Abstract
Background.
Adiponectin has anti-inflammatory properties, and its production is suppressed by inflammatory factors. Although elevated levels of adiponectin and inflammatory markers each predict mortality in older adults, the implications of their interdependent actions have not been examined.
Methods.
We investigated the joint associations of levels and interval changes in adiponectin, C-reactive protein (CRP), and interleukin 6 (IL-6) with risk of death in 840 older adults participating in a population-based study. Adiponectin, CRP, and IL-6 were measured in samples collected 8.9 (8.2–9.8) years apart, and all-cause mortality was subsequently ascertained (n = 176).
Results.
Interval changes and end levels of adiponectin, CRP, and IL-6 showed mostly positive, independent associations with mortality, without evidence of multiplicative interaction. Joint models, however, showed an U-shaped relationship between end level of adiponectin and outcome (hazard ratio [HR] [95% CI] = 0.72 [0.52–0.99] per standard deviation [SD] for levels <20.0 mg/L; HR = 1.91 [1.61–3.44] per SD for levels ≥20.0 mg/L). Participants with the greatest longitudinal increases (upper quartile) in both adiponectin and inflammatory markers had a higher risk of death (HR = 2.85 [1.78–4.58]) than those with large increases in adiponectin alone (HR = 1.87 [1.20–2.92]) (p = .043), but not inflammatory markers alone (HR = 2.48 [1.67–3.67]) (p = .55), as compared with smaller changes for both.
Conclusion.
Higher levels or interval change in adiponectin and inflammatory markers predict increased mortality in older persons independent of each other, although for adiponectin, the association appears inverse below 20 mg/L. These findings suggest that inflammatory and noninflammatory mechanisms governing aging-related decline operate in parallel and provide a potential explanation for paradoxical adiponectin–outcome associations reported previously.
Keywords: Adiponectin, C-reactive protein, Interleukin 6, Aging, Mortality
THE association between chronic inflammation and aging-related decline is well documented (1). According to the molecular inflammation hypothesis, aging is accompanied by a diminished capacity for maintenance of cellular redox balance (2). The resulting accumulation of reactive oxidative species leads to chronic upregulation of inflammatory molecules, which act both as markers of heightened oxidative stress and as direct mediators of further homeostatic dysregulation across organ systems (2). Such chronically elevated levels of inflammatory markers are a common feature not just of reduced survival in older cohorts but also of a variety of aging-associated diseases (1,2).
The adipocyte-derived hormone adiponectin has been demonstrated in laboratory studies to have insulin-sensitizing, antiatherogenic, and anti-inflammatory properties (3). On the anti-inflammatory side, adiponectin can reduce activation of nuclear factor kappa B, the principal orchestrator of proinflammatory cytokine gene expression (4,5) and promote clearance of proinflammatory apoptotic cells (6). In obesity-associated disorders, where inflamed adipose tissue is a forerunner of insulin resistance and atherosclerosis (7), a reciprocal relationship between adiponectin and proinflammatory mediators appears to be of major importance (3). Specifically, monocyte infiltration of the adipose compartment leads to increased secretion of tumor necrosis factor α and interleukin 6 (IL-6) (7), cytokines that directly inhibit adipocyte secretion of adiponectin (8). Removal of adiponectin's suppressive actions on the cytokines’ own production serves in turn to further amplify the underlying inflammatory process (7).
We have recently reported that adiponectin, for all its beneficial metabolic and inflammation-countering actions, increases longitudinally with age and that such an increase portends higher mortality (9). Our group has also shown that elevations in C-reactive protein (CRP) and IL-6, markers that account for incompletely overlapping aspects of the inflammatory response, are also adversely related to functional survival (10). These observations, however, leave unexplored the potential interrelation of adiponectin on the one hand and IL-6 and CRP on the other as determinants of all-cause mortality in the oldest old. Whether the pathways linking adiponectin and inflammatory mediators with mortality are partly or wholly interdependent, or even synergistic, remains unknown. We investigated these bioactive peptides jointly to determine whether they reflect separate or shared pathogenetic processes acting to shorten survival in late life.
METHODS
Study Population and Procedures
Cardiovascular Health Study (CHS) All Stars is an ancillary investigation of CHS that was undertaken to identify factors associated with successful aging. Procedures for both CHS and CHS All Stars have been described (11,12). Briefly, CHS is a population-based survey of risk factors for cardiovascular disease (CVD) in community-living adults 65 years of age or older. A cohort of 5,888 participants was recruited in 1989–1990 and 1992–1993 from Forsyth County, North Carolina; Sacramento County, California; Washington County, Maryland; and Allegheny County, Pennsylvania. The 1992–1993 examination included 5,553 individuals, of whom 1,677 (30.2%) survived and participated in the CHS All Stars follow-up examination in 2005–2006 (12). For 68% of participants, visits were conducted in person and a blood sample obtained. Blood specimens collected in 2005–2006 were paired with stored samples from the 1996–1997 examination to evaluate biochemical factors of potential relevance to aging.
