Part 7. Lipid Control in Type 2 Diabetes

Transcription

1 National Evidence Based Guidelines for the Management of Type 2 Diabetes Mellitus Part 7 Lipid Control in Type 2 Diabetes Prepared by the Australian Centre for Diabetes Strategies Prince of Wales Hospital, Sydney for the Diabetes Australia Guideline Development Consortium APPROVED BY THE NHMRC 16 SEPTEMBER 2004

2 Table of Contents PART Lipids Expert Working Group Guideline for Lipid Control in Type 2 Diabetes Introduction Issues for Lipid Control in Type 2 Diabetes Summary of Recommendations Recommendations...9 Section 1: Lipid abnormalities associated with Type 2 diabetes...9 Section 2: Effect of diet and exercise on lipids...42 Section 3: Effect of improved blood glucose control on lipids...78 Section 4: Effects of treatment with lipidmodifying agents on lipids...94 Section 5: Effects of treatment on outcomes References Evidence References General References Other References Identified Lipid Control Search Strategy and Yield Table...223

3 1.0 Lipid Control Guideline Expert Working Group Chairperson Content Experts Professor James Best Department of Medicine University of Melbourne MELBOURNE VIC Dr Harvey Newnham Department of Medicine Monash University MELBOURNE VIC A/Professor Richard O Brien Diabetes Unit Monash Medical Centre MELBOURNE VIC A/Professor Gerald Watts Department of Medicine Royal Perth Hospital PERTH WA A/Professor Tony Dart Baker Medical Research Institute MELBOURNE VIC ADEA Consumer RACGP Content and Methods Adviser Ms Gloria Kilmartin Department of Diabetes & Endocrinology Royal Melbourne Hospital MELBOURNE VIC Dr Graham Giles Cancer Control Research Institute Cancer Epidemiology Centre MELBOURNE VIC Dr Peter Harris School of Medical Education University of NSW SYDNEY NSW Professor Stephen Colagiuri Department of Endocrinology Prince of Wales Hospital SYDNEY NSW 3

4 Research Officers Dr Nasseem Malouf Diabetes Centre Prince of Wales Hospital SYDNEY NSW Ms Robyn Barnes Public Health Unit South Eastern Sydney Area Health Service SYDNEY NSW Ms Sarah Smith Human Nutrition Unit University of Sydney SYDNEY NSW Ms Caroline George Human Nutrition Unit University of Sydney SYDNEY NSW 4

5 2.0 Guideline for Lipid Control 2.1 Introduction Aim of the Guideline This guideline addresses issues relating to the control of lipid levels in Type 2 diabetes. Lipid abnormalities in people with in Type 2 diabetes can be broadly categorised into 2 groups: those common to the general population such as elevated total and LDL cholesterol and additional diabetes related abnormalities such as elevated triglycerides and reduced HDL cholesterol The aim of the guideline is to assist health practitioners, principally general practitioners, to effectively and efficiently detect and manage lipid abnormalities in those with Type 2 diabetes. Quality Assurance The methods used to identify and critically appraise the evidence in order to formulate the guideline recommendations are described in detail in Part 1 of the document. Additional steps were taken to assure the quality of the Lipid Control Guideline eg: the Project Management Team reviewed and checked each step of the methods process selected searches were repeated the chairman of the Lipids Control EWG double reviewed the majority of the articles used as evidence references the Medical Advisor reviewed and revised the entire final draft the Project Management Team checked all recommendations and evidence statements and compiled and checked the evidence tables, reference lists, and search strategy and yield tables Guideline Format Issues identified by the EWG and from the literature as critical to the control of lipids in Type 2 diabetes are shown in point (next page). Each of these issues is addressed in a separate section in a format presenting: Recommendation(s) Evidence Statements supporting the recommendations Background to issues for the guideline Evidence detailing and interpreting the key findings Summary of major evidence found Evidence Tables summarising the evidence ratings for the articles reviewed For all issues combined, supporting material appears at the end of the guideline topic and includes: Evidence references General references Other references identified Search Strategy and Yield Tables documenting the identification of the evidence sources. 5

6 The prevention of macrovascular disease is a major goal in the care of the person with Type 2 diabetes. Multifactorial intervention is the key to the prevention of macrovascular disease. This document should be considered in association with the Macrovascular Disease, Blood Pressure Control, and Blood Glucose Control Guidelines 6

7 2.2 Issues for Lipid Control in Type 2 diabetes What lipid abnormalities are associated with Type 2 diabetes and what are the consequences? What are the effects of diet and exercise on lipids in people with Type 2 diabetes? What is the effect of improved blood glucose control on lipids in Type 2 diabetes? What are the effects on lipids of treatment with lipidmodifying agents and hormone replacement therapy in Type 2 diabetes? Does treatment with lipid modifying agents or hormone replacement therapy improve outcomes in Type 2 diabetes? 7

8 2.3 Summary of Recommendations Recommendations People with Type 2 diabetes should have measurement of total cholesterol, triglycerides and HDL cholesterol levels and calculation of LDL cholesterol If feasible, lipid levels should be measured after a 10 to 12 hour overnight fast More than one lipid measurement should be used to make decisions to initiate or intensify lipid modifying therapy Specific dietary advice should be given to all people with Type 2 diabetes and elevated lipids which emphasises weight reduction in the overweight In people with Type 2 diabetes whose lipid levels are above target and who have unsatisfactory diabetes control, efforts should be made to improve their blood glucose control before considering lipid modifying medication Statins should be used to lower LDL cholesterol when triglycerides are normal or only slightly elevated. For severe hypercholesterolaemia, a bile acid binding resin or low dose nicotinic acid may be added. Fibrates should be used as first line therapy in people with predominantly elevated triglycerides and low HDL cholesterol with normal to slightly elevated LDL cholesterol levels. If treatment with a fibrate is not tolerated or if additional triglycerides lowering effect is required, fish oil can be used but the effect on diabetes control should be monitored. Treatment with a statin and fibrate should be considered in people with moderate to marked elevation of both LDL cholesterol and triglycerides. Because of the increased risk of myositis, the person should be fully informed and carefully monitored. People with Type 2 diabetes who have an LDL cholesterol >2.5mmol/L after interventions to modify lifestyle and improve blood glucose control, should be considered for statin therapy People with Type 2 diabetes who have triglycerides >2.0mmol/L after interventions to modify lifestyle and improve blood glucose control, should be considered for fibrate therapy 8

