Markers of Inflammation, Oxidative Stress, and Endothelial Dysfunction and the 20-year Cumulative Incidence of Early Age-related Macular Degeneration: The Beaver Dam Eye Study

Abstract

Importance

Modifying levels of factors associated with age-related macular degeneration (AMD) may decrease risk of visual impairment in older persons.

Objective

To examine the relationships of markers of inflammation, oxidative stress, and endothelial dysfunction to the 20-year cumulative incidence of early AMD.

Design

Longitudinal population-based cohort study.

Setting

Beaver Dam, Wisconsin.

Participants

A random sample of 975 persons in the Beaver Dam Eye Study without signs of AMD who participated in the baseline examination in 1988-1990 and up to four follow-up examinations in 1993-1995, 1998-2000, 2003-2005, and 2008-2010.

Exposures

Serum markers of inflammation (high sensitivity C-reactive protein [hsCRP], tumor necrosis factor-α receptor 2 [TNF-αR2], interleukin-6 [IL-6], and white blood cell count), oxidative stress (8-isoprostane and total carbonyl content), and endothelial dysfunction (soluble vascular cell adhesion molecule-1 [sVCAM-1] and soluble intercellular adhesion molecule-1) were measured. Interactions with Complement Factor H (rs1061170) and Age-Related Maculopathy Susceptibility 2 (rs10490924), C3 (rs2230199) and C2/CFB (rs4151667) were examined using multiplicative models. AMD was assessed from fundus photographs.

Main Outcome Measure

Early AMD defined by the presence of any size drusen and the presence of pigmentary abnormalities, or by the presence of large-sized drusen (≥125 μm diameter), in the absence of late AMD.

Results

The 20-year cumulative incidence of early AMD was 23.0%. Adjusting for age, sex, and other risk factors, hsCRP (odds ratio [OR] comparing 4^th to 1^st quartile 2.18, P=0.005), TNF-αR2 (1.78, P=0.04), and IL-6 (1.78, P=0.03) were associated with the incidence of early AMD. Increased incidence of early AMD was associated with sVCAM-1 (OR per standard deviation on the log ng/mL scale 1.21, P=0.04).

Conclusions and Relevance

We found modest evidence of relationships of serum hsCRP, TNF-αR2, and IL-6 and sVCAM-1 to the 20-year cumulative incidence of early AMD independent of age, smoking status, and other factors. It is not known whether these associations represent a cause and effect relationship or if other unknown confounders accounted for the findings. Even if inflammatory processes are a cause of early AMD, it is not known whether interventions that reduce systemic inflammatory processes will reduce the incidence of early AMD.

Introduction

Age-related macular degeneration (AMD), the most common cause of severe loss of vision in older persons of European ancestry, is a multifactorial disease with strong evidence of genetic determinants.^1-6 Age, smoking, physical activity, and obesity have been found in most studies to be related to the incidence of AMD.⁶ Inflammation, oxidative stress, and endothelial dysfunction are also among the many host and environmental influences that have been hypothesized to affect the incidence and progression of AMD.^7-25 There is a strong biological rationale supporting a role of inflammation, oxidative stress, and, to a lesser extent, endothelial dysfunction in the development and progression of AMD. There is accumulating evidence of a relationship of high sensitivity C-reactive protein (hsCRP) to late AMD but less consistent evidence of a similar relationship to the incidence of early AMD.^26-36 Fewer epidemiological studies have examined the relationships of other systemic markers of inflammation (e.g., tumor necrosis factor-alpha receptor 2 [TNF-αR2]),^32-34 oxidative stress,³⁷ and endothelial dysfunction^32,34,38 to the incidence of AMD.

In an earlier prospective substudy in the Beaver Dam Eye Study (BDES) cohort, we found no relationships of markers of systemic inflammation and endothelial dysfunction to the 10-year cumulative incidence of early AMD.³³ Since that report, we have genotyped AMD candidate gene single nucleotide polymorphisms (SNPs) including Complement Factor H (CFH rs1061170), Age-Related Maculopathy Susceptibility 2 (ARMS2 rs10490924), Complement Component 2/Complement Factor B (C2/CFB rs4151667) and Complement Component 3 (C3 rs2230199) and remeasured the same markers as well as systemic markers of oxidative stress present in a random sample of the BDES cohort. We hypothesized that elevated levels of markers of systemic inflammation in the presence of 1 or 2 variant alleles for CFH rs1061170, ARMS2 rs10490924, C2/CFB rs4151667 and C3 rs2230199 and higher levels of markers of oxidative stress and endothelial dysfunction would be associated with greater risk of developing early AMD.