During the 1996–1997 and 2005–2006 examinations, information on participants’ social and medical histories was updated. Determinations of blood pressure and body size were performed in standardized fashion, as were fasting laboratory measurements (11,13,14).
Paired measurements of adiponectin, CRP, and IL-6 were available in 988, 970, and 920 CHS All Stars participants, respectively. Since extreme elevations in CRP reflect acute instead of chronic inflammation (15), consistent with which end levels of CRP ≥ 20 mg/L showed an inverse association with mortality in graphical analyses, we excluded 55 participants with such levels. Remaining participants with available measurements for all three biomarkers (n = 865) were slightly younger and more often reported good-to-excellent health than those without (n = 812). Because of a significant influence of angiotensin–receptor blocker and thiazolidinedione use on adiponectin concentrations, individuals on such medications (n = 8 and n = 17, respectively) were excluded (9), leaving 840 eligible participants (546 women).
Definitions of Risk Factors
Hypertension, diabetes, and CVD were defined as previously reported (9). Clinical CVD was ascertained in the 1996–1997 and 2005–2006 examinations by combining the CHS questionnaire, medical record review, and physician confirmation (16). Follow-up surveillance and ascertainment entailed biannual interviews of participants (17). Potential incident events were investigated by review of medical records. These were initially classified by local physicians and subsequently adjudicated by a CHS committee using standardized criteria (11,17).
Outcome
All-cause mortality following the CHS All Stars visit was ascertained by telephone follow-up, proxy contact, and review of obituaries, available through June 2009.
Measurement of Biomarkers
All measurements were performed at the CHS Core Laboratory in May/June 2008 on serum stored at −70°C since venipuncture. Levels of adiponectin, CRP, and IL-6 were measured concurrently in paired samples from 1996–1997 and 2005–2006. Adiponectin was measured with an enzyme-linked immunosorbent assay (ELISA) (R&D Systems, Minneapolis, MN); intra- and interassay coefficients of variations were 2.5–4.7% and 5.8–6.9%, respectively. CRP was measured by BNII nephelometer (N High Sensitivity CRP; Dade Behring Inc., Deerfield, IL); intra- and interassay coefficients of variations were 0.7–3.1% and 3.5–6.0%, respectively. IL-6 was measured by ultrasensitive ELISA (Quantikine HS Human IL-6 Immunoassay; R&D Systems); intra- and interassay coefficients of variations were 2.9–8.7% and 7.3–9.0%, respectively.
Statistical Analysis
Owing to significant differences by sex in risk factors and mean adiponectin change, we examined levels of covariates according to sex-specific quartiles of adiponectin change and tested differences using chi-square or analysis of variance tests for trend. Pearson coefficients were computed to assess biomarker correlations. The functional form of the association of continuous levels of each biomarker with mortality was assessed with Lowess plots and cubic splines. Because there was no evidence of effect modification by sex in regression analyses using continuous splines, categorical analyses of adiponectin and inflammatory markers with outcome used overall, and not sex specific, quartiles of biomarkers in order to mirror the continuous associations. For binary comparisons, we selected a priori the cut-point for the highest quartile in biomarker distributions to permit comparison of high level or change (upper quartile) versus low level or change (lower three quartiles), with the exception of adiponectin level, for which previous analyses documented an inflection point at 20 mg/L in its relation with mortality (9). Associations of adiponectin and inflammatory markers with all-cause mortality were evaluated with Cox models. The Wald test was used to evaluate the significance of individual coefficients across categories defined by high and low levels or change. Modeling was done in stages, initially adjusting for age, sex, and race, adding significant clinical covariates thereafter and subsequently adjusting for interval weight change and biomarker sampling interval. Multiplicative interactions were tested by appropriate cross-product terms. Analyses were performed with STATA, version 11.0 (College Station, TX).
RESULTS
During a mean interval of 8.9 (range, 8.2–9.8) years, adiponectin concentrations increased by 2.7 ± 4.5 and 2.5 ± 4.6 mg/L to 13.3 ± 6.6 and 17.8 ± 7.4 mg/L in men and women, respectively (both p < .001). In the same period, CRP did not significantly change in men (p = .33), but declined in women by 0.7 ± 3.8 mg/L (p < .001), with end levels of 2.9 ± 3.1 and 3.1 ± 3.3 mg/L, respectively. In turn, IL-6 increased in both sexes (p < .001) by 0.9 ± 2.5 and 0.8 ± 2.3 pg/mL to 4.2 ± 2.3 pg/mL and 3.8 ± 2.2 pg/mL in men and women, respectively. There were accompanying interval decreases in body weight of 2.3 ± 5.8 in men and 4.0 ± 6.9 kg in women to 78.0 ± 12.3 and 65.1 ± 12.8 kg, respectively (both p < .001).
