Carbohydrate Metabolism in the Elderly

Carbohydrate metabolism in the elderly

D Elahi1,2* and DC Muller2

1Massachusetts General Hospital, Geriatric Research Laboratory, Boston MA 02114, USA; and 2Gerontology Research Center, National Institute on Aging, Laboratory of Clinical Investigation, Baltimore MD 21224, USA

In this short review we summarize the effect of age on glucose homeostasis. The concept of decreased glucose tolerance with increasing age is introduced, followed by evidence for this phenomenon. Speci®cally we review the evidence for changes in fasting glucose as a function of age and the effect of age on HbA1c. The role of age on hepatic glucose production and glucose uptake is then discussed in detail and we review the evidence that supports the concept that with advancing age hepatic glucose sensitivity to insulin is unaltered. We then review the large evidence for the role of age on the purported decrease in peripheral tissue sensitivity to insulin and conclude that the issue is unsettled. The decrease attributed to age is no longer signi®cant when confounders are controlled for, the largest being obesity. We next present evidence that b-cell sensitivity to glucose remains intact with aging. A review of age-related disorders due to hyperglycemia and confounding effects on the relationships of age and glucose tolerance is presented next. Finally we present new evidence that when the revised criteria for the diagnosis of type 2 diabetics as proposed by the American Diabetes Association and WHO are used, a greater percentage of the elderly will not be diagnosed. We conclude that, although glucose intolerance increases with aging, which is accompanied with other disorders, it is possible to ameliorate this effect with alteration of diet and exercise. Descriptors: aging; carbohydrate metabolism; glucose tolerance; diabetes European Journal of Clinical Nutrition (2000) 54, Suppl 3, S112±S120

Introduction US have abnormal glucose tolerance as de®ned by the American Diabetes Association. The reduction in whole-body carbohydrate metabolism in the elderly is one of the hallmarks of the aging process. Substantial evidence has been provided showing that Other evidence of carbohydrate intolerance in the increasing age is associated with decreased glucose toler- elderly ance (Andres, 1971; Broughton & Taylor, 1991; Davidson, 1979; DeFronzo, 1981; Reaven et al, 1989). Figure 1 shows Fasting glucose In the basal state, the decline in carbohydrate metabolism results of a 2 h glucose tolerance test in healthy men from as assessed by the fasting glucose is relatively small. A the Baltimore Longitudinal Study of Aging across the adult review of the literature (Davidson, 1979) had estimated that age span. There is a progressive decline in glucose tolerance from the third decade through the ninth decade of age. fasting glucose levels rise approximately 1 mg=dl per decade. Subsequent studies in populations that covered The 2 h plasma glucose level during an oral glucose the adult age span have produced more ambiguous results. tolerance test rises on average, 5.3 mg=dl per decade and the fasting plasma glucose rises on average, 1 mg=dl per In 263 healthy subjects aged 20 ± 69 y, age was positively related to fasting glucose even after adjustment for differ- decade (Davidson, 1979). This decline in glucose tolerance ences in adiposity (Berger et al, 1978). The Baltimore is also re¯ected in the NHANES III survey on the prevalence of diabetes and impaired fasting glucose and Longitudinal Study of Aging also has shown a positive cross-sectional correlation between age and fasting plasma impaired glucose tolerance in US adults (Harris et al, glucose in both men and women (Muller et al, 1996; 1998). Comparison of the percentage of physician-diag- Shimokata et al, 1991). Zavaroni et al (1986) demonstrated nosed diabetes in middle-aged adults (40 ± 49 y) and elderly an age-related increase in fasting glucose levels in Italian adults ( 75 y) reveals an increase of 3.9 ± 13.2%. Like- factory workers; this was signi®cantly reduced in women wise, the percentage of adults with undiagnosed diabetes after adjustments were made for the presence of other age- (de®ned as a fasting plasma glucose 126 mg=dl) rises from 2.5% to 5.7% and the percentage of adults with related variables. Small increases in fasting glucose levels with age were noted in 4170 men and women living in impaired fasting glucose (de®ned as a fasting plasma suburban California (0.7 mg=dl=decade in men; glucose of 110 ± 125 mg=dl) rises from 7.1% to 14.1%. Therefore, approximately a third of the elderly adults in the 2.0 mg=dl=decade in women) (Barrett-Connor, 1980). In a retirement community in California in subjects aged 47 ± 90 y, Maneatis et al (1982) found no signi®cant changes *Correspondence: D Elahi, Massachusetts General Hospital, Geriatric Research Laboratory, GRB SB 0015C, 55 Fruit Street, Boston, MA 02114, with age in fasting glucose levels in men but did observe a USA. small signi®cant positive correlation in women. However in E-mail: [email protected] other studies, age had little if any, effect on fasting glucose. Carbohydrate metabolism D Elahi and DC Muller S113

Figure 2 Effect of age and BMI on HbA1c in men. Values are means Æ s.e.m.; BMI 20 (d); 20.1 ± 23 (s); 23.1 ± 26 (j);>26 (n). (b) P < 0.01 vs group with BMI 20. (C) P < 0.01 vs group one decade younger. (CC) P < 0.05 vs group one decade younger.

