Higher cardiac vagal activity predicts lower peripheral resistance 6 years later in European but not African Americans

Abstract

African American (AA) individuals are at a greater risk for the development of cardiovascular complications, such as hypertension, compared with European Americans (EAs). Higher vagally mediated heart rate variability (HRV) is typically associated with lower blood pressure (BP) and total peripheral resistance (TPR). However, research has yet to examine the differential impact of HRV on longitudinal hemodynamic activity between AAs and EAs. We sought to rectify this in a sample of 385 normotensive youths (207 AAs, 178 EAs; mean age 23.16 ± 2.9 yr). Individuals participated in two laboratory evaluations spanning approximately 6 yr. Bioimpedance was used to assess HRV at time 1 and cardiac output at both time 1 and time 2. Mean arterial pressure (MAP) was measured at both time points via an automated BP machine. TPR was calculated as MAP divided by cardiac output. Results showed AAs to have higher BP and higher TPR at time 2 compared with EAs, independent of several important covariates. Also, higher HRV at time 1 significantly predicted both lower TPR and BP at time 2 among EAs only; these associations were attenuated and not significant in AAs. HRV did not significantly predict cardiac output at time 2 in the full sample or split by ethnicity. Our findings highlight that AAs show TPR mediated long-term increases in BP irrespective of resting HRV, providing a physiological pathway linking AAs with a greater risk for mortality and morbidity from hypertension and potentially other cardiovascular disease.

NEW & NEWSWORTHY African Americans and European Americans differ in hemodynamics underlying long-term blood pressure regulation. Over 6 yr, African Americans show total peripheral resistance-mediated increases in blood pressure compared with European Americans. Higher heart rate variability predicts lower blood pressure and total peripheral resistance 6 yr later in European Americans but not African Americans.

INTRODUCTION

African American (AA) individuals are at a greater risk for the development of cardiovascular complications, such as hypertension, compared with European Americans (EAs) (1). Elevated blood pressure (BP) is a risk factor for increased cardiovascular morbidity and mortality. Compared with EAs, AAs typically show higher resting blood pressure (2) marked by higher total peripheral resistance (TPR) (3). TPR-mediated increases in BP are particularly deleterious, as elevated TPR is associated with end-organ damage. However paradoxically, AAs also show higher resting vagally mediated heart rate variability (hereafter simply called HRV) (4). Given the cardioprotective nature of higher HRV (5), this paradoxical pattern of increased resting α-adrenergic sympathetic and parasympathetic activity in AAs has been termed the “cardiovascular conundrum” (3, 4). Interestingly, studies have shown that greater doses of acetylcholine are needed for AAs to produce the same degree of vasodilation as in EAs (6, 7). Another study showed AAs to have increased vascular α-adrenergic responsivity and suggest increased TPR is more easily achieved in AAs compared with EAs (8). Finally, pregnancy is a potent vasodilator; we recently showed pregnant AAs had greater TPR compared with pregnant EAs (9). These findings serve as direct evidence for impaired vasodilation in AAs relative to EAs.

There is growing interest in the role of the vagus nerve in cardiovascular disease (10, 11). Heart rate variability is a widely accepted method to assess autonomic activity with high-frequency HRV used to index vagally mediated HRV. We have recently shown that HRV was prospectively associated with both systolic BP and diastolic BP (12). It has also been reported that HRV is positively associated with endothelial function as indexed by brachial artery flow-mediated dilation in humans (13). The role of the vagus nerve in regulating heart rate (HR) and cardiac contractility (e.g., stroke volume), and thus cardiac output, has been extensively documented (14). Whereas the vagal innervation of the heart is well established, there is no evidence of direct vagal innervation of the vessels. Thus, the mechanism by which vagal activity might regulate BP and endothelial function and how this might differ between EAs and AAs is not established.

Therefore, research aimed at understanding the differential impact of vagal activity on longitudinal hemodynamic activity between AAs and EAs is warranted. In the present study, we sought to understand how vagal activity, as indexed by HRV, may predict hemodynamics underlying “long-term” blood pressure regulation differentially as a function of ethnicity. Mean arterial pressure (MAP) is determined by the product of cardiac output (CO) and TPR (MAP = CO × TPR). Therefore, although MAP typically increases with age, the hemodynamics (TPR or CO) underlying higher MAP may differ between AAs and EAs. Using secondary analysis of archival data from an empirical prospective study, we investigated hemodynamic patterns over the course of approximately 6 yr, with physiological measures taken at two time points: 1) baseline/initial visit (time 1); and 2) follow-up visit on an average of 6 yr later (time 2).

To our knowledge, this will be the first study to investigate if HRV may predict long-term hemodynamic activity differentially based on ethnicity. We aim to highlight how the impact of vagal activity on long-term blood pressure regulation may differ between AAs and EAs. Such results would have important implications for the course of elevated blood pressure in AA individuals.