The distribution of clinical characteristics by sex-specific quartiles of interval change in adiponectin is shown in Table 1. Increasing sex-specific quartiles of adiponectin change were significantly associated with older age, lower body weight, waist circumference, and CRP (but not IL-6) level, and less beta-blocker and statin use at the 2005–2006 visit, as well as greater decline in weight, waist, and CRP during the same interval. There were no significant correlations between levels or change of adiponectin and corresponding levels or change of CRP or IL-6. Adiponectin concentration was negatively correlated with body weight in 2005–2006 (r = −.38, p < .001), as was interval change in adiponectin with interval change in weight (r = −.31, p < .001). End level of IL-6, though not CRP, exhibited a modest positive correlation with weight (r = .11, p = .001), but no significant correlation was observed for change in either marker and change in weight. As expected, CRP and IL-6 concentrations and their corresponding interval changes were themselves strongly positively correlated (r = .45 and r = .48, p < .001 for both).
Table 1.
Associations of Interval Change in Adiponectin (1996–1997 to 2005–2006) With Clinical Characteristics at the 2005–2006 Examination or Their Interval Change*
| Variables | Sex-Specific Quartile† of Adiponectin Change |
p Trend | |||
| Q1, n = 211 | Q2, n = 209 | Q3, n = 211 | Q4, n = 209 | ||
| Age (y) | 84.8 ± 3.5 | 84.9 ± 3.5 | 85.0 ± 3.5 | 86.2 ± 3.8 | <.001 |
| Black race | 29 (13.7) | 42 (20.1) | 34 (16.1) | 21 (10.0) | .18 |
| Weight (kg) | 70.6 ± 14.0 | 71.9 ± 15.1 | 69.9 ± 13.2 | 66.0 ± 13.2 | <.001 |
| Waist circumference (cm) | 97.9 ± 14.0 | 100 ± 14.1 | 98.6 ± 12.0 | 94.1 ± 12.6 | .001 |
| Hypertension | 146 (69.5) | 150 (71.8) | 149 (70.6) | 149 (71.3) | .77 |
| Cardiovascular disease | 96 (45.5) | 78 (37.3) | 75 (35.6) | 83 (39.7) | .20 |
| Diabetes | 31 (14.8) | 27 (13.0) | 32 (15.3) | 30 (14.6) | .86 |
| Fasting glucose (mg/dL) | 102 ± 25 | 101 ± 23 | 103 ± 27 | 98 ± 23 | .22 |
| Total cholesterol (mg/dL) | 181 ± 40 | 189 ± 39 | 181 ± 38 | 182 ± 36 | .73 |
| Current smoking | 6 (2.8) | 5 (2.4) | 11 (5.3) | 3 (1.4) | .80 |
| Alcohol (drinks/wk) | 2.0 ± 4.8 | 1.3 ± 3.1 | 1.5 ± 3.6 | 2.1 ± 4.8 | .72 |
| Self-reported poor/fair health | 38 (18.0) | 37 (17.7) | 42 (20.0) | 47 (22.5) | .20 |
| Creatinine‡ (mg/dL) | 1.01 ± 0.31 | 1.00 ± 0.37 | 1.00 ± 0.57 | 1.04 ± 0.59 | .34 |
| Adiponectin (mg/L) | 13.6 ± 6.6 | 13.3 ± 7.1 | 16.1 ± 6.4 | 22.2 ± 6.0 | <.001 |
| Change in weight (kg) | −1.2 ± 5.7 | −2.2 ± 6.2 | −3.7 ± 5.3 | −6.6 ± 7.8 | <.001 |
| Change in waist (cm) | 2.5 ± 8.7 | 2.4 ± 9.5 | 0.8 ± 8.4 | −1.8 ± 9.5 | <.001 |
| CRP‡ (mg/L) | 2.1 ± 3.9 | 2.0 ± 2.7 | 2.0 ± 3.1 | 1.6 ± 3.1 | .004 |
| Change in CRP (mg/L) | 0.5 ± 3.9 | −0.9 ± 3.6 | −0.4 ± 3.5 | −0.9 ± 3.7 | .002 |
| IL-6‡ (pg/mL) | 3.5 ± 2.3 | 3.1 ± 2.0 | 3.5 ± 2.2 | 3.4 ± 2.3 | .92 |
| Change in IL-6 (pg/mL) | 1.0 ± 2.5 | 0.6 ± 2.1 | 1.0 ± 2.3 | 0.9 ± 2.5 | .98 |
| Medications | |||||
| Insulin | 7 (3.3) | 3 (1.4) | 7 (3.4) | 7 (3.4) | .68 |
| Oral hypoglycemic | 16 (7.6) | 17 (8.2) | 23 (11.0) | 18 (8.8) | .47 |
| Beta-blocker | 101 (48.1) | 78 (37.5) | 62 (29.7) | 59 (28.8) | <.001 |
| Angiotensin–receptor blocker | 22 (10.5) | 31 (14.9) | 36 (17.2) | 33 (16.1) | .08 |
| Angiotensin-converting enzyme inhibitor | 45 (23.8) | 47 (24.6) | 47 (24.5) | 49 (26.2) | .62 |
| Statin | 86 (41.0) | 91 (43.8) | 72 (34.4) | 70 (34.1) | .05 |
Notes: CRP = C-reactive protein; IL-6 = interleukin 6.