9 2.4 Recommendations Section 1: Lipids Issue What lipid abnormalities are associated with Type 2 diabetes and what are the consequences? Recommendations People with Type 2 diabetes should have measurement of total cholesterol, triglycerides and HDL cholesterol levels and calculation of LDL cholesterol If feasible, lipid levels should be measured after a 10 to 12 hour overnight fast More than one lipid measurement should be used to make decisions to initiate or intensify lipid modifying therapy 9

10 Evidence Statements Triglycerides levels are increased in Type 2 diabetes Evidence Level III2 HDL cholesterol is decreased in Type 2 diabetes Evidence Level III2 LDL cholesterol is similar in Type 2 diabetes and the general population, but LDL particle size is smaller Evidence Level III2 Total cholesterol is similar in Type 2 diabetes and the general population Evidence Level III2 There are intraindividual variations over time in lipid and lipoprotein levels in Type 2 diabetes Evidence level III2 Apolipoprotein A1 is reduced, apolipoprotein B is increased, and lipoprotein(a) is not altered in Type 2 diabetes Evidence Level III2 Lipid abnormalities are accentuated by increased body weight, poor glycaemic control and diabetic renal disease Evidence Level III2 There is insufficient evidence to draw any conclusion about the importance of lipoprotein oxidation in Type 2 diabetes Evidence Level III2 Total cholesterol and LDL cholesterol predict cardiovascular disease in Type 2 diabetes Evidence level III2 HDL cholesterol and triglycerides predict cardiovascular disease in Type 2 diabetes Evidence level III2 The contribution of increased lipoprotein (a) to risk of cardiovascular disease in Type 2 diabetes is uncertain Evidence level III2 There is limited evidence linking lipid levels and diabetic nephropathy Evidence level III2 There is limited evidence that lipid levels predict the development of hard exudates in diabetic retinopathy Evidence level III2 10

11 Background Lipid levels and Type 2 diabetes Lipid abnormalities are common in people with Type 2 diabetes and are a major contributor to the increase in cardiovascular disease experienced by people with Type 2 diabetes. Lipid abnormalities in Type 2 diabetes can be broadly categorised into 2 groups: Those which are common to the general population eg elevated total and LDL cholesterol Additional diabetes related abnormalities eg elevated triglycerides and reduced HDL cholesterol Insulin resistance and central obesity are two closely linked factors (Brunzell & Hokanson, 1999) that are important in determining the additional lipid abnormalities found in Type 2 diabetes. The role of insulin resistance in the pathogenesis of Type 2 diabetes was recognised by Himsworth (1936) over 60 years ago. Central obesity was recognised as a risk factor for the development of Type 2 diabetes over 40 years ago (Vague, 1956) and many subsequent studies have confirmed this association. However, it was not until quantitative tests for insulin resistance were available that the importance and prevalence of insulin resistance was fully appreciated (Reaven, 1988). Under the heading of Syndrome X, Reaven linked insulin resistance with increased verylowdensity lipoprotein (VLDL) triglycerides, decreased highdensity lipoprotein (HDL) cholesterol levels and hypertension. Subsequently, the presence of small, dense low density lipoprotein (LDL) particles has been added to the list of lipid abnormalities associated with insulin resistance (Reaven et al, 1993). The association of lipid abnormalities with insulin resistance has led to an appreciation that disordered lipid metabolism is a fundamental aspect of Type 2 diabetes (McGarry et al, 1992). There is also strong evidence linking central adiposity with insulin resistance (Carey et al, 1996; Karter et al, 1996) so that central obesity is now considered an integral part of the insulin resistance syndrome (Haffner, 1996). Insulin resistance and central obesity have the potential to cause lipid abnormalities by several mechanisms (Garg, 1996; Brunzell & Hokanson, 1999). Impaired insulin action allows greater free fatty acid release from an increased mass of intraabdominal adipose tissue, promoting hepatic triglycerides synthesis and VLDL production. At the same time, lipoprotein lipase activity, and therefore triglycerides clearance, is reduced by insulin resistance. Reduction of HDL cholesterol levels has been attributed to triglycerides enrichment by cholesterol ester transfer protein (CETP) and also to an increase of hepatic triglycerides lipase activity, related to insulin resistance. The concept that these lipid abnormalities are linked with insulin resistance and central adiposity rather than with Type 2 diabetes per se has been supported by the finding of elevated triglycerides and reduced HDL cholesterol levels in individuals who subsequently develop diabetes (Haffner et al, 1990; McPhillips et al, 1990). Reduced HDL cholesterol levels in firstdegree relatives of people with diabetes have also been found (Stewart et al, 1995; Shaw et al, 1999). There is abundant epidemiological and clinical trial evidence in people without diabetes that lipoproteins play a major role in the pathogenesis of atherosclerotic vascular disease. The Framingham study showed a strong, graded, continuous positive relationship between total cholesterol and all clinical manifestations of coronary heart disease (CHD) risk (Kannel et al, 1988). This study also showed a very similar direct relationship between LDL cholesterol and CHD risk, reporting that a 1% increase in LDL cholesterol is associated with slightly more than a 2% increase in CHD over 6 years (Wilson, 1990). Similarly the MRFIT (Multiple Risk 11