Methods

Population

Methods used to identify and descriptions of the population have appeared in previous reports.^39-43 Of the 5,924 eligible persons identified by a private census, 4,926 (83%) persons aged 43-86 years participated in the baseline examination in 1988-1990. Ninety-nine percent of the population was white. The cohort was re-examined at 5- (n=3,722), 10- (n=2,962), 15- (n=2,375) and 20-year (n=1,913) follow-up examinations.^40-43 There was greater than 80% participation among survivors at each examination.

All data were collected with Institutional Review Board approval from the University of Wisconsin-Madison in conformity with all federal and state laws, the work was compliant with the Health Insurance Portability and Accountability Act, and the study adhered to the tenets of the Declaration of Helsinki. Informed consent was obtained from each participant before every examination. Comparisons between participants and nonparticipants at each examination have appeared elsewhere.^39-43 In general, those who participated in the follow-up were more likely to be younger than nonparticipants who were alive or those who died, and, while adjusting for age, were less likely to have AMD.

Procedures

A standardized interview and examination were administered at each visit. Information on demographic characteristics, medication use including history of use of lipid-lowering drugs by type and use of steroidal and nonsteroidal anti-inflammatory drugs (NSAIDs), and history of smoking and physical activity was obtained by questionnaire. Body weight and height were measured. Similar procedures were followed at baseline and follow-up examinations.⁴⁴

Casual blood specimens were obtained at the baseline examination. An aliquot of blood was used immediately to determine the white blood cell (WBC) count. Remaining serum was stored for up to 17 years until being shipped on dry ice to the University of Minnesota laboratory for measurement of markers of inflammation (hsCRP, interleukin-6 [IL-6], and TNF-αR2), oxidative stress (8-isoprostane [8-ISO], an indicator of lipid oxidation, and total carbonyl content [TCC], an indicator of the amount of protein that has been oxidized by highly reactive free radicals), and endothelial dysfunction (soluble vascular cell adhesion molecule-1 [sVCAM-1] and soluble intercellular adhesion molecule-1 [sICAM-1]). The eMethods describe procedures to measure these markers and their inter-assay coefficients of variation as well as measurements of candidate gene SNPs.⁴⁵

Fundus Photography and Grading

Stereoscopic 30° color film fundus photographs centered on the macula (Diabetic Retinopathy Study standard field 2) were taken of each eye.^44,46,47 Gradings were performed for the pair of photographs of each macula at each examination using the Wisconsin Age-Related Maculopathy Grading System.^46-51 Graders were masked to any information regarding the participant and the fellow eye.

Definitions

The severity of AMD was determined using the 5-step Three Continent AMD Consortium Severity Scale.⁵² Individuals were considered not to have AMD if both eyes had either hard drusen or small soft drusen (<125 μm in diameter) only, regardless of area of involvement and no pigmentary abnormalities (defined as increased retinal pigment or retinal pigment epithelial [RPE] depigmentation present); or no definite drusen with any pigmentary abnormality. Early AMD was defined by the presence of any sized drusen and the presence of any pigmentary abnormality; or by presence of large-sized drusen (≥125 μm in diameter), regardless of area of involvement, in the absence of late AMD defined by the presence of pure geographic atrophy (GA) or exudative macular degeneration. When one eye was ungradable, it was assumed to have the same AMD severity as the fellow eye.

Persons at risk for developing early AMD were those without early AMD in either eye at baseline. Incidence of early AMD was defined by developing signs of early AMD in at least 1 eye when both eyes had no AMD at the baseline examination. Incidence was determined for signs of early AMD, e.g., large drusen size ≥125 μm, drusen type (soft indistinct/reticular), and pigmentary abnormalities (increased retinal pigment and RPE depigmentation). Due to limited power, we did not examine the relationship of the markers and risk of developing late AMD.

