2009 World Congress on the Insulin Resistance Syndrome: Cardiovascular Disease Concepts

Inflammation and insulin resistance

Steven Shoelson (Boston MA) discussed inflammation and the adaptive immune system in the pathogenesis of type 2 diabetes. The adaptive immune system has been thought to play roles in autoimmunity and in the pathogenesis of type 1 diabetes, but there is growing evidence of its role in type 2 diabetes and in atherosclerosis. Epidemiologic studies show that diabetes is associated with higher levels of circulating markers of inflammation—not only C-reactive peptide (CRP), but also the leukocyte count—and with levels of proinflammatory cytokines, such as tumor necrosis factor (TNF)-α and interleukin (IL)-6. A recently rediscovered literature from clinical experience beginning in 1875 describes therapeutic benefit of salicylates in what is now considered to be type 2 diabetes, giving impetus to Shoelson's study of the effect on diabetes of salsalate. The effect of salicylates was confirmed in animal models (7), with evidence that the molecular target of high-dose salicylates is not the COX-1 inhibition of platelet aggregation, as seen with 81-mg doses of aspirin, or COX-1/2 inhibition seen with 650-mg doses, but rather an effect on nuclear factor (NF)-κB seen at high (3–5 g daily) doses. NF-κB is activated by proinflammatory factors including IL-1, the toll-like receptor (TLR), TNF-α, and intracellular factors including oxidative and endothelial reticular stress, ceramides, and various protein kinase (PK)-Cs, with NF-κB in turn having multiple cytokine and receptor effects. Obesity induces inflammation and hence insulin resistance (IR), and salicylates reduce this in animal models. Shoelson asked whether this might be used in not only glycemic treatment, but also the prevention of atherosclerotic complications, showing a study of mice not expressing the LDL receptor, in which early atherosclerotic events seen with a high-fat diet decreased with salsalate treatment.

Salsalate is insoluble at acid pH, hydrolyzed and absorbed in the duodenum, generic, and inexpensive, with an excellent safety profile. The first stage of the Targeting INflammation using SALsalate in type 2 Diabetes (TINSAL-type 2 diabetes) trial included 120 people randomized to 0, 3, 3.5, and 4 g daily for 14 weeks, with baseline age 55, BMI 34 kg/m², diabetes duration 6 years, and A1C ∼7.6% (8). A1C decreased ∼0.4% at 8 and 14 weeks, and fasting blood glucose (FBG) decreased 10–20 mg/dl, while there were 0.1% and 10 mg/dl elevations in the placebo group. Triglycerides fell and adiponectin increased. There was no change in weight, liver function, or electrolytes, and no adverse gastrointestinal effect. Compliance was high, with tinnitus occurring infrequently. A 240-person 1-year trial is now ongoing, with planned coronary calcium as well as metabolic measurements.

Lipids and IR

Ronald Krauss (Berkley, CA) discussed lipoprotein abnormalities in the IRS, reminding the audience of the heterogeneity of LDL particle size and density, with smaller particles having greater atherogenicity. In the Quebec Cardiovascular study, the cholesterol circulating in small LDL particles added to coronary heart disease (CHD) risk independently of nonlipid risk factors, HDL and LDL cholesterol, triglyceride, lipoprotein (a), and apoB levels (9). Levels of small, but not of large, LDL were independently associated with CHD risk in the first 7 years of follow-up, but interestingly not during the subsequent 6 years of the study (10). Analysis of lipoprotein subfractions with nuclear magnetic resonance spectroscopy over 11 years of follow-up of 27,673 participants in the Women's Health Study showed that VLDL, IDL, small LDL, and large HDL were associated with CV risk (11). A systematic review of CV associations of LDL subfractions showed that higher LDL particle number is associated with CVD but failed to show evidence that measures of LDL subfractions add incremental benefit (12). Krauss suggested, however, that this reflects the high degree of intercorrelation of LDL particle measures with traditional risk, which obscures the extent to which these particles are risk mediators (13).

