Pre-hypertension is an emerging and remarkably common risk factor for not only hypertension, but also cardiovascular target organ complications. The term pre-hypertension was defined in 2003 by the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure as a systolic blood pressure (SBP) of 120 to 139 mm Hg, diastolic blood pressure (DBP) of 80 to 89 mm Hg, or both, although less than values defined as hypertension (SBP ≥ 140 mmHg or DBP ≥90 mm Hg) (1). In the United States alone, up to approximately 41.9 million men and approximately 27.8 million women may exhibit pre-hypertension (2). Pre-hypertension tends to progress to hypertension over a relatively short time and is a risk factor for development of microalbuminuria and cardiovascular disease, with consequently increased mortality (3- 7). However, the origins and pathogenesis of the syndrome are not yet well understood.

Given the serious prognosis for this condition, we undertook this evaluation of its pathophysiology, with the goals of not only prevention, but also discovery of novel target processes for treatment. We therefore used a twin pair design to examine the role of heredity in the origin of the pre-hypertension trait, as well as its hemodynamic and autonomic determinants.

Data were obtained from the University of California, San Diego, twin/family study population, which has been described previously (8). Demographics for the population sample in this study are shown in (Table 1). There were 812 total individuals, comprising 350 monozygotic twins, 198 dizygotic twins, 233 other siblings of twins, 17 parents of twins, 6 children of twins, and 8 other relatives. There were 572 females and 240 males. The Human Research Protection Program at University of California, Sand Diego, approved the protocol, and each subject gave informed written consent.

Data were obtained using a Dynapulse 5200A oscillometric noninvasive blood pressure monitor (Pulse-Metric, Vista, California), a device previously validated by us and others against invasive techniques, including blood pressure (BP) and cardiac output (CO) (9- 10). This monitor also noninvasively estimated heart rate (HR), mean arterial pressure, left ventricular (LV) contractility as the change in pressure divided by change in time (dP/dT), CO, stroke volume (SV), systemic vascular compliance (SVC), systemic vascular resistance (SVR), and brachial artery distensibility (i.e., compliance normalized to size). Cardiac index (CI = CO/body surface area [BSA]), systemic vascular resistance index (SVRI = SVR/BSA), and SV index (SV index = SV/BSA) were calculated using data obtained and BSA. Measurements were obtained in the seated position after at least 5 minutes of rest. The cuff was placed on the right arm with the arm supported at heart level. Measurements for each subject were obtained in triplicate, and the average of all 3 values were used, if within ±10%. Blood pressure values for each subject were estimated using adjustments to Dynapulse blood pressure readings using published data (11).

Triplicate-average blood pressures for each individual were age adjusted (by linear regression) to an age of 40 years. Normal, pre-hypertensive, and hypertensive blood pressures were defined by the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (1). Individuals were partitioned into the normotensive category if both SBP was <120 mm Hg and DBP was <80 mm Hg; into the pre-hypertensive category if SBP was 120 to 139 mm Hg, DBP was 80 to 89 mm Hg, or both; and into the hypertensive category if either SBP was more than 139 mm Hg or DBP was more than 89 mm Hg, or both, or they were taking antihypertensive medications. Forty-four individuals (5.4%) already were taking antihypertensive drugs and thus were placed in the hypertension group.

Plasma catecholamines were measured radioenzymatically in 680 individuals: samples were frozen quickly at −70°C, before a sensitive radioenzymatic assay based on catechol-O-methylation (12).

BP and HR were recorded continuously and noninvasively for 5 minutes in seated, resting subjects with a radial artery applanation device as well as thoracic electrocardiogram (ECG) electrodes and dedicated sensor hardware (Colin Pilot, Colin Instruments, San Antonio, Texas) and software (ATLAS, W. R. Medical Electronics, Stillwater, Minnesota; and Autonomic Nervous System, Tonometric Data Analysis [ANS-TDA] System, Colin Instruments, San Antonio, Texas). Effects of environmental stress (cold pressor test: immersion of 1 hand in ice water for 1 minute) on BP and HR were recorded in triplicate before and at the end of the stressor. Lorenz plots and analyses (13) using successive beat-to-beat intervals to evaluate sympathetic (cardiac sympathetic index; Lorenz L/T ratio) versus parasympathetic activity (cardiac vagal index; Lorenz log₁₀ [L × T]), were calculated from 5-minute resting cardiac monitoring data in seated subjects with ANS-TDA software in 772 individuals, as described (14).

