Arterial Wave Reflections and Incident Cardiovascular Events and Heart Failure: The Multiethnic Study of Atherosclerosis

Abstract

Background

Experimental and physiologic data mechanistically implicate wave reflections in the pathogenesis of left ventricular failure and cardiovascular disease, but their association with these outcomes in the general population is unclear.

Objectives

To assess the relationship between central pressure profiles and incident cardiovascular events.

Methods

Aortic pressure waveforms were derived from a generalized transfer function applied to the radial pressure waveform recorded non-invasively from 5,960 participants in the Multiethnic Study of Atherosclerosis (MESA). The central pressure waveform was separated into forward and reflected waves using a physiologic flow waveform. Reflection magnitude (RM=[reflected/forward wave amplitude] ×100), augmentation index (AIx=[second/first systolic peak] ×100) and pulse pressure amplification (PPA=[radial/aortic pulse pressure] ×100) were assessed as predictors of cardiovascular events (CVE) and congestive heart failure (CHF) during median 7.61 years of follow-up.

Results

After adjustment for established risk factors, aortic AIx independently predicted hard CVE (HR per 10%-increase=1.08; 95%CI=1.01-1.14; P=0.016), whereas PPA independently predicted all CVE (HR per 10%-increase=0.82; 95%CI=0.70-0.96; P=0.012). RM was independently predictive of all CVE (hazard ratio [HR] per 10%-increase=1.34; 95%CI=1.08-1.67; P=0.009), hard CVE (HR per 10%-increase=1.46; 95%CI=1.12-1.90; P=0.006) and strongly predictive of new-onset CHF (HR per 10%-increase=2.69; 95%CI=1.79-4.04; P<0.0001), comparing favorably to other risk factors for CHF as judged by various measures of model performance, reclassification and discrimination. In a fully-adjusted model, compared to non-hypertensive subjects with low RM, the HR for hypertensive subjects with low RM, non-hypertensive subjects with high RM and hypertensive subjects with high RM were 1.81 (95%CI=0.85-3.86), 2.16 (95%CI=1.07-5.01) and 3.98 (95%CI=1.96-8.05), respectively.

Conclusions

Arterial wave reflections represent a novel strong risk factor for CHF in the general population.

Introduction

Several considerations support a mechanistic role for central pressure profiles as causal determinants of cardiovascular disease.(1-3) The aortic pressure profile is determined by the interactions between the left ventricle (LV) and the load imposed by the arterial tree.(4) Wave reflections arising in peripheral arteries and returning to the proximal aorta during mid-to-late systole are important determinants of LV afterload. Both animal and human studies have demonstrated that a loading sequence characterized by prominent late systolic load adversely impacts LV structure and function.(5-8)

Central pulse pressure and arterial wave reflections can be assessed non-invasively with arterial tonometry. The aortic augmentation index (AIx), which depends on the pressure difference between the first and second systolic peaks (Fig.1), has been a widely used surrogate of wave reflections. However, AIx is not only influenced by the magnitude of wave reflections but also confounded by their timing, heart rate, body height and other factors.(9,10) When central pressure and flow waveforms are known, the aortic pressure wave can be separated into its forward and reflected components (wave separation analysis), allowing for measurement of reflection magnitude (RM), defined as the dimensionless ratio of the amplitudes of backward/forward waves. This computation does not depend on the calibration of the flow waveform and can be approximated using pressure information only, assuming a triangular or a physiologic flow waveform (Fig.1).(11,12)

Figure 1.

An issue of great interest is whether central pressure parameters are associated with incident cardiovascular events (CVE) independently of brachial pressures. A recent meta-analysis(3) suggested that central AIx independently predicts cardiovascular events (CVE). In contrast, more recent data from 2,232 Framingham Heart study participants indicated that carotid AIx or pulse pressure did not independently predict major CVE.(13) This study did not perform wave separation analysis and relayed on brachial arterial tonometry for computation of pulse pressure amplification, an approach that has been challenged by some investigators based on anatomic factors that may impede proper applanation tonometry of the brachial artery.(14) In another recent study among 1,272 Taiwanese subjects, reflected wave amplitude computed with a triangular flow waveform, but not carotid AIx, predicted all-cause mortality assessed from a National Death Registry 15 years later.(15)

Given the adverse impact of wave reflections on the LV(5-7) and the large public health burden of heart failure and cardiovascular disease, further data regarding the association between wave reflections and these outcomes are needed. In this ancillary study of the multiethnic study of atherosclerosis (MESA), a large community-based multiethnic cohort study that enrolled adults free of cardiovascular disease,(16) we aimed to assess the relationship between central pressure profiles and: (1) Incident hard CVE; (2) Incident congestive heart failure (CHF).