male Japanese workers (Hashimoto et al, 1995) aged 20 ± 59 was similar in both lean and obese (Figure 2). While there was no effect of gender on HbA1c levels in French men and women (Simon et al, 1989), a cross-sectional survey of Chinese men and women (Yang et al, 1997) found that women had lower levels before the age of 55 but Figure 1 Time course of plasma glucose level after oral glucose at older ages the levels were almost identical. administration in men of the Baltimore Longitudinal Study on Aging. Twenty minute values are presented by age decades where 2 represents values for 20 ± 29 year old subjects and 3 represents values for 30 ± 39 year Effect of age on hepatic glucose production and glucose old subjects, etc. uptake During the post-prandial state plasma glucose levels are Age was not independently related to fasting glucose in a maintained stable by coordinated balance between hepatic population survey of 740 Danes of the Second Generation glucose production and glucose uptake by peripheral tis- Fredericia Study (Vestbo et al, 1996). In a study of 710 sues (primary muscle). There are numerous techniques to healthy individuals, fasting glucose levels increased by assess the contribution of these two regulators of glucose only 8% and 6% over seven decades in men and women, homeostasis. However, many confounders, which are exa- respectively (Colman et al, 1995). cerbated in the elderly, make it dif®cult to delineate accurately the exact contribution of these two major reg- HbA1c ulators of glucose homeostasis. The two most commonly Glucose is known to bind irreversibly to the NH2 terminus used methods to quantitate regulation of glucose home- of the b-chains of globin protein in hemoglobin during the ostasis are the Min Mod (frequently sampled intravenous life span of the erythrocytes. It is this glycosylated species glucose tolerance test) (Bergman et al, 1987) and the that appears as HbA1c when hemoglobin is analyzed glucose clamp techniques (DeFronzo et al, 1979). Although by ion-exchange chromatography (Nuttall, 1998). HbA1c they both have their advantages and disadvantages, it is provides an objective means of quantifying the average generally agreed that the most rigorous and reproducible blood glucose concentration present in the weeks before test for examination of glucose homeostasis in the whole sampling. There have been many studies that examined body or a speci®c organ is the glucose clamp methodology. the effect of age on HbA1c levels. Most (Arnetz et al, 1982; The clamp technique has been more commonly used to Dunn et al, 1979; Graf et al, 1978; Hashimoto et al, examine differences between young and old, while fewer 1995; Kilpatrick et al, 1996; Nakashima et al, 1993; investigators have used the Min Mod to examine the issue Simon et al, 1989; Vestbo et al, 1996; Yang et al, 1997) in the elderly. Our own experience has been entirely with but not all studies (Kabadi, 1988; Van Wersch et al, 1991) the clamp technique and since this technique is considered have found elevations in glycohemoglobin levels with the reference method for assessment of glucose home- increasing age in non-diabetic subjects. In several of the ostasis a brief description will be provided. There are above studies (Hashimoto et al, 1995; Vestbo et al, 1996; several variants to this methodology and the most com- Yang et al, 1997), the increase in HbA1c was independent monly used ones are the hyperinsulinemic euglycemic form of other confounding effects on glucose tolerance (see and the hyperglycemic form. In all clamp protocols an below). The age-dependent increase in Hba1c in 7664 antecubital intravenous polyethylene catheter is inserted for

European Journal of Clinical Nutrition Carbohydrate metabolism D Elahi and DC Muller S114 infusion of insulin, and=or 20% dextrose, isotopes. A second polyethylene catheter is inserted in a retrograde fashion into a dorsal hand or wrist vein for obtaining blood samples. The hand is then enclosed in an insulated grounded warming chamber with the air heated to 68C to arterialize the blood. After a temperature equilibration period, at least four samples are obtained at 10 min intervals and immediately assayed to determine basal glucose level. In the euglycemic clamp at time zero an infusion of regular insulin is started in a falling 10 min primed-constant manner for 2 ± 5 h. The usual constant dose is 240 pmolm72 min71 (40 mUm72 min71), but doses as high as 6 nmolm72 min71 (1 Um72 min71) have been used. The usual dose rapidly creates a square wave of hyperinsulinemia at a level of 420 ± 600 pmol=l (70 ± 100 mU=ml). This requires infusion of glucose 4 min after the start of the insulin infusion in order to maintain plasma glucose levels at the level established during the basal period. The glucose infusion is adjusted approximately Figure 3 Dose ± response curves for suppression of hepatic glucose every 5 ± 10 min based on plasma glucose determination output by insulin in young and old participants. at the same frequency. Glucose utilization is slowly increased and reaches a plateau approximately 80 min Dose ± response relationships between plasma insulin level after the start of the insulin infusion. For the usual dose and HGP clearly demonstrate that in normal tolerant (240 pmol), the steady utilization rate is approximately individuals, the liver is exquisitely sensitive to insulin 30 ± 45 mmolkg71 min71 ( 5±8mgkg71 min71). In the and HGP is completely suppressed at insulin levels normal glucose tolerant individual, hepatic glucose produc- well below the commonly used insulin dose tion (HGP) is completely suppressed with this level of (240 pmolm72 min71). Additionally, as shown in Figure hyperinsulinemia, and if the plasma glucose is adequately 3, there is no difference in either the basal HGP or in the maintained, the glucose infusion rate is essentially equal to dose ± response curve of its suppression by insulin between glucose utilization and represents the amount of glucose young and old individuals. HGP is suppressed over 90% in metabolized (M). For insulin resistant states, such as aged individuals with insulin levels of 240 pmol=l (40 ml) obesity and diabetes, HGP is not totally suppressed and (Meneilly et al, 1987). The half-maximal effect is approxi- M must increase by the amount of glucose produced by the mately 150 pmol=l (25 mU=ml). This ®nding is in general liver. Since the hyperinsulinemic state is always the same, agreement with ®ndings of other investigators in both the higher the M, the more sensitive the individual and vice young and elderly volunteers (DeFronzo et al, 1979; Fink versa. et al, 1983; Rizza et al, 1981). In our studies (Meneilly et In the hyperglycemic clamp a square wave of hypergly- al, 1987) we have also observed that with low dose insulin cemia is rapidly produced and maintained for the duration infusions, HGP is more rapidly suppressed in the elderly. of the study. This is accomplished with a falling primed We have attributed this to a delayed suppression of endo- infusion of 20% dextrose, which usually lasts 12 min and genous insulin release in the elderly individuals (as not more than 15. Blood samples are obtained every 2 min assessed with C-peptide levels). Thus, early in the study, for the ®rst 10 min to estimate ®rst phase insulin response the liver is exposed to a higher insulin level in the old than and then every 5 min for the duration of the study. The in the young. The liver is exquisitely sensitive to portal plasma glucose level is maintained at the desired plateau insulin concentration and an increase of 30 ± 60 pmol=l(5± (usually 5.5 mmol=l above basal so that it is still in the U=ml) reduces HGP by 50% (DeFronzo et al, 1983). Thus, physiologic range) by adjustment of the glucose infusion. although total suppression of HGP does not differ with age, This adjustment is similar to that described for the eugly- reduction of HGP by insulin occurs more rapidly in the cemic clamp (ie, based on plasma glucose level and elderly at physiological levels. negative feedback control). In a normal individual, the square wave of hyperglycemia elucidates a biphasic insulin response, where the ®rst phase occurs over a 10 min period, Effect of age on peripheral tissue glucose utilization which is followed by a second phase. The latter increases gradually for at least four hours, despite stable hypergly- Muscle is the primary sink for glucose uptake as adipose cemia. This variant of the clamp assesses b-cell sensitivity tissue is relatively inert and accounts for only 2 ± 3% of to glucose. Furthermore, it is possible to analyze glucose glucose uptake (DeFronzo, 1981). Brain and splanchnic utilization, M, as in the euglycemic clamp. However, it glucose account for another 20 to 55% (DeFronzo, 1988). must be recognized that M in the hyperglycemic clamp is in The latter two sinks for glucose uptake are insulin inde- response to variable endogenous insulin release, whereas M pendent. The majority of reports examining the role of age in the euglycemic clamp is in response to a stable repro- on glucose uptake, utilizing a variety of techniques ducible and exogenous insulin concentration. (clamps, Min Mod, forearm infusion, glucose and insulin and somatostatin) have demonstrated decreased insulin sensitivity during hyperinsulinemia as summarized in the Hepatic glucose production report by the European Group for the Study of Insulin The role of the liver in the maintenance of plasma glucose Resistance (EGIR) (Ferrannini et al, 1996). However, this level has been best evaluated with the euglycemic clamp. decreased insulin sensitivity is not demonstrated for fasting