Entries are mean ± SD or n (%).
Sex-specific quartile cut-points were −0.3, 2.0, and 5.0 mg/L in women and 0.4, 2.3, and 5.2 mg/L in men.
Geometric mean.
During 3 years of follow-up, 176 participants (99 of 546 [18%] women and 77 of 294 [26%] men) died. Graphical assessment showed a fairly linear relation between end level of CRP and all-cause mortality after exclusion of participants with CRP ≥ 20 mg/L, but this was not true for end level of IL-6 or, as documented previously, adiponectin (9)(Supplementary Figure 1). Specifically, risk of death increased linearly for IL-6 for levels up to about 8.0 pg/mL, but plateaued above this value (n = 63). By contrast, interval changes showed no graphical evidence of departure from linearity for any of the biomarkers. Risk estimates for end concentrations and interval changes in biomarkers were similar at all levels of adjustment. Fully adjusted model results are shown in Table 2. Associations of individual biomarkers with mortality were unchanged when waist circumference and its change were added to the multivariable models. Significant adjusted associations with mortality were present for both end level and interval change of each analyte. When both the 2005–2006 concentration of each biomarker and its antecedent change were entered together in the same model however, only the end level of CRP and IL-6, but not their interval change, remained significantly associated with outcome. By contrast, when considered jointly, both the 2005–2006 level of adiponectin and its antecedent change continued to be significantly related to subsequent mortality in the fully adjusted model. Notably, as compared with the previously reported null association for adiponectin levels below 20 mg/L (9), the relation became significantly inverse in the lower portion of the distribution, while remaining positive for higher end levels or for interval change (hazard ratio [HR] = 0.67 per standard deviation [SD] <20.0 mg/L [0.48–0.93], HR = 1.91 per SD ≥20.0 mg/L [1.28–2.84], and HR = 1.21 per SD of change [1.01–1.45]). In addition, the previously noted (9) interaction of adiponectin level with race was no longer observed in the smaller subgroup evaluated here.
Table 2.
Association of Continuous Levels or Interval Changes of Adiponectin and Inflammatory Markers With Mortality
| Biomarker Range | Adjusted HR*,† (95% CI) | p | |
| End levels | |||
| Individual biomarkers | |||
| Adiponectin | <20 mg/L | 0.75 (0.54–1.02) | .07 |
| ≥20 mg/L | 2.11 (1.44–3.10) | <.001 | |
| CRP | All | 1.34 (1.19–1.52) | <.001 |
| IL-6 | <8 pg/mL | 1.66 (1.37–2.01) | <.001 |
| ≥8 pg/mL | 1.20 (0.71–2.02) | .50 | |
| Adiponectin and CRP together | |||
| Adiponectin | <20 mg/L | 0.75 (0.54–1.02) | .07 |
| ≥20 mg/L | 2.20 (1.50–3.22) | <.001 | |
| CRP | All | 1.36 (1.20–1.54) | <.001 |
| Adiponectin, CRP, and IL-6 together‡ | |||
| Adiponectin | <20 mg/L | 0.72 (0.52–0.99) | .044 |
| ≥20 mg/L | 2.35 (1.61–3.44) | <.001 | |
| CRP | All | 1.15 (0.99–1.33) | .06 |
| IL-6 | <8 pg/mL | 1.60 (1.30–1.96) | <.001 |
| ≥8 pg/mL | 1.18 (0.69–2.01) | .55 | |
| Interval change in levels | |||
| Individual biomarkers | |||
| Change in adiponectin | All | 1.23 (1.05–1.44) | .008 |
| Change in CRP | All | 1.37 (1.17–1.59) | <.001 |
| Change in IL-6 | All | 1.48 (1.28–1.71) | <.001 |
| ΔAdiponectin and ΔCRP together | |||
| Change in adiponectin | All | 1.26 (1.08–1.46) | .004 |
| Change in CRP | All | 1.40 (1.19–1.63) | <.001 |
| ΔAdiponectin, ΔCRP, and ΔIL-6 together§ | |||
| Change in adiponectin | All | 1.24 (1.06–1.46) | .007 |
| Change in CRP | All | 1.21 (1.01–1.45) | .035 |
| Change in IL-6 | All | 1.37 (1.16–1.61) | <.001 |
Notes: HR = hazard ratio; CI = confidence interval; CRP = C-reactive protein; IL-6 = interleukin 6. SD = standard deviation.
