In addition to these risk factors, there are abnormalities that particularly accompany diabetes and may be associated with atherosclerosis. In some cases, the evidence for these abnormalities being atherogenic is only pathophysiologic. In other cases, it is both pathophysiologic and epidemiologic. There is increasing epidemiologic evidence, although controversy still remains, that hyperglycemia is a risk factor for atherosclerosis (8–10). This may be because nonenzymatic glycation of proteins, such as those in the lipoproteins (11) or those in the artery wall, could be atherogenic. In addition, there is evidence that advanced glycation end products can produce changes in the artery that make it more susceptible to atherosclerosis (12). Many studies have shown that hyperinsulinemia or insulin resistance or both can increase or mark an increased risk of atherosclerosis (13–16).

Yet other risk factors are more likely to be present in those with diabetes than in those without diabetes. For example, renal disease, be it microalbuminuria or renal failure, is more prevalent in diabetic patients. Its presence is also accompanied by an increased risk of coronary artery disease (17,18). The same can be said of hypertension, dyslipoproteinemias, and alterations in coagulation factors.

The most common lipoprotein disorder in diabetes is an increase in the levels of the triglyceride-rich lipoproteins (19,20). Hypertriglyceridemia has been found at the time of diagnosis of type 2 diabetes (21) and has even been found in those who will develop diabetes before they manifest hyperglycemia (22). There is increasing evidence to support the coronary risk effect of hypertriglyceridemia, both in the general population (23) and in individuals with diabetes (24–26). There is still debate about whether this is a direct effect of the triglyceride-rich lipoproteins or if it is due to the frequently associated low levels of HDL. However, it should be recognized that in many individuals with diabetes, particularly those treated with insulin, the HDL levels need not be decreased (20).

The term dyslipoproteinemia has been used because the lipoprotein abnormalities in diabetes are not only in quantity but also in quality. Mention has already been made of the fact that the apolipoproteins can be nonenzymatically glycated. There have been many studies indicating that glycated LDL can be more atherogenic than native LDL (27). This may be because compared with native LDL, a greater proportion is bound to the "scavenger receptors" than to the classical "LDL receptors," because glycated LDL has a longer residence time in the circulation and because it is more likely to be oxidized. In addition to these changes in the LDL, in diabetic individuals, the particles are smaller and denser than in nondiabetic individuals (28). This is in part, but not entirely, accounted for by the hypertriglyceridemia that is frequently seen in diabetes. It is changes such as these that may make any given amount of LDL more atherogenic in diabetes. This could be one of the reasons that MRFIT found that at any given level of cholesterol, the person with diabetes is at a much greater risk for coronary artery disease. There are also changes in the lipid and protein compositions of the other lipoproteins. However, their link to atherogenesis has not been as well established.

There have also been a limited number of trials focusing on the effects of lowering plasma triglycerides and increasing HDL (34–37). These trials have also shown that lowering plasma triglycerides and increasing HDL reduces atherosclerotic cardiovascular disease. In general, it has not yet been possible to determine whether it is the combination of both lipid and lipoprotein alterations or one alone that is responsible for the cardiovascular benefit.

The relative benefit of correcting dyslipoproteinemias is greatest in individuals who already have clinical coronary artery disease (i.e., in secondary prevention). However, there is a larger population that has no clinical coronary artery disease and that benefits from such treatment (i.e., primary prevention). Hence, there may be a greater absolute benefit to primary prevention. One should recognize that the terms "primary prevention" and "secondary prevention" may be arbitrary. There are few adults who have absolutely no atherosclerosis, even if they do not have clinical symptoms. This is particularly true in diabetes, a condition in which myocardial ischemia is often silent.