European Journal of Clinical Nutrition Carbohydrate metabolism D Elahi and DC Muller S115 with other studies where several insulin doses were employed (DeFronzo et al, 1979; Fink et al, 1983). In addition, studies using various techniques, including the hyperglycemic clamp (DeFronzo et al, 1979; Elahi et al, 1993), the forearm glucose uptake technique (Jackson et al, 1982) and Min Mod (Chen et al, 1985) have revealed resistance to insulin-induced glucose disposal in aged volunteers. It is obvious that in addition to age various factors in¯uence insulin sensitivity including fat mass, fat distribution, physical ®tness, dietary composition and genetic factors. When these factors are taken into account, it is not clear that age, per se, still has an effect on peripheral glucose uptake.

The effect of age of b-cell sensitivity A major component of glucose homeostasis is the balance Figure 4 Regression of insulin action against age in the study population between insulin secretion and tissue sensitivity to insulin. (from 20 European Centers) grouped according to sex and obesity (de®ned as BMI>25 kg=m2). All regressions are adjusted by BMI and center. Again there are many reports that either support or refute a Dotted lines indicate regression lines with a slope not signi®cantly decline in b-cell sensitivity to glucose as a function of age. different from zero. Lean females, P < 0.002. However, very few investigators have examined b-cell secretion in a dose ± response manner. We have performed 230 hyperglycemic clamps for two hours at four doses (3.0, insulin levels (Coon et al, 1989; Ferrannini et al, 1991). 5.4, 7.9 and 12.8 mmol=l above basal glucose levels) in The EGIR report is an analysis of euglycemic clamps at young, middle aged and old healthy volunteers (Elahi et al, a single dose of 240 pmolm72 min71, from 20 European 1993). Deconvolution analysis for these studies suggests clinical research centers. In this study, a total of 1146 that insulin secretion is characterized by (1) a spike lasting clamps were performed (776 men and 380 women) for 2 h about one minute (®rst phase), (2) cessation of secretion in individuals with normal glucose tolerance and arterial from 0 to 15 minutes, (3) a step increase at 15 min to a level blood pressure. In addition, anthropometric data were above basal and (4) a gradually increasing secretion from available. Glucose utilization, M, was calculated during 15 ± 120 min. In these dose ± response studies as the hyper- the last 40 min of the study. In univariate analysis, age glycemic level increased, there was a gradual increase in was associated with a signi®cant decrease in insulin action insulin response within each age group. Furthermore, with (P 0.0002) 0.9 mmolkg71 decade of life, ( 0.2 mgkg71 each group, comparisons of insulin responses between decade of life) which was no longer signi®cant when doses were statistically signi®cant both for the early adjusted for BMI (P 0.08). However, BMI was strongly phase insulin response and for each of the succeeding 30 associated with a decrease in insulin actions min periods (30 ± 60, 60 ± 90 and 90 ± 120 min) of the late- (5 mmolkg71 10 kg of body weight, 0.9 mgkg71 10 kg of phase insulin response. However, in neither the early phase, body weight, P < 0.001), which remained signi®cant, even nor the later-phase insulin responses were there any statis- after adjustment for age. Furthermore, when the relation- tically signi®cant differences among the age groups. These ship of glucose utilization, M, was examined as a function results are consistent with the observation of DeFronzo of gender and obesity (BMI less than or equal to 25 or (1979) where hyperglycemic clamps were performed at greater than 25) an age-related decrease in insulin action 6.9 mmol=l above basal in young, middle-aged and old. was demonstrated only in lean women (Figure 4), without a Using the Min Mod technique, two studies (Chen et al, slope change after age 50 (ie, no menopause effect). Thus, 1985; Pacini et al, 1988) have also shown that neither ®rst this large study demonstrates that glucose uptake is not phase nor second phase insulin response differs signi®- altered as a function of age per se at this hyperinsulinemic cantly between young and old. Thus, the general consensus level. is that insulin secretion does not differ in aging. Despite the strength of the EGIR study, the issue of age- related insulin resistance is still controversial, because it has been argued that complete dose ± response curves are Hyperglycemia and age-related disorders necessary to resolve the issue. Several groups have con- The association between hyperglycemia and disease has ducted dose ± response studies as a function of age. The been examined in many clinical and epidemiological stu- studies are rather small (N 50), mainly in men, without a dies. A recent review (Pruess, 1997) has summarized the signi®cant difference in BMI between young and old. One evidence of the toxic effects of hyperglycemia (Table 1). study (Rowe et al, 1983) has examined the effect of age The author makes the case for the theory that glycosylation on glucose utilization over the insulin range of 60 ± is one of the driving forces behind physiological aging. 6000 pmol=l (10 ± 1000 mU=ml). An age-associated Since glycosylation is augmented by increasing the con- decrease in glucose utilization was demonstrated with centration of circulating glucose, even slightly, the glucose preservation of maximal glucose uptake (ie, a shift to the intolerance of the elderly would place them at higher risk right). The half-maximal glucose uptake occurred when for developing microvascular and macrovascular disorders. plasma insulin levels were 324 pmol=l (54 mU=ml) in the Another review that examines the strength of the associa- young and 678 pmol=l (113 mU=ml) in the old. When the tion between hyperglycemia and cardiovascular disease glucose utilization was plotted per kilogram of lean body (Haffner, 1998) presents evidence that dysglycemia is a mass, the relationship remained. This study is in agreement continuous risk factor for CVD. Possible mechanisms