Per SD increase. For adiponectin, end-level SD = 7.5 mg/L and change SD = 4.6 mg/L; for CRP, end-level SD = 3.2 mg/L and change SD = 3.7 mg/L; for IL-6, end-level SD = 2.2 pg/mL and change SD = 2.4 pg/mL.
Adjusted for age, sex, race, weight, diabetes, number of alcoholic drinks/wk, cardiovascular disease, fair or poor self-reported health, interval change in weight, and sample collection interval.
Corresponding HR (95% CI) for weight (per SD [14 kg] lower value) was 1.08 (0.98–1.18), p = .12.
Corresponding HR (95% CI) for interval change in weight (per SD [6.6 kg] decline) was 1.23 (1.09–1.45), p = .017.
When considered jointly, continuous levels of adiponectin and CRP were significantly associated with survival independently of each other (Table 2). Although CRP exhibited a positive association with mortality, there was a suggestion of an U-shaped relationship for adiponectin above and below its inflection point, though the inverse association at levels less than 20 mg/L was not significant. When IL-6 was added to the model, this biomarker was also positively related to increasing risk of death below 8.0 pg/mL, but its inclusion rendered the CRP association nonsignificant. Furthermore, inclusion of IL-6 made the inverse association of adiponectin statistically significant. By contrast, weight was not significantly associated with outcome in this model (Table 2). No significant effect–measure modification was observed in these joint analyses for race, age, sex, or prevalent CVD (all p > .20) or between adiponectin and CRP or IL-6 for levels below or above their respective inflection points (all p ≥ .10).
Evaluation of the joint influence of interval changes in adiponectin and inflammatory molecules revealed positive associations with future risk of death that were independent of each other, even when all three biomarkers were considered together (Table 2). The magnitudes of these associations were comparable to that for interval decline in weight. There were again no significant interactions by the aforementioned covariates (p ≥ .14) or between biomarker changes themselves (p ≥ .63).
We next examined whether participants who exhibited the greatest interval change for adiponectin and each inflammatory marker were at excess risk compared with those showing this for just one biomarker. As summarized in Table 3, individuals having both adiponectin and CRP increases in their respective upper quartiles tended to have the highest adjusted risk of death, but this was not significantly different from those for whom this was true only for CRP (p = .44) or adiponectin (p = .08). Likewise, the adjusted risk for mortality associated with high interval change in both adiponectin and IL-6 did not differ significantly from high interval change in IL-6 alone (p = .64), although a near-significant difference was observed compared with high change in adiponectin alone (p = .06) (Table 3).
Table 3.
Joint Associations With Mortality of Change in Adiponectin and Change in Inflammatory Markers According to the Upper Quartile of Their Distributions
| No. Deaths/No. at Risk | Incidence Rate* | Adjusted HR†(95% CI) | p | |
| ΔAdiponectin and ΔCRP‡ | ||||
| Low ΔAPN, low ΔCRP | 66/466 | 40.3 (31.7–51.4) | 1.00 (referent) | — |
| Low ΔAPN, high ΔCRP | 41/164 | 77.4 (57.0–105) | 1.98 (1.33–2.96) | .001 |
| High ΔAPN, low ΔCRP | 46/164 | 85.9 (64.4–114) | 1.53 (1.03–2.29) | .036 |
| High ΔAPN, high ΔCRP | 23/46 | 190 (126–286) | 2.48 (1.47–4.21) | .001 |
| Interaction p value | .56 | |||
| ΔAdiponectin and ΔIL-6‡ | ||||
| Low ΔAPN, low ΔIL-6 | 65/482 | 38.4 (30.1–48.9) | 1.00 (referent) | — |
| Low ΔAPN, high ΔIL-6 | 42/148 | 89.0 (65.8–120) | 2.20 (1.47–3.28) | <.001 |
| High ΔAPN, low ΔIL-6 | 38/148 | 78.1 (56.8–107) | 1.51 (1.00–2.32) | .053 |
| High ΔAPN, high ΔIL-6 | 31/62 | 183 (129–260) | 2.49 (1.54–4.03) | <.001 |
| Interaction p value | .38 | |||
Notes: CI = confidence interval; HR = hazard ratio; CRP = C-reactive protein; IL-6 = interleukin 6.
Per 1,000 person-years.
Compared with referent category having both biomarkers in the bottom three quartiles. Adjusted for same covariates as in Table 2.
Cut-points for high change based on upper quartile vs lower three quartiles: adiponectin (≥5.1 mg/L), CRP (≥0.9 mg/L), and IL-6 (≥1.9 pg/mL).