Genomic DNA was isolated from blood leukocytes, as has been described previously (15) using PureGene DNA extraction kits (Gentra Systems, Minneapolis, Minnesota). Genotyping was carried out as part of a single nucleotide polymorphism array (Ilumina 610-Quad, La Jolla, California) anchored by reference single nucleotide polymorphism data.

Estimates of heritability (h² = V_G/V_P, where V_G is additive genetic variance and V_P is total phenotypic variance), as well as shared environmental effects (environmental covariance) and pleiotropy (shared genetic determination or genetic covariance), were obtained from monozygotic versus dizygotic twin pairs, with variance-component methods implemented in Sequential Oligogenic Linkage Analysis Routines (16).

To test the effects of particular genes on traits in the twin/sibling sample, we used generalized estimating equations as in SAS software version 9.2 (SAS Institute, Cary, North Carolina), as previously described (8), in which correlated (e.g., within twinship) observations can be accounted for by establishing an exchangeable correlation matrix. To guard against potential artifactual conclusions as a result of genetic admixture, this analysis was confined to individuals self-identified as white.

Descriptive statistical analyses among the 3 groups (normotensive, pre-hypertensive, hypertensive) were carried out using SPSS software version 17.0 (SPSS, Inc., Chicago, Illinois). Because the participants were twins and therefore not genetically independent, the effective number of individuals for any characteristic with substantial heritability would be less than the actual number of participants, and the significance (but not the magnitude) of correlations may be overestimated; a conservative correction would be requirement of a threshold alpha of 0.025 (rather than the customary 0.05). Overall significance was determined by generalized linear models using BP group as a fixed factor and age and sex as covariates. Post-hoc pairwise analyses among the 3 individual groups were carried out with the conservative Bonferroni correction. All post-hoc p values were Bonferroni-corrected in SPSS. Spearman (nonparametric rho) and Pearson product moment (parametric R) correlations across traits were calculated in SPSS software as well. Results were plotted in Kaleidagraph software version 4.02 (Synergy Software, Reading, Pennsylvania), fitting linear or smoothed curves to the data, beginning with a Stineman function, whereupon the output of this function then has a geometric weight applied to the current point and ±10% of the data range, to arrive at the smoothed curve. Linear regression also was carried out using Kaleidagraph, which calculated the Pearson product moment correlations with R and p values.

Characteristics of the population sample in this study are shown in (Table 1), whereas noninvasive hemodynamic results are summarized in (Table 2). Average SBPs for normotensive, pre-hypertensive, and hypertensive individuals were 109.9 ± 0.49 mm Hg, 127.3 ± 0.46 mm Hg, and 138.0 ± 0.75 mm Hg, respectively (p ≤ 0.001), whereas DBPs in these groups were 66.3 ± 0.42 mm Hg, 74.4 ± 0.41 mm Hg, and 78.9 ± 0.67 mm Hg, respectively (p ≤ 0.001). Mean SBP and DBP in the hypertensive group were slightly lower than Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure definitions for hypertension because of the inclusion of 44 individuals taking antihypertensive medications. Mean arterial pressure (p ≤ 0.001) and pulse pressure (PP) (p ≤ 0.001) also rose across the BP categories. HR also differed (p ≤ 0.001), rising systematically with BP.

Biogeographic ancestries differed across the BP groups, with an increased frequency of pre-hypertension among African Americans (p ≤ 0.001), consistent with different population prevalence of hypertension by ancestry. Age was not different between normotensive and pre-hypertensive subjects, although it was substantially higher in the hypertensive subjects (p ≤ 0.001), consistent with the age-dependent penetrance of the hypertension trait. Men were more likely than women to exhibit pre-hypertension (p ≤ 0.001), also consistent with known hypertension demographics. Weight rose by approximately 15 kg across the groups (p ≤ 0.001), and body mass index also rose by approximately 3 kg/m² (p ≤ 0.001), whereas overall body size (as BSA) did not change (p = 0.155).

