Effect of Type 1 Diabetes on the Gender Difference in Coronary Artery Calcification: a Role for Insulin Resistance? : The Coronary Artery Calcification in Type 1 Diabetes (CACTI) Study

The data presented in this report were collected as part of the baseline examination of 1,420 participants in the Coronary Artery Calcification in Type 1 Diabetes (CACTI) study. Participants were 20–55 years of age and included 656 men and women with type 1 diabetes and 764 nondiabetic control subjects. All subjects were asymptomatic for CAD and had no history of coronary artery bypass graft, coronary angioplasty, or unstable angina. All patients with diabetes had been diagnosed when younger than 30 years, had been treated with insulin within 1 year of diagnosis, and had to have a disease duration of at least 10 years on enrollment. All nondiabetic control subjects had fasting blood glucose <110 mg/dl and were generally spouses, friends, and neighbors of the cases. All subjects provided informed consent and the study was approved by the Colorado Combined Institutional Review Board.

Participants completed the baseline examination between March 2000 and April 2002. Current height, weight, waist (measured at the smallest point between the 10th rib and the iliac crest over the bare skin) and hip (measured at the maximum circumference of the buttocks) were recorded, and BMI (weight/height²) and waist-to-hip ratio (WHR) were calculated. Resting systolic blood pressure (SBP) and fifth-phase diastolic blood pressure (DBP) were measured three times while the subjects were seated, and the second and the third measurements were averaged. Hypertension was defined as current SBP ≥135 mmHg or DBP ≥85 mmHg or current antihypertensive therapy. Participants completed a standardized questionnaire including medical history and medication inventory, Rose angina, current and past smoking status, physical activity, food frequency, daily insulin dose, and family history of diabetes, CAD, and hypertension.

After an overnight fast, blood was collected and centrifuged, and separated plasma was stored at 4°C until assayed. Total cholesterol and triglyceride levels were measured using standard enzymatic methods. HDL cholesterol was separated using dextran sulfate, and LDL cholesterol was calculated using the Friedewald formula. High-performance liquid chromatography was used to measure HbA_1c (HPLC; BioRad variant).

Insulin resistance was assessed as the inverse of the estimated glucose disposal rate (27), calculated according to the formula: estimated glucose disposal rate = 24.31 − 12.22*(WHR) − 3.29*(hypertension) − 0.568*(HbA_1c). The equation was derived from hyperinsulinemic-euglycemic clamps performed in 24 type 1 diabetic participants in the Pittsburgh Epidemiology of Diabetes Complications (EDC) Study (27).

An ultrafast Imatron C-150XLP EBCT scanner (Imatron, San Francisco, CA) was used to obtain two sets of high resolution, noncontrast, contiguous 3-mm tomographic images acquired at 100-ms exposure. Scanning started from near the lower margin of bifurcation of the main pulmonary artery with the subject breathholding for ∼35–45 s and proceeded caudally. Calcified coronary artery areas were identified as those with a minimum density of 130 Hounsfeld units (HU) and a minimum area of three pixels (1.03 mm²). A calcium score for each region was calculated by multiplying the area by the density score (1 for 130–199, 2 for 200–299, 3 for 300–399, and 4 for >399 HU). A total CAC score in Agatston units (AU) was calculated by adding up scores for all slices separately for left main, left anterior descending, circumflex, and right coronary arteries. The scanner was recalibrated every day with a phantom. No adjustments were made to noise due to excess BMI. Effective radiation dose for an EBCT sequence was 1.0 mSV for men and 1.3 mSV for women. A single technician obtained and scored all EBCT scans, and the average of two scores obtained 5-min apart was used.

An abdominal computed tomography scan at the L2-L3 levels was obtained on each participant. The L2-L3 disc space was located by counting the lumbar vertebra with L1 being the first non-rib-bearing vertebra. A single 6-mm thick image was obtained though the L2-L3 disc space during suspended respiration.