Last, we accounted for changes in CRP and IL-6 simultaneously, using a categorization of high inflammation change when either of the two markers fell within its upper quartile versus low inflammation change when both were within their bottom three quartiles. As shown in Figure 1A, the concurrence of high adiponectin change and high inflammation change was associated with a marked increase in all-cause mortality as compared with low adiponectin and inflammation change. The corresponding risk of death for high change in either adiponectin or inflammatory markers was intermediate. As compared with the referent category of low adiponectin and low inflammation changes, the HR (95% CI) for high change in both was 2.82 (1.70–4.70) after full adjustment. This was significantly higher than the adjusted HR for high adiponectin and low inflammation changes of 1.47 (0.91–2.37) (p = .043) but not the adjusted HR for high inflammation and low adiponectin changes (p = .55). Findings were similar (Figure 1B) when 2005–2006 levels of the three biomarkers were assessed with the categorization of high (≥20 mg/L for adiponectin [n = 250]; upper quartile for CRP [≥4.0 mg/L] and IL-6 [≥2.3 pg/mL]) versus low (<20 mg/L for adiponectin; lower three quartiles for CRP and IL-6). Full adjustment yielded HRs (95% CI) of 2.85 (1.78–4.58) for high adiponectin and high inflammation, 1.69 (1.05–2.79) for high adiponectin and low inflammation (p = .013 vs both high), and 2.48 (1.67–3.67) for high inflammation and low adiponectin (p = .33 vs both high), as compared with the referent category.
Figure 1.
(A) Change in adiponectin and inflammatory markers between 1996–1997 and 2005–2006 and cumulative mortality. p Value is for overall log-rank test. (B) Levels of adiponectin and inflammatory markers in 2005–2006 and cumulative mortality. p Value is for overall log-rank test.
DISCUSSION
In this community-based sample of adults surviving to advanced old age, we found that longitudinal increases in the concentrations of adiponectin, CRP, and IL-6 were independent predictors of all-cause mortality. There was no evidence of synergy in the positive associations of adiponectin and inflammatory markers with mortality, but concurrent longitudinal increases in levels of these peptides were associated with higher risk than increments in adiponectin alone. These findings contribute to the existing literature by highlighting, for the first time, that the adverse implications of age-related increases in adiponectin and inflammatory markers are largely unrelated.
Unlike adiponectin, interval changes of CRP and IL-6 were not significantly associated with mortality when their end levels were taken into account, but similar independent associations were observed when end levels of the biomarkers were considered. In contrast to interval changes, however, the positive linear associations of end concentrations of adiponectin and IL-6 with mortality did not apply for the full range of measured values.
In the case of IL-6, the relationship with outcome leveled off at the upper end of the biomarker's distribution, a feature that has not been noted in other studies of older adults linking this peptide with mortality (18,19). We excluded participants with extreme elevations of CRP from the present study in order to avoid distortion of the biomarker's relation to mortality by transient increases associated with acute illnesses (15). Whether acute, reversible inflammatory processes explain the plateauing of risk observed for IL-6, or this is due to the imprecision associated with a small number of participants with very high levels or instead relates to the accuracy of the assay is uncertain. Buttressed as these positive associations for single measurements of IL-6 and CRP are by the observation that their longitudinal increases similarly portend a higher risk of death (10,20), their mechanistic basis is not well defined. Accumulating evidence, however, supports the notion that progressive age-related inability to counter cellular oxidative stress activates redox-sensitive inflammatory genes and that the resulting low-grade inflammation sets in motion multiple deleterious processes that lead to aging-associated diseases and decline of organ function (1,2).
With regard to adiponectin, analyses revealed an U-shaped association with mortality not uncovered in our earlier investigation (9). Previous work highlights a paradox between the protective metabolic actions and associations of adiponectin in preclinical investigations and younger, disease-free populations and the adverse associations of this adipokine with cardiovascular outcomes and mortality in older cohorts and those with prevalent CVD (9,21). A leading explanation for the finding that longitudinal increases in adiponectin worsen survival in older adults is that these increases occur in response to homeostatic dysregulation and serve as a marker for such dysregulation and/or themselves compound the dysregulation through direct deleterious actions of the adipokine (9). The observation that adiponectin-raising genetic polymorphisms may be more common in people with exceptional longevity (22), however, would support the concept that the adipokine's favorable metabolic actions confer a net benefit over the life course and that increases late in life represent unsuccessful attempts to compensate for aging-related homeostatic decline (9).
Our finding of an U-shaped relationship for end level of adiponectin could reconcile the paradox noted earlier. Such a relationship shows that older adults with lower circulating adiponectin up to levels in the moderately high range are indeed at a health disadvantage. This is consistent with findings in younger cohorts, who typically have adiponectin concentrations in this range (3), wherein the adipokine's favorable cardiometabolic associations may be most manifest. Only at more elevated concentrations, starting here at the 70th percentile, does higher adiponectin associate with reduced survival. This suggests that very high levels, much like longitudinal increases in this age group, may predominantly signal and/or contribute to greater homeostatic dysregulation, accounting for their adverse prognostic implications. Previous studies of adiponectin and mortality in elders, however, have not documented differential associations by level, although analysis by cubic splines was not reported (23–25). Because this U-shaped relation, which emerged in our cohort only after exclusion of participants with acute illness, was only significant when accounting simultaneously for interval change or inflammatory markers, it will require independent replication.