European Journal of Clinical Nutrition Carbohydrate metabolism D Elahi and DC Muller S116 Table 1 Pathogenic consequences of hyperglycemiaa

Microvascular complications Nephropathy Neuropathy Retinopathy Basement membrane thickening Microvascular disease Protein glycosylation Impaired cellular immunity Cell cycle abnormalities Glucose toxicity: carbohydrate metabolism Impaired insulin secretion Impaired insulin sensitivity

aReproduced from Pruess (1997).

Figure 5 Risk factors of the metabolic syndrome. Each line represents a statistically signi®cant association between each risk factor that is found in most populations.

behind this association include alteration of protein kinase activity by glucose, oxidative stress induced by acute hyperglycemia, and the nonenzymatic glycosylation of proteins. These reviews suggest that type 2 diabetes melli- Figure 6 Age and sex differences in fatness, fat distribution, physical tus and physiological aging have somewhat similar etiolo- activity, and glucose tolerance. Two indices of each of these variables are gies and that asymptomatic hyperglycemia is not a benign presented. Fatness: body mass index (BMI) and percentage body fat (%fat; condition. computed from skinfolds); fat distribution: waist ± hip ratio (WHR) and subscapular ± triceps skin-fold thickness ratio (SSF=TSF); physical activity: activity by questionnaire and VO2max; glucose tolerance: fasting and Confounding effects on the relationship of age and 2 h glucose levels. Numbers of subjects in each decade are as follows

glucose tolerance Subset providing VO2max

Even though it is apparent that the elderly have impaired Age group (y) Men Women Men Women carbohydrate metabolism it still remains controversial whether this phenomenon is a result of aging per se or is 18 ± 34 71 60 27 33 35 ± 44 90 51 43 23 due to other age-related variables. The complexity of the 45 ± 54 77 36 23 16 inter-relationships of these variables is illustrated by Figure 55 ± 64 90 51 35 19 5. The lines connecting the different factors represent the 65 ± 74 74 44 26 19 statistically signi®cant associations that exist in the cross- 75 ± 92 65 33 19 15 sectional analyses of most populations. For example, most populations show a signi®cant decrease in glucose tolerance with increasing age (see above). However, there is an increase in adiposity and a decrease in physical activity or it acts through other as yet unidenti®ed associated with age and these factors are also associated with a mechanisms or risk factors. decrease in glucose tolerance. In most studies, aging is The elderly as a group, have more chronic diseases, take acting as a surrogate for mechanisms or risk factors that are more medications, are more obese, and are less physically the true causes underlying the decline in function. As these active. Since all of these factors are known to decrease mechanisms are identi®ed and accounted for, the strength glucose tolerance a proportion of the age-related decline of the effect of age, per se on function is lessened. When may be independent of age. Many studies have tried to age remains as an independent risk factor, either it is quantify the confounding effects of obesity and physical through biological mechanisms that need to be identi®ed inactivity on the age-associated decline in glucose toler-

European Journal of Clinical Nutrition Carbohydrate metabolism D Elahi and DC Muller S117 Table 2 Effect of diet on glucose tolerance

