Recently, there has been a growing tendency to think of diabetes and cardiovascular disease as sharing the same, or similar, environmental and genetic antecedents, that is, as "springing from a common soil." This idea was originated by Jarrett [1], who noted that cardiovascular complications were as common in newly diagnosed diabetic persons as in those whose diabetes was of long duration. Jarrett reasoned that if cardiovascular disease were a complication of diabetes, then its frequency would increase with increasing duration of diabetes. This correlation was not observed, however, at least not in the Whitehall Study [2]. Moreover, nondiabetic patients who were merely at increased risk by virtue of having impaired glucose tolerance seemed to have the same rate of cardiovascular disease as patients with long-established diabetes [2]. These findings led Jarrett to speculate that cardiovascular disease, rather than being a complication of diabetes, shared environmental and genetic antecedents with diabetes.

Early-Life Influences

Recently, Hales and colleagues [3] and Barker and associates [4] have called attention to early-life determinants of adult diseases. They have presented evidence indicating that low birth weight and low weight gain during the first year of life are associated with the occurrence of NIDDM and cardiovascular disease decades later [3, 4]. Hales and Barker [5] theorized that inadequate nutrition during fetal and early life might lead to inadequate development of pancreatic islet cells with a propensity toward islet cell failure and thus diabetes in adult life.

Low birth weight and low weight at age 1 year were also shown by Barker and colleagues [6] to be associated with the insulin resistance syndrome in adults, a finding recently confirmed not only in non-Hispanic whites but also in Mexican Americans Figure 1 [7]. This syndrome was initially described in lean persons by Reaven [8], who labeled it "syndrome X." As described by Reaven, the syndrome consisted of insulin resistance, hyperinsulinemia, dyslipidemia with high triglyceride and low high-density lipoprotein cholesterol levels, and hypertension [8]. Clearly, most persons who manifest features of the syndrome have, in addition, obesity, and in particular abdominal obesity. The type of obesity they have is probably visceral obesity, but because most studies characterize the obesity using anthropometric measurements such as waist circumference rather than abdominal computed tomography scans, it is not possible to distinguish definitively between visceral and subcutaneous abdominal adiposity. Therefore, although I use the term "abdominal obesity," the visceral adipose depot is in all likelihood the one directly linked to the insulin resistance syndrome and may be one of its determinants.

View larger version (55K): [in this window] [in a new window]   Figure 1. Prevalence of the insulin resistance syndrome by ethnicity and tertile of birth weight in 562 patients from the San Antonio Heart Study. The insulin resistance syndrome is defined as two or more of the following: hypertension; diabetes or impaired glucose tolerance; high triglyceride level (> 2.8 mmol/L); or low high-density lipoprotein cholesterol level (< 0.9 mmol/L). (From reference 7 with permission.).

As with the insulin resistance syndrome, low birth weight has also been shown to predict future development of abdominal obesity. This finding has been documented in whites in a United Kingdom study [9] and in both non-Hispanic whites and Mexican Americans in the San Antonio, Texas, population [7]. Perhaps inadequate nutrition in early life limits adipocyte development, and perhaps visceral adipocytes, because of the greater ease with which they undergo hypertrophy [10], become the storage depot of choice when excess calories in adults encounter limited adipose storage sites.

Can the theory be proved that inadequate nutrition in early life is the "common soil" that gives rise to both NIDDM and cardiovascular disease in adults? I have already alluded to a mechanism whereby this deficiency might lead to inadequate islet cell development, predisposing persons to later islet cell failure, and to a corresponding mechanism whereby the deficiency might also lead to abdominal obesity, which could in turn contribute to development of the insulin resistance syndrome in adults. If the insulin resistance syndrome could be shown to be an important risk factor for both NIDDM and cardiovascular disease, then the framework would exist for a comprehensive schema leading from nutritional deficiencies in early life to NIDDM and cardiovascular disease in adults. This hypothetical schema is shown in Figure 2. I will show that abundant evidence exists that each element of the insulin resistance syndrome is a risk factor for the future development of NIDDM. The evidence that the insulin resistance syndrome elements are also risk factors for future cardiovascular disease, however, is less conclusive.

View larger version (14K): [in this window] [in a new window]   Figure 2. Schema for the "common soil" hypothesis for NIDDM and cardiovascular disease.

The Insulin Resistance Syndrome: The Missing Link?

