Association of Circulating C-Reactive Protein and Interleukin-6 with Longevity into the 80s and 90s: The Rancho Bernardo Study

Abstract

Context: Few studies have examined whether the inflammatory markers IL-6 and C-reactive protein (CRP) are associated with exceptional longevity.

Objective: Our objective was to determine the association of serum CRP and IL-6 with adult lifespan.

Design, Setting, and Participants: This was a prospective, population-based study of 610 men and 743 postmenopausal women, mean age 73 yr, who had serum IL-6 and CRP measurements at baseline (1984–1987) and who were followed for mortality for up to 23 yr (through 2008). Participants must have been old enough at baseline (57 yr) to have the potential to achieve age 80 during follow-up.

Main Outcome Measures: Relative survival time and age at death (lifespan) were assessed.

Results: During follow-up, overall mortality was 69%. After adjustment for cardiovascular disease risk factors, in men, each sd increase in IL-6 was associated with a 15% decrease in survival time and 0.94-yr shorter lifespan (P < 0.001); corresponding values for CRP were a 12% decrease in survival time and a 1.00-yr reduction in lifespan (P < 0.001). Among women not using estrogen therapy (n = 532), survival time decreased 7% per sd higher IL-6 and lifespan was 1.35 yr shorter (P < 0.001). IL-6 was not related to lifespan among women using estrogen. CRP levels were not significantly associated with survival time or lifespan in women regardless of estrogen status.

Conclusions: Higher levels of inflammatory markers predicted reduced survival time and shorter lifespan among older men, whereas only IL-6 was associated with longevity in postmenopausal women and only among those not using estrogen.

Higher levels of C-reactive protein and IL-6 predict reduced survival time and shorter lifespan among older men; only IL-6 is associated with longevity in postmenopausal women.

C-reactive protein (CRP) and IL-6 are well-studied proinflammatory cytokines (1,2,3,4,5). In the atherosclerotic process, IL-6 promotes endothelial dysfunction and smooth muscle cell proliferation and migration and recruits and activates inflammatory cells (5). Additionally, IL-6 stimulates lipolysis and oxidation of fat and is the primary regulator of CRP synthesis and secretion (1). CRP induces the secretion of cellular adhesion molecules and tissue factors (1) and promotes adhesion and chemotaxis of monocytes via monocyte chemotactic protein-1 (4,5). CRP also mediates low-density lipoprotein (LDL) uptake by macrophages (3). Many epidemiological studies have reported an association of higher circulating CRP and IL-6 levels with increased risk of cardiovascular disease (CVD) (6,7,8,9) and all-cause mortality (10,11,12).

There is also evidence for IL-6 and CRP associations with obesity and insulin resistance. About 25–30% of IL-6 is secreted by adipose tissue, thought to imply a low-grade inflammatory response in white adipose tissue, which might lead to insulin resistance and type 2 diabetes (1). CRP is correlated with insulin levels (2,13), and levels of both cytokines are elevated in individuals with type 2 diabetes and insulin resistance (14,15,16,17).

As the average age of the U.S. population increases, interest has grown in factors associated with exceptional longevity. Although there is good evidence that longevity is moderately heritable with approximately 15–30% attributed to genetic factors (18,19,20,21), there are many potentially modifiable factors that can reduce the quantity and quality of life. Although increased levels of proinflammatory cytokines such as CRP and IL-6 are associated with type 2 diabetes and CVD, and these diseases are generally associated with decreased survival, the associations between these inflammatory markers with lifespan and extreme longevity into the eighth and ninth decades of life are unknown. We report here the associations of CRP and IL-6 with both survival and lifespan among community-dwelling men and women.