Authors Population n Age range Study type Effects

Lundgren et al (1989) Swedish women 1462 38 ± 60 12 y incidence No effect of diet on glucose of DM intolerance Feskens et al (1991a,b) Dutch men 175 64 ± 87 4 y incidence of IGTa or DM Carbohydrate-rich foods related to and women glucose intolerance; protective effect of ®sh consumption Lombrail et al (1992) French men 366 Ð DM prevalence study Independent effect of alcohol and women consumption Colditz et al (1992) US women 84360 34 ± 59 6 y incidence of DM No effect of diet on glucose intolerance Marshall et al (1994) US men 134 30 ± 74 IGTa to DM in 1 ± 3 y Fat consumption related to incidence and women of DM Feskens et al (1995) Dutch 338 50 ± 69 20 y incidence of DM Positive association of total fat intake and Finnish men and 2-hPGb Mooy et al (1995) Dutch men 2484 50 ± 74 DM prevalence study Positive association with heavy and women alcohol use in women; positive association with intake of total protein, animal protein and polyunsaturated fat in men; inverse association with energy intake in men; inverse association with moderate alcohol use aIGT impaired glucose tolerance. b2-hPG 2 h postload glucose level. ance. The cross-sectional relationships (Shimokata et al, Dietary factors may also confound the relationship of 1991) between age and obesity, physical activity, and age and glucose tolerance. It has been acknowledged that a glucose tolerance in the Baltimore Longitudinal Study of `westernized' diet and the possibility of a `thrifty genotype' Aging are shown in Figure 6. In this population, with aging, may be responsible for an epidemic of type 2 diabetes and men and women showed increasing fatness, a shift in fat cardiovascular disease in many cultures (Zimmet & distribution to the trunk of the body and especially to the Alberti, 1997). The increased adiposity and abdominal fat abdomen rather than the hips, decreasing ®tness, and distribution in these adult populations are postulated as the decreasing glucose tolerance. After statistically adjusting links in the causal chain. However, aging is also associated for the confounding effects of obesity, fat distribution and with a decline in energy intake, a loss of lean body mass physical activity on glucose tolerance the authors conclude and reduced physical activity (Flynn et al, 1992; Hallfrisch that in this population the decline in glucose tolerance from et al, 1990; Hoffman, 1993; Popkin et al, 1992; Roberts early-adult (17 ± 39 y) to middle-age (40 ± 59 y) is entirely et al, 1995; SENECA, 1996). Whether these age-related explained by the secondary in¯uences of fatness and changes in diet (particularly a decrease in dietary carbohy- ®tness. However, the decline from mid-life to old age drate) alter carbohydrate metabolism has been examined. (60 ± 92 y) was still in¯uenced by chronological age. In a Increasing carbohydrate intake did not ameliorate the population of Italian factory workers (Zavaroni et al, 1986) glucose intolerance in the elderly (Chen et al, 1987; age-related environmental factors (obesity, physical activ- Hughes et al, 1995). Population surveys that looked at ity, diabetogenic drug use) accounted for a large portion of the effect of dietary intake on glucose tolerance are the decline in glucose tolerance in women, but not in men. presented in Table 2. Two studies in women found no Maneatis et al (1982) found that after adjustment for effect of diet on glucose tolerance (Colditz et al, 1992; obesity and activity there was an independent effect of Lundgren et al, 1989). Three studies found that increased age on the decline in glucose tolerance in women. No effect fat intake was related to a worsening in glucose tolerance of age on glucose tolerance was noted for men. (Freskens et al, 1995; Marshall et al, 1994; Moody et al, In a study of 134 Italian women aged 18 ± 71 years 1995). Two studies found that alcohol intake was signi®- (Zamboni et al, 1997), after adjustment for visceral fat, no cantly related to glucose tolerance. In French men and signi®cant differences in glucose tolerance were found women there was a signi®cant positive independent asso- across ®ve age groups. In addition to population studies, ciation of alcohol intake and the prevalence of type 2 comparisons of the glucose tolerance of groups of young diabetes (Lombrail et al, 1992). The Hoorn Study (Mooy and old individuals have been made. Older master athletes et al, 1995) found a signi®cant inverse association of have similar glucose tolerance when compared to younger moderate alcohol intake and glucose intolerance in men athletic and sedentary individuals (Seals et al, 1984). and women and positive association of heavy alcohol Healthy, nonobese and physically active older subjects intake and glucose intolerance in women. Finally, one did not have signi®cantly different glucose tolerance than study found both a protective effect of ®sh consumption their younger counterparts in a study of Pacini et al (1988). (Feskens et al, 1991b) and a detrimental effect of the In another study of healthy, nonobese physically active consumption of carbohydrate-rich foods on the incidence individuals (Reaven et al (1989), the age-related decline in of glucose intolerance (Feskens et al, 1991a). glucose tolerance was reported to be modest in magnitude in the older subjects. Two other studies (Coon et al, 1992; Diagnostic criteria and the elderly Kohrt et al, 1993) have also suggested that abdominal obesity rather than age per se accounts for the insulin Recently the Expert Committee on the Diagnosis and resistance and glucose intolerance in the elderly. Classi®cation of Diabetes Mellitus (1997) of the American

