Author + information
- Received October 18, 2005
- Revision received February 1, 2006
- Accepted February 14, 2006
- Published online June 20, 2006.
- Allison B. Goldfine, MD⁎,†,⁎ (, )
- Joshua A. Beckman, MD†,
- Rebecca A. Betensky, PhD‡,
- Heather Devlin⁎,
- Shauna Hurley†,
- Nerea Varo, PhD†,2,
- Uwe Schonbeck, PhD†,3,
- Mary Elizabeth Patti, MD⁎ and
- Mark A. Creager, MD†,1
- ↵⁎Reprint requests and correspondence:
Dr. Allison B. Goldfine, Joslin Diabetes Center, One Joslin Place, Boston, Massachusetts 02215.
Objectives We evaluated whether endothelial dysfunction was present in nondiabetic persons with a family history (FH) of diabetes and assessed its relationship with insulin resistance and atherosclerosis risk factors.
Background Atherosclerosis is frequently present when type 2 diabetes (T2D) is first diagnosed. Endothelial dysfunction contributes to atherogenesis.
Methods Oral glucose tolerance and brachial artery flow-mediated, endothelium-dependent vasodilation (EDV) were assessed in 38 nondiabetic subjects; offspring of two parents with T2D (FH+) or with no first-degree relative with diabetes (FH−).
Results Although fasting glucose was higher in FH+ than FH− (5.3 ± 0.1 mmol/l vs. 4.9 ± 0.1 mmol/l, p < 0.03), glycemic burden assessed as 2-h or area-under-the-curve glucose after glucose load or glycosylated hemoglobin (HbA1c), and measures of insulin sensitivity or inflammation did not differ. Brachial artery flow-mediated EDV was reduced in FH+ (7.1 ± 0.9% vs. 11.7 ± 1.6%, p < 0.02), with no difference in nitroglycerin-induced endothelium-independent vasodilatation. In the combined cohort, only FH+ (r2= 0.12, p < 0.02) and HbA1c (r2= 0.14, p < 0.02) correlated with EDV. Insulin resistance, assessed by tertile of homeostasis model assessment of insulin resistance (HOMA-IR), was associated with impaired endothelium-dependent vasodilatation in FH− (p < 0.03, analysis of variance), but not in FH+, as even the most insulin-sensitive FH+ offspring had diminished endothelial function. In multiple regression analysis, including established cardiac risk factors, blood pressure and lipids, HbA1c, and HOMA-IR, FH remained a significant determinant of EDV (p = 0.04).
Conclusions Bioavailability of nitric oxide is lower in persons with a strong FH of T2D. Glycemic burden, even in the nondiabetic range, can contribute to endothelial dysfunction. Abnormalities of endothelial function may contribute to atherosclerosis before development of overt diabetes.
Cardiovascular disease is the leading cause of morbidity and mortality for patients with diabetes. Atherosclerosis is frequently present on diagnosis of type 2 diabetes (T2D), suggesting atherosclerotic processes begin before the onset of overt diabetes. Insulin resistance is highly associated with T2D, hypertension, dyslipidemia, obesity, and cardiovascular disease. Insulin resistance precedes and predicts both incident diabetes and cardiovascular disease. However, recent studies suggest insulin resistance per se may impart greater risk for development of diabetes in offspring of two parents with T2D than in persons with no family history (FH) of disease (1). Whether insulin resistance imparts increased cardiovascular risk in persons with a strong FH of diabetes remains unknown.
The endothelium participates in atherosclerotic pathogenesis. Attenuated function is considered an early marker of vascular disease. Although some studies demonstrated endothelial dysfunction in persons with FH of diabetes (2,3), this remains controversial (4). We hypothesized that vascular function among offspring of parents with diabetes is abnormal compared with persons with no FH of diabetes or coronary artery disease, even when traditional cardiac risk factors are similar. We characterized inter-relationships among insulin resistance, cardiac risk factors, and endothelial function according to FH of diabetes.
The study was approved by the institutional review board. Thirty-eight nondiabetic healthy subjects, 19 with two T2D parents (FH+) and 19 with no first-degree relative with diabetes or coronary artery disease (FH−), provided written informed consent. Family history was defined during medical interview by participant report of either diabetes in both biological parents (FH+), or in neither biological parent or any first-degree relative (FH −). The FH+ cohort has been previously described (5). The FH− cohort were recruited to be similar for age, gender, and body mass index. All were normotensive and non-smokers.
Fasting lipids and oral glucose tolerance, using a 100-g glucose load to maximize glucose and insulin excursion, were evaluated. Subjects were deemed nondiabetic using National Diabetes Data Group criteria (6,7). Areas under the curve for glucose and insulin were calculated by triangulation, and insulin resistance was determined by homeostasis model assessment (HOMA-IR).