METHOD

Participants and Experimental Design

In this study, 385 normotensive youths were enrolled in the Augusta Heart Study (15–17). Participants were initially between the ages of 15–32 yr old (mean age 23.16 ± 2.9 yr). There were 207 AAs (127 women) and 178 EAs (83 women). Individuals were participants in longitudinal studies of the development of cardiovascular risk factors in children with verified family histories of cardiovascular diseases (i.e., essential hypertension and/or premature myocardial infarction) (18–19). All individuals participated in two separate laboratory evaluations separated by an average of 6.32 yr (minimum of 2.1 yr, maximum of 8.2 yr). Smoking, medication histories, parents’ education levels, and physical activity were obtained via participant interview.

Subject’s height (in centimeters) and weight (in kilograms) were measured without shoes with a Health-O-Meter medical scale, which was calibrated daily. Body mass index (BMI) was calculated as weight/height². The study was approved by the institutional review committee, and informed consent was obtained from all participants on each study visit. All subjects were escorted to a quiet temperature-controlled room (20°C–22°C) and fitted with equipment for recording BP (Dinamap model 1846 SX, Critikon Inc.) as previously described (18–19). Blood pressure was assessed via the same arm for each participant both at baseline and the follow-up. The subject was placed in a supine position and given standardized instructions to relax as completely as possible for 15 min. Three hemodynamic measurements (at min 11, 13, and 15) throughout the 15-min baseline period were obtained and the final two measurements (at min 13 and 15) averaged to provide one reading of blood pressure per evaluation to derive mean arterial pressure (MAP; mmHG).

Heart rate (HR), cardiac output (CO; l/min), and stroke volume (SV; ml) were recorded using a BioZ impedance monitor (Cardio-Dynamics, San Diego, CA) as previously described (18–19). The raw RR interval data were first screened using the following two criteria: 1) RR intervals between 300 and 2,000 ms and 2) successive RR-interval ratios between 0.8 and 1.2. Kubios; HRV analysis software was used to generate HRV parameters from the recorded RR intervals (20). For the present study, we employed a frequency domain measure, high-frequency (HF) power (defined as the power between 0.15 and 0.40 Hz, using Fast Fourier Transformation), for all analyses¹. HF-HRV is a valid measure of predominantly parasympathetic (i.e., vagal) cardiac influence (21). To account for skewness, HRV was transformed using a natural logarithm. HRV measures were available at time 1 only; hemodynamic measures assessed via bio-impedance were available for both time points. Total peripheral resistance (TPR) was calculated as MAP divided by CO. Cardiac index (CI) and total peripheral index (TPI) we calculated by dividing CO and TPR by body surface area, respectively. Body surface area was calculated using the Du Bois and Du Bois (21a) calculation: body surface area = 0.007184 × weight^0.425 × height^0.725. CI and TPI were used as the primary measures of CO and TPR, respectively.

Statistical Analyses

Independent sample t tests were used to determine group differences in demographic and cardiovascular variables at time 1 (T1) and time 2 (T2) as a function of ethnicity independently. In addition, independent sample t tests were used to test the change in MAP, TPI, and CI between T1 and T2. To examine our primary question, a series of hierarchical regression models were conducted to examine whether HRV predicted MAP, CO, and TPR from T1 to T2 both in the whole sample and stratified by ethnicity. Age, sex, ethnicity, BMI, father’s education (proxy measure for socioeconomic status for our relatively young sample), and smoking were entered in model 1. For respective models, T1 MAP, CO, and TPR were added in the following step (model 2), followed by a final step including T1 HRV (model 3). Dependent measures at T2 included MAP, CO, and TPR. All analyses were conducted using SPSS (v. 20, IBM Armonk, NY); significance was set at P < 0.05. Unstandardized βs (b) and standard errors (in brackets), partial r coefficients, and P values are provided in the text for primarily associations. Standardized βs (b), t values, P values, zero-order correlations, and partial r correlations for all variables in all regression models are presented in Tables 2–4.

RESULTS

Means and standard deviations for the full sample, and stratified by ethnicity, are presented in Table 1. At both T1 and T2, AAs were significantly older by ∼6 mo compared with EAs (each P < 0.01) and had higher BMI (each P < 0.01), higher MAP (each P < 0.001), TPR (each P < 0.01), and TPI (each P = 0.001). AAs reported significantly lower father’s education (P < 0.001) at T1 compared with EAs. AAs also had significantly lower CO, CI, and SV at T2 (each P < 0.01). Finally, the increase in MAP and TPI from T1 to T2 was significantly greater in AAs compared with EAs (each P < 0.05). The increase in CI from T1 to T2 was significantly greater in EAs compared to AAs (P = 0.037).

Table 1.