One plausible theory is that insulin resistance places an extra demand on the pancreatic islet cells, compelling them to hypersecrete insulin, presumably for many years or even decades, and that this stress eventually leads to islet cell failure. In addition to theoretical considerations, however, empirical evidence also points to the insulin resistance syndrome as a precursor of diabetes. In fact, each element of the insulin resistance syndrome has been documented as a risk factor for NIDDM in at least two and up to ten prospective epidemiologic studies. Insulin resistance measured by the euglycemic insulin clamp technique in Pima Indians [11] and by the intravenous glucose tolerance test in whites [12] has been documented as a diabetes risk factor in prospective studies. Hyperinsulinemia, presumably reflecting compensatory hypersecretion of insulin, has also been documented to be a risk factor for NIDDM in at least six prospective epidemiologic studies [13-18]. Obesity and abnormal glucose tolerance are the classic risk factors for diabetes [13-24]. There is also evidence that abdominal obesity is an independent risk factor for diabetes [15-18, 22] and is perhaps more important than obesity itself. In the San Antonio Heart Study [25], overall obesity as reflected by body mass index (k/m²) is an independent risk factor for diabetes, but ceases to be so when the ratio of waist-to-hip circumference, a measure of abdominal obesity, is added to the predicting model (unpublished data). Evidence also suggests that the same is true for cardiovascular disease, namely, that abdominal obesity rather than generalized obesity is the stronger risk factor.

Hypertension and dyslipidemia with high triglyceride and low high-density lipoprotein levels are well-known concomitants of NIDDM. Perhaps less well known is that these two features of the insulin resistance syndrome are also precursors of diabetes [13, 15, 18-2023, 24]. Thus, all of the elements of the insulin resistance syndrome are risk factors for NIDDM. Because low birth weight is associated with development of the insulin resistance syndrome in adults [6, 7], the argument can be made that the insulin resistance syndrome is the metabolic derangement that links early nutritional deficiency and later NIDDM. This formulation implies that insulin resistance and compensatory insulin hypersecretion are the principal precursors of NIDDM. However, Hales and Barker [5] have argued that early nutritional deficiency leads to stunted islet cell development and a predisposition to islet cell failure. Is it possible to reconcile these two views?

The extensive epidemiologic data implicating hyperinsulinemia as a risk factor for diabetes is based on insulin assays that are now known to be nonspecific and to cross-react with proinsulin and other insulin precursors [26]. The relevance of this finding to the insulin resistance-insulin deficiency debate stems from the belief that the appearance of insulin precursors in the circulation signals a failing pancreas. We now generally accept that much, perhaps most, of the immunoassayable insulin in patients with established NIDDM is, in fact, not insulin but rather proinsulin [27-29]. The question at issue, however, is whether these findings extend to prediabetic persons. In other words, is diabetes preceded by a prolonged period of hyperinsulinemia reflecting a compensatory response to insulin resistance, or is it preceded by a prolonged period of hyperproinsulinemia reflecting islet cells on the verge of failure. Of note is the fact that high-risk populations such as Mexican Americans or patients with impaired glucose tolerance—populations that may be presumed to include many prediabetic persons—have elevated levels of both insulin and proinsulin [28-30]. However, the elevation of the latter is not disproportionate to the elevation of insulin itself. Thus, the ratio of proinsulin to insulin concentration is normal or at most minimally elevated [28-30]. These findings suggest that proinsulin plays at most a minor role during the prediabetic period. On the other hand, prospective data suggest that the fasting proinsulin concentration is a better predictor of future diabetes than is the specific insulin concentration (that is, a concentration measured with a radioimmunoassay that does not cross-react with proinsulin).

Additional data implying a role for insulin deficiency in the pathogenesis of NIDDM are based on the insulin response to an oral glucose load. Several studies have found that diminished 30-minute or 2-hour insulin concentrations, or both, after oral glucose are predictive of future diabetes [13, 17, 31-33]. Haffner and colleagues [34] have recently presented data from the San Antonio Heart Study indicating that a diminished 30-minute increment in the insulin-to-glucose ratio after an oral glucose load is predictive of the development of diabetes 7 to 8 years later.

One way to reconcile the insulin resistance-insulin deficiency controversy is to distinguish between risk factors that act early as opposed to late in the prediabetic period—on the "eve of conversion," so to speak. Picture the insulin resistance syndrome as acting during the early prediabetic period and those risk factors that signal a failing pancreas as acting during the late prediabetic period. Compatible with this view is the fact that several studies showing proinsulin concentration to be a more important risk factor for diabetes than insulin concentration had relatively short follow-up (3.5 and 5 years) [35, 36]. This is not true, however, of all studies implicating insulin deficiency as a risk factor, some of which had follow-ups as long as 10 to 20 years [37]. Even in the longer follow-up study, however, the elapsed time from risk factor measurement to conversion to diabetes was not reported. Thus, it is not known which patients converted to diabetes soon after the baseline examination and which converted later or whether the risk factors differed between the early and later converters. A prospective study of diabetes incidence is needed, with follow-up examinations at intervals of 2 or 3 years and a complete panel of risk factor measurements at each interval examination. Such a study would define the natural history of metabolic changes as they evolve throughout the prediabetic period. The recently funded Diabetes Prevention Trial for NIDDM may address some of these issues.