Differences in risk factors between subjects with type 1 diabetes (cases) and without diabetes (control subjects) in men and women were examined using ANOVA, adjusted for age. CAC scores were positively skewed with a high frequency of zero values. Therefore, logistic regression was used to examine the age-adjusted ORs of having any coronary calcification (CAC score >0) in diabetic patients versus control subjects, as well as the age-adjusted ORs of having any coronary calcification in men versus women. We tested the effect of gender on having any calcification (CAC score >0) and on having a CAC score ≥20, adjusted for covariates, in patients with type 1 diabetes and in nondiabetic control subjects. The covariates entered in the models were standard risk factors for CAD (age, total, HDL, and LDL cholesterol, triglycerides, smoking, SBP, DBP, and diabetes duration for type 1 diabetic patients only), estimated insulin resistance, individual components of the equation used to assess insulin resistance (HbA_1c, hypertension, and WHR), and insulin resistance syndrome-related factors (waist circumference, visceral fat measured at L23 level [IAF {intra-abdominal fat}L23], BMI) with different distributions by gender in type 1 diabetic patients and nondiabetic control subjects. To examine whether the ORs for diabetes-associated CAC were the same in men and in women, we included a diabetes-by-gender interaction and examined the effect of adjusting for CAD risk factors, estimated insulin resistance, and insulin resistance syndrome-related features on the interaction. Because triglyceride and visceral fat levels had skewed distributions they were logarithmically transformed before analysis.

Table 1 shows the baseline characteristics of type 1 diabetic patients and nondiabetic control subjects, stratified by gender and adjusted for age. Both men and women with type 1 diabetes had a significantly better fasting lipid profile (lower total and LDL cholesterol and triglyceride levels and higher HDL cholesterol levels) than nondiabetic control subjects, but the lipid profile was less beneficial with type 1 diabetes in women than in men. In men, type 1 diabetes was associated with slightly lower WHR and dramatically lower visceral fat (IAF L23), while women with type 1 diabetes had slightly higher BMI, WHR, and waist circumference but similar visceral fat compared with nondiabetic women. Estimated insulin resistance was higher in both men and women with type 1 diabetes compared with the nondiabetic control subjects of the same gender.

The distribution of CAC scores by diabetes status for each gender is shown in Fig. 1. The distribution was skewed, with a high frequency of 0 values. There were more diabetic than nondiabetic people with scores >0, especially among women (P = 0.01 for the age-adjusted diabetes by gender interaction). In each age-group examined, diabetes was associated with a higher prevalence of any calcification (Fig. 2, P < 0.001 for each gender, adjusted for age). In men, diabetes was associated with a 2.1-fold higher age-adjusted prevalence of CAC, while in women the age-adjusted effect of diabetes on CAC was 3.6 times higher (Table 2). In the nondiabetic group there was a larger gender difference in the prevalence of calcification than in type 1 diabetic subjects, so that the 4.6-fold higher age-adjusted OR for CAC in men versus women observed in control subjects was reduced to 2.5 in type 1 diabetes.

Table 3 shows the age-adjusted ORs for the association between CAC (>0) and risk factors in which the effect on diabetes-by-gender interaction was later examined, by diabetes status. The associations were significant for most factors, except for total and LDL cholesterol in patients with type 1 diabetes, and for smoking in both type 1 diabetic and nondiabetic participants.

In multivariate regression, estimated insulin resistance was associated with CAC in patients with type 1 diabetes and in nondiabetic control subjects, independent of age, gender, LDL and HDL cholesterol, smoking, and diabetes duration [ORs (95% CI) 1.6 (1.2–2.0) and 1.4 (1.2–1.7), respectively, P < 0.001 for 1-SD change in insulin resistance]. Among the components used to derive the insulin resistance equation, HbA_1c and WHR were associated with CAC in both type 1 diabetic patients and nondiabetic control subjects, as were waist and IAF L23, independent of age, gender, LDL and HDL cholesterol, smoking, and diabetes duration (data not shown).

