Adipose Endocrine Function, Insulin-Like Growth Factor-1 Axis, and Exceptional Survival Beyond 100 Years of Age

Abstract

Background. Observational studies have demonstrated similarities between the underpinning of frailty and biological features of centenarians, suggesting that adaptability to age-related multiple physiological decline may be a core component of successful aging. The aim of this study is to determine whether hormonal pathways potentially involved in energy homeostasis contribute to survival beyond 100 years of age.

Methods. We assessed a total of 252 centenarians (mean [standard deviation (SD)] age, 101.5 (1.8) years, range 100–108 years) using a complete set of biomarkers of adipose endocrine function and the insulin-like growth factor-1 (IGF-1) axis. Conventional risk factors at baseline were also assessed. The participants were followed up for all-cause mortality every 12 months by telephone contact.

Results. During 2253 days of follow-up, 208 centenarians (82.5%) died. The lowest tertile of leptin and the highest tertile of tumor necrosis factor-α were associated with higher mortality risk among centenarians after adjusting for age (per 6-month increase), sex, education, smoking, activities of daily living (ADL), cognitive function, and comorbidities (hazard ratio [HR] 1.6; 95% confidence interval [CI], 1.14–2.35; and HR 1.45; 95% CI, 1.00–2.08, respectively). The lowest tertiles of both IGF-1 and IGF binding protein 3 (IGFBP3) were also associated with increased mortality. The adipose risk score, indicating cumulative effects of adipokine dysregulation, was strongly associated with increased mortality risk; ADL; cognitive function; and levels of albumin, cholinesterase, high-density lipoprotein-cholesterol, C-reactive protein, interleukin 6, and IGF-1 at baseline.

Conclusions. The results suggested that preservation of adipose endocrine function and the IGF-1 axis may be potentially important for maintaining health and function and promoting survival at an extremely old age.

Centenarians, Leptin, Adiponectin, Tumor necrosis factor-alpha, Insulin-like growth factor 1, Mortality, Frailty, Metabolic network

WITH the recent exponential growth of the older population, frailty is increasingly recognized as a major threat to health maintenance and longevity (1–3). Although operationalization of frailty remains a challenge, diminution of functional reserve in multiple physiological systems is the most consistent platform that explains various clinical correlates of this syndrome. It has been proposed that inflammatory activation, hypercoagulability, hormonal decline, and chronic conditions such as atherosclerosis (4–8) are concomitantly involved in the pathophysiology of frailty. Surprisingly, these underpinnings of frailty share many similarities with the biological feature of centenarians, a possible representative of successful aging. Pioneering studies recognized this phenomenon as centenarians' paradox by demonstrating cumulative dysregulation in immune (9), metabolic (10), and coagulation systems (11) in apparently healthy centenarians. Recent studies of medical history and autopsy disclosed a wide spread of multiple pathologies and functional deficits among centenarians even in those enjoying good health just before death (12–17). Furthermore, demographic studies demonstrated that a cumulative measure of frailty discriminated survival chance in older adults, and this accumulation was delayed in the oldest old (18,19). These observations suggest that centenarians may provide a unique opportunity to investigate mechanisms of “successful adaptation,” by which individuals counteract age-related multiple chronic conditions and reduced organ capacity to prevent or postpone the onset of frailty until the very end of the maximum life span.

Aging itself constitutes a substantial part of frailty. At the molecular level, a wide variety of random damage to macromolecules in cells and tissues accumulates during aging—for example, DNA damage caused by oxidative stress, increased somatic and mitochondrial mutations, telomere attrition, and impairment of protein turnover (20). Nearly all the mechanisms for maintenance and repair of these damages require significant amounts of energy. Interestingly, energy deprivation was described as one of the fundamental components of frailty (21,22). In this context, the aging process at the ultimate stage of longevity can be described as diminution of energy reserves up to a certain threshold. Therefore, we hypothesized that centenarians could develop an optimum adaptive strategy designed to maintain energy homeostasis through multiple regulatory pathways and their network.

Whole-body energy homeostasis is largely regulated by the complex interaction of neuroendocrine systems. The recent epidemic of obesity has given prominence to adipose tissue as an active endocrine organ that regulates energy homeostasis by secreting a large number of bioactive substances termed adipokines (23). Because of its adverse metabolic consequences including insulin resistance, the endocrine function of adipose tissue has received maximum attention in the environment of excess energy and adiposity (obesity). However, accumulating evidence suggests that preservation of adipose endocrine function is critically important for maintaining various physiological functions in severe energy-deprived states (24,25). Therefore, it is worth investigating whether adipose endocrine function has a role in the maintenance of health and survival at the ultimate stage of longevity, where energy deprivation may constitute a dominant part of mortality pressure. In addition, the growth hormone (GH)–insulin-like growth factor-1 (IGF-1) axis is of particular interest in terms of the aging process and muscle energy metabolism (26). Age-associated decline in GH and IGF-1 secretion, known as somatopause, is associated with loss of exercise capacity and sarcopenia, which is one of the leading components of frailty (26). Therefore, in this study, we simultaneously examined the biomarkers of adipose endocrine function and the IGF-1 axis in relation to exceptional survival beyond 100 years of age and compared them with established risk factors.

