The Metabolic Syndrome: Time for a Critical Appraisal Joint Statement from the American Diabetes Association and the European Association for the Study of Diabetes

Reviews/Commentaries/ADA Statements ADA STATEMENT

The Metabolic Syndrome: Time for a Critical Appraisal Joint statement from the American Diabetes Association and the European Association for the Study of Diabetes

1 3 RICHARD KAHN, PHD ELE FERRANNINI, MD seemed clearly related to type 2 diabetes 2 4 JOHN BUSE, MD, PHD MICHAEL STERN, MD (15). This risk factor clustering, and its association with insulin resistance, led investigators to propose the existence of a unique pathophysiological condition, The term “metabolic syndrome” refers to a clustering of specific cardiovascular disease (CVD) called the “metabolic” (1–3) or “insulin risk factors whose underlying pathophysiology is thought to be related to insulin resistance. resistance” (11) syndrome. This concept Since the term is widely used in research and clinical practice, we undertook an extensive review was unified and extended with the land- of the literature in relation to the syndrome’s definition, underlying pathogenesis, and associa- mark publication of Reaven’s 1988 tion with CVD and to the goals and impact of treatment. While there is no question that certain CVD risk factors are prone to cluster, we found that the metabolic syndrome has been impre- Banting Medal award lecture (16). Reaven cisely defined, there is a lack of certainty regarding its pathogenesis, and there is considerable postulated that insulin resistance and its doubt regarding its value as a CVD risk marker. Our analysis indicates that too much critically compensatory hyperinsulinemia predis- important information is missing to warrant its designation as a “syndrome.” Until much needed posed patients to hypertension, hyperlip- research is completed, clinicians should evaluate and treat all CVD risk factors without regard to idemia, and diabetes and thus was the whether a patient meets the criteria for diagnosis of the “metabolic syndrome.” underlying cause of much CVD. Although obesity was not included in Reaven’s pri- Diabetes Care 28:2289–2304, 2005 mary list of disorders caused by insulin resistance, he acknowledged that it, too, was correlated with insulin resistance or or most of the 20th century, cardio- was first described, most notably the si- hyperinsulinemia, and that the obvious vascular disease (CVD) was identi- multaneous presence of obesity, type 2 “treatment” for what he termed “syn- F fied as the major cause of morbidity diabetes, hyperlipidemia, and hyperten- drome X” was weight maintenance (or and mortality in the developed world. sion (1–3). Although insulin resistance weight loss) and physical activity. During this period there was considerable (i.e., resistance to insulin-stimulated glu- Reaven’s seminal paper was followed effort to understand the underlying biol- cose uptake) as a feature of type 2 diabetes by many studies documenting the cluster- ogy of the disease and to identify the con- was first described many years earlier (4), ing of CVD risk factors and their relation- tributing risk factors. As risk factors were hyperinsulinemia was also found to be a ship to insulin resistance (17–25). identified, it became apparent that more key feature of type 2 diabetes (5,6), as well Indeed, since Reaven’s publication in than one were often present in the same as hyperlipidemia (7–9), obesity (10– 1988, a recent Medline search for articles individual. Toward the end of the cen- 13), and hypertension (12–14). In addi- using the key words “syndrome X” or “in- tury, the clustering of CVD risk factors tion, a cluster of heart disease risk factors sulin resistance syndrome” or “metabolic ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● syndrome” (conducted 28.01.2005) identified 4,646 citations, with 3,948 From the 1American Diabetes Association, Alexandria, Virginia; the 2Division of Endocrinology and of General Medicine & Clinical Epidemiology, University of North Carolina School of Medicine, Chapel Hill, studies performed on human subjects. North Carolina; the 3Department of Internal Medicine, University of Pisa School of Medicine, Pisa, Italy; and The term “metabolic syndrome” has the 4Division of Clinical Epidemiology, Department of Medicine, University of Texas Health Science Center now taken hold in the medical literature. at San Antonio, San Antonio, Texas. It has been defined and institutionalized, Address correspondence and reprint requests to Richard Kahn, PhD, American Diabetes Association, 1701 N. Beauregard St., Alexandria, VA 22311. E-mail: [email protected]. principally by the World Health Organi- This statement was reviewed and approved by the Professional Practice Committee of the American zation (WHO) (26) and the Third Report Diabetes Association and by an ad hoc committee of the European Association for the Study of Diabetes of the National Cholesterol Education (members: H. Beck-Nielsen [Odense, Denmark], K. Borch-Johnsen [Gentofte, Denmark], E.A.M. Gale [Bris- Program’s Adult Treatment Panel (ATP tol, U.K.], L. Groop [Malmo¨, Sweden], H.-U. Ha¨ring [Tu¨ bingen, Germany], R.J. Heine [Amsterdam, the III) (27,28), albeit with different defini- Netherlands], and D.R. Matthews [Oxford, U.K.]). Simultaneous publication: This article is being simultaneously published in 2005 in Diabetes Care and tions. In addition, other organizations Diabetologia by the American Diabetes Association and the European Association for the Study of Diabetes. have developed similar, but again not Abbreviations: ATP III, Third Report of the National Cholesterol Education Program’s Adult Treatment identical, definitions (29,30). The fact Panel; CHD, coronary heart disease; CRP, C-reactive protein; CVD, cardiovascular disease; IFG, impaired that a version of the metabolic syndrome fasting glucose; IGT, impaired glucose tolerance; IMT, intimal-medial thickness; WHO, World Health Organization. has its own ICD-9 code (277.7) also sug- © 2005 by the American Diabetes Association, Inc., and Springer-Verlag. Copying with attribution gests that it is well thought out (31,32). allowed for any noncommercial use of the work. In this review we examine the evi-

DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 2289 Critical appraisal of metabolic syndrome dence for its definition and underlying Table 1—Definitions of the metabolic syn- and specificity of the diagnosis and have pathogenesis, as well as analyzing the ev- drome undoubtedly led some physicians to diag- idence for its association with CVD. We ATP III definition (27,28) nose the metabolic syndrome in patients also discuss the evidence for the goals and Any three or more of the following who would not be labeled as such by impact of treatment. We mainly focused criteria: other providers. The impact of a false- our review on papers addressing the met- 1) Waist circumference Ͼ102 cm in men positive or a false-negative diagnosis has abolic syndrome as defined by ATP III, and Ͼ88 cm in women not been reported. the definition that appears to be used 2) Serum triglycerides Ն1.7 mmol/l Second, it is apparent that the defini- most often in the literature. Because the 3) Blood pressure Ն130/85 mmHg tions of the syndrome differ in the criteria ATP III and WHO definitions are some- 4) HDL cholesterol Ͻ1.0 mmol/l in men listed. For example, microalbuminuria is times used almost interchangeably or and Ͻ1.3 mmol/l in women listed in the WHO criteria but not in the compared with one another, we also ex- 5) Serum glucose Ն6.1 mmol/l (Ն5.6 ATP III; insulin resistance (as measured amined the literature that used the WHO mmol/l may be applicable) under hyperinsulinemic-euglycemic con- criteria. WHO definition (26) ditions) is relevant for WHO but not for For two reasons, we did not consider Diabetes, IFG, IGT, or insulin resistance ATP III. And while only an elevated fast- papers whose focus was on the ability of (assessed by clamp studies) and at least ing plasma glucose value is considered the metabolic syndrome to predict diabe- two of the following criteria: important in the ATP III definition, the tes. First, ample data show that the pres- 1) Waist-to-hip ratio Ͼ0.90 in men or WHO criteria recognize any measure ence of the metabolic syndrome is Ͼ0.85 in women whatsoever of insulin resistance. Al- effective in predicting the future risk of 2) Serum triglycerides Ն1.7 mmol/l or though it would promote better under- diabetes. That association, however, is HDL cholesterol Ͻ0.9 mmol/l in men standing of the justification for the criteria probably due to the fact that the definition and Ͻ1.0 mmol/l in women selected, no review of the clinical evi- of the syndrome includes glucose intoler- 3) Blood pressure Ն140/90 mmHg dence for inclusion or exclusion criteria ance, i.e., impaired fasting glucose (IFG) 4) Urinary albumin excretion rate Ͼ20 for either of the two definitions of the syn- or impaired glucose tolerance (IGT), ␮g/min or albumin-to-creatinine ratio drome has been published to date. themselves powerful predictors of future Ն30 mg/g Third and finally, the originally stated diabetes. Second, the practical use of di- rationale for the criteria is that the syn- agnosing metabolic syndrome has not drome components are associated with centered on its power to predict diabetes, 2) Does the treatment of the metabolic insulin resistance (26,27). But, as dis- but rather on its being a multivariate risk syndrome differ from the treatment of its cussed below, there is considerable doubt factor for CVD. individual components? whether all patients with the metabolic This review argues that the metabolic 3) What additional work should be done syndrome are indeed insulin resistant. syndrome is not nearly as well defined to improve our current knowledge of the More recently, a review of the ATP III def- and characterized as often assumed, and metabolic syndrome? inition (28) broadened the etiological ba- that the notion that it is a useful marker of sis for the syndrome from insulin CVD risk above and beyond the risk asso- Clarity of the existing definition resistance alone to include “obesity and ciated with its individual components is Table 1 shows the ATP III and WHO def- disorders of adipose tissue,” as well as a uncertain. In addition, although certain initions of the metabolic syndrome. These “constellation of independent factors that CVD risk factors undoubtedly occur to- definitions raise three important issues. indicate specific components of the met- gether more often than expected by First, some of the criteria used for defin- abolic syndrome.” However, it remains chance, the underlying pathophysiology ing the syndrome are ambiguous or in- unclear why some factors associated with of the syndrome is unclear. Moreover, the complete (26–28). For example, it is the latter two categories of abnormalities list of risk factors comprising the cluster is unclear whether the blood pressure defi- have not been included in the definition. not grounded by well-defined criteria. nition is systolic pressure Ն130 mmHg The fact that there are cut points for Therefore, this manuscript is intended as and diastolic Ն85 mmHg or whether it is the various risk factors implies that values a cautionary reminder to practitioners, either Ն130 mmHg or Ͼ85 mmHg. It is above the specified thresholds are associ- and as an urgent call for further research. also not defined whether a patient with ated with excess risk, yet the rationale for Our analysis addresses three key ques- hypertension who is now normotensive the specific cut points, as opposed to tions related to the metabolic syndrome: meets the blood pressure criterion, nor is higher or lower values, has never been 1) How clear is the existing definition of it specified how blood pressure should be delineated. Laaka et al. (33), in a study the metabolic syndrome for diagnostic measured, e.g., supine, sitting, mean of conducted in men with the metabolic purposes? two measurements. Given that there is no syndrome, showed that CVD and overall A) How useful is the syndrome defi- widespread agreement on the method for mortality was more consistently increased nition in predicting CVD risk? Do the in- measuring waist circumference, it is also using a waist circumference criterion of dividual components of the syndrome unclear how that factor should be deter- 102 cm rather than 94 cm. Other investi- convey “risk” differently from the syn- mined. Does a patient previously diag- gators (34) found that reducing the drome as a whole? nosed with IFG, but who now has normal threshold for IFG from 6.1 to 5.6 mmol/l B) Is the cluster of symptoms associ- glucose levels because of modest weight did not materially change the hazard ratio ated with the syndrome the result of a com- loss, meet the ATP III glucose criteria? (HR) for risk of coronary heart disease mon underlying pathological process? Such ambiguities affect the sensitivity (CHD), though it did increase the number

