of CRP for group 1 above
and below the 90th percentile and above and below the mean CRP plus 2 SDs.
Finally, to evaluate effect modification by cigarette smoking, additional
logistic regression analyses were conducted to determine the ORs for AMD in
6 subgroups defined by never and ever smoking and low, intermediate, and high
tertiles of CRP.
Of the 930 participants in this ancillary study, 61% were women and
39% were men, and the mean age was 69 years. Most participants (71%) had some
college or higher education. Forty-one percent of participants had never smoked,
51% were former smokers, and 8% were current smokers. The median CRP value
for all participants was 2.7 mg/L, with an overall range of 0.2 to 117 mg/L,
and the 90th percentile range was 0.2 to 10.6 mg/L. The CRP values did not
differ according to age groups (55-65, 66-70, and ≥71 years).
Table 1 displays the relationships
between baseline characteristics and maculopathy groups, unadjusted for other
variables. Significant differences (P<.05) between
individuals in maculopathy groups 3 and 4 and individuals in groups 1 and
2 included sex (lower proportion of women), smoking status (lower proportion
of never smokers), and education (lower proportion with a college degree).
Median baseline serum levels of CRP were higher among participants who
had more severe maculopathy (Table 2).
The difference between the median value for the most advanced maculopathy
group 4 (3.4 mg/L) and the median for maculopathy group 1 (2.7 mg/L) was statistically
significant (P = .02).
Table 3 displays the ORs
for risk of AMD according to the quartile of CRP, for maculopathy case groups
3 and 4 compared with groups 1 and 2, after adjustment for various other known
and potential factors associated with AMD. In an age- and sex-adjusted model,
persons above the highest quartile of CRP had higher risk of AMD (OR, 1.53;
95% CI, 1.03-2.28). The trend for an increase in risk of maculopathy with
increase in CRP was statistically significant (P =
.02). After adjustment for additional covariates, the trend for an increase
in risk remained significant for the highest quartile of CRP (OR, 1.65; 95%
CI, 1.07-2.55; P for trend = .02). In a separate
analysis (data not shown) that compared group 4 maculopathy with that of group
1, the effect estimate was similar for the highest level of CRP (OR, 1.72;
95% CI, 0.88-3.38) but was not significant, probably because of the reduced
sample size (P for trend = .09).
Table 4 displays the association
between CRP and maculopathy, with different cut points for values of CRP.
Persons with CRP levels above the 90th percentile had a significantly increased
risk, with an OR of 1.75 (95% CI, 1.12-2.75) for the age- and sex-adjusted
model and an OR of 1.92 (95% CI, 1.20-3.06) for the full multivariate model.
Persons with CRP values more than 2 SDs above mean levels for the study cohort
were also at increased risk, with an OR of 1.89 (95% CI, 0.98-3.66) for the
age- and sex-adjusted model and an OR of 2.03 (95% CI, 1.03-4.00) for the
full multivariate model.
To explore whether the effect of CRP was modified by cigarette smoking,
a consistently strong risk factor for AMD, we computed ORs for AMD in analyses
in which participants were stratified into 6 groups according to smoking (never,
ever) and tertile of CRP, as shown in Table
5. For smokers and never smokers, higher levels of CRP were associated
with higher risk of AMD. Persons in the high-risk group (current and past
smokers with the highest level of CRP) had a statistically significant, 2.16-fold
higher risk (95% CI, 1.33-3.49) of maculopathy compared with the low-risk
group (those who never smoked and had the lowest CRP level), after adjustment
for other factors. Among never smokers, the odds of developing AMD were 2.03
in the highest tertile of CRP (95% CI, 1.19-3.46) compared with individuals
in the lowest quartile of CRP, after adjustment for other factors. To evaluate
this relationship further, we analyzed the effect of smoking (ever, never)
stratified by tertile of CRP (data not shown). Cigarette smoking increased
risk of AMD more than 1.7-fold in the lower 2 tertiles of CRP—ORs, 1.79
(95% CI, 1.06-3.00) and 1.90 (95% CI, 1.12-3.22)—but there was no association
between smoking and AMD in the highest level of CRP (OR, 1.01; 95% CI, 0.61-1.69).
The highest levels of CRP appear to increase risk of AMD independent of smoking.
In this study, CRP levels were associated with AMD. After adjustment
for age, sex, and other variables, including smoking and body mass index,
CRP levels were significantly higher among individuals with intermediate and
advanced stages of AMD compared with controls. The magnitude of the association
ranged from an OR of 1.65 to 2.16 for the highest levels of CRP. Risk of AMD
was lowest among individuals who had low CRP values and never smoked. In contrast,
risk tended to be highest among smokers who also had higher levels of CRP.