Whatever the basis for the heightened mortality observed for longitudinal increases or very high levels of adiponectin, the current data support the premise that the mechanisms involved are to a considerable extent unrelated to inflammatory processes. Indeed, the independent associations of adiponectin and inflammatory changes with mortality occurred despite an inverse relation of higher quartiles of adiponectin change to interval change in CRP (but not IL-6), the latter consistent with the reciprocal relationship reported for adiponectin and inflammation (3). To be sure, the complex intersection of inflammatory and noninflammatory processes, and the presence of directionally opposite effects, does not allow their interdependence to be completely ruled out. But our observations do underscore that independent prognostic information is conveyed by adiponectin and inflammatory measures separately.
Several limitations deserve mention. CHS All Stars participants with available biomarker measurements reported better health status than those without, such that our findings apply only to healthier individuals surviving to advanced old age. Although study of the oldest old is by definition characterized by this survivorship bias, our focus on death following the second biomarker measurement safeguards the internal validity of our results. An additional limitation is the current lack of information on cause of death; analyses of cause-specific mortality will become possible with longer follow-up. Furthermore, our findings must be interpreted in the context of modest study power to detect interactions or additive effects between biomarkers or biomarkers and clinical covariates or to fully define the nature of the relationship between biomarkers and outcome throughout their distributions. Finally, we did not measure the high–molecular-weight adiponectin fraction, which may have stronger glycometabolic benefits and will require further investigation.
In conclusion, the current findings show that levels of adiponectin and inflammatory markers, and their interval change, exhibit significant positive associations with all-cause mortality that are independent of each other and of other potential confounders. Additional work is necessary to explore the nature of the adiponectin association at different ends of the adiponectin range and to determine whether concurrent measurement of adiponectin and inflammatory markers could improve risk prediction in this population.
SUPPLEMENTARY MATERIAL
Supplementary material can be found at: http://biomed.gerontologyjournals.org/.
FUNDING
National Institute on Aging (AG-023629 to CHS All Stars Study); the National Heart, Lung, and Blood Institute with additional contribution from the National Institute of Neurological Disorders and Stroke (http://www.chs-nhlbi.org/pi.htm) (contract numbers N01-HC-85079 through N01-HC-85086, N01-HC-35129, N01 HC-15103, N01 HC-55222, N01-HC-75150, and N01-HC-45133 and grant number U01 HL080295 to CHS); the National Institute on Aging (R01 AG-15928, R01 AG-20098, and AG-027058); National Heart, Lung and Blood Institute (R01 HL-075366); the University of Pittsburgh Claude. D. Pepper Older Americans Independence Center (P30-AG-024827 to A.B.N.); National Heart, Lung and Blood Institute (K23 HL-070854 and R01 HL-094555 to J.R.K.).
Supplementary Material
References
- 1.Franceschi C, Capri M, Monti D, et al. Inflammaging and anti-inflammaging: a systemic perspective on aging and longevity emerged from studies in humans. Mech Ageing Dev. 2007;128(1):92–105. doi: 10.1016/j.mad.2006.11.016. [DOI] [PubMed] [Google Scholar]
- 2.Chung HY, Cesari M, Anton S, et al. Molecular inflammation: underpinnings of aging and age-related diseases. Ageing Res Rev. 2009;8(1):18–30. doi: 10.1016/j.arr.2008.07.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Ziemke F, Mantzoros CS. Adiponectin in insulin resistance: lessons from translational research. Am J Clin Nutr. 2010;91(1):258S–261S. doi: 10.3945/ajcn.2009.28449C. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Ouchi N, Kihara S, Arita Y, et al. Adiponectin, an adipocyte-derived plasma protein, inhibits endothelial NF-kappaB signaling through a cAMP-dependent pathway. Circulation. 2000;102(11):1296–301. doi: 10.1161/01.cir.102.11.1296. [DOI] [PubMed] [Google Scholar]
- 5.Folco EJ, Rocha VZ, Lopez-Ilasaca M, et al. Adiponectin inhibits pro-inflammatory signaling in human macrophages independent of interleukin-10. J Biol Chem. 2009;284(38):25569–25575. doi: 10.1074/jbc.M109.019786. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Takemura Y, Ouchi N, Shibata R, et al. Adiponectin modulates inflammatory reactions via calreticulin receptor-dependent clearance of early apoptotic bodies. J Clin Invest. 2007;117(2):375–86. doi: 10.1172/JCI29709. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Gustafson B, Hammarstedt A, Andersson CX, et al. Inflamed adipose tissue: a culprit underlying the metabolic syndrome and atherosclerosis. Arterioscler Thromb Vasc Biol. 2007;27(11):2276–2283. doi: 10.1161/ATVBAHA.107.147835. [DOI] [PubMed] [Google Scholar]