12 Factor Intervention Trial) study showed a strong, graded, positive relationship between death from CHD and total cholesterol above levels of 4.65 mmol/l for 316,099 white men followed for an average of 12 years (Neaton & Wentworth, 1992). During the course of the study there were 6,327 deaths from CHD. Death rates for men in the three categories of total cholesterol levels below 4.65 mmol/l (3.1; 3.6; 4.1 mmol/l) were similar (ranging from 7.7 to 8.9 deaths per 10,000 personyears). For cholesterol levels above 4.65 mmol/l ageadjusted death rates due to CHD gradually increased to a peak of 54.5 deaths per 10,000 personyears for cholesterol levels of 8.28 mmol/l or above (Neaton & Wentworth, 1992). HDL cholesterol is inversely related to triglycerides levels and both parameters have been linked with CHD risk in large prospective studies, the data being more consistent for low HDL cholesterol level and increased risk of CHD than for high triglycerides. In 12 years of followup for 2,748 Framingham Heart Study participants the relative risk of death from CHD was increased 4.1fold for men with HDL cholesterol <0.9 mmol/l (lowest quintile) compared with men with HDL cholesterol >1.4 mmol/l (highest quintile). For women, death from CHD was increased 3.1fold when HDL cholesterol was <1.2 mmol/l compared with HDL cholesterol >1.8 mmol/l (Wilson et al, 1988). Over 6 years, a 1% lower HDL cholesterol level was associated with a 34% higher risk of CHD (Wilson, 1990). The importance of HDL cholesterol level as a predictor of CHD has also been shown in the MRFIT study (Multiple Risk Factor Intervention Trial Research Group, 1986), the placebo arm of the Helsinki Heart Study (Manninen et al, 1990) and the Prospective Cardiovascular Munster (PROCAM) study (Assmann & Funke, 1990). There has been divided opinion about the importance of plasma triglycerides levels as an independent risk factor for CHD. The evidence from three major prospective studies supports a role for triglycerides as a synergistic risk factor. The Framingham Heart Study showed that over a 14year period, triglycerides >1.7 mmol/l predicted increased CHD risk in men and women with HDL cholesterol <1.03 mmol/l after adjustment for age, systolic blood pressure, body mass index (BMI), low HDL, and electrocardiographically determined left ventricular hypertrophy (Castelli, 1992). For the 2,045 men in the placebo arm of the Helsinki Heart Study, triglycerides >2.3 mmol/l in conjunction with LDL cholesterol to HDL cholesterol ratio (LDL/HDL) >5 were associated with increased risk of death from CHD, with a relative risk (RR) of 3.82 (CI ), whereas in people with triglycerides 2.3 mmol/l and LDL/HDL >5, the increased risk was only 1.19 (CI ) (Manninen et al, 1992). The PROCAM study found very similar results (Assmann et al, 1998). People who had major coronary events (n=258) over 8 years had higher triglycerides levels and a greater LDL/HDL compared with those without major coronary events (n=4,381) (1.98 v 1.5. mmol/l, p<0.001; 4.6±1.6 v 3.4±1.2, p<0.001, respectively). Therefore, the combined data support a role for triglycerides level above 2.0 mmol/l as a risk factor when the LDL/HDL is > 5. In Type 2 diabetes the risk of CHD is significantly increased compared with the nondiabetic population. The key question regarding lipids and the increased CHD risk in Type 2 diabetes is whether the lipid and lipoprotein predictors of CHD in the general population operate in the same way in Type 2 diabetes. Lipid levels are not a universally recognised risk factor for cerebrovascular disease. A recent metaanalysis failed to find hypercholesterolaemia to be a risk factor for stroke in the general population (Prospective Studies Collaboration, 1995). On the other hand recent evidence indicates a reduced risk of stroke after lipidlowering interventions with statins in people with 12

13 preexisting CHD (Sacks et al, 1996; LIPID Study Group, 1998; White et al, 2000; Heart Protection Study Collaborative Group, 2002a). A link between lipid levels and diabetic retinopathy and progression of diabetic nephropathy has been suggested and this is also reviewed in this Section. Evidence Lipid levels and Type 2 diabetes Triglycerides levels are increased in Type 2 diabetes Elevation of triglycerides levels in people with Type 2 diabetes compared with those without diabetes has been consistently reported by most populationbased and clinicbased surveys (Table 1). In most of the studies summarised in Table 1, average triglycerides levels were similar in men and women, although some studies indicate that hypertriglyceridaemia may be more marked in women with Type 2 diabetes than in men (Siegel et al, 1996). In people with Type 2 diabetes, average triglycerides levels were approximately 2.2 mmol/l compared with 1.6 mmol/l in the nondiabetic population. The difference in prevalence of elevated triglycerides levels is illustrated by the following sample of studies. Data from the Second National Health and Nutrition Examination Survey (NHANES II) (Cowie et al, 1994) showed that the prevalence of hypertriglyceridaemia (>2.8 mmol/l) was 34% in men aged 4069 years with diabetes, compared with 11% in controls and 30% in women with diabetes compared with 6% in controls, but significance values were not stated. Another USbased population study, the Framingham Offspring Study, (Siegel et al, 1996) found that the prevalence of hypertriglyceridaemia (>2.75 mmol/l) was 23% in 120 men with diabetes compared with 9% in 1,878 controls (p<0.001) and 29% in 54 women with diabetes compared with 3% in 1,879 controls (p<0.001). Marked hypertriglyceridaemia (>5.5 mmol/l) was not more prevalent in men with diabetes (1.1% v 1.5% in controls), but was more prevalent in women with diabetes (11% v 0.2% in controls, p<0.001). In a Finnish populationbased study (Salomaa et al, 1992) triglycerides were elevated (>2.3 mmol/l) in 48% of 42 men with diabetes compared with 22% of 32 men with impaired glucose tolerance (IGT) and 15% of 136 controls (p=0.008). Elevated triglycerides were found in 52% of 54 women with diabetes compared with 26% of 70 women with IGT and 11% of 187 controls (p=0.0001). Overall studies show that the crude prevalence of elevated triglycerides levels in people with diabetes is approximately 30% which is some 3fold higher than the prevalence in people without diabetes. Studies of triglycerides levels after a fatty meal show increased levels in Type 2 diabetes (Cavallero et al, 1994; Curtin et al, 1994). In the diabetic group (n=14), people with hypertriglyceridaemia (n=8) had higher postprandial triglycerides levels compared with those with normotriglyceridaemia (n=6) (3.28±0.70 v 1.20±0.46 mmol/l, p<0.001), whereas diabetic people with normotriglyceridaemia (n=6) and controls (n=12) had a similar triglycerides levels after the test meal which contained 45 to 55% fat (Cavallero et al, 1994). Curtin et al (1994) also reported a significantly different pattern for triglycerides (p<0.05) following a fatrich meal in the diabetic group (n=6) compared with the control group (n=6). Triglyceridesrich lipoprotein fraction were significantly higher after the fatrich meal at all time points in people with diabetes (p<0.01). This postprandial hyperlipidaemia has been interpreted as indicating reduced chylomicron clearance and accumulation of triglyceridesrich chylomicron remnants in Type 2 diabetes. 13