All covariates were measured at baseline. Age was categorized into four groups: 43-54 years, 55-64 years, 65-74 years and 75 or more years. Body mass index (BMI) was calculated by dividing a participant's weight in kilograms by their height in meters squared. Obesity was defined as a BMI of 30 kg/m² or greater. Current smokers were identified as persons having smoked 100 or more cigarettes in their lifetime and smoking at the time of the examination. Participants were considered physically active if they engaged in physical activity long enough to work up a sweat at least once per week. Use of statin drugs, steroidal anti-inflammatory drugs, and NSAIDs were determined from self-report.

Statistical Analysis

All analyses were performed with SAS version 9.2 (SAS Institute, Cary, North Carolina, USA). Cumulative incidence was estimated by the product-limit method,⁵³ accounting for the competing risk of death.⁵⁴ Discrete logistic hazard regression⁵⁵ was used to estimate odds ratios (ORs) for associations between each marker of inflammation, oxidative stress, and endothelial dysfunction with incidence of early AMD, incidence of large drusen ≥125 μm in diameter, incidence of soft indistinct or reticular drusen, and incidence of pigmentary abnormalities. Each marker was examined using a natural logarithmic transformation and categorized into quartiles. P-values are reported per standard deviation (SD) increase on the logarithmic scale, for each higher quartile compared to the first quartile, and for a trend per increasing quartile. Models first adjust only for age and sex. Then models additionally adjust for smoking status, physical activity, BMI, statin use, and anti-inflammatory medication use. ORs were estimated for associations of having 1 and 2 versus 0 risk alleles of CFH and ARMS2 and having 1 or 2 versus 0 risk alleles for C3 and C2/CFB with the incidence of early AMD adjusting for the same factors. P-values were estimated for the relationship of age (older 2 versus younger 2 age groups), sex, obesity, current smoking, and physical activity to each marker using the Mann-Whitney U test. To test for interactions, we first modeled a multiplicative interaction between each inflammatory, oxidative stress, and endothelial dysfunction marker and having 0, 1, or 2 risk alleles for CFH and ARMS2 and having 0 or 1 or 2 risk alleles for C3 and C2/CFB; we then examined each relationship by stratifying by genotype for each SNP.

Change in area under the receiver operating characteristic curve (AUC) was used to measure improvement in prediction when a marker (e.g., hsCRP, modeled as trend per SD on the logarithmic scale) was added to a model based on traditional AMD risk factors and to a model based on traditional risk factors plus candidate SNPs using the method described by DeLong and colleagues.⁵⁶

Results

Of the 4,926 BDES participants at baseline, 1,793 were included as part of a random sample in a sub-study of chronic kidney disease. Age, sex, BMI, history of smoking status, history of sedentary lifestyle, history of use of NSAIDs, presence of early and late AMD, and the distributions of the AMD candidate genotypes did not differ between those included in the random sample and those excluded, except for CFH variant allele (59% in those included vs. 62% in those excluded, P=0.03, eTable 1). To be included in analyses, a participant from the random sample must have had measures of markers of inflammation (hsCRP, IL-6, TNF-αR2, and WBC count), oxidative stress (8-ISO and TCC), and endothelial dysfunction (sVCAM-1 and sICAM-1), relevant genetic data, and no AMD at baseline as assessed from 30° stereoscopic color fundus photographs. Each participant also must have had at least one follow-up visit with photographs where at least one eye was gradable for AMD. Characteristics of the 975 persons who met these criteria and were included in analyses and those excluded are described in Table 2.

Table 2.

Characteristics of Participants Included and Excluded from Analysis.

