Skip to main content
Clinical and Experimental Immunology logoLink to Clinical and Experimental Immunology
. 2003 Apr;132(1):24–31. doi: 10.1046/j.1365-2249.2003.02137.x

Predicting death from tumour necrosis factor-alpha and interleukin-6 in 80-year-old people

H BRUUNSGAARD *,, S LADELUND , A N PEDERSEN †,, M SCHROLL §, T JØRGENSEN , B K PEDERSEN
PMCID: PMC1808682  PMID: 12653832

Abstract

Ageing is associated with low-grade inflammation and markers such as IL-6 possess prognostic value. Tumour necrosis-alpha (TNF-α) initiates the inflammatory cascade and has been linked to several age-associated disorders. It remains, however, unknown if TNF-α is associated with mortality in old populations. The aim of the present study was to investigate if serum levels of TNF-α were associated with all-cause mortality independently of interleukin (IL)-6 in a prospective study of 333 relatively healthy 80-year-old people. A Cox regression model was used to explore effects of TNF-α and IL-6 on survival in the following 6 years. A total of 133 participants died during this follow-up period. TNF-α was associated with mortality in men, but not in women, whereas low-grade elevations in IL-6 were associated strongly with mortality in both sexes. TNF-α explained only 7% of the variability in IL-6 and effects of the two cytokines were independent of each other as well as of other traditional risk factors for death [smoking, blood pressure, physical exercise, total cholesterol, co-morbidity, body mass index (BMI) and intake of anti-inflammatory drugs]. These findings indicate that at least in old populations chronic elevated levels of TNF-α and IL-6 have different biological functions that trigger age-associated pathology and cause mortality.

Keywords: ageing, IL-6, inflammation, mortality, TNF

Introduction

Ageing is associated with increased circulating levels of tumour necrosis factor-alpha (TNF-α) interleukin (IL)-6, interleukin (IL)-1 receptor antagonist (IL-1Ra), soluble TNF receptors, acute phase proteins and high counts of neutrophils [1]. Plasma levels of TNF-α were correlated linearly with IL-6 and C reactive protein (CRP) in centenarians, indicating an interrelated activation of the entire inflammatory cascade in the oldest old [2]. It has been suggested that increased inflammatory activity in elderly populations (inflamm-ageing) results from a global reduction in the capacity to cope with a variety of stressors [3]. Although age-related increases in circulating cytokines and acute phase proteins are not very marked compared to levels in young individuals, cross-sectional studies have linked low-grade elevations in TNF-α and IL-6 to morbidity in elderly populations, e.g. circulating levels of both cytokines have been related to cardiovascular diseases [48] and frailty [9,10], TNF-α has been linked to Alzheimer's disease [2] and type 2 diabetes [11], whereas IL-6 has been demonstrated to be a strong predictor of mortality [6,12]. Consistent with this, TNF-α and IL-6 are multi-functional cytokines with important regulatory roles in immune processes, the metabolism of fat, proteins, carbohydrates and bones and the induction of procoagulant changes [1316]. TNF-α is a classical proinflammatory cytokine but it has also been suggested to have neuroprotective/anti-inflammatory effects during chronic neuropathological processes [17]. TNF-α is an important stimulator of IL-6 [18]. IL-6 is often categorized as a proinflammatory cytokine, but in fact it also has very important anti-inflammatory functions [19], e.g. IL-6 inhibits the synthesis of TNF-α [20], stimulates the production of IL-1Ra and induces the shedding of TNF receptors, which secondly bind and suppress the function of circulating TNF-α [21]. Furthermore, IL-6 induces the synthesis of acute phase proteins in the liver and many of these have anti-inflammatory effects including CRP [22].

To our knowledge it has not been examined previously if low-grade elevation in TNF-α is a prognostic marker of all-cause mortality in elderly populations. The cytokine cascade is initiated by TNF-α, suggesting that TNF-α may be a key link between inflammation and morbidity in old populations. Most age-related studies have focused so far on IL-6, which has been called a cytokine for gerontologists [23]. However, measurements of IL-6 may reflect indirectly activities of TNF-α, as the production of the two cytokines is tightly linked. Thus, it is possible that IL-6 simply represents a bystander phenomenon to increased production of TNF-α in elderly people.

The purpose of the present prospective study was to investigate if the circulating level of TNF-α was a predictor of all-cause mortality independently of IL-6 and other known risk factors in a cohort of 333 relatively healthy 80-year-old people.

Materials and Methods

The 1914 population, Center for Preventive Medicine

The study of the 1914 population started in 1964 as a survey of risk factors of coronary heart disease in all 50-year-old men and women born in 1914 and living in seven municipalities around Glostrup University Hospital [24]. The background population represent urban and suburban Danes [25]. Follow-up studies of the 1914 population have been performed with 5–10-year intervals since the initial survey and the cohort was supplemented with new people at the survey of 70-year-olds and the 75-year-olds. Subjects are followed through the Civil Registration System that registers all individuals with a residence in Denmark by a unique 10-digit number. The present study is based on 362 peopole from the 1914 population in Glostrup who participated in a home visit and in a health examination at Glostrup Hospital in 1995, when they were 80 years old. They constituted 55% of the invited population. Subject characteristics, including the prevalence of co-morbidity, are shown in Table 1. The participants in the present study were characterized by less limitation in the activity of daily living (ADL) or instrumental ADL than subjects who received a home visit but did not go to the hospital for a physical examination (Table 1). Most participants lived in their own homes and no one was demented. Twenty-nine subjects were omitted because they had experienced acute illness that could, potentially, cause temporary increases in TNF-α and IL-6: 21 individuals reported acute illness during the last 2 weeks; six individuals had an intake of antibiotics; seven individuals had polymyalgia rheumatica/giant cell arteritis according to the Danish National Register of Patients (see Disease categories). Six subjects fulfilled more than one of these criteria.