Methods

Participants

The Tokyo Centenarians Study is a prospective cohort study of centenarians living in the Tokyo metropolitan area. The recruitment, design, and procedures of the study have been described in detail elsewhere (14,15). We recruited 513 centenarians from a mailed survey, of whom 304 (65 men, 239 women, mean age 101.1 ± 1.7 years, range, 100–108 years) were enrolled in a home visit, in which our survey team (including a geriatrician, a psychologist, and a practical nurse) undertook a series of assessments at each centenarian's residence between July 2000 and May 2002. The male-to-female ratio of our participants was 1:3.6, which is comparable to the ratio in the total centenarian population in this area (1:3.8). Distributions of sex, self-reported education, and smoking status were not different between a subset of centenarians who responded to the mailed survey only and those who participated in the visiting survey (χ² = 2.51, p =.070; χ² = 0.245, p =.885; χ² = 0.118, p =.410, respectively). However, the age at enrollment was significantly higher in the participants of a home visit (mean age 101.0 ± 1.2 years in participants of the mailed survey only, p <.001 using the Mann–Whitney U test). Written informed consent was obtained from all the participants or by proxy. The present study was approved by the ethical committee of the Keio University School of Medicine.

Data Collection

For the purpose of this study, demographics including education, living arrangements, smoking status, physical and cognitive function, self-reported medical history and medications, body weight and height, supine blood pressure, physical examination, and electrocardiogram were examined at the each participant's residence (14,15). Physical function was assessed using the Barthel index (BI) (27). Cognitive function was evaluated according to the Mini-Mental State Examination (MMSE) (28) as well as the Clinical Dementia Rating (CDR) scale (29). The classification of self-reported medical condition was based on the International Classification of Diseases, 10th Revision (ICD-10) categories (30).

Biomarkers of Hormonal Pathways

Nonfasting blood samples were stored at −80° until subsequent assay. Initially, we identified three hormonal pathways that potentially regulate whole-body energy metabolism—adipose tissue function, IGF-1 axis, and thyroid system—as determinants of exceptional survival. Adiponectin and leptin were selected as biomarkers for adipose tissue function on the basis of previous epidemiologic and experimental evidence (31,32). Tumor necrosis factor-α (TNF-α), also known to be secreted from adipose tissue, was identified as a negative energy regulator through its roles in insulin resistance (33). Interleukin 6 (IL-6) is known to be secreted from adipose tissue; however, IL-6 is categorized as an inflammation marker because of its systemic roles on acute phase reactants and history as a cytokine for “gerontologist” (34). Plasma concentrations of high-molecular-weight (HMW) adiponectin and TNF-α were measured in duplicate using commercially available enzyme-linked immunosorbent assay (ELISA) kits [HMW Adiponectin kit (35); Fuji Rebio, Tokyo, and Quantikine HS, Human TNF-α; R&D Systems, Minneapolis, MN, respectively]. Plasma leptin concentrations were determined using a commercial radio immunoassay (SRL Limited, Tokyo, Japan). Coefficients of variation for assays of HMW adiponectin, TNF-α, and leptin were 0.542, 0.606, and 0.775, respectively. With regard to the IGF-1 axis, plasma concentrations of total IGF-1 were measured by radioimmunoassay, and those of IGF-binding protein 3 (IGFBP3) were determined by an immunoradiometric assay (SRL Limited).

Conventional Risk Factors

We selected three pathways as conventional risk factors associated with high mortality in the oldest old: nutrition, lipids, and inflammation. Serum concentration of albumin, which is the most established indicator for general health and nutrition among the elderly population, was measured using the bromcresol green (BCG) method. Retinal binding protein (RBP), which has a shorter half-life in the systemic circulation compared to albumin, and cholinesterase, a marker for liver synthesis and nutrition, were also examined. Plasma RBP concentration was measured by nephrometry (SRL Limited). The concentrations of total cholesterol were determined by automated enzymatic procedures. The concentrations of high-density lipoprotein-cholesterol (HDL-C) were determined after precipitation of apolipoprotein B-containing lipoproteins with phosphotungstic Mg²⁺. Non-HDL-C was calculated by subtracting values for HDL-C from those for total cholesterol. With regard to inflammation markers, plasma concentrations of IL-6 were measured in duplicate using a commercially available ELISA kit (Quantikine HS, Human IL-6; R&D Systems). C-reactive protein (CRP) was measured using a standard assay procedure in conjunction with a completely automated Hitachi 7170 system (Hitachi, Tokyo, Japan).

Assessment of the Cumulative Effects of Adipokine Dysregulation

To assess the cumulative effects of adipokine dysregulation, the adipose risk score (ARS) was created by simply summing the number of parameters for which the participants fell into the highest-risk tertiles; plasma HMW adiponectin < 10.1 mg/dL (lowest tertiles), plasma leptin concentration < 2.6 ng/mL (lowest tertiles), and TNF-α concentration > 3.80 pg/mL (highest tertiles).

Mortality

All-cause mortality was ascertained by telephone contact every 12 months until September 1, 2006, with a 223,373 person-day follow-up (median period, 768 days; range, 3–2253 days). The follow-ups of 297 centenarians (97.7% of the 304 original participants) were completed enabling confirmation of either the date of death or survival at the end of the observation period. Seven centenarians, who could not be traced for mortality information, were censored at the date of last contact (2.3%).

Statistical Analysis

Of the 304 centenarians who participated in the visiting survey of the Tokyo Centenarians Study, we obtained a complete data set of biomarkers at baseline for 252 participants (82.9%). Therefore, the following analyses were applied to these 252 centenarians. Participants' characteristics are expressed as means (standard deviation [SD]). Data are described as medians (interquartile ranges [IQR]) when they were not normally distributed, and logarithmically transformed for comparison with an unpaired t test or analysis of variance (ANOVA), or Kruskal–Wallis analysis was applied. The chi-square test was used to compare categorical variables. Cox proportional hazard models were used to test the association between biomedical risk factors and mortality. All the risk factors were coded into tertiles with the lowest risk group serving as a reference. A sequential approach was conducted whereby estimates were initially adjusted for age (per 6-month increase) and sex; adjusted for self-reported education, smoking status, BI category by tertile, CDR score, and the number of comorbidities as potential confounders; and then adjusted for the above covariates and conventional risk factors including serum levels of albumin (nutrient pathway), HDL-C (lipid pathway), and IL-6 (inflammation pathway). Originally we evaluated the full set of age-related diseases (14); however, the incidence of these comorbidities was considerably high, and only 3% of centenarians were free from them (14). Therefore, cardiovascular disease, non-skin cancer, diabetes, and renal disease were included in the models. Education was dichotomized as “high” for achievement of college, university or any schooling above high school, or “low” otherwise. Smoking was dichotomized as “never” or “ever.”