Hemodynamic determinants of BP were strikingly different across the 3 groups (Table 2). CO (p ≤ 0.001), CI (p ≤ 0.001), SV (p ≤ 0.001), and SV index (p ≤ 0.001) also were different by analysis of variance (ANOVA) across the 3 groups; these traits also differed between the normotensive and pre-hypertensive categories, although not significantly between pre-hypertensives and hypertensives on post hoc analysis (although values for these categories tended to increase as BP increased). Systemic vascular compliance (p = 0.001) and brachial artery distensibility (p ≤ 0.001) tended to decrease as BP increased; the overall changes were significantly different by ANOVA, although principally between normotensive and pre-hypertensive groups, rather than pre-hypertensives and hypertensives on post hoc analysis. SVR and SVRI were not significantly different among the 3 groups.

Within the heart, the maximum LV dP/dT (dP/dT_max; a measure of cardiac contractility) significantly (p ≤ 0.001) increased across the normotensive, pre-hypertensive, and hypertensive groups, with values of 1097.4 ± 8.89 mm Hg/s, 1263.8 ± 8.87 mm Hg/s, and 1347.6 ± 14.6 mm Hg/s, respectively. The increase in LV dP/dT_max correlated highly with SBP across the groups (R² = 0.528) ((Figure 1)A and Figure 1B), suggesting that up to approximately 53% of variation in SBP could be accounted for by the distribution of dP/dT.

PP (PP = SBP − DBP) also increased across the 3 groups ((Table 2), Figure 1C), from 56.2 ± 0.49 mm Hg, 61.8 ± 0.48 mmHg, and 68.0 ± 0.79 mm Hg, respectively (p ≤ 0.001). Among likely contributors across the BP groups, LV contractility increased (p ≤ 0.001), whereas systemic vascular compliance decreased as BP increased, with values of 1.31 ± 0.013 ml/mm Hg, 1.25 ± 0.013 ml/mm Hg, and 1.21 ± 0.022 ml/mm Hg (p = 0.001).

Results are summarized in (Table 3). Plasma norepinephrine levels increased with BP (p = 0.028), although only normotensive and hypertensive groups differed on post hoc analysis ((Figure 2)A and Figure 2B); plasma epinephrine did not differ across the groups. Lorenz beat-to-beat values probed cardiac sympathetic versus parasympathetic activity (13). Lorenz log₁₀ (L × T), a measure of parasympathetic activity, differed overall (p = 0.050), with the principal subgroup difference being between normotensives and hypertensives (p = 0.046). The scatterplot (Figure 2C) illustrates the contribution of cardiac parasympathetic activity to basal heart rate. The Lorenz L/T ratio, a measure of sympathetic activity, did not differ among the 3 groups. During environmental (cold) stress, changes in SBP, DBP, or HR did not differ among the 3 groups (in contrast to longitudinal reports [17]).

Trait heritability (h²) values are summarized in (Table 4) and are illustrated in (Figure 3)A. Using variance components from twin data, we estimated trait h² values at: SBP, 44.6 ± 6.7% (p ≤ 0.001); plasma norepinephrine, 65.2 ± 5.0% (p ≤ 0.001); HR, 62.2 ± 5.3% (p ≤ 0.001); Lorenz log₁₀ [L × T] (parasympathetic) index, 22.2 ± 8.0% (p = 0.004); LV dP/dT_max 35.3 ± 7.4% (p ≤ 0.001); CI, 60.5 ± 5.6% (p ≤ 0.001); and SVR, 57.3 ± 5.6% (p ≤ 0.001).

LV dP/dT, SVR, CI, HR, and plasma epinephrine and norepinephrine levels were examined for shared genetic determination with SBP, DBP, or both in the twin pairs ((Table 4), Figure 3B). Of these, only LV dP/dT and CI showed any significant genetic covariance overlap with SBP (CI: p ≤ 0.001, LV dP/dT: p ≤ 0.001). None of these traits showed heritable association with DBP. Also, SVR, CI, and LV dP/dT displayed environmental covariance with SBP (SVR: p = 0.005, CI: p = 0.007, LV dP/dT: p ≤ 0.001) (Figure 3B).

