Microvascular complications of diabetes: the role of angiotensin converting enzyme inhibitors Proceedings of a Round Table Meeting, June 1999, San Diego, USA

Abstract

Macrovascular disease including cardiovascular disease, stroke and peripheral vascular disease is three to five times as common in diabetic compared to non-diabetic subjects. It is necessary to adopt a multifactorial approach to prevent or slow the progression of the macrovascular disease in diabetes mellitus. Aggressive screening for glucose intolerance may help detect the early stages of diabetic macrovascular complications. Lifestyle modification may also contribute to the prevention of diabetic complications. In established diabetes, improved glucose control has some impact on slowing the progression of macrovascular disease. However, the most successful approach to manage macrovascular complications in diabetes is aggressive modification of each established risk factor such as hypertension, dyslipidaemia and hypercoagulability using specific lifestyle changes or modifications to each risk. Prevention of microvascular complications is very important with regard to the general protection of the diabetic subject. In the past, most emphasis was placed on the direct effects of hyperglycaemia on the microvascular disease. However, what really kills the majority of diabetic patients is heart disease and ischaemic cardiovascular and cerebrovascular disease. Though much progress has been made in controlling the microvascular component, more work is needed in preventing the increasing incidence of macrovascular disease, which is three to five times that found in the non-diabetic patient. There are numerous potentially reversible risk factors that contribute to macrovascular disease in the type 1 and 2 diabetic patients including hyperglycaemia, hypertension, dyslipidaemia and coagulation abnormalities. In type 2 diabetes, hyperinsulinaemia, insulin resistance and central obesity also contribute to the vascular disease (Table 1). Hyperglycaemia is an independent risk factor for cardiovascular complications in diabetes. The Honolulu Heart Study1 compared the risk of cardiovascular heart disease (CHD) across quintiles of blood glucose ranging from 40 to 532 mg/dl. There was a positive relationship for fatal CHD and for total CHD for each quintile of blood glucose (Figure 1). The Whitehall study examined 18403 males, and showed that CHD deaths rise abruptly once post-prandial glucose levels above the 95th response range are reached.2 In the Islington study of 223 males and females, there was a progressive increase in all CHD prevalence by quintiles of the 2 hour blood glucose and glycated haemoglobin.3 The Bedford survey showed that in 552 male and female subjects that all cause mortality and CHD mortality was higher in diabetic patients than normoglycaemic controls, but more importantly it also demonstrated that borderline diabetic patients were at significant mortality risk.4 Kuusisto et al.5 provided studies to show the same positive relationship between levels of HbA1c and fatal or non-fatal coronary events. In the Pathobiological Determinants of Atherosclerosis in Youth Study (PDAY), 6 a younger group of subjects (aged 15 to 34 years) who died were examined at autopsy. There was a twofold increase in fatty streak formation and a threefold increase in raised vascular lesions found at autopsy in those individuals with a HbA1c greater than 8%. This study suggests that, even in individuals in whom a diagnosis of overt diabetes has not been made, higher levels of glucose provide a setting for initiating the process of atherosclerosis. Honolulu Heart Study Despite the numerous positive studies relating glucose and CHD, it has been more difficult to show whether lowering of glucose in intensive versus standard insulin therapy slows the progression or prevents macrovascular disease in type 1 and type 2 diabetes mellitus. The landmark Diabetes Control and Complications Trial (DCCT)7 published in 1993 conducted in type 1 diabetic patients established for the first time that management of hyperglycaemia decreased the progression of diabetic microvascular disease (retinopathy, nephropathy and neuropathy). The DCCT trial, however, could make few conclusions on the impact of tight glucose control on macrovascular complications. In a similar study of a small population of type 2 diabetic patients in Japan (Kumomoto study), 8 intense insulin control duplicated the findings of the DCCT for prevention of microvascular complications but there was no difference in pooled cardiovascular events for macrovascular disease. The University Group Diabetes Program (UGDP) study9 of 414 diabetic patients with a mean age of 52.7 years, about 20% of the patients already had established cardiovascular disease at entry, showed no benefit of intensive insulin therapy to reduce cardiovascular risk in diabetic subjects. The Veterans Affairs Co-operative Study on Glycaemic Control and Complications in Type 2 