Diabetes Research and Clinical Practice, 18 (1992) 0 1992 Elsevier Science Publishers
173-184
173
B.V. All rights reserved 016%8227/92/$05.00
DIABET 00692
A cross-sectional evaluation of cardiovascular risk factors in coronary heart disease associated with Type 1 (insulin-dependent) diabetes mellitus Peter H. Winocour, Paul N. Durrington, Deepak Bhatnagar, Anthony Monica Ishola, Michael Mackness and Sharon Arrol
D. Mbewu,
The Universitv of Manchester Departments of Medicine. Hope Hospital, Salford and the Manchester Royal In$rmary, Manchester,
UK
(Received 2 September 1991) (Revision accepted 3 August 1992)
Summary
The contribution
from lipoproteins, blood pressure, albuminuria and demographic variables to coronary heart disease in 90 adult subjects with and 172 without Type 1 diabetes mellitus was examined in order to investigate whether risk factors were of equivalent importance in diabetic and non-diabetic coronary heart disease. Coronary heart disease (CHD) was present in roughly 25 % of subjects in each group. In Type 1 diabetes those with CHD had significantly higher levels of systolic blood pressure, albumin excretion, serum creatinine, triglycerides, VLDL cholesterol and C-peptide, and reductions in serum concentrations of HDL and HDL, cholesterol, in comparison to those without. However, the prevalence of smokers, and concentrations of Lp(a), ApoB and fibrinogen were comparable. Blood pressure and HDL cholesterol were higher in the CHD group with Type 1 diabetes in comparison to the nondiabetic group with CHD, although LDL concentrations and the prevalence of Lp(a) concentrations > 200 mg/l were lower. Logistic regression analysis revealed the strongest independent predictors of CHD in Type 1 diabetes were serum triglycerides, systolic blood pressure, age, serum LDL cholesterol, and the daily insulin dosage, whereas in the non-diabetic control group HDL, cholesterol, Lp(a), ApoAl and ApoB, total serum cholesterol and body mass index were additional predictors. CHD in Type 1 diabetes appears to be most closely associated with increasing age and levels of blood pressure and total serum lipids. Apolipoproteins and albuminuria did not seem to be important independent predictors of CHD in Type 1 diabetes, whereas the former were more clearly associated with CHD in non-diabetic controls. Key words: Type 1 diabetes; Coronary Albuminuria; Fibrinogen
Correspondence
to: Dr. P.H. Winocour,
heart
disease;
Blood
pressure;
Freeman Hospital, Freeman Road, Newcastle-upon-Tyne,
Lipids;
Lipoprotein(a);
NE7 7DN, UK.
174
Introduction Premature coronary heart disease (CHD) is one of the commonest causes of mortality in Type 1 (insulin-dependent) diabetes mellitus (IDDM) [ I-41 and its incidence is at least twice that of a non-diabetic population of similar age and ethnic origin [ 3-61. The precise reason for this remains unclear. Conventional cardiovascular risk markers (blood pressure, hypercholesterolaemia, and smoking) appear to confer risk in diabetic subjects [4,5,7-131, but there is some evidence that their relative contribution may differ from that observed in the general population [ 8-131. Proteinuria, serum triglyceride and HDL cholesterol concentrations are additional independent cardiovascular risk factors in diabetes, and may be better predictors of CHD than serum cholesterol [2,9,14-181. Recent evidence has suggested that fibrinogen [ 19-221 and lipoprotein Lp(a) [ 23-261 are strong predictors of CHD in the non-diabetic population. There are suggestions that their concentrations may be increased in IDDM complicated by nephropathy [27-301, although fibrinogen may not be an independent risk factor in diabetic CHD [ 3 11. It is unclear whether or not Lp(a) only exerts an influence when concentrations of LDL cholesterol are increased [ 25,32,33]. The majority of epidemiological studies of CHD risk factors in diabetes have focussed on non-independent diabetes mellitus (NIDDM) [4,11,12,14,15]. In the present study we have therefore examined the relative contributions of these various risk factors to CHD in subjects with IDDM in comparison to a normoglycaemic control group.
