Increases in Central Aortic Impedance Precede Alterations in Arterial Stiffness Measures in Type 1 Diabetes

The Wisconsin Diabetes Registry Project was a population-based cohort of individuals with type 1 diabetes (17–21). The Cardio-Diab Study cohort included 155 of these individuals with no known CVD. Each subject with type 1 diabetes invited a sex- and race-matched nondiabetic sibling or cousin, or if neither was available, a friend, within 5 (preferably) or 10 years of his or her age to serve as a control. Twenty consecutive pairs participating in the Cardio-Diab Study in June 2004 through January 2005 were enrolled. The primary end point was the difference in aortic characteristic impedance between groups. The sample size gave us >80% power to detect a 30% difference in aortic impedance (16). The protocol was approved by the University of Wisconsin Human Subjects Committee, and each participant gave separate informed consent for this substudy.

Participants were studied in the supine position after quiet rest (22–24). With a semiautomated, computer-controlled cuff, auscultatory blood pressure was obtained at 2-min intervals with a goal of obtaining three sequential readings within 5 mmHg for systolic blood pressure and diastolic blood pressure. Arterial tonometry was obtained from the brachial, radial, femoral, and carotid arteries in quick succession with a custom transducer. A limb-lead electrocardiogram (ECG) was recorded. Echocardiographic images of the left ventricular outflow tract were obtained in a parasternal long-axis view. Body surface measurements were assessed from suprasternal notch to brachial, radial, femoral, and carotid recording sites (22). Data were digitized during the primary acquisition, transferred to CD-ROM, and shipped to Cardiovascular Engineering (Waltham, MA) for analysis.

Noncontrast computerized tomography (CT) scans of the chest were performed using a GE HiSpeed Advantage spiral CT scanner (General Electric, Milwaukee, WI) using ECG gating. Scans were used to assess the outer diameter of the aorta 2 cm above the aortic valve. When the aorta was noncircular, the smallest dimension was used. A single timed urine collection was performed for determination of albumin concentration and creatinine clearance.

Tonometry waveforms were signal-averaged with the ECG used as a fiducial point (25). Average systolic and diastolic cuff pressures were used to calibrate the peak and trough of the signal-averaged brachial pressure waveform. Diastolic and integrated mean brachial pressures were used to calibrate other pressure tracings (26). Carotid-femoral PWV was calculated from transit time and body surface measurements corrected for parallel transmission as described previously (23). LVOT diameter was measured and used to compute cross-sectional area and to convert flow velocities to volume flows (27). The augmentation index was calculated using carotid waveforms (28). Aortic Z_c was estimated in the time domain as the pressure change associated with an increase in flow from 0 to 95% peak flow (ΔP/ΔQ). Proximal aortic compliance per unit length was calculated as described previously (29).

Baseline characteristics and hemodynamic data were compared using paired Student’s t tests and Pearson correlations. Multivariable regression analyses were performed two ways. Correlations between impedance and other clinical variables of interest were explored in a regression analysis using matched pairs and thus controlling for diabetes effect. An unmatched multivariable regression analysis was also performed to determine which variables contained independent predictive value with regard to impedance and pulse pressure. Classical assumptions of linear regression analysis were satisfied for all analyses.

Forty participants, 20 matched pairs, were sequentially studied. Complete arterial hemodynamic data were obtained from 17 pairs of participants. Seven type 1 diabetic subjects had related control subjects. The remaining 10 type 1 diabetic subjects had unrelated control subjects. Clinical characteristics of the participants are shown in Table 1. Although diabetes had been present on average for 15 years in the patients, the aggressiveness of diabetes management and control were demonstrated by the low prevalence of hypertension and microalbuminuria.

Arterial hemodynamic data are shown in Table 2. Systolic blood pressure and MAP were not statistically different in the two groups. Diastolic blood pressure was lower in the group with type 1 diabetes (P = 0.05). Both brachial and carotid pulse pressure were significantly elevated in the group with type 1 diabetes. Aortic Z_c was markedly elevated in patients with type 1 diabetes. Those with type 1 diabetes had larger forward pressure amplitude despite comparable peak flow (Fig. 1). PWV and the augmentation index, however, did not differ between type 1 diabetic and control subjects.