In a subset of 165 centenarians who were available for body mass index (BMI) measurement, we performed an additional analysis to see whether BMI affects the associations between hormonal biomarkers and mortality.

To assess the cumulative effects of adipokine dysregulation on mortality, Kaplan–Meier survival plots were constructed to illustrate the association between the number of adipose risk factors and survival probability using the log-rank test. ANOVA was used to compare the baseline characteristics across adipose risk categories.

Statistical analyses were conducted using the SPSS 13.0 software package (SPSS, Chicago, IL). The results were considered statistically significant at p <.05, and two-sided tests were applied.

Results

During the 2253 days (6.2 years) of follow-up, 208 of the 252 centenarians (82.5%) died. Of these centenarians, 158 (63.3%) were women. The baseline characteristics of the 252 centenarians are presented in Table 1. In general, centenarians were characterized by an array of indices for undernutrition.

Survival

The associations between mortality and the biomarkers of adipose endocrine function and the IGF-1 axis as well as conventional risk factors are presented in Table 2. The lowest tertile of leptin was consistently associated with a higher mortality risk in the initial model (hazard ratio [HR], 1.41; 95% CI, 1.01–1.99) and after adjustment for covariates (HR, 1.64; 95% CI, 1.14–2.35). The highest tertile of TNF-α was significantly associated with higher mortality (HR, 1.79; 95% CI, 1.27–2.50); however, the association was slightly attenuated after adjustment for covariates (HR, 1.45; 95% CI, 1.00–2.08). The lowest tertile of adiponectin appeared to be modestly associated with increased mortality risk, but the associations did not reach statistical significance. Regarding the IGF-1 system, the lowest tertile of IGF-1 was consistently associated with increased mortality in the initial and multivariate models (HR, 1.44; 95% CI, 1.04–2.01, and HR, 1.46; 95% CI, 1.04–2.06, respectively). The lowest tertile of IGFBP3 was also associated with higher mortality, but the association was attenuated and no longer statistically significant after adjustment for covariates (HR, 1.41; 95% CI, 1.00–2.01; and HR, 1.34; 95% CI, 0.95–1.91, respectively). Contrastingly, the IGF-1/IGFBP3 molar ratio was not associated with mortality among the centenarians.

When adjusted for the above-mentioned covariates as well as conventional risk factors, only the lowest tertile of leptin was significantly associated with a high mortality risk among centenarians (HR, 1.53; 95% CI, 1.04–2.24).

With the exception of non-HDL-C, all the risk factors from the nutrient, lipids, and inflammatory pathway were significantly associated with all-cause mortality among centenarians after adjustment for age and sex. However, in the multivariate model, the mortality risk remained significant only for the nutrient markers. Lower BMI was also significantly associated with higher mortality in the multivariate model (HR, 1.47; 95% CI, 1.02–2.13).

Because plasma concentrations of leptin and adiponectin were associated with fat mass, we performed additional analyses in a subset of 165 centenarians whose BMI were available to examine whether the BMI affects the associations between hormonal biomarkers and all-cause mortality (Table 3). The effects of leptin and adiponectin on mortality were remarkably emphasized in a subset of centenarians with a higher BMI (median, 22.2; range, 19.4–29.7) in all models. Interestingly, the effects of both IGF-1 and IGFBP3 on mortality risk were synergistically enhanced in centenarians with a higher BMI and attenuated in those with a lower BMI (median, 17.2; range, 13.0–19.3). In contrast, the effect of TNF-α was strikingly enhanced in persons with a lower BMI.

Adipose Tissue Function and Mortality

To evaluate the cumulative effects of adipokine dysregulation on mortality, an unadjusted Kaplan–Meier plot was constructed to illustrate the survival probability according to the numbers of adipose risk factors (Figure 1). The ARS was significantly associated with increased mortality risk as assessed by Cox regression analysis (HR, 1.49; 95% CI, 1.24–1.78; p <.001). This effect persisted even after adjustment for covariates and conventional risk factors (HR, 1.27; 95% CI, 1.03–1.55; p =.023).

Cross-sectional comparisons of baseline demographics, health and functional measurements, biomarkers, and conventional risk factors across ARS categories are shown in Table 4. According to the ARS categories, there was a gradual decrease in multiple biomarkers including HDL-C, adiponectin, leptin, albumin, IGF-1, and cholinesterase, as well as in physical and cognitive function at baseline. In contrast, a gradual increase was observed in TNF-α, IL-6, and CRP levels. Centenarians with no adipose risk were characterized by the highest scores of BI and MMSE; the highest concentrations of HDL-C, adiponectin, leptin, albumin, IGF-1, and cholinesterase; and the lowest concentrations of TNF-α, CRP, and IL-6. Although the population size was quite small, centenarians with three of the adipose risk factors were characterized as exhibiting multisystem deterioration-not only in hormonal pathways but also in terms of nutrition, inflammation, and lipids, as well as in physical and cognitive function.

Discussion

The present longitudinal study on centenarians describes the association of increased mortality with the decline in biomarkers of adipose endocrine function, as well as the IGF-1 axis. Additionally, we found that the ARS, reflecting cumulative dysregulation in multiple adipose-derived hormones, constitutes a strong marker of a poor prognosis independent of conventional risk factors. The results suggested that maintaining adipose endocrine function and the IGF-1 axis may be a contributing component of successful adaptation that counteracts age-related multisystem deterioration and promotes exceptional survival among centenarians.

Despite considerable uncertainty around the concept and operational definition of frailty, there is general agreement that the core feature of frailty is increased vulnerability to stressors due to decline in homeostatic reserve in multiple physiological systems (36). This process of frailty is inextricably linked to the aging process. For example, frailty index, defined by accumulation of deficits in comprehensive geriatric assessment, was consistently correlated with chronological age (37), and cross-sectional studies demonstrated that almost all centenarians had at least some functional deficit (15,16). Furthermore, we demonstrated that an array of contributing factors to frailty including poor nutrition, lower BMI, and cytokine upregulation predicts mortality among centenarians, suggesting that the mortality force of centenarians is analogous, if not identical to that in frail elderly persons. Therefore, uncovering mechanisms of successful adaptation at the final stage of aging could further our understanding of the biological underpinning of frailty. This may help us to develop better means of preventing or delaying some aspects of the frailty process.

Extensive investigation has focused on muscle wasting as a core component of frailty; however, little is known about the implication of the loss of adipose tissue mass and function. Our analysis stratified by BMI may evoke a protective role of adipose endocrine function against wasting syndrome in centenarians. As compared to centenarians with a higher BMI, those with a lower BMI exhibited attenuated associations of leptin and adiponectin with mortality risk. One possible explanation for these findings is that centenarians with lower BMI are more likely to suffer from malnutrition and cachexia, and this may have contributed to the higher mortality observed in this subgroup. The strikingly enhanced association of mortality risk with TNF-α, also known as cachectin in this subgroup, supports this view. Furthermore, this notion is compatible with the findings that upregulation of TNF-α, but not of IL-6 or CRP, predicted mortality among 126 Danish centenarians, possibly in relation to increased energy expenditure and cachexia (38). In contrast, in centenarians with higher BMI, the mortality effects of leptin and adiponectin were emphasized, but those of TNF-α were attenuated, suggesting that the defense mechanism mediated by adipose-derived hormones requires relevant fat mass.

Alternatively, the higher mortality in centenarians with a lower BMI as compared to their counterparts may be explained by the effect of weight loss itself. Weight loss is associated with increased mortality, particularly in older men (39). The negative effects of weight loss are mediated by numerous factors including protein energy malnutrition, immune dysfunction, bone loss and fracture, and anorexia, which could mask the associations between adipokines and mortality in the subset with a low BMI. Further exploration of dynamic change of BMI in relation to manifestations of weight loss is needed to better understand the ways in which adipokine dysregulation and/or weight loss itself affects the health status in centenarians.

Age-related decline in the GH/IGF-1 axis, along with poor nutrition and cytokine upregulation, has been proposed to contribute to sarcopenia, another form of muscle wasting. Experimental evidence suggests that age-associated mitochondrial dysfunction and the resulting energy depletion in muscle cells might play a prominent role in the development of sarcopenia (40). In this study, the decline in the IGF-1 axis was associated with increased mortality in centenarians, and this association was mediated by BMI in parallel with the case of leptin and adiponectin. Furthermore, our study demonstrated a significant association between the IGF-1 axis and the ARS, suggesting that these two hormonal pathways on the basis of sufficient fat might synergistically construct a defense mechanism against the wasting syndrome including sarcopenia and cachexia. However, this notion is challenged by an emerging entity of “sarcopenic obesity,” where excess fat and adipokine overproduction aggravates age-related muscle wasting (41). Our study provides little information on the association between excess adiposity and health effects of adipokines. Nevertheless, the absence of obese centenarians (BMI > 30.0) in this study might indicate that greater fat storage is not necessarily protective against age-related wasting syndrome. Further studies aiming at adipocyte size and metabolism, and its molecular regulators such as peroxisome proliferator-activated receptor gamma (PPAR-γ) are warranted to understand the controversy regarding the association among adiposity, adipokines, and aging.

Comprehensive analysis of multiple biomarkers across adipose risk categories provides an insight into the potential mechanism underlying the mediation of exceptional survival by adipose endocrine function among centenarians. We described herein a graded relationship of increasing the ARSs with decline in multiple key pathways of health maintenance, including physical and cognitive function, IGF-1 axis, HDL metabolism, and nutrient and liver synthesis, and with upregulation of inflammatory pathways. We propose a hypothetical interpretation for this finding as follows. At the ultimate stage of aging, each regulatory pathway does not function in isolation, but rather serves as a component of an interactive metabolic network that regulates whole-body energy homeostasis to maintain multiple physiological functions indispensable for survival. In this network, adipose-derived hormones, particularly leptin, serve as an integral feedback of peripheral nutrient, hormonal, and metabolic reserve to the central nervous system (CNS) sensing system. Therefore, stability of this metabolic network can be described as the principal component of health maintenance and longevity. Further research is obviously required to elucidate the detailed construction and gain a mechanistic insight into this metabolic network. However, this perspective reasonably fits with an emerging concept from current advances in obesity research, thereby delineating a highly integrated metabolic system, which comprises the hypothalamus as a CNS integrator, adipokines as afferent signals to CNS, and sympathetic nerves as a wiring link between key tissues to maintain whole-body energy homeostasis in response to nutritional availability and energy expenditure (42–46). In states of overnutrition, leptin has a critical role as an antiobesity hormone; however, when exposed to chronic nutritional excess, the system becomes overloaded, and eventually induces insulin resistance and obesity (47). Interestingly, adipose tissue deficiency or lipodystrophy is also associated with insulin resistance and metabolic dysregulation (48,49). Therefore, insulin sensitivity and/or resistance may be a surrogate marker for the effectiveness of this integrated metabolic system. In this view, Paolisso and colleagues (50) reported that preserved insulin sensitivity is one of the biological peculiarities of healthy centenarians. Although measurement of insulin resistance was lacking in the present study, the associations with the insulin-sensitizing hormones such as adiponectin and leptin with mortality in centenarians support the findings by Paolisso and colleagues. Moreover, insulin resistance was demonstrated as one of the precursors of the development of frailty in older adults (51). Collectively, the adaptive response to either energy excess or deprivation and protection against age-related multisystem decline that leads to frailty may share a common pathway mediated by the metabolic regulatory network. If our hypothesis is in the right direction, interventions to enhance the stability of this metabolic network, rather than those aiming at a single hormonal supplementation, should be encouraged for the prevention of frailty and the promotion of healthy aging.