Table 1.

Survey of the 1914 population of 80-year-olds

Participants Full examination programme, included in survival analysis n = 333 Participants Full examination programme, excluded in survival analysis n = 29 Non-participants, home visit n = 142 Non-participants, no contact n = 148 Total
Men 47% 48% 37% 39% 43%
Nursery homes/sheltered houses 7% 3% 15% 9%
ADL limitation1 14% 17% 58% 28%
IADL limitation2 22% 24% 58% 32%
BMI 26 (23–28) 24 (23–25) 25 (22–27)
Blood pressure
 Systolic 150 mmHg (136–164) 140 mmHg (130–157) 150 mmHg (136–164)
 Diastolic 82 mmHg (76–92) 80 mmHg (74–93) 82 mmHg (76–92)
Total cholesterol 6·1 mmol/l (5·3–6·8) 5·8 mmol/l (5·0–6·6) 6·1 mmol/l (5·3–6·7)
IL-6 3·3 pg/ml (2·3–5·5) 2·8 pg/ml (2·0–4·1) 3·3 pg/ml (2·3–5·4)
TNF-Α 4·1 pg/ml (3·2–5·1) 3·4 pg/ml (3·0–4·4) 4·1 pg/ml (3·2–5·0)
ABI 0·93 (0·75–1·03) 0·94 (0·82–1·02) 0·93 (0·76–1·03)
Chronic diseases
 CVD 19%
 Cancer 8%
 DM 4%
 CPD 3%
Current smokers 27%

Medians and interquartile range (25th−75th percentile) are shown for continuous variables. Percentages are shown for categorical data.

1

The ADLs (activity of daily living) included eating, getting in and out of bed, getting around inside, dressing, bathing, and using the toilet. A person had an ADL limitation if any ADL could not be performed, or was only performed with help.

2

A person had an IADL (instrumental activity of daily living) limitation if he or she could not get around outside or shop for groceries, or only got around outside or shopped for groceries with help. BMI, body mass index. ABI, ankle–arm arterial blood pressure index. CVD, cardiovascular diseases. DM, diabetes mellitus, CPD, chronic obstructive pulmonary diseases.

The local ethical commitee approved the study and the written informed consent of all participants were obtained.

TNF-α and IL-6 in serum

Circulating levels of TNF-α and IL-6 were measured in serum obtained in 1995 and stored at − 20°C until analysed by commercially available enzyme-linked immunosorbent assay (ELISA) kits (catalogue numbers: HSTA50 and HS600, R&D systems, Minneapolis, MN, USA). The used immunoassays measure the total amount of free TNF-α or IL-6 plus the amount bound to soluble receptors. All samples and standards were run as duplicates and the mean of duplicates was used in the statistical analyses. Detection limits were 0·1–0·2 pg/ml. In our hands, the coefficient of intra-assay variation is 15·7% for TNF-α (n = 12) and 8·7% for IL-6 (n = 11). With regard to the interassay/interday precision, the CV is 25·0% for TNF-α (n = 7) and 15·1% for IL-6 (n = 6) in our laboratory. All analyses were run within 8 weeks and kits with the same batch number were used in order to limit interassay variation.

Disease categories

Since 1977 all hospitals in Denmark has been obliged by law to report all discharge diagnoses in accordance with the International Classification of Diseases (ICD) to the Danish National Register of Patients. Diagnoses in this register in 1995 were used to isolate 80-year-old subjects with chronic diseases that would potentially influence circulating levels of TNF-α and IL-6. Ischaemic heart diseases were categorized as having a history of myocardial infarction, angina pectoris or unstable angina (I20–25). Cardiovascular diseases (CVD) were defined as ischaemic heart diseases, heart failure (I11, I13, I50), cerebrovascular diseases including stroke and transient cerebral ischemia (I60–69, G45) and peripheral arterial diseases (I70, I74). Diabetes mellitus (DM) included a non-fasting blood glucose more than 11·2 mmol/l at the examination in Glostrup Hospital or diagnosed type 1 (E10) or type type 2 DM (E11). Cancer was defined as having a history of a malignant neoplasm (C00–C43, C45–C97). Chronic pulmonary diseases (CPD) included chronic bronchitis, asthma or emhysema (J40–J47). Cancer, DM and CPD were pooled in the category called other chronic diseases in survival analyses due to the low number of people who presented these disorders.