14 Table 1: Lipid levels and Type 2 diabetes Reference Population Studied Mean Total Cholesterol Billingham, 1989 (UK) Burchfiel, 1990 (US) Byrne, 1994 (UK) Cowie, 1994 (US) Haffner, 1994 (US) Hughes, 1998 (Singapore: Chinese, Indians, Malays) Laakso, 1985 (East Finland) n=130 mean age 57yrs New Type 2 diabetes: 22 Established Diabetes Diet treated: 30 Sulphonylureas: 28 Insulin: 17 Control: 33 n=856, 2074yrs, Anglo (A) & Hispanic (H), Type 2 Diabetes: 121 M; 158 W Control: 219 M; 269 W n=1,156, age 4064yrs Type 2: 23 M; 28 W Control: 129 M; 181 W n= 751M; 1,987W NHANES II age 2074 yrs Type 2 diabetes Control San Antonio Heart Study Type 2: 33 M 62 W Control: 155 M 216 W Singapore Heart Study; age 3069yrs Type 2 diabetes 72M; 54W Control: 248 M; 282 W Age: 4564; Type 2 Diabetes (diet): 48 M; 40 W Sulphonylurea: 56M; 49W SU/Metformin: 14M; 15W Insulin: 17 M; 38 W Control: 65 M; 59 W Between groups comparison p<0.05, p<0.01, p<0.005, p< (v control) (v control) 5.7 M A, H 5.42, , 5.60 W A, H 5.90, , 5.82 Mean LDL Cholesterol M A, H 3.48, , W A, H 3.52, , 3.62 Lipid levels Mean HDL Cholesterol M A, H 0.96, , W A, H 1.24, , 1.44 Mean Triglycerides M A, H 2.51, , M M W W M M W W M M (all v control) W W (all v control) M M (all v control) W A, H 2.56, , W W (all v control)

15 Table 1: Lipid levels and Type 2 diabetes Reference Population Studied Mean Total Cholesterol Laakso, 1990 (Finland) Manzato, 1993 (Italy) Mattock, 1988 (UK) Nielsen, 1993 (Denmark) Niskanen, 1990 (Finland) age 4564yrs Type 2: 152 M; 153 W BMI <25: BMI 2527 BMI >27 Control: 65 M; 59 W BMI <25: BMI 2527 BMI >27 mean age 67yrs Type 2 Diet: 54 Sulphonylurea: 26 SU/Fenformin: 17 Insulin: 23 Control: 30 Type 2 diabetes; 82M; 59W Age 3564yrs 30M; 7W per gp age 61yrs Type 2+Normoalbuminuria +Microalbuminuria +Macroalbuminuria Control Type 2; mean age 56yrs + microalbuminuria: n=12 micoralbuminuria: n=101 Control M W (v control) M W 6.08 Age: 4564yrs M W Type 2: 42M; 54 W Salomaa, 1992 (v control) (v control) (Finland) IGT: 32 M; 70 W Control: 136 M; 187 W Age >65yrs Scheffer, 2003 Type 2 diabetes, n= (The Netherlands) Controls, n= Between groups comparison p<0.05, p<0.01, p<0.005, p< AU: microalbuminuria Lipid levels Mean LDL Mean HDL Cholesterol Cholesterol M M W W M (v BMI <25) M 0.98 M 1.1 (v control) (v AU) (v AU) W (v BMI <25) W 1.16 W 1.3 (v control) Mean Triglycerides M W (v BMI <25) (v BMI <25) M W (v control) 2.01 (v control) 1.09 M 2.7 (v control) (v AU) (v AU) W 2.7 (v control)

16 Table 1: Lipid levels and Type 2 diabetes Reference Population Studied Mean Total Cholesterol Siegal, 1996 (US) Suranti, 1992 (France) UKPDS 11, 1994 (UK) Uusitupa, 1986 (East Finland) Framingham offspring Diabetes (1+2): 120M; 54W Control: 1,878M; 1,879W Type 2 diabetes; n=380 normoalbuminuria microalbuminuria Age 52yra, New Type 2 Diabetes: 297 M; 210 W Control: 52 M; 143 W Age: 4564yrs Type 2: 70 M; 63 W Control: 62 M; 82 W Between groups comparison p<0.05, p<0.01, p<0.005, p<0.001 M M M W W W Lipid levels Mean LDL Mean HDL Cholesterol Cholesterol M M M W M W W M M W W W Mean Triglycerides M W M M W W