Measure	Included		Excluded		P valuea
	N	Mean ± SD, Median or %	N	Mean ± SD, Median or %
Mean age, years	975	58.1 ± 9.8	3951	63.0 ± 11.3	<.0001
Sex, % men	421	43.2	1743	44.1	0.60
Mean BMI, kg/m2	972	28.8 ± 5.6	3909	28.7 ± 5.4	0.92
Smoking status, %
Former	334	34.3	1413	35.8	0.63
Current	205	21.0	765	19.4	0.37
Sedentary lifestyle, %	707	72.5	3068	77.7	0.001
Using NSAIDs, %	303	31.1	1340	33.9	0.09
CFH genotype, %
C/T	408	44.4	1701	48.2	0.009
C/C	115	12.5	486	13.8	0.06
ARMS2 genotype, %
G/T	346	36.4	1245	34.5	0.45
T/T	35	3.7	193	5.3	0.05
C2FB rs4151667 genotype, %
T/A	42	6.4	223	9.9	0.006
A/A	0	0.0	8	0.4	0.97
C3 rs2230199 genotype, %
C/G	217	33.0	699	30.9	0.33
G/G	25	3.8	94	4.2	0.79
Median hsCRP, mg/L	956	1.9
Median TNF-αR2, pg/mL	950	2295.2
Median IL-6, pg/mL	944	2.1
Median WBC count, 1000/μL	957	7.0
Median 8-ISO, pg/mL	961	120.7
Median TCC, nmol/mg	971	0.1
Median sVCAM-1, ng/mL	973	758.8
Median sICAM-1, ng/mL	970	277.9

8-ISO, 8-isoprostane; ARMS2, age-related maculopathy susceptibility 2; C2/CFB, complement component 2/complement factor B; C3, complement component 3; CFH, complement factor H; hsCRP, high sensitivity C-reactive protein; IL-6, interleukin-6; NSAID, nonsteroidal anti-inflammatory drug; sICAM-1, serum intercellular adhesion molecule-1; sVCAM-1, serum vascular cell adhesion molecule-1; TCC, total carbonyl content; TNF-αR2, tumor necrosis factor alpha receptor 2; WBC, white blood cell.

^aAdjusted for age and sex.

Associations of markers with the 20-year incidence of early AMD

One hundred and ninety-eight of the 975 individuals developed early AMD. The 20-year cumulative incidence adjusting for the competing risk of death for early AMD was 23.0% (95% confidence interval [CI] 20.2%-25.8%). Log transformed serum hsCRP (P=0.004), IL-6 (P=0.02), and sVCAM-1 (P=0.04) were associated with the 20-year cumulative incidence of AMD while adjusting for age, sex, and other factors (Table 3). There were trends for increasing quartile of hsCRP (P for test of trend=0.01) and IL-6 (P for test of trend=0.04) and higher 20-year cumulative incidence of early AMD. Compared to those in the lowest quartile, those in the highest quartile for hsCRP (P=0.005), TNF-αR2 (P=0.04) and IL-6 (P=0.03) had greater odds of developing early AMD (Table 3). When all three of these markers plus sVCAM-1 were included in the same model, only hsCRP (P=0.04) and sVCAM-1 (P=0.04) remained associated with the incidence of early AMD. There were no other relationships of other markers to the 20-year cumulative incidence of early AMD (Table 3). Relationships for hsCRP and WBC count were similar when analyses were expanded to include all individuals in the population (data not shown). The relationships of the markers of inflammation, oxidative stress, and endothelial dysfunction to the incidence of drusen type and size and pigmentary abnormalities were similar to that of early AMD (eTable 4).

Table 3.

Relationship of Inflammatory, Oxidative Stress and Endothelial Dysfunction Markers to the 20-year Cumulative Incidence of Early Age-related Macular Degeneration.