Sample characteristics and ethnic comparisons

	Full Sample		EAs (n = 178)		AAs (n = 207)
	M	SD	M	SD	M	SD	P
Age, yr T1	23.16	2.85	22.66	2.88	23.58	2.77	0.001
Age, yr T2	29.47	2.96	28.93	2.94	29.93	2.91	0.001
Body mass index, kg/m2 T1	28.77	8.45	27.06	6.77	30.23	9.44	<0.001
Body mass index, kg/m2 T2	30.93	8.74	29.30	7.55	32.33	9.44	0.001
Father’s education, yr) T1	13.45	2.37	14.05	2.44	12.93	2.18	<0.001
Smoking T1	0.35	0.48	0.38	0.49	0.32	0.47	0.195
Physical activity T1	0.14	0.34	0.14	0.35	0.13	0.34	0.758
HF-HRV T1	7.17	1.06	7.13	1.11	7.21	1.00	0.475
MAP T1	81.63	8.40	78.95	7.26	83.94	8.64	<0.001
TPR T1	16.82	4.69	16.03	4.09	17.49	5.06	0.002
TPI T1	32.16	8.06	30.35	6.65	33.73	8.83	<0.001
CO T1	5.20	1.36	5.24	1.33	5.16	1.39	0.596
CI T1	2.67	0.58	2.71	0.55	2.63	0.61	0.184
HR T1	64.64	9.40	63.90	9.06	65.27	9.66	0.154
SV T1	80.97	20.07	82.31	18.97	79.82	20.95	0.226
MAP T2	86.30	10.39	82.60	8.95	89.48	10.50	<0.001
TPR T2	16.25	4.13	14.79	3.32	17.49	4.34	<0.001
TPI T2	32.24	8.10	29.18	6.22	34.87	8.61	<0.001
CO T2	5.60	1.34	5.82	1.26	5.42	1.38	0.003
CI T2	2.79	0.54	2.92	0.49	2.69	0.56	<0.001
HR T2	64.60	9.48	63.93	9.86	65.17	9.13	0.199
SV T2	87.88	21.16	92.26	20.47	84.11	21.06	<0.001
MAPΔ	4.66	8.88	3.65	8.83	5.54	8.85	0.024
TPIΔ	0.08	10.01	−1.16	8.31	1.15	11.17	0.043
CIΔ	0.13	0.72	0.21	0.66	0.06	0.77	0.037

This table depicts sample characteristics for the full sample (n = 385) and stratified by ethnicity. Means and standard deviations are presented on variables at both time 1 (subscript denoted as T1) and time 2 (subscript denoted as T2). Significant differences between African Americans (AAs) and European Americans (EAs) were evaluated and corresponding P values are presented. CO, cardiac output; CI, cardiac index; HF-HRV, high frequency heart rate variability; HR, heart rate; MAP, mean arterial pressure; SV, stroke volume; TPR, total peripheral resistance; TPI, total peripheral index. Δ Change. Significant P values (P < 0.05) are bold.

Predicting Mean Arterial Pressure at Time 2

Regression results for MAP at T2 are presented in Table 2A for the full sample. Model 1 accounted for 19.1% of the variance in T2 MAP with age, sex, and ethnicity emerging as significant predictors (each P < 0.01). In model 2, T1 MAP significantly accounted for an additional 19.1% of the variance [B = 0.60 (0.06), r_partial = 0.486, P < 0.001]. In model 3, HRV emerged as a significant inverse predictor [B = −1.34 (0.40), r_partial = −0.170, P = 0.001], explaining an additional 1.8% of the variance in T2 MAP.

Table 2.

Hierarchical regression models predicting time 2 mean arterial pressure for full sample and stratified by ethnicity

