Abstract
Though Coronary Artery Disease (CAD) is the leading cause of death in Type 1 Diabetes (T1D), the mechanisms responsible for the greatly increased risk are poorly understood. In particular the role of glycaemic control is controversial with one study suggesting it predicts CAD mortality but not incidence. In this analysis, of the Pittsburgh Epidemiology of Diabetes Complications study cohort of T1D, we examine whether risk factors differ for CAD morbidity and mortality, with a specific focus on HbA1c and insulin dose.
Participants (n=592) were followed for 18 years for incident non-fatal and fatal CAD. Cox stepwise regression was used to determine the independent risk factors for non-fatal and fatal CAD.
Mean age and diabetes duration at study baseline were 29 and 20 years, respectively. There were 109 incident non-fatal and 48 fatal CAD events. Baseline HbA1c was an independent risk factor for fatal CAD, along with duration of diabetes and albuminuria.
In contrast, baseline lower insulin dose was strongly predictive of non-fatal CAD, as was lower renal function, higher diastolic blood pressure, and lipids.
HbA1c predicts CAD mortality while lower insulin dose and standard CAD risk factors predict CAD morbidity.
Introduction
Though Coronary Artery Disease (CAD) is the leading cause of death in Type 1 Diabetes (T1D), the mechanisms responsible for the greatly (up to 10-fold) increased risk are poorly understood. While both insulin excess and deficiency have been implicated (1,2), standard CAD risk factors, e.g. blood pressure and lipids, are also clearly operational. The role of glycaemic (e.g. HbA1c) control however is less clear. On one hand, a small study of older onset T1D cases, without renal disease, reports a significant association (3) and there is clear evidence that intensive glucose management in the Diabetes Control and Complications Trial (DCCT) was associated with major reductions of cardiovascular disease (CVD) events in an extended follow up (4). On the other hand, however, three cohort studies, a follow up of patients from the Hvidore Hospital in Denmark (5), the Pittsburgh Epidemiology of Diabetes Complications (EDC) study (6) and the Eurodiab study (7) all failed to show a significant and/or independent association. Finally, while the Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR), has consistently reported a modest association, most recently (8) a multivariable RR of 1.23 per 1% HbA1c (95% CI 1.12-1.35) for cardiovascular mortality, the association of HbA1c with morbidity (myocardial infarction) was not significant (9). This raises the possibility that glycaemia may relate more strongly to CAD mortality than to morbidity. In this analysis, we therefore address, in the Pittsburgh EDC study (for the first time as we have not hitherto had a sufficient number of events), the question as to whether risk factors are different for CAD morbidity and mortality, with a specific focus on HbA1c and insulin dose.
Methods
The study population is the Pittsburgh EDC Study, which includes subjects with childhood (<17 years old) onset type 1 diabetes diagnosed between 1950 and 1980 at Children’s Hospital of Pittsburgh. The 658 subjects participating at baseline were examined between 1986 and 1988 and followed biennially. All study participants provided informed consent and the University of Pittsburgh Institutional Review Board approved the study protocol.
Outcomes were nonfatal CAD or fatal CAD. Nonfatal CAD was defined as myocardial infarction confirmed on medical records or Q-waves (Minnesota codes 1.1 or 1.2) (10), or revascularisation procedures, including coronary artery bypass graft, angioplasty, and coronary endarterectomy, or coronary artery stenosis ≥50% without revascularisation. All deaths were investigated by obtaining autopsy, coroner, or medical records. Two physicians, (TO, JF), classified deaths according to Diabetes Epidemiology Research International procedures (11). Those classified as having myocardial infarction or CAD as the underlying cause of death formed the second outcome.
Information was collected by questionnaire concerning demographic characteristics, medical history, Beck Depression Inventory (12), and health care behaviors, as previously described (13,14). A history of smoking was defined as having smoked at least 100 cigarettes in a lifetime. Participants were weighed in light clothing on balance beam scale. Height was measured using a wall-mounted stadiometer.
Fasting blood samples were assayed for lipids, lipoproteins, hemoglobin (HbA1), creatinine, fibrinogen, and white blood cell count (WBC), as previously described (13,14). Stable glycosylated hemoglobin A1 (HbA1) was originally measured in saline-incubated samples by microcolumn cation exchange chromatography (Isolab, Akron, Ohio, USA). On October 26, 1987, the method was changed to high-performance liquid chromatography (HPLC) (Diamat, Bio-Rad Laboratories, Hercule s, CA, USA). The two methods were highly correlated (r = 0.95; Diamat HbA1 = 0.18±1.00 Isolab HbA1). The original HbA1 values have been converted converted to a Diabetes Control and Complications Trial (DCCT) aligned HbA1c values using regression formulas derived from duplicate analyses (DCCT HbA1c = [0.83 * EDC HbA1] + 0.14). Blood pressure was measured by a random-zero sphygmomanometer according to a standardised protocol (15) after a 5-minute rest period. Hypertension was defined a blood pressure ≥140/90 mm Hg or the use of anti-hypertensive medications.