Other measures

Blood pressure was measured in a sitting position. Body mass index (BMI) was calculated as weight divided by height squared. Information about physical activity and smoking was based on a self-administered questionnaire that was given to the participants during the home visit and checked at the laboratory examination. In the statistical analyses current smokers included people who reported smoking every day, as well as people who reported smoking now and then. Similarly, the category ‘previous smokers’ included people who reported that they used to smoke every day and people who answered that they used to smoke now and then. Physical activity was described by six different levels in the questionnaire but was reduced into three groups in the analyses: (1) ‘mainly sitting’; (2) ‘light physical activity’ or ‘moderate physical activity about 3 h per week’; (3) ‘physical activity more than 4 h per week’ or ‘sports’ or ‘competitive sports’. Non-fasting cholesterol and non-fasting blood glucose were measured by standard laboratory procedures in Glostrup Hospital.

Follow-up

Vital status and time of death were obtained from the Civil Registration System in October 2001, 6 years after the survey of 80-year-olds.

Statistics

systat statistical software 8·0 (systat, Evanston, IL, USA) was used for initial statistical analyses and the open source statistical system R [26] was used for Cox regression analyses. Independent groups were compared by χ2 statistics (categorical variables) or by two-tailed analyses of variance (continuous variables). Linear relations between continuous variables were evaluated by Pearson's correlation analyses. TNF-α and IL-6 showed normal distributions after loge transformations in these analyses. A Cox regression model was used to explore the effects of TNF-α, IL-6 and other confounders on survival. All models fulfilled the assumption of proportional hazards evaluated by Schoenfeld residuals. Backward stepwise selection was used to remove unimportant variables. The removal of covariates was controlled afterwards by a forward-stepping procedure in the final models. Interactions with a potential physiological interest were also checked, e.g. TNF-α × IL-6, gender × cytokines, gender × BMI, gender × CVD, gender × other chronic diseases, cytokines × CVD and cytokines × other chronic diseases. P-values < 0·05 were considered significant in all analyses.

Results

Linear regression of IL-6 on TNF-α

A linear regression analysis was performed in order to evaluate how much of the variability in serum levels of IL-6 was explained by circulating TNF-α. TNF-α and IL-6 were correlated weakly and TNF-α explained only 7% of the variability in IL-6 (R = 0·28, n = 333, P < 0·0005).

Cytokines, co-morbidity and risk factors

Gender did not influence levels of the two cytokines (Table 2). Furthermore, co-morbidity including a history of CVD, a malignant neoplasm, DM or CPD were not associated with increased circulating levels of TNF-α or IL-6 and an intake of anti-inflammatory drugs did not result in decreased levels of the two cytokines (Table 2). In contrast, smokers had elevated IL-6 compared to non-smokers and physical inactivity was associated with increased levels of both cytokines. Furthermore, levels of IL-6 were correlated with diastolic blood pressure and total cholesterol. In these analyses men and women were pooled after checking that there were no interactions between gender and each of these parameters in a general linear model with TNF-α and IL-6 as dependent variables. In contrast, there was an interaction between gender and BMI. Thus, TNF-α was correlated with BMI in women, but not in men.

Table 2.

TNF-α, IL-6 and characteristics of the 1914 population

TNF-α pg/ml IL-6 pg/ml
Gender
 Men n= 158 4·1 (3·1–5·0) 3·5 (2·4–5·5)
 Women n = 175 4·2 (3·2–5·2) 3·2 (2·2–5·4)
P = 0·3 P = 0·5
Smoking status
 Never n = 109 4·2 (3·2–5·0) 3·0 (2·1–4·6)
 Previous n = 133 4·1 (3·0–5·1) 3·3 (2·2–5·5)
 Current n = 91 4·1 (3·2–5·1) 4·3 (2·6–5·9)
P = 0·6 P = 0·008
Physical activity
 Mainly sitting n = 25 5·0 (3·9–5·6) 6·2 (4·3–10·0)
 Light/moderate n = 248 4·2 (3·2–5·2) 3·3 (2·3–5·5)
 > Moderate n = 60 3·5 (2·9–4·4) 2·8 (2·0–4·4)
P = 0·01 P = 0·001
Total cholesterol n= 333 R = − 0·07; P = 0·2 R = − 0·26; P < 0·0005
 Blood pressure
 Systolic n = 333 R = − 0·039; P = 0·5 R = − 0·10; P = 0·06
 Diastolic n = 333 R = − 0·058; P = 0·3 R = − 0·12; P = 0·03
BMI
 Men n = 153 R = 0·12; P = 0·1 R = − 0·008; P = 0·9
 Women n = 161 R = 0·24; P = 0·002 R = 0·15; P = 0·07
CVD
 Yes n = 63 4·2 (3·5–5·1) 4·4 (2·6–5·9)
 No n = 270 4·1 (3·2–5·1) 3·1 (2·2–5·3)
P = 0·6 P = 0·4
Other chronic diseases
 Yes (pooled) n = 47 4·0 (3·2–5·6) 3·5 (2·3–5·1)
 No n = 286 4·1 (3·2–5·0) 3·3 (2·3–5·5)
P = 0·8 P = 0·4
Cancer
 Yes n = 25 4·5 (3·4–5·6) 3·3 (2·1–4·3)
 No n = 308 4·1 (3·2–5·0) 3·3 (2·3–5·5)
P = 0·4 P = 0·2
DM
 Yes n = 12 3·7 (3·2–5·3) 4·0 (2·7–5·5)
 No n = 321 4·1 (3·2–5·1) 3·3 (2·3–5·5)
P = 0·9 P = 1·0
CPD
 Yes n = 11 3·6 (2·8–5·0) 3·5 (2·3–4·6)
 No n = 322 4·1 (3·2–5·1) 3·3 (2·3–5·5)
P = 0·6 P = 0·09
Anti-inflammatory drugs
 Yes n = 133 4·1 (3·2–5·1) 3·6 (2·5–5·2)
 No n = 200 4·2 (3·2–5·0) 3·0 (2·2–5·5)
P = 0·7 P = 0·7