	b	t	P	r	rpartial	b	t	P	r	rpartial	b	t	P	r	rpartial
A: Full sample	Model 1 (R2 = 0.191)					Model 2 (R2change = 0.191)					Model 3 (R2change = 0.018)
Age T1	0.148	3.04	0.003	0.218	0.154	0.053	1.22	0.223	0.218	0.063	0.033	0.75	0.452	0.218	0.039
Sex	−0.208	−4.32	<0.001	−0.146	−0.217	−0.106	−2.46	0.014	−0.146	−0.126	−0.111	−2.59	0.010	−0.146	−0.133
Ethnicity	0.315	6.47	<0.001	0.330	0.316	0.168	3.76	<0.001	0.330	0.190	0.181	4.08	<0.001	0.330	0.206
BMI T1	0.094	1.88	0.061	0.170	0.096	0.123	2.81	0.005	0.170	0.143	0.119	2.76	0.006	0.170	0.141
Father’s education T1	−0.032	−0.66	0.511	−0.147	−0.034	−0.015	−0.34	0.734	−0.147	−0.017	−0.015	−0.36	0.719	−0.147	−0.019
Smoking T1	0.049	1.02	0.307	0.078	0.053	0.011	0.25	0.800	0.078	0.013	0.005	0.13	0.899	0.078	0.007
MAP T1						0.484	10.81	<0.001	0.571	0.486	0.472	10.66	<0.001	0.571	0.482
HF-HRV T1											−0.137	−3.35	0.001	−0.184	−0.170
B: EAs	Model 1 (R2 = 0.092)					Model 2 (R2change = 0.140)					Model 3 (R2change = 0.039)
Age T1	0.030	0.38	0.703	0.078	0.029	−0.032	−0.44	0.657	0.078	−0.034	−0.055	−0.76	0.446	0.078	−0.058
Sex	−0.222	−2.96	0.003	−0.184	−0.220	−0.086	−1.17	0.243	−0.184	−0.089	−0.094	−1.31	0.193	−0.184	−0.100
BMI T1	0.203	2.52	0.013	0.195	0.189	0.224	3.01	0.003	0.195	0.224	0.207	2.83	0.005	0.195	0.212
Father’s education T1	−0.056	−0.73	0.464	−0.101	−0.056	−0.040	−0.56	0.573	−0.101	−0.043	−0.052	−0.76	0.447	−0.101	−0.058
Smoking T1	0.036	0.49	0.623	0.091	0.037	0.005	0.07	0.947	0.091	0.005	−0.009	−0.13	0.894	0.091	−0.010
MAP T1						0.403	5.58	<0.001	0.422	0.393	0.394	5.57	<0.001	0.422	0.393
HF-HRV T1											−0.201	−3.01	0.003	−0.232	−0.225
C: AAs	Model 1 (R2 = 0.116)					Model 2 (R2change = 0.266)					Model 3 (R2change = 0.008)
Age T1	0.229	3.35	0.001	0.254	0.230	0.091	1.55	0.123	0.254	0.109	0.075	1.26	0.208	0.254	0.089
Sex	−0.211	−3.06	0.003	−0.229	−0.211	−0.148	−2.53	0.012	−0.229	−0.176	−0.149	−2.57	0.011	−0.229	−0.179
BMI T1	0.052	0.76	0.451	0.077	0.053	0.090	1.56	0.120	0.077	0.110	0.091	1.58	0.116	0.077	0.111
Father’s education T1	−0.005	−0.08	0.940	−0.057	−0.005	0.022	0.38	0.705	−0.057	0.027	0.026	0.46	0.647	−0.057	0.032
Smoking T1	0.048	0.70	0.482	0.119	0.050	0.000	0.00	0.999	0.119	0.000	0.000	0.00	0.999	0.119	0.000
MAP T1						0.544	9.28	<0.001	0.588	0.549	0.534	9.10	<0.001	0.588	0.542
HF-HRV T1											−0.093	−1.63	0.105	−0.192	−0.115

This table depicts hierarchical regression models predicting mean arterial pressure (MAP) at time 2 (T2) in all individuals. Table 2A give regression statistics for the full sample, and Table 2, B and C present regression statistics in European Americans (EAs) and African Americans (AAs). Age is measured in years. BMI, body mass index, kg/m²; HF-HRV, natural log transformed high frequency heart rate variability; T1, time 1. Significant predictors (P < 0.05) are bold at each step.

Regression results for MAP at T2 are presented in Table 2B for EAs. Model 1 accounted for 9.2% of the variance in T2 MAP with sex and BMI emerging as significant predictors (each P < 0.01). In model 2, T1 MAP significantly accounted for an additional 14.0% of the variance [B = 0.50 (0.09), r_partial = 0.393, P < 0.001]. In model 3, HRV emerged as a significant inverse predictor (B = −1.61 (0.54), r_partial = −0.225, P = 0.003), explaining an additional 3.9% of the variance in T2 MAP in EAs. Regression results for MAP at T2 are presented in Table 2C for AAs. Model 1 accounted for 11.6% of the variance in T2 MAP with sex and age emerging as significant predictors (each P < 0.01). In model 2, T1 MAP significantly accounted for an additional 26.6% of the variance [B = 0.50 (0.09), r_partial = 0.393, P < 0.001]. In model 3, HRV explained only an additional 0.8% of the variance and did not significantly predict T2 MAP in AAs [B = −0.97 (0.60), r_partial = −0.115, P = 0.105]. Figure 1 depicts a scatterplot of HRV at T1 and MAP at T2 stratified by ethnicity.

Figure 1.

Scatterplot between heart rate variability (HRV) at time 1 and mean arterial pressure at time 2. Note: This figure depicts a scatterplot between high-frequency heart rate variability (HF-HRV) at time 1 and mean arterial pressure (MAP) at time 2. Regression results showed HF-HRV to significantly predict MAP for European Americans (n = 178) (EAs; r_partial = −0.225, P = 0.003) but not African Americans (n = 207) (AAs; r_partial = −0.115, P = 0.105). *P < 0.05.

Predicting Total Peripheral Index at Time 2

Regression results for TPI at T2 are presented in Table 3A for the full sample. Model 1 accounted for 21.9% of the variance in T2 TPI with only BMI and ethnicity emerging as significant predictors (each P < 0.01). In model 2, T1 TPI significantly accounted for an additional 2.7% of the variance [B = 0.17 (0.05), r_partial = 0.187, P < 0.001]. In model 3, HRV emerged as a significant inverse predictor [B = −0.74 (0.35), r_partial = −0.107, P = 0.038), explaining an additional 1.8% of the variance in T2 TPI.