Subjects and Methods

Study participants

The Rancho Bernardo Study is a prospective cohort study of community-dwelling adults, established between 1972 and 1974, when 82% of the adult residents of a San Diego, CA, suburb participated in a survey of heart disease risk factors. Eighty percent of the surviving cohort from the 1972–1974 survey participated in an examination between 1984 and 1987. Participants included in the present analysis are 610 men and 743 women who participated in the 1984–1987 follow-up examination when blood for CRP and IL-6 assays was obtained and who were old enough at this examination to have the possibility of reaching age 80 during the 23-yr follow-up period. The Rancho Bernardo Study was approved by the Institutional Review Board of the University of California, San Diego; participants gave written informed consent.

CRP and IL-6

Morning fasting blood samples were obtained by venipuncture from 1984–1987 and frozen at −70 C until 2000, when IL-6 and CRP were measured on these never previously thawed samples. IL-6 was measured using a high-sensitivity amplified commercial ELISA with an alkaline phosphatase signal amplification system (R&D Systems, Minneapolis, MN). The intra- and interassay coefficients of variation of IL-6 ranged from 7–12% and from 8–13%, respectively, and sensitivity was 0.094 pg/ml. Duplicate measurements of IL-6 were performed, and the mean of the measurements was used for analysis. High-sensitivity CRP was measured with an automated, high-sensitivity immunonephelometry method (Dade Behring, Inc., Deerfield, IL). This CRP assay has a sensitivity of 0.2 mg/liter. The intraassay coefficients of variation for standard concentrations at 15, 25, and 60 mg/liter are 4, 2.4, and 4.4% (n = 20), respectively. The interassay coefficients of variation for standards at various concentrations are 2.6–5.7%. Those with CRP higher than 15 mg/liter were excluded from analysis because such high levels usually indicate acute illness or trauma.

Longevity and survival time

Vital status was determined for 97% of the original cohort through December 31, 2008, using annual mailings and telephone contacts and was confirmed by death certificates in most cases. Underlying cause of death was coded by a certified nosologist using the ninth revision of the International Classification of Diseases, Adapted.

Covariates

Information on age, sex, smoking status, alcohol use, physical activity, medical history, and medication use was obtained via standard self-administered or interviewer-administered questionnaires. Physical activity was assessed by asking whether the participant reported current exercise three or more times per week. Weight and height were measured with participants wearing light clothing and no shoes. Body mass index was calculated using weight in kilograms divided by height in meters squared. Waist circumference was measured at the level of the iliac crest. Systolic and diastolic blood pressure was measured twice in seated subjects at rest for at least 5 min, and the average of two readings was used.

Serum creatinine was measured by SmithKline Beecham Clinical Laboratories (King of Prussia, PA), and estimated glomerular filtration rate (GFR) was calculated by the modification of diet in renal disease equation (22). Fasting plasma glucose and glucose 2 h after a 75-g oral glucose tolerance test were analyzed using a glucose oxidase method in a hospital diagnostic laboratory. Fasting plasma insulin levels were measured by double-antibody RIA in a subset of 974 participants. Homeostasis model for insulin resistance (HOMA-IR) was calculated for this subset as [fasting glucose (millimoles per liter) × fasting insulin (milliunits per liter)]/22.5. Fasting total cholesterol and triglycerides were measured by enzymatic methods using the ABA-200 biochromatic analyzer (Abbott Laboratories, Abbott Park, IL). High-density cholesterol (HDL) was determined by precipitation analysis using a protocol from the Lipid Research Clinic. LDL cholesterol was calculated with the Friedewald equation (23).

Comorbidities included diabetes, hypertension, CVD, and cancer. Diabetes was defined as self-report of physician diagnosis, antidiabetes medication use, fasting glucose of 126 mg/dl or higher, or 2-h postchallenge glucose of 200 mg/dl or higher. Hypertension was defined as self-report of physician diagnosis, antihypertensive medication use, systolic blood pressure 130 mm Hg or higher, or diastolic blood pressure 85 mm Hg or higher. Prevalent CVD included self-reported myocardial infarction, stroke, transient ischemic attack, or coronary artery revascularization. Cancer was based on self-report, excluding nonmelanoma skin cancer. Number of medications was summed for each participant and included antihypertensive, antidiabetic, lipid-lowering, thyroid hormone, anticoagulants, antiangina, aspirin, cortisone, antiarrhythmic, and digitalis medications.

