Several large autopsy and epidemiological studies

Atherosclerotic Vascular Disease in Middle-Aged, Insulin-Treated, Diabetic Patients Association with Endogenous Insulin Secretion Capacity Tapani ROnnemaa, Markku Laakso, Pauli Puukka, Veikko Kallio, and Kalevi PySrala The prevalence of atherosclerotic vascular disease (ASVD) and Its risk factors were Investigated In 263 Insulin-treated diabetic patients, ages 45 to 64 years, who were older than 3 years when their diabetes was diagnosed. The patients were divided Into two groups based on the degree of endogenous Insulin secretion capacity: Group A: glucagon-stlmulated plasma C-peptlde <.2 nmol/1 and Group B: C-peptlde >.2 nmol/l. The age-adjusted prevalence of definite myocardlal Infarction was significantly higher In Group B than In Group A (16.8% vs. 5.2%, p<.1). A similar difference between Groups A and B was found for definite or possible coronary heart disease (54.6% vs. 32.9%, p<.1) and stroke (9.3% vs. 2.%, p<.5). In multlvarlate analysis, high glucagon-stlmulated plasma C-peptlde level (a.2 nmol/1) was positively associated with definite or possible coronary heart disease Independently of other cardiovascular risk factors. Our results Indicate that among Insulin-treated patients with a late onset of diabetes, the prevalence of ASVD Is markedly higher In those with persistent endogenous Insulin secretion (nonlnsultn-dependent diabetes) than In those with low or no Insulin secretion (Insulin-dependent diabetes). (Arteriosclerosis 8:237-244, May/June 1988) Several large autopsy and epidemiological studies have established that the prevalence of atherosclerotic vascular disease (ASVD) is higher in diabetic patients than in nondiabetic subjects. 1 Most of the studies, however, have not taken into consideration the type of diabetes even though it is probable that most of the diabetic patients in these studies dealing with middle-aged or elderly subjects had noninsulin-dependent diabetes (NIDDM). It has been shown that hyperinsulinemia is an independent risk factor for coronary heart disease (CHD) in nondiabetic subjects. 2 ' 3i A From cross-sectional studies, there is some evidence that hyperinsulinemia may be an independent risk factor for ASVD in diabetic patients also 5, e. 7 However, data concerning the predictive value of elevated plasma insulin levels for risk of ASVD in diabetic patients is, so far, scarce and controversial. 89 In patients with insulin-dependent diabetes mellitus (IDDM), insulin treatment results in elevated plasma free insulin levels in peripheral blood as compared to physiologic insulin levels in nondiabetics. Patients with NIDDM are often endogenously hyperinsulinemic due to their insulin resistance. 1 Thus, many patients with NIDDM who are From the Rehabilitation Research Centre of the Social Insurance Institution, Turku, and the Department of Medicine, Kuopk), University Central Hospital, Kuopio, Finland. An Academy of Finland grant to the authors from the University of Kuopio is gratefully acknowledged. Address for reprints: Tapani Ronnemaa, M.D., Rehabilitation Research Centre of the Social Insurance Institution, Peltolantie 3, SF-272 Turku, Finland. Received September 28, 1987; revision accepted December 16, 1987. treated with insulin may have considerable hyperinsulinemia and, accordingly, they may be at particularly high risk for ASVD. Because glucagon-stimulated C-peptide determination can differentiate between insulin-using NIDDM and IDDM patients, 11 a study of the prevalence of ASVD and C-peptide secretion capacity in insulin-treated patients may elucidate the possible difference in the ASVD risk between NIDDM and IDDM patients. Moreover, in patients being treated with similar doses of exogenous insulin, those who secrete more C-peptide probably have higher total (endogenous + exogenous) peripheral insulin levels. Therefore, C-peptide determinations might be used to study the association of hyperinsulinemia with the occurrence of ASVD in insulin-treated diabetic patients. The purpose of the present study was to investigate the prevalence of ASVD and its association with cardiovascular risk factors in a representative sample of insulin-treated diabetic patients whose onset of diabetes came at age 3 years or older and who had various degrees of endogenous insulin secretion determined by glucagon-stimulated C-peptide measurements. Subjects Methods A cross-sectional study to compare the prevalence of atherosclerotic vascular disease and its risk factors in middle-aged, maturity-onset, diabetic patients and in nondiabetic subjects was performed during 1982 to 1984 in East and West Finland. The detailed description of the procedure of obtaining representative study populations is re- 237