2290 DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 Kahn and Associates of individuals identified. Other compo- ing 20–30%), more Mexican-American the metabolic syndrome was 76%, and nents of the syndrome show a continuous men were classified as having the syn- those with the syndrome had HRs for all- relationship with CVD risk (35). Al- drome using the WHO definition, cause and CVD mortality that were no dif- though the thresholds defining the syn- whereas the ATP III criteria classified ferent from those of subjects without the drome are generally derived from other more Mexican-American women. De- syndrome. With frank diabetes of long well-established guidelines, we found no pending on the sex and ethnicity of the duration, the incremental risk attribut- study that systematically examined the populations, the prevalence of metabolic able, for example, to raised triglycerides impact of all the metabolic syndrome syndrome varied up to 24% between the or low HDL, is likely to be “swamped” by thresholds on the risk of CVD, nor did we two definitions. the presence of diabetes itself (48). The find a study that sought to optimize the The question of how to define a syn- fact that subjects were, on average, much positive predictive value of the definition drome (i.e., what factors comprise the older (mean age at baseline: 69 years) by changing the cut points of the risk syndrome) rests in large part on the pur- than in virtually all other studies, and that factors. pose of the construct. A syndrome can be hypertension was highly prevalent in the Some of the criteria (e.g., waist cir- defined on the basis of its ability to predict cohort may have masked the detrimental cumference, HDL) have sex-specific cut (a) future adverse event(s). Such a defini- effects of the syndrome. Another study points, implying that the relationship be- tion implies that the risk associated with was conducted in nondiabetic American tween the risk factor level and outcomes having the syndrome is greater than the Indians (52) and showed a nonsignificant differs between the sexes. However, we sum of its parts, and that the factors in- HR for risk of CVD in those with the syn- found no evidence that warrants estab- cluded have greater predictive power drome. The small number of events that lishing the sex-specific cut points used in than do other combinations. Alterna- occurred during the follow-up period, as the criteria as they relate to CVD risk. It is, tively, if the syndrome purports to iden- well as several other factors reviewed by for example, not known whether the same tify factors related to a unifying the authors, could have contributed to intra-abdominal fat mass carries a differ- pathological process (e.g., insulin resis- their borderline results. Finally, the present risk in men than in women. An anal- tance/hyperinsulinemia), then the defini- ence of the metabolic syndrome in a co- ogous argument can be made regarding tion should include all the factors clearly hort of women with suspected CVD who whether cut points should vary according associated with that underlying patho- had no angiographically significant coro- to race and ethnic groups. physiology, such that there is little ambigu- nary artery disease did not result in an There is ample evidence to show that ity regarding the etiology of the clustering. If increased 4-year risk of CVD, whereas the CVD risk is a function of the criteria cited the etiology is unclear, it becomes much presence of the syndrome resulted in sig- in the definitions of the metabolic syn- more difficult to decide what factors to in- nificantly higher risk in those who were drome, but it is unjustified to assume that clude in the definition, since the word “clus- angiographically positive (53). the optimal predictive power would be ter” itself can be ambiguous. Three studies have examined whether obtained by arbitrary dichotomies. Risk is In the case of the metabolic syn- the difference in prevalence between the a progressive function of, for example, hy- drome, the existing definition attempts to two definitions affects the predictive perglycemia and hypertension and can- bridge both constructs and, as will be power for subsequent development of not simply be regarded as present or pointed out in the following two sections, CVD (33,37,49). Two of these found the absent, depending on whether thresholds does not succeed very well with either. ATP III definition to be a slightly better are exceeded or not. predictor of all-cause and cardiovascular Although the WHO and ATP III def- Relationship between CVD risk and mortality (49) or CHD (37), whereas one initions generally identify the same indi- the metabolic syndrome (33) showed that the WHO definition viduals, important differences have been Many studies have shown that patients di- more consistently predicted CVD and all- found (36,37). Ford and Giles (36) agnosed with the metabolic syndrome, by cause mortality. The fact that all three showed that in the NHANES (National either the ATP III or WHO definition (or studies made modifications to one or both Health And Nutrition Examination Sur- by their modifications), have more prev- of the definitions, and that they included vey), a representative sample of the adult alent CVD or are at greater risk of devel- populations with dissimilar baseline char- U.S. population, about the same propor- oping it (33,38–50). In these studies, the acteristics, precludes drawing any con- tions were identified as having the syn- increased CVD risk in patients with the clusion as to which definition is superior. drome by the WHO or ATP III criteria syndrome ranged from 30 to 400%; this Nonetheless, individuals with meta- (25.1 vs. 23.9%, respectively). However, wide variation is probably due to the pop- bolic syndrome, however defined, have a ϳ15–20% of individuals were classified ulation studied, the precise definition of much higher CVD risk than subjects as having the syndrome by one definition the syndrome adopted, and the length of without the syndrome. This conclusion is but not the other, with equal discordance. follow-up. not surprising, since the individual com- Meigs et al. (37) determined the preva- There are three notable exceptions to ponents of the syndrome have long been lence of the syndrome, defined by ATP III the large body of evidence documenting known to be major cardiovascular risk or WHO criteria, in a population of non- the adverse impact of the metabolic syn- factors (54–59). Thus when they occur Hispanic whites and Mexican-American drome. One is a study by Bruno et al. (51) simultaneously, it is logical that adverse subjects in San Antonio and in subjects conducted in 1,565 elderly diabetic sub- outcomes should be more likely (60–62). participating in the Framingham Off- jects from the Italian town of Casale Mon- ATP III uses the term metabolic syn- spring Study. Although the syndrome ferrato, who were followed for a median drome to imply that certain risk factors was common in these populations (affect- of 8 years. At baseline, the prevalence of are associated with each other, and that

DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 2291 Critical appraisal of metabolic syndrome insulin resistance is the primary cause tion studies examined the relationship be- and consistent inverse association be- (27,28). They identify six components of tween CRP, the metabolic syndrome, and tween adiponectin and both insulin resis- the metabolic syndrome as “underlying,” incident cardiovascular events (68,73, tance and inflammation (70,82,83). In “major,” and “emerging” CVD risk factors 74). In all three, CRP was a strong inde- addition, adiponectin is also inversely as- (28). However, some risk factors associ- pendent predictor of events, and its sociated with other CVD risk factors such ated with insulin resistance in each of predictive value was equal to that of the as blood pressure, LDL cholesterol, and those categories are not included in the metabolic syndrome. In the two studies triglycerides (84,85). Moreover, several definition of the syndrome. For example, (68,73) that dichotomized CRP levels studies have shown adiponectin to be a physical inactivity is omitted as an under- (above and below 3.0 mg/dl), the age- strong (inverse) independent risk factor lying risk factor, while obesity is in- adjusted relative risk of future events was for CVD (86–89). cluded. Family history, sex, and age are no different in subjects with high CRP but Several other molecules have also major CVD risk factors that do not enter without the metabolic syndrome than it been found to be closely associated with into the definition, but hypertension is in- was in subjects with low CRP and with the insulin resistance, metabolic syndrome cluded. Some emerging risk factors asso- metabolic syndrome. However, in sub- risk factors, and the risk of CVD. These ciated with insulin resistance, e.g., certain jects with high CRP levels plus the meta- include plasminogen activator inhibitor proinflammatory and prothrombotic bolic syndrome, the relative risk of events (90–92) and fibrinogen (91–94). markers, are not included, but elevated virtually doubled from that found with All told, therefore, many candidate triglycerides and glucose intolerance are. either parameter alone, indicating that markers could be included in the meta- Interestingly, although the latter four CRP might be a valuable addition to the bolic syndrome. In combination with were designated “metabolic risk factors” definition of the syndrome. other markers related to CVD and insulin and a “component of the metabolic syn- Rutter et al. (74) also found that CRP resistance, more research may lead to a drome,” only elevated triglycerides and and the metabolic syndrome were inde- clearer understanding of the etiology of glucose intolerance are included in the of- pendent risk factors, but in contrast to the the syndrome and hence to a definition ficial list of components (28). The lack of two other reports, combining CRP and that has strong (or stronger) CVD predic- any standardized methodology or ratio- metabolic syndrome did not improve the tive value. nale for how the definition was con- predictive value of either used alone. Some investigators have compared structed, or can be modified, hampers its Reilly et al. (75) also found that CRP did the predictive value of the metabolic syn- optimization and utility. not add significantly to the metabolic syndrome with that of the Framingham risk It is not known whether the substitu- drome, but their study did not include prediction model. For example, a recent tion or addition of any other well-known, CVD outcomes. It is unclear why some post hoc analysis of the placebo-treated conventional CVD risk factor(s) would studies show great value when CRP is groups in the 4S (Scandinavian Simvasta- improve the predictive value of the syn- added, while others do not. The discrep- tin Survival Study) and AFCAPS/ drome. In studies demonstrating that ant results have not, however, deterred TexCAPS (Air Force/Texas Coronary metabolic syndrome was associated with some investigators from advocating that Atherosclerosis Prevention Study) trials higher CVD risk (33,38–50), this excess CRP be included in the definition of the (47) showed that the increased event rate risk remained after adjustment for other metabolic syndrome (76). in subjects with the metabolic syndrome conventional risk factors. This would sug- There is also an association between remained significant after adjustment for gest that if other risk factors are included other markers of inflammation and insu- the Framingham 10-year risk score in the definition, the predictive value of lin resistance/hyperinsulinemia (70,72), (which uses both dichotomized and con- the syndrome may improve. However, we as well as inflammation and obesity (77– tinuous variables), suggesting that the found no study that examined the impact 79), leading some investigators to con- syndrome carries risk not captured by of substituting another CVD risk factor clude that inflammation is integrally Framingham risk scoring. It should be for one already included in the definition. related to the components of the meta- noted, however, that this analysis omitted The issue of whether the risk factors act bolic syndrome (76,77,80). CRP is also diabetes (or any other measure of glucose synergistically has also not been analyzed. strongly associated with adipose-derived intolerance) from the metabolic syndrome Conversely, there are many studies cytokines including interleukin-6 and tu- definition, thereby requiring patients to suggesting that relatively new indexes re- mor necrosis factor ␣ (81), and is more meet three of the remaining four factors to lated to both insulin resistance and CVD likely to be elevated in obese insulin- qualify and removing what may be a critical may also be useful predictive tools (or resistant, but not obese insulin-sensitive, criterion from the definition. This modifica- useful additions to the syndrome defini- subjects (71). Because obesity (particu- tion may have biased their findings. Also, tion). Since it is now well accepted that larly in the visceral compartment) is asso- they dichotomized the Framingham score inflammation plays a major role in athero- ciated with insulin resistance, and these (i.e., Ͼ20% risk vs. Յ20%), so a precise genesis (63), it is not surprising that adipose-derived inflammatory markers determination of the predictive ability of markers of inflammation might be used to have been linked to dyslipidemia, hyper- Framingham versus metabolic syndrome predict CVD events. One such marker, tension, and insulin action (70,72), there could not be determined. C-reactive protein (CRP), has been stud- is a heightened interest in markers from In contrast, Wilson et al. (95), using ied in great detail, and has been found to adipose tissue that are predictive of CVD data from Framingham, found no advan- be an independent CVD risk factor (64– (81). tage in risk assessment above the 68) and an independent marker of insulin One such marker is adiponectin. It is Framingham algorithm (i.e., age, sex, resistance (69–72). Three large popula- now well established that there is a strong smoking, blood pressure, total choles-