Statistical methods

Because there is evidence that CRP and IL-6 levels differ between men and women (24), which may result in differential associations with longevity, all models were stratified by sex. Sex-specific characteristics were compared between those who survived and those who died using t test, χ², or Wilcoxon tests as appropriate.

Associations of IL-6 and CRP with survival were evaluated using accelerated failure time (AFT) models with a Weibull distribution. AFT models are parametric survival models that can be used as an alternative to Cox models and provide parameter estimates in terms of percent relative survival time (25). Associations of IL-6 and CRP with age at death (i.e. adult lifespan) among those who died were assessed using linear regression models. Univariate associations of CRP and IL-6 with both longevity and age at death were evaluated first, followed by staged models used to examine potential confounders. CRP and IL-6 were also entered together into a single model. For women, interactions of estrogen use with IL-6 and CRP in fully adjusted survival time and lifespan models were also examined.

To assess the appropriate functional form of IL-6 and CRP with age at death (lifespan), generalized additive models with a cubic B-spline function were used to construct splines. Generalized additive models extend the generalized linear model by allowing fit of nonparametric functions to estimate the associations of predictors and outcomes. To examine the functional form of IL-6 and CRP with survival/longevity, penalized cubic B-splines were constructed, which add a smoothness penalty to the Cox likelihood.

Splines were constructed in SPlus version 8.1 (Tibco Software, Inc., Seattle, WA); all other analyses were conducted in SAS version 9.1.3 (SAS Institute, Cary, NC). P values < 0.05 were considered statistically significant.

Results

Baseline characteristics

At baseline, the mean ± sd age of men and women was similar (73.0 ± 7.5 and 73.0 ± 7.4 yr, respectively). During follow-up, 935 of 1353 (69%) participants died, 465 of 610 (76%) men and 470 of 743 (63%) women. The median follow-up time was 13.3 yr, with a maximum follow-up of 23 yr. Average age at death was 84.7 yr for men and 86.5 yr for women. Mean ± sd CRP and IL-6 levels for women on estrogen were 3.4 ± 3.0 mg/liter and 2.9 ± 2.8 pg/ml, respectively; for women not on estrogen, mean ± sd CRP and IL-6 levels were 2.7 ± 2.7 mg/liter and 3.2 ± 2.5 pg/ml, respectively.

Splines indicated approximately linear associations of both CRP and IL-6 with age at death and longevity for both sexes. Among men, those who survived differed significantly from those who died by age, physical activity, diabetes, CVD, hypertension, systolic blood pressure, estimated GFR, number of medications, and number of comorbidities (Table 1). Among women, age, alcohol use, physical activity, CVD, hypertension, systolic blood pressure, HDL cholesterol, number of medications, estrogen use, and number of comorbidities differed between those who died and those who did not (Table 1). IL-6 levels differed by vital status for both men and women; CRP levels differed for men but not women (Table 1).

Table 1.