238 ARTERIOSCLEROSIS VOL 8, No 3, MAY/JUNE 1988 ported elsewhere. 12 Trie basic study population was chosen from the national register of the Social Insurance Institution which includes all Finnish citizens who receive antidiabetic medication. Briefly, the inclusion criteria were the following: 1) age 45 to 64 years, 2) diabetes diagnosed at the age of 3 years or later, and 3) area of residence and place of birth in the Kuopio University Central Hospital district (East Finland) or in the Turku University Central Hospital district (West Finland). Altogether, 567 diabetic patients from East Finland and 639 diabetic patients from West Finland participated; participation rates were 83% and 73%, respectively. Of these diabetic patients, 12 from East Finland and 158 patients from West Finland were having insulin treatment at the time of examination. C-peptide measurement after intravenous glucagon was performed in 112 patients from East Finland and in 151 patients from West Finland (i.e., in 93% and 96%, respectively, of the insulin-treated patients). These 263 patients, 142 men and 121 women, formed the final study population. The nondiabetic persons came from a random sample of all persons, ages 45 to 64 years, registered in Kuopio and Turku University Central Hospital districts. Altogether, 313 nondiabetic men from East Finland, 325 men from West Finland, 336 nondiabetic women from East Finland, and 399 women from West Finland participated in the study. All subjects gave their informed consent to the study. Methods The study program was carried out during one outpatient visit to the Rehabilitation Research Centre of the Social Insurance Institution in Turku or to the Clinical Research Unit of the University of Kuopio. The study protocol was in accordance with the appropriate institutional guidelines. myocardial infarction (Ml) was defined as follows: Minnesota Code 13 1.1 or 1.2 in resting electrocardiogram (ECG) or definite myocardial infarction based on medical records according to the World Health Organization (WHO) criteria. 14 or possible coronary heart disease (CHD) was defined as follows: definite or possible Ml based on medical records, 14 or typical angina pectoris as assessed by Rose's cardiovascular questionnaire 13 or Minnesota Code 1.1 to 1.3, 4.1 to 4.3, 5.1 to 5.3, or 7.1 in resting ECG ("ischemic ECG changes"). Intermittent claudication was assessed by Rose's cardiovascular questionnaire. 13 The WHO criteria for definite stroke were used to ascertain previous stroke. 15 Body mass index (BMI) was calculated from the formula: BMI = weight (kg)/height (m) 2. Physical activity was defined as regular physical activity during leisure time for at least 3 minutes at least twice a week. Smoking and alcohol consumption were defined as current smoking or weekly alcohol use. Hypertension was defined as the use of antihypertensive drugs, actual systolic blood pressure >16 mm Hg, or diastolic blood pressure &95 mm Hg. The procedure for blood pressure measurement has been described elsewhere. 12 The level of elevated serum creatinine was taken as 14 /xmol/l. Proteinuria was defined as urinary protein concentration >.5 g/l in a random sample. Biochemical Methods All blood specimens were taken at 8 a.m. after a 12-hour fast with no interruption of drug treatment. The analyses were performed in duplicate except for glucosylated hemoglobin A v Glucosylated hemoglobin A, was determined by affinity chromatography (Isolab Incorporated, Akron, Ohio). Serum cholesterol and triglycerides were determined enzymatically (Boehringer, Mannheim, West Germany). The day-to-day variation in cholesterol measurement was 2.2% and in triglyceride measurement, 4.4%; the intraassay variation was 1.6% in cholesterol and 2.6% in triglycerides. In all statistical analyses, triglycerides were handled after logarithmic transformation. Serum high density lipoprotein (HDL) cholesterol was determined enzymatically after precipitation of low density lipoprotein (LDL) and very bw density lipoprotein (VLDL) with dextran sulphate. 16 The day-to-day and intra-assay variations for HDL cholesterol were 3.8% and 1.7%, respectively. Serum apolipoproteins A-l and B were determined by radioimmunoassay. 17 The day-to-day and intra-assay variations for apoprotein A-l were 8.6% and 6.2%, respectively. The respective figures for apoprotein B were 5.9% and 4.7%. The endogenous insulin secretion capacity was assessed by plasma C-peptide measurement 6 minutes after intravenous injection of 1 mg of glucagon. 18 The test was performed 1.5 hours after breakfast and a normal morning dose of insulin to avoid hypoglycemia during the test. Hypoglycemia is associated with erroneously low, stimulated C-peptide values, 19 but none of our patients had plasma glucose below 4. mmol/l during the test. Plasma C-peptide was assessed by radioimmunoassay (antiserum M 123, Novo Industri, Copenhagen, Denmark). 