17 HDL cholesterol is decreased in Type 2 diabetes A reduced HDL cholesterol in people with Type 2 diabetes has been demonstrated in many populationbased and clinicbased surveys. Of the studies summarised in Table 1, average HDL cholesterol levels were approximately mmol/l higher in women than in men. In people with Type 2 diabetes, average HDL cholesterol levels were approximately 0.2 mmol/l lower than in the nondiabetic population. The difference in prevalence of decreased HDL cholesterol levels is illustrated by the following sample of studies. In the Framingham Offspring Study (Siegal et al, 1996) the prevalence of low HDL cholesterol (<0.9 mmol/l) was 44% in diabetic men compared with 20% in controls (p<0.001) and 38% in diabetic women compared with 9% in controls (p<0.001). NHANES II data (Cowie et al, 1994) showed HDL cholesterol was <0.9 mmol/l in 28% of diabetic men compared with 14% of controls and 10% in diabetic women compared with 6% in controls. In a Finnish populationbased study (Salomaa et al, 1992), HDL cholesterol was <1.02 mmol/l in 55% of men with diabetes compared with 25% of men with IGT and 20% of controls. HDL cholesterol was <1.2 mmol/l in 52% of women with diabetes compared with 31% of women with IGT and 18% of controls. In general studies show that the crude prevalence of reduced HDL cholesterol levels in people with diabetes is approximately 2fold higher than in people without diabetes. LDL cholesterol is similar in Type 2 diabetes and the general population, but LDL particle size is smaller LDL cholesterol concentration can be determined in the following ways: Calculated: The approach most frequently used in the reviewed studies was to separate VLDL and HDL particles by ultracentrifugation or by precipitation of nonhdl particles. Cholesterol content of the VLDL and HDL fractions was measured and LDL cholesterol concentration was calculated by subtracting VLDL cholesterol and HDL cholesterol from total cholesterol concentration (Laakso et al, 1985; Billingham et al, 1989; UKPDS 11, 1994; Siegel et al, 1996). The main alternative approach is to calculate LDL concentration using the Friedewald equation (Burchfiel et al, 1990; Hughes et al, 1998; Tsalamandris et al, 1998). This method estimates the cholesterol content of the triglyceridesrich lipoproteins as triglycerides concentration in mmol/l 2.2 (providing triglycerides 4.5 mmol/l) and calculates LDL cholesterol by subtracting this value and HDL cholesterol from total cholesterol (Friedewald et al, 1972). Direct measurement: Less commonly, LDL particles are separated by ultracentrifugation and cholesterol concentration measured directly (Manzato et al, 1993). There is a good correlation between calculated and measured LDL cholesterol in people with Type 2 diabetes (Branchi et al, 1998), but the more direct measures are preferred for epidemiological studies. Most studies have found similar LDL cholesterol levels in people with Type 2 diabetes and people without diabetes (Table 1). Average LDL cholesterol levels in the Caucasian populations were 3.6 mmol/l for men with diabetes, 3.8 mmol/l for women with diabetes, 3.6 mmol/l for nondiabetic men and 3.7 mmol/l for nondiabetic women. Also the prevalence of elevated LDL cholesterol is similar. In the Framingham Offspring Study 17

18 (Siegel et al, 1996), LDL cholesterol levels 4.2 mmol/l were present in 22% of men with diabetes compared with 27% of controls and in 35% of women with diabetes compared with 22% of controls (p=ns). NHANES II data (Cowie et al, 1994) showed LDL cholesterol was 4.2 mmol/l in 38% of diabetic men compared with 39% of controls and 52% in diabetic white compared with 35% in controls (the 95% CI were very wide for each of these figures). However, consistent with the strong inverse correlation between triglycerides level and LDL particle size, diabetes is associated with smaller LDL particle size. In the Framingham Offspring Study a weighted LDL particle score 3.5 (small, dense particles) was present in 49% of women with diabetes compared with 33% of controls (p<0.05) and in 40% of men with diabetes compared with 11% of controls (p<0.001). The presence of small dense LDL particles was significantly associated with diabetes and hypertriglyceridaemia in both sexes (OR 1.79, p=0.002 for men; OR 5.27, p< for women) (Siegel et al, 1996). In the San Antonio Heart Study of mainly MexicanAmerican subjects (Haffner et al, 1994), LDL particle size was smaller in diabetic men and women, compared with nondiabetic men and women (252.2±1.8 v 256.1±0.8 Å, p=0.048 for men; 254.7±1.3 v 259.7±0.7 Å, p=0.005 for women). Also the percentage of small LDL particles was higher in people with diabetes (men 52%; 49% women) than in people without diabetes (38%; 33%, respectively). After adjustment for triglycerides and HDL cholesterol levels, LDL size was only significantly smaller in women with diabetes than in women without diabetes (p=0.043). In a casecontrol study, Scheffer et al (2003) examined the relationship of in vitro LDL oxidisability (ie lag time) and circulating in vivo oxidised LDL with LDL particle size in 58 elderly people (aged >65 years) with wellcontrolled Type 2 diabetes and 58 agematched controls with normal glucose metabolism. Mean LDL cholesterol level was similar in people with and without diabetes (3.7±0.9 v 3.9±0.9 mmol/l, p=0.36), but the LDL particle size was significantly smaller in people with diabetes (21.3±0.5 v 21.7±0.3 nm, p<0.001). Although lag time did not differ between the diabetes and control group (71.5±9.6 v 71.7±7.7 min, p=0.94), the diabetes group had a nonsignificantly higher level of circulating in vivo oxidised LDL (87.2±26.9 v 81.0±22.1 U/L, p=0.18). The LDL particle size was found to be inversely associated with oxidised LDL in people with diabetes (r= 0.35, p=0.007). This relation remained strong after controlling for LDL cholesterol level (r= 0.52, p<0.001). Total cholesterol is similar in Type 2 diabetes and the general population Total cholesterol concentration represents the sum of LDL, HDL and VLDL cholesterol, with contributions usually ranking in that order. Lower HDL cholesterol levels in Type 2 diabetes reduce total cholesterol, while higher triglycerides levels are associated with more VLDL cholesterol and so increase total cholesterol. The balance of total cholesterol represents LDL cholesterol, the concentration of which is usually unaltered by diabetes. Because diabetes induced changes in HDL cholesterol and VLDL cholesterol are in opposite directions and LDL cholesterol is unaltered, total cholesterol levels in diabetic subjects are similar to nondiabetic populations in most studies (Table 1). Of the studies summarised in Table 1, average total cholesterol levels in the Caucasian populations were 6.2 mmol/l for men with diabetes, 6.6 mmol/l for women with diabetes, 6.2 mmol/l for nondiabetic men and 6.4 mmol/l for nondiabetic women. Overall the prevalence of elevated total cholesterol is similar in diabetic and nondiabetic people. For example, NHANES II data (Cowie et al, 1994) showed total cholesterol was 6.2 mmol/l in 46% of diabetic men compared with 35% of controls and 49% in diabetic women 18