Table 3.

Hierarchical regression models predicting time 2 total peripheral resistance index for the full sample and stratified by ethnicity

	b	t	P	r	rpartial	b	t	P	r	rpartial	b	t	P	r	rpartial
A: Full sample	Model 1 (R2 = 0.219)					Model 2 (R2change = 0.027)					Model 3 (R2change = 0.009)
Age T1	0.088	1.83	0.068	0.212	0.094	0.083	1.75	0.080	0.212	0.090	0.066	1.39	0.165	0.212	0.072
Sex	−0.042	−0.90	0.370	0.039	−0.046	−0.019	−0.41	0.683	0.039	−0.021	−0.018	−0.38	0.702	0.039	−0.020
Ethnicity	0.286	5.98	<0.001	0.351	0.294	0.248	5.14	<0.001	0.351	0.256	0.250	5.21	<0.001	0.351	0.260
BMI T1	0.264	5.38	<0.001	0.347	0.267	0.268	5.56	<0.001	0.347	0.275	0.266	5.55	<0.001	0.347	0.275
Father’s education T1	−0.043	−0.90	0.368	−0.186	−0.046	−0.040	−0.84	0.402	−0.186	−0.043	−0.039	−0.84	0.403	−0.186	−0.043
Smoking T1	0.047	1.00	0.316	0.053	0.052	0.049	1.07	0.285	0.053	0.055	0.045	0.98	0.326	0.053	0.051
TPI T1						0.171	3.69	<0.001	0.233	0.187	0.189	4.04	<0.001	0.233	0.204
HF-HRV T1											−0.096	−2.08	0.038	−0.081	−0.107
B: EAs	Model 1 (R2 = 0.131)					Model 2 (R2change = 0.034)					Model 3 (R2change = 0.024)
Age T1	0.044	0.57	0.567	0.149	0.044	0.047	0.63	0.532	0.149	0.048	0.029	0.39	0.701	0.149	0.030
Sex	−0.028	−0.38	0.703	0.019	−0.029	−0.013	−0.17	0.862	0.019	−0.013	−0.014	−0.19	0.847	0.019	−0.015
BMI T1	0.326	4.13	<0.001	0.344	0.300	0.332	4.27	<0.001	0.344	0.311	0.320	4.15	<0.001	0.344	0.303
Father’s education T1	0.010	0.13	0.897	−0.091	0.010	0.017	0.23	0.816	−0.091	0.018	0.008	0.12	0.907	−0.091	0.009
Smoking T1	0.099	1.37	0.173	0.136	0.104	0.107	1.50	0.137	0.136	0.114	0.097	1.37	0.174	0.136	0.104
TPI T1						0.185	2.64	0.009	0.167	0.198	0.214	3.03	0.003	0.167	0.226
HF-HRV T1											−0.159	−2.22	0.028	−0.166	−0.168
C: AAs	Model 1 (R2 = 0.110)					Model 2 (R2change = 0.029)					Model 3 (R2change = 0.003)
Age T1	0.121	1.77	0.079	0.186	0.124	0.112	1.66	0.099	0.186	0.116	0.100	1.45	0.150	0.186	0.102
Sex	−0.050	−0.72	0.475	−0.035	−0.050	−0.020	−0.28	0.777	−0.035	−0.020	−0.018	−0.26	0.795	−0.035	−0.018
BMI T1	0.258	3.73	<0.001	0.288	0.254	0.261	3.83	<0.001	0.288	0.261	0.262	3.84	<0.001	0.288	0.263
Father’s education T1	−0.080	−1.18	0.240	−0.133	−0.083	−0.079	−1.18	0.238	−0.133	−0.083	−0.076	−1.14	0.257	−0.133	−0.080
Smoking T1	0.012	0.17	0.865	0.048	0.012	0.012	0.18	0.857	0.048	0.013	0.012	0.17	0.864	0.048	0.012
TPI T1						0.174	2.61	0.010	0.179	0.182	0.186	2.72	0.007	0.179	0.190
HF-HRV T1											−0.057	−0.83	0.406	−0.059	−0.059

This table depicts hierarchical regression models predicting total peripheral index at time 2 (T2) in all individuals. Table 3A give regression statistics for the full sample, and Table 3, B and C present regression statistics in European Americans (EAs) and African Americans (AAs). Age measured in years, BMI, body mass index, kg/m²; HF-HRV, natural log transformed high frequency heart rate variability; TPI, total peripheral index; T1, time 1. Significant predictors (P < 0.05) are bold at each step.