Characteristics by mortality and sex of Rancho Bernardo Study participants

	Men			Women
	Survived n = 145	Died n = 465	P valuea	Survived n = 273	Died n = 470	P valuea
Age (yr)	67.8 ± 6.0	74.7 ± 7.1	<0.001	69.2 ± 6.1	75.2 ± 7.2	<0.001
Current smoking [n (%)]	15 (10.3)	48 (10.3)	0.99	36 (13.2)	68 (14.5)	0.63
Daily alcohol use [n (%)]	81 (55.9)	245 (52.7)	0.50	125 (45.8)	167 (35.5)	0.006
Exercise ≥ 3×/wk [n (%)]	133 (91.7)	393 (84.5)	0.03	231 (84.6)	360 (76.6)	0.009
Body mass index (kg/m2)	26.0 ± 2.8	25.5 ± 3.2	0.17	24.5 ± 3.8	24.1 ± 3.8	0.11
Waist circumference (cm)	93.1 ± 8.2	93.5 ± 8.8	0.66	78.6 ± 9.4	79.2 ± 9.8	0.42
Diabetes [n (%)]	13 (9.0)	94 (20.2)	0.002	38 (13.9)	61 (13.0)	0.72
CVD [n (%)]	22 (15.2)	158 (34.0)	<0.001	38 (13.9)	129 (27.5)	<0.001
Hypertension [n (%)]	86 (59.3)	358 (77.0)	<0.001	179 (65.6)	354 (75.3)	0.004
Systolic blood pressure (mm Hg)	134.0 ± 17.6	143.3 ± 20.8	<0.001	136.1 ± 18.3	145.8 ± 23.1	<0.001
Diastolic blood pressure (mm Hg)	78.3 ± 8.6	62.8 ± 14.2	0.22	74.6 ± 9.4	74.8 ± 9.5	0.74
Estimated GFR (ml/min · 1.73 m2)	65.2 ± 11.0	62.8 ± 14.2	0.02	59.2 ± 11.5	58.6 ± 14.2	0.11
HDL cholesterol (mg/dl)	53.7 ± 14.3	53.3 ± 14.4	0.73	70.2 ± 18.7	66.5 ± 18.6	0.009
LDL cholesterol (mg/dl)	134.4 ± 33.2	132.5 ± 34.1	0.59	139.2 ± 38.6	140.0 ± 38.9	0.80
Total cholesterol (mg/dl)	211.7 ± 35.8	209.3 ± 37.9	0.52	233.1 ± 39.9	229.9 ± 40.2	0.29
Triglycerides (mg/dl)	122.2 ± 84.1	123.1 ± 84.0	0.67	122.1 ± 75.5	122.7 ± 73.8	0.72
Fasting glucose (mg/dl)	100.7 ± 12.6	105.3 ± 22.8	0.21	99.4 ± 22.2	98.5 ± 21.7	0.32
Fasting insulin (mU/liter)b	12.0 (8.0, 15.0)	12.0 (8.0, 16.0)	0.63	11.0 (8.0, 14.0)	11.0 (8.0, 16.0)	0.47
HOMA-IRb	2.8 (1.9, 3.8)	2.9 (1.9, 4.2)	0.47	2.7 (1.8, 3.4)	2.6 (1.9, 3.9)	0.69
No. of medications	0.8 ± 1.1	1.1 ± 1.2	<0.001	0.8 ± 0.9	1.2 ± 1.2	<0.001
Estrogen use [n (%)]				105 (38.5)	106 (22.6)	<0.001
No. of comorbidities	1.6 ± 1.2	2.4 ± 1.5	<0.001	1.8 ± 1.2	2.1 ± 1.2	<0.001
IL-6 (pg/ml)	2.6 ± 1.6	3.6 ± 2.8	<0.001	2.6 ± 2.4	3.2 ± 2.5	<0.001
CRP (mg/liter)	2.0 ± 2.1	2.8 ± 2.7	<0.001	2.8 ± 2.7	3.1 ± 3.1	0.22

^aP value obtained by t test, Wilcoxon test, or χ² test as appropriate. 

^bMedian (quartile 1, quartile 3). 

Associations of IL-6 and CRP with survival (Table 2)

Table 2.