2 Statistical Methods Statistical analyses of the data were carried out using OSIRIS computer programs (1981) as well as programs developed at the Social Insurance Institution. Student's f test (continuous variables) or % 2 test (categorical variables) was used to test the significance of the difference between the two groups. In the comparison of more than two groups, analysis of variance (ANOVA) or x 2 test was used. Analysis of covariance was used in the adjustment of confounding factors when comparing the four patient groups formed on the basis of C-peptide response to glucagon. Pearson or Spearman correlation coefficients were calculated to indicate the correlations between continuous variables. Age-adjusted prevalence rates were calculated by a direct standardization method 21 over four 5-year age groups using the total 198 Finnish population, ages 45 to 64 years, as a standard population. The Ftest was used to test differences of prevalences, controlling for the effect of age. 22 Logistic analyses based on the maximum likelihood method and the Wald test were calculated using SAS computer programs. Univariate logistic analyses were used to assess the association of a single cardiovascular risk factor with the occurrence of different atherosclerotic manifestations. Multivariate logistic analyses were used to assess the

ATHEROSCLEROSIS IN DIABETIC PATIENTS Rflnnemaa et al. 239 independent contribution of the risk factors to the occurrence of ASVD. Logistic models were also computed by using a backward stepwise procedure, which first fits the model to all variables and then eliminates variables one by one on the basis of maximum likelihood statistics. In backward stepwise logistic models, a statistic called R, which measures the predictive ability of the model (including those variables having an independent association with the dependent variable), was used. The R statistic is similar to the multiple correlation coefficient in the normal setting, after a correction for the number of variables estimated. 23 Odds ratios and their 95% confidence limits were calculated using the data from logistic analyses to estimate the relative risks. 2124 Approval of Ethics Committees Approval for this study was given by the Ethics Committees of the University Central Hospital of Turku and the University of Kuopio. Results Table 1 shows the characteristics of the diabetic patients according to sex and area of residence. The only significant East-West difference in the background variables was the higher prevalence of hypertension in women and the higher cholesterol and apoprotein A-l in men in the East. The distribution of stimulated plasma C-peptide values differed to some degree between the groups. Because no essential regional differences between diabetic patients were observed, the patients from East and West Finland were pooled in the subsequent data analyses. Table 2 (men) and Table 3 (women) show the risk factor levels in diabetic patients according to the degree of endogenous insulin secretion and in nondiabetic subjects. The four C-peptide groups differed from each other with respect to age, BMI, and duration of diabetes. Subjects in the highest C-peptide group (> 1. nmol/l) were slightly older and heavier and had a somewhat shorter duration of diabetes than patients in the group with lowest C-peptide secretion. The four C-peptide groups did not differ in glycemic control, physical activity, use of alcohol, serum cholesterol, or in prevalence of elevated serum creatinine or proteinuria (analysis of variance or x 2 test). No patient had a serum creatinine level above 2 /xmol/1. The insulin dose per kilogram of body weight was slightly lower in men with stimulated C-peptide of so.2 nmol/l than in men with stimulated C-peptide of <.2 nmol/l. BMI and serum triglycerides in both sexes and apoprotein B and the proportion of hypertensives in women increased with increasing C-peptide secretion. Serum HDL cholesterol and apoprotein A-l decreased with increasing endogenous insulin secretion. The differences between the four patient groups in HDL cholesterol, triglycerides, apoprotein A-l, and apoprotein B persisted after the adjustment for age and BMI (analysis of covariance), except for triglycerides in men and apoprotein B in women. As compared to diabetic patients in the lowest C-peptide group, nondiabetic subjects in both sexes had lower HDL cholesterol and apoprotein A-l, high- Table 1. Characteristics of Diabetic Patients by Sex and Area of Residence Men Women Age (yrs) BMI (kg/m 2 ) Duration of diabetes (yrs) GHbA, (%) Insulin dose (U/day) Insulin dose (U/kg/day) Stimulated C-peptide + <.2 nmol/l (%).2 to.59 nmol/l (%).6 to 1. nmol/l (%) > 1. nmol/l (%) (mmol/l) HDL cholesterol (mmol/l) Trigtycerides (mmol/l) Apo A-l (g/l) ApoB(g/l) Smokers (%) Hypertensives (%) Alcohol users (%) Physically active (%) Elevated serum creatinine (%) Proteinuria (%) East Finland (n = 53) 55.1 ±.7 25.3 ±.4 12.6±.7 1.1 ±.2 44.9 ±1.9.6 ±.3 57 21 13 9 6.86±.23t 1.51 ±.7 1.75 ±.15 1.16±.3t 1.18 +.5 26 47 62 32 8 17 West Finland (n = 89) 55.6 ±.5 25.7±.4 12.7±.7 9.9 ±.2 47.5+1.9.6 ±.2 42 31 6 21 6.12±.13 1.37 +.4 1.5 +.1 1.6 ±.2 1.1±.3 27 33 61 37 2 16 East Finland (n = 59) 57.5 ±.8 26.2±.5 12.1 ±.6 1.3 ±.3 39.4 ±1.8.62 ±.3 36 1 15 39 7.29 ±.27 1.5 +.6 2.77 +.6 1.15±.3 1.22 ±.4 3 71* 15 39 3 17 West Finland (n = 62) 56.7 ±.7 26.9 ±.6 12.8 +.8 1.8 ±.2 45.6 ±2.3.65±.3 63 11 8 18 6.97±.19 1.62 ±.6 1.88 ±.23 1.17±.2 1.19±.4 14 46 32 4 2 1 Values are means + SEM. *p<.5, tp<.1 for the difference between East and West Finland. $The distribution is different between East and West Finland in men. (p<.5) and women (p<.1).