2292 DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 Kahn and Associates terol, diabetes, HDL cholesterol) when followed for many years. There are stud- (IFG/IGT/diabetes) in the syndrome defi- some of the unique metabolic syndrome ies, however, that partially address this nition was demonstrated by Malik et al. factors (obesity, triglycerides) were added issue. Malik et al. (48) reported that, com- (48) in their study on NHANES II partic- or substituted. Further, when taking an pared with individuals with no risk fac- ipants. They observed that diabetes alone elevated blood glucose level out of the tors, those with one to two syndrome conveyed a much greater risk of CHD metabolic syndrome definition, the 10- factors had an HR of 2.1 for CHD mortal- (HR ϭ 5), CVD (3.6), or overall mortality year risk for CHD did not achieve the ity and 3.5 if they had the full syndrome (2.1) than the presence of the metabolic threshold for ATP III’s CHD risk equiva- (i.e., three to five risk factors). Other in- syndrome (3.5, 2.7, and 1.5, respectively) lent (27), suggesting that glucose intoler- vestigators (34,73,97) also found that the according to definitions that included ance is a critical component contributing risk for CVD increased with the number subjects with and without IFG/IGT/ to the predictive power of the syndrome. of factors present. diabetes. Adding preexisting CVD to dia- Additional evidence for the greater pre- Other studies, using multivariate betes was an even more powerful dictive value of the Framingham scoring analysis, have shown that the individual predictor of mortality (11.3, 7.9, and 2.9, system has been provided by Stern et al. risk factors comprising the syndrome respectively) over the 13-year follow-up (96) in a study of the Hispanic and non- each carried a different odds ratio for pre- period. Similarly, Stern et al. (100) Hispanic whites who participated in the dicting prevalent CHD, incident CHD, or showed that, among patients with preva- San Antonio Heart Study (n ϭ 2,570) and CVD mortality. In addition to hypergly- lent CVD, the excess risk for all-cause and were free of diabetes and CVD at baseline cemia, low HDL cholesterol and hyper- CVD mortality associated with the meta- and followed for 7–8 years. Their analysis tension usually conferred a significantly bolic syndrome was entirely driven by the showed that the Framingham score had greater risk compared with the presence inclusion of diabetes in the definition, and significantly higher sensitivity for predict- of obesity or high triglycerides (44,49), once diabetes was controlled for, the pres- ing events than the presence of the meta- although McNeill et al. (34) found that ence of the metabolic syndrome no longer bolic syndrome, and when used in only an elevated blood pressure and low conferred excess risk. Finally, Hunt et al. combination, the predictive value did not HDL cholesterol were significantly asso- (49) also showed that the presence of IFG improve. In a multivariate analysis for ciated with CHD. Golden et al. (98) as- (Ͼ6.1 mmol/l) alone was a stronger pre- predicting CVD, using a model that incor- sessed carotid intimal-medial thickness dictor of CVD or all-cause mortality in a porated the Framingham equation and (IMT) related to 57 combinations of six general population than either the syn- the metabolic syndrome, the former had factors related to insulin resistance. In drome as a whole or any of its individual an HR of 7.9 (95% CI 5.3–11.7) com- their analysis, 29 of the 57 groupings components. These reports raise the pared with 1.5 (1.0–2.2) for the latter, were associated with excess carotid IMT. question of why glucose intolerance (par- confirming the superiority of the Fra- The difference in excess IMT between in- ticularly diabetes) is included in the defi- mingham equation. Expressed differ- dividuals with two, three, or four factors nition of the metabolic syndrome, since it ently, they found that the metabolic was minimal, but those with five or six appears to account for most, if not all, of syndrome predicts CVD with a sensitivity factors showed an appreciable increase in the CVD predictive value. of 55% and a false-positive rate of 22%, excess IMT. Hypertension and hypertri- Since the metabolic syndrome does whereas the Framingham risk score had a glyceridemia were the two factors that not include all known CVD risk factors, it significantly higher sensitivity (69%) most contributed to the excess IMT. should convey risk independently of when the false-positive rate was held to Taken together, these studies suggest that other conventional risk factors (e.g., LDL, 22%. In another study that compared re- not all combinations that lead to the diag- age, smoking, family history); however, ceiver operating characteristic curves nosis of the syndrome convey equal risk, the proportion of the global CVD risk cap- (which denote the effectiveness of an as- although the actual hierarchy of risk pre- tured by the syndrome is unknown. It say or test), the metabolic syndrome pro- dictability for each of the syndrome com- would be invaluable to know, from a list vided identical risk prediction to that binations remains unknown. of all known CVD risk factors, the hierar- achieved by the Framingham score (34). The studies also illustrate another chy of combinations with the highest pre- In the studies reviewed so far, a per- likely shortcoming of the current ap- dictive value. Then, a true comparison son was diagnosed with the metabolic proach to diagnosing the syndrome. Both between the metabolic syndrome, other syndrome if he or she had any three or the ATP III and WHO definitions weigh models using different risk factors more of the five criteria (Table 1). (To each risk component equally, yet it is (72,73), or perhaps some new combina- meet the WHO definition, three of five clear that some risk factors included in the tion would tell us what is the best CVD possible criteria must be present, one of definition have greater CVD predictive predictive model. which is mandatory.) Thus, there are 16 importance than others. This fact is high- Another important question is the de- possible combinations that will meet the lighted in other algorithms used to pre- gree to which the presence of the syn- ATP III definition of the syndrome and 11 dict CVD risk using regression drome in itself adds to CVD prediction for the WHO definition. Do all these com- coefficients to assign different weights to beyond the contribution of the compo- binations portend the same CVD risk? risk factors (95,99), and it is apparent nent risk factors. In other words, is the This question has not yet been an- from studies that examined the risk of whole greater than the sum of its parts? If swered, and may never be, since an ex- CVD in persons with one or two compo- the syndrome conveyed no additional tremely large population that includes nents of the syndrome versus three or risk beyond its components, then clini- sufficient numbers of people with each more (97). For example, the dispropor- cians would have little reason to treat combination of criteria would have to be tionate impact of glucose intolerance cases of the syndrome rather than ad-

DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 2293 Critical appraisal of metabolic syndrome dressing each risk factor as it was identi- ease (e.g., diabetes, hypertension, pro- (35) analyzed the results of insulin clamp fied. At least five studies address this teinuria, clinical CVD), as well as with far measurements in a large number of non- issue. One is the study by Golden et al. milder forms of the same conditions, it is diabetic and diabetic subjects. An analysis (98) reviewed above, which examined all likely that there is a risk gradient for CVD of the distribution of insulin-mediated possible combinations of six factors re- events among people with the syndrome. glucose disposal, as measured by the lated to insulin resistance. Individuals Thus, the definition will capture a spec- clamp, showed evidence of bimodality, with any four-, five-, or six-component trum of severities, and it is highly likely and the optimal cut point classified 33% groupings had no greater excess IMT than that a person who satisfies the diagnostic of nondiabetic subjects and 93% of dia- the sum of the same factors taken sepa- criteria with risk factor levels just over the betic subjects as insulin resistant. Some rately. The cross-sectional studies by Al- cut point will have a much lower CVD investigators have chosen a cut point exander et al. (44) and Yarnell et al. (101) risk than another individual with the based on the relationship between insulin showed that the impact of the syndrome same combination but higher risk factor resistance and the frequency of adverse on CVD was greatly attenuated in a mul- levels. This problem stands in contrast to outcomes; however, their sample sizes tivariate analysis by controlling for certain the Framingham (95) and UKPDS (U.K. were very small (114,127). of its components, thereby suggesting Prospective Diabetes Study) (99) risk Although many nondiabetic adult that the whole is not greater than its parts. models, in which the spectrum of severi- subjects with a wide range of age and Also, in a prospective study of diabetic ties is weighted, so that it is clear who may body mass are hyperinsulinemic and in- and nondiabetic subjects free of CVD and be at greater or lesser risk. sulin resistant (ϳ50%), ϳ25% are insulin followed for an average of 11 years, the Finally, people with diabetes and resistant but without hyperinsulinemia risk of incident CHD associated with the clinical CVD should be excluded from the and the same proportion are hyperinsu- syndrome was no greater than that ex- definition of metabolic syndrome, since linemic but without insulin resistance plained by the presence of its components their inclusion provides no additional (126). The relationship between insulin (34). Finally, in the secondary analysis of clinically useful information to guide resistance and hyperinsulinemia, re- the prospective WOSCOPS (West of Scot- treatment beyond current guidelines. viewed in detail by Ferrannini and Balkau land Coronary Prevention Study), Satter (126), is complex, and although both pa- et al. (73) showed that the metabolic syn- Does the syndrome reflect a single rameters will capture individuals with the drome was not a significant predictor of underlying pathological process? metabolic syndrome, each makes an inde- CHD when adjusted for its component When the concept of the syndrome was pendent contribution to the clinical find- factors in a multivariate model. Thus, first proposed, insulin resistance and/or ings associated with the syndrome these studies suggest that the syndrome hyperinsulinemia were initially thought (128,129). Thus, hyperinsulinemia and itself conveys no greater information than to be the primary etiological process, insulin resistance each partially identify the sum of its component risk factors. since most subjects with the syndrome different groups of individuals, they each In summary, if the metabolic syn- had one or the other abnormality. Consis- cluster with various CVD risk factors, and drome is a multicomponent risk factor for tent with the primary adverse outcome of individuals with the metabolic syndrome CVD, the components of which may be the metabolic syndrome (CVD), many may have either, both, or none of these more or less strongly linked with insulin studies (11,105–116), but not all (117– “insulin-related” abnormalities. resistance, neither the ATP III nor the 119), have shown that insulin resistance Even though most people who have WHO definitions consider the many or hyperinsulemia is a CVD risk factor. the metabolic syndrome are insulin resis- other similarly related CVD risk factors, Although the reasons for the disparity in tant, as discussed earlier, this is probably such as age, physical activity, or history of results have been debated elsewhere due to the fact that almost all people with CVD events. Some of these and other risk (120–122), it is important to note that the an elevated blood glucose value (the most factors are included in the Framingham measurement of insulin itself, whether to prevalent characteristic among those with algorithm (95), which has been shown to determine the presence of insulin resis- the syndrome) are insulin resistant. Con- be in general a more powerful tool for tance or hyperinsulinemia, is fraught with versely, many studies have shown that predicting future CVD events. However, errors and inconsistency (123,124) and only a minority of nondiabetic individuals even the Framingham risk equation does that the methods used vary considerably with insulin resistance (but who may have not include important CVD risk factors from laboratory to laboratory. Thus, IFG or IGT) will have the metabolic syn- (e.g., previous CVD events, family his- many subjects who are insulin resistant or drome. In a study of 260 nondiabetic, tory), and has been shown to be much less hyperinsulinemic in one institution may overweight/obese individuals, McLaugh- useful than other risk equations in pre- not be classified as such when tested in lin et al. (130) found that 78% of those dicting future CVD events in people with another setting, because the measure- with metabolic syndrome were insulin re- diabetes (102–104). Other newly identi- ment of insulin is not standardized. More- sistant, but only 48% with insulin resis- fied CVD risk factors have been shown to over, insulin-mediated glucose disposal tance had metabolic syndrome. Liao et al. be strongly associated with insulin resis- varies six- to eightfold in apparently (131) reported that 39% of 74 over- tance and CVD, but it is unclear if they healthy, nondiabetic men (125), absolute weight/obese nondiabetic adults were in- should be added to the syndrome and insulin concentrations vary widely sulin resistant, and 31% with insulin given equal or greater weight than the (16,126), and there is no absolute crite- resistance met ATP III criteria. Moreover, current components. rion with which to classify individuals as the ATP III–negative/insulin-resistant in- Because the criteria for the syndrome being insulin resistant or insulin sensitive. dividuals had CVD risk factor profiles that will capture individuals with frank dis- Very recently, however, Stern et al. were significantly worse than the ATP III–