19 compared with 40% in controls. In the Finnish survey by Salomaa and colleagues (1992), total cholesterol was >6.9 mmol/l in 33% of men with diabetes compared with 31% of men with IGT and 21% of controls. Total cholesterol >7.3 mmol/l was found in 11% of women with diabetes, 37% with IGT and 24% of controls (Salomaa et al, 1992). However some studies have noted some differences in total cholesterol levels in diabetes. The Framingham Offspring Study (Siegel et al, 1996) reported total cholesterol 6.2 mmol/l in 18% of men with diabetes compared with 23% of controls (p=ns) and in 41% of women with diabetes compared with 23% of controls (p<0.001). Implications for Clinical Practice People with Type 2 diabetes frequently have higher triglycerides levels and lower HDL cholesterol levels compared with the general population. Total and LDL cholesterol levels are similar in people with and without diabetes and the prevalence of elevated total and HDL cholesterol levels are also similar. As reviewed in other Sections of this Guideline, elevated total cholesterol, LDL cholesterol and triglycerides, and a low HDL cholesterol all predict an increased risk for coronary heart disease in Type 2 diabetes. Therefore, a complete assessment of lipid levels in people with Type 2 diabetes requires the measurement of total cholesterol, triglycerides and HDL cholesterol and calculation of LDL cholesterol. There are intraindividual variations over time in lipid and lipoprotein levels in Type 2 diabetes. Intraindividual variations of lipid and lipoprotein levels are observed in people with Type 2 diabetes. This intraindividual variability is derived from both analytical variability (measurement error) and biological variability with the latter accounting for most of the variability (Smith et al, 1993). A study of biological variability in 135 people with diabetes (both Type 1 and Type 2) (Tsalamandris et al, 1998) found that total variability expressed as a coefficient of variation was 8.8±0.4% for total cholesterol, 23.9±1.5% for triglycerides, 10.2±0.5% for HDL cholesterol and 12.0±0.5% for calculated LDL cholesterol over an average followup period of 5.1 years. Greater variability was observed in men compared with women for total cholesterol (9.5±0.5% v 7.9±0.5%, p<0.01) and for triglycerides (26.5±2.2% v 20.4±1.4%, p=0.03). Most of the variability in these parameters was related to biological factors, not measurement error (Tsalamandris et al, 1998). In the United Kingdom Prospective Diabetes Study (UKPDS), the baseline triglycerides concentration in people allocated to diet correlated with the values six months later (r s =0.72; CI 0.68 to 0.76) while the correlation between repeated measures of HDL cholesterol was lower (r s =0.52; CI 0.45 to 0.58) (Turner et al, 1998). An important source of biological variability for some lipid fractions is whether or not lipids are measured in the fasting state, particularly in relation to increasing the variability in the triglycerides levels. Diurnal and monthly variability of serum lipids were examined in 11 healthy subjects aged 3263 years (Wasenius et al, 1990). Blood sample were drawn at 8am after 8 hour fasting, 12pm, 3pm, 6pm and 9pm for diurnal measurements, and were drawn weekly at 8am after 8 hour fasting for monthly measurements. The mean diurnal variability was 2.4% for total cholesterol, 3.5% for HDL cholesterol, 5.1% for LDL cholesterol, and 29.5% for triglycerides. The corresponding monthly variabilities were 4.2%, 4.1%, 5.2%, 20.7%, respectively. Implications for Clinical Practice Epidemiological and intervention studies in Type 2 diabetes have measured lipid levels after a 10 to 12 hour overnight fast and therefore evidence relating to lipid abnormalities and their treatment is based on this sampling method. Despite this, recommendations between 19

20 organisations have varied somewhat on this point. The National Cholesterol Education Program (NCEP) (The National Cholesterol Education Program Expert Panel, 2002) and the Heart Foundation of Australia (National Heart Foundation of Australia & The Cardiac Society of Australia and New Zealand, 2001) recommend sampling after a 12 hour fast when screening for lipid abnormalities. The UK Clinical Guidelines on Lipid Management for Type 2 diabetes (Collaborative Programme between: Royal College of General Practitioners, Diabetes UK, Royal College of Physicians, Royal College of Nursing) recommend fasting measurement if feasible (McIntosh et al, 2002). However it should be noted that there is little change in total cholesterol, HDL cholesterol and LDL cholesterol postprandially, so these lipid parameters can be measured satisfactorily in the nonfasting state. Taking the available information into account, it is recommended that lipid levels be assessed after an overnight fast of 1012 hours. However it is recognised that only triglycerides (and to a small extent calculated LDL cholesterol) are affected by nonfasting. Depending on individual and practice circumstances it may not be feasible to measure lipids in the fasting state, however if lipid levels are not measured fasting, caution is required in interpreting an elevated triglycerides level and calculated LDL cholesterol. Because there is intraindividual variation over time in lipids in Type 2 diabetes, management decisions should generally be based on more than one measurement. There are no clinical trials on optimal frequency for reassessment of lipid levels. The Clinical Practice Recommendations of the American Diabetes Association (ADA, 2003) state that levels of LDL, HDL, total cholesterol, and triglycerides should be measured every year in adult patients. If values fall in the lowerrisk levels, assessment may be repeated every 2 years. The UK Clinical Guideline for Type 2 Diabetes on Lipid Management recommends annual assessment of lipid levels as part of the annual review (McIntosh et al, 2002). As with all recommendations, the frequency of reassessment should be modified by clinical judgment. However in general, it is considered that lipid levels should be measured at the time of diagnosis in people with Type 2 diabetes and annually as part of the annual review. In people with lipid levels above target, levels should be reevaluated 3 months after recommended lifestyle (diet/exercise) changes (See Section 2) and after improving glycaemic control (See Section 3), and after commencing lipidlowering agents (See Sections 4 & 5). In people on long term lipidlowering therapy, lipid levels would usually be measured at 6 monthly intervals. Apolipoprotein A1 is reduced, apolipoprotein B is increased, and lipoprotein(a) is not altered in Type 2 diabetes Apolipoprotein A1 (Apo A1) is the major apolipoprotein in HDL particles and consistent with the lower levels of HDL cholesterol, several studies have shown lower Apo A1 levels in Type 2 diabetes. In the Framingham Offspring Study (Siegel et al, 1996) the mean Apo A1 level was 134 mg/dl in 54 people with diabetes compared with 158 mg/dl in 1,879 people without diabetes (p<0.001). Apo A1 levels were also consistently lower in people with diabetes compared with people without diabetes for both sexes (100 v 105 mg/dl, p<0.01 in men; 107 v 116 mg/dl, p<0.01 in women) in a population based study in Finland (Rönnemaa et al, 1989). In a clinic based study (Billingham et al, 1989), the mean Apo A1 level was 94 mg/dl in 13 men at diagnosis of diabetes, compared with 114 mg/dl in 17 men without diabetes (p<0.05). In 9 women the mean Apo A1 level at diagnosis of diabetes was 107 mg/dl, compared with 125 mg/dl in 16 women without diabetes (p<0.05). However, Dean et al (1990) found that people with Type 2 diabetes (n=35) had a similar Apo A1 level compared with 35 control subjects (121.4±24.1 v 131.4±36.5 mg/dl, p=ns). Another casecontrol study showed that there was no significant difference in Apo A1 level between