Marker	N at Risk	N Events	Model 1a			Model 2b
			OR	95% CI	P-value	OR	95% CI	P-value
hsCRP (mg/L)
Per SD on log scale	875	178	1.27	1.08-1.50	0.003	1.29	1.08-1.55	0.004
≤1.0	237	33	1.00			1.00
>1.0 - 2.0	226	54	1.77	1.12-2.80	0.02	1.94	1.19-3.14	0.007
>2.0 - 4.5	247	50	1.55	0.98-2.46	0.06	1.76	1.07-2.87	0.03
>4.5	165	41	2.14	1.31-3.48	0.002	2.18	1.27-3.74	0.005
Trend over quartiles	875	178	1.22	1.06-1.42	0.007	1.23	1.05-1.45	0.01
TNF-αR2 (pg/mL)
Per SD on log scale	886	179	1.18	1.00-1.40	0.06	1.19	1.00-1.42	0.06
≤2000	252	36	1.00			1.00
>2000 - 2500	320	71	1.55	1.02-2.36	0.04	1.56	1.01-2.42	0.05
>2500 - 3000	176	37	1.48	0.91-2.41	0.12	1.44	0.87-2.39	0.16
>3000	138	35	1.76	1.04-2.99	0.04	1.78	1.03-3.08	0.04
Trend over quartiles	886	179	1.17	1.00-1.38	0.05	1.17	0.99-1.38	0.07
IL-6 (pg/mL)
Per SD on log scale	880	176	1.22	1.05-1.42	0.009	1.22	1.03-1.44	0.02
≤1.6	274	43	1.00			1.00
>1.6 - 2.4	244	52	1.50	0.98-2.29	0.06	1.58	1.00-2.50	0.05
>2.4 - 3.7	192	42	1.45	0.92-2.29	0.11	1.57	0.96-2.56	0.07
>3.7	170	39	1.69	1.06-2.69	0.03	1.78	1.05-3.02	0.03
Trend over quartiles	880	176	1.17	1.01-1.35	0.03	1.19	1.01-1.39	0.04
WBC count (1000/μL)
Per SD on log scale	892	181	1.23	1.05-1.45	0.01	1.13	0.95-1.35	0.16
≤6.0	241	40	1.00			1.00
>6.0 - 7.2	245	58	1.49	0.97-2.28	0.07	1.45	0.93-2.25	0.10
>7.2 - 8.5	200	39	1.26	0.79-2.01	0.34	1.13	0.69-1.84	0.63
>8.5	206	44	1.75	1.11-2.75	0.02	1.41	0.86-2.31	0.17
Trend over quartiles	892	181	1.16	1.01-1.33	0.04	1.08	0.92-1.26	0.34
8-ISO (pg/mL)
Per SD on  log scale	879	175	0.95	0.81-1.12	0.55	0.92	0.78-1.08	0.30
≤97	232	43	1.00			1.00
>97 - 124	229	54	1.37	0.89-2.10	0.15	1.43	0.92-2.21	0.11
>124 - 164	201	46	1.41	0.90-2.19	0.13	1.53	0.97-2.42	0.07
>164	217	32	0.79	0.49-1.27	0.32	0.71	0.43-1.17	0.18
Trend over quartiles	879	175	0.94	0.82-1.08	0.41	0.92	0.80-1.07	0.28
TCC (nmol/mg)
Per SD on log scale	888	180	0.98	0.83-1.15	0.80	0.97	0.82-1.15	0.75
≤0.1	255	51	1.00			1.00
>0.1 - 0.14	182	35	0.94	0.60-1.47	0.78	0.87	0.54-1.39	0.56
>0.14 - 0.2	213	54	1.26	0.84-1.90	0.27	1.19	0.78-1.82	0.42
>0.2	238	40	0.88	0.57-1.36	0.56	0.88	0.56-1.38	0.57
Trend over quartiles	888	180	0.99	0.87-1.14	0.92	0.99	0.86-1.14	0.92
sVCAM-1 (ng/mL)
Per SD on log scale	890	180	1.17	0.99-1.39	0.06	1.21	1.01-1.44	0.04
≤660	256	44	1.00			1.00
>660 - 790	258	50	1.14	0.74-1.74	0.55	1.19	0.76-1.85	0.45
>790 - 950	211	42	1.13	0.72-1.78	0.58	1.14	0.72-1.81	0.58
>950	165	44	1.43	0.90-2.27	0.13	1.58	0.98-2.56	0.06
Trend over quartiles	890	180	1.11	0.96-1.29	0.16	1.14	0.98-1.33	0.09
sICAM-1 (ng/mL)
Per SD on log scale	888	180	1.04	0.89-1.23	0.61	0.98	0.82-1.17	0.81
≤240	248	45	1.00			1.00
>240 - 280	207	48	1.36	0.88-2.09	0.17	1.49	0.95-2.34	0.08
>280 - 330	212	43	1.07	0.69-1.66	0.77	1.08	0.68-1.72	0.74
>330	221	44	1.26	0.81-1.96	0.30	1.10	0.68-1.78	0.70
Trend over quartiles	888	180	1.05	0.91-1.20	0.50	1.00	0.86-1.16	0.99

8-ISO, 8-isoprostane; AMD, age-related macular degeneration; hsCRP, high sensitivity C-reactive protein; IL-6, interleukin-6; OR, odds ratio; sICAM-1, serum intercellular adhesion molecule-1; sVCAM-1, serum vascular cell adhesion molecule-1; TCC, total carbonyl content; TNF-αR2, tumor necrosis factor alpha receptor 2; WBC, white blood cell.