European Journal of Clinical Nutrition Carbohydrate metabolism D Elahi and DC Muller S118 Diabetes Association has provided new criteria that sim- impaired status classi®cation of 60 ± 74 y old Mexican- plify the method of detection of undiagnosed diabetes. For American men. In general, this discrepancy between the clinical diagnosis in asymptomatic individuals the ADA WHO and ADA classi®cations becomes more pronounced recommends that diabetes be de®ned as a fasting plasma with increasing age for all demographic groups. Although glucose value greater than or equal to 7.0 mmol=l the use of fasting plasma glucose alone for diabetes (126 mg=dl). This recommendation revises the 1980 ± diagnosis may simplify testing, the WHO criteria would 1985 World Health Organization (WHO) diagnostic criteria identify a much greater percentage of elderly subjects with (World Health Organization, 1985) for diabetes, which diabetes or impaired glucose tolerance. relied on both fasting glucose and the 2 h oral glucose tolerance test. A fasting plasma glucose 7.8 mmol=l (140 mg=dl) and=or a 2 h plasma glucose value 11.1 Conclusions mmol=l (200 mg=dl) was diagnostic of diabetes by the Age has profound effects on glucose homeostasis. Using WHO criteria. The ADA criteria also provided recommen- the ADA criteria and the NHANES III data, approximately dations for two other diagnostic classes. Impaired fasting 18.8% of the US population between the ages of 60 and 74 glucose (IFG) is de®ned as a fasting plasma glucose are diabetic. Another 14% have impaired fasting glucose. 6.1 mmol=l (110 mg=dl) and < 7.0 mmol=l (126mg=dl); The prevalence of diabetes in the US remains high even normal fasting glucose is de®ned as a fasting plasma though the prevalence of hypertension and the incidence of glucose < 6.1 mmol=l. The WHO criteria had de®ned and mortality from heart disease and stroke are declining. A impaired glucose tolerance (IGT) as a fasting plasma sedentary lifestyle and easy access to a high caloric diet are glucose < 7.8 mmol=l and a 2 h plasma glucose characteristic of Western civilization. The elderly popula- 7.8 mmol=l and < 11.1 mmol=l. The NHANES III tion has had a prolonged exposure to these harmful envir- survey (Harris et al, 1998) used the fasting plasma glucose onmental conditions and therefore on average has a higher and the 2 h glucose value from the oral glucose tolerance percentage of body weight as fat and a more detrimental fat test data to compare the percentage of the population distribution. These age-related changes in body composi- meeting the ADA and WHO diagnostic criteria by age, tion are responsible for a large portion of the decline in sex and ethnic groups. Table 3 gives a summary by age glucose tolerance seen in the elderly. They are also poten- groups of the percentages of the population with undiag- tially modi®able through a prudent combination of diet and nosed diabetes and impaired status (either IFG or IGT) for exercise. both the WHO and ADA diagnostic criteria. For all demographic groups, the WHO criteria identi®ed a larger number of subjects than the ADA criteria; the exceptions References were the undiagnosed diabetes classi®cation for the 40 ± Andres R (1971): Aging and diabetes. Med. Clin. N. Am. 55, 835 ± 845. 49 y old groups for non-Hispanic White men, non-Hispanic Arnetz B, Kallner A & Theorell T (1982): The in¯uence of aging on Black women, and Mexican-American men and the hemoglobin A1c (HbA1c). J. Gerontol. 37, 648 ± 650.

Table 3 Comparison of the percentages of the US population 40 ± 74 y of age (NHANES III) with undiagnosed diabetes or impaired statusa by 1980 ± 1985 WHO and 1997 ADA diagnostic criteriab

Age (y)

Race Sex Criteria 40 ± 49 50 ± 59 60 ± 74

Undiagnosed diabetes Non-Hispanic White Men WHO 2.3 6.8 11.8 ADA 2.9 3.5 8.2 Women WHO 1.5 4.3 10.4 ADA 1.3 4.4 4.3 Non-Hispanic Black Men WHO 5.8 10.4 9.7 ADA 4.3 3.0 6.6 Women WHO 2.9 8.1 7.8 ADA 3.7 8.5 8.5 Mexican-American Men WHO 5.8 16.4 14.3 ADA 6.8 12.9 6.3 Women WHO 6.6 16.0 9.0 ADA 4.9 7.5 3.5 Impaired status Non-Hispanic White Men WHO 11.7 11.4 19.5 ADA 10.2 9.1 15.6 Women WHO 10.5 15.0 22.1 ADA 3.2 6.4 12.5 Non-Hispanic Black Men WHO 9.4 15.7 19.4 ADA 6.7 9.3 15.4 Women WHO 9.4 12.9 16.2 ADA 7.2 10.5 9.8 Mexican-American Men WHO 18.6 13.0 22.5 ADA 15.6 8.6 24.5 Women WHO 15.6 26.0 24.9 ADA 7.9 8.6 10.1

aImpaired status impaired glucose tolerance for WHO and impaired fasting glucose for ADA. bData from Harris et al (1998).