21 people with diabetes and 30 control subjects (149±27 v 151±28.2 mg/dl) (Manzato et al, 1993). One apolipoprotein B (Apo B) molecule is present in each LDL and VLDL particle, so that Apo B level is a measure of the combined total number of these particles. In the Framingham Offspring Study (Siegel et al, 1996) mean Apo B level was 106 mg/dl in those with diabetes, compared with 83 mg/dl in controls (p<0.001). Apo B levels were also consistently higher in people with diabetes than in the controls (123 v 117 mg/dl in men, p<0.05; 125 v 119 mg/dl in women, p<0.05) in a population based study in Finland (Rönnemaa et al, 1989). In a clinic based study (Billingham et al, 1989) mean Apo B level was 95 mg/dl in 22 men and women at diagnosis of diabetes, compared with 75 mg/dl in 33 controls (p<0.05). Another clinic based study (Manzato et al, 1993) found mean Apo B level of 150 mg/dl in 120 people with diabetes compared with 135 mg/dl in 30 control subjects (p=0.02). In contrast, the clinic based study by Dean et al (1990) found no difference in Apo B between 35 people with and 35 people without diabetes (90.9±24.0 v 96.2±22.5 g/dl, p=ns). The higher Apo B levels in Type 2 diabetes indicate higher numbers of both VLDL and LDL particles, consistent with higher triglycerides levels and normal LDL cholesterol levels, but smaller, denser LDL particles. Apolipoprotein (a) is an apolipoprotein with structural homology to plasminogen. Bound to the Apo B moiety of a lipoprotein that closely resembles a LDL particle, it forms lipoprotein (a) [Lp(a)] that is potentially atherogenic and thrombogenic. Most studies of Lp(a) in Type 2 diabetes indicate that levels are not different to those found in the nondiabetic population (Haffner et al, 1992a; Csaszar et al, 1993; Velho et al, 1993; Chang et al, 1995). A population based study (Haffner et al, 1992a) showed a similar Lp(a) concentration for both men and women in people with and without diabetes (13.6±1.5 v 16.1±1.4 mg/dl; 12.6±0.8 v 15.9±1.3 mg/dl, respectively) (p=0.361). In a case control study (Csaszar et al, 1993), the mean Lp(a) concentration among people with Type 2 diabetes did not differ from those of control subjects in both Hungarian (12.2±17.9 v 10.5±13.5 mg/dl) and Austrian groups (14.6±21.0 v 12.2±16.5 mg/dl). Another case control study in a Chinese population also showed no difference in the Lp(a) levels between people with and without diabetes (22.3±2.2 v 20.7±1.9 mg/dl) (Chang et al, 1995). Velho et al (1993) compared Lp(a) levels in people with Type 2 diabetes, their normoglycaemic relatives, and healthy controls and found that Lp(a) levels were not significantly different (27.2±19.3 v 27.1±18.2 v 23.1±15.1 mg/dl, p=0.38). Lipid abnormalities are aggravated by increased body weight, poor glycaemic control, and diabetic renal disease In people with Type 2 diabetes, greater disturbance of lipid metabolism has been reported in association with increasing body weight, poor glycaemic control, diabetic renal disease and female gender. Increasing body mass index (BMI) and waist:hip ratio (WHR) within the Type 2 diabetes population have been linked with hypertrigyceridaemia and reduced HDL cholesterol in a number of studies (Table 1) in both men and women (Laakso et al, 1985; Uusitupa et al, 1986; Laakso & Pyorala, 1990; Byrne et al, 1994). Laakso et al (1985) reported that BMI had a statistically significant negative correlation with HDL levels (p<0.001 for both men & women) and a positive correlation with triglycerides (p<0.001 in men; p<0.05 in women) in people with Type 2 diabetes. Uusiputa et al (1986) also found a negative correlation of BMI with HDL cholesterol levels (p=0.014 in men; p=0.066 in women) and a positive correlation of BMI with triglycerides (p=0.023; p=0.046, respectively) in a population of 133 people with Type 2 diabetes. In a case control study, obesity was associated with a lower HDL cholesterol level (BMI >27.0 v <25.0 kg/m 2 : 1.06 v 1.33 mmol/l, p<0.001 in men; 1.09 v 21

