We studied 91 subjects (52 men, age 52 ± 14 years) with and without cardiovascular disease. Of these patients, 33 were normal controls (NL), 28 were patients with isolated systolic hypertension (HTN), 10 were normotensive patients with diabetes mellitus (DM), 10 were patients with systolic heart failure (CHF), and 10 were patients who had undergone renal transplantation (RT). Clinical data, blood pressure, and cardiovascular risk factors were gathered and all patients were then studied with simultaneous applanation tonometry and pulsed-wave Doppler and then had 2D ultrasound and tissue Doppler performed on their carotid artery.

Applanation tonometry was performed on the right carotid artery in all patients after they had been allowed to rest for 5–10 minutes. Tonometry uses a transcutaneously applied, micro-manometer-tipped probe which is placed against an arterial wall. Application of sufficient pressure to distort, or applanate the artery creates a signal which approximates instantaneous arterial pressure (Figure 1). This is then digitized and reconstructed on a personal computer. Blood pressure was measured using a standard sphygmomanometer on the right brachial artery, after the patient was allowed to rest 5–10 minutes. Calibration of the tonometric waveform was performed by assuming equivalence of mean [(2*DBP + SBP)/3] and diastolic brachial cuff pressure.

Total arterial compliance (TAC) was derived using the pulse pressure method, using a combination of stroke volume (derived by echocardiography from the dimensions and pulsed-wave Doppler measurement of flow in the left ventricular outflow tract (LVOT)) and the pressure waveform obtained from applanation tonometry of the radial artery 14. These waveforms were obtained simultaneously with the pulsed-wave Doppler, digitized (WaveBook 512, IOTech Inc., Cleveland, OH), and transferred to a laptop computer, where they were synchronized using the R wave of the electrocardiogram. Using specialized acquisition software, three sets of gated data (ECG, tonometry, Doppler) were acquired and stored for off-line analysis (figure 1). Depending on heart rate, this was usually 20–30 cardiac cycles per patient. Echocardiographic images and pulsed Doppler were acquired using a Philips HDI5000 ultrasound system (Philips Inc, Bothell, WA) with a 1.7 MHz harmonic imaging probe.

For calculation of total arterial compliance, the binary files were processed and analyzed using a custom analysis program written in MatLab 4. Between five and ten cardiac cycles of tonometry and Doppler were chosen from the raw dataset based on data quality and averaged for analysis. With the use of cursors to identify the beginning and end of aortic ejection, peak ejection, and peak pulse pressure on both the reconstructed Doppler and central pressure waveforms, the analysis program determined mean values for pressure and flow and calculated a mean aortic pressure (figure 2). Based on the pressure and flow data derived from the central pressure waveform and the Doppler, the analysis program generated values for total arterial compliance, augmentation index, and other hemodynamic indices. We have previously reported the intra-observer variation (mean TAC 1.17 ± 0.02 ml/mmHg) and coefficient of variation of this method (1.7%) 15.

The carotid arteries were scanned longitudinally in the anterior, lateral and posterior aspects 2–10 cm below the bifurcation, and digital cine loops are acquired for offline analysis. The image was then optimized for TDI using the smallest possible region of interest (ROI) box to achieve the highest frame rate, which was usually between 160–220 frames per second. The 2D and color Doppler settings were also customized for extraction of vessel displacement: for 2D imaging – dynamic range of 150 dB, the 2D option is set to penetration, persistence is set at low and frame rate is set at maximum; for the color tissue Doppler – the gain was set at 100%, with a PRF of >200 Hz. Care was taken not to include any discrete plaques in the tissue Doppler measurements. Loops of 3–5 cardiac cycles with TDI were then acquired digitally for offline analysis (figure 3).

The TDI images were analyzed offline using custom-written software (AWM v1.05, Philips Medical Systems, Bothell, WA) which extracts the velocity information for the ROI area over the cardiac cycle, and with a processing algorithm, generates values for vessel wall displacement (in microns) over time 16. These data were then saved and exported in numerical format for analysis (figure 4).

Both the extracted TDI displacement curves and the tonometric curves were then imported into a custom written MatLab program for analysis and once calibrated, a comparison was done between the waveforms (figure 5). Carotid displacement curves (in μm) derived from the TDI recordings were transformed into approximated pulse pressure curves (in mmHg) following the methodology described by Van Bortel et al based on the observation that the mean blood pressure is constant throughout the large artery tree 17, as well as the diastolic pressure 18. The tonometric recordings were calibrated setting the average integrated curve equal to the mean blood pressure [derived from (2*DBP+SBP)/3] or if used, from an average automatic BP recording system. The minimum diastolic tonometric value was set equal to the diastolic BP. Similarly, a two-point calibration was used for the carotid distension curves, integrated with the mean set to the mean BP, and the minimum distension set to the diastolic BP.