For EAs, regression results for TPI at T2 are presented in Table 3B. Model 1 accounted for 13.1% of the variance in T2 TPI with only BMI emerging as a significant predictor (P < 0.001). In model 2, T1 TPI significantly accounted for an additional 3.4% of the variance [B = 0.17 (0.06), r_partial = 0.198, P = 0.009]. In model 3, higher HRV significantly predicted lower T2 TPI [B = −0.89 (0.40), r_partial = −0.168, P = 0.028), explaining an additional 2.4% of the variance in T2 TPI in EAs. Regression results for TPI at T2 are presented in Table 3C for AAs. Model 1 accounted for 11.0% of the variance in T2 TPI with sex and age emerging as significant predictors (each P < 0.01). In model 2, T1 TPI significantly accounted for an additional 2.9% of the variance [B = 0.17 (0.07), r_partial = 0.182, P = 0.010]. In model 3, HRV explained only an additional 0.3% of the variance and did not significantly predict T2 TPI in AAs [B = −0.49 (0.59), r_partial = −0.059, P = 0.406). Figure 2 depicts a scatterplot of HRV at T1 and TPR at T2 stratified by ethnicity.

Figure 2.

Scatterplot between heart rate variability (HRV) at time 1 and total peripheral resistance at time 2. Note: This figure depicts a scatterplot between high frequency heart rate variability (HF-HRV) at time 1 and total peripheral resistance (TPR) at time 2. Regression results showed HF-HRV to significantly predict TPR for European Americans (n = 178) (EAs; r_partial = −0.168, P = 0.028) but not African Americans (n = 207) (AAs; r_partial = −0.059, P = 0.406). *P < 0.05.

Predicting Cardiac Index at Time 2

Regression results for CI at T2 are presented in Table 4A for the full sample. Model 1 accounted for 10.5% of the variance in T2 CI with ethnicity and BMI emerging as significant predictors (each p < 0.01). In model 2, T1 CI significantly accounted for an additional 2.7% of the variance (B = 0.15 (0.05), r_partial = .173, p = 0.001). In model 3, HRV did not significantly predict T2 CI (B = -0.12 (0.03), r_partial = −.023, p = 0.655), explaining virtually no variance in T2 CI. Regression results for CI at T2 are presented in Table 4B for EAs. Model 1 accounted for 6.7% of the variance in T2 CI with BMI emerging as a significant predictor (P = 0.002). In model 2, T1 CI significantly accounted for an additional 2.3% of the variance [B = 0.18 (0.07), r_partial = 0.206, P < 0.001]. In model 3, HRV did not significantly predict T2 CI [B = 0.01 (0.03), r_partial = 0.017, P = 0.826), explaining virtually no variance in T2 CI. Regression results for CI at T2 are presented in Table 4C for AAs. Model 1 accounted for 6.5% of the variance in T2 CI with BMI as a significant predictor (each P = 0.003). In model 2, T1 CI significantly accounted for an additional 2.3% of the variance [B = 0.14 (0.06), r_partial = 0.158, P = 0.025]. In model 3, HRV explained only an additional 0.4% of the variance and did not significantly predict T2 CI [B = −0.36 (0.04), r_partial = −0.063, P = 0.371] in AAs.

Table 4.

Hierarchical regression models predicting time 2 cardiac index for full sample and stratified by ethnicity