European Journal of Clinical Nutrition Carbohydrate metabolism D Elahi and DC Muller S119 Barrett-Connor E (1980): Factors associated with the distribution of fasting Haffner S (1998): The importance of hyperglycemia in the nonfasting state plasma glucose in an adult community. Am. J. Epidemiol. 112, 518 ± to the development of cardiovascular disease. Endocr. Rev. 19, 583 ± 523. 592. Berger D, Crowther R, Floyd J, Pek S & Fajans S (1978): Effect of age on Hallfrisch J, Muller D, Drinkwater D, Tobin J & Andres R (1990): fasting plasma levels of pancreatic hormones in man. J. Clin. Endocri- Continuing diet trends in men: the Baltimore Longitudinal Study of nol. Metab. 47, 1183 ± 1189. Aging (1961 ± 1987). J. Gerontol. 45, M186 ± M191. Bergman RN, Prager R, Volund A & Olefsky JM (1987): Equivalence of Harris MH, Flegal KM, Cowie CC, Eberhardt MS, Goldstein DE, Little the insulin sensitivity index in man derived by the minimal model RR, Wiedmeyer H-M & Byrd-Holt DD (1998): Prevalence of diabetes, method and the euglycemic glucose clamp. J. Clin. Invest. 79, 790 ± impaired fasting glucose, and impaired glucose tolerance in U.S. adults. 800. Diabetes Care 21, 518 ± 524. Broughton DL & Taylor R (1991): Review: deterioration of glucose Hashimoto Y, Futamura A & Ikushima M (1995): Effect of aging on tolerance with age: The role of insulin resistance. Age Ageing 20, HbA1c in a working male Japanese population. Diabetes Care 18, 221 ± 225. 1337 ± 1340. Chen M, Bergman RN, Pacini G & Porte DJ (1985): Pathogenesis of age- Hoffman N (1993): Diet in the elderly. Needs and risks. Med. Clin. N. Am. related glucose intolerance in man: insulin resistance and decreased b- 77, 745 ± 756. cell function. J. Clin. Endocrinol. Metab. 60, 13 ± 20. Hughes V, Fiatarone M, Fielding R, Ferrara C, Elahi D & Evans W (1995): Chen M, Halter JB & Porte DJr (1987): The role of dietary carbohydrate in Long-term effects of high-carbohydrate diet and exercise on insulin the decreased glucose tolerance of the elderly. J. Am. Geriatr. Soc. 35, action in older subjects with impaired glucose tolerance. Am. J. Clin. 417 ± 424. Nutr. 62, 426 ± 433. Colditz G, Manson J, Stampfer M, Rosner B, Willett W & Speizer F Jackson R, Blix P, Matthews J, Hamling J, Din B, Brown D, Belin J, (1992): Diet and risk of clinical diabetes in women. Am. J. Clin. Nutr. Rubenstein A & Nabarro J (1982): In¯uence of ageing on glucose 55, 1018 ± 1023. homeostasis. J. Clin. Endocrinol. Metab. 55, 840 ± 848. Colman E, Toth M, Katzel L, Fonong T, Gardner A & Poehlman E (1995): Kabadi U (1988): Glycosylation of proteins. Lack of in¯uence of aging. Body fatness and waist circumference are independent predictors of the Diabetes Care 11, 429 ± 432. age-associated increase in fasting insulin levels in healthy men and Kilpatrick E, Dominiczak M & Small M (1996): The effects of ageing on women. Int. J. Obes. 19, 798 ± 803. glycation and the interpretation of glycaemic control in type 2 diabetes. Coon P, Bleecker E, Drinkwater D, Meyers D & Goldberg A (1989): QJM 89, 307 ± 312. Effects of body composition and exercise capacity on glucose tolerance, Kohrt WM, Kirwan JP, Staten MA, Bourey RE, King DS & Holloszy JO insulin, and lipoprotein lipids in healthy older men: a cross-sectional (1993): Insulin resistance in aging is related to abdominal obesity. and longitudinal intervention study. Metabolism 38, 1201 ± 1209. Diabetes 42, 273 ± 281. Coon PJ, Rogus EM, Drinkwater D, Muller DC & Goldberg A (1992): Lombrail P, Lang T, Durrieu A, Bernard J, Betouigt H, Dupouy M, Font D, Role of body fat distribution in the decline in insulin sensitivity and Viel E, Parant C & Vexiau P (1992): Alcohol: an underscored risk glucose tolerance with age. J. Clin. Endocrinol. Metab. 75, 1125 ± 1132. factor for diabetes mellitus. Eur. J. Med. 1, 324 ± 328. Davidson M (1979): The effect of aging on carbohydrate metabolism: a Lundgren H, Bengtsson C, Blohme G, Isaksson B, Lapidus L, Lenner RA, review of the English literature and a practical approach to the diagnosis Saaek A & Winther E (1989): Diatary habits and incidence of of diabetes mellitus in the elderly. Metabolism 28, 688 ± 705. noninsulin-dependent diabetes mellitus in a population study of DeFronzo R (1979): Glucose intolerance and aging. Evidence for tissue women in Gothenburg, Sweden. Am. J. Clin. Nutr. 49, 708 ± 712. insensitivity to insulin. Diabetes 28, 1095 ± 1101. Maneatis T, Condie R & Reaven G (1982): Effect of age on plasma DeFronzo R (1981): Glucose intolerance and aging. Diabetes Care 4, glucose and insulin responses to a test mixed meal. J. Am. Geriatr. Soc. 493 ± 501. 30, 178 ± 182. DeFronzo RA (1988): Lilly lecture 1987. The triumvirate: beta-cell, Marshall J, Hoag S, Shetterly S & Hamman R (1994): Dietary fat predicts muscle, liver. A collusion responsible for NIDDM. Diabetes 37, conversion from impaired glucose tolerance to NIDDM. Diabetes Care 667 ± 687. 17, 50 ± 56. DeFronzo R, Tobin J & Andres R (1979): Glucose clamp technique: a Meneilly GS, Minaker KL, Elahi D & Rowe JW (1987): Insulin action in method for quantifying insulin secretion and resistance. Am. J. Physiol. aging man: evidence for tissue-speci®c differences at low physiological 237, E214 ± E223. insulin levels. J. Gerontol. 42, 196 ± 201. DeFronzo RA, Ferrannini E, Hendler R, Felig P & Wahren J (1983): Mooy J, Grootenhuis P, De Vries H, Valkenburg H, Bouter L, Kostense P Regulation of splanchnic and peripheral glucose uptake by insulin and & Heine R (1995): Prevalence and determinants of glucose intolerance hyperglycemia in man. Diabetes 32, 35 ± 45. in a Dutch Caucasian population. The Hoorn Study. Diabetes Care 18, Dunn P, Cole R, Soeldner J & Gleason R (1979): Reproducibility of 1270 ± 1273. hemoglobin Aic and sensitivity to various degrees of glucose intoler- Muller D, Elahi D, Tobin J & Andres R (1996): Insulin response during the ance. Ann. Intern. Med. 91, 390 ± 396. oral glucose tolerance test: the role of age, sex, body fat and the pattern Elahi D, Muller DC, McAloon-Dyke M, Tobin JD & Andres R (1993): The of fat distribution. Aging 8, 13 ± 21. effect of age on insulin response and glucose utilization during four Nakashima K, Nishizaki O & Andoh Y (1993): Acceleration of hemoglo- hyperglycemic plateaus. Exp. Gerontol. 28, 393 ± 409. bin glycation with aging. Clin. Chim. Acta 215, 111 ± 118. Ferrannini E, Haffner SM, Mitchell BD & Stern MP (1991): Hyperinsu- Nuttall FQ (1998): Comparison of percent total GHb with percent HbA1c linaemia: the key feature of a cardiovascular and metabolic syndrome. in people with and without known diabetes. Diabetes Care 21, 1475 ± Diabetologia 34, 416 ± 422. 1480. Ferrannini E, Vichi S, Beck-Nielsen H, Laakso M, Paolisso G & Smith U Pacini G, Beccaro VF, Cobelli C & Crepaldi G (1988): Insulin sensitivity (1996): Insulin action and age. Diabetes 45, 947 ± 953. and beta-cell responsivity are not decreased in elderly subjects with Feskens E, Bowles C & Kromhout D (1991a): Carbohydrate intake and normal OGTT. J. Am. Geriatr. Soc. 36, 317 ± 323. body mass index in relation to the risk of glucose intolerance in an Popkin B, Haines P & Patterson R (1992): Dietary changes in older elderly population. Am. J. Clin. Nutr. 54, 136 ± 140. Americans, 1977 ± 1987. Am. J. Clin. Nutr. 55, 823 ± 830. Feskens E, Bowles C & Kromhout D (1991b): Inverse association between Pruess H (1997): Effects of glucose=insulin perturbations on aging and ®sh intake and risk of glucose intolerance in normoglycemic elderly chronic disorders of aging: the evidence. J. Am. Coll. Nutr. 16, 397 ± men and women. Diabetes Care 14, 935 ± 940. 403. Feskens E, Virtanen S, RaÈsaÈnen L, Tuomilehto J, StengaÊrd J, Pekkanen J, Reaven GM, Chen N, Hollenbeck C & Chen YDI (1989): Effect of age on Nissinen A & Kromhout D (1995): Dietary factors determining diabetes glucose tolerance and glucose uptake in healthy individuals. J. Am. and impaired glucose tolerance. A 20-year follow-up of the Finnish and Geriatr. Soc. 37, 735 ± 740. Dutch cohorts of the Seven Countries Study. Diabetes Care 18, 1104 ± Rizza RA, Mandarino LJ & Gerich JE (1981): Dose ± response character- 1112. istics for effects of insulin on production and utilization of glucose in Fink RI, Kolterman OC, Grif®n J & Olefsky JM (1983): Mechanisms of man. Am. J. Physiol 240, E630 ± E639. insulin resistance in aging. J. Clin. Invest. 71, 1523 ± 1535. Roberts S, Fuss P, Heyman M & Young V (1995): In¯uence of age on Flynn M, Nolph G, Baker A & Krause G (1992): Aging in humans: a energy requirements. Am. J. Clin. Nutr. 62, 1053S ± 1058S. continuous 20-year study of physiologic and dietary parameters. J. Am. Rowe JW, Minaker KL, Pallotta JA & Flier JS (1983): Characterization of Coll. Nutr. 11, 660 ± 672. the insulin resistance of aging. J. Clin. Invest. 71, 1581 ± 1587. Graf R, Halter J & Porte DJ (1978): Glycosylated hemoglobin in normal Seals D, Hagberg J, Allen W, Hurley B, Dalsky G, Ehsani A & Holloszy J subjects and subjects with maturity-onset diabetes. Evidence for a (1984): Glucose tolerance in young and older athletes and sedentary saturable system in man. Diabetes 27, 834 ± 839. men. J. Appl. Physiol. 56, 1521 ± 1525.