Even among individuals who never smoked, the risk of AMD was increased 2-fold
among those with the highest category of CRP compared with the lowest level
of CRP as the referent category.
To our knowledge, this is the first report of an association between
CRP values and AMD in a large cohort of cases and controls. We found only
one other study18 of early age-related maculopathy
that showed no association with CRP. However, in that study, cigarette smoking
was also not associated with maculopathy among the 29 variables assessed,
and the authors suggested that this might be due to a "substantial rate of
Our findings have important implications and lend support to an evolving
hypothesis that inflammation is associated with the pathogenesis of AMD.11,19,20 Several mechanisms
are potentially involved that could lead to inflammatory responses, including
oxidative stress caused by known risk factors for AMD, such as smoking,1,2 insufficient antioxidants in the diet,3,5 dietary fat,4,5 and
obesity.8 Smoking is one of the most consistent
risk factors for AMD, yet many individuals who have never smoked develop AMD.
In fact, we found that higher CRP values increased risk of AMD among smokers
and individuals who never smoked, independent of the other risk factors in
the model. Therefore, factors other than smoking in these individuals might
create an adverse milieu or damage the RPE-retina-choroidal complex in some
way, which in turn could lead to an inflammatory stimulus and increase CRP
values. Elevated CRP levels have been found in other chronic diseases associated
with aging, including coronary heart disease, stroke, and Alzheimer disease,12,13,15 so AMD may have similar
pathogenetic processes. CRP is a measure of systemic inflammation but may
also be associated with local (intraocular) inflammatory factors or immune
function. These results provide insight into potentially important mechanisms
and will stimulate additional epidemiologic and basic science research to
sort out primary (causal) and secondary events.
Unique features of this study include the evaluation of a systemic biomarker
for inflammation in a large and well-characterized population of patients
with and without maculopathy from 2 geographic areas in the United States.
Further strengths of this study include the standardized collection of risk
factor information, including direct measurements of blood pressure and body
mass index, and classification of maculopathy by means of standardized ophthalmologic
examinations and fundus photography. Misclassification was unlikely because
CRP values were quantified by using objective laboratory methods without knowledge
of the participants' maculopathy status, and AMD grade was assigned without
knowledge of CRP status.
Residual confounding is a concern in many epidemiologic studies. We
controlled for known AMD risk factors and those associated with AMD in this
study cohort. For example, obesity and cigarette smoking are related to AMD
and are also related to increased levels of CRP and other systemic inflammatory
markers.21 CRP was significantly and independently
related to AMD in this study after adjustment for these confounding factors.
Although some unmeasured and therefore uncontrolled factors might still be
confounding this relationship, they would have to be highly associated with
CRP and be a strong risk factor for AMD to explain these results.
Our study population consists of patients with a range of maculopathy
and some individuals without AMD who participated in a randomized trial of
nutritional supplements. Results were not altered after adjustment for assignment
to antioxidants within the randomized trial. Controls were more likely to
be women, to be nonsmokers, and to have more education, so selection bias
should be considered. However, these analyses adjusted statistically for these
differences, and previous case-control analyses of the entire AREDS cohort,
as well as this subset at 2 centers, demonstrate an association with known
risk factors for AMD similar to that of other study populations. Although
this is a selected population, the cases likely represent the typical patient
with AMD, and the overall population is comparable to others in this age range
in terms of smoking status and prevalence of obesity. The large sample size
and well-characterized study population provided a unique opportunity to evaluate
our hypotheses. Moreover, the biological effects of CRP are not likely to
differ in major ways among various populations of patients with AMD.
Measures of CRP were taken from single fasting blood specimens that
were stored in a repository at −140°C until analyzed. These are
standard methods that are in use in several large-scale epidemiologic studies
throughout the country. In fact, these are the same methods used in the studies12,13 that established CRP as a marker
for CVDs. The medians and ranges of CRP in the various quartiles in our study
are similar to those of other published studies12,13 of
CRP and CVDs. Because a single measurement was taken, we cannot evaluate the
effects of changes in the levels of this biomarker over time. Follow-up studies
In summary, to the best of our knowledge, this is the first study to
implicate CRP as a systemic inflammatory marker for the development of AMD.
Higher CRP values were found to be significantly related to AMD independent
of established risk factors, including smoking and obesity. Among smokers
and nonsmokers, higher baseline CRP levels were associated with an increased
risk of AMD. These results may shed light on the mechanisms and pathogenesis
of AMD development and prognosis. Moreover, CRP levels may add clinically
relevant predictive information about risk of AMD in addition to known risk
factors. Anti-inflammatory agents might have a role in preventing AMD, and
inflammatory biomarkers such as CRP may provide a method of identifying individuals
for whom these agents and other therapies would be more or less effective.
These results and hypotheses should be evaluated further with prospective
studies and possibly randomized trials.