Pearson's correlation was used to determine the concordance between central pressure determined by TDI and by tonometry, paired t-tests were used to determine the mean differences between the two techniques and Bland-Altman analysis was used to determine the differences from the mean for central pressure between the two techniques.

Table 1 shows clinical data, medications, blood pressure, flow and compliance for the whole patient group. Patients had an average of 1 ± 1 cardiovascular risk factors, and were on 1 ± 1 antihypertensive medications. However, blood pressure, cardiac output and TAC were all within the normal range.

Correlation between central pressure by TDI and tonometry was excellent for maximum pressure (r = 0.97, p < 0.0001). In paired t-tests the mean differences between central pressures derived by the two techniques was minimal, although there was a 5 mmHg difference between maximum pressure derived by TDI and tonometry (table 2). Bland-Altman analysis was then performed to determine the concordance between the two techniques. For diastolic pressure the differences were negligible (1.20 ± 1.2 mmHg) but this was expected since both sets of data were calibrated with diastolic cuff pressure (figure 6). For systolic pressure however the difference from the mean was greater (5.36 ± 5.5 mmHg) with the TDI underestimating maximum pressure (figure 6).

Because of the calibration with mean and diastolic pressure, only maximum pressure was compared in a subgroup analysis. The HTN group had the highest maximum pressure and the CHF patients had the lowest pressure, as would be expected. The controls, DM group and RT groups all had maximum pressure within a normal range. In paired t-tests comparing the TDI and tonometry the HTN group had the highest paired difference (9 ± 6 mmHg; p < 0.0001) and the controls had the smallest paired difference (2 ± 2 mmHg; p < 0.0001) (table 3).

The TDI technique underestimated systolic pressure in about a third of cases resulting in a higher than acceptable difference in the Bland-Altman analysis. The causes of this appear to be related to inappropriate TDI settings and other technical problems, and therefore this may be avoided with experience and guidelines. Failure to use a sufficiently high pulse repetition frequency caused aliasing of the velocity signals and the resulting displacement waveforms and calibration by the technique described above produced a pressure waveform which does not reflect accurate maximum pressure when compared to the carotid tonometry in several cases (Figure 7). Excessive motion of the carotid artery may lead to failure of the edge detection mechanism. This excessive motion has affected the reproducibility of carotid tonometry in other studies and can be seen here as well (Figure 7) and may be controlled by a simple breath hold or transducer immobilization. Other causes of failure of the edge detection with TDI are reverberation artifacts, excessive noise in the vessel lumen and excessive vessel movement due to breathing or high mobility of the carotid artery (Figure 8). Once again these may be able to be corrected with breath holds, patient positioning, and careful attention to 2D and Doppler settings. Thus, with attention to technical aspects and patient preparation this technique is likely to be highly feasible.

Total arterial compliance is reduced with age, vascular disease and hypertension 192021, and is linked with the sequellae of these disorders. Reduction of compliance leads to an increase in afterload on the heart, an increase in pulse pressure and in turn, left ventricular hypertrophy 22 and diastolic dysfunction. Increased pulse pressure is a determinant of cardiovascular risk and mortality 2232425. Lower compliance also leads to lower diastolic pressure 26, resulting in a decreased coronary perfusion pressure 27. Several studies have shown a correlation between compliance and the presence of significant CAD 282930 as might well be expected due to common cardiovascular risk factors. Moreover, arterial compliance has been shown to be a contributor to the provocation of ischemia at stress testing 31, and an association has been documented between reduced exercise capacity and the presence of a lower ischemic threshold 3233.

While brachial blood pressure has been shown to be a strong predictor of cardiovascular morbidity, mortality and outcome 2223435 the role of central pressure in clinical outcomes is increasingly recognized 35. Independent of brachial blood pressure, central pulse pressure predicts left ventricular structure 3637 and the extent of coronary artery disease. The CAFÉ substudy of the ASCOT trial showed that central pressure and pressure augmentation were significantly associated with clinical endpoints 35. Despite insignificant differences between brachial pressures in the study groups over a four year follow-up period, the group with lower central pulse pressure showed a significant decrease in cardiovascular events. From the results of this study, assessment of central blood pressure appears to be paramount in the treatment of patients at risk for cardiovascular disease.

While applanation tonometry is considered the "gold standard" for use in assessing arterial stiffness it is not without limitations. The major limitation of pressure measurement with tonometry is the use of brachial cuff pressure to calibrate the signals. Past studies have shown poor inter-observer variability in taking brachial pressure and a poor correlation between reported and actual pressure 3839. Operator error and a learning curve with acquisition of tonometry also play a role in its usefulness as a clinical tool. And while several studies have shown the limitations of using transfer functions with radial tonometry {4, 5, 6, 7, 14}, the issues of performing carotid tonometry remain–lack of bony support underneath to achieve true applanation, obesity, respiratory changes and mobility of the artery