How then, Krauss asked, can we improve clinical risk prediction? He hypothesized that there are distinct clusters of lipoprotein subfractions representing specific mechanisms which independently confer risk. Using a new method of ion mobility to determine lipoprotein particle sizes and concentrations (14), he showed an analysis of the prospective Mälmo cohort study of 4,594 persons who also had traditional lipid profiles with mean follow-up 12.2 years (15). HDL was most strongly associated with risk, then triglyceride, and non-HDL cholesterol, and, in this analysis, LDL itself was not a significant risk factor. Using principal component analysis to make weighted linear combinations of correlated variables to maximize the variation explained, the first component, including 41% of risk, was made up of LDL, IDL, and VLDL and gave apoB-related risk; the second component, with 24% of risk, comprised small and medium LDL and lower levels of large HDL; and the third component, giving 11% of the risk, was principally composed of HDL. The second component, Krauss explained, “captures the phenotype of IR,” both in men and women, and was independent of LDL but not HDL. He asked whether these risk components represent biological pathways, using genetics to analyze their relationships. In genome-wide association studies, selecting single nucleotide polymorphisms (SNPs) associated with lipid changes, he analyzed changes in the three components with SNP genotype, suggesting that if a SNP aligns strongly with one component, it would suggest this gene to be related. Indeed, the first component was related to a set of ApoB, apoE, and LDL-receptor SNPs; the second component with hepatic lipase, LPL, apoA5, and cholesterol ester transfer protein (CETP) SNPs; and the third component with a CETP SNP, suggesting that the components are indeed biologically meaningful, and that the pathways involved in the second component appear to be distinct (16). “Ultimately,” he said, “these will be part of our diagnostic testing.” The second component was strongly predictive of CV risk in both sexes, independent of LDL cholesterol. HDL was related to risk by two of the components, suggesting its actions in taking up cholesterol from macrophages, and, independently, its role as a marker of the IR syndrome.

Thomas Bersot (San Francisco, CA) discussed approaches to recognizing and managing CV risk in insulin resistant (IR) patients with low HDL cholesterol levels. “It's clear,” he said, “that the way we eat and we live in this country is killing us.” In a metaanalysis of multiple studies, mortality was increased by approximately one third during the first year after acute coronary syndrome in patients with diabetes (17). The “Get With The Guidelines” group studied 136,905 persons admitted after a CHD event in 2000–2006. 45% had prior vascular disease or diabetes, but only 21% were taking a lipid-lowering drug, with the LDL 94 in those who were and 108 in those not receiving treatment, while HDL was 39.6 in both groups (18), leading Bersot to observe that “the defining risk factor is low HDL cholesterol concentration.” “Patients are not being identified and appropriately treated,” he said, while “use of lipid-lowering drugs by only 21% … is a tragedy.” Bersot reviewed data from the National Health and Nutrition Examination Survey showing average levels of LDL and of HDL cholesterol 123 and 49.6 mg/dl, respectively, in the general population. Over a 6 year period, LDL decreased from 108 to 103, but HDL fell as well, from 43 to 39, which may reflect the increasing prevalence of obesity.

An important question concerns appropriate lipid targets for persons with diabetes or IR. In the Treating to New Targets (TNT) study, the lesser benefit with atorvastatin 10 mg than with 80 mg daily suggests LDL <70 mg/dl as the target for secondary prevention. Comparing diabetic versus nondiabetic persons in TNT, RR reductions were 20% in both, but absolute risk reduction was 3% versus 1.5%. As low HDL is one of the mediators of risk in diabetic patients, non-HDL cholesterol and the ratio of total-to-HDL cholesterol (goal <3.0) may be the best parameters to follow in reducing CV risk. The ApoB-to-ApoA1 ratio, the absolute value of ApoB (goal <80 mg/dl), and LDL particle number may also be useful in risk assessment.

Bersot recommended use of the most effective statins at maximum doses to reduce LDL and triglyceride-rich lipoproteins in the endeavor to attain LDL and non-HDL targets, and suggested consideration of additional LDL reduction with ezetimibe, niacin, or bile acid sequestrants, and additional triglyceride reduction to lower non-HDL cholesterol with niacin, fibrates, or omega 3 fatty acids. The lipid arm of ACCORD in which simvastatin was administered with either fenofibrate or placebo has been completed, and the possible benefit in the high triglyceride/low HDL cholesterol subgroup somewhat confirms these recommendations (19). Results of the ongoing Atherothrombosis Intervention in Metabolic syndrome with low HDL/High triglyceride and Impact on Global Health Outcomes (AIM-HIGH) and other studies of addition of niacin to a statin will not be available for years. Of course, a crucial aspect of care is change in lifestyle, and it is infrequent for us to successfully encourage a patient to walk 40 min and eat three to four servings of fruits and vegetables daily.