The Insulin Resistance Syndrome: The Link to Cardiovascular Disease

The final step in establishing the insulin resistance syndrome as the link between early nutritional deficiency and the subsequent development of diabetes and cardiovascular disease is to confirm the elements of the insulin resistance syndrome as cardiovascular risk factors. Here the evidence shows mixed results. Several elements of the insulin resistance syndrome are undoubted cardiovascular risk factors. Others are more questionable. Obesity, for example, is a well-established cardiovascular risk factor, at least in univariate analyses [38, 39], and in a long-term follow-up of the Framingham Heart Study, it appears to be an independent cardiovascular risk factor as well [40]. As with diabetes, the evidence suggests that it is not obesity per se but rather abdominal obesity that constitutes the strongest adiposity-related risk factor [41, 42]. Hypertension is also a well-established cardiovascular risk factor [43, 44], as are low levels of high-density lipoprotein cholesterol [45, 46]. Hypertriglyceridemia, on the other hand, has long been controversial as a cardiovascular risk factor [47]. This controversy has recently been clarified by the discovery that hypertriglyceridemia is associated with a shift in the density spectrum of low-density lipoproteins (LDLs) toward smaller, denser particles [48-50]. These small, dense LDL particles are thought to be more atherogenic than ordinary LDL particles [51-53], perhaps because they are more vulnerable to oxidation [54], and oxidized LDL particles appear to be more atherogenic than native LDL particles [55]. Hypertriglyceridemia is also associated with increased concentrations of plasminogen activator inhibitor-1 [56], which favors the atherothrombotic process by inhibiting fibrinolysis [57].

The cardinal features of the insulin resistance syndrome, however, are insulin resistance and hyperinsulinemia. Does any evidence exist that these features are cardiovascular risk factors? No study has directly tested the hypothesis that insulin resistance per se is a cardiovascular risk factor. Several prospective studies in the 1980s purported to show that hyperinsulinemia was associated with increased risk for cardiovascular disease [58-60]. Several more recent prospective studies, however, have not confirmed these findings [61-65]. Data from animal studies and in vitro data suggest that insulin is atherogenic [66]. Thus far, however, it must be concluded that these findings have not been confirmed in humans. This point is of more than theoretical interest because it bears on exogenous insulin use in the treatment of diabetes.

A summary of the schema presented thus far is as follows: An adverse environment in early life, which probably includes nutritional deficiencies, constitutes the "common soil" predisposing a person to both NIDDM and cardiovascular disease as an adult. The insulin resistance syndrome is the link between the early-life exposures and the eventual disease manifestations many decades later. Evidence for the linking role of the insulin resistance syndrome is that low birth weight is associated with the later development of this syndrome and that all elements of the syndrome are risk factors for NIDDM and most are risk factors for cardiovascular disease as well.

The Common Antecedent Theory: Observations That Do Not Quite Fit

Although the above formulation is appealing, certain facts remain that are not readily accommodated. Chief among these is that certain populations, most notably Pima Indians, have very high rates of NIDDM but very low rates of cardiovascular disease [67]. Of course, many factors other than nutrition in early life and the insulin resistance syndrome influence these two conditions, so identical relationships in all populations are not expected. Even among Pima Indians, where the "soil" is evidently not conducive to the development of cardiovascular disease, the rates are higher in diabetic than in nondiabetic persons [67]. Moreover, in the Strong Heart Study [68], the disjunction between the rates of diabetes and cardiovascular disease was seen only among Native Americans from Arizona. Among Native Americans from Oklahoma and North and South Dakota, the rates of both diseases were high [68]. Interestingly, the percentage of white genetic admixture is also higher in the latter populations [69], suggesting that the link between diabetes and cardiovascular disease is strongest in white populations. However, high rates of NIDDM and cardiovascular disease also occur among South Asians living in the United Kingdom [70] and indicate that this link can be strong in non-white populations as well.

The disjunction between diabetes and cardiovascular disease may also exist in Mexican Americans who share Native American and white genetic admixture (approximately 40% and 60%, respectively) [71]. Mexican Americans have a prevalence of NIDDM approximately three times higher than among non-Hispanic whites [72]. Yet their cardiovascular mortality, at least among men, is about 15% to 20% lower than in non-Hispanic whites [73, 74]. They also have a lower prevalence of myocardial infarction than do non-Hispanic whites [73], although this difference could be attributed to a higher case fatality rate. In fact, some evidence exists that this is true; one study found a higher admission rate to coronary care units [75] and higher case fatality rates [76] among Mexican Americans than among non-Hispanic whites. Although the exact rate of cardiovascular disease in Mexican Americans is still somewhat uncertain, it does not seem to be nearly as excessive as is the case for NIDDM.