METHODS

Study population

MESA enrolled 6,814 men and women aged 45-84 years who identified themselves as white, African-American, Hispanic, or Chinese and were free of clinically apparent cardiovascular disease, from six US communities between 2000-2002.(16) The study was approved by the institutional review boards of participating centers and participants gave informed consent.

Data collection

Standardized questionnaires were used to obtain information about cardiovascular risk factors and medication use. Resting blood pressure was measured in triplicate using a Dinamap-Pro100 oscillometric sphygmomanometer (GE Medical Systems; Waukesha,WI). The average of the last 2 measurements was used. Hypertension was defined as systolic blood pressure (SBP) ≥140 mmHg, diastolic blood pressure (DBP) ≥90 mmHg or antihypertensive medication use.(17) Serum total cholesterol, high-density-lipoprotein (HDL)-cholesterol, and glucose were measured after a 12-hour fast. Diabetes mellitus was defined as fasting glucose ≥126 mg/dL or hypoglycemic medication use.(18)

Hemodynamic measurements

Radial arterial waveform 30-second recordings were obtained at baseline in the supine position using the HDI/PulseWave-CR2000 tonometry device (Hypertension Diagnostics, Eagan,MN), digitized at 200Hz and exported for offline-processing using custom-designed software written in Matlab (The Mathworks; Natick,MA).

A generalized transfer function(19) was applied to the radial pressure waveform to obtain a central pressure waveform. Aortic-radial pulse pressure amplification (PPA) was computed as (radial pulse pressure/aortic pulse pressure) ×100 (Fig.1). This computation does not depend on calibration of the radial pressure waveform.

The first and second systolic peaks were identified in the aortic pressure waveform as previously described(20). AIx was computed as (second/first systolic peak) ×100 (Fig.1). A physiologic flow waveform(11) was used for separation of the pressure wave into forward and reflected waves.(21) RM was calculated as $\frac{\text{backward wave amplitude}}{\text{forward wave amplitude}}\times 100$ (Fig.1). Further details about our analysis methods can be found in the online supplemental section available at http://content.onlinejacc.org.

Event adjudication

In addition to 3 on-site examinations, a telephone interviewer contacted participants every 9-12 months to inquire about incident CVE. Two physicians independently reviewed copies of medical records and death certificates for hospitalizations and outpatient cardiovascular diagnoses, for blinded endpoint classification using pre-specified criteria(22) as summarized in the online supplemental section available at http://content.onlinejacc.org. The following endpoints were defined a priori for this study: (1) Hard CVE, defined myocardial infarction, resuscitated cardiac arrest, coronary heart disease death, stroke, or stroke death; (2) All CVE, defined as any hard CVE, angina, other atherosclerotic death, or other cardiovascular disease death; (3) CHF, which required clinical symptoms (e.g., dyspnea) or signs (e.g., edema), a physician CHF diagnosis and medical treatment for CHF in addition to: (a) pulmonary edema/congestion by chest X-ray and/or (b) dilated ventricle or poor LV function by echocardiography or ventriculography, or evidence of LV diastolic dysfunction. Whereas the 2 former are composite endpoints commonly used in cardiovascular risk studies, congestive heart failure was defined as a specific endpoint a priori based on previous experimental data.