Patients and Protocol The study was approved by the local area Ethical Committee, and written consent obtained from all participants. Initially, 200 subjects with insulin-treated diabetes were recruited from the diabetic clinic at random for a study of home blood glucose mon-
itoring. Of these, 90 individuals had stable Type 1 diabetes (non-ketotic, HbA, < 13%), and preserved renal function (serum creatinine I 180 pmol/l in all cases, < 120 pmol/l in 84), and fulfilled the following criteria. All were Caucasian, aged between 17 and 70 yr and had a duration of diabetes of at least 5 yr. The diagnosis of Type 1 diabetes was based on clinical grounds recognising that residual insulin secretion may be a feature of adults with Type 1 diabetes [34]. They had received l-3 daily injections of insulin continually within 1 yr of diagnosis of diabetes, and had circulating C-peptide concentrations less than 0.60 nmol/l (< 0.10 nmol/l in 73 cases). Ketoacidosis was previously documented in 75 “6 of cases, and diabetes was diagnosed at an age of 35 or less in SSo/ of cases. None were taking drugs known to affect lipid metabolism at the time of the study, other than insulin, and all had normal serum albumin levels. Calcium antagonists or nitrates were taken by 8 subjects, loop diuretics in 2, and aspirin and ACE inhibitors in 2 and 3 cases. respectively. Beta-blockers were withdrawn from those patients taking them prior to the study and substituted with calcium antagonists, ACE inhibitors or nitrates. All patients had received standard dietary advice (fixed amounts of carbohydrate and calories derived from 50% carbohydrate, 30% fat and 20% protein, isocaloric with their energy requirements). Proliferative (new vessel formation) retinopathy (subsequently confirmed on fluorescein angiography) was sought by fundoscopy after pupillary dilation. Coronary heart disease was classified on the basis of documented myocardial infarction at least 6 months previously (typical clinical history with contemporary positive electrocardiographic and cardiac enzyme evidence), typical angina pectoris (exertional chest pain relieved by glyceryl trinitrate) or pronounced resting ECG Q wave abnormalities and/or T wave inversion and STsegment depression, (Minnesota codes 1- 1, 1-2, 8- 1,8-3). All those classified as free of CHD had normal ECGs and no history compatible with angina or myocardial infarction. Peripheral vascular disease was diagnosed by a history of in-
175
termittent claudication associated with one or more absent foot pulses. Non-diabetic Caucasian controls with and without coronary heart disease from a similar social background to the diabetic patients (comparable distribution of social class) were recruited from two sources. Those without coronary heart disease were either healthy men working in local industry in an adjacent area of Greater Manchester who underwent routine health checks at work, or men and women attending a general practice health screening programme in South Manchester. The bulk of non-diabetic cases with CHD presented consecutively for coronary angiography. None were diabetic or had had a myocardial infarction or unstable angina within 6 months of assessment. Betablockers were being taken by 22 patients, calcium antagonists by 21, longacting nitrates by 15, aspirin/persantin anticoagulants by 16, and diuretics or ACE inhibitors by 4 and 2 cases, respectively. All diabetic subjects were interviewed and assessed by one of us (P.H.W.), whilst non-diabetic controls were reviewed independently (D.B., P.N.D., A.D.M.). Symptoms of macrovascular disease and tobacco and alcohol consumption were assessed by interview with the aid of the WHO Rose questionnaire [35]. Blood pressure was measured with a sphygmomanometer after at least 10 min rest whilst seated.
