Invited Commentary: Hypertension and Arterial Stiffness—Origins Remain a Dilemma

Abstract

In this issue of the Journal, Chen et al. (Am J Epidemiol. 2016;183(7):599–608) present repeated measures of aorto-femoral pulse wave velocity, capacitive compliance (C₁), and oscillatory compliance (C₂) in the Bogalusa Heart Study, with the purpose of addressing which comes first: blood pressure elevation or arterial stiffening. After an average follow-up period of 7 years (2000–2010), the authors found that blood pressure at a mean age of 36 years predicted change in arterial stiffening by a mean age of 43 years, but not the reverse. Essential hypertension results from a mosaic of pathological mechanisms. It has been theorized that biological pathways involving increased sympathetic tone, insulin resistance, renin-angiotensin-aldosterone activation, and inflammation lead to hyperkinetic circulation, volume overload, and vascular remodeling. The resultant accelerated vascular aging may be assessed by determining the degree of arterial stiffness. The findings of Chen et al. add important empirical information to the literature but do not solve the dilemma regarding the origins of essential hypertension, partly because there are many techniques for estimating the many aspects of arterial stiffness which are not fully understood. Mathematical features of estimates might not be uniform across the age range. There is a need for tracking blood pressure and different aspects of arterial stiffness from childhood onward, with an aim of preventing hypertension in adult life.

Despite numerous studies, the directionality of the association between essential hypertension and arterial stiffness remains unclear (1). One theory is that vascular changes precede elevated blood pressure (BP): Such changes lead to stiffer arteries, resulting in higher systolic BP, accompanied by a higher pulse pressure. An alternative theory is that elevated BP precedes vascular changes: Higher arterial BP per se has a detrimental influence on the vascular wall, leading to stiffer arteries, which results in atherosclerosis, fibrosis, and progressive BP rise. In this discussion and controversy, a wide array of techniques that are proposed to estimate arterial stiffness have been used (2). These either approach the arteries in a general way or focus on an arterial cross-section of a certain vascular area or blood vessel. The majority of these studies have been performed in middle-aged and older adults, where the classical aging process is relevant.

Chen et al. (3) studied the question of directionality in the Bogalusa Heart Study (Bogalusa, Louisiana). They followed the change in BP and arterial stiffness 7 years later, on average (2000–2010), in a white and black sample of participants in early middle age (mean age 36 years at first examination and maximum age 51 years at follow-up). They selected 3 different estimates of general arterial function. The first measure was aorto-femoral pulse wave velocity (afPWV), based on the distance from the aortic arch to the femoral artery pulse divided by pulse wave transit time—the difference in arrival time of the pulse wave at the femoral artery minus arrival time at the aortic arch. The other approach was based on diastolic pulse contour analysis of the radial artery waveform using a modified windkessel model, leading to the parameters C₁ and C₂. C₁ has been conceptualized as proximal arterial compliance, capacitive compliance, or large-artery elasticity, while C₂ has been conceptualized as distal arterial compliance, oscillatory compliance, or small-artery elasticity. Both C₁ and C₂ contain an ad hoc estimate of systemic vascular resistance (R), which in turn includes mean arterial BP measured at the brachial artery as a component. To minimize introduction of BP into their estimates of C₁ and C₂, Chen et al. multiplied both C₁ and C₂ by R, leading to C₁R and C₂R, both of which are highly correlated with C₁ and C₂ (3). They used a cross-lagged path analysis to study the temporal relationship between BP and the arterial function measures (afPWV, C₁R, and C₂R).

Their results indicated that during early middle age, higher BP at baseline preceded increased afPWV and decreased C₁R, both conceptualized as large-artery stiffness, but large-artery stiffness did not precede increases in BP. Thus, the pattern was that of a 1-directional relationship. They also reported a bidirectional relationship for C₂R, conceptualized as small-artery stiffness, with higher baseline BP being associated with less reduction in C₂R and lower baseline C₂R being associated with less increase in BP, but neither of these associations achieved statistical significance in this sample of about 400 people (3). The heterogeneous results for progression of BP and arterial stiffness may suggest that the relationship appears to be modest and that the theory “vascular changes precede elevated BP” is supported, but that it is difficult to find a consistent phenomenon. In particular, the authors considered the possibility that the direction of the relationship (arterial stiffening leading to increasing BP vs. increasing BP leading to arterial stiffening) might be different in early middle age than at older ages.