24 ARTERIOSCLEROSIS VOL 8, No 3, MAY/JUNE 1988 Table 2. Characteristics and Coronary Heart Disease Risk Factors of Diabetic Men by Endogenous Insulin Secretion and of Nondlabetic Men Stimulated plasma C-peptide (nmol/l) Age (yrs) BMI (kg/m 2 ) Duration of diabetes (yrs) GHbA, (%) Insulin dose (U/day) Insulin dose (U/kg/day) (mmol/1) HDL cholesterol (mmol/1) Trigrycerides (mmol/1) Apo A-l (g/l) Apo B (g/l) Smokers (%) Hypertensives (%) Alcohol users (%) Physically active (%) Elevated serum creatlnlne (%) Proteinuria (%) <.2 (n = 67) 53.8±.7 24.2 ±.3 14.1 ±.8 1. ±.2 49.1 ±1.9.67 ±.3 6.48±.18 1.59 ±.5 1.34 ±.9 1.18±.3 1.9 ±.4 36 31 67 39 1 16.2 to.59 (n = 39) 55.2 ±.6 25.4 ±.5* 12.6 +.9 1.1+.2 44.4 ±3..57 ±.3* 6.23 ±.26 1.35 ±.6+ 1.58±.18 1.6 ±.2* 1.13 +.6 18 36 56 26 5 15.6 to 1. (n = 12) 56.2 ±1.4 27.4 ±.9* 1.1 ±1.4* 9.5±.3 42. ±3.6.51 ±.4+ 6.75 ±.41 1.3±.11* 1.88±.39 1.11 ±.4 1.23 ±.7 8 58 5 42 17 25 >1. (n = 24) 57.5±.9f 28.7 ±.6* 1.2±1.f 9.7 ±.3 45.1 ±3.8.51 ±.4+ 6.27 ±.22 1.11 ±.7* 2.19 ±.24*.95 ±.3* 1.2 ±.5 25 54 63 33 4 13 ANOVA or x 2 test p<.5 p<.5 p<.1 p<.1 Nondiabetics (n = 638) 54.2 ±.2 26.1 ±.1* 6.64 ±.5 1.35 ±.1* 1.54 ±.3* 1.8±.1t 1.15±.1 31 29 72 33 2* Values are means ± SEM. *p<.5, tp<.1, *p<.u1 as compared to groups with <.2 nmol/l. The differences in serum lipids and lipoproteins persisted after adjustment for age. After adjustment for BMI, the differences persisted except for trigiyce rides. Table 3. Characteristics and Coronary Heart Disease Risk Factors of Diabetic Women by Endogenous Insulin Secretion and of Nondlabetic Women Stimulated plasma C-peptide (nmol/l) Age (yrs) BMI (kg/m 2 ) Duration of diabetes (yrs) GHbA,(%) Insulin dose (U/day) Insulin dose (U/kg/day) (mmol/1) HDL cholesterol (mmol/1) Triglycertdes (mmol/1) Apo A-l (g/l) Apo B (g/l) Smokers (%) Hypertensives (%) Alcohol users (%) Physically active (%) Elevated serum creatinine (%) Proteinuria (%) <.2 (n = 6) 55.5 ±.7 24.8 ±.5 13.8 ±.8 1.7 ±.2 41. ±2.2.64 ±.4 6.96±.19 1.8 ±.5 1.31 ±.7 1.24 ±.2 1.12±.4 17 42 28 47 7.2 to.59 (n = 13) 54.3 ±1.7 26.8 ±1.4 1.8±1.4 1.2 +.4 31.2 + 3.1*.46 +.3* 7.15 ±.37 1.8±.14 1.37±.17 1.21 ±.6 1.17±.7 62 46 46 8.6 to 1. (n = 14) 58.6 ±1.3 28.2±1.1f 12.9 ±.6 11.1 ±.8 47.9±5.1.65 ±.6 7.39 ±.42 1.31 ±.7* 3.3 ±.63+ 1.7 ±.5+ 1.35 ±.1* 7 71* 14 29 31 >1. (n = 34) 6.3 ±.8* 29. ±.8* 1.3 ±.6* 1.3 ±.3 47.5 ±2.2.68 ±.4 7.3 ±.42 1.16 ±.6* 4.16±1.1* 1.3 ±.3* 1.31 ±.6+ * 82* 12 29 9 21 ANOVA or x 2 test p<.1 p<.1 p<.5 p<.1 p<.5 Nondiabetics (n = 735) 54.3 ±.2 27. +.2* 6.97 ±.5 1.62 ±.1* 1.33 ±.2 1.17±.1f 1.14 +.1 1 35 28 31* 1* Values are means ± SEM. *p<.5, tp<.1, * as compared to groups with <.2O nmol/l. The differences in serum lipids and lipoproteins persisted after adjustment for age. After adjustment for BMI, the differences persisted except for apoprotein B. er BMI, and lower prevalence of proteinuria. In addition, nondiabetic men had higher triglyceride levels and nondiabetic women were less active physically than the corresponding diabetic patients. Postglucagon plasma C-peptide concentration correlated positively with serum triglycerides and apoprotein B and negatively with serum HDL cholesterol and apoprotein A-l in both sexes (Table 4). C-peptide secretion did not correlate significantly with serum cholesterol. Apoprotein B correlated strongly with cholesterol and apoprotein A-l cone-

ATHEROSCLEROSIS IN DIABETIC PATIENTS Ronnemaa et al. 241 lated with HDL cholesterol in both sexes. The Spearman correlation coefficient between stimulated plasma C-peptjde and insulin dose/kg was -.18 (p<.1). For study of the association of endogenous insulin secretion with ASVD, the patients were divided into two groups: those with stimulated C-peptide level below.2 nmol/l and those with C-peptide level of ==.2 nmol/l. This limit was chosen because patients with glucagon-stlmulated C-peptide below the limit of.2 nmol/l have a high probability of insulin-dependent diabetes mellitus. 11 The age-adjusted prevalences of definite myocardial infarction, definite or possible coronary heart disease, and intermittent claudication were higher in those with considerable persistent endogenous insulin secretion (Table 5). The differences were more clear-cut in women than in men. The results were similar for patients from West and East Finland (data not shown). The prevalences of definite Ml and definite or possible CHD were similar in diabetics in the low C-peptide group and in nondiabetic subjects, and the prevalences of Intermittent claudication and stroke tended to be slightly lower in nondiabetic subjects than in patients with low or absent C-peptide secretion. Table 4. Correlation Coefficients (Pearson) between Stimulated Plasma C-Peptlde and Serum Llpld and Llpoproteln Concentrations Men (n = 142) HDL cholesterol Log triglycerides ApoA-l ApoB Women (n = 121) HDL cholesterol Log triglycerides ApoA-l ApoB.1 -.43*.35$ -.4*.17*.9 -.58*.55$ -.4$.28t C-peptJde.14.5$.16.84$..57$.1.85$ HDL cholesterol -.45f.8$ -.29$ -.61$.69$ -.33$ Log triglyce rides -.28$.72$ -.43$.7$ Apoprotein A-l -.18* -.18* *p<.5, fp<.1, $ for the significance of the correlation coefficient. Because the association of high C-peptide secretion with the prevalence of ASVD was evident in both sexes and because the distribution of patients with stimulated C- peptide below.2 nmol/l and >.2 nmol/l was similar in both sexes (47% and 53% in men, 5% and 5% in women, difference by x 2 test, ), men and women were combined in subsequent logistic regression analyses. In univariate logistic regression analyses, C-peptide secretion was significantly associated with definite Ml, definite or possible CHD, and definite stroke (Table 6). Insulin dose showed no association with ASVD. Among the classic risk factors for CHD, cholesterol, triglycerides, and apoprotein B showed the most consistent positive, and HDL cholesterol, the most consistent negative, associations with ASVD. In multjvariate logistic regression analyses that included common cardiovascular risk factors, apoproteins A-l and B were not included because of their strong correlation with HDL cholesterol and cholesterol, respectively (Table 4). In these analyses, C-peptide secretion was still strongly associated with the prevalence of definite or possible CHD (Table 7). and smoking tended to show a positive association with Ml, and HDL cholesterol showed an inverse association with Ml. Age was positively associated with definite or possible CHD. Triglycerides, physical activity, and residence in East Finland were positively associated, and insulin dose per kilogram was inversely associated, with intermittent claudication. Alcohol intake was inversely, and male sex was positively, associated with stroke. In backward stepwise logistic models, the results were similar to those in the multivariate logistic analyses presented above. However, in the backward stepwise model, the association of stimulated plasma C-peptide with definite Ml was statistically significant and the association with definite or possible CHD was even more apparent. In addition, the association of triglycerides with definite or possible CHD and with definite stroke was statistically significant. The R statistic was.26 for definite stroke and was somewhat higher, between.33 and.36, for other manifestations of ASVD. The relative risk estimates (odds ratios) for definite myocardial infarction and definite or possible CHD in patients with glucagon-stimulated C-peptlde level &.2 nmol/l as compared to those with low or absent secretion were 3. Table 5. Age-Adjusted Prevalences of Atherosclerotic Manifestations In Insulin-Treated Diabetic Subjects by Sex and Stimulated Plasma C-Peptlde Level and In Nondiabetic Subjects Ml or possible CHD Intermittent claudication stroke Men Diabetic C-peptide (nmol/l) <.2 a.2 (n = 67) (n = 75) 7.4 (5) (12) 33.8 (23) (39) 9. 3.2 (6) (2) 16.7 49.3* 11.3 13.2* (7) (9) Nondiabetic (n = 638) 7.4 (48) 29.9 (191) 4.1* (27) 1..1 (7) Women Diabetic C-peptide (nmol/l) <.2O ao.2 (n = 6) (n = 61) 2.5 (2) 16.4$ (11) 31.5(2) 65.4 (48) 3.6 1.2 (3) (D 16.8t 3.5 (11) (3) Values are percentages. The number of cases is shown In parentheses. *p<.1, tp<.5, $p<.1, as compared to the low C-peptide group. Ml = myocardial infarction; CHD = coronary heart disease. Nondiabetic (n = 735) 1.9 31.7 1.3t.3 (14) (234) (9) (2) Men + women Diabetic C-peptide (nmot/1) <.2 ao.2 (n=127) (n = 136) 5.2 32.9 6.8 2. (7) (43) (9) (3) 16.8$ (23) 54.6 (87) 13.6 (18) 9.3t(12) Nondiabetic (n = 1373) 4.5 3.8 2.6*.6t (62) (425) (36) (9)

242 ARTERIOSCLEROSIS VOL 8, No 3, MAY/JUNE 1988 and 2.91, respectively (Table 8). The corresponding figures after adjustment for all common risk factors for ASVD were 2.69 and 2.38. The association between C-peptide response and ASVD was also analyzed separately in persons with stimulated C-peptide values of <.2 nmol/l or a.2 nmol/l. Among persons with C-peptide values of <.2 nmol/l, patients with a C-peptide level above the detection limit (.17 nmol/l) had a higher age-adjusted prevalence of definite or possible CHD and claudication than did patients with undetectable C-peptide (prevalences for CHD were 5.5% vs. 27.4%, p<.5, and those for claudication were 18.5% vs. 3.9%, p <.5). Similarly, among persons with a C-peptide value of a.2 nmol/l, patients with a C-peptide level above 1. nmol/l had a higher age-adjusted prevalence of definite or possible CHD and claudication than did patients with a C-peptide level betweeni.2 and 1. nmol/l (prevalences for CHD ' were 54. 6% vs. 48.7%, p<.5 and those for claudication were 16.% vs. 9.9%, C-peptidel HDL cholesterol Trigtyce rides Apo A-l ApoB Age Duration of diabetes Insulin dose/kg BMI Smoking Hypertension Alcohol intake Physical activity Male sex Residence in East Finland Ml + 2.77+. + 2.58+. -3.75 + 4.6-1.5 + 4.41 + 2.19t +.82 -.45 + 1.23 + 1.19 + 1.2 -.91-1.65* +.31 +.48 or possible CHD + 4.8 + 2.17f -3.16* + 4.42-2.72* + 3.59 + 4.49-2.8+ +.76 + 3.* -1.33 + 2.63t -1.94* -.62-2.2+ +.41 Table 6. Association between Atherosclerotic Manlfestatlons and Their Risk Factors In Insulin-Treated Diabetic Subjects Intermittent claudication + 1.6 + 3.5* -2.f + 4.13 -.85 + 3.99 +.83 +.46 -.97 + 1.9 -.3 + 2.41 + -1.24 -.4 -.66 + 3.25* stroke + 2.11 + +.44-2.5+ + 2.35+ -1.35 + 1.36 +.42 -.81 -.19 + 1.35 + 1.47 + 1.46-1.4 -.32 + 1.5 -.21 Values are (values (beta/standard error) of univariate logistic regression analysis. *p<.1, tp<.5, *p<.1, ; anaiyzed as a dichotomized variable (glucagon-stimulated value 2=.2 nmol/l). Ml = myocardial infarction; CHD = coronary heart disease. p<.5). These associations between C-peptide level and ASVD persisted after adjustment for other cardiovascular risk factors (multivariate logistic regression analysis, data not shown) in both the high (>.2 nmol/i) and low (<.2 nmol/l) C-peptide subgroups. Table 7. Association between Atherosclerotic Manifestations and Their Risk Factors In Insulin-Treated Diabetic Subjects C-peptidell HDL cholesterol Triglycerides Age Duration of diabetes Insulin dose/kg BMI Smoking Hypertension Alcohol intake Physical activity Male sex Residence in East Finland Ml + 1.61 + 1.59-2.27+ +.11 + 1.26 + 2.9+ -1.25 -.44 + 2.8+ -.52 -.73 -.8 +. or possible CHD + 2.57* +.78 -.56 + 1.5 + 2.87* -1.87* + 1.33 +.1 +.3 -.8 -.63 +.47 -.79 +.34 -.21 Backward stepwise logistic model C-peptidell + 2.1 + + 3.14* + 2.47+ HDL cholesterol -2.9* Triglycerides + 2.69* Age + 3.6* Duration of diabetes + 2.1+ Smoking + 2.31 + Hypertension Alcohol intake Male sex Residence in East Finland R statistic.356.331 Significance Intermittent claudication +.22 +.49 -.59 + 2.19+ -.41 + 1.68* -2.16+ -.2 +.88 +.37 -.66 +1.78* +.35 + 2.74* + 3.67 + 2.79*.363 stroke + 1.4 -.22 -.4 +.98 +. -.12 +.25 +.73 + 1.48 +.64-1.98+ + 1.5 + 2.9+ -.6 + 2.66* -2.2+ + 2.54+.263 p=.3 Values are rvalues (beta/standard error) of multivariate logistic regression analysis. *p<.1, fp<.5, *p<.1, ; analyzed as a dichotomized variable (glucagon-stimulated value a.2 nmol/l). Ml = myocardial infarction; CHD = coronary heart disease. Table 8. Atherosclerotic Manifestations in Insulin-Treated Diabetic Subjects. Odds Ratios between Patients with Persistent and Low or No Endogenous Insulin Secretion Capacity Odds ratio A B Ml 3. (1.22 to 7.39) 2.69 (.8 to 9.) or possible CHD 2.91 (1.72 to 4.92) 2.38 (1.23 to 4.6) Intermittent claudication 1.89 (.8 to 4.48) 1.15 (.35 to 3.78) stroke 4.8 (1.9 to 15.23) 2.45 (.46 to 13.15) The odds ratios and, In parentheses, the 95% confidence intervals (patients with glucagon-stimulated C-peptide 2:.2 nmol/l vs. those with C-peptide <.2 nmol/l) are as follows: A = after age-adjustment, B = after adjustment for cholesterol, HDL cholesterol, triglycerides, age, duration of diabetes, insulin dose/kg, BMI, smoking, hypertension, alcohol Intake, physical activity, sex, and area of residence.