Methods Fasting blood glucose concentration was measured by a glucose oxidase method (Yellow Springs Analyser, Clandon Scientific, UK) and HbA, by ion exchange chromatography (Boehringer, Mannheim, Germany) (normal range 5.08.0%, between assay variability 3.0-6.02). C-peptide levels were measured following a standard mixed meal as previously described [ 361. All fasting lipid and lipoprotein estimations were carried out in duplicate. The methodology been described in detail has previously [ 9,26,37,38]. Lipoproteins were isolated by ultra-
centrifugation. Our laboratory currently participates in the UK national quality control scheme for cholesterol estimation. The within batch coefficients of variation for cholesterol and triglyceride assays were 1.5 y0 and 2.1%) respectively, and for lipoprotein isolation 4.0-5.0%. In view of the possibility that some cholesterol present in Lp(a) might be included in the HDL and HDL, cholesterol determined by ultracentrifugation, HDL was also isolated by precipitation of other lipoproteins, including Lp(a), with sodium phosphotungstate and magnesium chloride [ 39,401. Total serum apolipoprotein B and Al (the latter in controls only) concentrations were determined by immunoelectrophoresis as described previously [ 26,411. The within batch coefficient of variation of the assay was 5.4%. Serum Lp(a) concentrations were measured in samples stored at -70 “C, using a two-site radioimmunometric assay (Pharmacia Diagnostics AB, Uppsala, Sweden) as previously described [26,29]. The within batch coefficient of variation was 6%. In all lipid and lipoprotein analyses, high and low quality control sera were included. None of the values reported was obtained from assays in which the quality control sera gave results which did not lie within 10% of the original mean value. Serum creatinine was measured on a multi channel auto-analyser, and urinary albumin excretion by an ELISA method from 24-h collections on 3 separate occasions. The average values are reported. The lower limit of detection of the assay was 0.5 pg/l and the upper limit of normal for the albumin excretion rate (AER), based on 50 healthy non-diabetic controls, was 15 pg/min. Fibrinogen was measured in the plasma of diabetic subjects by a nephelometric method as previously described [42]. A commercial standard and pooled normal serum were assayed in each batch for quality control. The between assay variability was consistently less than 10%. The distributions of albumin excretion rates, alcohol intake, triglycerides and Lp(a) were positively skewed, and were logarithmically transformed for all analyses except the Spearman rank
176
correlation coefficient. Differences between the groups were compared by the Mann-Whitney Uor Student’s unpaired t-tests, depending on the distribution of the data. Forward stepwise logistic regression analysis was employed to examine independent factors associated with CHD in those with and without IDDM, following univariate logistic regression analysis. A standardised estimate of relative risk (SERR) was calculated from logistic regression coefficients, as previously described [ 121. The SERR gives relative risk for a change in 1 unit of standard deviation for a particular valuable. All variables were entered subsequently into the stepwise logistic regression analyses. In order to examine whether the impact of Lp(a) on CHD in IDDM was dependent on LDL, attempts were made to case-match each of those with IDDM and CHD for (in order of se-
TABLE Clinical
Number
in Type
1 diabetic
and control
Coronary heart disease was documented in 23 of the 90 subjects with IDDM and in 40 of the 172 without diabetes. Alcohol consumption and sex ratios were comparable between the groups (Table 1). The group with IDDM and CHD were compared to those with IDDM without CHD. Those with CHD had significantly higher levels of systolic blood pressure (p
173-184
173
B.V. All rights reserved 016%8227/92/$05.00
DIABET 00692
A cross-sectional evaluation of cardiovascular risk factors in coronary heart disease associated with Type 1 (insulin-dependent) diabetes mellitus Peter H. Winocour, Paul N. Durrington, Deepak Bhatnagar, Anthony Monica Ishola, Michael Mackness and Sharon Arrol
D. Mbewu,
The Universitv of Manchester Departments of Medicine. Hope Hospital, Salford and the Manchester Royal In$rmary, Manchester,
UK
(Received 2 September 1991) (Revision accepted 3 August 1992)
Summary
The contribution
from lipoproteins, blood pressure, albuminuria and demographic variables to coronary heart disease in 90 adult subjects with and 172 without Type 1 diabetes mellitus was examined in order to investigate whether risk factors were of equivalent importance in diabetic and non-diabetic coronary heart disease. Coronary heart disease (CHD) was present in roughly 25 % of subjects in each group. In Type 1 diabetes those with CHD had significantly higher levels of systolic blood pressure, albumin excretion, serum creatinine, triglycerides, VLDL cholesterol and C-peptide, and reductions in serum concentrations of HDL and HDL, cholesterol, in comparison to those without. However, the prevalence of smokers, and concentrations of Lp(a), ApoB and fibrinogen were comparable. Blood pressure and HDL cholesterol were higher in the CHD group with Type 1 diabetes in comparison to the nondiabetic group with CHD, although LDL concentrations and the prevalence of Lp(a) concentrations > 200 mg/l were lower. Logistic regression analysis revealed the strongest independent predictors of CHD in Type 1 diabetes were serum triglycerides, systolic blood pressure, age, serum LDL cholesterol, and the daily insulin dosage, whereas in the non-diabetic control group HDL, cholesterol, Lp(a), ApoAl and ApoB, total serum cholesterol and body mass index were additional predictors. CHD in Type 1 diabetes appears to be most closely associated with increasing age and levels of blood pressure and total serum lipids. Apolipoproteins and albuminuria did not seem to be important independent predictors of CHD in Type 1 diabetes, whereas the former were more clearly associated with CHD in non-diabetic controls. Key words: Type 1 diabetes; Coronary Albuminuria; Fibrinogen
Correspondence
to: Dr. P.H. Winocour,
heart
disease;
Blood
pressure;
Freeman Hospital, Freeman Road, Newcastle-upon-Tyne,
Lipids;
Lipoprotein(a);
NE7 7DN, UK.
