Endothelial-Dependent Vasodilation and Incidence of Type 2 Diabetes in a Population of Healthy Postmenopausal Women

Patients enrolled in the study were selected from women referred to the Bene Essere Donna center, an institution dedicated to the study of menopause-related disorders. This center is open to all postmenopausal women aged ≤60 years. Menopause was defined as the absence of menstruation from >6 months or a plasma level of 17β-estradiol <120 pmol/l and/or follicle-stimulating hormone >40 IU/l.

At baseline, each participant underwent fasting blood testing for levels of total cholesterol and triglycerides as well as a 75-g oral glucose tolerance test (OGTT), as required by our protocol. Participants were eligible for this study if they had a normal OGTT (fasting plasma glucose <110 mg/dl and 2-h plasma glucose <140 mg/dl) (14). Physical examination variables measured at baseline included body weight, waist circumference, and systolic and diastolic blood pressure. Patient history, 12-lead electrocardiogram, and echocardiogram were used to exclude past or present cardiac diseases. At baseline, all women underwent a vascular reactivity test on the brachial artery to measure the endothelium-dependent vasodilatation using a noninvasive ultrasound method. Participants provided questionnaire data concerning lifestyle practices and potential risk factors for cardiovascular disease.

Hypertension (blood pressure ≥140/90 mmHg or use of antihypertensive medications), hyperlipidemia (total cholesterol plasma levels ≥200 mg/dl and/or triglyceride levels ≥170 mg/dl), smoking habits, and obesity (BMI ≥30 kg/m²) were the exclusion criteria of the study; women treated with hormone replacement therapy were also excluded.

A total of 840 healthy postmenopausal women (mean age 53 ± 6 years) with normal glucose tolerance satisfied the above-mentioned criteria. Enrollment started in 1997, and the results were finally elaborated at the end of April 2004, when the last women to enroll had been followed for at least 0.5 years; the mean follow-up was 3.9 ± 0.7 years (range 0.5–6.9). All participants gave written informed consent to participate in this prospective study, which was approved by the ethics committee of the University of Modena and Reggio Emilia.

The technique for assessing brachial artery flow-mediated dilation (FMD) has been described in detail elsewhere (15–18). Briefly, FMD was assessed in the subject’s right arm in the recumbent position after a 15-min equilibration period in a temperature-controlled room (22–25°C) by means of a Acuson 128 XP/10 mainframe (Acuson, Mountain View, CA) with a 7.0-MHz linear array transducer. The brachial artery was longitudinally imaged ∼5 cm proximal to the antecubital crease, where the clearest image was obtained, and the diameter at end-diastole was measured (the mean of three measurements was used in the analysis). A pressure cuff, placed on the forearm (proximal to the target artery), was inflated until no blood flow was detected through the brachial artery with the Doppler probe. After 5 min, the cuff was released, and this was followed by an increase in blood flow. This phenomenon increased shear stress, which served as the stimulus to induce vasodilation. After cuff release, the diameter of the brachial artery was measured at 45, 60, 90, 180, and 300 s. The maximum diameter in any of these measurements was used in the calculation of FMD according to the following formula: (maximum diameter during reactive hyperemia − diameter at baseline)/diameter at baseline × 100%.

Measurement reproducibility was assessed in 50 randomized women who were examined twice, 1 h apart, by two different investigators. The intraobserver intersession coefficient of variation (CV) (evaluated as the SD of the mean difference/√2 × pooled mean) (19) was 3.3% for basal brachial artery diameter and 12.4% for FMD.

All women were seen in our outpatient clinics at regular intervals (every 6 months). Telephone contact was used every 3 months to reduce the drop-out rate. At baseline and every 6-month follow-up visit, women underwent an interview, an examination, and blood collection. Fasting serum glucose measurements were performed at each visit. New cases of diabetes were identified in accordance with the criteria used in the Atherosclerosis Risk in Communities (ARIC) study (20,21), as follows: 1) self-reported use of hypoglycemic medications, 2) fasting (>8 h) serum glucose level >126 mg/dl, 3) nonfasting serum glucose level >200 mg/dl, or 4) self-report of physician diagnosis. For individuals classified by physician diagnosis or medication use, the date of the onset of diabetes was considered to be the midpoint between the last visit when a woman was not diabetic and the first visit when a woman was diabetic. For those diagnosed by fasting or nonfasting glucose level, the date of the onset of diabetes was the estimated date at which blood glucose levels crossed the above-mentioned threshold, assuming a linear increase in glucose levels between visits. All of the patients with diabetes were telephoned in August through September 2004 in order to avoid false diagnosis of diabetes.