This study has several limitations. First, it is important to note that population studies of extremely old people may sometimes be affected by a selection bias, because relatively healthy people tend to be selected. In the present study, 59% of the participants who took the mailed survey subsequently agreed to a visiting survey. Although distributions of sex, self-reported education, and smoking status were not different between responders and nonresponders to the visiting survey, we cannot exclude the possibility that we had sampled relatively healthier centenarians in this survival analysis. Second, assessment of risk factors relied on a single measurement at baseline; this could likely underestimate their impact on all-cause mortality. Neither a single measurement of BMI at one point in time can capture the dynamic complex of weight change, nor may BMI be a reliable indicator of adiposity in the oldest old. Third, the risk factors studied did not include many important factors the impact of which on health status, disability, and survival in the oldest old are of general interest (e.g., the hypothalamo-pituitary-adrenal axis [adrenopause] and sex hormones). Finally, the generalizability of our findings from centenarians needs to be verified in other oldest-old populations. Further studies with proper interpretation of weight history, adiposity, and fat distribution in relation to health and mortality in older adults would help to overcome some of these limitations.

Conclusion

Despite several limitations, our results demonstrated a potential contribution of adipose endocrine function and the IGF-1 axis to exceptional survival beyond the age of 100 years. Adipose-derived hormones, particularly leptin, are of great importance because of their role as integral components of the metabolic network that maintains health and promotes survival against the mortality force associated with frailty. Understanding the interactions and synergism between these metabolic pathways is likely to be important in developing effective interventions for the promotion of healthy aging and longevity.

Decision Editor: Luigi Ferrucci, MD, PhD

Figure 1.

The adipose risk score was created by simply summing the number of parameters for which the participants fell into the highest-risk tertiles as follows: plasma high-molecular-weight adiponectin < 10.1 mg/dL (lowest tertile), plasma leptin concentration < 2.6 ng/dL (lowest tertile), and tumor necrosis factor-α concentration > 3.80 pg/mL (highest tertile)

Table 1.

Baseline Characteristics of Participants.

Characteristics	All Participants (N = 252)
Sociodemographics
Age, mean (SD), y	101.5 (1.8)
Women, n (%)	196 (77.8)
Higher education, n (%)	52 (20.6)
Institutionalized, n (%)	75 (29.8)
Current or former smoker, n (%)	42 (16.7)
Health and function
Barthel Index, median, (IQR)	45 (15–80)
Clinical Dementia Rating
Normal cognitive function, n (%)	66 (26.2)
Mild cognitive impairment, n (%)	39 (15.5)
Mild dementia, n (%)	53 (21.0)
Moderate dementia, n (%)	23 (9.1)
Severe dementia, n (%)	71 (28.2)
MMSE*, median, (IQR)	15 (7–21)
Body mass index,† median (IQR)	19.4 (17.2–21.8)
SBP, mean (SD), mmHg	143 (23)
DBP, mean (SD), mmHg	77 (13)
Comorbidities
Cardiovascular disease, n (%)	81 (26.6)
diabetes, n (%)	18 (5.9)
Renal disease, n (%)	34 (13.5)
Non-skin cancer, n (%)	23 (9.2)
Biomarkers of hormonal pathways
Adiponectin, median, (IQR), mg/dL	12.7 (1.1–48.5)
Leptin, median, (IQR), ng/mL	3.6 (0.9–21.9)
TNF-α, median, (IQR), pg/mL	3.28 (1.56–27.42)
IGF-1, mean (SD), ng/mL	62.5 (26.6)
IGFBP3, mean (SD), μg/mL	1.7 (0.6)
IGF-1/IGFBP3 molar ratio, mean (SD)	0.15 (0.06)
Conventional risk factors
Total cholesterol, mean (SD), mg/dL	170 (34)
HDL cholesterol, mean (SD), mg/dL	51 (13)
Non-HDL cholesterol, mean (SD), mg/dL	117 (32)
Albumin, mean (SD), g/dL	3.6 (0.4)
RBP, mean (SD), mg/dL	3.5 (1.4)
Cholinesterase, mean (SD), IU/L	216 (56)
CRP, median, (IQR), mg/dL	0.15 (0.00–15.87)
IL-6, median, (IQR), pg/mL	2.90 (1.18–40.70)

Notes: *Data were obtained from 227 participants.

†Data were obtained from 165 participants.

SD = standard deviation; IQR = interquartile range; MMSE = Mini-Mental State examination; SBP = systolic blood pressure; DBP = diastolic blood pressure; TNF-α = tumor necrosis factor-α; IGF-1 = insulin-like growth factor-1; IGFBP3 = IGF-binding protein-3; HDL = high-density lipoprotein; RBP = retinol binding protein; CRP = C-reactive protein; IL-6 = interleukin-6.

Table 2.

Biomarkers of Hormonal Pathways and Conventional Risk Factors in Relation to All-Cause Mortality Beyond 100 Years of Age.