Existing theories of the origin of essential hypertension are based on a mosaic of pathological mechanisms and associated risk factors/contributors (4). As noted above, one theory is that there is a temporal sequence of vascular structural and functional changes in the pathogenesis of arterial hypertension, with endothelial dysfunction leading to a decrease in small-artery elasticity and increased peripheral vascular resistance, resulting in increased arterial wave reflections from the periphery to the heart, which cause more damage in the larger arteries (5). This theory is supported by a finding from the Multi-Ethnic Study of Atherosclerosis (MESA), in which Peralta et al. (6) studied 2,512 participants with an average age of 58 years who were free of clinical cardiovascular disease and nonhypertensive at baseline. Several estimates of arterial changes were studied in relation to the 545 cases of incident hypertension that were identified after an average of 4.3 years of follow-up. Two measures of arterial structural change were studied: coronary artery calcification and common carotid intima-media thickness (cIMT). Neither the presence nor the amount of coronary artery calcified atherosclerotic plaque predicted future hypertension (6). Calcification is a late metabolic stage in atherosclerotic plaque formation, whereas change in cIMT represents an earlier stage of atherosclerosis. In MESA, cIMT was a strong predictor of future hypertension in a model with highly adjusted results (6). The 3 measures of arterial functional change were aortic distensibility, C₁, and C₂. Among cIMT and these 3 arterial functional predictors, C₂ was clearly the strongest (hence our interest in the findings for C₂R in the Bogalusa Heart Study, nonsignificant but in a much smaller sample (3)). Shimbo et al. (7) previously showed in MESA that lower aortic distensibility, C₁, and C₂ at baseline were associated with increased BP variability (which correlates well with hypertension). Also in MESA, Shimbo et al. did not find an association for another measure of arterial function, flow-mediated vasodilation with future hypertension, even though lower values of flow-mediated vasodilation are often considered to reflect endothelial dysfunction and the development of hypertension (8).

Increased sympathetic tone along with a hyperdynamic circulation also relates to hypertension. Arterial BP is cardiac output × total peripheral vascular resistance. Prehypertension and hypertension are accompanied by increased heart rate due to increased sympathetic tone leading to increased cardiac output (9). Moreover, the Tecumseh Study found an association in participants with borderline elevated blood pressure of increased insulin resistance with sympathetic overactivity, which may affect the vascular wall, leading to more structural changes and more atherosclerotic plaque, resulting in stiffer arteries (9). In a Japanese study, Tomiyama et al. (10) found that both baseline heart rate and changes in heart rate were positively associated with changes in brachial ankle pulse wave velocity among 1,795 people with a mean age of 39 years and 5–6 years of follow-up, similar to Chen et al. (3).

The renin-angiotensin-aldosterone (RAAS) system is also a key pathological factor in hypertension. The RAAS system plays a crucial role in sodium and water handling. This not only leads to an increased circulatory volume, which translates to a hyperkinetic circulation, but also promotes vascular remodeling and fibrosis (11). These will lead at a later stage to arterial stiffness beyond an already high BP. Moreover, pathological effects of the RAAS system will promote inflammation.

Beyond the complexity of the possible origins of essential hypertension and the many different aspects of arterial stiffness, other considerations relate to the nature of the measures of arterial structure and function. Relatively few studies have used repeated measurements of arterial stiffness. In the Bogalusa Heart Study, within-person correlation of afPWV (tracking correlation) was 0.13, suggesting that afPWV was not stable over time, which could detract from the ability to predict future BP. C₁R had similarly low tracking correlation (r = 0.22), while C₂R was considerably more stable over time (tracking correlation: r = 0.40). These tracking correlations may be interpreted in light of published reproducibility or shorter-term tracking correlations. In the Bogalusa Heart Study, Chen et al. reported repeat testing correlations for afPWV, C₁, and C₂ that were all at least 0.68 for testing on different days (3). Carotid femoral pulse wave velocity had a test-retest correlation of 0.91, while carotid radial pulse wave velocity had a corresponding correlation of 0.76 in measurements taken 5 minutes apart at an average age of 40 years (computed as 1 − variance of the difference/(2 × variance of a single measure)) (12). The intraclass correlation coefficient for carotid femoral pulse wave velocity in children with measurements taken 3 days apart was 0.54 (13). Among participants who were generally older than those in the Bogalusa Heart Study (3), the same-day reproducibility of C₁, C₁R, C₂, and C₂R was in the range of 0.6–0.8 (14, 15). