Insulin Resistance Syndrome in the Elderly Assessment of Functional, Biochemical, Metabolic, and Inﬂammatory Status

Cardiovascular and Metabolic Risk ORIGINAL ARTICLE

Insulin Resistance Syndrome in the Elderly Assessment of functional, biochemical, metabolic, and inﬂammatory status

1,2 2 WILLIAM A. BANKS, MD DAVID R. THOMAS, MD nosing IRS is difficult. As such, diagnosis 3 1,2 LISA M. WILLOUGHBY, PHD JOHN E. MORLEY, MB, BCH is often made based on the presence of secondary criteria, such as hypertension, history of glucose intolerance, or elevated OBJECTIVE — Hyperinsulinemic euglycemia, or insulin resistance syndrome (IRS), is asso- BMI (2). As these are also associated with ciated with increased morbidity and mortality. Although thought to be associated with proin- diabetes and likely are sequella develop- flammatory states, little work has been done in this area. Here, we determined the impact of IRS ing from IRS, such a diagnostic approach on functional, biochemical, metabolic, and inflammatory status in a high-risk population: el- makes it difficult to ascertain incidence of derly women in nursing homes. IRS or to study early pathological events that may be unique to it. These difficulties RESEARCH DESIGN AND METHODS — Functional, biochemical, metabolic, and are magnified in the elderly, as many of inflammatory parameters were measured in 100 consecutive ambulatory, elderly women who resided in nursing homes. Diabetic subjects and residents with fasting blood glucose Ն110 mg/dl the secondary criteria are independently were excluded. Remaining residents were classified as insulin resistant (IR) (insulin Ͼ100 associated with aging. pmol/l) or non-IR (NIR). Here, we studied a population of elderly, ambulatory women living in long- RESULTS — A total of 16 residents were IR and 53 NIR. No differences in functional status, term care facilities in a Midwestern BMI, renal function, C-reactive protein, or immune cell levels were found. Fasting blood glucose metropolitan area. We assessed in the eu- was higher in IR subjects ([means Ϯ SD] 94.1 Ϯ 8.1 vs. 87.9 Ϯ 8.2, P Ͻ 0.05), indicating a very glycemic subset their functional, bio- mild glucose intolerance. Serum C-peptide (P Ͻ 0.05), amylin (P Ͻ 0.01), and leptin (P Ͻ 0.01), chemical, metabolic, and inflammatory but not adiponectin or resistin, were higher in IR subjects. Higher leptin-to-BMI and insulin– parameters as a function of endogenous to–C-peptide ratios suggested an increased percent body fat mass and altered clearance of serum insulin levels. insulin, respectively. Eleven of 13 cytokines had arithmetic elevations, but only tumor necrosis factor-␣ (TNF) reached statistical significance (P Ͻ 0.01). TNF and insulin levels were highly correlated. RESEARCH DESIGN AND CONCLUSIONS — IRS in the healthiest of long-term care residents is relatively rare but is METHODS — Studies were approved associated with mild glucose intolerance, increased percent body fat, altered insulin clearance, by the local institutional review board, and a proinflammatory status as evidenced by an elevated TNF. and consent forms were obtained on all participants. One hundred elderly ambu- Diabetes Care 30:2369–2373, 2007 latory women aged Ն65 years were recruited from eight long-term care facilities nsulin resistance syndrome (IRS) is IRS is defined on the basis of serum insu- in the St. Louis metropolitan area. Exclu- characterized by a decreased tissue lin levels regardless of glucose status. The sion criteria for the study were nonambu- I sensitivity to the action of insulin, lead- population of IRS that clearly does not latory status or a contraindication for ing to a compensatory increase in insulin overlap with diabetes is that with eugly- exercise. A fasting morning blood sample secretion (1). It is thought that most cemic hyperinsulinemia. Residents with was drawn and measured for glucose, adults with IRS maintain normal glucose IRS, despite euglycemia, are at increased electrolytes, blood urea nitrogen, creati- levels and will never develop overt type 2 risk for hypertension, stroke, polycystic nine, blood urea nitrogen–to–anion gap diabetes but are nonetheless at increased ovary syndrome, and nonalcoholic ste- ratio, anion gap, C-reactive protein, pre- risk for cardiovascular disease. Whereas atohepatitis (2). albumin, albumin, transferrin, white diabetes is defined on the basis of serum Because measuring serum insulin is blood cells (further assessed as percent- glucose levels regardless of insulin status, not part of routine clinical practice, diag- ages of neutrophils, lymphocytes, eosino- phils, and segments), platelets, red cell ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● indexes, and hemoglobin. Blood samples From the 1Geriatric Research, Education and Clinical Center, Veterans Affairs Medical Center, St. Louis, 2 were also measured blinded with multi- Missouri; the Division of Geriatrics, Department of Internal Medicine, St. Louis University School of plex kits (Millipore, St. Charles, MO) for Medicine, St. Louis, Missouri; and the 3Center for Outcomes Research, Department of Internal Medicine, St. Louis University School of Medicine, St. Louis, Missouri. insulin, C-peptide, leptin, total amylin, Address correspondence and reprint requests to William A. Banks, MD, 915 N. Grand Blvd., St. Louis, adiponectin, resistin, interleukin (IL)-1␣, MO 63106. E-mail: [email protected]. IL-1␤, IL-1 receptor antagonist, IL-2, Received for publication 3 April 2007 and accepted in revised form 21 May 2007. IL-4, IL-6, IL-8, IL-10, interferon-␥, tu- Published ahead of print at http://care.diabetesjournals.org on 29 May 2007. DOI: 10.2337/dc07-0649. ␣ Abbreviations: IL, interleukin; IR, insulin resistant; IRS, insulin resistance syndrome; MAP, macrophage mor necrosis factor (TNF)- , eotaxin, inflammatory protein; NIR, non–insulin resistant; RANTES, regulated on activation, normal T-cell expressed macrophage inflammatory protein and secreted; TNF, tumor necrosis factor. (MAP)-1␣, and regulated on activation, A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion normal T-cell expressed and secreted factors for many substances. (RANTES). Any multiplex levels above © 2007 by the American Diabetes Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby the detection limit were diluted and as- marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. sayed again. Any multiplex levels below

DIABETES CARE, VOLUME 30, NUMBER 9, SEPTEMBER 2007 2369 Insulin resistance syndrome in the elderly

Table 1—Resident characteristics elevated in the IR group (Table 2). Adi- ponectin showed a statistical trend (P ϭ IR group NIR group P 0.08) toward being lower in the IR group. The proinflammatory cytokine TNF was n 16 53 ϳ50% higher (P Ͻ 0.01) (Table 2); al- Age (years) 81.9 Ϯ 6.8 83.8 Ϯ 6.6 0.43 though the other cytokines except for Height (cm) 155.7 Ϯ 5.5* 156.7 Ϯ 7.3 0.61 MIP-1␣ and RANTES were arithmetically Weight (kg) 65.6 Ϯ 9.3* 62.3 Ϯ 13.5 0.20 higher in the IR group, none reached sta- BMI (kg/m2) 26.6 Ϯ 2.9* 25.1 Ϯ 5.3 0.09 tistical significance. C-reactive protein, Charlson index score 2.8 Ϯ 0.8 2.9 Ϯ 1.4 0.89 immune cell measures, and neither of the Smoker status 0.6 proinflammatory indexes were signifi- Current smoker 0 (0) 3 (6)† cantly different between the IR and NIR Never smoked 5 (31) 14 (28) groups. Former smoker 11 (69) 33 (66) Ratios of leptin to BMI (P Ͻ 0.01), Consumes alcohol 0 (0) 10 (20)† 0.052 insulin to C-peptide (P Ͻ 0.001), insulin Mini–mental state examination 22.3 Ϯ 6.4* 21.4 Ϯ 7.2‡ 0.82 to BMI (P Ͻ 0.001), and insulin to TNF Functional independence measure 99.4 Ϯ 20.4 94.2 Ϯ 20.6 0.23 (P Ͻ 0.001) were significantly higher in Mini–nutritional assessment 21.1 Ϯ 4.1‡ 21.7 Ϯ 4.4 0.77 the IR group (Table 3). The ratios for in- ADL 12.2 Ϯ 4.1 12.8 Ϯ 4.5 0.79 sulin to leptin, TNF to BMI, and insulin to GDS 8.5 Ϯ 7.0† 7.1 Ϯ 5.2§ 0.65 amylin were not different between the Get-up-and-go test 30.6 Ϯ 21.4† 32.7 Ϯ 18.2 0.31 two groups. At 6 meters 13.9 Ϯ 6.7† 17.4 Ϯ 14.4§ 0.39 Significant positive correlations were At 6 min 148.9 Ϯ 34.2‡ 138.8 Ϯ 67.9¶ 0.45 found between BMI and leptin (P Ͻ 0.001, Data are means Ϯ SD or n (%). *Based on n Ϫ1. †Based on n Ϫ 3. ‡Based on n Ϫ 2. §Based on n Ϫ 5. Based r ϭ 0.68), TNF and insulin (P Ͻ 0.01, r ϭ on n Ϫ 7. ¶Based on n Ϫ 13. ADL, activities of daily living; GDS, geriatric depression scale. 0.36), leptin and insulin (P Ͻ 0.01, r ϭ 0.31), and insulin and the leptin-to-BMI ratio (P Ͻ 0.01, r ϭ 0.34). Figure 1 shows the detection limit were assigned the and values ranged from 0 to 4. The second the relation between TNF and log insulin value of 1.0. Residents were weighed, inflammatory score excluded C-reactive (Y ϭ 9.21x ϩ 5.36, r ϭ 0.533, n ϭ 69, P Ͻ their height measured, and their BMI cal- protein for a measure based entirely on 0.01). Adiponectin and leptin were in- culated as weight (in kilograms) divided cytokine measures. versely correlated (P Ͻ 0.001, r ϭ by the square of height (in meters). Mini– Ϫ0.50). There were trends for associations mental status examination (3), mini– Analysis of data between C-reactive protein versus TNF nutritional assessment (4), activities of Means are reported with their SDs and n. (P ϭ 0.07) and insulin versus BMI (P ϭ daily living (5), the geriatric depression Results were analyzed using the SPSS ver- 0.06). TNF and leptin did not correlate. scale (6), functional independence mea- sion 14 package. IR and NIR subjects sures (7), the comorbidity scale of Charl- were compared using ␹2 tests for categor- CONCLUSIONS — IRS is increas- son (8,9), the timed get-up-and-go test ical variables and Mann-Whitney U tests ingly recognized as a harbinger of classical (10), the timed 6-m walk (11), the 6-min for continuous variables. Rank-order cor- disease states, including diabetes and walk (12), smoking history, and alcohol relations (Spearman’s r) were used to inves- metabolic syndrome X. It has been postu- consumption history were assessed at the tigate associations. Statistical significance lated that insulin resistance, diabetes, and time of enrollment. was determined at the 0.05 level and sta- metabolic syndrome X are all associated Residents on any form of medication tistical trend defined as 0.05 Ͻ P Ͻ 0.10. with a proinflammatory state, but it has for the treatment of diabetes or with a fast- been difficult to determine whether the ing blood glucose of Ն110 mg/dl were RESULTS — The mean insulin levels proinflammatory state is a cause or effect excluded from further analysis. One resi- were 280 Ϯ 313 pmol/l (IR group, n ϭ of these conditions. Furthermore, it is un- dent with abnormally high values for C- 16) and 58.4 Ϯ 18.51 pmol/l (NIR group, clear what proinflammatory cytokines are peptide (Ͼ4,500 pmol/l), insulin (2,280 n ϭ 53). Table 1 compares resident char- most likely to be elevated early in the pmol/l), and total amylin (Ͼ4,500 acteristics. There were no differences in course of these conditions, especially in pmol/l) was excluded from the study. The age, morphometric measures, or func- insulin resistance. All these conditions 69 remaining nondiabetic, euglycemic tional, nutritional, or mental status, al- also become increasingly common with residents were classified as insulin resis- though there was a statistical trend (P ϭ aging. Here, we examined functional, tant (IR) (fasting serum insulin Ͼ100 0.09) toward an elevated BMI in the IR chemical, metabolic, and inflammatory pmol/l, n ϭ 16) or non-IR (NIR) (fasting group. Table 2 shows that glucose was status in a relatively healthy and uniform serum insulin Ͻ100 pmol/l, n ϭ 53). slightly elevated in the NIR group (P Ͻ population of elderly women. We ex- Two inflammatory scores were calcu- 0.05), but there were no differences in the cluded any person taking diabetes medi- lated by first determining the median val- other general chemistries. Renal function cations and used strict criteria for ues for a parameter and then calculating as measured by creatinine or the blood euglycemia (i.e., fasting glucose Ͻ110 the number of subjects who had values urea nitrogen–to–creatinine ratio was mg/dl). higher than the median. The first inflam- similar between the two groups. These residents had significant im- matory score included the parameters of C-peptide (P Ͻ 0.05), amylin (P Ͻ pairments in their abilities to care for C-reactive protein, TNF, IL-6, and IL-1␤, 0.001), and leptin (P Ͻ 0.01) were all themselves, requiring nursing home

2370 DIABETES CARE, VOLUME 30, NUMBER 9, SEPTEMBER 2007 Banks and Associates

Table 2—Laboratory values in IR and NIR subjects have IRS, fasting hyperglycemia, or frank diabetes, as it is generally assumed that IR group NIR group P residents this debilitated would have near-universal insulin resistance (13). n 16 53 Those with IR did not differ from the Insulin (pmol/l) 280.1 Ϯ 312.5 58.4 Ϯ 18.5 Ͻ0.001 NIR group in age, height, weight, smok- Na (mmol/l) 139.5 Ϯ 2.9 140.3 Ϯ 2.8 0.45 ing history, drinking history, or nutri- K (mmol/l) 4.2 Ϯ 0.3 4.4 Ϯ 0.4 0.14 tional risk, as measured by the mini– Cl (mmol/l) 104.7 Ϯ 3.3 104.1 Ϯ 4.6 0.67 nutritional assessment tool. Importantly, Ϯ Ϯ CO2 (mmol/l) 25.8 2.2 27.1 3.0 0.08 the two groups did not differ in BMI, al- BUN (mg/dl) 16.1 Ϯ 3.9 19.1 Ϯ 7.0 0.27 though the IR group had a BMI that was Creatinine (ng/dl) 0.9 Ϯ 0.2 0.9 Ϯ 0.3 0.79 ϳ6% greater (P ϭ 0.09). The groups did Ca (mg/dl) 9.2 Ϯ 0.5 10.9 Ϯ 12.6 0.89 not differ on measures of severe illness as Cholesterol (mg/dl) 189.6 Ϯ 37.3 193.4 Ϯ 46.9* 0.86 measured by the comorbidity scale of Agap (mmol/l) 13.3 Ϯ 3.7 13.6 Ϯ 3.6 0.63 Charlson, a medical record–based inven- BUN/creatinine 18.5 Ϯ 7.2 20.6 Ϯ 6.2 0.19 tory of 19 conditions associated with in- PreA (mg/dl) 21.7 Ϯ 4.9 21.6 Ϯ 6.0 0.72 patient mortality. Functional status did Albumin (g/dl) 3.4 Ϯ 0.4 3.4 Ϯ 0.4* 0.95 not differ between the two groups, as Transferrin (mg/dl) 230.1 Ϯ 33.7 229.3 Ϯ 50.3 0.90 measured by four different measures: the C-reactive protein (mg/dl) 0.5 Ϯ 0.4 0.9 Ϯ 1.9 0.40 functional index measure, the activities of WBC (k/mm3) 6.8 Ϯ 2.3 6.1 Ϯ 1.7 0.38 daily living, the 6-m walk, and the 6-min RBC (m/mm3) 4.2 Ϯ 0.4 4.0 Ϯ 0.6 0.19 walk. This lack of difference suggests that HGB (g/dl) 13.0 Ϯ 0.9 12.2 Ϯ 1.4 0.11 the muscle wasting, sarcopenia, and HCT (%) 38.7 Ϯ 3.0 36.7 Ϯ 4.2 0.13 frailty associated with insulin resistance PLT (k/mm3) 238.8 Ϯ 48.0 249.9 Ϯ 68.9 0.78 (13) is confounded by aging and other MPV (fl) 8.6 Ϯ 1.1* 9.0 Ϯ 1.1† 0.10 associated factors and is more likely a later MCV (fl) 91.5 Ϯ 3.1 91.5 Ϯ 6.2 0.96 association with IRS. Levels of depres- MCH (pg) 30.7 Ϯ 1.1 30.6 Ϯ 2.4 0.80 sion, as measured by the geriatric depres- MCHC (g/dl) 33.5 Ϯ 0.8 33.4 Ϯ 0.9 0.59 sion scale, or dementia, as measured by RWD (%) 14.1 Ϯ 1.3 14.9 Ϯ 2.2 0.19 the mini–mental status examination, Neutro (%) 55.8 Ϯ 9.6 58.7 Ϯ 9.1 0.17 were similar between the groups. Balance Lymph (%) 31.4 Ϯ 7.9 28.2 Ϯ 8.3 0.18 and risk of falls as measured by the timed Mono (%) 8.1 Ϯ 1.8 9.0 Ϯ 2.6 0.21 get-up-and-go test was also similar. This Eos (%) 4.4 Ϯ 3.1 3.5 Ϯ 3.0 0.17 lack of difference shows that any deleteri- Baso (%) 0.7 Ϯ 0.3* 0.6 Ϯ 0.4† 0.11 ous effects of insulin resistance had not IL-1␤ (pg/ml) 12.9 Ϯ 24.0 6.2 Ϯ 21.4 0.44 yet manifested themselves in common in- IL-2 (pg/ml) 47.8 Ϯ 109.2 10.4 Ϯ 20.9 0.49 dexes of health status. IL-1r␣ (pg/ml) 333.3 Ϯ 478.3 187.3 Ϯ 226.2 0.15 These groups did not differ in the in- IL-4 (pg/ml) 563.7 Ϯ 1,379.2 318.0 Ϯ 861.6 0.10 dicators of general inflammation, includ- IL-6 (pg/ml) 186.2 Ϯ 355.8 67.8 Ϯ 118.3 0.94 ing C-reactive protein, prealbumin, IL-8 (pg/ml) 76.7 Ϯ 105.0 43.8 Ϯ 66.6 0.30 albumin, transferrin, or immune cell lev- IL-10 (pg/ml) 30.9 Ϯ 66.0 21.2 Ϯ 82.4 0.18 els, or in hemoglobin or hematocrit. They IL-1␣ (pg/ml) 310.4 Ϯ 852.6 218.1 Ϯ 560.2 0.89 also did not differ in cholesterol levels. Interferon-␥ (pg/ml) 7.3 Ϯ 14.4 2.6 Ϯ 3.4 0.66 Hyperlipidemia is a hallmark of metabolic TNF-␣ (pg/ml) 15.3 Ϯ 6.6 10.8 Ϯ 4.6 0.008 syndrome X, whereas low cholesterol Eotaxin (pg/ml) 173.3 Ϯ 105.7 151.4 Ϯ 58.9 0.87 (Ͻ160 mg/dl) is used in geriatrics as an MIP-1a (pg/ml) 39.7 Ϯ 70.8 49.4 Ϯ 87.3 0.32 index of malnutrition. Overall, therefore, RANTES (pg/ml) 39,264.0 Ϯ 17,184.7 39,952.8 Ϯ 16,926.6 0.92 insulin resistance had little impact on im- Leptin (pmol/l) 2,828.2 Ϯ 1,961 1,675.6 Ϯ 1,722.5 0.005 mediate general health, giving us the op- T-Amylin (pmol/l) 34.6 Ϯ 58.1 8.0 Ϯ 7.2 Ͻ0.001 portunity to examine an early stage of IRS. C-peptide (pmol/l) 721.0 Ϯ 378.5 628.0 Ϯ 714.2 0.03 Although the study included only Adiponectin (ng/ml) 29,796.2 Ϯ 17,293.5 40,151 Ϯ 21,827.2 0.08 those with a fasting blood glucose of Resistin (pg/ml) 13,801.1 Ϯ 3,493.8 15,918.6 Ϯ 13,289.1 0.53 Ͻ110 mg/dl, glucose was higher (P Ͻ Glucose (mg/dl) 94.1 Ϯ 8.4 87.9 Ϯ 8.2 0.02 0.05) in the IR (94.2 mg/dl) versus the Data are means Ϯ SD. *Based on n Ϫ 1. †Based on n Ϫ 2. Agap, anion gap; Baso, basophil; BUN, blood urea NIR group (87.9 mg/dl). The IR group nitrogen; Eos, eosinophil; Lymph, lymphocyte; MIP, macrophage inflammatory protein; Mono, monocytes; fasting glucose was well within normal Neotro, neutrophil; PreA, prealbumin; T-amylin, total amylin. limits and only 7% higher than in the NIR group. Nevertheless, it is a clear indica- placement. They were, however, the pairments. Mean insulin levels were 280 tion that even this early stage of IRS is healthiest members of the nursing home pmol/l in the IR group and 58.4 pmol/l in associated with impairments in glucose population, having on average mild to the NIR group. It is significant that the handling. moderate cognitive and functional im- majority of recruited residents did not In contrast to the lack of differences in

DIABETES CARE, VOLUME 30, NUMBER 9, SEPTEMBER 2007 2371 Insulin resistance syndrome in the elderly

Table 3—Ratios and hyperglycemia, possibly through its ability to regulate proinflammatory cyto- IR group NIR group P kines (14–17). However, adiponectin did not distinguish insulin resistance from n 16 53 non–insulin resistance, being decreased Leptin to BMI 106.3 Ϯ 67.2* 62.4 Ϯ 59.0 0.002 in the IR group by ϳ25% (P ϭ 0.08). TNF-␣ to leptin 0.01 Ϯ 0.02 0.02 Ϯ 0.02 0.11 Evidence for a proinflammatory state Insulin to C-peptide 0.6 Ϯ 0.7 0.1 Ϯ 0.1 Ͻ0.001 was most clearly revealed by a significant Ϯ Ϯ Insulin to leptin 0.2 0.3 0.1 0.2 0.11 (P Ͻ 0.01) elevation in TNF. TNF levels Ϯ Ϯ Ͻ Insulin to BMI 9.2 10.3* 2.4 0.9 0.001 were 50% higher in IR compared with ␣ Ϯ Ϯ Ͻ Insulin to TNF- 17.1 14.2 6.4 3.8 0.001 NIR residents. Arithmetic increases that ␣ Ϯ Ϯ TNF- to BMI 0.5 0.2* 0.4 0.2 0.14 did not reach statistical significance were Ϯ Ϯ Insulin to amylin 35.3 101.9 20.1 21.5 0.87 evident in every other proinflammatory Data are means Ϯ SD. *Based on n Ϫ 1. and anti-inflammatory cytokine measured with the exceptions of MIP-1␣ and routine laboratories, several hormone lev- the NIR group. However, BMI is a cruder RANTES. The lack of statistical signifi- els were elevated in the IR group. As ex- measure of adiposity than serum leptin, as cance in the elevations of some of the cy- pected, insulin C-peptide was elevated BMI does not distinguish between lean tokines may have been caused in part by but only by ϳ15%. Interestingly, the in- and adipose tissue. The leptin-to-BMI ra- the large variance within groups. To fur- sulin/C-peptide level was about fivefold tio was also ϳ70% higher in the IR group ther assess a global trend of cytokine eleva- higher in the IR group, suggesting that in comparison with the NIR group (P Ͻ tion and to negate variance, we computed insulin or C-peptide clearance or degra- 0.005). This suggests that either those in two proinflammatory indexes, both of dation was affected more than secretion. the IR group were hypersecreting leptin which included TNF that assessed eleva- Any difference in insulin clearance cannot per unit of fat mass or that a greater percent tions on nonparametric scoring. How- be ascribed to renal failure, as serum cre- of their body weight was adipose tissue. ever, neither of these summations was atinine did not differ between the IR and The latter seems more likely, considering different between the two populations. NIR groups. Amylin, cosecreted with in- the recent description of the sarcopenic This suggests that even with the power of sulin in a 1:1 molar ratio from the pan- obese. These residents, despite having a including a statistically significant com- creas, was also elevated over fourfold in normal or high BMI, have decreased mus- ponent (TNF), there is no evidence for a the IR population. The ratio of insulin and cle mass, and so a greater proportion of statistically meaningful global cytokine amylin did not differ between the two their BMI is adipose tissue. Such residents increase. This supports TNF as being the groups, suggesting that the two peptides are at an increased risk for mortality and primary cytokine whose elevation is rele- were cleared and secreted in a similar morbidity. The current findings suggest vant in early insulin resistance. fashion. that insulin resistance could be an early The role of serum TNF is further sup- Serum insulin levels were highly cor- component of sacropenic obesity. ported by a strong correlation with serum related with serum leptin levels, consis- The strong associations of leptin and insulin levels. This correlation between tent with insulin resistance being associated leptin/BMI with serum insulin and their TNF and insulin was even stronger than with obesity. As expected from numerous segregation between the NIR and IR that between leptin and insulin. One other studies, leptin also correlated posi- groups suggest a strong association be- source of serum TNF could be from adi- tively with BMI and inversely with adi- tween obesity and insulin resistance. pose tissue. However, serum TNF levels ponectin. Leptin was increased ϳ70% in However, BMI only showed a trend to- did not correlate with serum leptin levels. the IR population (P Ͻ 0.01). This was ward association with insulin (P ϭ 0.06), In some studies, TNF and leptin levels unexpected, given that the average BMI whereas the leptin-to-BMI ratio corre- correlate, presumably because both can for the IR group was only marginally lated better with serum insulin than did be secreted from fat. However, TNF has (6%), and not significantly, higher than in leptin alone. This again suggests that se- sources in addition to adipose tissue, rum levels of leptin are much more sensitive whereas adipose tissue is essentially the at inventorying the metabolic parameters sole source of serum leptin. Therefore, the associated with insulin resistance than is lack of correlation between leptin and BMI. This is further suggested by the find- TNF suggests that the source of elevated ing that the leptin-to-insulin ratio did not TNF was not adipose tissue. Further sup- differ between the IR and NIR groups, porting this were the findings that the even though each of these components TNF-to-BMI ratio did not distinguish be- alone were very different. Thus, leptin tween IR and NIR groups the way leptin- and insulin overlap to some degree in to-BMI ratio did, and the TNF-to-insulin their abilities to distinguish the IR and ratio, unlike the leptin-to-insulin ratio, NIR groups but likely retain independent still distinguished the IR and NIR groups. predictive abilities. Therefore, TNF and leptin behaved very Figure 1—TNF and insulin. A nonlinear rela- In contrast, serum adiponectin levels differently, making it unlikely that adi- tion existed between serum levels of TNF and correlated even more strongly with leptin pose tissue was the source of the elevation serum levels of insulin. **P Ͻ 0.01. ϩ, NIR than did insulin. Adiponectin is secreted in TNF. TNF did show a statistical trend group; E, IR group. from fat and can reverse insulin resistance with C-reactive protein (P ϭ 0.07), sug-

2372 DIABETES CARE, VOLUME 30, NUMBER 9, SEPTEMBER 2007 Banks and Associates gesting that it might have arisen from a 10. Mathias S, Nayak US, Isaacs B: Balance in References elderly patients: the “get-up and go” test. generalized proinflammatory condition. 1. Einhorn D: American College of Endocri- In conclusion, we examined early in- Arch Phys Med Rehabil 67:387–389, 1986 nology position statement on the insulin 11. Nebard-Rothe K, Sobush DC, Bousamra sulin resistance in a population of elderly resistance syndrome: executive summary. M, Haasler GB, Lipchik RJ: Self-selected women living in a long-term care facility Endocr Pract 9 (Suppl. 2):6–8, 2003 walking velocity for functional ambula- before debility ascribable to insulin resis- 2. American College of Endocrinology position in patients with end-stage emphy- tion statement on the insulin resistance tance had arisen. Although no functional sema. J Cardiopulm Rehabil 17:85–91, 2007 syndrome: position statement. Endocr impairments were found between the IR 12. Enright PL, McBurnie MA, Bittner V, Pract 9 (Suppl. 2):9–21, 2003 and NIR groups, we found that other pa- Tracy RP, McNamara R, Newman AB, the rameters were already changing. Despite 3. Folstein MF, Folstein SE, McHugh PR: “Mini-mental state:” a practical method Cardiovascular Health Study: The 6-min exclusion of any person with a serum fast- walk test: a quick measure of functional Ն for grading the cognitive state for the cli- ing glucose 110, a slight, statistically nician. J Psychiatr Res 12:189–192, 1975 status in elderly adults. Chest 23:387– significant increase in serum glucose of 4. Guigoz Y, Lauque S, Vellas BJ: Identifying 398, 2003 7% demonstrated an early glucose intol- the elderly at risk for malnutrition: the 13. Abbatecola AM, Paolisso G: Is there a re- erance in the IR group. Differences in C- mini nutritional assessment. Clin Geriatr lation between insulin resistance and peptide and the insulin–to–C-peptide Med 18:737–757, 2002 frailty syndrome? Curr Pharm Des.In ratio suggested an increased insulin secre- 5. Katz S: Assessing self-maintenace: activi- press tion and a decreased insulin clearance in ties of daily living, mobility and instru- 14. Engeli S, Feldpausch M, Gorzelniak K, Hartwig F, Heintze U, Janke J, Mohlig M, the IR group. Increased leptin and leptin- mental activities of daily-living. J Am Geriatr Soc 31:721–726, 1983 Pfeiffer AF, Luft FC, Sharma AM: Associ- to-BMI ratios in the absence of differences ation between adiponectin and mediators in BMI suggest that the IR group had a 6. Sheikh JI, Yesavage JA: Geriatric depression scale: recent evidence and develop- of inflammation in obese women. Diabe- higher percent of body fat mass and that ment of a shorter version. Clin Gerontol tes 52:942–947, 2003 IR could be related to sarcopenia of obe- 5:165–172, 1986 15. Qi Y, Takahashi N, Hileman SM, Patel, HR, sity. TNF was the one proinflammatory 7. Granger CV, Hamilton BB, Keith RA: Ad- Berg AH, Pajvani UB, Scherer PE, Ahima RS: cytokine statistically elevated in the IR vances in functional assessment for med- Adiponectin acts in the brain to decrease group. The lack of correlation of TNF ical rehabilitation. In Topics in Geriatric body weight. Nat Med 10:524–529, 2004 with BMI or leptin suggests that the TNF Rehabilitation. Lewis CB, Ed. Baltimore, 16. Spranger J, Verma S, Gohring I, Bobbert did not originate from fat. Thus, early in- MD, Aspen Publishing, 1986, T, Seifert J, Sindler AL, Pfeiffer A, Hileman sulin resistance is associated with eleva- 8. O’Connell RL, Lim LL: Utility of the SM, Tschop M, Banks WA: Adiponectin tions of leptin, even after correction of Charlson comorbidity index computed does not cross the blood-brain barrier but modifies cytokine expression of brain BMI and a proinflammatory state, as evi- from routinely collected hospital dis- charge diagnosis codes. Methods Inf Med endothelial cells. Diabetes 55:141–147, denced by elevated levels of serum TNF. 39:7–11, 2000 2006 9. Charlson ME, Pompei P, Ales KL, Mac- 17. Wulster-Radcliffe MC, Ajuwon KM, Kenzie CR: A method of classifying Wang J, Christian JA, Spurlock ME: Adi- Acknowledgments— This study was sup- prognostic comorbidity in longitudinal ponectin differentially regulates cytokines ported by a Veterans Affairs merit review and studies: development and validation. J in porcine macrophages. Biochem Biophys National Institutes of Health Grant NS050547. Chronic Dis 40:373–383, 1987 Res Commun 316:924–929, 2004

DIABETES CARE, VOLUME 30, NUMBER 9, SEPTEMBER 2007 2373