Complications were assessed as previously described (13,14) and overt nephropathy (ON) defined as albumin excretion rate >200 μg/min in 2 of 3 timed urine samples (16). Estimated glomerular filtration rate was based on the Cockcroft-Gault formula (17).
Although we have previously shown differences in CAD risk factors by sex, even though there was no difference in the risk of CAD (18), formal tests of interaction did not reveal a significant difference between sex and HbA1c or most other significant univariate predictors of the 18-year incidence of fatal or nonfatal CAD. Nevertheless, as sex interactions with albumin excretion rate (p<0.05) and diastolic blood pressure (p=0.07) were observed for nonfatal CAD, we have chosen to present results overall and gender-specifically. Additionally, approximately 10% of our population (n=66) was under the age of 18 years at baseline. As adolescence is a time of variable change in anthropometric measurements, i.e. waist circumference and BMI, as well as hormonal status resulting in increased insulin requirements due to insulin resistance, and as this age group is likely to have the longest event free follow-up time, the pediatric population was excluded from analyses.
Participants were followed were followed from 1986 to 1988, when the baseline examinations were conducted until January 31, 2007 when mortality was censored. Continuous data were compared across groups by the Student’s t test. Categorical data were compared across groups by Pearson’s χ2 test. Kaplan-Meier analyses and univariate Cox proportional hazards models were used to explore the association of HbA1c and daily insulin dose/kg of body weight with the fatal and nonfatal CAD. Mutivariable Cox proportional hazards modeling with stepwise selection was used to determine the independent predictors of each of the two outcomes: nonfatal and fatal CAD. Covariates that were univariately significant or of biological plausibility were allowed to enter the model. Given the strong correlation between age and duration (r=0.84) only duration was used in multivariable analyses. Both systolic blood pressure (SBP) and diastolic blood pressure (DBP) were univariately significant predictors of each outcome; however, in separately fit multivariable models, only DBP was selected as an independent variable. Therefore, since both variables were highly correlated (r=0.70), in multivariable analyses, only DBP was allowed in the models. Models allowed for sex, diabetes duration, insulin dose/kg body weight, HbA1c, waist circumference, (log) AER, (log) eGFR, DBP, use of hypertension medication, white blood cell count, HDLc, non-HDLc, (log) fibrinogen, and a history of smoking. Cox modeling using updated mean variables (the average value of a variable during follow-up) was also conducted. Analyses examining Beck Depression Inventory and physical activity were limited to smaller populations (n=525 and 550, respectively). Albumin excretion rate, estimated glomerular filtration rate, fibrinogen, the Beck Depression Inventory score, and physical activity were natural logarithmically transformed before analyses. The criterion for statistical significance was P < 0.05. Statistical analysis was conducted using SAS version 9.1 (Cary, North Carolina).
Results
During 18 years of follow-up (mean=15.0 yrs), 127 of the 572 participants remaining after exclusions (i.e. 13 with prevalent CAD, 7 individuals with incomplete data, and the 66 aged less than 18 years), had an incident CAD event; 109 of those were nonfatal (22% of men and 16% of women). There were also 128 deaths, 48 due to CAD. For both endpoints (Tables 1 and 2), cases were significantly older and of longer diabetes duration, had poorer kidney function (estimated glomerular filtration rate) and greater kidney damage (albumin excretion rate) and higher white blood cell count. Cases were also more likely to have hypertension, a worse lipid profile, a history of smoking, marginally lower physical activity levels, a higher Beck Depression Inventory score, or other co-morbid conditions. Body mass index was not significantly different between cases and non-cases for either nonfatal or fatal CAD. Waist circumference, however, was significantly higher in those subsequently experiencing either a nonfatal or fatal CAD event, but statistical significance disappeared after stratification by sex. A history of severe hypoglycaemia was more common in those who experienced a nonfatal CAD event, while this difference did not exist for fatal CAD. Mean HbA1c did not differ between nonfatal CAD cases and non cases (8.9 v 8.7%, p=0.25 respectively); however, it was predictive of CAD mortality (9.2 v 8.7% p=0.03). The converse was true for daily insulin dose/kg of body weight. Mean daily insulin dose/kg of body weight at baseline was significantly lower in those who progressed to nonfatal CAD (0.71 vs 0.78, p=0.009), but not in those who died of CAD (0.77 vs 0.76, p=0.91). There was no difference by sex.
Table 1.