174
Introduction Premature coronary heart disease (CHD) is one of the commonest causes of mortality in Type 1 (insulin-dependent) diabetes mellitus (IDDM) [ I-41 and its incidence is at least twice that of a non-diabetic population of similar age and ethnic origin [ 3-61. The precise reason for this remains unclear. Conventional cardiovascular risk markers (blood pressure, hypercholesterolaemia, and smoking) appear to confer risk in diabetic subjects [4,5,7-131, but there is some evidence that their relative contribution may differ from that observed in the general population [ 8-131. Proteinuria, serum triglyceride and HDL cholesterol concentrations are additional independent cardiovascular risk factors in diabetes, and may be better predictors of CHD than serum cholesterol [2,9,14-181. Recent evidence has suggested that fibrinogen [ 19-221 and lipoprotein Lp(a) [ 23-261 are strong predictors of CHD in the non-diabetic population. There are suggestions that their concentrations may be increased in IDDM complicated by nephropathy [27-301, although fibrinogen may not be an independent risk factor in diabetic CHD [ 3 11. It is unclear whether or not Lp(a) only exerts an influence when concentrations of LDL cholesterol are increased [ 25,32,33]. The majority of epidemiological studies of CHD risk factors in diabetes have focussed on non-independent diabetes mellitus (NIDDM) [4,11,12,14,15]. In the present study we have therefore examined the relative contributions of these various risk factors to CHD in subjects with IDDM in comparison to a normoglycaemic control group.
Patients and Protocol The study was approved by the local area Ethical Committee, and written consent obtained from all participants. Initially, 200 subjects with insulin-treated diabetes were recruited from the diabetic clinic at random for a study of home blood glucose mon-
itoring. Of these, 90 individuals had stable Type 1 diabetes (non-ketotic, HbA, < 13%), and preserved renal function (serum creatinine I 180 pmol/l in all cases, < 120 pmol/l in 84), and fulfilled the following criteria. All were Caucasian, aged between 17 and 70 yr and had a duration of diabetes of at least 5 yr. The diagnosis of Type 1 diabetes was based on clinical grounds recognising that residual insulin secretion may be a feature of adults with Type 1 diabetes [34]. They had received l-3 daily injections of insulin continually within 1 yr of diagnosis of diabetes, and had circulating C-peptide concentrations less than 0.60 nmol/l (< 0.10 nmol/l in 73 cases). Ketoacidosis was previously documented in 75 “6 of cases, and diabetes was diagnosed at an age of 35 or less in SSo/ of cases. None were taking drugs known to affect lipid metabolism at the time of the study, other than insulin, and all had normal serum albumin levels. Calcium antagonists or nitrates were taken by 8 subjects, loop diuretics in 2, and aspirin and ACE inhibitors in 2 and 3 cases. respectively. Beta-blockers were withdrawn from those patients taking them prior to the study and substituted with calcium antagonists, ACE inhibitors or nitrates. All patients had received standard dietary advice (fixed amounts of carbohydrate and calories derived from 50% carbohydrate, 30% fat and 20% protein, isocaloric with their energy requirements). Proliferative (new vessel formation) retinopathy (subsequently confirmed on fluorescein angiography) was sought by fundoscopy after pupillary dilation. Coronary heart disease was classified on the basis of documented myocardial infarction at least 6 months previously (typical clinical history with contemporary positive electrocardiographic and cardiac enzyme evidence), typical angina pectoris (exertional chest pain relieved by glyceryl trinitrate) or pronounced resting ECG Q wave abnormalities and/or T wave inversion and STsegment depression, (Minnesota codes 1- 1, 1-2, 8- 1,8-3). All those classified as free of CHD had normal ECGs and no history compatible with angina or myocardial infarction. Peripheral vascular disease was diagnosed by a history of in-
175
termittent claudication associated with one or more absent foot pulses. Non-diabetic Caucasian controls with and without coronary heart disease from a similar social background to the diabetic patients (comparable distribution of social class) were recruited from two sources. Those without coronary heart disease were either healthy men working in local industry in an adjacent area of Greater Manchester who underwent routine health checks at work, or men and women attending a general practice health screening programme in South Manchester. The bulk of non-diabetic cases with CHD presented consecutively for coronary angiography. None were diabetic or had had a myocardial infarction or unstable angina within 6 months of assessment. Betablockers were being taken by 22 patients, calcium antagonists by 21, longacting nitrates by 15, aspirin/persantin anticoagulants by 16, and diuretics or ACE inhibitors by 4 and 2 cases, respectively. All diabetic subjects were interviewed and assessed by one of us (P.H.W.), whilst non-diabetic controls were reviewed independently (D.B., P.N.D., A.D.M.). Symptoms of macrovascular disease and tobacco and alcohol consumption were assessed by interview with the aid of the WHO Rose questionnaire [35]. Blood pressure was measured with a sphygmomanometer after at least 10 min rest whilst seated.