Statistical analysis

A more detailed description of our statistical methods can be found in the online supplemental section available at http://content.onlinejacc.org. We examined the association between hemodynamic measures and time to CVE or CHF using the Kaplan-Meier method and Cox regression. Model goodness-of-fit was assessed with the Akaike’s information criterion (AIC) and Bayesian information criterion (BIC).(23,24) Model discrimination was assessed with the Harrel’s c-index (which is analogous to the area under the receiver-operator-characteristic curve).(23,25) Calibration was assessed with the Hosmer-Lemeshow test. Improvements in subject reclassification was further assessed using the net reclassification improvement (NRI),(23,25) which depends on the increased probability that a new model will categorize case subjects as higher-risk and decreased probability that it will categorize control subjects as lower-risk, compared to a base model, as explained in more detail in the online supplemental section. We computed 2 versions of the NRI: a category-based NRI based on usual categories for 10-year coronary heart disease risk (adapted at 5 years as <2.5%, 2.5-<5%, 5-<10% and ≥10%). Since no established categories exist that guide clinical decisions for CHF risk, we computed category-free reclassification measures, which are independent on arbitrarily defined risk thresholds(26). These include the category-free NRI and the relative integrated discrimination improvement (rIDI), which expresses the relative improvement in discrimination slopes (difference in mean predicted probabilities between case and control participants) between the base model and new model.(23,25-27) Various indices of model performance were used to: (1) Assess the added predictive value of central pressure indices; (2) Compare the predictive value of aortic hemodynamic indices to that of well-established risk factors, as previously described.(23)

All tests were 2-sided with α=0.05. Analyses were performed using SPSS v19 (SPSS Inc., Chicago,IL).

RESULTS

Of 6,336 participants who underwent radial tonometry, 6153 (97.1%) had technically-adequate data. Central waveforms from 164 subjects had no discernible inflections (due to merging of the first and second peaks) or more than one inflection, impeding adequate identification of the first and second systolic peaks. Twenty-nine participants had no follow-up information, leaving 5,960 participants in the analysis. Table 1 shows baseline characteristics of subjects included in this study.

Table 1. Baseline characteristics of study participants (n=5,960).

Characteristic	Median (IQR) or Count (percentage)
Number of events
Heart Failure	104 (1.7)
Myocardial Infarction	112 (l.9)
Hard coronary heart disease events	148 (2.5)
All coronary heart disease events	258 (4.3)
Stroke	100 (1.7)
Any hard cardiovascular event	241 (4.0)
Any cardiovascular event	356 (6.0)
Age, Years	62 (53-70)
Sex
Male	2862 (48)
Female	3096 (52)
Ethnicity
White	2240 (37.6)
African American	1620 (27.2)
Chinese-American	728 (12.2)
Hispanic-American	1370 (23.0)
Body height, cm	166 (159-174)
Body weight, kg	77.1 (66.2-89.4)
Body Mass Index, kg/m2	27.5 (24.6-31.2)
Total Cholesterol, mg/dL	192 (171-215)
LDL-Cholesterol, mg/dL	116 (96-136)
HDL-Cholesterol, mg/dL	48 (40-59)
Triglycerides, mg/dL	112 (78-162)
Diabetes Mellitus	755 (12.7)
Current Smoking	2151 (36.1)
Hypertension	2667 (44.8)
Estimated Glomerular Filtration Rate, mL·min−1 1.73 m−2	80 (70-92)
Hypertension medication use	2207 (37)
Brachial SBP, mmHg	124 (111-140)
Brachial DBP, mmHg	72 (65-78.5)
Pulse Pressure amplification, %	1.10 (1.05-1.17)
Aortic Augmentation index, %	145 (135-159)
Reflection Magnitude, %	84 (81-87)
Heart rate, bpm	63 (57-70)

Cardiovascular events

During a median follow-up of 7.61 (interquartile range: 7.34-7.78) years, 407 subjects experienced a first CVE, 281 subjects experienced a first hard CVE and 117 experienced a first episode of CHF.

Hazard ratios for incident CVE and CHF associated with a 10-point increase in AIx, RM or PPA are shown in Table 2. Table 2 also shows standardized hazard ratios (those associated with 1 standard deviation increase in the predictors). In adjusted analyses, AIx was independently associated with hard CVE (model 3, Table 2; HR per 10%-increase=1.08; 95%CI=1.01-1.14; P=0.016). The category-free NRI, relative IDI and increase in c-index achieved by adding PPA to the other variables in this model were 0.036, 0.002 and 0.004, respectively. PPA was independently associated with all CVE (model 3, Table 2; HR per 10%-increase=0.82; 95%CI=0.70-0.96; P=0.012). The category-free NRI, relative IDI and increase in c-index achieved by adding PPA to the other variables in this model were 0.10, 0.02 and 0.002, respectively. PPA and AIx were not independently associated with incident CHF.