The other two studies, the Cholesterol and Recurrent Events (CARE) Trial in patients with "average" cholesterol levels (30,31) and the Scandinavian Simvastatin Survival Study (4S) in patients with elevated cholesterol levels (32), were secondary intervention studies that respectively used pravastatin or simvastatin, two drugs of the hydroxymethylglutaryl-CoA reductase inhibitor group. In post hoc analyses of their diabetic subgroups, each study found a reduction in coronary events in individuals treated with the active drug. In contrast to the total population, individuals with diabetes did not have a reduction in mortality. However, this was probably due to the smaller number of individuals in the diabetic population. The 4S study (41) had an obvious bias toward individuals with diabetes who were selected for this study. This group made up only 5% of the whole 4S study population, a proportion much smaller than the percentage of those with diabetes in the same age-group in the same region. The low percentage of diabetes in the 4S population undoubtedly reflected the entry and exclusion criteria for the study. One of these criteria was to exclude individuals whose triglyceride levels exceeded 2.5 mmol/l. Because the most frequent form of hyperlipidemia in diabetes is hypertriglyceridemia, this presents an obvious limitation to the direct applicability of the data to the hyperlipidemia seen in diabetes. The CARE Trial had a higher triglyceride exclusion value (>3.95 mmol/l). However, to find a significant reduction in events, the CARE Trial had to expand its primary end point to include not only coronary heart disease deaths and nonfatal myocardial infarction, but also coronary artery bypass and angioplasty (31). Despite such limitations and the cautions that must be exercised with all post hoc subgroup analyses, these studies strongly suggest that treating hypercholesterolemia in diabetes will reduce the risk of recurrent cardiac events in individuals with preexisting coronary artery disease.

No studies published to date have been designed to examine the effects of lipid intervention on coronary artery disease specifically in diabetes. Four are currently under way. Three are clinical end point studies that have just started, and their protocols have not yet been published. The Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) Study is examining the effect of treatment with micronized fenofibrate on total and fatal coronary artery disease events in men and women with type 2 diabetes, some of whom will have known coronary disease. Their lipid entry criteria are as follows: triglyceride >1.0 mmol/l and total cholesterol <5.5 mmol/l if there is prior coronary heart disease or <6.5 mmol/l if not. This study is being conducted in Finland, Australia, and New Zealand (A. Keech, personal communication). The second clinical end point study is the Collaborative Atorvastatin Diabetes Study (CARDS), which is examining the effects of a minimum of 4 years' treatment with atorvastatin versus placebo in 2,120 patients with type 2 diabetes and no established cardiovascular disease. Their entry lipid levels must be LDL <4.14 mmol/l and triglyceride <6.78 mmol/l. The study is being undertaken by 90 centers in the U.K. and the Republic of Ireland (D.J. Betteridge, personal communication). The third clinical end point study is the Lipids in Diabetes Study (LDS). It is also a study on a population of 4,000 men and women with type 2 diabetes who have normal or near-normal lipid levels and are not thought to have had previous cardiovascular disease. This study is also being undertaken in the U.K. It is planned with a 2 X 2 factorial design in which various pairs of placebo, micronized fenofibrate, and cerivastatin are used to modify the participants' plasma lipids (R.R. Holman, personal communication).

The angiographic Diabetes Atherosclerosis Intervention Study (DAIS) was the first of the lipid intervention studies specifically designed in diabetes and is now nearing completion. The protocol has been published (43). In summary, the primary aim of DAIS is to determine whether long-term treatment with micronized fenofibrate will alter the progression or regression of angiographically determined coronary artery disease in men and women with type 2 diabetes. The lipid entry criteria allow for randomization of individuals with either mild hypertriglyceridemia, mild hypercholesterolemia, or both. Quantitative coronary angiograms (44) are conducted at entry and at the end of the treatment period. Treatment continues for a minimum of 3 years after the last randomized person has entered the study. Because of the duration of the recruitment period, the average duration of treatment will be slightly over 4 years. The study is conducted in Canada, Finland, France, and Sweden. Calculations suggested that when allowing for a 20% dropout rate, the sample size required for the study would be 260. A margin of 40 was added to this calculation, thus giving a recruitment target of 300. In fact, 418 people (315 men and 113 women) were randomized. Of these individuals, 218 had no prior clinical evidence of coronary artery disease and 200 had prior clinical evidence of coronary artery disease. The population baseline characteristics have been described (45). The group mean ± SD level for plasma cholesterol was 5.57 ± 0.70 mmol/l, for plasma triglyceride was 2.42 ± 1.00 mmol/l, for LDL cholesterol was 3.43 ± 0.70 mmol/l, and for HDL cholesterol was 1.03 ± 0.20 mmol/l. At entry, patients had acceptable glycemic control, with fasting plasma glucose levels of 8.80 ± 2.50 mmol/l and HbA_1c values of 7.51 ± 1.20%. There were no regional or sex differences in these values.