2294 DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 Kahn and Associates negative/insulin-sensitive group, imply- ship between a set of observed variables hyperinsulinemia is itself a cardiovascular ing that many presumably high-risk (in this case the clinical features of the risk factor, many investigators have individuals will be not be identified by metabolic syndrome) and a smaller set of sought to identify ways to better and more screening for metabolic syndrome. Also, unknown underlying variables (e.g., the simply identify persons with insulin resis- they found that the sensitivity, specificity, etiology) termed “factors.” The factors tance apart from the diagnosis of the met- and positive predictive value for predict- ideally represent unique, independent abolic syndrome. Indeed, easier, simpler, ing insulin resistance in nondiabetic indi- domains that have not been directly mea- and at least equally effective ways are now viduals with three or more metabolic sured but give rise to the observed vari- available to identify insulin-resistant sub- syndrome traits were 20, 92, and 50%, ables. Thus an array of CVD risk variables, jects. Laws and Reaven (151) showed that respectively, denoting poor clinical util- occurring more than would be expected a high triglyceride and low HDL choles- ity. Cheal et al. (132) determined that by chance (25), may be related to one un- terol concentration is a strong indicator of 16% of 443 healthy, nondiabetic subjects derlying factor, thereby supporting a sin- insulin resistance, and when expressed as were insulin resistant and/or positive for gle unifying etiology, or they may be a ratio (130), the optimal cut points in metabolic syndrome, with a sensitivity, related to two or more factors, suggesting overweight/obese individuals resulted in specificity, and positive predictive value a relationship between the underlying do- a sensitivity, specificity, and positive pre- for metabolic syndrome as predictor of mains. Alternatively, more than one un- dictive value for insulin resistance of 64, insulin resistance of 46, 93, and 76%, re- derlying factor suggests a more complex 68, and 67%, respectively. The addition spectively. This study also showed that etiology, and often the analysis cannot ac- of extra measurements (i.e., blood glu- very few of the possible three-, four-, or count for all the variability, thereby pro- cose, blood pressure, BMI) was less sensi- five-factor combinations occurred in the viding additional evidence that more than tive (52%) but more specific (85%) in nondiabetic patients classified with the one physiological process underlies the predicting insulin resistance. Thus, both syndrome. expression of the observed correlations. the metabolic syndrome and this abbrevi- As noted above, most investigators Although factor analysis was devel- ated index have only a moderate likeli- use the phrase “insulin resistance ” to de- oped some time ago (139–141), there is hood of identifying the person with scribe the hallmark of the metabolic syn- no standardized methodology and its use insulin resistance. More recently, Stern et drome, even though insulin resistance or and interpretation is often problematic al. (35) collected euglycemic clamp (the hyperinsulinemia may not be present in (142). Nonetheless, factor analysis is an gold standard for measuring insulin resis- subjects with the syndrome. Further- intriguing exploratory, somewhat subjec- tance) data from Ͼ2,000 lean and over- more, the extent to which an elevated risk tive and qualitative approach toward un- weight/obese individuals and used a of CVD is due to insulin resistance itself, derstanding the root cause(s) of the decision tree classification scheme to de- versus isolated hyperinsulinemia, or ver- metabolic syndrome. The results of many velop decision rules for identifying insu- sus some other related factor is still un- of these studies are shown in Fig. 1. lin-resistant individuals based on clear. Some investigators turn to studies As shown, all of the studies (24,33,38, common clinical measurements. In their on the relationship of insulin resistance to 143–149) found that at least two and usu- study, decision rules based on either ho- the etiology of atherosclerosis (133–136) ally three to four factors underlie the meostasis model assessment of insulin re- and to the underlying etiology of type 2 overall correlation between risk variables, sistance and BMI or BMI and family diabetes (137) as evidence that insulin re- even though different factor analysis history of diabetes had sensitivities and sistance is the more important abnormal- methods were used. Shen et al. (150), us- specificities in the range of 80%. Thus, if the ity (138). Unfortunately, we could find no ing “confirmatory factor analysis,” which aim is to identify insulin resistance in either study that has compared insulin resis- provides some advantages and is comple- lean or overweight/obese subjects, there are tance, as measured by sensitive and spe- mentary to exploratory factor analysis, simpler ways to do so than by identifying cific methods (e.g., euglycemic insulin proposed a correlated four-factor model those with the metabolic syndrome. clamp), with fasting insulin levels to that nicely depicts the major factors re- In summary, the attempt to define the determine which variable is a better pre- lated to the syndrome (Fig. 2). Thus, it is metabolic syndrome as the result of a sin- dictor of cardiovascular events in nondi- clear that more than one distinct patho- gle (or even major) unifying pathophysi- abetic individuals. physiological process underlies the clini- ological process, e.g., insulin resistance, is Perhaps most important is the fact cal expression of the syndrome, but problematic. Although insulin resistance that the multitude of reports relating in- insulin resistance/hyperinsulinemia ap- or hyperinsulinemia is clearly an impor- sulin resistance to any risk factor or CVD pears related in some fashion to most. tant feature of the syndrome, many other are all association studies. It may well be Also, and equally important, all the fac- as yet unidentified factors are also impor- that there is a more basic defect that can tors can account for no more than about tant. Insulin resistance may simply be one result in insulin resistance and/or other two-thirds of the total variance observed of many abnormalities linked to a more CVD risk factors. The uncertainty sur- in the clustering, suggesting that the syn- fundamental, truly unifying pathophysi- rounding the causative factor(s) that give drome may be even more complex than ology. Moreover, the definition of the syn- rise to the syndrome has prompted many that inferred by factor analyses. These drome includes risk factors that are only investigators to perform a “factor analysis” studies, therefore, again call into question weakly related to insulin resistance or hy- as an approach toward understanding the the appropriateness of implying that a perinsulinemia (e.g., blood pressure) and fundamental cause of the clustering. Fac- handful of CVD risk factors have a com- excludes others that are closely related tor analysis is a multivariate correlation mon underlying pathophysiology. (e.g., CRP, adiponectin). Finally, al- method that seeks to explain the relation- To the extent that insulin resistance/ though many clinical values are signifi-

DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 2295 rtclapaslo eaoi syndrome metabolic of appraisal Critical 2296 D IABETES C ARE , VOLUME 28, NUMBER ,S 9, EPTEMBER 2005 Figure 1—Results from factor analyses of CVD risk variables. Kahn and Associates

Table 2—Treating the metabolic syndrome

Characteristic Patient A Patient B Patient C Waist circumference (cm) 110 103 114 Triglycerides (mmol/l) 1.62 0.34 1.34 Systolic/diastolic blood pressure (mmHg) 170/95 135/90 125/80 HDL (mmol/l) 1.06 1.68 1.29 Fasting plasma glucose (mmol/l) 5.28 6.1 7.22 LDL (mmol/l) 4.65 1.81 1.94 Other Patient smokes; None; taking no Patient had previous taking no drugs drugs MI 4 years ago; taking a ␤ blocker and aspirin All three patients are 50 years old, white males, with no symptoms of CVD and no family history of diabetes, CHD, or stroke. They present for a routine physical examination. Based on the findings above, what factor(s) should be treated and what is the goal of therapy? cantly associated with insulin resistance/ derlying pathological process, it must be factors is related to abnormalities in vis- hyperinsulinemia, the strength of their remembered that the clustering of CVD ceral adipose tissue (152) or an altered association (which has not exceeded a risk factors has been well documented, inflammatory state; other factors may be correlation coefficient of 0.7 and is usu- and thus it is likely (but not assured) that associated with the cluster because they ally 0.3–0.6) is not particularly there is some underlying etiology. It may relate indirectly to one of its components. impressive. be that insulin resistance is simply a risk Although the studies reviewed above factor not unlike other metabolic syn- Treatment of patients who have the question the hypothesis that insulin resis- drome components, and that the under- metabolic syndrome tance/hyperinsulinemia is the major un- lying etiology for some of the syndrome When a person is identified as having the metabolic syndrome, it is not always clear what should be the treatment of choice or the goals of therapy. In a post hoc analysis of the 4S, Pyorala et al. (153) found that simvastatin reduced CVD events to the same degree in nondiabetic patients with or without the metabolic syndrome. We found no other controlled trial results ex- amining the value of a specific pharmacological therapy on patients with the metabolic syndrome. It should again be

Table 3—Summary of concerns regarding the metabolic syndrome 1) Criteria are ambiguous or incomplete. Rationale for thresholds are ill defined. 2) Value of including diabetes in the definition is questionable. 3) Insulin resistance as the unifying etiology is uncertain. 4) No clear basis for including/excluding other CVD risk factors. 5) CVD risk value is variable and dependent on the specific risk factors present. 6) The CVD risk associated with the “syndrome” appears to be no greater than the sum of its parts. 7) Treatment of the syndrome is no different than the treatment for each of its components. Figure 2—Factor structure of the metabolic syndrome. Adapted with permission from Shen et al. 8) The medical value of diagnosing the (150). syndrome is unclear.

DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 2297 Critical appraisal of metabolic syndrome remembered that the current definitions therefore, should neither rely on, nor re- syndrome may divert attention away from of the syndrome capture many people quire a diagnosis of, metabolic syndrome addressing risk factors that are present. with frank disease (e.g., diabetes, hyper- to prescribe and encourage what is now a tension, clinical CVD), as well as those fundamental tenet of medicine—weight Further research needed who have milder conditions or “normal” maintenance (or reduction), exercise, and In the preceding sections, we identified values that, while qualifying them for the a healthy meal plan. many unanswered questions related to diagnosis of the syndrome, are not high The conundrum of treating the meta- the metabolic syndrome. Many are unre- enough to warrant specific therapy. Thus, bolic syndrome is illustrated in the case solved fundamental issues that raise con- it is important to distinguish the approach studies shown in Table 2. Patient A is siderable doubt about the construct itself. and value of treating various metabolic obese and has severe hypertension; the All this should prompt an aggressive syndrome combinations. However, no likely treatment is lifestyle modification research agenda, and based on what is studies have examined the value of tailor- (exercise and weight loss) to include highlighted by this review as missing, ing the treatment algorithm to the partic- smoking cessation counseling, an antihy- should give pause to those in medical ular combination of criteria that resulted pertensive drug, and aspirin. Patient B is practice or in industry. Importantly, there in the diagnosis of the syndrome. obese, has prehypertension and pre- have yet to be any controlled randomized Many of the syndrome characteristics diabetes; his likely treatment is aggressive trials, or systematic, prospective, longitu- are acknowledged to be closely related to lifestyle modification and aspirin. Patient dinal studies that carefully document the insulin resistance or hyperinsulinemia C is also obese, has diabetes and a history clinical value of treating a disease label and their correlates. Yet at the same time it of acute myocardial infarction; his treat- now being given out to a huge number of is unknown whether treating “insulin re- ment would also be aggressive lifestyle our population. sistance” itself would be of value in pre- modification with perhaps a glucose- At the very least, we suggest that the venting CVD in all, or a subset, of lowering agent and aspirin. following is urgently needed: metabolic syndrome patients. Although Does it matter that only patient B has 1) A critical analysis of how the syndrome some studies suggest that the newer insu- the metabolic syndrome (by ATP III crite- is defined. Are all risk factors equally im- lin-sensitizing agents (i.e., thiazo- ria)? If that were immediately apparent, portant? Do some combinations (of two, lidinediones) improve glycemic control, would the treatment change? Who will three, or four factors) portend greater reduce CVD risk factors, and generally re- have the next CVD event, patient A, B, or CVD risk than others? sult in a beneficial CVD profile (154– C? We submit that at this time, the diag- 2) A definition of the syndrome, in which 161), at the time of writing no controlled nosis of the metabolic syndrome itself, or variables have defined lower and upper studies have shown that thiazolidinedio- the lack of it, adds virtually nothing to the cut points or that uses continuous vari- nes reduce CVD events even in the setting treatment of one or more CVD risk factors ables in a multivariate score system (e.g., of diabetes, although one major trial that in a given patient. If, however, it were Framingham/UKPDS risk engine). will help address this issue will be re- known that insulin resistance causes CVD 3) An evidence-based analysis assessing ported shortly (i.e., the PROactive [Pro- and there was a sensitive way to measure the rationale and value of adding (or re- spective Pioglitazone Clinical Trial in and treat it, or if we knew the relative risk placing) other CVD risk factors (e.g., age, Macrovascular Events] study); studies us- of CVD among various configurations of CRP, family history, a direct measure of ing metformin are equivocal. Since thia- risk factors, or if we knew that a combi- insulin resistance) to the definition. zolidinediones affect a wide variety of nation of risk variables found at border- 4) An assessment of CVD risk in subjects parameters, even favorable trial results line disease levels elevates one’s CVD risk with combinations of intermediate phe- will not prove that reducing insulin resis- considerably, then knowing that a patient notypes only (e.g., IFG/IGT, mildly ele- tance itself is the critical factor. Moreover, has the metabolic syndrome could also be vated triglycerides, blood pressure 120– even if positive trial evidence were to useful. But at present, none of those cri- 140 mmHg) and who have, or don’t have, emerge relatively soon, other important teria have been fulfilled. insulin resistance or hyperinsulinemia. issues have been identified (162), such as These case studies also raise addi- 5) An aggressive research agenda to iden- how will insulin resistance be measured, tional concerns. For patients with type 2 tify the underlying cause(s) of the CVD what is the cut point to begin treatment, diabetes, a comprehensive set of evi- risk factor clustering. and is the target population only patients dence-based prevention services should similar to those included in the trials? be provided (e.g., regular eye and foot ex- Conclusion Thus, our knowledge base is such that we ams). In patients who are diagnosed with As a construct that denotes risk factor cannot yet contemplate drug treatment the metabolic syndrome and who also clustering, the metabolic syndrome has for insulin resistance, let alone the meta- have diabetes, the importance and treat- been a useful paradigm. That is, it draws bolic syndrome. ment of the diabetes may, in contrast to attention to the fact that some CVD risk Other modifiers of insulin resistance what might be expected, take a backseat factors tend to cluster in patients so pre- include weight reduction and exercise, relative to the syndrome, with patients or disposed. The teaching point implied by and they have been identified as key ele- their doctors possibly neglecting or over- the term, and explicitly stated by ATP III, ments in the treatment of the metabolic looking essential factors of diabetes man- is that the identification of one of the risk syndrome (27,31,32). But they are also agement. Also, for patients who fail to variables in a patient should prompt a key elements in the treatment of all com- meet the necessary number of criteria to search for others. ponents of the syndrome when they occur diagnose the syndrome (e.g., only two of At the time these relationships were in isolation (27,163,164). Clinicians, the factors are present), the absence of the first documented, the advent of a phrase

