Reservoir-Excess Pressure Parameters Independently Predict Cardiovascular Events in Individuals With Type 2 Diabetes

This is an ancillary study of the SUMMIT-VIP study (Surrogate Markers for Micro- and Macro-Vascular Hard End Points for Innovative Diabetes Tools-Vascular Imaging Prediction).

Participants were 832 individuals recruited for the SUMMIT-VIP study from Exeter, Dundee (both United Kingdom), and Malmö (Sweden) sites. Of those, 306 had T2DM with cardiovascular disease (DMCVD, 70.4±7.8 years, 94F), 348 had T2DM without CVD (diabetes mellitus[DM], 67.7±8.4 years, 142F) and 178 had neither T2DM nor CVD ([CTRL], 67.2±8.9 years, 83F). A detailed description of the SUMMIT-VIP study, including inclusion/exclusion criteria, has been described elsewhere^4,27 and summarized in the Data Supplement. UK National Research Ethics Service South West Committee, East of Scotland Research Ethics Service, and the Institutional Ethics Committee at the University of Lund approved all study procedures, and written informed consent was obtained from all participants.

Applanation tonometry of the radial artery was performed using a SphygmoCor system (AtCor Medical Pty Ltd, West Ryde, Australia). Participants lay supine on an examination bed and rested for 10 minutes before assessment. Right radial artery pressure waveforms were recorded over 10 seconds with a high-fidelity micromanometer (SPT-304, Millar Instrument, Houston, TX) attached to the SphygmoCor system. Waveforms were then processed through dedicated software (SphygmoCor version 8.2) to calculate an ensemble-averaged radial pressure waveform calibrated by brachial systolic and diastolic pressures (as per the manufacturer’s suggestion), and a corresponding aortic pressure waveform was derived using a previously validated generalized transfer function.^28,29 The following central artery hemodynamic parameters were obtained for analysis in this study: central systolic and diastolic BP, central PP, central augmented pressure, and central augmentation index adjusted for a heart rate of 75 bpm. Three separate waveform data were acquired for each participant, and the average of these acquisitions was used as representative for statistical analysis.

A supine brachial BP was measured 3× at 1-minute intervals using validated semiautomated oscillometric devices (Omron M6, Hoofddorp, the Netherlands). The average of the last 2 measurements was used as a representative brachial systolic pressure, diastolic pressure, and PP.

Reservoir-excess pressure parameters were calculated based on a pressure-alone approach from the ensemble-averaged radial pressure waveforms without the application of a generalized transfer function. In the reservoir-excess pressure analysis,¹⁷ the measured pressure waveform can be separated into (1) a reservoir pressure component which varies in magnitude through changes in the resistance to outflow from the reservoir, the reservoir compliance, and the asymptotic pressure,³⁰ and (2) an excess pressure component which is the difference between the measured pressure waveform and reservoir pressure. The calculation of the reservoir pressure depends on determination of 2 rate constants: the systolic rate constant (SRC) which is the inverse of the product of the constant of proportionality between the excess pressure and the arterial inflow and the total arterial compliance; and the diastolic rate constant (DRC) which is the inverse of the product of the peripheral vascular resistance and the total arterial compliance. Figure 1 shows a schematic example of the reservoir-excess pressure separation. A review of the method, including its theoretical basis and validation, has been published recently.¹⁷ The following reservoir-excess pressure parameters were obtained for analysis in this study: reservoir pressure integral (INTPR), peak reservoir pressure (MAXPR), excess pressure integral (INTXSP), SRC, and DRC.

The participants with T2DM irrespective of the CVD status at baseline (n=654) were followed up for a period of 3 years. Of those, 12 participants were excluded from the final analysis (n=642) due to the missing data that were used to adjust for multivariate logistic regression analysis (Figure S1 in the Data Supplement). Clinical events that occurred during this period were recorded as defined in the SUMMIT-VIP study.²⁷ These included cardiovascular mortality, myocardial infarction, unstable angina, cardiac arrest, revascularization procedures, peripheral arterial disease, cerebrovascular events (stroke and transient ischemic attacks), and noncardiovascular mortality.