Table 2. Results of Cox proportional hazards models examining the relationship between hemodynamic variables at baseline and the risk of cardiovascular events and heart failure during follow-up.

	Crude (n=5960)			Adjusted model 1 (n=5937) *			Adjusted Model 2 (n=5934)*
	Hazard Ratio (95%CI) per 10% increase	Standardized Hazard Ratio (95%CI)	P value	Hazard Ratio (95%CI) per 10% increase	Standardized Hazard Ratio (95%CI)	P value	Hazard Ratio (95%CI) per 10% increase	Standardized Hazard Ratio (95%CI)	P value
Reflection Magnitude †
All CVE	1.48 (1.2-1.81)	1.21 (1.09-1.34)	<0.0001	1.38 (1.11-1.71)	1.17 (1.05-1.30)	0.004	1.34 (1.08-1.67)	1.15 (1.04-1.28)	0.009
Hard CVE	1.66 (1.29-2.13)	1.28 (1.13-1.45)	<0.0001	1.52 (1.17-1.98)	1.23 (1.08-1.40)	0.002	1.46 (1.12-1.90)	1.20 (1.06-1.37)	0.006
Heart Failure	2.61 (1.75-3.88)	1.60 (1.32-1.94)	<0.0001	2.75 (1.83-4.12)	1.64 (1.35-2.00)	<0.0001	2.69 (1.79-4.04)	1.62 (1.33-1.98)	<0.0001
Augmentation Index ‡
All CVE	1.05 (1.004-1.09)	1.11 (1.01-1.21)	0.031	1.03 (0.99-1.08)	1.07 (0.98-1.19)	0.19	1.05 (1.00-1.11)	1.12 (1.00-1.25)	0.052
Hard CVE	1.07 (1.02-1.13)	1.17 (1.05-1.30)	0.005	1.05 (0.99-1.11)	1.12 (0.99-1.26)	0.058	1.08 (1.01-1.14)	1.17 (1.03-1.33)	0.016
Heart Failure	1.06 (0.98-1.14)	1.14 (0.97-1.34)	0.14	1.03 (0.95-1.12)	1.07 (0.89-1.28)	0.50	1.08 (0.99-1.17)	1.17 (0.97-1.42)	0.11
Pulse Pressure Amplification §
All CVE	0.92 (0.82-1.02)	0.92 (0.83-1.02)	0.09	0.98 (0.88-1.10)	0.99 (0.89-1.10)	0.78	0.82 (0.70-0.96)	0.82 (0.71-0.96)	0.012
Hard CVE	0.92 (0.81-1.04)	0.92 (0.81-1.04)	0.19	1.02 (0.89-1.16)	1.02 (0.90-1.15)	0.81	0.87 (0.72-1.05)	0.87 (0.73-1.05)	0.15
Heart Failure	0.91 (0.75-1.11)	0.92 (0.76-1.11)	0.37	1.05 (0.86-1.28)	1.04 (0.86-1.27)	0.67	0.82 (0.61-1.1)	0.82 (0.62-1.09)	0.18

^*Adjusted model 1 includes age, gender, total cholesterol, HDL-cholesterol, smoking, systolic and diastolic blood pressure, diabetes mellitus. Adjusted model 2 further adjusts for ethnicity, body height, body weight, antihypertensive medication use, heart rate and estimated glomerular filtration rate. Complete covariate data was available from 5,934 subjects.

^†Reflection magnitude = (reflected wave amplitude/forward wave amplitude) × 100.

^‡Augmentation index = (second systolic peak /first systolic peaks) × 100.

^§Pulse Pressure Amplification = (radial pulse pressure/aortic pulse pressure) × 100.

Hazard ratios correspond to a 10%-increase (or a 10-point increase in the indices shown between parentheses in the formulas above).