	b	t	P	r	rpartial	b	t	P	r	rpartial	b	t	P	r	rpartial
A: Full sample	Model 1 (R2 = 0.105)					Model 2 (R2change = 0.027)					Model 3 (R2change = 0.000)
Age T1	−0.005	−0.11	0.916	−0.102	−0.005	−0.020	−0.39	0.700	−0.102	−0.020	−0.023	−0.44	0.657	−0.102	−0.023
Sex	−0.063	−1.24	0.214	−0.121	−0.064	−0.073	−1.46	0.145	−0.121	−0.075	−0.073	−1.46	0.146	−0.121	−0.075
Ethnicity	−0.152	−2.97	0.003	−0.211	−0.151	−0.139	−2.74	0.006	−0.211	−0.140	−0.138	−2.72	0.007	−0.211	−0.139
BMI T1	−0.225	−4.29	0.000	−0.272	−0.215	−0.220	−4.26	0.000	−0.272	−0.214	−0.221	−4.27	0.000	−0.272	−0.215
Father’s education T1	0.030	0.59	0.555	0.120	0.030	0.029	0.57	0.569	0.120	0.029	0.029	0.57	0.569	0.120	0.029
Smoking T1	−0.007	−0.14	0.885	0.003	−0.007	−0.018	−0.37	0.711	0.003	−0.019	−0.019	−0.38	0.702	0.003	−0.020
CI T1						0.165	3.42	0.001	0.172	0.173	0.160	3.19	0.002	0.172	0.162
HF-HRV T1											−0.023	−0.45	0.655	−0.051	−0.023
B: EAs	Model 1 (R2 = 0.067)					Model 2 (R2change = 0.040)					Model 3 (R2change = 0.000)
Age T1	0.018	0.23	0.821	−0.075	0.017	0.003	0.03	0.973	−0.075	0.003	0.004	0.06	0.956	−0.075	0.004
Sex	−0.052	−0.69	0.492	−0.090	−0.052	−0.045	−0.61	0.545	−0.090	−0.046	−0.045	−0.60	0.551	−0.090	−0.046
BMI T1	−0.252	−3.08	0.002	−0.253	−0.229	−0.253	−3.15	0.002	−0.253	−0.235	−0.252	−3.12	0.002	−0.253	−0.233
Father’s education T1	−0.008	−0.10	0.918	0.063	−0.008	−0.012	−0.16	0.871	0.063	−0.012	−0.011	−0.15	0.881	0.063	−0.011
Smoking T1	−0.009	−0.12	0.905	−0.028	−0.009	−0.029	−0.39	0.696	−0.028	−0.030	−0.028	−0.38	0.705	−0.028	−0.029
CI T1						0.201	2.75	0.007	0.192	0.206	0.204	2.73	0.007	0.192	0.205
HF-HRV T1											0.017	0.22	0.826	0.003	0.017
C: AAs	Model 1 (R2 = 0.065)					Model 2 (R2change = 0.023)					Model 3 (R2change = 0.004)
Age T1	−0.020	−0.28	0.779	−0.067	−0.020	−0.036	−0.51	0.609	−0.067	−0.036	−0.047	−0.66	0.510	−0.067	−0.047
Sex	−0.076	−1.07	0.284	−0.095	−0.076	−0.098	−1.38	0.171	−0.095	−0.097	−0.096	−1.34	0.180	−0.095	−0.095
BMI T1	−0.217	−3.06	0.003	−0.238	−0.211	−0.209	−2.97	0.003	−0.238	−0.206	−0.209	−2.97	0.003	−0.238	−0.206
Father’s education T1	0.058	0.84	0.403	0.085	0.059	0.060	0.87	0.385	0.085	0.061	0.063	0.92	0.361	0.085	0.065
Smoking T1	−0.003	−0.04	0.965	0.002	−0.003	−0.010	−0.14	0.887	0.002	−0.010	−0.010	−0.14	0.888	0.002	−0.010
CI T1						0.154	2.26	0.025	0.141	0.158	0.136	1.90	0.059	0.141	0.133
HF-HRV T1											−0.065	−0.90	0.371	−0.084	−0.063

This table depicts hierarchical regression models predicting cardiac index (CI) at time 2 (T2) in all individuals. Table 4A give regression statistics for the full sample, and Table 4, B C present regression statistics in European Americans (EAs) and African Americans (AAs). Age is measured in years. BMI, body mass index, kg/m²; HF-HRV, natural log transformed high frequency heart rate variability; T1, time 1. Significant predictors (P < 0.05) are bold at each step.

DISCUSSION

Results from the current study showed AAs to have higher MAP marked by higher TPR at T2 compared with EAs, independent of several important covariates including BMI, sex, socioeconomic status (fathers’ education), and respective hemodynamic measures at T1. The major novel findings of this study showed that higher HRV at T1 significantly predicted both lower TPR and MAP at T2 among EAs only; these associations were attenuated and not significant in AAs, especially HRV predicting TPR at T2. HRV did not significantly predict CO in the full sample or split by ethnicity.

Overall, the current investigation is the first to investigate how cardiac vagal tone (i.e., HRV) may differentially predict hemodynamics underlying long-term BP regulation based on ethnicity. Recent research showed HRV to predict long-term systolic and diastolic BP in EAs but not AAs (22), and these findings are in line with the current results; here, we show HRV to predict long-term MAP in EAs however this association was attenuated in AAs. Novel results in the current investigation extend prior findings: HRV differentially predicts the “hemodynamics” underlying such BP elevations over time between AAs and EAs. This is important, as BP elevations via increased TPR are particularly deleterious and associated with end organ damage. In all, we highlight that greater vagal activity may be beneficial in reducing BP (MAP) via reductions in TPR in EAs but not in AAs. This result in EAs is consistent with the finding that HRV was associated with endothelial function as indexed by brachial artery flow mediated dilation in humans (13).

There is increasing interest in the role of the vagus in cardiovascular disease in general and hypertension in particular (10, 11, 23, 24). Whereas there is ample evidence for direct vagal innervation of the heart, there is currently no evidence of direct vagal innervation of the vessels (14, 25). Therefore, the exact way in which the vagus is related to peripheral resistance is not clear. It has been reported that in rodents, activation of preganglionic vagal neurons in the dorsal motor nucleus of the vagus is associated with decreased ventricular contractility (26). This has led to the speculation that the impact on peripheral resistance may be secondary to prolonged vagally meditated decreases in CO (25). However in the current study, we found that HRV was not a significant predictor of CO in either EAs or AAs in our sample of young adult humans. Another speculation is that, at least in rodent models, acetylcholine released from the vagus nerve and carried through the bloodstream may act on muscarinic receptors on endothelial cells to produce vasodilation (27, 28). However, to date, there is no evidence for this mechanism in humans.