Medians and interquartile range (25th−75th percentile) are shown. R, Pearson's correlation coefficient. CVD, cardiovascular diseases. Other chronic diseases are defined as one of the following diagnoses: cancer, diabetes mellitus (DM) or chronic obstructive pulmonary diseases (CPD). Anti-inflammatory drugs: daily intake of non-steroid anti-inflammatory drugs, acetyl salicylic acid or corticosteroids.

Cytokines and mortality risk

After 6 years, 133 participants had died. Non-survivors were characterized by elevated serum levels of IL-6, lower total cholesterol, physical inactivity, a higher prevalence of current smokers or a history of smoking and a higher prevalence of CVD and cancer compared to survivors (Table 3). No difference was detected in circulating levels of TNF-α. A Cox regression model was used for more detailed analyses of associations between circulating cytokines and mortality. Initial fitting of the model revealed that all the included continuous variables could appropriately be scored linearly except IL-6. With regard to IL-6, there was a linear association between circulating levels in the range 0–20 pg/ml and the hazard function, whereas there was no association between the mortality risk and IL-6 levels in 15 individuals who had levels beyond 20 pg/ml. Different unsatisfying transformations were tried. Finally, we decided to separate IL-6 into two variables in order to limit the strong influence of the relatively few subjects with high levels: (1) IL-6 (0–20 pg/ml), values = < 20·0 pg/ml scored linearly otherwise 0; (2) IL-6 (>20 pg/ml), values> 20 pg/ml scored linearly otherwise 0. In a model including TNF-α, IL-6 (0–20 pg/ml) and IL-6 (>20 pg/ml) only IL-6 in the low range was associated strongly with mortality (model A in Table 4). Thus, the risk of dying increased with 13% when IL-6 in serum increased 1 pg/ml in the range 0–20 pg/ml adjusted for the effect of TNF-α, whereas TNF-α itself or IL6 > 20 pg/ml did not influence the hazard function. The separation of IL-6 into two variables represents a mathematical solution of a biological dilemma. On one hand, IL-6 levels more than 20 pg/ml may not represent low-grade inflammatory activity. On the other hand, this study is a population-based study and, accordingly, we thought it was disturbing to exclude outliers without a very good biological argument. There was no difference in the prevalence of chronic diseases or other risk factors in the group with IL-6 > 20 pg/ml compared with the remaining cohort. Moreover, high levels were repeatedly found after a re-analysis of the ELISA with new dilutions. These people probably represent a very heterogeneous population including acute unreported harmless illness, serious unrecognized chronic diseases, a different genotype, etc. but there were no facts that justified exclusions.

Table 3.

Survivors and non-survivors in the 1914 population

Survivors n = 200 Non-survivors n = 133 Statistics
Gender
 Men 41·0% 57·1% P = 0·004
 Women 59·0% 42·9%
IL-6 pg/ml 2·8 (2·0–4·6) 4·4 (2·9–6·6) P < 0·0005
TNF-α pg/ml 4·0 (3·1–5·0) 4·2 (3·2–5·4) P = 0·1
Total cholesterol mmol/l 6·2 (5·5–6·8) 5·9 (5·0–6·6) P = 0·05
BMI 26 (23–28) 25 (23–28) P = 0·3
Blood pressure
 Systolic 150 (138–164) 150 (132–164) P = 0·4
 Diastolic 82 (76–92) 84 (76–90) P = 0·7
Physical activity
 Mainly sitting 4·0% 12·8% P < 0·0005
 Light/moderate 73·0% 76·7%
 > Moderate 23·0% 10·5%
Smoking status
 Never 39·5% 22·6% P = 0·005
 Previous 36·5% 45·1%
 Current 24·0% 32·3%
Cardiovascular diseases 13·5% 27·1% P = 0·002
Cancer 5·0% 11·3% P = 0·03
Diabetes mellitus 2·5% 5·3% P = 0·2
Chronic pulmonary disease 4·0% 2·3% P = 0·4
Anti-inflammatory drugs 36·5% 45·1% P = 0·1

Medians and interquartile range (25th−75th percentile) are shown.

Table 4.

Cox regression models in the 1914 population

Hazard ratio SE P
Model A
 IL-6 (0–20 pg/ml) 1·13 0·0196 <0·0005
 TNF-α 1·02 0·0316 0·5
Model B
 IL-6 (0–20 pg/ml) 1·09 0·0237 0·0007
 TNF-α 1·13 0·0450
 TNF-α/men 1·15 0·0817 0·03
Model C
 IL-6 (0–20 pg/ml) 1·10 0·0242 0·0005
 TNF-α 1·13 0·0437
 TNF-α/men 1·16 0·0845 0·03

Model A: unadjusted model. IL-6 > 20 pg/ml had no significant influence on the hazard function and were accordingly removed from the model. Model B: TNF/men denotes an interaction between TNF-α and gender with women as the reference. Adjusted for the effect of gender, physical exercise, BMI and BMI × gender (interaction). Smoking history and blood pressure had no significant influence on the hazard function and were accordingly removed from the final model. Model C: adjusted for the same parameters as model B as well as for the effect of a history of cardiovascular diseases and a history of other chronic diseases. Cholesterol and intake of anti-inflammatory drugs did not affect the hazard function and were accordingly removed from the final model.