Table 4 shows which covariates explain the effect of gender (male versus female) on CAC prevalence (CAC >0) in type 1 diabetic patients and nondiabetic control subjects. In model 1, adjustment for LDL and HDL cholesterol and diabetes duration (in type 1 diabetic patients only) in addition to age, explained some of the male excess in CAC in patients with type 1 diabetes and ∼40% of the male excess in CAC in control subjects. In model 2, addition of estimated insulin resistance further reduced, but did not completely eliminate, the male CAC excess in both type 1 diabetic patients and control subjects. In models 3–5 individual components of the insulin resistance equation are added to model 1: neither HbA_1c, nor hypertension appeared to contribute substantially to the male excess. However, WHR (the third component of the insulin resistance formula) explained a large part of the male CAC excess in both type 1 diabetic patients and nondiabetic control subjects (model 5). Adjustment for WHR (model 5), waist circumference (model 6), or IAF L23 (model 7) reduced the male CAC excess by a similar magnitude in type 1 diabetic patients and control subjects. As a result, the male CAC excess in diabetic subjects became statistically nonsignificant, although after accounting for these factors men were still 30–40% more likely to have CAC. Simultaneous adjustment for age, diabetes duration, HDL and LDL cholesterol, smoking, hypertension, HbA_1c, waist circumference, and visceral fat (model 8) did not further reduce the gender difference. Other standard CAD risk factors (total cholesterol, triglycerides, and BMI) explained little of the gender difference in CAC in both groups. Similar results were obtained with a different cut point for calcification (CAC >20 vs. <20) (data not shown).

Figure 3 shows the ORs for the CAC association with diabetes in women relative to men (the diabetes-by-gender interaction on CAC) for eight different models. The first model shows a 1.9-fold (95% CI 1.2–3.4, P = 0.01) higher OR for CAC associated with diabetes in women than in men, after controlling only for age. The next models show the effect of adjustment for insulin resistance-related factors on the significantly higher OR for calcification associated with diabetes in women than in men. On adjustment for HbA_1c (model 2) or hypertension (model 3), the diabetes-associated OR for CAC still remains significantly higher in women than in men. On adjustment for WHR (model 4), waist (model 5), visceral fat (model 6), or HDL and LDL cholesterol (model 7), the diabetes-by-gender interaction loses significance. Finally, in model 8, HDL and LDL cholesterol together with visceral fat and HbA_1c explain almost all of the increased diabetes-associated CAC in women versus men (model 8).

This study confirms earlier reports from the U.K. (15) showing that type 1 diabetes reduces the gender difference in CAC, or that type 1 diabetes has a greater effect on calcification in women than in men. In the British study, the gender difference in CAC was absent in patients with type 1 diabetes, whereas in CACTI the age-adjusted OR for CAC in men versus women with type 1 diabetes was 2.5. The factors explored in the British study had little effect in explaining the gender-diabetes interaction. In CACTI, men had higher estimated insulin resistance, more abdominal fat, higher WHR and waist circumference, and higher LDL and lower HDL cholesterol than women among both patients with type 1 diabetes and nondiabetic control subjects. The factors that contribute to the gender difference in calcification in the control population may be different from those in patients with diabetes. However, in CACTI, estimated insulin resistance assessed according to the equation developed by the EDC Study, LDL and HDL cholesterol, as well as body fat distribution, and each substantially explained the gender difference in CAC in control subjects and abolished the already reduced gender difference in calcification in subjects with diabetes (Table 4). In addition, type 1 diabetes had an unfavorable effect on fat distribution in women but not in men, while the beneficial lipid profile with type 1 diabetes was less pronounced in women than in men (Table 1). Therefore, on adjustment for either waist, WHR, visceral fat, or LDL and HDL cholesterol, the significant diabetes-by-gender interaction on coronary calcification disappeared (Fig. 3). The effect of WHR on the diabetes-by-gender interaction was not accounted for by BMI.

Recent studies have suggested that insulin resistance, relatively newly documented in type 1 diabetes (20–23), may contribute to the increased CAD risk in these patients (24–26,28). Excess coronary calcification has been reported in subjects with impaired glucose tolerance and insulin resistance (19). Our cross-sectional data show that estimated insulin resistance is independently associated with coronary calcification in patients with type 1 diabetes and substantially explains the gender difference in CAC in patients with type 1 diabetes and nondiabetic control subjects. Most of this effect is accounted for by the WHR, a component of the equation used to estimate insulin resistance (27) and a measure of abdominal fat (29). Visceral fat accumulation has been associated with reduced HDL cholesterol and raised LDL cholesterol, hypertension, and insulin resistance (30). Although the difference in WHR could statistically explain the gender difference in CAC, it may simply represent another way of determining sex. However, this explanation does not account for the association between WHR and CAC within each sex group or the fact that the association is independent of sex, in both patients with type 1 diabetes and control subjects.