RM was independently associated with incident CVE. After adjustment for age, ethnicity, gender, SBP, DBP, antihypertensive medication use, height, weight, diabetes mellitus, total cholesterol, HDL-cholesterol, current smoking, heart rate and glomerular filtration rate, a 10%-increase in RM was predictive of a higher risk of CVE (HR=1.34; 95%CI=1.08-1.67; P=0.009), hard CVE (HR per 10%-increase=1.46; 95%CI=1.12-1.90; P=0.006). The category-free NRI, relative IDI and increase in c-index for the prediction of all CVE achieved by adding RM to the other variables shown in model 3 (Table 2) were 0.15 and 0.05 and 0.002 respectively. The category-free NRI, relative IDI and increase in c-index for the prediction of hard CVE when RM was added to the other variables shown in model 3 (Table 2) were 0.13, 0.08 and 0.002 respectively.

RM was strongly predictive of incident CHF (HR per 10%-increase=2.69; 95%CI=1.79-4.04; P<0.0001; Table 2). Table 3 shows independent predictors of incident CHF in a fully adjusted model along with standardized hazard ratios for each term, in order to allow an easier comparison between various predictors. The full model is similar to the one used in Table 2 except for height and weight, which was replaced by body mass index to more intuitively assess the independent contribution of obesity to CHF risk prediction in the model. Table 3 also shows improvements in model performance observed when individual predictors were added to a model containing all other variables except the predictor in question. RM was associated with a standardized hazard ratio of 1.61 (95%CI=1/32-1.96), the largest Wald statistic and the greatest decrease in AIC and BIC (indicating improvement in model fit) and the greatest increase in IDI and rIDI. With the exception of age, a non-modifiable risk factor, RM was associated with the greatest NRI. Of note, these improvements in model performance were also superior to changes induced by adding SBP and DBP together.

Table 3. Predictors of incident heart failure in multivariate analysis (n=5934)^†.

Full model with adjusted Hazard Ratios (c-index=0.802; AIC=1893; BIC=1943)				Effects of adding individual predictors to a model containing all other variables
	Standardized Hazard Ratio (95%CI)	Wald Statistic	P Value	Change in BIC†	Change in AIC†	Change in c-index ‡	NRI ‡	IDI ‡	rIDI ‡
Age (10-years)	1.62 (1.26-2.08)	14.44	<0.0001	−10.10	−12.87	0.020	0.47 **	0.010 **	0.22 **
Male Gender	1.74 (1.38-2.21)	21.37	<0.0001	−17.09	−19.85	0.015	0.34 **	0.017 **	0.44 **
Body Mass Index (10 kg/m2)	1.26 (1.03-1.55)	4.83	0.028	0.15	−2.62	0.007	0.32**	0.002 NS	0.050 NS
Diabetes Mellitus	1.24 (1.07-1.44)	8.37	0.004	−3.09	−5.86	0.010	0.019 NS	0.003 NS	0.061 NS
Systolic BP (10 mmHg)	1.69 (1.33-2.13)	18.97	<0.0001	−13.10	−15.86	0.013	0.31 **	0.011 **	0.25 **
Diastolic BP (10 mmHg)	0.67 (0.52-0.86)	9.71	0.002	−4.94	−7.70	0.006	0.15 **	0.007 **	0.14 **
Reflection Magnitude (10%)	1.61 (1.32-1.96)	22.03	<0.0001	−17.79	−20.55	0.011	0.38 **	0.018	0.48 **
Systolic and diastolic BP added together §	---	---	---	−8.46	−13.98	0.013	0.28**	0.011**	0.25**

AIC=Akaike information criterion; BIC=Bayesian information criterion; NRI=net reclassification improvement (category-free); IDI=integrated discrimination improvement; rIDI=relative integrated discrimination improvement.

^*Only significant predictors of CHF are shown. However, all models are also adjusted for ethnicity, antihypertensive medication use, total cholesterol, HDL-cholesterol, current smoking, heart rate and estimated glomerular filtration rate. All hazard ratios are standardized.

^†Larger decreases (changes with negative sign) indicate a larger improvement in model fit.

^‡Larger increases indicate a larger improvement in performance in reclassification or discrimination.

^§This row presents improvements in model performance when both SBP and DBP are added to a model containing all other variables contained in the full model.