A decrease in aortic diameter could explain increased Z_c with no difference in PWV, given the strong dependence of impedance on vessel diameter. LVOT diameter determined by echocardiography and proximal aortic diameter determined by CT scan were both smaller in the group with type 1 diabetes, with the LVOT measurement reaching statistical significance (Table 2).

Correlation analysis was used to determine relations between both aortic impedance and pulse pressure and other clinical variables. Initial correlations were performed with matched pairs being preserved in the analysis, in effect controlling for diabetes. In addition to the presence of diabetes, urine albumin excretion, A1C, and female sex were significantly correlated with Z_c (Table 3). Multivariable analysis was performed using unmatched pairs but including diabetes as a variable. Urine albumin and LVOT diameter had independent predictive power in this model, with an F statistic of 5.3 (P = 0.003); however, none of the individual P values in this model met the Bonferroni standard of P = 0.0125 for four comparisons. In multivariable analysis, only diabetes, impedance, and BMI remained independent predictors of central pulse pressure.

In this study we investigated changes in the large and medium arteries of young people with type 1 diabetes and demonstrated an increase in Z_c in type 1 diabetes, independent of MAP. Increased Z_c was associated with a higher level of pressure pulsatility in the diabetic aorta for a given level of flow. This increase in Z_c occurred before changes in more traditional measures of central arterial stiffness such as PWV, which has been shown to be elevated in more advanced stages of diabetes (6–8). The increase in Z_c was associated with increased central and peripheral pulse pressure in the patients with diabetes and was accompanied by a decrease in LVOT diameter. Although an increase in PWV and earlier return of reflected waves to the proximal aorta can cause an increase in central pulse pressure, particularly in younger people, premature wave reflection was not the mechanism of increased central or peripheral pulse pressure in our type 1 diabetic patients. Rather, the primary finding was an increase in incident pressure wave amplitude, indicating changes in the proximal aorta, leading to a greater pressure change for a given flow from the left ventricle.

Increased central arterial stiffness has been demonstrated previously in patients with type 2 diabetes (8–10). Early changes in diabetes and in type 1 diabetes in particular have not been as well characterized (11). Postmortem analysis demonstrated that aortas from individuals with type 1 diabetes were intrinsically stiffer than those from age-matched control subjects (12). In another study of participants with type 1 diabetes, diabetes was associated with increased aortic augmentation index in men but not in women (6). However, use of generalized transfer functions to calculate central pressure waveforms in diabetic subjects has been questioned, and changes in PWV and the augmentation index have not always correlated in studies of patients with diabetes (13,15). Furthermore, none of the foregoing studies measured Z_c.

Increased pulse pressure is known to be a marker of adverse clinical events in patients with diabetes (2,3), and pulse pressure is increased when impedance rises. From the waterhammer equation, Z_c is directly proportional to PWV and inversely proportional to vessel area (Z_c = PWV × ρ/area, where ρ is the density of blood). Impedance will thus be sensitive to changes in vessel diameter. Because PWV was not different between the two groups in our study, we suggest that a small decrease in aortic diameter could explain our data. This suggestion is supported by the decrease in LVOT area noted in the diabetic group in this study, which maintained significance in multivariable modeling. The failure to note a decrease in aortic diameter may result from use of noncontrast CT scans, allowing visualization of only the external vessel diameter. Eutrophic aortic remodeling results in a decreased internal lumen diameter due to vessel wall thickening, without a change in outer vessel diameter. If this sort of aortic remodeling were to occur in diabetes, leading to increased pulsatile pressure in the proximal aorta, it would be missed on a noncontrast scan. Decreases in internal aortic diameter have been noted previously in patients with diabetes (14) and those with hypertension (16). This decrease in aortic diameter with pathological conditions contrasts with the modest age-related increase in aortic diameter that has been observed in relatively healthy individuals. With aortic wall stiffening and aortic diameter increases, as in healthy aging, a predominant increase in PWV is observed because the increase in diameter offsets the effect of wall stiffening on Z_c (16). The present findings suggest that factors responsible for proper matching between aortic diameter and resting flow may be impaired in type 1 diabetes, indicating an active and potentially plastic vascular remodeling process at work in young patients with this disease process.