ATHEROSCLEROSIS IN DIABETIC PATIENTS Rflnnemaa et al. 243 Discussion The present study showed that the risk for ASVD in middle-aged, insulin-treated, diabetic patients whose diabetes had been diagnosed after the age of 3 years was higher in patients with persistent endogenous insulin secretion than in those with low or no secretion. Patients with postglucagon C-peptide values below.2 nmol/l have a high probability of having IDDM, 11 while most patients with higher C-peptide secretion have NIDDM. Therefore, the results suggest that among insulin-treated patients with a relatively late onset of diabetes, those who have NIDDM are more prone to atherosclerosis than those who have IDDM. The prevalence of atherosclerotic disease in patients with stimulated C-peptide levels of s.2 nmol/l was similar to those in our study of NIDDM patients, most of whom were receiving treatment with oral antidiabetic drugs. 12 On the other hand, the prevalences of ASVD in patients with C-peptide levels of <.2 nmol/l were close to those in nondiabetics of the same age. In IDDM patients who have an onset of diabetes before the age of 2 years and a 2- to 4-year duration of diabetes, the CHD mortality and morbidity is several times higher than in nondiabetic subjects. 25 This excessive occurrence of CHD in patients with juvenile onset IDDM begins to manifest after age 3 years and is particularly marked in patients with diabetic nephropathy. 25 In our patients with true IDDM, but with a late onset of diabetes, the mean duration of diabetes was only 14 years and the prevalence of nephropathy was relatively tow. Accordingly, our results do not exclude the possibility that IDDM patients with onset of diabetes after the age of 3 years also have an increased risk for CHD when their diabetes has lasted for more than 2 years or if they have developed nephropathy. A marked regional difference in the prevalence of atherosclerotic diseases has been found between NIDDM patients living in East or West Finland; patients in East Finland are more prone to ASVD. 12 This difference was not found in the present study of insulin-treated patients, but the limited number of cases does not allow conclusions to be drawn about living in East Finland as a risk factor for ASVD in Insulin-treated diabetic patients. The insulin dose was inversely related to C-peptide response in men, but not in women. The absence of this relation in women may be an indication of higher insulin resistance in women with high endogenous insulin secretion as compared to men. Diabetic men with low or no endogenous insulin secretion had a higher prevalence of definite Ml and claudication than did women, but there was no apparent sex difference in patients with higher C-peptide response. Earlier studies have shown that diabetes is associated with diminution of the normal sex difference for risk of CHD. 28 Our results are concordant with this and suggest that the abolishment of the sex difference is confined to insulin-treated diabetics with a high endogenous insulin secretion. In the group of patients with postglucagon plasma C- pepttde below.2 nmol/l (all IDDM patients), the presence of detectable C-peptide was positively associated with ASVD. This is in contrast to findings concerning microangiopathy, which have suggested that persistent C- peptide secretion may protect against proliferative retinopathy in diabetic patients. 27 There are several possible reasons for the different prevalences of atherosclerotic diseases in the two patient groups formed according to the degree of endogenous insulin secretion capacity. Differences in age, smoking, alcohol, and glycemic control are probably excluded as possible reasons. The duration of diabetes was approximately 3 years shorter in patients with persistent insulin secretion than in those with low or no insulin secretion, suggesting that the difference in ASVD prevalence is not related to the duration of diabetes. However, patients with NIDDM may have had asymptomatic diabetes and, before that, abnormal glucose tolerance, for several years before diabetes was diagnosed. By contrast, the time lag between the onset and the diagnosis of diabetes is much shorter in patients with IDDM. Therefore, the results of the effect of the duration of diabetes should be interpreted with caution. Because the present study was cross-sectional, a possible difference in fatalities during Ml or stroke between patients with persistent and those with low or no insulin secretion may have led to bias affecting the results. However, to our knowledge, there are no data that compare the fatalities during Ml or stroke in patients with IDDM and those with NIDDM. The higher prevalence of ASVD in patients with persistent insulin secretion may, in part, be explained by their more atherogenic serum lipid profile (i.e., low HDL cholesterol and high triglyceride levels) and also by the more frequent occurrence of hypertension in this group of patients as compared to patients with low or no endogenous insulin secretion. In the murtivariate analyses including C- peptide secretion capacity, the variables, cholesterol, triglycerides, tow HDL cholesterol, and smoking, were associated with ASVD, indicating that general cardiovascular risk factors are operative in insulin-treated, late-onset, diabetic patients. High postglucagon plasma C-peptide levels, however, showed an independent association with ASVD occurrence in multivariate analysis. The odds ratios between patients with high and those with low or no C- peptide secretion were positive for all manifestations of ASVD, although the ratios were statistically significant only for definite or possible CHD. Several possible explanations for the independent association between high endogenous insulin secretion and ASVD must be considered. First, postglucagon plasma C- peptide may be only an indicator of diabetes type, and the difference in ASDV prevalence in patients with high C- peptide response (patients with NIDDM) and persons with low or no C-peptide response (patients with IDDM) may be related to the metabolic differences between these two types of diabetes. The second possibility is hyperinsulinemia. Although plasma C-peptide concentrations do not necessarily strongly correlate with circulating endogenous insulin levels due to differences in the extraction of insulin and C- peptide by the liver, plasma C-peptide concentrations roughly reflect endogenous peripheral insulin levels in insulin-treated patients. Accordingly, it is likely that among the insulin-treated diabetic patients of the present study, those diabetics showing high postglucagon plasma C-pep-