Association of IL-6 and CRP with survival among Rancho Bernardo Study participants

			Women
	Men		Estrogen use		No estrogen use
	Time ratio (95% CI)	P value	Time ratio (95% CI)	P value	Time ratio (95% CI)	P value
IL-6 per sda
Unadjusted	0.82 (0.79, 0.86)	<0.001	0.93 (0.86, 1.01)	0.07	0.85 (0.80, 0.89)	<0.001
Age adjusted	0.85 (0.81, 0.88)	<0.001	0.95 (0.87, 1.03)	0.21	0.90 (0.86, 0.94)	<0.001
+Body size/LS/comorbiditiesb	0.85 (0.81, 0.89)	<0.001	0.97 (0.89, 1.06)	0.49	0.90 (0.86, 0.95)	<0.001
+Lipids/blood pressurec	0.85 (0.81, 0.89)	<0.001	0.98 (0.90, 1.06)	0.56	0.92 (0.87, 0.96)	<0.001
+CRP	0.87 (0.83, 0.92)	<0.001	0.97 (0.89, 1.06)	0.54	0.92 (0.87, 0.97)	0.003
CRP per sda
Unadjusted	0.85 (0.8, 0.89)	<0.001	0.95 (0.86, 1.06)	0.37	0.97 (0.92, 1.03)	0.29
Age adjusted	0.86 (0.82, 0.91)	<0.001	0.97 (0.88, 1.06)	0.46	0.96 (0.91, 1.01)	0.14
+Body size/LS/comorbiditiesb	0.89 (0.84, 0.93)	<0.001	0.98 (0.89, 1.09)	0.76	0.96 (0.92, 1.02)	0.17
+Lipids/blood pressurec	0.88 (0.84, 0.93)	<0.001	0.99 (0.89, 1.11)	0.91	0.91 (0.93, 1.03)	0.47
+IL-6	0.91 (0.86, 0.97)	0.003	1.01 (0.89, 1.14)	0.90	1.01 (0.96, 1.07)	0.63

CI, Confidence interval; LS, lifestyle. 

^aThe sd values are 2.6 pg/ml for IL-6 and 2.6 mg/liter for CRP in men and 2.5 pg/ml for IL-6 and 2.9 mg/liter for CRP in women. 

^bModels include age, body mass index, waist circumference, daily alcohol use, current smoking, diabetes, CVD, hypertension, cancer, estimated GFR, and total number of medications. 

^cModels include previous model variables plus systolic blood pressure, diastolic blood pressure, HDL cholesterol, LDL cholesterol, and triglycerides. 

In unadjusted AFT models in men, each sd increase in IL-6 was associated with an 18% decrease in survival time. This association was not materially changed after adjustment for lifestyle, comorbidities, and traditional CVD risk factors; in the adjusted model, each sd increase in IL-6 was associated with a 15% decrease in survival time (P < 0.001). For CRP, each sd increase was associated with a 15% decrease in survival time for men (P < 0.001); the association remained significant although somewhat attenuated after adjustment for potential confounders, with a 12% decrease (P < 0.001).

In women, there were no significant interactions of CRP or IL-6 with estrogen use for survival time (P = 0.47 and P = 0.12, respectively). However, to maintain consistency with other analyses and because of well-known differences in CRP levels in women using oral estrogen compared with those who do not use estrogen, survival time results in women were stratified by estrogen use. In an unadjusted model, each sd increase in IL-6 was associated with a 15% decrease in survival time in women not on estrogen (P < 0.001) and not significantly related to survival time in women on estrogen (P = 0.07). The association in women not on estrogen was attenuated to an 8% decrease with adjustment for lifestyle, comorbidites, and traditional CVD risk factors (P = 0.003). Regardless of estrogen use status, CRP was not significantly associated with survival time in women in unadjusted or fully adjusted models.

Additional adjustment of CRP models for IL-6 (and vice versa) did not substantially alter results for either sex (Table 2). Associations were also unaltered in sensitivity analyses stratifying by age 75 at baseline and in separate models adjusting for use of antihypertensive medications, use of cholesterol-lowering medications, or HOMA-IR (data not shown). The dependence of CRP-survival associations on short-term mortality was examined in separate fully adjusted models by eliminating deaths that occurred during the first 3 and 5 yr of follow-up. The association of CRP with reduced survival time in men was essentially unchanged in these analyses [time ratios = 0.91 (95% confidence interval = 0.88–0.95) and 0.93 (95% confidence interval = 0.90–0.97), for eliminating the first 3 and 5 yr, respectively], and the absence of a significant CRP association for women was not altered regardless of estrogen status (data not shown).