Diabetes (VACSDM)10 enrolled 153 older patients with a long duration of their disease. The results were surprising in that they showed that intensive glucose treatment actually increased the pooled cardiovascular events. It was concluded that it might be better to modify insulin therapy in the later stages of diabetes or in subjects who already have multiple complications. On the other hand, immediate insulin therapy after acute myocardial infarction in a diabetic patient may have beneficial effects. In the DIGAMI Study11 it was shown that the very high death rates in the first 3–4 years after MI in diabetes (44%) could be significantly lowered by early intensive insulin management (82% type 2, 19% type 1). Because the effect was quite rapid, it was proposed that insulin's effect might be on platelet function, thrombosis or myocardial dysfunction rather than directly on glucose uptake. It was anticipated that a positive effect of intense glucose control on macrovascular complications would be found in the United Kingdom Prospective Diabetes Study (UKPDS).12 However, the main results showed no difference in pooled cardiovascular events in the intensive glycaemic versus standard control groups. There was a trend towards a reduction in myocardial infarction in the intensive control group but this did not reach significance. Interestingly, in the UKPDS, a sub-group analysis of obese individuals placed on metformin showed a 42% reduction in mortality and a 39% reduction in myocardial infarction. As metformin produced the same degree of glucose lowering as in the insulin/oral agents group, this agent may have beneficial effects independent of glucose lowering. Though it is established that hyperglycaemia is an independent risk factor, in a recent 15 year follow-up study in Finland in type 2 diabetes, patients at 5 and 10 years still had a very high incidence of cardiovascular disease despite tight control of their blood glucose.13 These results suggest that although hyperglycaemia may be a powerful predisposing factor for coronary heart disease, lowering blood glucose alone is insufficient to ensure complete prevention of these complications. Other risk factors must be considered. In diabetes, 40–50% of subjects have an abnormal lipid profile. The most characteristic lipid pattern in diabetes is high serum triglyceride level and a low HDL cholesterol (HDL-C). In addition, there are several associated abnormalities in lipoprotein composition such as increased low density LDL particles, which are atherogenic, as well as increased glycation and oxidation of lipoproteins. Studies such as the Paris Prospective study14 indicate that high triglycerides are a risk marker for cardiovascular disease in diabetic and non-diabetic subjects. Explanations for the risk of high triglycerides include the association with low density LDL and oxidised LDL as well as intermediate density lipoproteins. There are few outcome studies on the effects of triglyceride lowering on cardiovascular risk reduction in diabetes. The Helsinki Heart Study15 examined the effect of the fibric acid derivative gemfibrozil versus placebo on reduction on cardiovascular events in 4081 males (135 diabetic patients). There was a 27% reduction in triglyceride levels and a 1.5% increase in HDL-C during the gemfibrozil treatment period. There was a reduction in MI and cardiovascular death in the gemfibrozil group. The Veterans Affairs HDL Intervention Trial (VA-HIT)16 studied gemfibrozil effect in patients with a low HDL-C and showed a significant risk reduction (22%) for coronary heart disease and non-fatal MI in men with a low HDL cholesterol. Results in diabetic patients from this study are being analysed correctly. The Framingham Heart Study compared fasting lipid profiles in diabetic and non-diabetic men and women.17 Although triglyceride and HDL-C levels differ in diabetes, the levels of total cholesterol and LDL cholesterol levels usually do not differ much between subjects with and without diabetes. Yet, the Framingham Heart Study analysis of diabetic patients also found a much higher risk of CHD in men and women at 30-year examination, which was especially striking for CHD in women aged 35 to 64 years of age.17 Thus, there is probably benefit in treating average or moderately elevated LDL cholesterol in diabetic subjects. Several large intervention studies focused at lowering LDL cholesterol have shown benefit to the diabetic patient. Studies of secondary prevention using the HMGCoA reductase inhibitors, which primarily lower LDL cholesterol, have provided some information on the diabetic patient.18-20 The Scandinavian Simvastatin Study (4S)18 examined 3617 males and 827 females, of whom 202 were diabetic, compared simvastatin to placebo. Simvastatin lowered the LDL-C by 35%, and showed a significant 55% reduction in risk for major CHD events and a 37% reduction for any cardiovascular events in the diabetic patients (Figure 2). The Cholesterol and Recurrent Events (CARE) study18 