Baseline Predictors of Nonfatal Coronary Artery Disease (CAD), mean (SD) or % (n)
| Nonfatal CAD (n=109) |
Non-cases (n=463) |
p-value | |
|---|---|---|---|
| Sex (female) | 42.2 (46) | 51.4 (238) | 0.09 |
| Age (years) | 33.1 (6.3) | 28.0 (6.6) | <0.0001 |
| Duration (years) | 24.8 (6.4) | 19.2 (6.9) | <0.0001 |
| BMI (kg/m2) | 24.0 (3.3) | 23.8 (3.0) | 0.50 |
| Waist (cm) | 82.1 (8.7) | 79.7 (8.9) | 0.01 |
| males | 85.4 (6.7) | 83.5 (7.9) | 0.08 |
| females | 77.5 (9.2) | 76.1 (8.2) | 0.30 |
| HbA1c (%) | 8.9 (1.6) | 8.7 (1.5) | 0.25 |
| %High ( > mean (8.7%)) | 17.6 (56) | 82.5 (263) | 0.39 |
| %Low (≤ mean) | 20.4 (51) | 79.6 (199) | |
| Daily insulin dose/kg body wt | 0.71 (0.21) | 0.78 (0.24) | 0.009 |
| %High ( > mean (0.77/kg/day)) | 19.4 (31) | 87.7 (219) | 0.0002 |
| %Low ( ≤ mean) | 25.1 (74) | 74.9 (221) | |
| AER* (μg/min) | 177.5 (13.3-935.6) | 14.8 (7.6-100.6) | <0.0001 |
| eGFR* (mg/min/1.73 m2) | 98.8 (65.1-120.6) | 110.4 (89.4-135.3) | <0.0001 |
| WBC (x 103/mm2) | 7.4 (2.1) | 6.5 (1.9) | <0.0001 |
| SBP (mm Hg) | 123.3 (21.8) | 112.9 (14) | <0.0001 |
| DBP (mm Hg) | 79.1 (12.4) | 72.3 (10.2) | <0.0001 |
| Hypertension | 39.5 (43) | 12.1 (56) | <0.0001 |
| Hypertension medication | 17.9 (19) | 3.0 (13) | <0.0001 |
| HDLc (mg/dl) | 50.6 (11.4) | 54.5 (12.9) | 0.005 |
| males | 46.3 (8.2) | 49.4 (10.0) | 0.03 |
| females | 56.4 (12.6) | 59.3 (13.4) | 0.18 |
| Non-HDLc (mg/dl) | 161.3 (42.0) | 133.9 (40.7) | <0.0001 |
| males | 167.9 (44.3) | 136.4 (41.2) | <0.0001 |
| females | 152.5 (37.3) | 131.5 (40.2) | 0.001 |
| Fibrinogen (mg/dl)* | 323.1 (99.8) | 281.3 (85.8) | <0.0001 |
| Beck Depression Inventory* (n=506) | 7 (3-14) | 5 (2-10) | 0.02 |
| History of smoking** | 54.2 (58) | 38.2 (172) | 0.003 |
| Physical activity* (kcal) (n=532) | 924 (420-2296) | 1512 (616.5-2912.0) | 0.09 |
| Overt Nephropathy (n=525) | 50.9 (55) | 24.9 (104) | <0.0001 |
| Proliferative Retinopathy | 65.7 (71) | 25.5 (118) | <0.0001 |
|
Symptomatic Autonomic
Neuropathy (n=459) |
19.8 (19) | 6.6 (24) | <0.0001 |
|
History of hypoglycaemia resulting
in unconsciousness (%) (n=544) |
58.1 (61) | 41.9 (44) | 0.004 |
AER=albumin excretion rate eGFR=estimated glomerular filtration rate WBC=white blood cell count SBP= systolic blood pressure DBP=diastolic blood pressure
Natural logarithmically transformed before analysis
Having smoked at least 100 cigarettes in a lifetime
Table 2.
Baseline Predictors of Mortality from Fatal Coronary Artery Disease, mean (SD) or % (n)
| Fatal CAD (n=48) | Non-cases (n=544) |
p-value | |
|---|---|---|---|
| Sex (female) | 45.8 (22) | 49.5 (269) | 0.63 |
| Age (years) | 33.8 (6.1) | 28.7 (6.7) | <0.0001 |
| Duration (years) | 25.4 (6.3) | 20.0 (7.1) | <0.0001 |
| BMI (kg/m2) | 24.3 (3.6) | 23.8 (3.0) | 0.27 |
| Waist (cm) | 82.9 (9.3) | 80.0 (8.9) | 0.03 |
| males | 86.0 (8.0) | 83.8 (7.7) | 0.18 |
| females | 79.1 (9.5) | 76.1 (8.2) | 0.09 |
| HbA1c (%) | 9.2 (1.6) | 8.7 (1.5) | 0.03 |
| %High ( > mean (8.7%)) | 5.8 (19) | 94.2 (309) | 0.03 |
| %Low (≤ mean) | 10.8 (28) | 89.2 (232) | |
| Daily insulin dose/kg body wt | 0.77 (0.25) | 0.76 (0.24) | 0.91 |
| %High ( > mean (0.76 /kg/day)) | 7.0 (18) | 93.0 (240) | 0.43 |
| %Low ( ≤ mean) | 8.8 (27) | 91.2 (280) | |
| AER* (μg/min) | 509.1 (71.6-1006.9) | 16.6 (7.9-153.6) | <0.0001 |
| eGFR* (mg/min/1.73 m2) | 90.4 (56.9-116.1) | 109.3 (87.0-134.2) | 0.002 |
| WBC (x 103/mm2) | 8.0 (2.1) | 6.6 (1.9) | <0.0001 |
| SBP (mm Hg) | 124.0 (21.9) | 114.1 (15.4) | 0.004 |
| DBP (mm Hg) | 78.3 (13.5) | 73.3 (10.7) | 0.01 |
| Hypertension (%) | 37.5 (18) | 16.2 (88) | 0.0002 |
| Hypertension medication (%) | 12.8 (6) | 5.4 (28) | 0.04 |
| HDLc (mg/dL) | 50.8 (13.1) | 53.9 (12.6) | 0.11 |
| males | 46.8 (9.9) | 48.8 (9.7) | 0.32 |
| females | 55.2 (14.9) | 59.1 (13.1) | 0.20 |
| Non-HDLc (mg/dl) | 166.1 (52.4) | 137.9 (42.0) | 0.0008 |
| males | 169.5 (60.8) | 142.1 (42.5) | 0.04 |
| females | 162.4 (42.4) | 133.6 (41.2) | 0.002 |
| Fibrinogen (mg/dl)* | 328.2 (84.7) | 287.5 (92.1) | 0.001 |
| Beck Depression Inventory* (n=525) | 8 (5-14) | 6 (2-10) | 0.04 |
| History of smoking** | 61.7 (29) | 40.2 (213) | 0.004 |
| Physical activity* (kcal) (n=550) | 1201 (574-2016) | 1428 (616-2816) | 0.07 |
| Overt Nephropathy (%) (n=541) | 67.4 (31) | 28.1 (139) | <0.0001 |
| Proliferative Retinopathy (%) | 72.9 (35) | 31.0 (168) | <0.0001 |
|
Symptomatic Autonomic
Neuropathy (%) (n=468) |
19.4 (7) | 9.1 (39) | 0.04 |
|
History of hypoglycaemia resulting
in unconsciousness (%) (n=563) |
44.4 (20) | 46.0 (238) | 0.85 |
AER=albumin excretion rate eGFR=estimated glomerular filtration rate WBC=white blood cell count SBP= systolic blood pressure DBP=diastolic blood pressure
Natural logarithmically transformed before analysis
Having smoked at least 100 cigarettes in a lifetime
Figure 1 presents the Kaplan Meier hazard curves for nonfatal and fatal CAD by mean split of HbA1c and daily insulin dose/kg of body weight. The unadjusted relative risk, in a time to event Cox model, per 1% increase in HbA1c was 1.15 (1.01-1.30), p=0.04 for non fatal CAD and 1.32 (1.09-1.59), p=0.004 for fatal CAD. The unadjusted per unit daily insulin dose/kg of body weight this was 0.27 (0.10-0.71, p=0.008) for nonfatal CAD and 0.92 (0.23-3.69, p=0.91) for fatal CAD. There was no relationship between Hba1c and insulin dose/kg of body weight/per day (r= −0.005, p=0.90 for non-fatal CAD vs non cases; r= −0.02, p=0.71 for fatal CAD vs non cases).
Figure 1.
Nonfatal and fatal coronary artery disease by HbA1c and Insulin Dose
In multivariate analyses (Table 3), diabetes duration, insulin dose/kg of body weight, estimated glomerular filtration rate (renal function), diastolic blood pressure, HDLc, non-HDLc, and white blood cell count were significant predictors of the 18-year incidence of nonfatal CAD events. HbA1c was not. Sex-specific analyses revealed that in men diabetes duration, non-HDLc, and renal function (estimated glomerular filtration rate), but not damage (albumin excretion rate), were predictive. In females, diabetes duration and non-HDLc were also predictive, but kidney function was not. White blood cell count and diastolic blood pressure were also independent predictors in women.
Table 3.