The association between mortality and IL-6 in the range 0–20 pg/ml was still significant when we also adjusted for the effect of traditional risk factors including gender, BMI, BMI × gender, physical activity, systolic blood pressure, smoking history and current smoking status, although the hazard ratio decreased slightly (model B in Table 4). In this model we also found a significant interaction between TNF-α and gender. The addition of this interaction resulted in an almost ninefold increase in the hazard ratio of TNF-α and this effect was explained abundantly by an excess risk in men. Thus, TNF-α was associated with mortality in men but not in women. Among interactions with a potential physiological interest (defined in Materials and methods) only TNF × gender and BMI × gender had significant effects and for that reason only these variables are included in the presented models. Furthermore, the blood pressure and the smoking status had no independent effect on the hazard function and were therefore omitted from the model. Finally, we adjusted for the effect of a history of CVD, other chronic diseases, intakes of anti-inflammatory drugs and total cholesterol. A history of CVD and other chronic diseases influenced the hazard function (data not shown) but effects of TNF-α and IL-6 were remarkably unaffected (model C in Table 4). Intakes of anti-inflammatory drugs and total cholesterol were not associated with increased mortality risk in this model and were accordingly removed.

Discussion

The novel finding in the present study was that enhanced serum levels of TNF-α in men and low-grade elevations of IL-6 in both sexes were independent of each other associated with all-cause mortality in relatively healthy 80-year-old people after the adjustment for known risk factors and co-morbidity. We found no relation between circulating cytokines and the prevalence of chronic diseases. In contrast, high levels of both cytokines were associated with classical risk factors such as smoking, physical inactivity and BMI. These findings indicate that low-grade elevations in circulating IL-6 and TNF-α cause age-associated pathology and mortality.

In this cohort, we did not confirm the hypothesis that systemic TNF-α was a better predictor of the mortality risk than IL-6. TNF-α in serum explained only a small part of the variation (7%) in IL-6 in the present cohort, indicating that the major part of the systemic anti-inflammatory activity seemed to be triggered independently of systemic proinflammatory activity. It is probable that IL-6 is a better marker of the sum of ongoing inflammation than systemic levels of TNF-α because a local production of TNF-α may not escape to the circulation, although it induces a strong systemic IL-6 response. The fact that TNF-α and IL-6 were associated with the mortality risk independently of each other suggests different biological effects of the two cytokines. Both cytokines affect several risk factors in cardiovascular disorders, which is the largest disease category in octogenarians. Endothelial dysfunction is considered to be one of the first steps in atherosclerosis [27] and it has been demonstrated that TNF-α, but not IL-6, impairs the endothelium-dependent relaxation in humans [28] and TNF-α, but not IL-6, causes directly endothelial up-regulation of cellular adhesion molecules [29]. Furthermore, TNF-α causes insulin resistance [30]. IL-6 predicted the development of type 2 DM in large population-based studies [31,32], but IL-6 may have acted as a surrogate marker for TNF-α in these studies. Thus, mice bearing IL-6-secreting tumours have hypoglycaemia [33] whereas IL-6 knock-out mice develop late-onset obesity, insulin resistance and hyperglycaemia [34]. However, IL-6 induces procoagulant changes by increasing the production of fibrinogen, tissue factor, factor VIII, von Willebrand factor and platelets [35]. Moreover, TNF-α and IL-6 both cause dyslipidaemia [14,33]. It has also been suggested that frailty is a result of a metabolic imbalance caused by overproduction of catabolic cytokines such as TNF-α and by diminished availability or action of anabolic hormones, resulting form ageing itself and the presence of associated chronic conditions [9]. Basal studies have indeed demonstrated that TNF-α causes anorexia, muscle wasting [36] and bone loss [37], but although IL-6 is a strong prognostic marker of functional disability [38], the direct role of IL-6 in these processes is unresolved. IL-6 causes also anorexia but mice bearing IL-6-secreting tumours had an unaltered percentage of lean body mass [33] and osteoporosis was not a feature of transgenic mice over-expressing IL-6 [37]. In this light, it is probable that increased circulating levels of TNF-α promote atherogenesis, type 2 DM and frailty, whereas circulating IL-6 is a risk marker of thromboembolic manifestations and has a more uncertain role in the syndrome of frailty. Consistent with the finding that IL-6 is an independent risk factor in the present study, the IL6–174G > C polymorphism is an important regulator of IL-6 production during ageing [39] and provides a gender-dependent frailty gene beyond the age of 100 years [40].

It is unclear why TNF-α was associated only with mortality in men, although no difference in circulating levels were detected between the two sexes. It is possible that this effect was not detected in women because a lower proportion died during the follow-up period. On the other hand, men may be more susceptible to the biological effects of TNF-α for unknown reasons or the underlying explanation may be a mixture. Indeed, young men had larger TNF-α production following lipopolysaccharide (LPS) stimulation in vitro compared to young women, but this difference was blurred with ageing, suggesting a selection of individuals with low TNF-α production in longevity and/or an effect of sex hormones [41]. To our knowledge, little is known about biological factors underlying the gender-associated difference in longevity.