22 1.48 mmol/l in women) and higher triglycerides levels (3.36 v 1.72 mmol/l, p<0.001; 3.63 v 1.74 mmol/l, p<0.001, respectively) in 305 people with Type 2 diabetes (Laakso & Pyorala, 1990). Byrne et al (1994) found that WHR had a positive association with triglycerides concentrations in people with diabetes (p<0.001 for men; p<0.01 for women). A correlation between measures of glycaemic control and HDL cholesterol has been reported (Laakso et al, 1985). After adjusting for age, physical activity and BMI, fasting plasma glucose (FPG) was significantly associated with HDL cholesterol in both men (p=0.003) and women (p<0.001) with Type 2 diabetes (Laakso et al, 1985). The influence of glycaemic control has important implications for determining decisions relating to the initiation of lipidmodifying therapy. Treatment of hyperglycaemia should generally precede initiation of treatment for lipid abnormalities especially when triglycerides are elevated or HDL cholesterol is reduced (see Lipids Section 3). The development of diabetic nephropathy is associated with a further reduction of HDL cholesterol and increase of triglycerides in Type 2 diabetes (Mattock et al, 1988; Niskanen et al, 1990; Suraniti et al, 1992; Nielsen et al, 1993). A crosssectional study of 141 people with diabetes found there were significant correlations between albumin excretion rate and total triglycerides (r=0.214, p<0.05) and HDL cholesterol (r=0.243, p<0.01) (Mattock et al, 1988). Among 123 people with diabetes, there were no differences in triglycerides and HDL cholesterol in people with (n=21) and without microalbuminuria (n=92) at baseline. However, at the 5year examination, people with microalbuminuria had a significantly higher triglycerides levels (4.24±0.90 v 2.38±0.16 mmol/l, p<0.05) and a lower HDL cholesterol level (0.92±0.05 v 1.07±0.03 mmol/l, p<0.05) compared with those without microalbuminuria (Niskanen et al, 1990). A negative correlation of urinary albumin excretion to HDL cholesterol (p=0.0001) was found by Suraniti et al (1992) in 380 people with Type 2 diabetes. In a case control study, people with Type 2 diabetes and macroalbuminuria (>300 mg/24h) had a higher median triglycerides levels than in people with microalbuminuria ( mg/24h) and people with normoalbuminuria ( 30 mg/24h) (2.01 [ ] v 1.82 [ ] v 1.30 [ ] mmol/l, p<0.01) (Nielsen et al, 1993). There is insufficient evidence to draw any conclusion about the importance of lipoprotein oxidation in Type 2 diabetes Results of studies of lipoprotein oxidation in Type 2 diabetes have been variable. Some studies have reported increased lipid peroxidation products in plasma (Nacitarhan et al, 1995; Ceriello et al, 1997a; Freitas et al, 1997) but others have not (MacRury et al, 1993; Haffner et al, 1995). Nacitarhan et al (1995) measured malondialdehyde (MDA), a marker of lipid peroxidation, among people with Type 2 diabetes (n=78, including n=34 with hyperlipidaemia), people with hyperlipidamia (n=38), and healthy controls (n=28) and found that serum MDA level was significantly higher in people with diabetes (7.1±2.5; 6.2±1.7 nmol/l, respectively) and hyperlipidaemia (5.7±1.01 nmol/l) than in healthy controls (4.8±0.8 nmol/l) (all p<0.001). In the diabetic group, people with hyperlipidaemia had a higher serum MDA level than normolipidaemic people (p<0.02). Urine MDA levels in the diabetic group were higher than that of the nondiabetic hyperlipidaemic group (5.7±2.6; 5.7±2.3 nmol/l, respectively v 4.5±1.1 nmol/l, p<0.01). Ceriello et al (1997a) reported that total radicaltrapping antioxidant parameter (TRAP), a marker of plasma antioxidant capacity, was significantly reduced in people with diabetes (n=46) compared with control subjects (n=47) (677±58 v 950.1±16.0 μmol/l, p<0.001). However, a study conducted by MacRury et al (1993) showed that the level of plasma antioxidant, caeruloplasmin, was 22

23 higher in people with diabetes than control subjects (19.3±6.9 v 14.2±5.1 μmol/l, p<0.01). A number of studies have reported increased lipid susceptibility to oxidation using plasma (Haffner et al, 1995) or LDL (Dimitriadis et al, 1995; Leonhardt et al, 1996) from subjects with Type 2 diabetes. Ceriello et al (1997b) reported the existence of lower antioxidant defenses in people with Type 2 diabetes. TRAP measured by fluorescencebased method and calculated by a mathematical formula were both lower in diabetic people (n=40) than controls (n=40) (690.4±16.5 v 961.9±16.7 μmol/l, p<0.001; 615.2±15.2 v 669.3±11.8 μmol/l, p<0.005, respectively). However, a study of 72 people with Type 2 diabetes and 94 nondiabetic controls did not find any difference in LDL susceptibility to oxidation, (Leinonen et al, 1998a). Another study by the same group of antioxidant defenses in 81 people with Type 2 diabetes and 102 controls showed no difference (Leinonen et al, 1998b). Nearly all studies have used nonspecific measures of lipoprotein oxidation or have examined in vitro susceptibility to oxidation or antioxidant capacity. None of these measures gives a direct indication of the importance of lipoprotein oxidation in Type 2 diabetes. The results of these indirect tests have been variable and it is therefore not possible to draw any evidencebased conclusion about the importance of lipoprotein oxidation in Type 2 diabetes. Total cholesterol and LDL cholesterol predict cardiovascular disease in Type 2 diabetes Data from crosssectional studies provide information on the association between lipid levels and cardiovascular disease while prospective cohort studies provide information on the relationship of lipid levels and the future development of cardiovascular disease. Crosssectional studies Crosssectional studies provide limited evidence for total and LDL cholesterol as risk factors for cardiovascular disease (CVD) in Type 2 diabetes (see Table 2). Nearly all studies have used a composite of macrovascular disease indicators and few studies have included more than 100 subjects. Some studies have found higher total and LDL cholesterol in association with CVD in Type 2 diabetes. The Milan Study on Atherosclerosis and Diabetes excluded people on insulin therapy, as well as those with clinical CVD or diabetes complications (MiSAD Group, 1997). In order to detect silent myocardial ischaemia, 925 people aged 40 to 65 years underwent exercise ECG, followed by exercise thallium scintigram if the ECG was abnormal. Total cholesterol was 5.83±1.12 mmol/l in the 59 subjects with both tests positive compared with 5.48±1.04 mmol/l in the other 866 subjects (p=0.014). Another hospital based study in London assessed people with Type 2 diabetes for CHD using resting ECG, history and standard questionnaire and for peripheral vascular disease using questionnaire and ankle/arm systolic blood pressure ratio (Seviour et al, 1988). Men with macrovascular disease (n=48) had higher total cholesterol and LDL cholesterol levels compared with men without evidence of macrovascular disease (n=47) (5.98±1.20 v 5.15±0.98 mmol/l, p=0.0001; 4.29±0.95 v 3.57±0.74 mmol/l, p=0.0001, respectively). The corresponding figures among women were 6.46±1.51 v 5.71±1.01 mmol/l (p=0.04); and 4.68±1.26 v 3.82±0.96 mmol/l (p=0.008), respectively. Two North American studies have also shown higher LDL cholesterol in association with prevalent CHD in Type 2 diabetes. Of 227 subjects with Type 2 diabetes in the Rochester Diabetic Neuropathy Study (O'Brien et al, 1994), 96 were defined as having CHD on the basis of history and ECG, abnormal noninvasive cardiovascular tests or abnormal coronary angiography. Total cholesterol was 5.26±1.03 mmol/l in this group, compared with 23