Predictors of the 18 Year Incidence of Nonfatal or Fatal Coronary Artery Disease in Type 1 Diabetes
| Nonfatal CAD Events 109 cases; 463 non-cases |
CAD as the Primary Cause of Death 48 cases; 544 non-cases |
|||
|---|---|---|---|---|
| Risk Factors |
Unadjusted
HR (95% CI) |
Adjusted
HR (95% CI) |
Unadjusted
HR (95% CI) |
Adjusted
HR (95% CI) |
| Diabetes duration (years) |
1.12 (1.08-1.15) | 2.07 (1.64-2.61) | 1.11 (1.06-1.16) | 1.95 (1.40-2.71) |
| HbA1c (%) | 1.15 (1.01-1.30) | NS | 1.32 (1.09-1.59) | 1.55 (1.14-2.10) |
| Daily insulin dose/kg body weight |
0.34 (0.13-0.89) | 0.75 (0.57-0.97) | 0.92 (0.23-3.67) | |
| AER (μg/min)* | 1.37 (1.25-1.49) | NS | 1.51 (1.32-1.74) | 1.89 (1.38-2.57) |
| eGFR (mg/min/1.73 m2)* |
0.29 (0.21-0.39) | 0.73 (0.60-0.88) | 0.30 (0.19-0.48) | |
| DBP (mm Hg) | 1.06 (1.04-1.08) | 1.52 (1.25-1.85) | 1.05 (1.03-1.08) | |
| HDLc (mg/dl) | 0.97 (0.95-0.99) | 0.74 (0.59-0.92) | 0.98 (0.95-1.00) | |
| Non-HDCc (mg/dl) | 1.01 (1.01-1.02) | 1.44 (1.20-1.73) | 1.01 (1.01-1.02) | |
| WBC (x 103/mm2) | 1.27 (1.17-1.38) | 1.42 (1.20-1.68) | 1.35 (1.21-1.52) | 1.48 1.16-1.88) |
Natural logarithmically transformed before analyses
NS=not selected AER=albumin excretion rate eGFR=estimated glomerular filtration rate DBP=diastolic blood pressure SBP=systolic blood pressure HDLc=high density lipoprotein cholesterol Non-HDLc=non-high density lipoprotein cholesterol WBC=white blood cell count
Models allowed for sex, diabetes duration, insulin dose/kg body weight, HbA1c, waist circumference, (log) AER, (log) eGFR, DBP, use of hypertension medication, WBC, HDLc, non-HDLc, (log) fibrinogen, and a history of smoking.
There were forty-eight fatal CAD events, twenty-five (52%) with a prior history of CAD. Significant independent predictors of CAD mortality (Table 3), like nonfatal CAD, included diabetes duration, and white blood cell count. However, HbA1c and renal damage (albumin excretion rate) were now also predictive, while daily insulin dose/kg of body weight, blood pressure, lipids, and renal function were not. Sex-specific analyses revealed that diabetes duration, HbA1c, estimated glomerular filtration rate (renal function), and WBC were significant predictors in men. For women, the significant predictors were diabetes duration, albumin excretion rate (renal damage), and smoking. When allowed for, neither physical activity nor Beck Depression Inventory score were selected for either outcome either overall or sex-specifically.
Finally, Cox models were repeated using updated means rather than baseline variables and confirmed the above associations, except that HbA1c also was a predictor of nonfatal events (HR=1.18, 1.01-1.38) along with insulin dose (inversely), albumin excretion rate, diastolic blood pressure, non-HDLc, and WBC. In the fatal CAD model with updated means, diastolic blood pressure medication was protective, and HbA1c, diabetes duration, and WBC remained significant predictors (data not depicted).
Discussion
These data, though generally consistent with our previous reports using combined fatal and non fatal CAD events, suggest that predictors of these two components of CAD differ significantly. In particular a lower daily insulin dose, but not HbA1c, independently predicts nonfatal CAD, while HbA1c, but not insulin dose, predicts fatal CAD. We also confirm that traditional risk factors continue to predict CAD over the long term (18 years) and confirm previous reports that though absolute CAD risk does not vary by gender, CAD risk factors do (18).
HbA1c has previously been shown to predict CAD in type 1 diabetes by some, (8,19,20) but not all (6,7,9), prior studies as recently reviewed (21). Since that review, Shankar et al (8) have reported an update from the WESDR study showing that elevated glycated hemoglobin remains predictive of both CVD mortality, and interestingly, all cause mortality, a finding also seen in the general population (22). The current results thus may help, in part, resolve the conflicts between the epidemiologic studies (6-9, 19) as the EDC study now shows an association with CAD mortality, consistent with WESDR (8,20), but not for nonfatal CAD, also consistent with WESDR (9).
The reason why HbA1c may be more closely related to CAD mortality than to nonfatal CAD is likely to be multifactorial. First, this may be mathematical as all the risk factors examined are closely related and in the time to event analyses, HbA1c was univariately associated with both outcomes. It is therefore reasonable to conclude there is a stronger association of HbA1c with mortality than with morbidity. However, the multivariable models were stable and multicollinearity was not strong. Secondly, as HbA1c is a strong predictor of microvascular complications (23,24), this will increase the risk of significant co-morbidities (e.g. renal disease), which may increase case fatality. This is, however, unlikely the full explanation as such complications, particularly renal disease, also increase the risk of nonfatal CAD (Table 1).