In accordance with the present data, IL-6 was associated with mortality both in those with and without CVD in a study of healthy older people with an equivalent prevalence of morbidity as in the 1914 population, and similar results were found for cardiovascular and non-cardiovascular causes of death [12]. We assessed that the present study was not large enough to test for cause-specific mortality. In contrast to the study by Harris et al. [12], as well as the present data, IL-6 was associated only with mortality in women with a history of CVD among 620 disabled older women [6]. However, the latter report [6] studied frail people with a high prevalence of co-morbidity and this could explain this discrepancy easily, because IL-6 probably acted as a static disease marker to a higher extent in such a population. Other inflammatory markers including high levels of CRP, decreased albumin and low cholesterol have also been related to functional status and mortality risk in old populations [12,4244], but these associations may simply reflect activities of TNF-α and IL-6. For instance, TNF-α and IL-6 together stimulate the liver production of CRP [18], and IL-6 is a stronger predictor of mortality than CRP when both parameters are included simultaneously in a Cox regression model [12]. Chronic elevations in systemic IL-6 also caused low total cholesterol in mice [33] as well as in monkeys [45]. In accordance with this, IL-6 was correlated negatively with total cholesterol in the present study and cholesterol had no effect on the hazard function when IL-6 was included in the survival analysis, although non-survivors had low cholesterol.

Limitations of the present data should be considered. Serum samples were stored at −20°C for a long time period and we cannot exclude the possibility of protein degradation. Such an effect is unlikely, however, because the distributions of the two cytokines are similar to studies that used fresh blood samples [2] or stored serum at −80°C [3,4]. Furthermore, even if such an effect were present it would not affect the validity of the results because blood samples were collected during the same time period and handled identically, and accordingly protein degradation would happen to the same extent in all sera.

The present data suggest that it may be important to discuss intervention strategies in elderly populations in order to decrease systemic levels of TNF-α and IL-6 in the future. For instance, this could include a more aggressive treatment of infections and/or supplementations. However, an isolated inhibition of the IL-6 production is questionable because systemic IL-6 may have an important role in the control and limitation of local proinflammatory processes.

In conclusion, serum levels of TNF-α in men and low-grade elevations of IL-6 in both sexes were independent of each other associated with all-cause mortality in healthy octogenarians, indicating different biological effects of the two cytokines. It is probable that increased circulating levels of TNF-α promote type 2 DM and atherosclerosis, whereas circulating IL-6 is a risk marker of thromboembolic manifestations. Associations between TNF-α, IL-6 and mortality were also independent of traditional risk factors and co-morbidity, demonstrating that low-grade elevations in circulating cytokines are independent risk factors. Accordingly, the present study supports the hypothesis that systemic IL-6 and TNF-α trigger age-associated pathology and mortality.

Acknowledgments

This investigation was supported by a grant from the Danish Research Council. We gratefully acknowledge the technical assistance of Ruth Rousing and Hanne Villumsen.