^NSP>0.05;

^**P<0.05 for NRI, IDI or rIDI<>0

The addition of RM to a model containing all other variables shown in Table 3 resulted in a category-free NRI of 0.38 and a rIDI of 0.48, indicating a 48%-relative increase in the discrimination slope achieved by all variables in the base model. The addition of RM to the model resulted in a category-based NRI of 0.17. A category-dependent reclassification table can be found in the online supplemental section available at http://content.onlinejacc.org (supplemental Table 1). The full model containing RM demonstrated adequate calibration (Hosmer-Lemeshow χ²=8.50; P=0.49).

Finally, in order to more directly compare the value of brachial pulse pressure vs. RM as predictors of CHF, RM or brachial pulse were separately added to a base model that included age, gender, body mass index, diabetes mellitus, ethnicity, antihypertensive medication use, total cholesterol, HDL-cholesterol, current smoking, heart rate and estimated glomerular filtration rate. When RM was added to this base model, the standardized HR for RM was 1.54 (1.26-1.88; P<0.0001), with an achieved NRI, relative IDI and increase in c-index of 0.32, 0.42 and 0.011, respectively. When PP was added to this base model, the standardized HR for RM was 1.42 (1.18-1.70; P<0.0001), with an achieved NRI, relative IDI and increase in c-index of 0.27, 0.20 and 0.011, respectively.

Figure-2A shows cumulative hazard curves for incident CHF among subjects stratified by RM tertiles. There was a progressive increase in CHF risk with increasing RM tertile (P<0.0001). In order to further compare the risk of CHF associated with hypertension versus increased RM, we stratified the population according to the presence (prevalence=45%) or absence of hypertension. We also stratified the population into those above or below the 55th RM percentile to obtain a prevalence of “high” RM identical to the prevalence of hypertension (45%), allowing a more direct comparison of the contribution of each factor to CHF risk in the population. Figures 2B-2C show adjusted cumulative hazard functions for each sub-stratum according to the presence or absence of hypertension and high RM. As shown, non-hypertensive subjects with low RM were at lowest risk, hypertensive subjects with high RM were at highest risk, whereas hypertensive subjects with low RM and non-hypertensive subjects with high RM were at intermediate risk for incident CHF.

Figure 2.

DISCUSSION

Our study identifies an important novel independent risk factor for new-onset CHF among adults in the general population who are free of clinically apparent cardiovascular disease. In our large multiethnic sample, the magnitude of arterial wave reflections assessed non-invasively via radial arterial tonometry was a strong predictor of incident CHF. RM was also a significant predictor of incident CVE, but its association with this composite endpoint was not as strong as its association with CHF.

Physiologic principles(28) and experimental studies(4-6) directly implicate wave reflections in the pathogenesis of LV failure, but this is the first study to assess their association with the risk of incident CHF in humans. The pressure (and flow) wave generated by the LV (forward wave) is transmitted by conduit vessels and partially reflected at sites of impedance mismatch, such as points of branching or change in wall diameter or material properties along the arterial tree as well as at the junction of small conduit arteries with high resistance arterioles in which mean pressure falls precipitously.(4,29) Multiple small reflections travel back to the proximal aorta and merge into a “net” reflected wave. It has been long known that arterial wave reflections profoundly affect LV afterload.(4,28-30) Due to the finite wave transit time from the heart to reflection sites and back to the proximal aorta, in adults beyond youth wave reflections typically increase LV afterload in mid-to-late systole.(4,29) For any given level of SBP, a pattern characterized by prominent late-systolic load has been unequivocally demonstrated to exert deleterious effects on LV structure and function in animal models,(4-6) observations that have been supported by human studies.(7,8) Consistent with these mechanistic data, our study indicates that wave reflections are important predictors of CHF risk. Wave reflections are thought to be generated predominantly near muscular (medium-sized) and smaller arterial segments and are highly modifiable. Vasoactive drugs such as organic nitrates profoundly decrease wave reflections despite small effects on brachial pressures.(31) Therefore, our findings not only have implications for CHF risk assessment but may also identify a potential novel therapeutic target for CHF prevention. However, we note that our study does not prove that this is a suitable therapeutic strategy, and this issue will need to be addressed in appropriately designed clinical trials.