A third likely explanation maybe that inadequate glycaemic control at the time of an acute coronary syndrome (ACS) may jeopardise the injured myocardium and thus lead to greater mortality (case fatality). The first DIGAMI study (25) and other studies (26,27) that show better outcomes in patients with ACS who are in better glycaemic control would be consistent with this argument, although in the current study we do not have glucose values available from the time of event.
A failure of baseline HbA1c to predict CAD overall (i.e. combined fatal and non fatal CAD) has been a consistent finding in our population and has led to the proposal that part of the explanation maybe that hyperglycaemia, which leads to AGE formation and protein cross-linking, may yield a more “stable” type of atherosclerosis, which would be less likely to rupture and cause an ACS (21). This would help explain a weak relationship with CAD events despite their being increased atherosclerosis in T1D. The extensive evidence for this ‘glucose stabilisation’theory has also recently been reviewed (21). The current analyses extend these observations further and suggest that A1c (glycemic control) may influence the natural course of CAD events in addition to their occurance.
Another important finding in the current analysis is the finding that insulin dose, is closely, and inversely, related to nonfatal CAD (but shows no association with mortality). This observation is also consistent with the hypothesis that adequate insulin dosage is critical to avoid accelerated atherosclerosis and further suggests that possibly some of the contribution of better ‘glycaemic’ levels, in those studies with a strong association, may result from better insulin dosage and regulation in addition to lower glycaemic exposure alone. Unfortunately, with the exception of Rossing et al (5), insulin dose has not been reported in these epidemiologic studies, and even in Rossing’s study, was not used in multivariable models.
Daily insulin dose,/kg of body weight may therefore be more closely related to nonfatal CAD than HbA1c as it may be a better measure of overall insulin mediated glucose homeostasis and metabolic regulation than HbA1c which measures average blood glucose, capturing both hypoglycaemia and hyperglycaemia, over approximately a three month time span.
It is possible that nonfatal CAD cases had greater glucose variability despite having similar average glucose compared to non-cases. Interestingly, in the current analyses severe hypoglycaemia (i.e. resulting in unconsciousness) was predictive of nonfatal CAD (HR=1.8, p=0.004), but not fatal CAD (HR=0.94, p=0.85) until adjusted for diabetes duration, suggesting greater excursions to the lower range at least. Glucose instability, or oscillations, have been shown to be more highly associated with oxidative stress and angiogenesis than sustained hyperglycaemia (28).
The dramatic findings of the DCCT/EDIC study (4), showing a 57% reduction of nonfatal myocardial infarction, stroke and CVD death in the former intensively treated group would initially appear to be at variance with our results, for the positive findings are largely based on non fatal events for which the current study suggests a weaker glycaemic association. There are insufficient fatal cases (3 in the intensive and 4 in the conventional groups) in DCCT/EDIC to permit a similar analysis to that presented here. However, at this time it is unclear how this intensive therapy effect is mediated, as full risk factor modeling has not been reported and thus it is possible that some of the benefit of intensive management is related to nonglycaemic effects (e.g. multiple insulin dose regimen, per se), as well as to level of glycaemic control.
We have also demonstrated that the traditional cardiovascular disease risk factors are, in general, predictive of CAD in type 1 diabetes, despite the differences discussed above. Dyslipidemia and blood pressure, predictive of cardiovascular disease in the general population, were also predictive of CAD in our type 1 diabetes population. As Shankar et al (8) point out, proteinuria and dyslipidemia are consequences of hyperglycaemia and therefore may be intermediates in the causal pathway between hyperglycaemia and mortality. Interestingly, age-adjusted HbA1c was predictive of nonfatal CAD events in our population until the association was adjusted for non-HDL cholesterol, which would be consistent with low insulin dosage being the underlying derangement resulting in both HDL non-HDL cholesterol and high blood glucose. Change over time may also be critical as evidenced by our time dependent, i.e. updated mean, modeling, where HbA1c was also predictive of morbidity along with lower insulin dose (although, like at baseline, insulin was not predictive of mortality).
Consistent with previous findings in type 1 diabetes (4,18), neither fatal nor non fatal CAD risk differed by gender. We did, however, find gender differences in risk factors, consistent with previous findings in this population (18). These sex differences are important to confirm in other populations as they may significantly affect the design of optimal intervention strategies.
A number of the limitation of the study is that participants have not been followed from diagnosis, which may have been as early as 1950. There is therefore a considerable early mortality/survival effect. Further analyses will address this issue when event numbers permit.
In conclusion, the current data further confirm the complexity of the associations of glycaemic control and diabetes management with cardiovascular disease.
The observations that HbA1c more strongly predicts CAD mortality than morbidity and insulin dose relates more strongly to CAD morbidity than mortality provides important insights into our understanding of the aetiology of CAD in type 1 diabetes. Further evaluation of these observations may lead to important clinical implications. The closer relationship of HbA1c with mortality may also reflect the benefit of good glyaecemic control at the time of myocardial insult thus enabling more efficient and less toxic myocardial metabolism. The converse observation that a low insulin dose predicts non fatal events should also help allay fears that insulin itself is strongly atherogenic and encourage clinicians to use adequate insulin dosage to achieve good control.