References

  • 1.Bruunsgaard H, Pedersen M, Pedersen BK. Aging and proinflammatory cytokines. Curr Opin Hematol. 2001;8:131–6. doi: 10.1097/00062752-200105000-00001. [DOI] [PubMed] [Google Scholar]
  • 2.Bruunsgaard H, Andersen-Ranberg K, Jeune B, Pedersen AN, Skinhoj P, Pedersen BK. A high plasma concentration of TNF-alpha is associated with dementia in centenarians. J Gerontol Med Sci. 1999;54A:M357–64. doi: 10.1093/gerona/54.7.m357. [DOI] [PubMed] [Google Scholar]
  • 3.Franceschi C, Bonafé M, Valensin S, et al. Inflamm-aging. An evolutionary perspective of immunosenescence. Ann NY Acad Sci. 2000;908:244–54. doi: 10.1111/j.1749-6632.2000.tb06651.x. [DOI] [PubMed] [Google Scholar]
  • 4.Ridker PM, Rifai N, Stampfer MJ, Hennekens CH. Plasma concentration of interleukin-6 and the risk of future myocardial infarction among apparently healthy men. Circulation. 2000;101:1767–72. doi: 10.1161/01.cir.101.15.1767. [DOI] [PubMed] [Google Scholar]
  • 5.Ridker PM, Rifai N, Pfeffer M, Sacks F, Lepage S, Braunwald E. Elevation of tumor necrosis factor-alpha and increased risk of recurrent coronary events after myocardial infarction. Circulation. 2000;101:2149–53. doi: 10.1161/01.cir.101.18.2149. [DOI] [PubMed] [Google Scholar]
  • 6.Volpato S, Guralnik JM, Ferrucci L, et al. Cardiovascular disease, interleukin-6, and risk of mortality in older women: the women's health and aging study. Circulation. 2001;103:947–53. doi: 10.1161/01.cir.103.7.947. [DOI] [PubMed] [Google Scholar]
  • 7.Bruunsgaard H, Skinhøj P, Pedersen AN, Schroll M, Pedersen BK. Ageing, TNF-alpha and atherosclerosis. Clin Exp Immunol. 2000;121:255–60. doi: 10.1046/j.1365-2249.2000.01281.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Skoog T, Dichtl W, Boquist S, et al. Plasma tumour necrosis factor-alpha and early carotid atherosclerosis in healthy middle-aged men. Eur Heart J. 2002;23:376–83. doi: 10.1053/euhj.2001.2805. [DOI] [PubMed] [Google Scholar]
  • 9.Hamerman D. Toward an understanding of frailty. Ann Intern Med. 1999;130:945–50. doi: 10.7326/0003-4819-130-11-199906010-00022. [DOI] [PubMed] [Google Scholar]
  • 10.Ferrucci L, Harris TB, Guralnik JM, et al. Serum IL-6 level and the development of disability in older persons. J Am Geriatr Soc. 1999;47:639–46. doi: 10.1111/j.1532-5415.1999.tb01583.x. [DOI] [PubMed] [Google Scholar]
  • 11.Paolisso G, Rizzo MR, Mazziotti G, et al. Advancing age and insulin resistance: role of plasma tumor necrosis factor-alpha. Am J Physiol Endocrin Metab. 1998;38:E294–E299. doi: 10.1152/ajpendo.1998.275.2.E294. [DOI] [PubMed] [Google Scholar]
  • 12.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;106:506–12. doi: 10.1016/s0002-9343(99)00066-2. [DOI] [PubMed] [Google Scholar]
  • 13.Jaattela M. Biologic activities and mechanisms of action of tumor necrosis factor-alpha/cachectin. Lab Invest. 1991;64:724–42. [PubMed] [Google Scholar]
  • 14.Grunfeld C, Palladino M. Tumor necrosis factor: immunologic, antitumor, metabolic, and cardiovascular activities. Adv Intern Med. 1990;35:45–71. [PubMed] [Google Scholar]
  • 15.Woods A, Brull DJ, Humphries SE, Montgomery HE. Genetics of inflammation and risk of coronary artery disease: the central role of interleukin-6. Eur Heart J. 2000;21:1574–83. doi: 10.1053/euhj.1999.2207. [DOI] [PubMed] [Google Scholar]
  • 16.Yudkin JS, Kumari M, Humphries SE, Mohamed-Ali V. Inflammation, obesity, stress and coronary heart disease: is interleukin-6 the link? Atherosclerosis. 2000;148:209–14. doi: 10.1016/s0021-9150(99)00463-3. [DOI] [PubMed] [Google Scholar]
  • 17.Liu J, Marino MW, Wong G, et al. TNF is a potent anti-inflammatory cytokine in autoimmune-mediated demyelination. Nat Med. 1998;4:78–83. doi: 10.1038/nm0198-078. [DOI] [PubMed] [Google Scholar]
  • 18.Richards C, Gauldie J. Role of cytokines in the acute-phase response. In: Aggarwal BB, Puri RK, editors. Human cytokines: their roles in disease and therapy. Cambridge, MA: Blackwell Science; 1995. [Google Scholar]
  • 19.Tilg H, Dinarello CA, Mier JW. IL-6 and APPs: anti-inflammatory and immunosuppressive mediators. Immunol Today. 1997;18:428–32. doi: 10.1016/s0167-5699(97)01103-1. [DOI] [PubMed] [Google Scholar]
  • 20.Xing Z, Gauldie J, Cox G, et al. IL-6 is an antiinflammatory cytokine required for controlling local or systemic acute inflammatory responses. J Clin Invest. 1998;101:311–20. doi: 10.1172/JCI1368. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Tilg H, Trehu E, Atkins MB, Dinarello CA, Mier JW. Interleukin-6 (IL-6) as an anti-inflammatory cytokine: induction of circulating IL-1 receptor antagonist and soluble tumor necrosis factor receptor p55. Blood. 1994;83:113–8. [PubMed] [Google Scholar]
  • 22.Suffredini AF, Fantuzzi G, Badolato R, Oppenheim JJ, O'Grady NP. New insights into the biology of the acute phase response. J Clin Immunol. 1999;19:203–14. doi: 10.1023/a:1020563913045. [DOI] [PubMed] [Google Scholar]
  • 23.Ershler WB. Interleukin-6: a cytokine for gerontologists. J Am Geriatr Soc. 1993;41:176–81. doi: 10.1111/j.1532-5415.1993.tb02054.x. [DOI] [PubMed] [Google Scholar]