Vascular function studies
Endothelial function was determined using high-resolution ultrasonography (Toshiba Powervision 8000, 7.5 MHz linear-array probe, Toshiba America Medical Systems, Inc., Tustin, California) triggered by electrocardiogram “R” wave with Data Translation frame-grabber videocard (Dataviz, Trumbull, Connecticut) as previously described (5). A sphygmomanometric cuff placed above the antecubital fossa was inflated to suprasystolic pressure for 5 min. Flow-induced, endothelial-dependent vasodilation (EDV) was determined 1 min after cuff deflation. Endothelium-independent vasodilation (EIV) was determined 3 min after nitroglycerin, 0.4 mg sublingually, which was withheld for systolic blood pressure <100 mm Hg (four subjects in each group). Arterial diameter was measured using edge detection software (Brachial Tools 4.2.2, Medical Imaging Applications LLC, Iowa City, Iowa).
Glucose, lipids, and glycohemoglobin were measured in the Joslin clinical laboratory. Immunoassays were performed in duplicate and included serum insulin (Diagnostic Systems Laboratories, Webster, Texas), high-sensitivity C-reactive protein (Alpha Diagnostic International, San Antonio, Texas), interleukin-6 (R&D Systems, Inc., Minneapolis, Minnesota), FFA (Wako Chemicals Inc., Richmond, Virginia), plasma plasminogen activator inhibitor-1 (American Diagnostica, Greenwich, Connecticut), soluble intercellular adhesion molecule-1 (R&D Systems), soluble CD40 ligand (BenderMedSystems, Vienna, Austria), and adiponectin (Linco Research, Inc., St. Charles, Missouri).
The unpaired ttest was used for comparison of FH+ and FH−. Pearson’s correlation, analysis of variance (ANOVA), and simple, stepwise, and multiple regression analyses were performed using StatView (SAS Institute Inc., Cary, North Carolina). Independent variables were assessed directly and after logarithmic transformation for skewed distribution. Because no data became uniquely significant after logarithmic transformation, all data are presented as natural variables. Results are considered significant with two-tailed p values <0.05.
No subject had diabetes. Glycosylated hemoglobin (HbA1c) was within normative range (4% to 6%) in all participants. Ten subjects in each group had 2-h glucose concentrations between 7.8 to 11.1 mmol/l, suggesting some glucose intolerance. Demographic and metabolic characteristics are summarized (Table 1).Although fasting glucose was higher in FH+ than FH− (5.3 ± 0.1 mmol/l vs. 4.9 ± 0.1 mmol/l, respectively, p < 0.03), no significant differences existed in all other measures of glycemia including 2-h glucose, area-under-the-curve glucose, or HbA1c. Plasminogen activator inhibitor-1 tended to be higher in FH+ (p = 0.07), largely due to one FH+ subject. All other metabolic and inflammatory measures did not differ between groups.
EDV and FH of diabetes
Endothelium-dependent vasodilation was 38% lower in FH+ than FH− (7.1 ± 0.9% vs. 11.7 ± 1.6%, p < 0.02). There was no difference between groups in EIV (18.9 ± 1.3% vs. 18.3 ± 1.7%, p = 0.8) (Fig. 1).Baseline diameter of the brachial artery was similar (3.6 ± 0.2 mm vs. 3.5 ± 0.1 mm, p = 0.8, FH+ vs. FH−, respectively).
EDV and insulin resistance
To assess the relationship between HOMA-IR and EDV, the entire cohort was divided into tertiles of insulin resistance (<1.3, 1.3 to 2.6, and >2.6). In two-way ANOVA, FH remained an important determinant of EDV (p = 0.03). There was no difference in EDV in the two most insulin-sensitive subgroups, but the most resistant tertile had significantly lower EDV (Fig. 2).The same boundary levels of HOMA-IR were then applied to FH+ and FH−; thus, groups were similar in magnitude of insulin resistance. Each tertile was represented in FH+ and FH−. The same pattern was seen for insulin resistance and EDV in FH− as in the whole cohort. Specifically, the most insulin-resistant persons had abnormal EDV. However, in FH+, EDV was impaired in all three tertiles of insulin sensitivity, and there was no relationship between EDV and HOMA-IR. Similar relationships were obtained using a general linear model analysis with HOMA-IR and FH as covariates, accounting for multiple comparisons.
Although insulin resistance was demonstrated to interact with endothelial function, HOMA-IR associated more strongly with traditional cardiac risk factors, obesity, cholesterol, and blood pressure than with EDV (Table 2).Using Bonferroni correction for multiple comparisons, correlations between HOMA-IR and measures of obesity, high-density lipoprotein, and triglycerides remained significant.