Data are presented as means±SD, means (95% CI), or number (%). Skewed data were log-transformed for statistical analysis. A χ² test was used to compare categorical parameters between groups. An ANOVA/ANCOVA with a Bonferroni post hoc test was used for between-group comparisons of continuous data (after adjusting for age and sex for variables other than age and sex). Due to the unavailability of time to composite cardiovascular event data, univariate and multivariate logistic regression analyses were used to determine associations between baseline values of reservoir-excess pressure parameters, and conventional central and peripheral artery hemodynamic parameters and incident cardiovascular events during the 3-year follow-up period in individuals with T2DM, and reported as odds ratio (OR) with 95% CI. Reservoir-excess pressure parameters and conventional central and peripheral artery hemodynamic parameters were standardized before entering into the logistic regression analysis to allow comparisons across the parameters. Covariates were chosen a priori as being established risk factors for CVD and complete case analysis was performed as levels of missingness were minimal (<10%). All statistical analysis was conducted using IBM SPSS Statistics 25 (IBM, Armonk, NY) and statistical significance was set at P<0.05.

Characteristics of the study participants are presented in Table 1.

Reservoir-excess pressure parameters stratified by group are presented in Figure 2. INTPR was lower in DMCVD (85.8 [83.9–87.6] mm Hg·s) and DM (87.7 [86.0–89.4] mm Hg·s) than CTRL (95.9 [93.5–98.3] mm Hg·s). MAXPR was lower in DMCVD (104.6 [103.0–106.2] mm Hg) compared with DM and CTRL (107.9 [106.5–109.4] mm Hg and 109.8 [107.7–111.8] mm Hg, respectively). INTXSP was greater in DMCVD (7.6 [7.4–7.9] mm Hg·s) compared with DM and CTRL (7.0 [6.8–7.2] mm Hg·s and 6.7 [6.4–7.0] mm Hg·s, respectively). SRC was lower in a stepwise manner among the groups (DMCVD, 6.4 [6.2–6.6] 1/s, DM, 7.0 [6.8–7.3] 1/s, CTRL, 7.6 [7.3–7.9] 1/s). DRC was greater in DMCVD (2.4 [2.3–2.5] 1/s) than CTRL (2.2 [2.1–2.3] 1/s) but was similar between DMCVD and DM (2.3 [2.3–2.4] 1/s).

Conventional central and peripheral artery hemodynamic parameters stratified by group are also presented in Table 1. Central diastolic BP was lower and central PP was higher in DMCVD compared with DM and CTRL. DM showed a trend for lower central augmented pressure compared with DMCVD and CTRL. Brachial PP was higher in DMCVD and DM compared with CTRL. Other parameters were similar between the groups.

In the subgroup of individuals with T2DM (n=642), 14 deaths (6 cardiovascular and 9 noncardiovascular causes) and 108 composite cardiovascular events occurred during a 3-year follow-up period (see Table S1 for the breakdown of cardiovascular events that occurred). Table 2 shows selected baseline characteristics of individuals with T2DM stratified by the incidence of cardiovascular event during the 3-year follow-up period. In those who had a cardiovascular event during the follow-up period, hemoglobin A1c level was higher and heart rate was faster than in those who had no event. In the group that had had a cardiovascular event during the follow-up period, smoking, a previous history of CVD, and insulin treatment were more prevalent than in the other group. Other parameters were similar between the groups.

Reservoir-excess pressure parameters of individuals with T2DM stratified by the incidence of cardiovascular event during the 3-year follow-up period are presented in Table 2. At baseline, INTPR was lower, and DRC was greater in those with cardiovascular events than those without. INTXSP tended to be greater in those with cardiovascular events than in those without.

Conventional central and peripheral artery hemodynamic parameters of individuals with T2DM stratified by the incidence of cardiovascular event during the 3-year follow-up period are also presented in Table 2. Brachial PP was higher in those with cardiovascular events than in those without. Central PP tended to be higher in those with cardiovascular events during the follow-up period than in those without. Other parameters were similar between the groups.