In CACTI, type 1 diabetes was associated with adverse features related to central adiposity (higher WHR, larger waist circumference, and more intra abdominal fat) in women but not in men, i.e., women with type 1 diabetes had a more android deposition of adipose tissue (Table 1). By contrast, men with type 1 diabetes had lower WHR and IAF than nondiabetic men. The CACTI data therefore suggest that the diabetes-associated increase in coronary calcification, and possibly in CAD, may be accounted for by different factors, or may follow somewhat different pathways, in women than in men.

An inverse relationship between glycemic control and insulin sensitivity has been demonstrated in type 1 diabetes (31). CACTI patients had a fair glycemic control, as determined by an HbA_1c mean of 7.8 ± 1.2. Whether there is a threshold in HbA_1c levels associated with impaired hepatic insulin sensitivity, whether less than optimal glycemic control over a period of time longer than that determined by a single measurement contributes more to impaired insulin action, and whether these relationships are similar in men and women, is still unknown.

This study is larger than previous reports concerning coronary calcification in type 1 diabetes and includes a sizable nondiabetic control population. We obtained consistent results with different measures of central adiposity, with HDL and LDL cholesterol, with different calcification cut points, and with different statistical approaches. Nevertheless, our report has several important limitations. This is a cross-sectional association phase of the study and any inferences concerning causality have to await confirmation during prospective follow-up of the cohorts. Also, no direct assessment of insulin resistance is available for the study participants that would be reliable and valid in both type 1 diabetic patients and nondiabetic control subjects. Existing validated measures of insulin sensitivity/resistance such as euglycemic clamps or the insulin-boosted minimal model are not practical for large epidemiological studies. The estimated glucose disposal rate used in this study is an equation derived from euglycemic-hyperinsulinemic clamp studies performed on subjects with type 1 diabetes, which was significantly related to measured glucose disposal rate (r = 0.64) (27). However, it is hard to evaluate to what extent the formula measures insulin resistance in other cohorts of patients with type 1 diabetes. The equation has not been validated in nondiabetic control subjects, and any comparison of insulin resistance between the two groups in this study should be made carefully. WHR, although a component of the equation and reportedly a marker of insulin resistance in the general population (32), is primarily a measure of abdominal fat. While both men and women with type 1 diabetes had higher estimated insulin resistance than nondiabetic control subjects, diabetes was associated with lower central adiposity in men. Whether type 1 diabetic men and women are more insulin resistant than nondiabetic control subjects and whether this is the reason for their increased coronary calcification remains to be answered.

Similar to previous studies (15), the CACTI study excluded individuals with diagnosed CAD at entry. If CAD is undetected more often in women than in men, the exclusion of individuals with known CAD may have resulted in an underestimate of the gender difference in CAC prevalence in both type 1 diabetic and nondiabetic CACTI participants.

The validity of CAC as a measure of plaque burden in the general population is well established (9). There is concern that some of the signal detected by EBCT in type 1 diabetes represents medial rather than intimal calcification. Although pathology studies suggest that calcification in coronary arteries primarily involves the intimal layer (33), more research is needed to assess the potential of EBCT to discriminate between medial and intimal calcification in diabetes. Importantly for this study, however, coronary calcification has been shown to be related to coronary atherosclerosis on autopsy similarly in men and women (8,34).

In conclusion, the CACTI study has shown that type 1 diabetes increases the prevalence and the severity of coronary calcification and reduces the gender difference in CAC. Gender differences in insulin resistance-associated fat deposition, and HDL and LDL cholesterol distribution may explain why diabetes increases coronary calcification and likely coronary artery disease, in women relatively more than in men.

Support for this study was provided by the National Heart, Lung and Blood Institute (National Institutes of Health/NHLBI R01 HL61753), the Adult General Research Center (National Institutes of Health M01 RR00051), and Diabetes and Endocrinology Research Center Clinical Investigation Core (P30 DK57516).

This study was performed at the Adult General Research Center at the University of Colorado Health Sciences Center, at the Barbara Davis Center for Childhood Diabetes in Denver, Colorado, at the Colorado Heart Imaging Center in Denver, Colorado.