used pravastatin 40 mg/day in 3583 male and 567 female subjects with CHD and average elevations of in LDL cholesterol. The relative risk of CHD decreased by 25%. The study included 586 diabetic patients, who achieved the same benefits of LDL cholesterol lowering as the non-diabetic group. The Long-Term Intervention with Pravastatin (LIPID) study20 included over 9000 men and women with CHD treated with pravastatin 40 mg/day or placebo. The results showed a 24% decrease in deaths from coronary heart disease in subjects on active therapy. There were 782 diabetic patients in this study. Reduction in cholesterol with simvastatin reduces major coronary events Several primary prevention studies using statins in patients with no history of coronary heart disease have also shown major reductions in cardiovascular event rates. In the West of Scotland Study (WOSCOPS), 21 which included 6595 male subjects treated with pravastatin 40 mg daily, lowering LDL-C by 26%, resulted in a 31% reduction in primary coronary events, of baseline. These results indicated that the agent pravastatin was effective in primary prevention of CHD. In the Air Force/Texas Coronary Atherosclerosis Prevention Study (AFCAPS/TexCAPS)22 5608 males and 997 females with no evidence of heart disease were treated with the HMGCoA reductase inhibitor lovastatin for the primary prevention of CHD. Lovastatin lowered LDL-C by 26% and showed a 37% risk reduction for the development of a first major coronary event. This study included only 155 diabetic patients, but similar trends to those seen in WOSCOPS were observed. The ADA guidelines have recommended that an LDL-C below 100 mg/dl should be achieved in diabetic subjects with existing cardiovascular disease and an LDL-C below 130 mg/dl should be the goal in diabetic patients without cardiovascular disease. Moreover, diabetes itself represents a risk and some feel that diabetic subjects, whether they have heart disease or not, should be treated with a goal of decreasing LDL-C below 100 mg/dl and of lowering triglyceride levels to 200 mg/dl or below. Type 2 diabetes is characterised by an increase in atheroembolic complications related to platelet activation and aggregation. These changes decrease fibrinolytic activity. The pro-coagulant state in type 2 diabetes is characterised by increased levels of platelet aggregation, thromboxane release and decreased fibrinolytic activity, leading to an increase in plasminogen activator inhibitor (PAI-1). This suggests that multiple factors in the diabetic state such as cytokines, insulin, pro-insulin, high glucose and modified LDL and VLDL along with activated platelets act to increase PAI-1. Increased PAI-1 then inhibits tPA, leading to a reduction in plasmin formation with a resultant reduction in fibrinolysis and fibrin deposits (Figure 3). Postulated mechanisms of altered fibrinolysis in diabetes The cyclooxygenase enzyme system is quite sensitive to inhibition, so aspirin and other inhibitors of this system should reverse many features of the pro-coagulant state found in the diabetic subject. Many large population studies included substantial subgroups of diabetic subjects and have demonstrated a benefit in reducing cardiovascular events in both diabetic and normoglycaemic patients. The US Physicians Health Service Study (USPHS)23 was a primary prevention study that used aspirin and showed a highly significant reduction in MI. In the Early Treatment of Diabetic Retinopathy Study (ETDRS), 24 which was a mixed primary and secondary prevention study, there was also a significant reduction for MI as the major end point. The Hypertension Optimal Treatment (HOT) Study25 examined the association of level of diastolic blood pressure lowering on cardiovascular events and included aspirin along with the antihypertensive drug regimen. The study had a large diabetic sub-group in whom the combination of aspirin and progressive reduction of blood pressure offered more protection from cardiovascular events than in the non-diabetic patients. The Anti-Platelet Trial (APT)26 was a meta-analysis of secondary prevention studies with aspirin on vascular events showing a highly significant effect of aspirin in patients at high risk for stroke and MI. Overall, most type 2 diabetic patients benefit from both primary and secondary prevention with aspirin in the same way that the non-diabetic subject would. One difference may be that in the diabetic patient there is a further risk reduction in MI. In diabetes, the correct dose of aspirin is still under debate, studies ranging from using very low to extremely high doses (75 to 975 mg/day). Low dose aspirin is safer and seems to be as effective as higher doses in diabetic subjects. Two recent blood pressure intervention studies that included significant diabetic sub-groups25, 27 have offered even further and stronger proof with regard to protection achieved by controlling hypertension. Additionally, the UKPDS study28 examined the effect of two antihypertensive