EDV and glycemic burden
In secondary analysis evaluating glycemia and traditional cardiac risk factors with endothelial function, relationships between EDV, FH, and each metabolic variable were evaluated using simple regression analysis. In the combined cohort, only FH (r2= 0.12, p < 0.02) and HbA1c (r2= 0.14, p < 0.02) correlated inversely with EDV (Fig. 3),suggesting an important interaction between chronic glycemia and EDV, even in the normative range.
Next, we evaluated FH+ and FH− subgroups independently. In FH−, HbA1c (r2= 0.15, p = 0.05), cholesterol (r2= 0.38, p < 0.005), and systolic blood pressure (r2= 0.23, p < 0.04) each correlated inversely with EDV. In FH+, only high-density lipoprotein directly associated with EDV (r2= 0.31, p < 0.02), as previously reported (5). In contrast with FH−, there was no association between EDV and cholesterol or systolic blood pressure in FH+; EDV was attenuated across the normative range of these variables (Fig. 4).Although significant relationships might become evident with larger cohorts, these data suggest the strength of the relationship between EDV and traditional cardiac risk factors may differ in FH+ and FH−. Furthermore, FH+ demonstrate blunted endothelial function across the range of cholesterol and systolic pressure, suggesting protective benefits of low blood pressure or low cholesterol on EDV is attenuated in FH+.
In multiple regression analysis of the entire cohort, incorporating only independent variables significantly correlated in the whole cohort or in either FH+ or FH− subgroups (FH, HbA1c, HOMA-IR, cholesterol, high-density lipoprotein cholesterol, and systolic blood pressure); only FH (p = 0.04) remained a significant determinant of EDV. We fit bivariate models with FH and each major cardiac risk factor as predictors, along with their interaction with FH and found significant interactions between FH and both cholesterol (p = 0.004) and systolic blood pressure (p = 0.03). Additionally, the main effects of the risk factors of cholesterol and blood pressure were significant (p = 0.0009 and p = 0.01, respectively) as was FH status (p < 0.01 in each model), indicating they are indeed independently predictive of EDV even in the presence of FH, and that FH is an independent predictor of EDV. These data suggest adverse effects of FH on EDV are mediated in part, but not solely, through blood pressure and lipids, even when these measures are normative. Due to differences in fasting glucose between groups, the analysis was repeated including fasting glucose in the model. Family history remained a significant determinant of EDV (p < 0.03).
Inflammation and endothelial function
Relationships between EDV and inflammatory cytokine/adipokine measures of inflammatory mediators did not correlate with EDV in the whole cohort, or in either FH+ or FH− subgroups.
Our study demonstrates reduced EDV in nondiabetic individuals with strong FH compared with persons with no FH of diabetes. Differences cannot be explained by confounding variables including age, gender, ethnicity, obesity, lipids, blood pressure, glycemia, or insulin resistance. Although we did not find associations between EDV and age or obesity previously reported in population-based studies (8), ranges of these variables were smaller in our healthy cohort. In multiple regression analysis, only FH remained a significant determinant of EDV. Furthermore, endothelial dysfunction was demonstrated in even the most insulin-sensitive offspring of diabetic parents. Although the pleiotropic metabolic disturbances of the pre-diabetic state may contribute additively or synergistically to atherosclerosis pathogenesis, our findings suggest a strong FH of diabetes is independently associated with diminished EDV and may contribute to cardiovascular risk in advance of overt diabetes. Frequent presence of vascular pathology at the time of diagnosis of diabetes suggests importance for identification of pre-diabetic persons with diminished endothelial function and early atherosclerotic disease, and FH of diabetes is a recognized risk for development of diabetes.
Several previous investigations also found endothelial dysfunction in offspring of diabetic parents (2,3), though this remains controversial (4). In these studies, offspring cohorts differed from control subjects with respect to post-load glucose and insulin, cholesterol, severity of insulin resistance, body mass index, blood pressure, and other factors (2–4). Also, relationships between glucose metabolic clearance rate and endothelial function were not assessed in control subjects (3). These confounding differences make it difficult to assess independent effects of FH of diabetes on endothelial function. Our FH+ were offspring of two T2D parents, contrasting with studies with just one of two parents with diabetes, thereby enriching our group for familial differences. Moreover, our cohorts were similar for multiple confounding variables, yet attenuation of EDV was significant in offspring.