Descriptive statistics were tabulated as the mean ± SD or frequency percentage. Differences in baseline characteristics between the groups were examined by an ANOVA and χ² test, when appropriate. We used the Cox proportional hazards regression model to analyze the association between endothelial function and incident diabetes. Person-time was calculated from enrollment until the date of diabetes onset, death, drop out, or the end of the study, whichever occurred first. FMD was evaluated in the following two ways: divided into tertiles and as continuous term. We computed crude and multiple-adjusted hazard ratios (and 95% CIs) as a measure of the relative risk (RR) of incident diabetes for decreasing FMD, with the highest tertile as the reference. Adjusted estimates of risk were calculated by use of the Cox proportional hazards regression model, which controlled for a range of potential confounders, including age (continuous), family history of diabetes (yes/no), duration of postmenopausal period (continuous), alcohol consumption (never, 1–5 drinks per month, 2–6 drinks per week, and ≥1 drink per day), physical activity (never, 1–2 times per week, and ≥3 times per week), baseline systolic and diastolic blood pressure (continuous), BMI (continuous), waist circumference (continuous), fasting plasma levels of glucose, and total cholesterol and triglycerides (continuous), as well as the changes from baseline to the last follow-up visit (continuous). A two-sided P value <0.05 was considered statistically significant.

A total of 840 women were followed for 3.9 ± 0.7 years (3,252 person-years). During follow-up, 102 incident cases of diabetes occurred (equivalent of 7.9 cases/1,000 women-years). Of these, 58 of 102 (56.9%) patients were diagnosed by their own personal physician, 41 of 102 (40.2%) patients were diagnosed at our clinic, and 3 (2.9%) patients stated that they were taking hypoglycemic drugs. All diabetic patients were informed by telephone, and none died. All of them were being administered at least one hypoglycemic drug; of 102 patients, 27 (26.5%) were taking an association of two hypoglycemic drugs and only 6 (5.9%) were being treated with insulin. Six patients among the general population died during follow-up, three from non-cardiac-related causes (one accident and two neoplasm) and three from acute myocardial infarction. We lost contact with four patients.

Table 1 gives the baseline characteristics of the participants according to FMD tertiles. No significant differences were registered across tertiles. In Table 2, the changes of the parameters detected both at baseline and during follow-up are shown. Only systolic blood pressure revealed a significant increase in every tertile considered. Nevertheless, we found no significant differences in the intertertile comparisons.

The RRs of diabetes, according to FMD tertiles, are summarized in Table 3. The RR decreased steadily across FMD categories compared with the reference tertile (FMD ≥5.6%). Adjustment for age and various confounders attenuated the RR only slightly. When FMD was examined as a continuous variable, each 1-unit decrease of FMD was associated with a significant 32% (95% CI 22–48%) increase in the multiple-adjusted RR of incident diabetes.

The results of our prospective study clearly demonstrate that endothelial function significantly influences the future development of diabetes, independently of age and several other well-known diabetes risk factors. In our opinion, this is a very important tool because it radically changes the way endothelial dysfunction is considered. Endothelial dysfunction is usually explained as being the consequence of the endothelium being exposed to damaging factors, e.g., high blood pressure, high cholesterol, high blood glucose, smoking, etc.—the response-to-injury theory (3). Our data revolutionize the concept because they indicate that endothelial dysfunction may influence the development of diabetes. The present results have been obtained by studying a population of postmenopausal women who represent a unique model of studying endothelial dysfunction consequences. In fact, the decrease in estrogens that physiologically follows menopause does in itself compromise the endothelial function in women, even in the absence of other cardiovascular risk factors (1,2).

The current data support the conclusions of Meigs et al. (13), who demon-strated, in the large cohort of postmenopausal women examined in the Nurses’ Health Study, that elevated plasma levels of molecular biomarkers of endothelial dysfunction were significant predictors of incident diabetes. The difference between this and the present study is that endothelial function has been evaluated, in our experience, directly on a large conduit artery rather than indirectly by evaluating the concentration of spillover markers whose overproduction reveals a suffering of the endothelial cells in the capillary and arteriolar microcirculation. The concentration of the previously mentioned biomarkers shows a modest but also meaningful correlation with the endothelial function assessed by brachial artery FMD (22,23). The data of our study are, therefore, in line with the hypothesis that endothelial dysfunction, however it is determined and independent of which vascular district is involved, is always associated with a meaningful risk of developing diabetes in postmenopausal women.