European Journal of Clinical Nutrition Carbohydrate metabolism D Elahi and DC Muller S120 SENECA (1996): Longitudinal changes in the intake of energy World Health Organization (1985): Diabetes Mellitus: Report of a WHO and macronutrients of elderly Europeans. Eur. J. Clin. Nutr. 50, study group. Technical Report Series, no. 727. Geneva: WHO. S67 ± S76. Yang Y-C, Lu F-W, Wu J-S & Chang C-J (1997): Age and sex effects on Shimokata H, Muller DC, Fleg JL, Sorkin J, Ziemba AW & Andres R HbA1c. A study in a healthy Chinese population. Diabetes Care 20, (1991): Age as an independent determinant of glucose tolerance. 988 ± 991. Diabetes 40, 44 ± 51. Zamboni M, Armellini F, Harris T, Turcato E, Micciolo R, Bergamo- Simon D, Senan C, Garnier P, Saint-Paul M & Papoz L (1989): Epide- Andreis I & Bosello O (1997): Effects of age on body fat distribution miological features of glycated haemoglobin A1c-distribution in a and cardiovascular risk factors in women. Am. J. Clin. Nutr. 66, 111 ± healthy population. Diabetologia 32, 864 ± 869. 115. The Expert Committee on the Diagnosis and Classi®cation of Diabetes Zavaroni I, Dall'Aglio E, Brushi F, Bonora E, Alpi O, Pezzarossa A & Mellitus (1997): Report of the expert committee on the diagnosis and Butturini U (1986): Effect of age and environmental factors on glucose classi®cation of diabetes mellitus. Diabetes Care 20, 1183 ± 1197. tolerance and insulin secretion in a worker population. J. Am. Geriatr. Van Wersch J, Westerhuis L & Venekamp W (1991): HbA1c and serum Soc. 34, 271 ± 275. fructosamine in diabetic patients: relationship to age, clotting and Zimmet P & Alberti K (1997): The changing face of macrovascular ®brinolysis parameters and urinary microalbumin excretion. Clin. disease in non-insulin-dependent diabetes mellitus: an epidemic in Chim. Acta 201, 99 ± 104. progress. Lancet 350,1±4. Vestbo E, Damsgaard E, Frùland A & Mogensen C (1996): Clinical features in persons with a family history of diabetes compared to controls (The Second Generation Frederica Study). J. Int. Med. 240, 381 ± 387.

European Journal of Clinical Nutrition