Associations of IL-6 and CRP with adult lifespan (Table 3)

Table 3.

Association of IL-6 and CRP with lifespan among those who died during follow-up in the Rancho Bernardo Study

			Women
	Men		Estrogen use		No estrogen use
βa (95% CI)	P value	βa (95% CI)	P value	βa (95% CI)	P value
IL-6 per sdb
Unadjusted	−0.66 (−1.21, −0.11)	0.02	0.41 (−0.91, 1.72)	0.54	−1.01 (−1.70, −0.32)	0.004
Age adjusted	−1.06 (−1.46, −0.65)	<0.001	−0.10 (−0.99, 0.80)	0.83	−1.56 (−2.02, 1.09)	<0.001
+Body size/LS/comorbiditiesc	−0.94 (−1.34, −0.54)	<0.001	0.13 (−0.90, 1.16)	0.81	−1.41 (−1.89, −0.94)	<0.001
+Lipids/blood pressured	−0.94 (−1.35, −0.53)	<0.001	0.12 (−0.93,1.17)	0.82	−1.35 (−1.84, −0.86)	<0.001
+CRP	−0.69 (−1.13, −0.26)	0.002	−0.02 (−1.10, 1.06)	0.97	−1.45 (−2.06, −0.85)	<0.001
CRP per sdb
Unadjusted	−1.38 (−1.94, −0.83)	<0.001	0.55 (−0.40, 1.94)	0.44	−0.65 (−1.37, 0.06)	0.07
Age adjusted	−1.25 (−1.66, −0.83)	<0.001	0.42 (−0.52, 1.36)	0.38	−0.41 (−0.90,0.09)	0.11
+Body size/LS/comorbiditiesc	−0.92 (−1.33, −0.51)	<0.001	0.87 (−0.14, 1.89)	0.09	−0.20 (−0.71, 0.32)	0.45
+Lipids/blood pressured	−0.89 (−1.31, −0.47)	<0.001	0.79 (−0.33, 1.91)	0.16	−0.17 (−0.68, 0.34)	0.52
+IL-6	−0.61 (−1.06, −0.16)	0.008	0.79 (−0.36, 1.95)	0.17	0.46 (−0.11, 1.02)	0.11

CI, Confidence interval; LS, lifestyle. 

^aβ-Coefficient corresponds to years of life lost or gained. 

^bsd values are 2.6 pg/ml for IL-6 and 2.6 mg/liter for CRP in men, 2.5 pg/ml for IL-6 and 2.9 mg/liter for CRP in women who did not use estrogen, and 2.6 pg/ml for IL-6 and 3.0 mg/liter for CRP for women who were on estrogen. 

^cModels include age, body mass index, waist circumference, daily alcohol use, current smoking, diabetes, CVD, hypertension, cancer, estimated GFR, and total number of medications. 

^dModels include previous model variables plus systolic blood pressure, diastolic blood pressure, HDL cholesterol, LDL cholesterol, and triglycerides. 

In linear regression analyses among the 465 men who died during follow-up, each sd increase in IL-6 corresponded to approximately 0.7 yr shorter adult lifespan in unadjusted models (P = 0.02). After adjustment for covariates, this association was strengthened to an approximately 1-yr reduction in lifespan (P < 0.001). Each sd increase in CRP levels in men was associated with a 1.4-yr shorter lifespan in unadjusted models (P < 0.001) and with approximately 1 yr of reduced lifespan after multivariate adjustment (P < 0.001). CRP and IL-6 levels were moderately correlated in men (r = 0.37; P < 0.001). To examine the independence of CRP and IL-6 associations, we included both in the same multiply adjusted model. Adjusting for the other cytokine reduced associations to 0.7 (P < 0.001) and 0.6 (P < 0.003) years of life lost, respectively, for CRP and IL-6.