We further explored pleiotropic genetic effects on SBP with its more proximal determinants: either CI or LV dP/dT. Because the L-type voltage operated calcium channel controls not only vascular smooth muscle tone, but also cardiac contractility and sympathetic transmitter release, we began with the characteristic alpha-subunit of voltage operated calcium channel that confers L-type specificity: alpha-1C (CACNA1C) (18), which displayed several single nucleotide polymorphisms associated with both LV dP/dT_max and SBP. To illustrate the association, we focus on the CACNA1C intron-3 variant rs2239050 allele, because it has been associated previously with both BP and dihydropyridine response (18). CACNA1C genetic variation (Figure 3C) predicted both LV dP/dT (p = 0.008) and SBP (p ≤ 0.001), and multivariate ANOVA also indicated a pleiotropic or coordinate effect of the gene on both traits (p ≤ 0.001). At this G/C variant, our population sample displayed 91% G/9% C alleles, in Hardy Weinberg equilibrium (p = 0.681), and consistent with European populations reported in the HapMap (19). Given the minor allele frequency of 9%, the number of minor allele homozygotes (C/C) was small (n = 4), and hence we combined C/C and G/C individuals, in effect testing a C-allele dominant model for each trait.

Our results suggest that increases in BP in pre-hypertension result from a pathway initiated by heritable disturbances displaying joint genetic determination, and then actuated by autonomic and hemodynamic events in series (Figure 4).

The twin sample provided us with an unusual opportunity to probe for early heritable origins of pre-hypertension. Heritability (h²) is the fraction of trait variance attributable to genetic variation (20). The h² values document substantial genetic determination of not only SBP and DBP, but also several associated pathogenic traits: CI, plasma norepinephrine levels, cardiac parasympathetic index (Lorenz log₁₀ [L × T]), SVR, and cardiac contractility measured by LV dP/dT_max. Genetic covariance (or pleiotropy) analyses indicate that these correlated traits share heritable determination (i.e., coordinate gene action on more than 1 trait). The findings thus illustrate that genetic predisposition to pre-hypertension (and hence hypertension) is initiated through pleiotropic gene effects, initially on pathogenic autonomic traits (e.g., parasympathetic and sympathetic tone), and later on hemodynamic traits (e.g., cardiac contractility or LV dP/dT_max). Not all traits associated or correlated with SBP and DBP in our sample displayed genetic covariance; indeed, some (SVR, LV dP/dT_max, CI) evinced shared environmental determination, in the form of environmental covariance. Thus, genes alone are not adequate to explain risk for development of pre-hypertension or hypertension itself; indeed, gene-by-environment interactions also may be at work (21).

The full spectrum of genetic variation underlying hypertension is only now beginning to emerge and remains incompletely understood (22- 25). Allelic variation at CACNA1C has been associated with BP response to calcium channel blockers (18,26), or even BP itself (27). Association of CACNA1C variant rs2239050 predicted both cardiac contractility and SBP (Figure 3C), and the multivariate ANOVA confirmed a pleiotropic effect of the locus on both traits. Variant rs2239050 lies within an intron, and thus is unlikely to be the causal variant; however, systematic polymorphism discovery at CACNA1C (28) revealed 46 polymorphisms across the approximately 300-kbp locus, including several potentially functional variants.

There was an overall increase in plasma norepinephrine levels as BP and HR increased; the increment in norepinephrine is a novel finding in pre-hypertension, whereas epinephrine was unchanged. Several previous reports suggest catecholamine increments in hypertension (29- 33). Age and obesity also can affect catecholamine levels, even independent of BP (29,34). The increased norepinephrine could account in part for the increase we saw in contractility. Increased norepinephrine secretion also may contribute to unchanged SVR, which otherwise would be expected to decrease reflexively in the face of elevated CO (35).

To understand why, we used Lorenz plots to generate selective physiological indices of cardiac sympathetic versus parasympathetic activity. Parasympathetic activity (as Lorenz log₁₀ [L × T]) was decreased in hypertension, with intermediate values in pre-hypertension, consistent with the increase in HR seen in our data and other studies, rather than an increase in cardiac sympathetic activity (36- 37); indeed, decline in parasympathetic tone seemed to account for up to approximately 11% of HR variance (R² = 0.111) (Figure 2C). An increase in HR during pre-hypertension has been reported in white persons and American Indians, although not in African Americans (36).

Because BP is the algebraic product of CO (or CI = CO/BSA) and SVR (SVRI = SVR/BSA; i.e., mean arterial pressure = CO × SVR), we evaluated these traits in pre-hypertension and found that increased CO, rather than SVR, accounted for BP elevation in pre-hypertension. In turn, given that CO = SV × HR, the elevation of CO in pre-hypertension resulted from increments of both SV and HR. The increment in SV was driven by elevated cardiac contractility (LV dP/dT_max). Finally, the elevated HR was attributed primarily to a decline in parasympathetic tone, as evidenced by the cardiac vagal index, although norepinephrine results also pointed to an elevation in sympathetic activity.