therapies on several cardiovascular outcomes in type 2 diabetes. The HOT study25 examined three target levels for lowering diastolic blood pressure to 90, 85 and 80 mm Hg and their relation to cardiovascular events. The diabetic sub-group showed an even better relationship between level of DBP lowering and reduction in cardiovascular events than did the non-diabetic group. In the diabetic patients there was a significant fall in the numbers of cardiovascular events for each 5 mm Hg reduction in DBP from 90 to 85 to 80 mm Hg. The Treatment of Systolic Hypertension in Europe (SYST-EUR) study29 involved a large number of elderly diabetic patients with isolated systolic hypertension treated with a long-acting dihydropyridine calcium antagonist. The SYST-EUR showed a more striking reduction in all morbid events and in the total mortality, cardiovascular mortality as well as in all cardiovascular end-points, strokes and all cardiac end-points in the diabetic versus the non-diabetic subject. These very striking results from both studies emphasise the importance of blood pressure control in limiting macrovascular complications in all hypertensive diabetic subjects. The threshold level for blood pressure control in diabetes for many years was 140/90 mm Hg but now the recommended guidelines have been revised 130/85 mm Hg.30, 31 Professor Anthony Barnett Professor of Medicine, University of Birmingham, and Clinical Director of Diabetes and Endocrinology, Birmingham Heartlands Hospital, UK The BRILLIANT and EUCLID studies emphasise the potentially important role for ACE inhibitor therapy in the management of diabetic microvasular disease in hypertensive type 2 diabetic patients and in normotensive type 1 diabetic patients. Indeed, care guidelines for type 1 diabetes should now include the treatment of early stage renal disease (and perhaps also retinopathy) with ACE inhibitors even in normotensive type 1 diabetic subjects. Diabetes is a major public health problem for two reasons. Firstly, there are about 200 million people with diabetes worldwide and the numbers are increasing almost exponentially. The other reason is the long-term vascular complications. Myocardial infarction, for example, is two to three times as common in diabetic as in non-diabetic persons, stroke two to six times; peripheral vascular disease also much more common in diabetic patients. Diabetic microvascular disease also causes much morbidity and mortality and in the western world; diabetic eye disease is the commonest cause of blindness in the working population. Diabetic kidney disease affects about 25% of diabetic patients and is the commonest reason for renal failure and renal dialysis and is also associated with massively increased cardiovascular risk. There are a number of factors associated with pathogenesis of diabetic nephropathy. Haemodynamic factors include an increase in glomerular filtration rate associated with development of glomerulosclerosis and proteinuria. Metabolic factors include glycation/glycosylation of long-lived tissue proteins such as collagen, a possible role for growth factors particularly IGF-1 in abnormal collagen synthesis and also a reduction in proteoglycans and sialic acid, which results in loss of perm-selectivity and the development of proteinuria. Albuminuria and also microalbuminuria are strongly associated with hypertension. In type 1 diabetes patients this is associated with significantly increased risks of renal failure and also very significant increased risks of cardiovascular disease. In type 2 diabetes, albuminuria with hypertension is less strongly associated with later development of renal failure but very strongly associated with cardiovascular risk, with increases of 25- to 100-fold. Interestingly, in non-diabetic subjects, this is also associated with increased cardiovascular risk, suggesting that albuminuria is indicative of a generalised process of damage to blood vessels. Albuminuria, including microalbuminuria, is a predictor of mortality in both diabetic and non-diabetic persons and there is a correlation between blood pressure and urinary albumin excretion rate. Hypertension hastens the progression of renal failure, and reduction of blood pressure retards this and correlates with a reduction in urinary albumin excretion rate and consequent reduction in progression of the disease process (Figure 4). Schematic block diagram of factors associated with diabetic nephropathy Angiotensin II has a role in increased intra-glomerular pressure and renal loss of albumin. In animal models, accumulation of collagen in renal tissue with hypertension is characterised by an activated renin angiotensin system. With ACE inhibition and microalbuminuria in hypertensive type 1 diabetic patients there is a reduction in urinary albumin excretion. This effect is also observed in both normotensive type 1 and hypertensive type 2 diabetic patients. If a type 2 diabetic patient is normotensive with albuminuria, then another cause for albuminuria