244 ARTERIOSCLEROSIS VOL 8, No 3, MAY/JUNE 1988 tide levels averaged higher plasma insulin levels than did those with low or absent C-peptide secretions. This is probable because the mean daily insulin dose was similar in these two groups of diabetics; in the former group, the increase of plasma insulin level in the peripheral circulation caused by exogenous insulin treatment added to higher endogenous circulating plasma insulin levels. On the other hand, insulin may have direct atherogenic effects, for instance, by stimulating the proliferation of aortic smooth muscle cells. 28 The third possibility is that insulin resistance and related factors features characteristic of patients with NIDDM may be even more important underlying factors than hyperinsulinemia in the excessive occurrence of ASVD in insulin-treated patients with NIDDM. Increasingly, insulin treatment of patients with NIDDM in inadequate metabolic control is being recommended. 29 In the light of our results, the possibility that ASVD risk could be increased by insulin treatment of patients with persistent endogenous insulin secretion rrujst be considered. In conclusion, among insulin-treated diabetic patients who have a late onset of the disease, ASVD is more prevalent in those who have persistent endogenous insulin secretion capacity (patients with NIDDM) than in patients with low or no endogenous insulin secretion (patients with IDDM). This difference may be partly explained by the more atherogenic cardiovascular risk factor profile in patients with NIDDM, but most of this difference may be due to hyperinsulinemia, insulin resistance, and related factors characterizing patients with this type of diabetes. Acknowledgments We thank Jeremiah Stamler, Northwestern University Medical School, Chicago, Illinois, for valuable comments during the manuscript preparation. References 1. Pyorala K, Laakso M, Uusttupa M. Diabetes and atherosclerosis: an epidemiologic view. Diabet Metab Rev 1987;3: 463-524 2. Welborn TA, Wearne K. Coronary heart disease Incidence and cardiovascular mortality In Busselton with reference to glucose and Insulin concentration. Diabetes Care 1979;2: 154-16 3. Duclmetlere P, Eschwege E, Papoz L, Richard JL, Claude JR, Rosselln G. Relationship of plasma insulin level to the incidence of myocardial infarction and coronary heart disease mortality in a middle-aged population. Dlabetologia 198; 1925-21 4. Pyorala K, Savolalnen E, Kaukola S, Haapakoskl J. Plasma insulin as coronary heart disease risk factor: relationship to other risk factors and predictive value during 9Vi-year follow-up of the Helsinki Policemen Study population. Acta Med Scand 1985;(suppl 71):38-52 5. Kashyap ML, Maglll F, Rojaa L, Hoffman MM. Insulin and non-esterified fatty acid metabolism in asymptomatic diabetics and atherosclerotic subjects. Can Med Assoc J 197; 12:1165-1169 6. Santen RJ, Willis PW, Fajans SS. Atherosclerosis in diabetes mellitus. Correlations with serum lipid levels, adiposity and serum insulin level. Arch Intern Med 1972;13:833-843 7. Standl E, Janka HU. High serum insulin concentrations In relation to other cardiovascular risk factors In macrovascular disease of type 2 diabetes. Horm Metab Res 1985;15(suppl): 46-51 8. Jarrett RJ, McCartney P, Keen H. The Bedford Study: Ten year mortality rates In newly diagnosed diabetics, borderline diabetics and normoglycaemic controls and risk indices for coronary heart disease in borderline diabetics. Diabetologia 1982;22:7-84 9. Hlllson RM, Hockaday TDR, Mann Jl, Newton DJ. Hyperinsulinaemia is associated with development of electrocardiographic abnormalities In diabetics. Diabetes Res 1984;1: 143-149 1. Kolterman OG, Gray RS, Griffin J, et al. Receptor and postreceptor defects contribute to the insulin resistance in noninsulin-dependent diabetes mellitus. J Clin Invest 1981;68: 957-969 11. Madsbad S, Albertl KQMM, Binder C, et al. Role of residual Insulin secretion in protecting against ketoacidosis in Insulindependent diabetics. Br Med J 1979;2:1257-1259 12. Laakso M, Rdnnemaa T, Pyorala K, Kalllo V, Puukka P, Penttlla I. Atherosclerotic vascular disease and its risk factors in noninsulin-dependent diabetic and nondiabetic subjects in East and West Finland. Diabetes Care 1988 (in press) 13. Rose G, Blackburn H, Glllum RF, Prineas RJ. Cardiovascular survey methods. 2nd ed. Geneva: World Health Organization, 1982 14. World Health Organization. Proposal for the multinational monitoring of trends and determinants in cardiovascular disease and protocol (MONICA Project) Unpublished WHO document. WHO/MNC/82.1, Rev 1, Geneva: World Health Organization, 1983 15. Werifeld FD, ed. The national survey of stroke. Stroke 1981 ;12 (suppl l):32-35 16. Kostnar GM. Enzymatic determination of cholesterol In high density lipoprotein fractions prepared by polyanion precipitation. Clin Chem 1976;22:695 17. Lehtonen A, Marnletnl J, Inberg M, Maatela J, Alanen E, Nllttymakl K. Levels of serum lipids, apolipoprotelns A-l and B and pseudocholinesterase, and their discriminate values in patients with coronary bypass operation. Atherosclerosis 1986:59:215-221 18. Faber OK, Binder C. C-peptide response to glucagon. A test for the residual beta cell function. Diabetes 1977;26:65-61 19. ROnnemaa T. Practical aspects in performing the glucagon test in the measurement of C-peptide secretion in diabetic patients. Scand J Clin Lab Invest 1986;46:345-349 2. Hedlng LG. Radioimmunological determination of human C- peptide in serum. Diabetologia 1975; 11:541-548 21. Armltage P. Statistical methods in medical research. Oxford: Blackwell scientific publication, 1971:384-391, 426-^33 22. Searle SR. Linear models. New York: John Wiley & Sons, 1971:286-326 23. Atkinson AC. A note on the generalized information criterion for choice of a model. Biometrika 198;67:413-418 24. Lee J. Covariance adjustment of rates based on the multiple logistic regression model. J Chronic DIs 1981 ;34:415426 25. Krolewskl AS, Koslnskl EJ, Warram JH, et al. Magnitude and determinants of coronary artery disease in juvenile-onset, insulin-dependent diabetes mellitus. Am J Cardlol 1987; 59:75-755 26. Kannel WB, McGee DL. Diabetes and cardiovascular disease. The Framingham Study. JAMA 1979;241535-238 27. Mosler MA. Circulating C-peptide and diabetic retinopathy. Diabetes Res 1984;1:151-154 28. Stout RW, Blerman EL, Ross R. Effect of insulin on the proliferation of cultured primate arterial smooth muscle cells. Circ Res 1975;36.319-327 29. O'Rahllly SP, Nugent Z, Rudenskl AS, et al. Beta-cell dysfunction, rather than insulin insensitivity, is the primary defect in familial type 2 diabetes. Lancet 1986;2:36-364