Diabetes and Cardiovascular Disease: Direct Links

Even if NIDDM and cardiovascular disease have common genetic and environmental antecedents, it does not preclude the possibility that the onset of clinical diabetes (that is, significant hyperglycemia) further increases the risk for cardiovascular disease. To the extent that this is true, cardiovascular disease can be said to be truly a complication of diabetes. This possibility is also shown in Figure 2. Several studies reported in the 1980s, however, failed to find a relation between the degree of hyperglycemia in diabetic patients and cardiovascular risk [77-79]. More recently, studies from Finland have suggested that glycated hemoglobin levels were univariate predictors of cardiovascular end points in diabetic patients [80]. Not only have the epidemiologic results been mixed, but no compelling mechanism has been found whereby hyperglycemia might play a role in atherogenesis. Recent work by Bucala and colleagues [81], however, may have provided such a mechanism. These researchers recently described an advanced glycosylation end-product modification of LDL [81] that has diminished fractional clearance in transgenic mice expressing the human LDL receptor [82]. They have also shown that circulation levels of this type of modified LDL are elevated in diabetic patients [81] and can be lowered by administering the advanced glycosylation inhibitor aminoguanidine [82]. The diminished clearance of the modified LDL could contribute to diabetic dyslipidemia and atherosclerosis, thereby providing a mechanism linking hyperglycemia to atherosclerotic cardiovascular disease.

The Question of Insulin Toxicity

The question of whether endogenous insulin levels are a cardiovascular risk factor has major clinical implications because it raises questions about whether exogenous insulin administration might also increase cardiovascular risk. This issue becomes particularly salient with the recent publication of the Diabetes Control and Complications Trial (DCCT) results [83]. This study provided highly persuasive (some would say conclusive) evidence that tight control of the blood glucose level with insulin in patients with IDDM markedly reduces their risk for microvascular complications of diabetes. Adverse side effects of the intensive insulin regimen were weight gain and a threefold greater risk for hypoglycemic episodes. Concern has therefore been expressed about whether the benefits achieved in the DCCT can be extrapolated to patients with NIDDM who, because of their older age and greater likelihood of comorbidity, might suffer graver consequences from hypoglycemia and weight gain [84]. Nevertheless, the DCCT results probably argue in favor of a more aggressive approach to lowering blood sugar levels in patients with NIDDM. However, many clinicians are now questioning whether overinsulinization of these patients might expose them to an increased risk for cardiovascular disease. In a recent series of articles on the insulin resistance syndrome [85-88], a somewhat reassuring consensus has emerged that this syndrome is a cardiovascular risk factor, not because of insulin itself, but because of its other, noninsulinemic manifestations.

The possible atherogenic potential of exogenous insulin administration can only be confirmed or refuted with well-designed and well-executed randomized clinical trials. The much maligned University Group Diabetes Program (UGDP) is somewhat reassuring on this topic. Patients were randomly assigned to placebo or to one of two insulin regimens (fixed and variable) [89]. Neither insulin regimen was associated with a high rate of cardiovascular mortality relative to placebo. Unfortunately, reduced cardiovascular mortality was not associated with insulin therapy either. Nevertheless, if the DCCT results can be extrapolated to UGDP-like patients, they would at least be expected to have less microvascular disease-related morbidity, if not reduced cardiovascular mortality. The DCCT itself offered some evidence in favor of reduced cardiovascular disease in the intensively managed group, although the number of end points was small and the result was statistically insignificant, owing presumably to the young age of the DCCT participants. Also, the cardiovascular end points included peripheral vascular disease and amputations, which have a microvascular (that is, neuropathic) as well as a macrovascular component. On the other hand, the pilot phase of the Veterans Affairs Cooperative Study on Glycemic Control and Complications in Diabetes Mellitus (VACSDM), which involved 153 patients with NIDDM who were followed for 27 months, gave less reassuring results. Major cardiovascular events were almost twice as common, and total cardiovascular events were approximately 50% more common in the intensively managed group compared with the standard therapy group (see "The Feasibility of Intensive Insulin Management in NIDDM: Implications of the Veterans Affairs Cooperative Study on Glycemic Control and Complications in NIDDM"). These results, however, were not statistically significant, probably because of the small number of events. It is hoped that the complete VACSDM trial will now be done to see if this result holds up. Finally, the United Kingdom Prospective Diabetes Study [90] (see "United Kingdom Prospective Diabetes Study 17: A 9-Year Update of a Randomized, Controlled Trial on the Effect of Improved Metabolic Control on Complications in NIDDM"), still in progress, has the potential to shed additional light on this vexing issue.

At present, sufficient evidence does not exist to indict insulin as a cardiovascular risk factor or to withhold optimum insulinization to minimize hyperglycemia in patients with NIDDM in an effort to prevent microvascular complications. Further clinical trial data, however, would certainly be welcome to provide a definitive answer to this urgent clinical question.