should be looked for, other than diabetic nephropathy. The effects of ACE inhibitors on urinary albumin excretion rate appear to be independent, in part, of effects on systemic blood pressure. It is thought that the ACE inhibitors are acting locally on the efferent arteriole causing relaxation and hence reduction in intra-glomerular pressure, thereby producing a reduction in proteinuria and renal damage, in addition to their systemic blood pressure lowering effect (Table 2). There are two studies which feature larger numbers of patients and which are longer term than many previous studies. The BRILLIANT study (Blood Pressure, Renal Effects, Insulin Control, Lipids, Lisinopril and Nifedipine Trial)32 was a 12 month double-blind, randomised, multicentre, multinational, parallel-group comparison of lisinopril and sustained release nifedipine in 239 male and 96 female hypertensive (DBP 90–100 mm Hg) type 2 diabetic patients with microalbuminuria or early nephropathy (albumin excretion rate 20–200 µg/min). The aim of treatment was to achieve a diastolic pressure less than 90 mm Hg and to evaluate the effects of these two agents on urinary albumin excretion rate over 12 months. A sub-group of patients also had ambulatory blood pressure monitoring. Blood pressure was measured regularly throughout the 12 month period of the trial, together with urinary electrolytes and urinary albumin excretion rates at 6 months and 12 months. There was a significant fall in blood pressure within one month of starting treatment with both lisinopril and nifedipine, although there was no difference between the two groups throughout the length of the study. In addition, there was also a reduction in urinary albumin excretion rate (µg/min) with lisinopril and nifedipine at 6 months, which was greater in the lisinopril group. At 12 months, the urinary albumin excretion rate had fallen further in the lisinopril group, whilst in the nifedipine group there was a rise back towards baseline. At 12 months, the between treatment difference was 20 µg/min in favour of lisinopril, a highly significant difference (p=0.006). All patients included in BRILLIANT had baseline albumin excretion rates above 20 µg/min and at the end of the study almost 30% of the patients in the lisinopril group were normalbuminuric, i.e. urinary albumin excretion rate less than 20 µg/min, compared with only 12% in the nifedipine group. It is concluded from this study that lisinopril treatment had a greater beneficial effect on urinary albumin excretion rate than nifedipine. The difference could not be explained by changes in blood pressure since these were identical in the two groups, nor could they be explained by differences in metabolic effects of the drugs because again they were also identical. It is believed that the observed difference between treatments may result from specific effects of ACE inhibitors on renal function (Figure 5). BRILLIANT study data The EUCLID study (EURODIAB Controlled Trial of Lisinopril in Insulin-Dependent Diabetes)33 was a European, multicentre, two year, randomised, parallel-design clinical trial of lisinopril and placebo in normotensive patients aged between 20 and 59 years who had type 1 diabetes (defined as diagnosis before the age of 36 and the need for continuous insulin therapy within one year of diagnosis) and who were either normo- or microalbuminuric. According to the WHO criteria they were normotensive with systolic blood pressure less than 155 mm Hg and diastolic blood pressure between 75 and 90 mm Hg. The strength of this study was that it was very large and included 530 patients, of whom two-thirds had baseline and follow-up retinal photographs taken. Patients were classified according to whether they were normoalbuminuric or microalbuminuric: 85% had albumin excretion rates in the normal range, i.e. less than 20 µg/min compared with 15% who were in the microalbuminuric range, 20–200 µg/min. Blood pressure was measured regularly and urinary albumin excretion rate was evaluated every 6 months throughout the 2 years of the study. Diastolic blood pressure in these normotensive patients was reduced both in the placebo group and in the lisinopril group, although more so in the lisinopril group. The treatment difference, 3 mm Hg in favour of lisinopril, was statistically significant and was maintained throughout the rest of the trial. For glycated haemoglobin there was no treatment difference either comparing baseline to the end of the study or between the treatment groups. There was a fall in albumin excretion rate both in the placebo group and in the lisinopril group although this was greater in the lisinopril group; the treatment difference was 18.8% (p=0.03) (Figure 6). However, importantly, in the group who were microalbuminuric at baseline, at 2 years the albumin excretion rate was 49.7% lower in the lisinopril group than in the placebo group, a significant clinical difference. The conclusion from the EUCLID study was that lisinopril was of clinical