^aModel 1 adjusted for age and sex.

^bModel 2 adjusted for age, sex, smoking habits, physical activity, statin use, anti-inflammatory medication use, and body mass index.

We examined the relationship of serum TNF-αR2 to the incidence of early AMD by sex and, in women, by menopausal status. The TNF-αR2 relationship was similar in women (OR per 1 SD increase on log scale 1.22, P=0.08) and men (OR 1.16, P=0.34), and was stronger in women who had gone through menopause (OR 1.45, P=0.01) compared to women who had not (OR 1.07, P=0.91) while adjusting for age, BMI, smoking status, physical activity levels, and use of statins and anti-inflammatory medications.

Associations of candidate gene SNPs with the 20-year incidence of early AMD and interactions with the markers

Both CFH (P=0.003) and ARMS2 (P=0.006) with 2 risk alleles were associated with the 20-year incidence of early AMD. However, neither C3 nor C2/CFB were related to the 20-year incidence of early AMD (eTable 5). The relationships of each marker per SD on the log scale, stratified by having 0, 1, and 2 risk alleles for CFH and ARMS2, to the incidence of early AMD after adjustment for age, sex, smoking status, and other factors at baseline are presented in the Figure. There were no interactions between C3 or C2/CFB and any of the markers (data not shown).

Figure.

Relationship of markers of inflammation, endothelial dysfunction, and oxidative stress to the 20-year incidence of early age-related macular degeneration in the Beaver Dam Eye Study (1988-1990 to 2008-2010) stratified by A. CFH rs1061170 genotype; B. ARMS2 rs10490924 genotype.

Risk assessment

The models including smoking, physical activity, BMI, age, and sex discriminated poorly in predicting the incidence of early AMD (Table 6). The largest increase and incremental gain in the AUC occurred after including hsCRP in the model that included traditional risk factors for incidence of early AMD; however, it was not statistically significant (P=0.42, Table 6).

Table 6.

Effects of Markers of Inflammation, Endothelial Dysfunction, and Oxidative Stress on the Risk of Age-Related Macular Degeneration in Risk Assessment Models.

Markera	Traditional Risk Factor Model				Traditional Risk Factors + Genetic Factors Model
	AUC (95% CI)		Change in AUC, %	P value	AUC (95% CI)		Change in AUC, %	P value
	Without Markers in Model	With Marker in Model			Without Markers in Model	With Marker in Model
hsCRP	0.6714 (0.6212, 0.7216)	0.6786 (0.6295, 0.7277)	1.07	0.42	0.6923 (0.6411, 0.7435)	0.6997 (0.6492, 0.7502)	1.06	0.32
TNF-αR2		0.6715 (0.6217, 0.7214)	0.02	0.97		0.6926 (0.6417, 0.7435)	0.04	0.93
IL-6		0.6778 (0.6287, 0.7269)	0.95	0.47		0.6992 (0.6483, 0.7500)	0.98	0.38
WBC count		0.6747 (0.6239, 0.7255)	0.49	0.59		0.6962 (0.6445, 0.7479)	0.56	0.48
8-ISO		0.6726 (0.6221, 0.7232)	0.18	0.69		0.6927 (0.6412, 0.7443)	0.06	0.86
TCC		0.6750 (0.6251, 0.7249)	0.54	0.67		0.6953 (0.6443, 0.7462)	0.42	0.65
sVCAM-1		0.6720 (0.6215, 0.7226)	0.09	0.92		0.6956 (0.6437, 0.7474)	0.46	0.61
sICAM-1		0.6719 (0.6214, 0.7225)	0.08	0.82		0.6909 (0.6393, 0.7425)	-0.20	0.45

8-ISO, 8-isoprostane; AUC, area under the receiver operating characteristic curve; CI, confidence interval; hsCRP, high sensitivity C-reactive protein; IL-6, interleukin-6; TCC, total carbonyl content; TNF-αR2, tumor necrosis factor alpha receptor 2; sICAM-1, soluble intercellular adhesion molecule-1; sVCAM-1, soluble vascular cell adhesion molecule-1; WBC, white blood cell.