- 8.Bruun JM, Lihn AS, Verdich C, et al. Regulation of adiponectin by adipose tissue-derived cytokines: in vivo and in vitro investigations in humans. Am J Physiol Endocrinol Metab. 2003;285(3):E527–E533. doi: 10.1152/ajpendo.00110.2003. [DOI] [PubMed] [Google Scholar]
- 9.Kizer JR, Arnold AM, Strotmeyer ES, et al. Change in circulating adiponectin in advanced old age: determinants and impact on functional outcome and mortality. The Cardiovascular Health Study All Stars Study. J Gerontol A Biol Sci Med Sci. 2010;65A(11):1208–1214. doi: 10.1093/gerona/glq122. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Strotmeyer ES, Jenny NS, French B, et al. Does IL-6 track with functional decline? [abstract] Gerontologist. 2009;49(suppl 2):454. [Google Scholar]
- 11.Fried LP, Borhani NO, Enright P, et al. The Cardiovascular Health Study: design and rationale. Ann Epidemiol. 1991;1:263–276. doi: 10.1016/1047-2797(91)90005-w. [DOI] [PubMed] [Google Scholar]
- 12.Newman AB, Arnold AM, Sachs MC, et al. Long-term function in an older cohort—the Cardiovascular Health Study All Stars Study. J Am Geriatr Soc. 2009;57(3):432–440. doi: 10.1111/j.1532-5415.2008.02152.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Cushman M, Cornell ES, Howard PR, et al. Laboratory methods and quality assurance in the Cardiovascular Health Study. Clin Chem. 1995;41(2):264–270. [PubMed] [Google Scholar]
- 14.Macy EM, Hayes TE, Tracy RP. Variability in the measurement of C-reactive protein in healthy subjects: implications for reference intervals and epidemiological applications. Clin Chem. 1997;43(1):52–58. [PubMed] [Google Scholar]
- 15.Pearson TA, Mensah GA, Alexander RW, et al. Markers of inflammation and cardiovascular disease: application to clinical and public health practice. Circulation. 2003;107(3):499–511. doi: 10.1161/01.cir.0000052939.59093.45. [DOI] [PubMed] [Google Scholar]
- 16.Psaty BM, Kuller LH, Bild D, et al. Methods of assessing prevalent cardiovascular disease in the Cardiovascular Health Study. Ann Epidemiol. 1995;5:270–277. doi: 10.1016/1047-2797(94)00092-8. [DOI] [PubMed] [Google Scholar]
- 17.Ives DG, Fitzpatrick AL, Bild DE, et al. Surveillance and ascertainment of cardiovascular events. The Cardiovascular Health Study. Ann Epidemiol. 1995;5:278–285. doi: 10.1016/1047-2797(94)00093-9. [DOI] [PubMed] [Google Scholar]
- 18.Harris TB, Ferrucci L, Tracy RP, et al. Associations of elevated interleukin-6 and C-reactive protein levels with mortality in the elderly. Am J Med. 1999;106(5):506–512. doi: 10.1016/s0002-9343(99)00066-2. [DOI] [PubMed] [Google Scholar]
- 19.Stork S, Feelders RA, van den Beld AW, et al. Prediction of mortality risk in the elderly. Am J Med. 2006;119(6):519–525. doi: 10.1016/j.amjmed.2005.10.062. [DOI] [PubMed] [Google Scholar]
- 20.Alley DE, Crimmins E, Bandeen-Roche K, et al. Three-year change in inflammatory markers in elderly people and mortality: the Invecchiare in Chianti study. J Am Geriatr Soc. 2007;55(11):1801–1807. doi: 10.1111/j.1532-5415.2007.01390.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kizer JR, Barzilay JI, Kuller LH, et al. Adiponectin and risk of coronary heart disease in older men and women. J Clin Endocrinol Metab. 2008;93(9):3357–3364. doi: 10.1210/jc.2008-0640. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Atzmon G, Pollin TI, Crandall J, et al. Adiponectin levels and genotype: a potential regulator of life span in humans. J Gerontol A Biol Sci Med Sci. 2008;63(5):447–453. doi: 10.1093/gerona/63.5.447. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Wannamethee SG, Whincup PH, Lennon L, et al. Circulating adiponectin levels and mortality in elderly men with and without cardiovascular disease and heart failure. Arch Intern Med. 2007;167(14):1510–1517. doi: 10.1001/archinte.167.14.1510. [DOI] [PubMed] [Google Scholar]
- 24.Laughlin GA, Barrett-Connor E, May S, et al. Association of adiponectin with coronary heart disease and mortality: the Rancho Bernardo study. Am J Epidemiol. 2007;165(2):164–174. doi: 10.1093/aje/kwk001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Poehls J, Wassel CL, Harris TB, et al. Association of adiponectin with mortality in older adults: the Health, Aging, and Body Composition Study. Diabetologia. 2009;52(4):591–595. doi: 10.1007/s00125-009-1261-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.