There are important implications of increased central arterial pulsatility early in the disease process of diabetes. The increased pulsatility with the change in Z_c is carried into the periphery, as manifest by increased brachial pulse pressure. The presence of higher pulse pressure in the central and peripheral circulation may lead to transmission of pressure pulsatility farther into the microcirculation than normally found and may precipitate microvascular damage. Evidence for microvascular damage is suggested by the correlation between increased Z_c and urine albumin levels in our study. Increased glomerular perfusion pressure, as might be seen if pressure pulsatility increases, is known to lead to renal microvascular damage (30). Our data suggest that changes in the central aorta may precipitate renal microvascular change or, conversely, that early microvascular damage in the kidney may contribute to increases in pressure and flow pulsatility at all levels of the circulation. The fact that the correlation with urine albumin is seen in the absence of overt microalbuminuria suggests that the aortic changes, which were substantial, may precede and contribute to the renal changes. In addition, our data suggest that once microalbuminuria develops, the relationship between arterial hemodynamics and renal vascular function may be altered, a concept supported by others (31). A similar relationship between pulse pressure or measures of central aortic stiffness and urine albumin excretion has been noted previously in type 2 diabetes (7,32,33), whereas urine albumin does not necessarily correlate with hyperinsulinemia (34), again suggesting the importance of hemodynamic changes in the development of renal microvascular disease. The link between pulse pressure or central arterial stiffness and urine albumin levels has also been demonstrated in patients with hypertension (16,35–41) and in healthy adults (42,43). In addition to damaging microvessels, the higher degrees of pressure pulsatility seen in type 1 diabetes may also increase tensile and shear stresses in the large and medium vessels and could contribute to early atherogenesis.

There are limitations to our study. The group studied was small and may not be representative of people with type 1 diabetes. In addition, by necessity, this exploratory analysis of stiffness involved multiple comparisons, and spurious significant values are possible. The study was powered for the primary end point only, and other significant findings must be interpreted as exploratory rather than definitive. With the small sample size, outliers may significantly influence the data and lead to spurious findings. The urine albumin determination is based on a single timed collection rather than on an average of repeated collections. Our conclusion that aortic diameter is reduced because Z_c is increased but PWV is not assumed that Z_c and PWV are measured at the same point. Z_c assesses the proximal aortic root whereas carotid-femoral PWV evaluates the spatially averaged properties of the descending thoracic and abdominal aorta and iliac artery. Thus, regionally heterogeneous changes in arterial properties, rather than a difference in aortic diameter, may explain elevated Z_c with unchanged carotid-femoral PWV. These data are cross-sectional. Longitudinal data on a larger group of subjects would be required to demonstrate aortic remodeling and the relationship to changes in impedance and other measures of stiffness over time.

In summary, we have shown increased pulse pressure in young individuals with type 1 diabetes, resulting from changes in aortic characteristic impedance at similar MAPs. The changes in Z_c may be related to decreased aortic diameter. Z_c correlates with urine albumin levels, which suggests a link between central arterial properties, increased central and peripheral pulse pressures, and microvascular damage at an early stage in the pathophysiology of diabetic vascular disease. We suggest that increased vessel pulsatility, delivered distally in the circulation, may expose organs to greater degrees of small vessel pulsatile flow than is optimal, leading to microvascular damage and organ dysfunction.

The Cardio-Diab Study was funded by the National Heart, Lung and Blood Institute Grant HL62897. N.K.S. is funded in part through National Institutes of Health Grant AG01022 K23. This work was funded by the University of Wisconsin Department of Medicine, Madison, Wisconsin.