2298 DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 Kahn and Associates to capture the prevalence of the clustering at cut points indicative of frank disease Insulin action and resistance in obesity was likely a helpful reminder to clini- (e.g., blood pressure Ͼ140/90 mmHg, and non-insulin-dependent type II dia- cians, and certainly served to open a wide fasting plasma glucose Ն7.0 mmol/l), betes mellitus. Am J Physiol 243:E15– avenue of research into its etiology and treatment should correspond to estab- E30, 1982 impact. Now, however, it has taken on lished guidelines (27,163,168). 11. Stern MP, Haffner SM: Body fat distribution and hyperinsulinemia as risk factors meaning and import greater than is justi- 3) Providers should avoid labeling for diabetes and cardiovascular disease. fied by our current knowledge. Indeed, patients with the term “metabolic syn- Arteriosclerosis 6:123–130, 1986 only recently the International Diabetes drome,” as this might create the impres- 12. Modan M, Halkin H, Almog S, Lusky A, Federation developed yet a new defini- sion that the metabolic syndrome denotes Eshkol A, Shefi M, Shitrit A, Fuchs Z: tion, which suggests that the key element a greater risk than its components, or that Hyperinsulinemia: a link between hy- is central obesity (165). In their paper, it is more serious than other CVD risk pertension, obesity, and glucose intoler- however, no data were reviewed indicat- factors, or that the underlying pathophys- ance. J Clin Invest 75:809–817, 1985 ing the impact or benefit derived from iology is clear. 13. Haffner SM, Fong D, Hazuda HP, Pugh their new definition. 4) All CVD risk factors should be individ- JA, Patterson JK: Hyperinsulinemia, up- As shown in this review (Table 3), ually and aggressively treated. per body adiposity, and cardiovascular risk factors in non-diabetics. Metabolism there is much fundamental, clinically im- 5) Until randomized controlled trials have 37:338–345, 1988 portant, and critically missing informa- been completed, there is no appropriate 14. Ferrannini E, Buzzigoli G, Bonadonna R, tion about the metabolic syndrome to pharmacological treatment for the meta- Giorico MA, Oleggini M, Graziadei L, warrant a more serious examination of bolic syndrome, nor should it be assumed Pedrinelli R, Brandi L, Bevilacqua S: In- whether medical science is doing any that pharmacological therapy to reduce sulin resistance in essential hyperten- good by drawing attention to (166) and insulin resistance will be beneficial to pa- sion. N Engl J Med 317:350–357, 1987 labeling millions of people with (36) a tients with the metabolic syndrome. 15. Wingard DL, Barrett-Connor E, Criqui presumed disease that does not stand on MH, Suarez L: Clustering of heart dis- firm ground. In particular, patients with ease risk factors in diabetic compared to diabetes or clinical CVD should be ex- non-diabetic adults. Am J Epidemiol 117: References 19–26, 1983 cluded from the case definition of meta- 1. Avogaro P, Crepaldi G, Enzi G, Tiengo 16. Reaven GM: Role of insulin resistance in bolic syndrome, as they provide no A: Associazione di iperlidemia, diabete human disease. Diabetes 37:1595–1607, additional understanding of risk or treat- mellito e obesita di medio grado. Acto 1988 ment recommendations that are other- Diabetol Lat 4:36–41, 1967 17. Zavaroni I, Bonora E, Pagliara M, wise not currently recommended. 2. Haller H: Epidemiology and associated Dall’Aglio E, Luchetti L, Buonnano G, Medical science usually defines a syn- risk factors of hyperlipoproteinemia. Z Bonati PA, Bergonzani M, Gnudi L, Pas- drome as an “aggregate of symptoms and Gesamte Inn Med 32:124–128, 1977 seri M, et al.: Risk factors for coronary signs associated with any morbid process, 3. Singer P: Diagnosis of primary hyperli- artery disease in healthy persons with and constituting together the picture of poproteinemias. Z Gesamte Inn Med 32: hyperinsulinemia and normal glucose the disease” (167). The specific signs and 129–133, 1977 tolerance. N Engl J Med 320:502–506, 4. Himsworth H: Diabetes mellitus: a dif- 1989 symptoms are usually caused by a unify- ferentiation into insulin-sensitive and 18. Ferrannini E, Haffner SM, Mitchell BD, ing underlying pathology, and their com- insulin-insensitive types. Lancet 1:127– Stern MP: Hyperinsulinaemia: the key bination confers a risk that is different 130, 1936 feature of a cardiovascular and meta- from the sum of the parts. In almost every 5. Shen S-W, Reaven GM, Farquhar JW: bolic syndrome. Diabetologia 34:416– way—from the term itself, to the under- Comparison of impedance to insulin 422, 1991 lying pathophysiology, to the variables in- mediated glucose uptake in normal and 19. Haffner SM, Valdez RA, Hazuda HP, cluded or excluded, to the value of diabetic subjects. J Clin Invest 49:2151– Mitchell BD, Morales PA, Stern MP: Pro- making the diagnosis, and finally to its 2160, 1970 spective analysis of the insulin-resis- treatment—the metabolic syndrome re- 6. Ginsburg H, Kimmerling G, Olefsky JM, tance syndrome (syndrome X). Diabetes quires much more study before its desig- Reaven GM: Demonstration of insulin 41:715–722, 1992 resistance in untreated adult onset dia- 20. Mykkanen L, Kuusisto J, Pyorala K, nation as a “syndrome” is truly warranted betic subjects with fasting hyperglyce- Laakso M: Cardiovascular disease risk and before its clinical utility is adequately mia. J Clin Invest 55:454–461, 1975 factors as predictors of type 2 (non-insu- defined. We hope this reappraisal gives 7. Reaven GM, Lerner R, Stern M, Farquhar lin-dependent) diabetes mellitus in el- pause to the growing use of the term, as JW: Role of insulin in endogenous hy- derly subjects. Diabetologia 36:553–559, well as stimulates urgently needed re- pertriglyceridemia. J Clin Invest 46: 1993 search. 1756–1767, 1967 21. Schmidt MI, Duncan BB, Watson RL, Consequently, in addition to the re- 8. Olefsky JM, Farquhar JW, Reaven GM: Sharrett AR, Brancati FL, Heiss G: A met- search suggested above, our recommen- Reappraisal of the role of insulin in hy- abolic syndrome in whites and African- dations to clinicians are: pertriglyceridemia. Am J Med 57:551– Americans: the Atherosclerosis Risk in 1) Adults with any major CVD risk factor 560, 1974 Communities baseline study. Diabetes 9. Orchard TJ, Becker DJ, Bates M, Kuller Care 19:414–418, 1996 should be evaluated for the presence of LH, Drash AL: Plasma insulin and li- 22. Laakso M, Sarlund H, Mykkanen L: In- other CVD risk factors. poprotein concentrations: an athero- sulin resistance is associated with lipid 2) Patients with CVD risk variables above genic association? Am J Epidemiol 118: and lipoprotein abnormalities in sub- the cut point for normal should receive 326–337, 1983 jects with varying degrees of glucose counseling for lifestyle modification, and 10. Olefsky JM, Kolterman OG, Scarlett JA: tolerance. Arteriosclerosis 10:223–231,

DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 2299 Critical appraisal of metabolic syndrome