Among the 470 women who died during follow-up, the association of CRP and IL-6 with adult lifespan varied by estrogen use status (P for interaction = 0.001 for IL-6, and 0.04 for CRP). Lifespan analyses stratified by estrogen use were significantly different. For women not using estrogen, each sd increase in IL-6 was associated with an approximately 1-yr shorter lifespan (P = 0.004), which persisted after adjustment for covariates. In women using estrogen, levels of IL-6 were not significantly related to lifespan. CRP was not significantly related to adult lifespan in women regardless of estrogen status. Inclusion of CRP and IL-6 in the same adjusted models increased the strength of the IL-6-lifespan association in women not on estrogen but did not alter the null association for CRP or the absence of any cytokine-lifespan association in women using estrogen.

Additional adjustment for HOMA-IR, antihypertensive medications, and cholesterol-lowering medications separately did not change results for IL-6 or CRP (data not shown). Sensitivity analyses stratifying by age 75 at baseline did not alter the significance of the associations between either CRP or IL-6 and lifespan for either men or women (data not shown).

Associations of IL-6 and CRP with cardiovascular and noncardiovascular mortality

Overall, 465 of 935 deaths (49%) in this cohort were attributed to CVD, 227 of which were in men and 229 in women. We examined whether sex differences in IL-6 and CRP mortality associations were specific to cardiovascular deaths. Figure 1 displays IL-6 and CRP associations with CVD vs. non-CVD mortality in men and women based on AFT models. In men, IL-6 and CRP were significantly and positively associated with both CVD and non-CVD death. For women on estrogen, IL-6 and CRP were not significantly associated with either CVD death or non-CVD death. For women not using estrogen, IL-6 but not CRP, was significantly associated with both CVD death and non-CVD death.

Figure 1.

Results of association of CRP and IL-6 with CVD and non-CVD mortality by sex and estrogen status (women). The x-axis shows the non-CVD and CVD mortality by sex/estrogen status, and the y-axis displays the corresponding time ratios with 95% confidence intervals from the AFT models. Models are adjusted for age, body mass index, waist circumference, daily alcohol use, current smoking, diabetes, CVD, hypertension, cancer, estimated GFR, total number of medications, systolic and diastolic blood pressure, LDL and HDL cholesterol, and triglycerides.

Discussion

CRP and IL-6 have been associated with CVD and all-cause mortality in a variety of populations; however, whether they are associated with exceptional longevity has not been well studied. Overall, this study found that higher levels of inflammatory markers were associated with reduced survival and shorter adult lifespan in men and women who were 57 yr or older at the time of the cytokine measurements, but with important differences by sex and by use of estrogen therapy in women. In men, higher levels of IL-6 and CRP were associated with both a decrease in survival time and adult lifespan independent of lifestyle, comorbidities, and other traditional cardiovascular risk factors including insulin resistance. For women in this study, higher IL-6 levels were significantly associated with decreased survival time and shorter lifespan, but only among women who were not using estrogen therapy at the time of the cytokine measurements. CRP was not significantly associated with either survival time or lifespan in women with or without estrogen therapy. Almost half of all deaths were attributed to CVD. These results did not vary by CVD vs. non-CVD mortality for either sex.

There is some evidence that the association of CRP with mortality may wane over time, i.e. the further the mortality event is from the time of CRP measurement, and that CRP may in fact be a better near-term mortality predictor (26,27,28). This has been indicated for both general population studies (27) as well as in those with existing vascular disease (26,28). In the present study, the association of higher CRP with reduced longevity in men persisted when early deaths (those occurring during the first 3 and 5 yr of follow-up) were eliminated from the analyses, implying that CRP may possibly predict longer-term survival in men who have already lived beyond middle age.