The explanation of this has not been completely clarified and is still a focus of research. One of the most interesting hypotheses, in our opinion, is that of Pinkney et al. (11), who identified the endothelium as the principal controlling factor of insulin concentrations in the interstitium and, therefore, the amount of insulin effectively reaching the target cells. This hypothesis considers the endothelium as a “barrier,” which, when dysfunctional, limits the contact between the insulin and the insulin-sensitive cells. In other words, normal endothelial function would be a fundamental prerequisite for normal insulin action, and therefore endothelial dysfunction would be accompanied by a progressive insulin resistance. Moreover, experimental and clinical studies suggest that a major determinant of insulin effectiveness is the trans endothelium insulin transport (24,25). Contrarily, studies in humans and animals and in vitro experiences have generated many hypotheses on the possible link between insulin resistance and endothelial dysfunction. Insulin stimulates NO production in cultured endothelial cells extracted from human umbilical vein (26). However, insulin is a vasodilator that stimulates endothelial NO production in humans (27,28). Stimulation of NO production by insulin is mediated by signaling pathways involving phospatidylinositol 3-kinase (PI3K) and activation of endothelial NO synthase (eNOS) (29,30) and mitogen-activated protein kinase (MAPK) (31). These data provide a molecular basis for the dependency of insulin action on endothelial function or vice versa. In vitro, inhibition of PI3K blocks the ability of insulin to stimulate increased expression of eNOS and increases the expression of vascular cellular adhesion molecule-1 and E-selectin, as well as the rolling interaction of monocytes with endothelial cells (30). The MAPK pathway appears to mediate not only the ability of monocytes to migrate, but also the expression of the prothrombotic, profibrotic factor plasminogen activator inhibitor-1 (32).

It is well known that FMD induced by reactive hyperemia is endothelium dependent (33,34); in other words, it depends on the ability of the endothelium to produce vasodilator-endowed substances, mainly NO. It is well demonstrat-ed that postmenopausal status is associated with a significant reduced arterial NO activity (35). It is reasonable to conclude that our patients with a lower level of FMD had worse endothelial dysfunction, implying a lower production of NO. Some important biological phenomenon, such as the glucose uptake from the insulin target cells and the production of glucose-stimulated insulin secretion, are influenced by mechanisms involving the production of NO from the endothelial cells (36,37). A mutation of the eNOS able to determine a lower production of NO is associated, in fact, with a reduction in glucose uptake, hyperlipidemia, and hypertension in animals (36), and the administration of an inhibitor of the eNOS leads to a significant reduction in glucose uptake in humans (37). On the other hand, it is well known that the activity of glucokinase, the enzyme that plays a key role in glucose-stimulated insulin secretion from pancreatic β-cells, is controlled in its action from endothelium-derived NO (38). These data explain why drugs able to improve endothelial function, such as pravastatin and ramipril, are also able to reduce the risk of developing diabetes (39,40). For the same reason, some insulin-sensitizing drugs, such as troglitazone and metformin, even when used in patients without diabetes, improve endothelium-dependent FMD and reduce the risk of developing diabetes (41,42).

Our study has some limitations in that it is an observational, nonrandomized study. For this reason, results must be considered with caution. We cannot exclude, for example, that endothelial dysfunction may be simply an epi-phenomenon, resulting from a hidden confounding etiological cause, leading to both diabetes and endothelial dysfunction. However, our study presents several strengths, including the prospective design and the large and homogeneous cohort. All of these factors encourage us to hypothesize a causal link between impairment of endothelium-dependent vasodilation and incident diabetes.

In conclusion, our data support the idea that endothelial dysfunction is a fundamental step in the diabetogenesis process. Today it is possible to study, in a noninvasive way, the endothelial function through ultrasound. This allows us to prospectively identify those patients with a poor endothelium-dependent FMD. The latter should be regarded as a marker of diabetes risk in postmenopausal women. Close follow-up and “aggressive” management of other diabetic risk factors should be justified in these subjects.

This study was partially supported by the Italian MURST (Min-istero dell’Università e della Ricerca Scientifica e Tecnologica) and was sponsored by WOCDA (Woman’s Cardiovascular Disease Association).