Logistic regression analysis was performed to determine whether reservoir-excess pressure parameters and central and peripheral artery hemodynamic parameters would independently predict the incidence of cardiovascular events during the 3-year follow-up period in individuals with T2DM. In a minimally adjusted (age and sex) logistic regression model (Figure 3A), INTPR (OR, 0.699 [95% CI, 0.545–0.897], P=0.005), INTXSP (OR, 1.238 [95% CI, 1.011–1.628], P=0.041), and DRC (OR, 1.499 [95% CI, 1.217–1.846], P<0.001) were significant independent predictors of the incidence of cardiovascular events during the follow-up period. In addition, in the same minimally adjusted logistic regression model (Figure 3B), central PP (OR, 1.281 [95% CI, 1.001–1.640], P=0.049) and brachial PP (OR, 1.332 [95% CI, 1.042–1.703], P=0.022) were significant independent predictors of the incidence of cardiovascular events during the follow-up period.

In a multivariate logistic regression model after adjusting for age, sex, total and HDL (high-density lipoprotein) cholesterol, current smoking, systolic BP, pharmacological hypertensive treatment, and study center (Figure 3C), INTPR (OR, 0.594 [95% CI, 0.446–0.792], P<0.001) and DRC (OR, 1.602 [95% CI, 1.246–2.059], P<0.001) independently predicted the incidence of cardiovascular events during the follow-up period. In the same logistic regression model (Figure 3D), however, none of the conventional central and peripheral artery hemodynamic parameters independently predicted the incidence of cardiovascular events during the follow-up period. These observations remained unaltered after further adjustments for body mass index, resting heart rate, hemoglobin A1c, previous history of cardiovascular disease, estimated glomerular filtration rate, duration of T2DM, and insulin treatment (Table 3 and Figure S2 through S4).

When INTPR and DRC were forced into the same multivariate logistic regression model together with age, sex, total and HDL cholesterols, current smoking, systolic BP, pharmacological hypertensive treatment, and study center, odds ratio were negligibly altered, and INTPR (OR, 0.642 [95% CI, 0.464–0.888,] P=0.007) and DRC (OR, 1.434 [95% CI, 1.101–1.870], P=0.008) independently predicted the incidence of cardiovascular events during the follow-up period. The strength of these associations was essentially unaltered by further adjustments for body mass index, resting heart rate, hemoglobin A1c, previous history of cardiovascular disease, estimated glomerular filtration rate, duration of T2DM and insulin treatment (INTPR [OR, 0.566 [95% CI, 0.319–1.005], P=0.052), and DRC (OR, 1.397 [95% CI, 1.060–1.840], P=0.018), although the 95% CIs of the estimates were widened slightly.

Each reservoir-excess pressure parameter studied at baseline in this study has a different association with T2DM and the presence of CVD: INTPR was lower in T2DM than in CTRL irrespective of the presence of CVD. MAXPR was lower and INTXSP was greater in DMCVD compared with DM and CTRL. Interestingly, SRC and DRC exhibited stepwise alterations among DMCVD, DM and CTRL. From these observations, the information contained in the pressure waveform of the individuals with T2DM can be explained as follows. First, the lower INTPR in T2DM compared with CTRL indicates that a buffering capacity of central arteries, especially the aorta, is attenuated in T2DM due to stiffened central arteries, and this decline in the buffering capacity will result in a smaller proportion of stroke volume stored in systole and a faster discharge of blood from the aorta in diastole (consistent with the higher DRC). The proportionate increase in INTXSP compared with MAXPR in DMCVD may be interpreted as indicating an increase in excess work by the left ventricle and maybe an index of circulatory dysfunction.¹⁶ Indeed, the association of INTXSP with subclinical biomarkers of target organ damage¹⁸ and the prognostic utility of INTXSP has been demonstrated in several cohorts although not previously in people with diabetes.^18,21–23 Finally, the stepwise alterations in SRC and DRC with increasing cardiovascular risk (cf. DM versus DM and CVD) is indicative of how the rate of reservoir filling and emptying linked to central artery stiffness may influence different aspects of the ventricular-vascular interaction. These alterations in SRC and DRC provide insight into the accumulation of cardiovascular risk and may have a utility for cardiovascular risk stratification in T2DM. The potential utility of SRC for risk stratification has been demonstrated in people with hypertension¹⁹ and the general population,²⁴ and an association between DRC and kidney function has been reported.³¹ Taken together with our observations that DRC and INTPR independently predicted the incidence of cardiovascular events over 3 years in T2DM, it is likely that the parameters derived from the diastolic phase, ie, DRC and INTPR, may be more clinically relevant parameters than those derived from the systolic phase, at least in T2DM.