The endothelium and IR

Jorge Plutzky (Boston, MA) discussed endothelial regulation of metabolic as well as vascular function, and hence that there is the potential that endothelial dysfunction itself may cause the abnormalities of IR. Energy balance, particularly of glucose and of lipids/fatty acids, has major consequences in obesity and in the control of glucose and lipid metabolism, and it is noteworthy that atherosclerosis may occur consequence of both. The endothelium has the largest surface area of any organ in the body, and is involved in both systemic and local responses and in a variety of disease processes. The endothelium may then, Plutzky suggested, act as transducer of circulating factors and tissue response—including mechanical forces such as blood pressure, shear stress, pulsatile flow, the cellular blood components, erythrocytes, leukocytes, and platelets—and circulating nutrients, particularly lipids, and toxins. A particularly important mediator released by the endothelium is nitric oxide, having paracrine effect on vasodilation, inflammation, and effects of adhesion molecules. People at risk of developing type 2 diabetes show reduced microvascular hyperaemia (20) and decreased brachial artery reactivity (21), manifestations of abnormal endothelial function with decreased vasodilation (22). Both PPAR-α and -γ are present in the endothelium. Activation of either receptor improves flow-mediated vasodilation as well as inflammatory markers (23). Plutzky described a complex experimental model with specific deletion of PPAR-γ in the endothelium, which had no effect on body weight (or on adipocyte PPAR-γ) but decreased adipocyte size and adipocyte PPAR-γ response, with decreased skeletal muscle but increased liver triglyceride, the latter presumably related to the liver's fenestrated epithelium allowing passive uptake of fatty acids with the animals having high circulating fatty acid and triglyceride levels and abnormal vasomotor function on a high-fat diet (24).

IR and thrombosis

Peter Grant (Leeds, U.K.) discussed the association between diabetes and thrombotic risk, suggesting that diabetes and CVD represent aspects of the same condition underpinned by an “inflammatory atherothrombotic insulin resistance syndrome” related to obesity, with inflammation and thrombosis further enhanced by hyperglycemia. Inflammation and thrombosis should, Grant observed, be considered protective and related processes, with both diabetes and heart disease occurring as results of normal responses in abnormal settings. IR presumably evolved primarily as a defense for energy conservation, and it is interesting that IR causes a hypercoagulable state.

Grant described interesting biologic characteristics of the horseshoe crab, which has existed on earth for 250 million years, with a single vascular cell, the amebocyte, functioning both as a platelet and phagocyte, with bacterial endotoxin releasing amebocyte clotting factors leading to simultaneous clearance of bacteria and rapid clot formation, suggesting a primal linkage of these two processes. The mammalian coagulation cascade, platelet activation process, and immune cells have much more complex characteristics. Fibrinolysis is inhibited by plasminogen activator inhibitor (PAI)-1, while C3 is the central regulatory protein of the complement cascade. Inflammation and thrombosis strongly interact to protect against injury and infection, with the IR state associated with a complex of endothelial dysfunction, macrophage abnormalities, and formation of thrombus involving complement, clotting factors, and platelets. Atherectomy specimens from diabetic patients have more lipid-rich material, with greater degrees of macrophage infiltration and thrombus formation (25). In IR, thrombotic risk is related to adipocyte factors, including local inflammation caused by release of free fatty acids. IR occurs not only in the liver, fat, and skeletal muscle, but there is increasing recognition of its effect also in the endothelium, in macrophages, in platelets, and in cardiac myocytes, all contributing to CVD. Endothelial dysfunction leads to reduction in generation of nitric oxide (NO), which acts as a vasodilator but also decreases platelet stabilization. The IR macrophage develops into a foam cell, and although thrombosis is a late manifestation, increased levels of PAI-1, the primary inhibitor of plasminogen activator, are seen earlier. Adipocyte production of PAI-1 is reduced by weight loss and can be inhibited pharmacologically with pentoxifylline, which inhibits TNF-α. Normoglycemic first-degree relatives of persons with diabetes have doubling of PAI-1 and increased fibrinogen levels (26). Clot structure-function relationships in these persons show progressively longer clot lysis time due to a denser, more tightly branched clot structure, as IR progresses to type 2 diabetes, with clot abnormalities also demonstrated in studies of metabolic syndrome (27). Such denser fibrin clots are associated with worse survival after acute events. Other effects of diabetes include increased antiplasmin binding and decreased plasmin generation, further reducing clot lysis. Grant pointed out that in part these abnormalities are related to glycemia, perhaps related to nonenzymatic glycation of clotting factors, and can be reversed with improved control. Platelet activation is increased by inflammation, potentially leading to a positive feedback cycle with enhanced clot formation. C3, which alters fibrin structure and prolongs the time to clot lysis, increasing thrombotic potential, is associated with IR and CVD, as is C-reactive protein (CRP), which alters endothelial cell function and increases thrombus formation.