Traditional risk factors = age, sex, smoking status, physical activity, and body mass index.

Genetic factors = CFH rs1061770 and ARMS2 rs10490924.

^aMarkers modeled per standard deviation on the logarithmic scale.

Discussion

In the BDES, higher levels of serum hsCRP, TNF-αR2, IL-6, and sVCAM-1 were modestly associated with the 20-year cumulative incidence of early AMD independent of age, sex, smoking, physical activity, obesity status, and history of use of statins and anti-inflammatory drugs.

Most studies have shown a consistent relationship between serum hsCRP and late AMD.^{26-28,31,33-36} There is less consistency in studies that have examined the relationship of hsCRP to the long-term incidence of early AMD; five studies did not find a relationship^{26,28,30,31,34} and two did.^27,36 Our findings are consistent with those from a recent meta-analysis of five large studies that showed that participants with high hsCRP levels (>3 mg/L) had an increased risk of incident early AMD (OR 1.49; 95% CI 1.06-2.08) compared to participants with low hsCRP levels (<1 mg/L).³⁶ When similar analyses were performed in the BDES cohort, while adjusting for age, sex, and other factors, those with high levels of hsCRP had greater odds of developing early AMD (OR >3 mg/dL vs. <1 mg/dL 1.69; P=0.04).⁵⁷

The pathogenetic mechanisms underlying the role of hsCRP and other inflammatory biomarkers in the development of AMD are complex and not fully understood.²⁶ Johnson and colleagues speculated that elevations of hsCRP during acute phase reactions over a lifetime in individuals homozygous for the CFH rs1061170 risk alleles resulted in increasing tissue damage to Bruch's membrane and the RPE, further increasing the risk of AMD compared to those homozygous for the wild type of CFH.⁵⁸ There is emerging evidence that the association of elevated levels of hsCRP with early AMD is not due to hsCRP directly damaging the RPE and Bruch's membrane. Instead, when hsCRP levels increase (e.g., during an acute phase reaction), it has been shown that CRP is more likely to bind more strongly to the CFH gene site when 1 or 2 risk alleles are present compared to when no risk alleles are present.^26,59,60 The stronger binding of hsCRP is thought to block the regulatory function of CFH in deactivating surface-bound C3b, a key factor in the response of the complement immune system to inflammatory stimuli.^59,60 The finding in the BDES of a borderline multiplicative interaction of CFH with higher levels of hsCRP for the 20-year cumulative incidence of early AMD when 1 and 2 risk alleles are present is consistent with these observations.

Our study showed that TNF-αR2 was associated with the development of early AMD independent of BMI, smoking, and other factors. Their relationship was no longer statistically significant when hsCRP was included in the model. TNF-αR2 was not previously shown to be related to the prevalence of any AMD or the progression to late AMD.^32,34 TNF-α is a cytokine involved in cell activation, differentiation and apoptosis, and has been shown to be related to AMD in some studies.^61,62 The receptor TNF-αR2 is expressed in the choroid vascular cells, RPE, and Mueller cells in the retina. Its role in the pathogenesis of AMD is poorly understood, as is the reason the relationship was stronger in post-menopausal women than in pre-menopausal women. The reason for the inverse interaction in the BDES group with ARMS2 is not understood. It may be a chance finding.