Our findings underscore the importance of adequate insulin dosage as well as the management of all standard risk factors for CAD, including blood pressure and lipids, in type 1 diabetes.
Acknowledgements
This work was funded by NIH grand DK34818. We would also like to thank John Fuller for his help in classifying the mortality cases. Finally we are indebted to the EDC study participants for their dedication and cooperation in the advancement of knowledge in the scientific community.
Footnotes
Conflict of interest All authors declare that the answer to the questions on your competing interest form are all No and therefore have nothing to declare.
Contributor Information
Baqiyyah Conway, 3512 Fifth Ave Pittsburgh, PA 15217 The USA Graduate Student Research Assistant.
Tina Costacou, 3512 Fifth Ave Pittsburgh, PA 15217 The USA Visiting Assistant Professor.
Trevor Orchard, 3512 Fifth Ave Pittsburgh, PA 15217 The USA Professor of Epidemiology.
References
- 1.Stout R. Insulin and atheroma. 20-yr perspective. Diabetes Care. 1990;13(6):631–54. doi: 10.2337/diacare.13.6.631. [DOI] [PubMed] [Google Scholar]
- 2.Kraemer F. Insulin deficiency alters cellular cholesterol metabolism in murine macrophages. Diabetes. 1986;35(7):764–770. doi: 10.2337/diab.35.7.764. [DOI] [PubMed] [Google Scholar]
- 3.Krolewski A, Kosinski E, Warram J, Leland O, Busick E, Asmal A, Rand L, Christlieb A, Bradley R, Kahn C. Magnitude and determinants of coronary artery disease in juvenile-onset, insulin dependent diabetes mellitus. Am J Cardiol. 1987;59:750–755. doi: 10.1016/0002-9149(87)91086-1. [DOI] [PubMed] [Google Scholar]
- 4.Nathan D, Cleary P, Backlund J, Genuth S, Lachin J, Orchard T, Raskin P, Zinman B, the Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) Study Research Group Intensive diabetes treatment and cardiovascular disease in patients with type 1 diabetes. N Engl J Med. 2005;353:2643–2653. doi: 10.1056/NEJMoa052187. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Rossing P, Hougaard P, Borch-Johnsen K, Parving H. Predictors of mortality in insulin dependent diabetes: 10 year observational follow up study. BMJ. 1996;313(7060):779–84. doi: 10.1136/bmj.313.7060.779. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Orchard T, Olson J, Erby J, Williams K, Forrest K, Kinder L, Ellis D, Becker D. Insulin resistance-related factors, but not glycemia, predict coronary artery disease in type 1 diabetes. Diabetes Care. 2003;26:1374–1379. doi: 10.2337/diacare.26.5.1374. [DOI] [PubMed] [Google Scholar]
- 7.Soedamah-Muthu S, Chaturvedi N, Toeller M, Ferriss B, Reboldi P, Michel G, Manes C, Fuller J, the EURODIAB Prospective Complications Study Group Risk factors for coronary artery disease in type 1 diabetic patients in Europe: the EURODIAB Prospective Complications Study. Diabetes Care. 2004;27:530–537. doi: 10.2337/diacare.27.2.530. [DOI] [PubMed] [Google Scholar]
- 8.Shankar A, Klein R, Klein B, Moss S. Association between glycosylated hemoglobin level and cardiovascular and all-cause mortality in type 1 diabetes. Am J of Epidemiol. 2007;166(4):393–402. doi: 10.1093/aje/kwm096. [DOI] [PubMed] [Google Scholar]
- 9.Klein B, Klein R, McBride P, Cruickshanks K, Palta M, Knudtson M, Moss S, Reinke J. Cardiovascular disease, mortality, and retinal microvascular characteristics in type 1 diabetes: Wisconsin Epidemiologic Study of Diabetic Retinopathy. Arch Intern Med. 2004;164:1917–1924. doi: 10.1001/archinte.164.17.1917. [DOI] [PubMed] [Google Scholar]
- 10.Prineas R, Crow R, Blackburn H. The Minnesota Code Manual of Electrocardiographic Findings. John Wright-PSG, Inc.; Littleton, MA: 1982. [Google Scholar]
- 11.Diabetes Epidemiology Research International Mortality Study Group International Evaluation of Cause-Specific Mortality and IDDM. Diabetes Care. 1991;14(1):55–60. doi: 10.2337/diacare.14.1.55. [DOI] [PubMed] [Google Scholar]
- 12.Beck A, Garbin M. Psychometric properties of the Beck depression inventory: 25 years of evaluation. Clin Psychol Rev. 1988;8:77–100. [Google Scholar]