Hyperglycemia and endothelial function
Both hyperglycemia and insulin resistance could contribute to atherosclerosis. In the absence of overt diabetes, multiple studies have found that either impaired fasting or 2-h post-load glucose predicts cardiovascular events (9,10). Fasting hyperglycemia diminishes microvascular hyperemia (11) and inversely correlates with EDV (8), and short-term hyperglycemia attenuates endothelial function in healthy persons (12). Hyperglycemia may adversely affect vascular function through multiple mechanisms including increased flux through the polyol pathway, increased oxidative stress, activation of protein kinase C-beta, and advanced-glycation end products formation (13). We evaluated nondiabetic cohorts with normal-to-mild glucose intolerance enriching our cohorts for risk of diabetes and early atherosclerotic pathology. This is the first study to document an association between normative HbA1c and EDV in nondiabetic persons further implicating an important and adverse effect of very mild elevations in glucose on endothelial function. All subjects had normative fasting glucose, yet differences in glycemic burden assessed by fasting glucose existed between groups. Therefore, we cannot discount the possibility that a marginal increase in blood glucose contributes to attenuation in endothelial function in offspring.
Insulin resistance and endothelial function
Insulin resistance is associated with endothelial dysfunction, cardiovascular risk (14), and incident cardiovascular events in epidemiology studies (15). Although our sample size is small, we found insulin resistance associates with impaired EDV. The relationship was predominant in FH−, with diminished endothelial function in the most insulin-sensitive FH+ offspring. In Native Americans, another high-risk group for development of diabetes, HOMA-IR was predictive of incident diabetes but not predictive of incident cardiovascular disease after adjustment for established cardiac risk factors including body mass index, waist circumference, blood pressure, and lipids (16). Consistently, we demonstrate HOMA-IR is more strongly related to traditional cardiac risk factors than with EDV.
Subclinical inflammation is associated with insulin resistance (17) and cardiovascular disease (18). In our study, endothelial dysfunction in FH+ offspring was not reflected by differences in multiple inflammatory measures. Nonetheless, other cytokines/adipokines could underlie attenuated EDV in offspring.
Although insulin resistance precedes development of diabetes in high-risk cohorts including offspring of diabetic parents, Pima Indians, and other ethnic minority populations, (6,19,20) insulin resistance per se may not be sufficient for development of diabetes in Caucasian persons without FH of disease who are at lower risk (1). A FH of diabetes, therefore, may impart additional factor(s) permissive for progression to disease. As FH of diabetes remains significantly correlated with endothelial function, it is interesting to speculate whether similar environmental or genetic risk factor(s) not assessed in current measures underlies both endothelial dysfunction and predisposition to diabetes.
We demonstrate that nondiabetic offspring of diabetic parents have impaired EDV. These data suggest bioavailability of nitric oxide is lower in offspring of two T2D parents and may contribute to cardiovascular risk in advance of development of overt diabetes. Modest hyperglycemia, even within the normative range, may contribute to attenuated endothelial function. Finally, as endothelial dysfunction is present in nondiabetic offspring of diabetic parents, strong FH of T2D could be considered an additional cardiac risk factor.
↵1 Dr. Creager is the Simon C. Fireman Scholar in Cardiovascular Medicine at Brigham and Women’s Hospital.
↵2 Dr. Varo is now at Clinica Universitaria de Navarra, Pamplona, Spain.
↵3 Dr. Schonbeck is now with the Department of Cardiovascular Disease, Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, Connecticut.
Supported by NIH P01 HL48743, K23-DK02795, P30 DK36836, M01 RR001032.
- Abbreviations and Acronyms
- endothelium-dependent vasodilation
- endothelium-independent vasodilation
- family history
- subjects with both parents having type 2 diabetes
- subjects with no first-degree relative with diabetes or coronary artery disease
- glycosylated hemoglobin
- homeostasis model assessment of insulin resistance
- type 2 diabetes
- Received October 18, 2005.
- Revision received February 1, 2006.
- Accepted February 14, 2006.
- American College of Cardiology Foundation
- Goldfine A.B.,
- Bouche C.,
- Parker R.A.,
- et al.
- Caballero A.E.,
- Arora S.,
- Saouaf R.,
- et al.
- Balletshofer B.M.,
- Rittig K.,
- Enderle M.D.,
- et al.
- National Diabetes Data Group
- Vita J.A.,
- Keaney J.F. Jr..,
- Larson M.G.,
- et al.
- Balkau B.,
- Shipley M.,
- Jarrett R.J.,
- et al.
- Williams S.B.,
- Goldfine A.B.,
- Timimi F.K.,
- et al.
- Hsueh W.A.,
- Quinones M.J.
- Hanley A.J.,
- Williams K.,
- Stern M.P.,
- Haffner S.M.
- Resnick H.E.,
- Jones K.,
- Ruotolo G.,
- et al.
- Festa A.,
- D’Agostino R. Jr..,
- Howard G.,
- Mykkanen L.,
- Tracy R.P.,
- Haffner S.M.
- Blake G.J.,
- Ridker P.M.