In the BDES, there was no relation of two markers of oxidative stress, serum 8-ISO and TCC, to the incidence of AMD. This is consistent with the lack of a protective effect of antioxidant vitamins for the incidence of early AMD in the AREDS 1.⁶³ The RPE has been shown to be vulnerable to oxidative damage by radical-catalyzed lipid peroxidation.^64-66 The lack of an association may be due to oxidative stress not being related to incident early AMD or that the two markers do not reflect oxidative stress occurring at the cellular level at the RPE. The variability of these two oxidative stress measures may have affected our ability to find a relationship if it were present. Few other epidemiological studies have examined the relationships of these measures of oxidative stress to AMD. In one, a prospective case-control study involving 77 AMD patients and 75 controls, plasma F2 isoprostane was not related to AMD after adjustment for age, sex, and smoking.³⁷

In the BDES, while adjusting for age, sex, smoking and other factors, sVCAM-1 but not sICAM-1 was associated with the incidence of early AMD. Both of these cellular adhesion molecules are transmembrane cell surface proteins with immunoglobulin superfamily domains. They regulate inflammation by attracting WBCs and controlling their migration into the blood vessel wall.⁶⁷ Increased expression of these molecules in the cellular wall is reflected by increases in soluble forms of these molecules in the plasma. Increases in the number of WBCs have been shown in the choroid of eyes with early and late AMD.^68-70 Complement mediated activation of choroidal endothelial secretion of sICAM-1 has been hypothesized to play a role in the pathogenesis of AMD.⁷¹ However, few associations were found in the studies that have examined these relationships.^31,32,72

Our findings suggest that while there are statistically significant, clinically meaningful relationships of the inflammatory markers in persons without AMD, they have limited prognostic value for predicting the incidence of early AMD, independent of age, sex, smoking history, and other traditional risk factors. The increase in AUC of 1.02% for inclusion of hsCRP in the risk prediction model was small and not statistically significant. It compares unfavorably with other potential predictive factors used for other endpoints (e.g., hsCRP and serum high-density lipoprotein cholesterol levels) when added to the Framingham risk score for coronary heart disease in the Atherosclerosis Risk in Communities study.^73-75

There are many strengths to our study, including the use of standard protocols to measure AMD from fundus photographs over a 20-year period in a representative population-based study. There are also limitations. First, the analyses were performed in a randomized sample of the cohort in an effort to minimize bias. It is possible that this sample may not be representative of the cohort. However, randomization appeared to minimize this possibility; there were few differences between those randomized and those not randomized. Additionally, hsCRP and WBC count were measured in the whole cohort at baseline and the findings were similar to those reported in the smaller randomized cohort. Second, selective survival may have obscured relationships if people with high levels of serum 8-ISO or TCC who developed early AMD were more likely to die before being examined than those with low levels of these markers. Those with higher levels of serum 8-ISO were not more likely to die than to be observed with or without early AMD (OR per SD on the log scale 0.98; 95% CI 0.80-1.15; P=0.80). However, those with higher levels of TCC were more likely to die than to be observed with or without early AMD (OR 1.22; 95% CI 1.06-1.38; P=0.02) after adjusting for age, sex, smoking status, physical activity, BMI, and anti-inflammatory medication use. Third, a single measure of a marker, e.g., hsCRP, may not be representative of lifetime exposure. However, Nash and colleagues evaluated the 10-year percent agreement between groups for levels of IL-6 (50.8%) and hsCRP (53.4%), and their data suggest that the levels of these inflammatory markers track over time and are fairly stable.⁷⁶ Data from another study suggest modest variability of inflammatory markers over time, dependent partially upon changes in cardiovascular risk factors, eg, obesity, physical activity, and smoking status as people age.⁷⁷ Fourth, the long period between freezing and measurement of samples may have resulted in the greater variability found in serum 8-ISO and TCC levels, reducing our ability to find a relationship. Serum samples were stored at -80°C. These tests were found to be stable with essentially no evidence of auto-oxidation in a pilot study from the Nurses' Health Study.⁷⁸ Further, Schwedhelm and colleagues reported long-term storage of blood samples in prospective studies at -80°C to be stable with respect to these markers of oxidative stress.⁷⁹

In summary, inflammatory markers and one marker of endothelial dysfunction were modestly associated with the 20-year cumulative incidence of early AMD in the BDES. These data provide further support for the role of inflammation in the pathogenesis of early AMD. It may be one of many mechanisms involved in the development of this complex multifactorial disease. It is unknown whether these associations represent a cause-and-effect relationship or if other unknown confounders accounted for the findings. Even if inflammatory processes are a cause of early AMD, it is unknown whether interventions that reduce systemic inflammatory processes will reduce the incidence of early AMD.