The most promising hypothesis put forward to explain the effects of vagal activity on the peripheral vasculature involves the cholinergic anti-inflammatory pathway (29–31). It has been proposed that this pathway is dysregulated in hypertension (30). Rodent models have shown that vagal activation leads to down regulation of proinflammatory cytokines in the serum and vasculature (11, 32). In humans, we have reported in both cross-sectional and prospective studies, that greater HRV is associated with lower levels of inflammatory markers (33–35). Of special relevance to the current findings, it has also been reported that AAs have both greater HRV and greater levels of proinflammatory cytokines relative to EAs (4, 36). Thus, the lack of association between HRV and TPR in the current study may be due to a dysregulation in the cholinergic anti-inflammatory pathway. However, additional research is needed to fully explore this potential mechanism.

Previous cross-sectional studies have examined ethnic differences in HRV and TPR (e.g., 3, 4, 9, 37), however the present study is the first investigation to examine ethnic differences in the longitudinal association between HRV and TPR 6 yr later. Sherwood and Hinderliter (8) showed that in response to the α-adrenergic agonist, phenylephrine, AAs had greater vasoconstriction (i.e., greater TPR) compared with EAs. Moreover, as mentioned, pregnancy is a potent vasodilator, and yet, pregnant AA women showed higher resting TPR and higher HRV compared with EA women, suggesting impaired vasodilation in AAs (9). Taken together, evidence suggests impaired vasodilation in response to both exogenous (e.g., Ref. 8) and endogenous (e.g., Ref. 9) vasodilatory agents. The current results are in line with these data, as higher HRV is not associated with lower TPR in AAs. Novel results from the current investigation suggest that this HRV and TPR association is evident longitudinally in EAs, but not AAs, thereby providing a physiological mechanism linking AAs with increased risk for cardiovascular disease.

Indeed, these data have major implications for health disparities research working to understand physiological mechanisms underpinning poorer cardiovascular health in AAs compared with EAs. To put this into a clinical context, the magnitude of the difference between AAs and EAs in TPR is equivalent to the difference between dippers and nondippers reported by Sherwood and colleagues (38). Nondipping is known to be associated with greater left ventricular hypertrophy, myocardial infarction, and stroke. Thus, the reported magnitude of the effect in the present study could be associated with the greater incidence of these adverse factors in AAs.

As early as 1948, the baroreflex was thought to play a role in both long-term and short-term BP regulation (39). Despite some conflicting empirical studies (e.g., Ref. 40) due to potential methodological issues (41), decades of research suggest that the baroreflex, mediated by the vagus and associated brain regions, is a major determinant of long-term BP regulation (see Ref. 42 for review). Therefore, given the current results regarding less effective vagal activity in AAs in decreasing TPR, the baroreflex may play an important role in the differential hemodynamics underlying long-term BP regulation between AAs and EAs. Overall, the current data support the idea of long-term BP regulation by the vagally mediated baroreflex (41, 42) and this influence of the baroreflex may differ between AAs and EAs; however future studies are needed to examine this directly.

The current study is not without its limitations. Although there was a high prevalence of smoking among our sample, smoking did not emerge as a significant predictor in any of our analyses. Also, the level of regular physical activity was generally low across the entire sample, however importantly, there were no significant differences in physical activity as a function of ethnicity. Results remain the same when controlling for the change in BMI from T1 to T2 (not a significant predictor in any model). In addition, whereas father’s education, as a proxy for socioeconomic status in this young adult sample, was significantly higher in EAs, it was not a significant predictor in any of the regression models. In the current study, impedance cardiography was used to provide an indirect measure of peripheral blood flow. Future studies should examine more direct measures such as brachial artery flow mediated dilation as used by Pinter et al. (13).

In summary, the current results suggest that in normotensive individuals, higher HRV predicts decreased MAP and TPR 6 yr later in EAs. In contrast, there was no association between HRV and hemodynamics 6 yr later in AAs. HRV did not predict CO in either EAs or AAs. Thus, it is unlikely that the effect of HRV on TPR was secondary to HRV effects on CO. In closing, our data highlight that AAs show TPR mediated long-term increases in BP irrespective of resting HRV, providing a physiological pathway linking AAs with a greater risk for mortality and morbidity from hypertension and potentially other cardiovascular disease.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

D.P.W., J.D.H., X.W., and G.K. conceived and designed research; J.D.H. performed experiments; D.P.W. analyzed data; D.P.W., J.F.T., and G.K. interpreted results of experiments; D.P.W. prepared figures; D.P.W. drafted manuscript; D.P.W., J.F.T., J.D.H., X.W., and G.K. edited and revised manuscript; D.P.W., J.F.T., J.D.H., X.W., and G.K. approved final version of manuscript.