The norepinephrine increment also suggested that an increase in sympathetic activity contributed to (or perhaps even accounted for) the rise in LV dP/dT ((Table 3), Figure 2B). An overall schema integrating our results into a likely pathogenic sequence of events for pre-hypertension is given in (Figure 4).

PP (PP = SBP − DBP) is an emerging independent risk factor for cardiovascular morbidity (7), and widening of PP in the past has been attributed to either increased cardiac contractility or diminished vascular compliance (38). Our data indicate that PP already is increased in pre-hypertension and that the change can be attributed to both elevated contractility and diminished vascular compliance, even at this early stage.

Our observation of an increase in cardiac contractility, as measured by LV dP/dT_max, with a failure of SVR to decrease appropriately. Although this has been seen in hypertensive patients previously (39), our data are unique in that they shows that there may be a progressive increase in cardiac contractility with BP. Similar data in pre-hypertension have been reported in the past for CO: an increase in CO in pre-hypertensives may occur without significant change in SVR (36- 37,40). However, measures of contractility have not been obtained previously: our data show a striking correlation with increasing LV dP/dT_max and increasing SBP. Other factors involved in CO, such as SV and HR, seem to become altered at the pre-hypertension phase, but do not change further as individuals progress to hypertension.

We did not detect differences in SVR (or SVRI) across the 3 groups. It should be noted that the expected homeostatic response to an increase in cardiac contractility and CO would be a decline in SVR to restore BP to normal. Thus, a failure of SVR decline can contribute pathogenically to elevated BP.

The emerging cardiovascular risk factor PP (PP = SBP − DBP) (7) already was elevated (by approximately 17.5%), even in pre-hypertension. Why did this occur? We noted 2 plausible mechanisms: both elevated cardiac contractility and decreased arterial distensibility.

First, the twin study design allowed us to probe the role of heredity in both hypertension and its genetically codetermined precursor events. Our study also is comprehensive in that it couples multiple hemodynamic and autonomic indices across BP strata.

Limitations in the study include its cross-sectional (rather than longitudinal) design. Finally, although we focused on autonomic and hemodynamic determinants of BP in this report, our pre-hypertensives demonstrated a body mass index increment of approximately 2.9 kg/m² over their normotensive counterparts; in other reports, an early elevation of body mass index in pre-hypertension, with its associated metabolic traits, also may be central to the syndrome (41).

Our findings indicate that the onset of pre-hypertension may be genetically determined at least in part and the result of joint heritability among a cluster of BP-correlated autonomic and hemodynamic traits. In the autonomic realm, an increment in sympathetic tone, coupled with a decline in parasympathetic tone, may be pathogenic. Within hemodynamic determinants of BP, an elevation in cardiac contractility (augmenting SV) seems to drive an increase in CO, accounting for elevated BP, perhaps assisted by lack of homeostatic decline in SVR. The increment in PP seems to result from pathogenic changes in both LV contractility and arterial distensibility. The role of genetic pleiotropy in trait determination is reinforced by the coordinate effects of genetic variation at a single locus on both contractility and BP.

Thus far, deliberate treatment approaches to pre-hypertension have been focused mainly on angiotensin receptor blockade (42- 43) or nonpharmacological means (44), such as dietary modification (DASH [Dietary Approaches to Stop Hypertension] diet) (45) or weight reduction. Our results document several additional points for logical and rational therapeutic intervention in pre-hypertension, including elevated sympathetic outflow (suggesting alpha-2-adrenergic agonist treatment), increased cardiac contractility (suggesting beta-adrenergic antagonist treatment), decreased arterial distensibility, and participation of CACNA1C (Figure 3C) (suggesting L-type voltage operated calcium channel antagonists). Nonetheless, because adrenergic pathways subserve many functions in addition to BP, antiadrenergic drugs may exhibit more frequent central nervous system side effects than newer drugs targeting the angiotensin system (46). A recent meta-analysis indicates that a variety of cardiovascular agents used during secondary prevention trials even in subjects without hypertension may be effective in diminishing risk of subsequent vascular events, although investigators agree that more data on individual drugs would be useful (43).