1990 108:1422–1425, 2003 prospective and cross-sectional evalua- 23. Schmidt MI, Watson RL, Duncan BB, 32. Grundy SM, Hansen B, Smith SC Jr, Clee- tion. Atherosclerosis 165:285–292, 2002 Metcalf P, Brancati FL, Sharrett AR, man JI, Kahn RA: Clinical management 44. Alexander CM, Landsman PB, Teutsch Davis CE, Heiss G: Clustering of dyslip- of metabolic syndrome. Circulation 109: SM, Haffner SM, Third National Health idemia, hyperuricemia, diabetes, and 551–556, 2004 and Nutrition Examination Survey hypertension and its association with 33. Lakka HM, Laaksonen DE, Lakka TA, (NHANES III), National Cholesterol fasting insulin and central and overall Niskanen LK, Kumpusalo E, Tuom- Education Program (NCEP): NCEP- obesity in a general population: Athero- ilehto J, Salonen JT: The metabolic syn- defined metabolic syndrome, diabetes, sclerosis Risk in Communities Study drome and total and cardiovascular and prevalence of coronary heart disease Investigators. Metabolism 45:699–706, disease mortality in middle-aged men. among NHANES III participants age 50 1996 JAMA 288:2709–2716, 2002 years and older. Diabetes 52:1210– 24. Meigs JB, D’Agostino RB, Wilson PWF, 34. McNeill AM, Rosamond WD, Girman 1214, 2003 Cupples LA, Nathan DM, Singer DE: CJ, Golden SH, Schmidt MI, East HE, 45. Katzmarzyk PT, Church TS, Blair SN: Risk variable clustering in the insulin re- Ballantyne CM, Heiss G: The metabolic Cardiorespiratory fitness attenuates sistance syndrome: the Framingham syndrome and 11-year risk of incident the effects of the metabolic syndrome on Offspring Study. Diabetes 46:1594– cardiovascular disease in the Atheroscle- all-cause and cardiovascular disease 1600, 1997 rosis Risk in Communities study. Diabe- mortality in men. Arch Intern Med 164: 25. Liese AD, Mayer-Davis EJ, Haffner SM: tes Care 28:385–390, 2005 1092–1097, 2004 Development of the multiple metabolic 35. Stern SE, Williams K, Ferrannini E, De- 46. Ford ES: The metabolic syndrome and syndrome: an epidemiologic perspec- fronzo RA, Bogardus C, Stern MP: Iden- mortality from cardiovascular disease tive. Epidemiol Rev 20:157–172, 1998 tification of individuals with insulin and all-causes: findings from the Na- 26. World Health Organization: Definition, resistance using routine clinical mea- tional Health and Nutrition Examination Diagnosis, and Classification of Diabetes surements. Diabetes 54:333–339, 2005 Survey II Mortality Study. Atherosclerosis Mellitus and its Complications: Report 36. Ford ES, Giles WH: A comparison of the 173:309–314, 2004 of a WHO Consultation. Geneva, World prevalence of the metabolic syndrome 47. Girman CJ, Rhodes T, Mercuri M, Health Org., 1999 using two proposed definitions. Diabetes Pyorala K, Kjekshus J, Pedersen TR, 27. Expert Panel on the Detection, Evalua- Care 26:575–581, 2003 Beere PA, Gotto AM, Clearfield M, the 4S tion, and Treatment of High Blood Cho- 37. Meigs JB, Wilson PW, Nathan DM, Group, the AFCAPS/TexCAPS Research lesterol in Adults: Executive summary of D’Agostino RB Sr, Williams K, Haffner Group: The metabolic syndrome and the Third Report of the National Choles- SM: Prevalence and characteristics of the risk of major coronary events in the terol Education Program (NCEP) Expert metabolic syndrome in the San Antonio Scandinavian Simvastatin Survival Panel on Detection, Evaluation, and Heart and Framingham Offspring Stud- Study (4S) and the Air Force/Texas Cor- Treatment of High Blood Cholesterol in ies. Diabetes 52:2160–2167, 2003 onary Atherosclerosis Prevention Study Adults (Adult Treatment Panel III). 38. Lempiainen P, Mykkanen L, Pyorala K, (AFCAPS/TexCAPS). Am J Cardiol 93: JAMA 285:2486–2497, 2001 Laakso M, Kuusisto J: Insulin resistance 136–141, 2004 28. Grundy SM, Brewer HB Jr, Cleeman JI, syndrome predicts coronary heart dis- 48. Malik S, Wong ND, Franklin SS, Kamath Smith SC Jr, Lenfant C, National Heart, ease events in elderly non-diabetic men. TV, L’Italien GJ, Pio JR, Williams GR: Lung, and Blood Institute, American Circulation 100:123–128, 1999 Impact of the metabolic syndrome on Heart Association: Definition of meta- 39. Kekalainen P, Sarlund H, Pyorala K, mortality from coronary heart disease, bolic syndrome: report of the National Laakso M: Hyperinsulinemia cluster cardiovascular disease and on all causes Heart, Lung, and Blood Institute/Ameri- predicts the development of type 2 dia- in United States adults. Circulation 110: can Heart Association conference on betes independent of a family history of 1245–1250, 2004 scientific issues related to definition. Cir- diabetes. Diabetes Care 22:86–92, 1999 49. Hunt KJ, Resendez, Williams K, Haffner culation 109:433–438, 2004 40. Pyorala M, Miettinen H, Halonen P, SM, Stern MP: National Cholesterol Edu- 29. Balkau B, Charles MA, Drivsholm T, Laakso M, Pyorala K: Insulin resistance cation Program versus World Health Or- Borch-Johnsen K, Wareham N, Yudkin syndrome predicts the risk of coronary ganization metabolic syndrome in relation JS, Morris R, Zavaroni I, van Dam R, heart disease and stroke in healthy mid- to all-cause and cardiovascular mortality Feskins E, Gabriel R, Diet M, Nilsson dle-aged men: the 22-year follow-up re- in the San Antonio Heart Study. Circula- P, Hedblad B, European Group For sults of the Helsinki Policemen Study. tion 110:1251–1257, 2004 The Study Of Insulin Resistance (EGIR): Arterioscler Thromb Vasc Biol 20:538– 50. Scuteri A, Najjar S, Morrell C, Lakatta E: Frequency of the WHO metabolic 544, 2000 The metabolic syncrome in older indi- syndrome in European cohorts, and 41. Kuusisto J, Lempiainen P, Mykkanen L, viduals: prevalence and prediction of an alternative definition of an insulin re- Laakso M: Insulin resistance syndrome cardiovascular events. Diabetes Care 28: sistance syndrome. Diabetes Metab 28: predicts coronary heart disease events in 882–887, 2005 364–376, 2002 elderly type 2 diabetic men. Diabetes 51. Bruno G, Merletti F, Biggeri A, Bargero 30. Einhorn D, Reaven GM, Cobin RH, Ford Care 24:1629–1633, 2001 G, Ferrero S, Runzo C, Prina Cerai S, E, Ganda OP, Handelsman Y, Hellman 42. Isomaa B, Almgren P, Tuomi T, Forsen Pagano G, Cavallo-Perin P; Casale Mon- R, Jellinger PS, Kendall D, Krauss RM, B, Lahti K, Nissen M, Taskinen MR, ferrato Study: Metabolic syndrome as a Neufeld ND, Petak SM, Rodbard HW, Groop L: Cardiovascular morbidity and predictor of all-cause and cardiovascular Seibel JA, Smith DA, Wilson PW: Amer- mortality associated with the metabolic mortality in type 2 diabetes: the Casale ican College of Endocrinology position syndrome. Diabetes Care 24:683–689, Monferrato Study. Diabetes Care 27: statement on the insulin resistance syn- 2001 2689–2694, 2004 drome. Endocr Pract 9:237–252, 2003 43. Onat A, Ceyhan K, Basar O, Erer B, To- 52. Resnick HE, Jones K, Ruotolo G, Jain 31. Wilson PWF, Grundy SM: The meta- brak S, Sansoy V: Metabolic syndrome: AK, Henderson J, Lu W, Howard BV, bolic syndrome: practical guide to ori- major impact on coronary risk in a pop- Strong Heart Study: Insulin resistance, gins and treatment: part I. Circulation ulation with low cholesterol levels: a the metabolic syndrome, and risk of in-

2300 DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 Kahn and Associates

cident cardiovascular disease in non-di- with coronary heart disease: meta-anal- in the Framingham Offspring Study. abetic American Indians: the Strong yses of prospective studies. JAMA 279: Circulation 110:380–385, 2004 Heart Study. Diabetes Care 26:861–867, 1477–1482, 1998 75. Reilly MP, Wolfe ML, Rhodes T, Girman 2003 65. Ridker PM, Hennekens CH, Buring JE, C, Mehta N, Rader DJ: Measures of insu- 53. Marroquin OC, Kip KE, Kelley DE, Rifai N: C-reactive protein and other lin resistance add incremental value to Johnson BD, Shaw LJ, Bairey Merz CN, markers of inflammation in the predic- the clinical diagnosis of metabolic syn- Sharaf BL, Pepine CJ, Sopko G, Reis SE, tion of cardiovascular disease in women. drome in association with coronary ath- Women’s Ischemia Syndrome Evalua- N Engl J Med 342:836–843, 2000 erosclerosis. Circulation 110:803–809, tion Investigators: Metabolic syndrome 66. Ridker PM: Clinical applications of C-re- 2004 modifies the cardiovascular risk associ- active protein for cardiovascular disease 76. Ridker PM, Wilson PW, Grundy SM: ated with angiographic coronary artery detection and prevention. Circulation Should C-reactive protein be added to disease in women: a report from the 107:363–369, 2003 metabolic syndrome and to assessment Women’s Ischemia Syndrome Evalua- 67. Ridker PM, Cannon CP, Morrow D, Rifai of global cardiovascular risk? Circulation tion. Circulation 109:714–721, 2004 N, Rose LM, McCabe CH, Pfeffer MA, 109:2818–2825, 2004 54. Fuller JH, Shipley MJ, Rose G, Jarrett RJ, Braunwald E: C-reactive protein levels 77. Yudkin JS, Kumari M, Humphries SE, Keen H: Coronary heart disease risk and and outcomes after statin therapy: Prav- Mohamed-Ali V: Inflammation, obesity, impaired glucose tolerance: the White- astatin or Atorvastatin Evaluation and stress and coronary heart disease: is in- hall Study. Lancet 1:1373–1376, 1980 Infection Therapy-Thrombolysis in terleukin-6 the link? Atherosclerosis 148: 55. Lippel K, Tyroler H, Eder H, Gotto A, Myocardial Infarction 22 (PROVE IT- 209–214, 2000 Vahouny G: Relationship of hypertri- TIMI 22). N Engl J Med 352:20–28, 2005 78. Xu H, Barnes GT, Yang Q, Tan G, Yang glyceridemia to atherosclerosis. Athero- 68. Ridker PM, Buring JE, Cook NR, Rifai N: D, Chou CJ, Sole J, Nichols A, Ross JS, sclerosis 1:406–417, 1981 C-reactive protein, the metabolic syn- Tartaglia LA, Chen H: Chronic inflam- 56. Miller GJ, Miller NE: Plasma high den- drome, and risk of incident cardiovascu- mation in fat plays a crucial role in the sity lipoprotein concentration and de- lar events: an 8-year follow-up of 14,719 development of obesity-related insulin velopment of ischemic heart disease. initially healthy American women. Cir- resistance. J Clin Invest 112:1821–1830, Lancet 1:16–19, 1975 culation 107:391–397, 2003 2003 57. Keys A, Aravanis C, Blackburn H, Van 69. Pradhan AD, Cook NR, Buring JE, Man- 79. Wellen KE, Hotamisligil GS: Obesity-in- Buchem FS, Buzina R, Djordjevic BS, son JE, Ridker PM: C-reactive protein is duced inflammatory changes in adipose Fidanza F, Karvonen MJ, Menotti A, independently associated with fasting tissue. J Clin Invest 112:1785–1788, Puddu V, Taylor HL: Coronary heart dis- insulin in nondiabetic women. Arterio- 2003 ease: overweight and obesity as risk fac- scler Thromb Vasc Biol 23:650–655, 80. Pearson TA, Mensah GA, Alexander RW, tors. Ann Intern Med 77:15–27, 1972 2003 Anderson JL, Cannon RO 3rd, Criqui M, 58. Deckert T, Feldt-Rasmussen B, Jensen T, 70. Yudkin JS, Stehouwer CD, Emeis JJ, Fadl YY, Fortmann SP, Hong Y, Myers Kofoed-Enevoldsen A: Albuminuria re- Coppack SW: C-reactive protein in GL, Rifai N, Smith SC Jr, Taubert K, flects widespread vascular damage: the healthy subjects: associations with obe- Tracy RP, Vinicor F, Centers for Disease Steno hypothesis. Diabetologia 32:219– sity, insulin resistance, and endothelial Control and Prevention, American Heart 226, 1989 dysfunction: a potential role for cyto- Association: Markers of inflammation 59. Lewington S, Clarke R, Qizilbash N, kines originating from adipose tissue? and cardiovascular disease: application Peto R, Collins R: Age-specific relevance Arterioscler Thromb Vasc Biol 19:972– to clinical and public health practice: a of usual blood pressure to vascular mor- 978, 1999 statement for healthcare professionals tality: a meta-analysis of individual data 71. McLaughlin T, Abbasi F, Lamendola C, from the Centers for Disease Control and for one million adults in 61 prospective Liang L, Reaven G, Schaaf P, Reaven P: Prevention and the American Heart studies: Prospective Studies Collabora- Differentiation between obesity and in- Association. Circulation 107:499–511, tion. Lancet 360:1903–1913, 2002 sulin resistance in the association with 2003 60. Kannel WB, McGee DL: Diabetes and C-reactive protein. Circulation 106: 81. Kershaw EE, Flier JS: Adipose tissue as cardiovascular risk factors in the Fra- 2908–2912, 2002 an endocrine organ. J Clin Endocrinol mingham study. Circulation 59:8–13, 72. Festa A, D’Agostino R Jr, Howard G, Metab 89:2548–2556, 2004 1979 Mykkanen L, Tracy RP, Haffner SM: 82. Weyer C, Funahashi T, Tanaka S, Hotta 61. Rosengren A, Welin L, Tsipigianni A, Chronic subclinical inflammation as K, Matsuzawa Y, Pratley RE, Tataranni Wilhelmsen L: Impact of cardiovascular part of the insulin resistance syndrome: PA: Hypoadiponectinemia in obesity risk factors on coronary heart disease the Insulin Resistance Atherosclerosis and type 2 diabetes: close association and mortality among middle-aged dia- Study (IRAS). Circulation 102:42–47, with insulin resistance and hyperinsu- betic men: a general population study. 2000 linemia. J Clin Endocrinol Metab 86: BMJ 299:1127–1131, 1989 73. Sattar N, Gaw A, Scherbakova O, Ford I, 1930–1935, 2001 62. Stamler J, Vaccaro O, Neaton JD, Went- O’Reilly DS, Haffner SM, Isles C, Macfar- 83. Chandran M, Phillips SA, Ciaraldi T, worth D: Diabetes, other risk factors, lane PW, Packard CJ, Cobbe SM, Shep- Henry RR: Adiponectin: more than just and 12-year cardiovascular mortality for herd J: Metabolic syndrome with and another fat cell hormone? Diabetes Care men screened in the Multiple Risk Fac- without C-reactive protein as a predictor 26:2442–2450, 2003 tor Intervention Trial. Diabetes Care 16: of coronary heart disease and diabetes in 84. Matsubara M, Maruoka S, Katayose S: 434–444, 1993 the West of Scotland Coronary Preven- Decreased plasma adiponectin concen- 63. Ross R: Atherosclerosis: an inflammation Study. Circulation 108:414–419, trations in women with dyslipidemia. tory disease. N Engl J Med 340:115–126, 2003 J Clin Endocrinol Metab 87:2764–2769, 1999 74. Rutter MK, Meigs JB, Sullivan LM, 2002 64. Danesh J, Collins R, Appleby P, Peto R: D’Agostino RB Sr, Wilson PW: C-reac- 85. Kazumi T, Kawaguchi A, Sakai K, Association of fibrinogen, C-reactive tive protein, the metabolic syndrome, Hirano T, Yoshino G: Young men with protein, albumin, or leukocyte count and prediction of cardiovascular events high-normal blood pressure have lower

DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 2301 Critical appraisal of metabolic syndrome

serum adiponectin, smaller LDL size, a population-based study. Diabetes Care mortality in Busselton with reference to and higher elevated heart rate than those 21:649–654, 1998 glucose and insulin concentrations. Di- with optimal blood pressure. Diabetes 95. Wilson PW, D’Agostino RB, Levy D, Be- abetes Care 2:154–160, 1979 Care 25:971–976, 2002 langer AM, Silbershatz H, Kannel WB: 107. Fontbonne A, Charles MA, Thibult N, 86. Kumada M, Kihara S, Sumitsuji S, Prediction of coronary heart disease us- Richard JL, Claude JR, Warnet JM, Ros- Kawamoto T, Matsumoto S, Ouchi N, ing risk factor categories. Circulation 97: selin GE, Eschwege E: Hyperinsulin- Arita Y, Okamoto Y, Shimomura I, 1837–1847, 1998 aemia as a predictor of coronary heart Hiraoka H, Nakamura T, Funahashi T, 96. Stern MP, Williams K, Gonzalez-Villal- disease mortality in a healthy popula- Matsuzawa Y, Osaka CAD Study Group: pando C, Hunt KJ, Haffner SM: Does the tion: the Paris Prospective Study, 15- Association of hypoadiponectinemia metabolic syndrome improve identifica- year follow-up. Diabetologia 34:356– with coronary artery disease in men. tion of individuals at risk of type 2 dia- 361, 1991 Arterioscler Thromb Vasc Biol 23:85–89, betes and/or cardiovascular disease? 108. Moller LF, Jespersen J: Fasting serum in- 2003 Diabetes Care 27:2676–2681, 2004 [er- sulin levels and coronary heart disease in 87. Kojima S, Funahashi T, Sakamoto T, ratum in Diabetes Care 28:238, 2005] a Danish cohort: 17-year follow-up. Miyamoto S, Soejima H, Hokamaki J, 97. Klein BE, Klein R, Lee KE: Components J Cardiovasc Risk 2:235–240, 1995 Kajiwara I, Sugiyama S, Yoshimura M, of the metabolic syndrome and risk of 109. Perry IJ, Wannamethee SG, Whincup Fujimoto K, Miyao Y, Suefuji H, Kita- cardiovascular disease and diabetes in PH, Shaper AG, Walker MK, Alberti KG: gawa A, Ouchi N, Kihara S, Matsuzawa beaver dam. Diabetes Care 25:1790– Serum insulin and incident coronary Y, Ogawa H: The variation of plasma 1794, 2002 heart disease in middle-aged British concentrations of a novel, adipocyte de- 98. Golden SH, Folsom AR, Coresh J, Shar- men. Am J Epidemiol 144:224–234, rived protein, adiponectin, in patients rett AR, Szklo M, Brancati F: Risk factor 1996 with acute myocardial infarction. Heart groupings related to insulin resistance 110. Despres JP, Lamarche B, Mauriege P, 89:667–672, 2003 and their synergistic effects on subclini- Cantin B, Dagenais GR, Moorjani S, Lu- 88. Zoccali C, Mallamaci F, Tripepi G, cal atherosclerosis: the Atherosclerosis pien PJ: Hyperinsulinemia as an inde- Benedetto FA, Cutrupi S, Parlongo S, Risk in Communities study. Diabetes 51: pendent risk factor for ischemic heart Malatino LS, Bonanno G, Seminara G, 3069–3076, 2002 disease. N Engl J Med 334:952–957, Rapisarda F, Fatuzzo P, Buemi M, Nico- 99. Stevens RJ, Kothari V, Adler AI, Stratton 1996 cia G, Tanaka S, Ouchi N, Kihara S, Fu- IM; United Kingdom Prospective Diabe- 111. Pyorala M, Miettinen H, Laakso M, nahashi T, Matsuzawa Y: Adiponectin, tes Study (UKPDS) Group: The UKPDS Pyorala K: Hyperinsulinemia predicts metabolic risk factors, and cardiovascu- risk engine: a model for the risk of cor- coronary heart disease risk in healthy lar events among patients with end-stage onary heart disease in type II diabetes middle-aged men: the 22-year follow-up renal disease. J Am Soc Nephrol 13:134– (UKPDS 56). Clin Sci (Lond) 101:671– results of the Helsinki Policemen Study. 141, 2002 679, 2001 [erratum in Clin Sci (Lond) Circulation 98:398–404, 1998 89. Pischon T, Girman CJ, Hotamisligil GS, 102:679, 2002] 112. Ducimetiere P, Eschwege E, Papoz L, Rifai N, Hu FB, Rimm EB: Plasma adi- 100. Stern MP, Williams K, Hunt KJ: Impact Richard JL, Claude JR, Rosselin G: Rela- ponectin levels and risk of myocardial on diabetes/metabolic syndrome in pa- tionship of plasma insulin levels to the infarction in men. JAMA 291:1730– tients with established cardiovascular incidence of myocardial infarction and 1737, 2004 disease. Atheroscler Suppl 6:3–6, 2005 coronary heart disease mortality in a 90. Potter van Loon BJ, Kluft C, Radder JK, 101. Yarnell JWG, Patterson CC, Bainton D, middle-aged population. Diabetologia Blankenstein MA, Meinders AE: The car- Sweetnam PM: Is metabolic syndrome a 19:205–210, 1980 diovascular risk factor plasminogen ac- discrete entity in the general population? 113. Folsom AR, Szklo M, Stevens J, Liao F, tivator inhibitor type 1 is related to Evidence from the Caerphilly and Smith R, Eckfeldt JH: A prospective insulin resistance. Metabolism 42:945– Speedwell population studies. Heart 79: study of coronary heart disease in rela- 949, 1993 248–252, 1998 tion to fasting insulin, glucose, and dia- 91. Haffner SM, D’Agostino R Jr, Mykkanen 102. Yeo WW, Yeo KR: Predicting CHD risk betes: the Atherosclerosis Risk in L, Tracy R, Howard B, Rewers M, Selby J, in patients with diabetes mellitus. Diabet Communities (ARIC) study. Diabetes Savage PJ, Saad MF: Insulin sensitivity Med 18:341–344, 2001 Care 20:935–942, 1997 in subjects with type 2 diabetes: rela- 103. Song SH, Brown PM: Coronary heart 114. Yip J, Facchini FS, Reaven GM: Resis- tionship to cardiovascular risk factors: disease risk assessment in diabetes mel- tance to insulin-mediated glucose dis- the Insulin Resistance Atherosclerosis litus: comparison of UKPDS risk engine posal as a predictor of cardiovascular Study. Diabetes Care 22:562–568, 1999 with Framingham risk assessment func- disease. J Clin Endocrinol Metab 83: 92. Festa A, D’Agostino R Jr, Mykkanen L, tion and its clinical implications. Diabet 2773–2776, 1998 Tracy RP, Zaccaro DJ, Hales CN, Haffner Med 21:238–245, 2004 115. Uusitupa MI, Niskanen LK, Siitonen O, SM: Relative contribution of insulin and 104. McEwan P, Williams JE, Griffiths JD, Ba- Voutilainen E, Pyorala K: 5-year inci- its precursors to fibrinogen and PAI-1 in gust A, Peters JR, Hopkinson P, Currie dence of atherosclerotic vascular disease a large population with different states of CJ: Evaluating the performance of the in relation to general risk factors, insulin glucose tolerance: the Insulin Resistance Framingham risk equations in a popula- level, and abnormalities in lipoprotein Atherosclerosis Study (IRAS). Arterio- tion with diabetes. Diabet Med 21:318– composition in non-insulin dependent scler Thromb Vasc Biol 19:562–568, 1999 323, 2004 diabetic and non-diabetic subjects. Cir- 93. Ernst E, Resch KL: Fibrinogen as a car- 105. Pyorala K: Relationship of glucose toler- culation 82:27–36, 1990 diovascular risk factor: a meta-analysis ance and plasma insulin to the incidence 116. Hu G, Qiao Q, Tuomilehto J, Eliasson M, and review of the literature. Ann Intern of coronary heart disease: results from Feskens EJ, Pyorala K, DECODE Insulin Med 18:956–963, 1993 two population studies in Finland. Dia- Study Group: Plasma insulin and cardio- 94. Imperatore G, Riccardi G, Iovine C, Riv- betes Care 2:131–141, 1979 vascular mortality in non-diabetic Euro- ellese AA, Vaccaro O: Plasma fibrinogen: 106. Welborn TA, Wearne K: Coronary heart pean men and women: a meta-analysis a new factor of the metabolic syndrome: disease incidence and cardiovascular of data from eleven prospective studies.