- 13.Orchard TJ, Dorman JS, Maser RE, Becker DJ, Ellis D, LaPorte RE, Kuller LH, Wolfson S, Drash AL. Factors associated with the avoidance of severe complications after 25 years of insulin-dependent diabetes mellitus: Pittsburgh Epidemiology of Diabetes Complications Study-I. Diabetes Care. 1990;13(7):741–7. doi: 10.2337/diacare.13.7.741. [DOI] [PubMed] [Google Scholar]
- 14.Orchard TJ, Dorman JS, Maser RE, Becker DJ, Drash AL, Ellis D, LaPorte RE, Kuller LH. The prevalence of complications in insulin-dependent diabetes mellitus by sex and duration: Pittsburgh Epidemiology of Diabetes Complications Study - II. Diabetes. 1990;39:1116–1124. doi: 10.2337/diab.39.9.1116. [DOI] [PubMed] [Google Scholar]
- 15.Borhani N, Kass E, Langford H, Payne G, Remington R, Stamler J. The Hypertension Detection and Follow-up Program. Prev Med. 1976;5:207–215. [Google Scholar]
- 16.Ellis D, Buffone G. A new approach to the evaluation of proteinuric states. Clin Chem. 1977;23:666–670. [PubMed] [Google Scholar]
- 17.Cockcroft D, Gault M. Prediction of creatinine clearance from serum creatinine. Nephron. 1976;16:31–41. doi: 10.1159/000180580. [DOI] [PubMed] [Google Scholar]
- 18.Lloyd C, Kuller L, Ellis D, Becker D, Wing R, Orchard T. Coronary artery disease in IDDM: gender differences in risk factors but not risk. Arterioscler Thromb Vasc Biol. 1996;16:720–726. doi: 10.1161/01.atv.16.6.720. [DOI] [PubMed] [Google Scholar]
- 19.Lehto S, Ronnemaa T, Pyorala K, Laakso M. Poor glycemic control predicts coronary heart disease events in patients with type 1 diabetes without nephropathy. Arterioscler Thromb Vasc Biol. 1999;19:1014–1019. doi: 10.1161/01.atv.19.4.1014. [DOI] [PubMed] [Google Scholar]
- 20.Moss S, Klein R, Klein B. Cause-specific mortality in a population-based study of diabetes. Am J Public Health. 1991;81:1158–1162. doi: 10.2105/ajph.81.9.1158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Orchard T, Costacou T, Kretowski A, Nesto R. Type 1 diabetes and coronary artery disease. Diabetes Care. 2006;29(11):2528–2538. doi: 10.2337/dc06-1161. [DOI] [PubMed] [Google Scholar]
- 22.Khaw K, Wareham N, Bingham S, Luben R, Welch A, Day N. Association of hemoglobin A1c with cardiovascular disease and mortality in adults: the European prospective investigation into cancer in Norfolk. Ann Intern Med. 2004;141(6):413–420. doi: 10.7326/0003-4819-141-6-200409210-00006. [DOI] [PubMed] [Google Scholar]
- 23.The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. The Diabetes Control and Complications Trial Research Group. N Engl J Med. 1993;329:977–986. doi: 10.1056/NEJM199309303291401. [DOI] [PubMed] [Google Scholar]
- 24.N Engl J Med. 2000;342:381–389. [Google Scholar]
- 25.Malmberg K, Rydén L, Efendic S, Herlitz J, Nicol P, Waldenström A, Wedel H, Welin L. Randomized trial of insulin-glucose infusion followed by subcutaneous insulin treatment in diabetic patients with acute myocardial infarction (DIGAMI study): effects on mortality at 1 year. J Am Coll Cardiol. 1995;26(1):57–65. doi: 10.1016/0735-1097(95)00126-k. [DOI] [PubMed] [Google Scholar]
- 26.Bhadriraju S, Ray KK, DeFranco AC, Barber K, Bhadriraju P, Murphy SA, Morrow DA, McCabe CH, Gibson CM, Cannon CP, Braunwald E. Association between blood glucose and long-term mortality in patients with acute coronary syndromes in the OPUS-TIMI 16 trial. Am J Cardiol. 2006;97(11):1573–7. doi: 10.1016/j.amjcard.2005.12.046. [DOI] [PubMed] [Google Scholar]
- 27.Foo K, Cooper J, Knight C, Suliman A, Ranjadayalan K, Timmis A. A single serum glucose measurement predicts adverse outcomes across the whole range of acute coronary syndromes. Heart. 2003;89:512–516. doi: 10.1136/heart.89.5.512. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Monnier L, Mas E, Ginet C, Michel F, Villon L, Cristol J, Colette C. Activation of Oxidative Stess by Glucose Fluctuations Compared with Sustained Chronic Hyperglycemia in Patients with Type 2 Diabetes. JAMA. 2006;495(4):1681–1687. doi: 10.1001/jama.295.14.1681. [DOI] [PubMed] [Google Scholar]