Methods Fasting blood glucose concentration was measured by a glucose oxidase method (Yellow Springs Analyser, Clandon Scientific, UK) and HbA, by ion exchange chromatography (Boehringer, Mannheim, Germany) (normal range 5.08.0%, between assay variability 3.0-6.02). C-peptide levels were measured following a standard mixed meal as previously described [ 361. All fasting lipid and lipoprotein estimations were carried out in duplicate. The methodology been described in detail has previously [ 9,26,37,38]. Lipoproteins were isolated by ultra-
centrifugation. Our laboratory currently participates in the UK national quality control scheme for cholesterol estimation. The within batch coefficients of variation for cholesterol and triglyceride assays were 1.5 y0 and 2.1%) respectively, and for lipoprotein isolation 4.0-5.0%. In view of the possibility that some cholesterol present in Lp(a) might be included in the HDL and HDL, cholesterol determined by ultracentrifugation, HDL was also isolated by precipitation of other lipoproteins, including Lp(a), with sodium phosphotungstate and magnesium chloride [ 39,401. Total serum apolipoprotein B and Al (the latter in controls only) concentrations were determined by immunoelectrophoresis as described previously [ 26,411. The within batch coefficient of variation of the assay was 5.4%. Serum Lp(a) concentrations were measured in samples stored at -70 “C, using a two-site radioimmunometric assay (Pharmacia Diagnostics AB, Uppsala, Sweden) as previously described [26,29]. The within batch coefficient of variation was 6%. In all lipid and lipoprotein analyses, high and low quality control sera were included. None of the values reported was obtained from assays in which the quality control sera gave results which did not lie within 10% of the original mean value. Serum creatinine was measured on a multi channel auto-analyser, and urinary albumin excretion by an ELISA method from 24-h collections on 3 separate occasions. The average values are reported. The lower limit of detection of the assay was 0.5 pg/l and the upper limit of normal for the albumin excretion rate (AER), based on 50 healthy non-diabetic controls, was 15 pg/min. Fibrinogen was measured in the plasma of diabetic subjects by a nephelometric method as previously described [42]. A commercial standard and pooled normal serum were assayed in each batch for quality control. The between assay variability was consistently less than 10%. The distributions of albumin excretion rates, alcohol intake, triglycerides and Lp(a) were positively skewed, and were logarithmically transformed for all analyses except the Spearman rank
176
correlation coefficient. Differences between the groups were compared by the Mann-Whitney Uor Student’s unpaired t-tests, depending on the distribution of the data. Forward stepwise logistic regression analysis was employed to examine independent factors associated with CHD in those with and without IDDM, following univariate logistic regression analysis. A standardised estimate of relative risk (SERR) was calculated from logistic regression coefficients, as previously described [ 121. The SERR gives relative risk for a change in 1 unit of standard deviation for a particular valuable. All variables were entered subsequently into the stepwise logistic regression analyses. In order to examine whether the impact of Lp(a) on CHD in IDDM was dependent on LDL, attempts were made to case-match each of those with IDDM and CHD for (in order of se-
TABLE Clinical
Number
in Type
1 diabetic
and control
Coronary heart disease was documented in 23 of the 90 subjects with IDDM and in 40 of the 172 without diabetes. Alcohol consumption and sex ratios were comparable between the groups (Table 1). The group with IDDM and CHD were compared to those with IDDM without CHD. Those with CHD had significantly higher levels of systolic blood pressure (p