Most previous studies assessing the association between central pressures and incident CVE included populations with manifest cardiovascular or renal disease.(3) Studies in the general population (such as the Strong Heart Study(32) or Framingham Heart Study(13)) have been restricted to a single ethnic group and examined composite CVE endpoints. Furthermore, previous studies assessed central pulse pressure and AIx, rather than RM assessed with wave separation analysis. Several studies suggest that AIx is predictive of composite CVE endpoints among subjects with established cardiac or renal disease.(3) However, central AIx and pulse pressure did not predict CVE independently of brachial pressures among 2,232 participants enrolled in the Framingham Heart Study. In our study, the adjusted relationship between AIx/PPA and incident CVE was not consistent. Although AIx was an independent predictor of hard CVE, whereas PPA was an independent predictor of all CVE in multivariate analyses, knowledge of these variables yielded negligible increases in reclassification and discrimination for these endpoints. Unlike RM, PPA/AIx were not predictive of incident CHF. We emphasize that these findings do not indicate that central pulse pressure is irrelevant for cardiovascular risk, but rather indicate that, when considering the broad distribution of brachial pressures within a population such as ours, knowledge of the smaller difference between peripheral and central pulse pressure (PPA) does not provide substantial incremental risk prediction for hard CVE/CHF.

The different predictive performance of AIx compared to RM has several potential explanations. AIx is not only influenced by RM but also strongly dependent on reflection timing, heart rate, body height and hemodynamic phenomena unrelated to wave reflections.(4,9,10,33) Computation of AIx is dependent on the accuracy of high-frequency components of measured pressure (which produce sharp inflections in the waveform), which may be less accurately reproduced with a generalized transfer function due to their higher inter-individual variability.(34) In contrast, RM is less sensitive to confounding factors listed above and its computation depends predominantly on low-frequency pressure harmonics, which relate more consistently between the aortic and radial sites among different individuals(34) and may therefore be better reproduced with a generalized transfer function.

Our study is the largest to date to assess the association between central pressure profiles and incident CVE. Other strengths of our investigation include the multiethnic community-based sample, standardized assessments and careful event adjudication using hard criteria for CHF/CVE. However, it is important to acknowledge some limitations. Our observational study cannot prove a causal link between wave reflections and CHF/CVE. We did not measure central flow but assumed a physiologic flow waveform, an imperfect approach that, although better related to incident CHF/CVE than AIx, provides only an approximation of true RM. This raises the possibility that the predictive ability of RM may be improved if more accurately measured with subject-specific flow waveform data. Since participants had no known cardiovascular disease at baseline, this cohort represents a particularly healthy sample of the population at large which is, however, ideal for examining early vascular changes predisposing to CHF. As a consequence, the absolute risk of CHF in this cohort was relatively low and the observed number of events was insufficient for gender-specific sub-analyses, which however should be feasible in the future as more events accumulate in the cohort. Despite our large, multiethnic sample, it would be desirable to replicate our findings in other populations to better establish their generalizability. Finally, we did not measure aortic pulse wave velocity, an index of aortic wall stiffness, which also impacts LV afterload. However, we note that aortic pulse wave velocity affects predominantly the timing (rather than the magnitude) of wave reflections and that the LV senses only operating load and not large arterial material properties per se. It is possible that aortic pulse wave velocity and RM provide complementary information about cardiovascular risk. This should be addressed in future studies.

In summary, in an ethnically-diverse population free of cardiovascular disease at baseline, RM was independently associated with incident CVE and strongly associated with incident CHF. Arterial wave reflections represent an important novel risk factor for CHF and a potential therapeutic target for primary CHF prevention.

Abbreviations

LV

Left Ventricle

AIx

Augmentation Index

CVE

Cardiovascular Events

CHF

Congestive Heart Failure

SBP

Systolic Blood Pressure

DBP

Diastolic Blood Pressure

PPA

Pulse Pressure Amplification

AIC

Akaike’s information criterion

BIC

Bayesian information criterion

Net reclassification improvement

Net Reclassification Improvement

rIDI

Relative integrated discrimination improvement

RM

Reflection Magnitude