2302 DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 Kahn and Associates

Diabetologia 47:1245–1256, 2004 type 2 diabetes independent of insulin don, Chapman and Hall, 1994, p. 93– 117. Welin L, Eriksson H, Larsson B, Ohlson resistance: evidence for a pathogenic 106 LO, Svardsudd K, Tibblin G: Hyperinsu- role of relative hyperinsulinemia. Diabe- 142. Lawlor DA, Ebrahim S, May M, Davey linaemia is not a major coronary risk fac- tes 49:2094–2101, 2000 Smith G: (Mis)use of factor analysis in tor in elderly men: the study of men born 130. McLaughlin T, Abbasi F, Cheal K, Chu J, the study of insulin resistance syn- in 1913. Diabetologia 35:766–770, 1992 Lamendola C, Reaven G: Use of meta- drome. Am J Epidemiol 159:1013–1018, 118. Orchard TJ, Eichner J, Kuller LH, Becker bolic markers to identify overweight in- 2004 DJ, McCallum LM, Grandits GA: Insulin dividuals who are insulin-resistant. Ann 143. Gray RS, Fabsitz RR, Cowan LD, Lee ET, as a predictor of coronary heart disease: Intern Med 139:802–809, 2003 Howard BV, Savage PJ: Risk factor clus- interaction with apolipoprotein E phe- 131. Liao Y, Kwon S, Shaughnessy S, Wallace tering in the insulin resistance syn- notype: a report from the Multiple Risk P, Hutto A, Jenkins AJ, Klein RL, Garvey drome: the Strong Heart Study. Am J Factor Intervention Trial. Ann Epidemiol WT: Critical evaluation of adult treat- Epidemiol 148:869–878, 1998 4:40–45, 1994 ment panel III criteria in identifying in- 144. Lehto S, Ronnemaa T, Pyorala K, Laakso 119. Ferrara A, Barrett-Connor EL, Edelstein sulin resistance with dyslipidemia. M: Cardiovascular risk factors clustering SL: Hyperinsulinemia does not increase Diabetes Care 27:978–983, 2004 with endogenous hyperinsulinaemia the risk of fatal cardiovascular disease in 132. Cheal KL, Abbasi F, Lamendola C, predict death from coronary heart dis- elderly men or women without diabetes: McLaughlin T, Reaven GM, Ford ES: ease in patients with type II diabetes. the Rancho Bernardo Study, 1984– Relationship to insulin resistance of Diabetologia 43:148–155, 2000 1991. Am J Epidemiol 140:857–869, the Adult Treatment Panel III diagnostic 145. Sakkinen PA, Wahl P, Cushman M, 1994 criteria for identification of the meta- Lewis MR, Tracy RP: Clustering of pro- 120. Stern MP: The insulin resistance syn- bolic syndrome. Diabetes 53:1195– coagulation, inflammation, and fibrino- drome: the controversy is dead, long live 1200, 2004 lysis variables with metabolic factors in the controversy! Diabetologia 37:956– 133. Laakso M, Sarlund H, Salonen R, Su- insulin resistance syndrome. Am J Epide- 958, 1994 honen M, Pyorala K, Salonen JT, Kar- miol 152:897–907, 2000 121. Wingard DL, Barrett-Connor EL, Fer- hapaa P: Asymptomatic atherosclerosis 146. Maison P, Byrne CD, Hales CN, Day NE, rara A: Is insulin really a heart disease and insulin resistance. Arterioscler Wareham NJ: Do different dimensions risk factor? Diabetes Care 18:1299– Thromb 11:1068–1076, 1991 of the metabolic syndrome change to- 1304, 1995 134. Agewall S, Fagerberg B, Attvall S, Wen- gether over time? Evidence supporting 122. McKeigue P, Davey G: Associations be- delhag I, Urbanavicius V, Wikstrand J: obesity as the central feature. Diabetes tween insulin levels and cardiovascular Carotid artery wall intima-media thick- Care 24:1758–1763, 2001 disease are confounded by comorbidity. ness is associated with insulin-mediated 147. Hanley AJ, Karter AJ, Festa A, Diabetes Care 18:1294–1298, 1995 glucose disposal in men at high and low D’Agostino R Jr, Wagenknecht LE, Sav- 123. Robbins DC, Andersen L, Bowsher R, coronary risk. Stroke 26:956–960, 1995 age P, Tracy RP, Saad MF, Haffner S: Chance R, Dinesen B, Frank B, Gin- 135. Howard G, O’Leary DH, Zaccaro D, Factor analysis of metabolic syndrome gerich R, Goldstein D, Widemeyer HM, Haffner S, Rewers M, Hamman R, Selby using directly measured insulin sensitiv- Haffner S, Hales CN, Jarett L, Polonsky JV, Saad MF, Savage P, Bergman R: In- ity: the Insulin Resistance Atherosclero- K, Porte D, Skyler J, Webb G, Gallagher sulin sensitivity and atherosclerosis: the sis Study. Diabetes 51:2642–2647, 2002 K: Report of the American Diabetes As- Insulin Resistance Atherosclerosis Study 148. Wang JJ, Qiao Q, Miettinen ME, Lappa- sociation’s Task Force on standardiza- (IRAS) Investigators. Circulation 93: lainen J, Hu G, Tuomilehto J: The meta- tion of the insulin assay. Diabetes 45: 1809–1817, 1996 bolic syndrome defined by factor 242–256, 1996 136. Haffner SM, Mykkanen L, Festa A, Burke analysis and incident type 2 diabetes in a 124. Wallace TM, Levy JC, Matthews DR: Use JP, Stern MP: Insulin-resistant predia- Chinese population with high postpran- and abuse of HOMA modeling. Diabetes betic subjects have more atherogenic dial glucose. Diabetes Care 27:2429– Care 27:1487–1495, 2004 risk factors than insulin-sensitive predi- 2437, 2004 125. Yeni-Komshian H, Carantoni M, Abbasi abetic subjects: implications for prevent- 149. Ford ES: Factor analysis and defining the F, Reaven GM: Relationship between ing coronary heart disease during the metabolic syndrome. Ethn Dis 13:429– several surrogate estimates of insulin re- prediabetic state. Circulation 101:975– 437, 2003 sistance and quantification of insulin- 980, 2000 150. Shen BJ, Todaro JF, Niaura R, McCaffery mediated glucose disposal in 490 137. DeFronzo RA: Lilly Lecture 1987: The JM, Zhang J, Spiro A 3rd, Ward KD: Are healthy non-diabetic volunteers. Diabe- triumvirate: ␤-cell, muscle, liver: a col- metabolic risk factors one unified syn- tes Care 23:171–175, 2000 lusion responsible for NIDDM. Diabetes drome? Modeling the structure of the 126. Ferrannini E, Balkau B: Insulin: in 37:667–687, 1988 metabolic syndrome X. Am J Epidemiol search of a syndrome. Diabet Med 19: 138. Reaven GM: Pathophysiology of insulin 157:701–711, 2003 724–729, 2002 resistance in human disease. Physiol Rev 151. Laws A, Reaven GM: Evidence for an in- 127. Facchini FS, Hua N, Abbasi F, Reaven 75:473–486, 1995 dependent relationship between insulin GM: Insulin resistance as a predictor of 139. Mardia KV, Kent JT, Bibby JM: Factor resistance and fasting plasma HDL-cho- age-related diseases. J Clin Endocrinol analysis. In Multivariate Analysis. Mardia lesterol, triglyceride and insulin concen- Metab 86:3574–3578, 2001 KV, Kent JT, Bibby JM, Eds. London, trations. J Intern Med 231:25–30, 1992 128. Mykkanen L, Haffner SM, Ronnemaa T, Academic Press, 1979, p. 255–280 152. Carr DB, Utzschneider KM, Hull RL, Ko- Bergman RN, Laakso M: Low insulin 140. Cureton EE, D’Agostino RB: Factor Anal- dama K, Retzlaff BM, Brunzell JD, Shofer sensitivity is associated with clustering ysis: An Applied Approach. Hillsdale, NJ, JB, Fish BE, Knopp RH, Kahn SE: Intra- of cardiovascular disease risk factors. Lawrence Erlbaum Associates, 1986 abdominal fat is a major determinant of Am J Epidemiol 146:315–321, 1997 141. Krzanowski WJ: Explaining observed the National Cholesterol Education Pro- 129. Weyer C, Hanson RL, Tataranni PA, Bo- associations: latent-variable models. In gram Adult Treatment Panel III criteria gardus C, Pratley RE: A high fasting Principles of Multivariate Statistical Meth- for the metabolic syndrome. Diabetes 53: plasma insulin concentration predicts ods: A Primer. Krzanowski WJ, Ed. Lon- 2087–2094, 2004

DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005 2303 Critical appraisal of metabolic syndrome

153. Pyorala K, Ballantyne CM, Gumbiner B, Metab 2:iv, 2002] Cushman WC, Green LA, Izzo JL Jr, Lee MW, Shah A, Davies MJ, Mitchel YB, 158. Aronoff S, Rosenblatt S, Braithwaite S, Jones DW, Materson BJ, Oparil S, Pedersen TR, Kjekshus J, Scandinavian Egan JW, Mathisen AL, Schneider RL: Wright JT Jr, Roccella EJ, Joint National Simvastatin Survival Study (4S): Reduc- Pioglitazone hydrochloride mono- Committee on Prevention, Detection, tion of cardiovascular events by simva- therapy improves glycemic control in Evaluation, and Treatment of High statin in nondiabetic coronary heart the treatment of patients with type 2 di- Blood Pressure, National Heart, Lung, disease patients with and without the abetes: a 6-month randomized placebo- and Blood Institute, National High metabolic syndrome: subgroup analyses controlled dose-response study: the Blood Pressure Education Program Co- of the Scandinavian Simvastatin Survival Pioglitazone 001 Study Group. Diabetes ordinating Committee: Seventh report Study (4S). Diabetes Care 27:1735– Care 23:1605–1611, 2000 of the Joint National Committee on Pre- 1740, 2004 159. Freed MI, Ratner R, Marcovina SM, Krei- vention, Detection, Evaluation, and 154. UK Prospective Diabetes Study der MM, Biswas N, Cohen BR, Brunzell JD, Treatment of High Blood Pressure. Hy- (UKPDS) Group: Effect of intensive Rosiglitazone Study 108 Investigators: Ef- pertension 42:1206–1252, 2003 blood-glucose control with metformin fects of rosiglitazone alone and in combi- 164. Smith SC Jr, Jackson R, Pearson TA, on complications in overweight patients nation with atorvastatin on the metabolic Fuster V, Yusuf S, Faergeman O, Wood with type 2 diabetes (UKPDS 34). Lancet abnormalities in type 2 diabetes mellitus. DA, Alderman M, Horgan J, Home P, 352:854–865, 1998 Am J Cardiol 90:947–952, 2002 Hunn M, Grundy SM: Principles for na- 155. Knowler WC, Barrett-Connor E, Fowler 160. Haffner SM, Greenberg AS, Weston tional and regional guidelines on cardio- SE, Hamman RF, Lachin JM, Walker EA, WM, Chen H, Williams K, Freed MI: Ef- vascular disease prevention: a scientific Nathan DM, Diabetes Prevention Pro- fect of rosiglitazone treatment on non- statement from the World Heart and gram Research Group: Reduction in the traditional markers of cardiovascular Stroke Forum. Circulation 109:3112– incidence of type 2 diabetes with life- disease in patients with type 2 diabetes 3121, 2004 style intervention or metformin. N Engl mellitus. Circulation 106:679–684, 165. International Diabetes Federation: The J Med 346:393–403, 2002 2002 IDF Consensus worldwide definition of 156. Shadid S, Jensen MD: Effects of pioglita- 161. Satoh N, Ogawa Y, Usui T, Tagami T, the metabolic syndrome [article online]. zone versus diet and exercise on meta- Kono S, Uesugi H, Sugiyama H, Sug- Available from www.idf.org. Accessed 2 bolic health and fat distribution in upper awara A, Yamada K, Shimatsu A, Kuzuya June 2005 body obesity. Diabetes Care 26:3148– H, Nakao K: Antiatherogenic effect of 166. Stein R: New diagnosis for overweight: 3152, 2003 pioglitazone in type 2 diabetic patients major risk factors add up to “metabolic 157. Lebovitz HE, Dole JF, Patwardhan R, irrespective of the responsiveness to its syndrome.” Washington Post, 8 February Rappaport EB, Freed MI, Rosiglitazone antidiabetic effect. Diabetes Care 26: 2005: Sect. A, p. 1 Clinical Trials Study Group: Rosiglita- 2493–2499, 2003 167. Stedman’s Medical Dictionary. 27th ed. zone monotherapy is effective in pa- 162. Davidson MB: Is treatment of insulin re- Baltimore, MD, Lippincott, Williams and tients with type 2 diabetes. J Clin sistance beneficial independent of gly- Wilkins, 2000, p. 1746 Endocrinol Metab 86:280–288, 2001 [er- cemia? Diabetes Care 26:3184–3186, 168. American Diabetes Association: Stan- ratum in J Clin Endocrinol Metab 2003 dards of medical care in diabetes. Diabe- 86:1659, 2001 and J Clin Endocrinol 163. Chobanian AV, Bakris GL, Black HR, tes Care 28 (Suppl. 1):S4–S36, 2005

2304 DIABETES CARE, VOLUME 28, NUMBER 9, SEPTEMBER 2005