DRC and INTPR were both predictors of cardiovascular events in T2DM during the 3-year follow-up period, independent of conventional cardiovascular risk factors, and the prognostic ability of these parameters remained unaltered even after both parameters were forced into the same multivariate logistic regression model. These observations indicate that each parameter might possess unique prognostic information on the incidence of cardiovascular events in T2DM.

In the minimally (age and sex) adjusted logistic regression model, both central and brachial PP were significantly associated with the incidence of cardiovascular events during the 3-year follow-up period. However, neither parameter was associated with the incidence of cardiovascular events after adjusting for conventional risk factors and study center, and this was unaffected by further adjustment for other potential cofounders. Compared to these observations, earlier studies have reported that brachial PP is an independent risk predictor for cardiovascular events in T2DM.^11–13 The reason for these divergent results is unclear but could relate to well-controlled BP with pharmacological treatment in our T2DM participants. Nevertheless, given the prognostic utility of reservoir-excess pressure parameters along with the lack of prognostic utility with conventional central and peripheral hemodynamic parameters demonstrated in this study, these observations clearly suggest that the conventionally obtained central and peripheral artery hemodynamic parameters may not be capable of capturing clinical information from the pressure waveform sufficient to provide prognostic utility in T2DM. The lack of prognostic utility of conventional parameters analyzed in this study could arise from the fact that they are derived either from extreme points on the pressure waveform (eg, systolic and diastolic BP) or from derivatives calculated from specific points on the waveform (eg, PP and augmentation index), rather than extracting information from the waveform morphology as a whole (eg, DRC and INTPR). Conversely, the waveform analysis, such as reservoir-excess pressure analysis, could be useful in people with T2DM, even in those whose BP is well-controlled, as it may identify more subtle hemodynamic abnormalities.

Data on left ventricular function by echocardiography are not available in this study. Because the systolic phase of ventricular-vascular interaction and hence INTXSP and SRC may be dependent on myocardial contractility and stroke volume, our findings should be interpreted in this context. Additionally, the data on peripheral vascular resistance are not available. However, the majority of our follow-up T2DM participants were taking vasodilating antihypertensive medications, and accordingly the inter-participant variability on vasomotor tone might have been limited. Therefore, the relative dependence of DRC is shifted further toward central artery stiffness in our cohort and thus we think that our interpretation of the results is plausible. Lack of time to event data for the composite cardiovascular outcome meant that the evaluation of prognostic significance had to be performed by logistic regression analysis rather than survival analysis; however, given the low proportion of events, the loss of efficiency due to the use of logistic regression would be expected to have been very small.³⁷ Finally, all participants who were followed up over 3 years were older people with T2DM. Therefore, the findings may not be applicable to other populations.

We demonstrate that each reservoir-excess pressure parameter has a different association with T2DM or CVD. Furthermore, we present evidence that reservoir-excess pressure parameters independently predict the incidence of cardiovascular events during the 3-year follow-up period in individuals with T2DM. Conversely, conventional central and peripheral artery hemodynamic parameters did not demonstrate the prognostic utility in the same cohort. These results support the concept that the conduit artery pressure waveform contains clinically meaningful information for cardiovascular risk stratification and also suggest that reservoir-excess pressure analysis may provide an additional tool for cardiovascular risk stratification in individuals with T2DM, over and above conventional interpretation of pressure waveform.