  • 24.Schroll M. A ten-year prospective study, 1964–74, of cardiovascular risk factors in men and women from the Glostrup population born in 1914. Dan Med Bull. 1982;29:213–52. [PubMed] [Google Scholar]
  • 25.Schroll M, Jorgensen T, Ingerslev J. The Glostrup Population Studies, 1964–92. Dan Med Bull. 1992;39:204–7. [PubMed] [Google Scholar]
  • 26.Ihaka R, Gentleman R. A language for data analysis and graphics. J Computational Graph Statistics. 1996;5:299–314. [Google Scholar]
  • 27.Ross R. Atherosclerosis − an inflammatory disease. N Engl J Med. 1999;340:115–26. doi: 10.1056/NEJM199901143400207. [DOI] [PubMed] [Google Scholar]
  • 28.Bhagat K, Vallance P. Inflammatory cytokines impair endothelium-dependent dilatation in human veins in vivo. Circulation. 1997;96:3042–7. doi: 10.1161/01.cir.96.9.3042. [DOI] [PubMed] [Google Scholar]
  • 29.Meager A. Cytokine regulation of cellular adhesion molecule expression in inflammation. Cytokine Growth Factor Rev. 1999;10:27–39. doi: 10.1016/s1359-6101(98)00024-0. [DOI] [PubMed] [Google Scholar]
  • 30.Hotamisligil GS. The role of TNF-alpha and TNF receptors in obesity and insulin resistance. J Intern Med. 1999;245:621–5. doi: 10.1046/j.1365-2796.1999.00490.x. [DOI] [PubMed] [Google Scholar]
  • 31.Pradhan AD, Manson JE, Rifai N, Buring JE, Ridker PM. C-reactive protein, interleukin 6, and risk of developing type 2 diabetes mellitus. JAMA. 2001;286:327–34. doi: 10.1001/jama.286.3.327. [DOI] [PubMed] [Google Scholar]
  • 32.Schmidt MI, Duncan BB, Sharrett AR, et al. Markers of inflammation and prediction of diabetes mellitus in adults (Atherosclerosis Risk in Communities study): a cohort study. Lancet. 1999;353:1649–52. doi: 10.1016/s0140-6736(99)01046-6. [DOI] [PubMed] [Google Scholar]
  • 33.Metzger S, Hassin T, Barash V, Pappo O, Chajek-Shaul T. Reduced body fat and increased hepatic lipid synthesis in mice bearing interleukin-6-secreting tumor. Am J Physiol Endocrinol Metab. 2001;281:E957–E965. doi: 10.1152/ajpendo.2001.281.5.E957. [DOI] [PubMed] [Google Scholar]
  • 34.Wallenius V, Wallenius K, Ahren B, et al. Interleukin-6-deficient mice develop mature-onset obesity. Nat Med. 2002;8:75–9. doi: 10.1038/nm0102-75. [DOI] [PubMed] [Google Scholar]
  • 35.Kerr R, Stirling D, Ludlam CA. Interleukin 6 and haemostasis. Br J Haematol. 2001;115:3–12. doi: 10.1046/j.1365-2141.2001.03061.x. [DOI] [PubMed] [Google Scholar]
  • 36.Reid MB, Li YP. Tumor necrosis factor-alpha and muscle wasting: a cellular perspective. Respir Res. 2001;2:269–72. doi: 10.1186/rr67. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Pacifici R. Aging and cytokine production. Calcif Tissue Int. 1999;65:345–51. doi: 10.1007/s002239900710. [DOI] [PubMed] [Google Scholar]
  • 38.Cohen HJ, Pieper CF, Harris T, Rao KK, Currie MS. The association of plasma IL-6 levels with functional disability in community dwelling elderly. J Gerontol Med Scien. 1997;52:M201–8. doi: 10.1093/gerona/52a.4.m201. [DOI] [PubMed] [Google Scholar]
  • 39.Olivieri F, Bonafe M, Cavallone L, et al. The -174 C/G locus affects in vitro/in vivo IL-6 production during aging. Exp Gerontol. 2002;37:309–14. doi: 10.1016/s0531-5565(01)00197-8. [DOI] [PubMed] [Google Scholar]
  • 40.Bonafe M, Olivieri F, Cavallone L, et al. A gender-dependent genetic predisposition to produce high levels of IL-6 is detrimental for longevity. Eur J Immunol. 2001;31:2357–61. doi: 10.1002/1521-4141(200108)31:8&#60;2357::aid-immu2357&#62;3.0.co;2-x. [DOI] [PubMed] [Google Scholar]
  • 41.Bruunsgaard H, Pedersen AN, Schroll M, Skinhoj P, Pedersen BK. Impaired production of proinflammatory cytokines in response to LPS stimulation in elderly humans. Clin Exp Immunol. 1999;118:235–41. doi: 10.1046/j.1365-2249.1999.01045.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Reuben DB, Cheh AI, Harris TB, et al. Peripheral blood markers of inflammation predict mortality and functional decline in high-functioning community-dwelling older persons. J Am Geriatr Soc. 2002;50:638–44. doi: 10.1046/j.1532-5415.2002.50157.x. [DOI] [PubMed] [Google Scholar]
  • 43.Weijenberg MP, Feskens EJ, Souverijn JH, Kromhout D. Serum albumin, coronary heart disease risk, and mortality in an elderly cohort. Epidemiology. 1997;8:87–92. doi: 10.1097/00001648-199701000-00014. [DOI] [PubMed] [Google Scholar]
  • 44.Rosenthal AJ, McMurtry CT, Sanders KM, Jacobs M, Thompson D, Adler RA. The soluble interleukin-2 receptor predicts mortality in older hospitalized men. J Am Geriatr Soc. 1997;45:1362–4. doi: 10.1111/j.1532-5415.1997.tb02937.x. [DOI] [PubMed] [Google Scholar]
  • 45.Ettinger WHJ, Sun WH, Binkley N, Kouba E, Ershler W. Interleukin-6 causes hypocholesterolemia in middle-aged and old rhesus monkeys. J Gerontol A Biol Sci Med Sci. 1995;50:M137–40. doi: 10.1093/gerona/50a.3.m137. [DOI] [PubMed] [Google Scholar]

Articles from Clinical and Experimental Immunology are provided here courtesy of British Society for Immunology

RESOURCES