benefit in normotensive patients with type 1 diabetes with little or no signs of renal disease, although the greatest effect was seen in those with microalbuminuria, and the results suggest that care guidelines for type 1 diabetes should now include the treatment of early stage renal disease with ACE inhibitors even in normotensive patients. EUCLID progression of at least one level of retinopathy by retinopathy level at baseline For retinopathy, retinal neovascularisation is the most worrying feature and this is thought to be due to a response to ischaemia of the retina. Neovascularisation may be induced by local and systemic growth factors, which may include angiotensin II, which is certainly present in the vitreous and in the retina. The mechanism of development of retinopathy includes blood flow changes, capillary leakage and angiogenesis, all of which may be influenced by ACE inhibition. Moreover, if the pathogenesis of nephropathy is similar to retinopathy, one might expect benefit in retinopathy also. Part of the EUCLID study33 evaluated the effect of the ACE inhibitor lisinopril on retinopathy in type 1 diabetic patients. The trial looked at progression of retinopathy; progression to proliferative retinopathy and also the incidence of retinopathy (patients with no retinopathy at the start of the trial who subsequently developed retinopathy). In 354 type 1 normotensive diabetic patients aged between 20 and 59 years from 16 European centres, retinal photography was performed at the beginning and end of the study to enable a comparison after 24 months on lisinopril versus placebo. At baseline, patients were classified according to the EURODIAB Hammersmith Grading System, which lists five levels of retinopathy: no retinopathy, minimal non-proliferative, moderate non-proliferative, severe non-proliferative and proliferative retinopathy or photocoagulation scars. The method used involved assessment of separate lesions against one or two standard photographs. Progression of retinopathy was defined as a change of one level in the worst eye, a change of two levels in the worst eye or as a change to proliferative retinopathy or photocoagulation scars, which were combined for the analysis. The groups were reasonably balanced for severity of retinopathy and baseline characteristics by the treatment allocation. The EUCLID retinopathy data showed that for progression of retinopathy by at least one level, 23% (39/166) of patients in the placebo group progressed by at least one level compared with only 13% (21/159) in the lisinopril group, giving an odds ratio of 0.5 (p=0.02). For progression of retinopathy by at least two levels, which represents a marked progression of retinopathy, 7% (11/166) of patients in the placebo group progressed by at least two levels compared with only 2% (3/157) in the lisinopril group, giving an odds ratio of 0.27 (p=0.05). Dramatically, for progression to proliferative retinopathy, 7% (11/166) of patients in the placebo group progressed to proliferative retinopathy compared with only 1% (2/159) in the lisinopril group, giving a very low odds ratio of 0.18 (p=0.03). The incidence of retinopathy in those with no retinopathy at baseline was 24% (15/62) of patients in the placebo group and only 18% (13/72) in the lisinopril group, giving an odds ratio of 0.69 in favour of lisinopril (p=0.4) (Figure 7). EUCLID progression of at least one level of retinopathy by retinopathy level at baseline Looking at the results by retinopathy status at baseline, progression of at least one level of retinopathy in those who started with no retinopathy was 24% in the placebo group compared with 18% in the lisinopril group. However, in those with mild, non-proliferative retinopathy at baseline, 22% of the placebo group progressed by at least one level of retinopathy compared with only 9% of the lisinopril group. Finally, in the group with moderate and severe non-proliferative retinopathy at baseline, the differences were even more striking: 26% progressed in the placebo group compared with only 10% in the lisinopril group. The greatest benefit from use of lisinopril was in those who had good glycaemic control. In the well controlled group (glycated haemoglobin less than 7%) compared with a less well controlled group, 18% of patients in the placebo group progressed compared with only 7% in the lisinopril group (odds ratio 0.34) (Figure 8). EUCLID progression of at least one level of retinopathy by glycaemic control at baseline Thus, in summary, lisinopril halved the progression of retinopathy in type 1 diabetic patients over 2 years; progression to proliferative retinopathy was also reduced and the incidence of new retinopathy was reduced by 30% (not a significant difference; p=0.4). In conclusion, there is now strong evidence that ACE inhibitors can reduce the rate of progression of diabetic renal disease in both type 1 and type 2 diabetes. In addition, evidence from the EUCLID trial suggests they may also be beneficial in slowing the