	Adjusted for Age and Sex			Adjusted for Covariates*			Adjusted for Covariates and Conventional Risk Factors†
Biomarkers	HR (95% CI)	p Value	HR (95% CI)	p Value	HR (95% CI)	p Value
Adipose endocrine function
Adiponectin, tertiles
>15.7	1.00		1.00		1.00
10.1–15.7	0.99 (0.71–1.39)	.952	1.06 (0.74–1.52)	.738	0.97 (0.68–1.40)	.879
<10.1	1.24 (0.88–1.73)	.216	1.23 (0.86–1.76)	.252	1.27 (0.87–1.89)	.214
Trend	1.11 (0.94–1.32)	.217	1.11 (0.93–1.33)	.249	1.13 (0.93–1.37)	.232
Leptin, tertiles
>5.2	1.00		1.00		1.00
2.6–5.2	1.09 (0.77–1.51)	.632	1.34 (0.93–1.91)	.113	1.43 (0.99–2.06)	.056
<2.6	1.41 (1.01–1.99)	.047	1.64 (1.14–2.35)	.008	1.53 (1.04–2.24)	.030
Trend	1.19 (1.00–1.42)	.049	1.28 (1.07–1.53)	.008	1.24 (1.03–1.50)	.025
TNF-α, tertiles
<2.93	1.00		1.00		1.00
2.93–3.80	1.21 (0.86–1.71)	.276	1.28 (0.89–1.83)	.181	1.32 (0.92–1.90)	.131
>3.80	1.79 (1.27–2.50)	.001	1.45 (1.00–2.08)	.047	1.31 (0.89–1.92)	.177
Trend	1.34 (1.13–1.59)	.001	1.20 (1.00–1.44)	.046	1.15 (0.95–1.39)	.151
IGF-1 system
IGF-1, tertiles
>70	1.00		1.00		1.00
50–70	0.91 (0.64–1.30)	.615	0.94 (0.64–1.36)	.727	0.95 (0.65–1.39)	.779
<50	1.44 (1.04–2.01)	.030	1.46 (1.04–2.06)	.029	1.28 (0.89–1.83)	.182
Trend	1.21 (1.02–1.44)	.031	1.21 (1.01–1.44)	.035	1.12 (0.94–1.35)	.210
IGFBP3, tertiles
>1.99	1.00		1.00		1.00
1.45–1.99	0.94 (0.66–1.33)	.709	0.90 (0.63–1.30)	.589	1.04 (0.72–1.50)	.853
<1.45	1.41 (1.00–1.43)	.048	1.34 (0.95–1.91)	.098	1.24 (0.86–1.79)	.255
Trend	1.20 (1.01–1.43)	.044	1.17 (0.97–1.40)	.093	1.11 (0.92–1.34)	.265
Molar ratio, tertiles
>0.161	1.00		1.00		1.00
0.125–0.161	0.96 (0.68–1.35)	.820	0.98 (0.69–1.39)	.926	0.88 (0.62–1.26)	.481
<0.125	1.20 (0.85–1.71)	.305	1.30 (0.90–1.87)	.165	1.16 (0.80–1.69)	.437
Trend	1.10 (0.92–1.31)	.320	1.13 (0.94–1.36)	.192	1.07 (0.88–1.30)	.489
Nutrition
Albumin, tertiles
>3.8	1.00		1.00
3.5–3.8	1.34 (0.94–1.93)	.109	1.09 (0.73–1.61)	.677	NA
<3.5	2.58 (1.81–3.67)	<.001	2.10 (1.42–3.10)	<.001	NA
Trend	1.63 (1.36–1.95)	<.001	1.48 (1.21–1.81)	<.001	NA
RBP, tertiles
>4.0	1.00		1.00
2.9-4.0	1.26 (0.89–1.77)	0.192	1.21 (0.85–1.74)	0.293	NA
<2.9	1.71 (1.20–2.44)	0.003	1.61 (1.09–2.36)	0.016	NA
Trend	1.31 (1.09–1.57)	0.003	1.27 (1.05–1.54)	0.015	NA
Cholinesterase, tertiles
>237	1.00		1.00
190-237	1.10 (0.77–1.57)	.597	1.15 (0.80–1.66)	.451	NA
<190	2.19 (1.54–3.11)	<.001	1.80 (1.24–2.60)	.002	NA
Trend	1.51 (1.25–1.81)	<.001	1.35 (1.12–1.63)	.002	NA
BMI, dichotomized‡
≥19.4	1.00		1.00
<19.4	1.39 (0.97–1.98)	.071	1.47 (1.02–2.13)	.039	NA
Lipid
HDL cholesterol, tertiles
>58	1.00		1.00
46–58	1.29 (0.92–1.82)	.146	1.23 (0.85–1.78)	.458	NA
<46	1.70 (1.22–2.37)	.002	1.36 (0.95–1.95)	.099	NA
Trend	1.30 (1.10–1.54)	.002	1.16 (0.97–1.39)	.100	NA
Non-HDL cholesterol, tertiles
<101	1.00		1.00
101–123	0.93 (0.66–1.299)	.649	0.97 (0.68–1.38)	.865	NA
>123	0.89 (0.63–1.25)	.507	0.85 (0.60–1.21)	.359	NA
Trend	0.94 (0.80–1.12)	.507	0.92 (0.77–1.10)	.358	NA
Inflammation
CRP, tertiles
<0.08	1.00		1.00
0.08–0.31	1.01 (0.72–1.41)	.949	0.88 (0.63–1.25)	.480	NA
>0.31	1.47 (1.05–2.04)	.023	1.26 (0.89–1.78)	.198	NA
Trend	1.21 (1.02–1.43)	.029	1.11 (0.93–1.34)	.242	NA
IL-6, tertiles
<2.50	1.00		1.00
2.50-3.74	1.45 (1.03–2.04)	.035	1.29 (0.89–1.86)	.180	NA
>3.74	1.61 (1.15–2.26)	.005	1.39 (0.98–1.97)	.068	NA
Trend	1.27 (1.07–1.49)	.005	1.17 (0.99–1.40)	.070	NA

Notes: *Adjusted for age, sex, education, smoking, Barthel Index, Clinical Dementia Rating (CDR) scale, number of comorbidities.