Grant discussed approaches to thrombotic risk treatment in diabetes. Improving glycemic control is important, with evidence of pleiotrophic effect of some glucose-lowering medications. Thiazolidinediones reduce thrombin-induced expression of the platelet surface antigen CD40L, reduce PAI-1 levels, reduce platelet activation, reduce CRP, and reduce macrophage CD36 expression, all of which would be expected to improve clinical outcome, although Grant observed that the evidence that this occurs is limited. The role of anticoagulant treatment in diabetes is uncertain. The cyclogenase (COX)-1 inhibitor aspirin inhibits the thromboxane A2 platelet aggregation pathway. A metaanalysis of six primary prevention trials of 95,000 persons did not show an effect on mortality, but in 16 secondary prevention trials of 17,000 persons, there was a 23% reduction in events, from 8.2 to 6.7% (28). The effect of aspirin among diabetic patients is unclear. A study of 2,499 diabetic persons with acute coronary syndrome failed to show evidence of benefit of aspirin administration (29). Among 58,000 Swedish diabetic persons treated with aspirin, there was evidence of increased mortality due to serious bleeding, causing a net of 107 excess deaths in persons without CVD, although preventing 164 deaths among those with CVD (30), leading Grant to state, “We should be very cautious in patients with diabetes for primary prevention.” He reviewed the effect of clopidogrel in the Clopidogrel versus Aspirin in Patients at Risk of Ischemic Events (CAPRIE) substudy of 4,000 diabetic persons, showing evidence of greater effect than aspirin, although the agent should only be used in very high-risk primary prevention. The use of clopidogrel is complicated by evidence that there is a rebound increase in mortality on its withdrawal, which is particularly seen in diabetic patients. Prasugrel, a similar agent, showed evidence of benefit in the subgroup of persons with diabetes (31). Future approaches with novel anticoagulants and anti-inflammatory agents appear promising, with Grant mentioning antiplatelet agents, such as the thromboxane inhibitor picotamide; inhibitors of p2y12 (32); and the thrombin receptor protease-activated receptor-1 (PAR-1) (33).

Glycemia and CVD

Darren McGuire (Dallas, TX) discussed “fallacy, faith, and fact” pertaining to the relationship of glucose control to CVD. In the UK Prospective Diabetes Study (UKPDS), among some 4,000 type 2 diabetic persons followed for a decade, nearly one in four had an atherosclerotic event (34). Inversely, 30–50% of persons with CVD have diabetes, in many cases not diagnosed. CHD risk begins to increase at low glucose levels, well below those used for diagnosis of diabetes. Does glucose-lowering then improve CV risk? In the University Group Diabetes Program (UGDP), cumulative mortality was greatest in persons receiving tolbutamide, leading to a black box warning in the product information for all sulfonylureas.

McGuire noted that earlier studies showed no evidence of macrovascular benefit of glycemic treatment of type 2 diabetes, and that although there was nearly significant reduction in myocardial infarction in the UKPDS, there was a comparable degree of increase in stroke. The UKPDS did suggest significant CV benefit of metformin. Furthermore, McGuire considered PROactive to suggest CV benefit of pioglitazone treatment. Combining PROactive, ACCORD, ADVANCE, VADT, and the follow-up studies of the UKPDS, there is a trend to CV benefit of glycemic control started late in the natural history of diabetes, with a suggestion that greater benefit can be obtained with treatment from the time of diabetes diagnosis. Observational data shows improved outcome among persons receiving insulin sensitizers after acute myocardial infarction (35), but McGuire observed discordance between rosiglitazone versus pioglitazone in being associated with risk versus benefit, respectively. “Beware the product label” he concluded, those of sulfonylureas and rosiglitazone implying risk, while metformin seems sturdily associated with benefit (he did not discuss the sulfonylurea plus metformin UKPDS-subset study, in which CV risk was greater than that of sulfonylureas alone [36]). He also questioned whether glycemia as a surrogate marker is truly associated with reduction in CVD, noting that in the UKPDS, the drug associated with greater CV benefit, metformin, was associated with lesser A1C reduction.

McGuire also presented mixed evidence pertaining to in-hospital glycemic control. Certainly, diabetes and high glucose are associated with adverse outcome, both in general (37) and with increasing blood glucose at the time of hospitalization for acute coronary syndrome (38). The DIGAMI study should not be used as evidence favoring glycemic control, because it aimed for blood glucose levels between 126 and 199 mg/dl (39). Van den Berghe's study showed benefit of intensive glycemic control for patients in a surgical intensive care unit (ICU) (40), but McGuire interpreted the medical ICU study as being negative (41), and the multicenter NICE-SUGAR ICU study showed a significant increase in mortality (42), which, McGuire said, “has led us to make a hasty retreat.” Hypoglycemia is associated with increased mortality (43) and certainly is “a marker of adverse outcome” (if not a mediator). McGuire interpreted the studies to suggest an in-hospital target glucose <180 mg/dl. More intensive glycemic treatment may be reasonable after CABG surgery, although he termed “the data […] not as strong as the cardiovascular surgeons would like us to believe.”