Few studies have evaluated the association of CRP or IL-6 with exceptional longevity. Arai et al. (29) did not find either CRP or IL-6 associated with all-cause mortality in men and women centenarians in models adjusted for traditional cardiovascular risk factors, physical function, and total number of comorbidities. In a group of men and women nonagenarians, Jylhä et al. (30) also found no association of CRP or IL-6 with all-cause mortality in models adjusted for traditional risk factors. In another study of nonagenarians, Kravitz et al. (31) found that higher CRP was associated with increased risk of all-cause mortality but not dementia. However, these studies had small sample sizes (208–285 participants) and much shorter follow-up periods (∼4–6 yr) and concentrated on cohorts that had already achieved 90+ or 100+ yr old, so inflammatory markers were measured much more proximal to time of death. In contrast, the present study has a much larger sample size and longer follow-up of participants who were younger at baseline (median age 73 yr). Additionally, we employed statistical methods that gave us percent decrease in survival time or years of decrease in lifespan as opposed to hazard ratios, which indicate only relative risk for CRP and IL-6 with all-cause mortality. These results indicate that inflammation measured years earlier may play an important role in survival to age 80 and beyond, particularly for men.

The present study found no significant associations of CRP with survival and lifespan in women regardless of estrogen use status. Recent studies suggest that it is unlikely that CRP is causally related to CVD (32,33) and may more appropriately be defined as an indicator of overall inflammatory state or immune response. Although it is acknowledged that chronic inflammation is a threat to survival (34), inflammatory processes are an essential component of the protective immune response (35). Low levels of inflammatory cytokines in an elderly woman could indicate immune deficiency, and in turn, higher levels could possibly indicate an intact and/or healthy immune system. Our discrepant results for CRP in men and women could also indicate the presence of an unknown or unadjusted confounder for women that counteracts detrimental effects of higher CRP on longevity. Because adiposity is associated with levels of inflammatory markers (36), and adipose tissue distributions vary by sex (37), we adjusted for waist circumference and body mass index, but more precise measures of regional adiposity such as visceral fat or sc fat were not available.

Among women who were estrogen users, there was a significant association of higher IL-6 with shorter lifespan, which was not seen in women on estrogen therapy. Previous reports of an association between IL-6 with exogenous estrogen have been inconsistent with some studies finding estrogen use associated with higher levels of IL-6, whereas other studies have found no association or an inverse association (38,39). And although the association of CRP with lifespan in women was not significant, there was a suggestion of association of higher CRP with longer lifespan among estrogen users and shorter lifespan among non-estrogen users. This association could be an artifact of estrogen use affecting CRP levels and not due to any underlying inflammatory state per se. It is well established that women who use oral hormone replacement therapy have higher levels of CRP (39,40,41,42) and that women who use estrogen tend to live longer, possibly due to healthy-user bias.

Our study has many strengths but also some limitations. The Rancho Bernardo Study is a large prospective cohort who is well characterized for covariates and has nearly complete follow-up for vital status. Even though plasma samples were collected at a 1984–1987 examination, the measurement of CRP and IL-6 occurred in 2000, allowing the use of high-sensitivity methods. The Rancho Bernardo cohort comprises participants of primarily Caucasian and middle to upper-middle class socioeconomic status, which limits generalizability but concurrently greatly reduces unmeasured confounding due to these factors. As in all observational studies, residual confounding is possible. We did not incorporate information on estrogen dose or perform drug-specific analyses; however, more than 80% of women taking estrogen at this time were taking unopposed oral conjugated equine estrogen at a daily dose of 0.625 mg.

Our results indicate that higher levels of IL-6 and CRP predict reduced survival and lifespan even among older men who live an exceptionally long life, with a significant IL-6 association only in women not taking estrogen. Inflammation is often associated with chronic diseases such as diabetes and CVD, which are in turn associated with decreased survival. Replication of these findings is needed as well as further investigation of the association of CRP with survival and lifespan in women. Studying factors associated with exceptional longevity is of utmost importance as the size of the elderly population increases.