†Adjusted for age, sex, education, smoking, Barthel Index, CDR scale, numbers of comorbidities, and serum levels of albumin, HDL-C, and IL-6.

‡Data were obtained from 165 participants.

HR = hazard ratio; CI = confidence interval; NA = not applicable; TNF-α = tumor necrosis factor-α; IGF-1 = insulin-like growth factor-1; IGFBP3 = IGF-binding protein 3; RBP = retinol binding protein; HDL = high-density lipoprotein; CRP = C-reactive protein; IL-6 = interleukin-6; BMI = body mass index.

Table 3.

Adipokines, IGF-1 Axis, and All-Cause Mortality Stratified by Body Mass Index (N = 165).

	Lower BMI (Median, 17.2; Range, 13.0–19.3) (N = 82)					Higher BMI (Median, 22.2; Range, 19.4–29.7) (N = 83)
Biomarkers	Age and Sex Adjusted HR (95% CI)	Multivariate Adjusted* HR (95% CI)	Multivariate Plus Conventional Risk Factors Adjusted† HR (95% CI)	Age and Sex Adjusted HR (95% CI)	Multivariate Adjusted* HR (95% CI)	Multivariate Plus Conventional Risk Factors Adjusted† HR (95% CI)
Adipose endocrine function
Adiponectin, tertiles
>15.7	1.00	1.00	1.00	1.00	1.00	1.00
10.1–15.7	0.96 (0.50–1.72)	0.88 (0.41–2.81)	0.79 (0.37–1.71)	1.43 (0.69–2.94)	1.65 (0.73–3.72)	1.52 (0.65–3.58)
<10.1	1.49 (0.80–2.76)	1.14 (0.46–1.85)	1.30 (0.51–3.32)	1.78 (0.89–3.56)	2.08 (0.97–4.43)	1.94 (0.81–4.64)
Trend	1.23 (0.89–1.70)	1.09 (0.69–1.73)	1.18 (0.73–1.92)	1.34 (0.93–1.94)	1.42 (0.99–2.06)	1.34 (0.90–2.14)
Leptin, tertiles
>5.2	1.00	1.00	1.00	1.00	1.00	1.00
2.6–5.2	0.99 (0.44–2.20)	1.18 (0.53–2.79)	1.64 (0.62–4.35)	0.93 (0.51–1.70)	0.94 (0.48–1.83)	0.72 (0.34–1.45)
<2.6	1.26 (0.56–2.80)	1.28 (0.56–3.06)	1.54 (0.54–4.37)	2.20 (1.07–4.55)‡	2.40 (1.02–5.66)‡	3.12 (1.19–8.23)‡
Trend	1.17 (0.80–1.70)	1.10 (0.74–1.63)	1.14 (0.71–1.82)	1.33 (0.94–1.88)	1.35 (0.91–2.01)	1.30 (0.84–2.02)
TNF-α, tertiles
<2.93	1.00	1.00	1.00	1.00	1.00	1.00
2.93–3.80	0.98 (0.55–1.75)	1.17 (0.61–2.25)	1.28 (0.64–2.58)	1.14 (0.58–2.22)	1.13 (0.55–2.32)	1.00 (0.45–2.22)
>3.80	1.82 (1.01–3.28)‡	3.26 (1.58–6.96)‡	2.97 (1.21–7.27)‡	1.91 (0.97–3.78)	1.67 (0.78–3.59)	1.46 (0.60–3.56)
Trend	1.33 (0.98–1.81)	1.74 (1.18–2.56)‡	1.64 (1.06–2.52)‡	1.39 (0.98–1.98)	1.29 (0.84–1.91)	1.20 (0.77–1.88)
IGF-1 axis
IGF-1, tertiles
>70	1.00	1.00	1.00	1.00	1.00	1.00
50–70	1.02 (0.57–1.80)	1.27 (0.65–2.47)	1.01 (0.50–2.03)	1.04 (0.54–2.02)	1.13 (0.54–2.34)	1.14 (0.55–2.37)
<50	1.28 (0.69–2.39)	1.28 (0.66–2.47)	1.06 (0.52–2.19)	2.26 (1.17–4.51)‡	2.52 (1.20–5.38)‡	2.13 (0.97–4.66)
Trend	1.13 (0.82–1.57)	1.13 (0.81–1.57)	1.03 (0.72–1.48)	1.50 (1.06–2.12)‡	1.62 (1.08–2.42)‡	1.43 (0.96–2.14)
IGFBP3, tertiles
>1.99	1.00	1.00	1.00	1.00	1.00	1.00
1.45–1.99	0.90 (0.50–1.63)	0.85 (0.46–1.60)	0.90 (0.47–1.72)	0.87 (0.44–1.70)	0.93 (0.45–1.93)	1.03 (0.49–2.17)
<1.45	1.20 (0.68–2.14)	1.04 (0.54–2.02)	1.01 (0.48–2.12)	2.30 (1.17–4.51)‡	2.54 (1.20–5.38)‡	2.32 (1.01–5.34)‡
Trend	1.09 (0.81–1.47)	1.02 (0.73–1.42)	1.00 (0.69–1.45)	1.55 (1.06–2.24)‡	1.62 (1.08–2.42)‡	1.46 (0.95–2.24)

Notes: *Adjusted for age, sex, education, smoking, Barthel Index, Clinical Dementia Rating (CDR) scale, and number of comorbidities.

†Adjusted for age, sex, education, smoking, Barthel Index, CDR scale, numbers of comorbidities, and serum levels of albumin, high-density lipoprotein cholesterol, and interleukin-6.

‡p <.05.

BMI = body mass index; HR = hazard ratio; CI = confidence interval; TNF-α = tumor necrosis factor-α; IGF-1 = insulin-like growth factor-1; IGFBP3 = IGF-binding protein 3.

Table 4.

Clinical Phenotypes of Centenarians According to the Adipose Risk Score.

	Adipose Risk Score‡
	0 (N = 69)	1 (N = 121)	2 (N = 58)	3 (N = 4)	p Value 1	p Value 2
Demography
Age,* median (IQR), y	100.4 (100.2–101.7)	100.8 (100.2–102.2)	100.8 (100.2–102.6)	102.3 (100.5–105.5)	.205	.141
Women, n (%)	56 (81.2)	98 (81.0)	40 (69.0)	2 (50.0)	.134	.091
Health and function
Barthel Index,* median (IQR)	60 (30–92.5)	40 (15–75)	32.5 (10–80)	10 (1.25–26.25)	.013	.016
MMSE,*† median (IQR)	17 (11–23)	14 (6–21)	10 (6–20)	6 (1–16.25)	.024	.015
CDR,* median, (IQR)	0.5 (0.0–2.0)	1.0 (0.0–3.0)	1.0 (0.5–3.0)	2.5 (0.875–3.0)	.024	.013
SBP, mean (SD), mmHg	142 (22)	145 (24)	140 (24)	126 (18)	.204	.198
DBP, mean (SD), mmHg	78 (12)	77 (14)	77 (12)	74 (5)	.942	.890
Biomarkers of hormonal pathways
Adiponectin, median, (IQR), mg/dL	15.0 (12.2–18.6)	13.1 (8.4–17.8)	9.3 (6.7–12.9)	8.5 (5.3–9.6)	<.001	<.001
Leptin, median, (IQR), ng/mL	4.6 (3.3–7.4)	3.4 (2.0–6.2)	2.5 (1.9–5.6)	1.6 (1.3–1.9)	<.001	<.001
TNF-α, median, (IQR), pg/mL	2.93 (2.60–3.20)	3.24 (2.63–3.89)	4.25 (3.90–4.93)	4.20 (3.96–4.36)	<.001	<.001
IGF-1, mean (SD), ng/mL	69 (29)	62 (26)	57 (24)	43 (13)	.030	.020
IGFBP3, mean (SD), μg/mL	1.8 (0.6)	1.8 (0.6)	1.6 (0.6)	1.2 (0.3)	.078	.079
IGF-1/IGFBP3 molar ratio, (SD)	0.16 (0.08)	0.15 (0.04)	0.15 (0.05)	0.15 (0.07)	.344	.244
Conventional risk factors
Total cholesterol, mean (SD), mg/dL	172 (29)	172 (36)	163 (35)	140 (28)	.125	.141
HDL-cholesterol, mean (SD), mg/dL	58 (13)	53 (13)	49 (12)	37 (5)	<.001	<.001
Non-HDL cholesterol, mean (SD), mg/dL	114 (24)	120 (34)	115 (35)	104 (31)	.456	.341
Albumin, mean (SD), g/dL	3.8 (0.3)	3.6 (0.4)	3.5 (0.4)	3.2 (0.6)	<.001	<.001
RBP, mean (SD), mg/dL	3.6 (1.3)	3.5 (1.6)	3.4 (1.2)	2.9 (0.9)	.779	.729
Cholinesterase, mean (SD), IU/L	237 (47)	228 (51)	197 (62)	192 (52)	.002	.001
CRP, median, (IQR), mg/dL	0.07 (0.04–0.25)	0.17 (0.06–0.55)	0.20 (0.08–0.72)	1.05 (0.38–2.74)	.001	.001
IL-6, median, (IQR), pg/mL	2.61 (2.08–3.38)	2.99 (2.35–4.33)	3.40 (2.55–5.89)	4.16 (3.06–6.20)	.003	.001

Notes: p values 1 are for the comparison between 0, 1, 2, and 3 adipose risk score category by analysis of variance; p values 2 are for the comparison between 0, 1, and 2 or more ARS category by analysis of variance (ANOVA).

*Differences were calculated by Kruskal–Wallis analysis.

†Data were obtained in 227 participants.

‡Calculated by summing the number of parameters corresponded to the following values; high-molecular-weight adiponectin < 10.1 mg/dL, leptin < 2.6 ng/mL, and tumor necrosis factor-α > 3.80 pg/mL.

SD = standard deviation; IQR = interquartile range; MMSE = Mini-Mental State Examination; CDR = clinical dementia rating; SBP = systolic blood pressure; DBP = diastolic blood pressure; TNF-α = tumor necrosis factor-α; IGF-1 = insulin-like growth factor-1; IGFBP3 = IGF-binding protein-3; RBP = retinol binding protein; HDL = high-density lipoprotein; CRP = C-reactive protein; IL-6 = interleukin-6.

This study was funded by a grant from the Ministry of Health, Welfare, and Labor for the Scientific Research Project for Longevity (Multidisciplinary approach to centenarians and international comparison, and Research on healthy aging: Semisupercentenarians and long-lived sibling study, principal investigator, Nobuyoshi Hirose) a grant from Keio Health Consulting Center, and a grant from the Foundation for Total Health Promotion (2006). The funders did not have any role in the design and conduct of the study, in the collection, analysis, and interpretation of the data, or in the preparation, review, or approval of the manuscript.

Author notes

1Division of Geriatric Medicine, Department of Internal Medicine, and 2Department of Preventive Medicine and Public Health, Keio University School of Medicine, Tokyo, Japan.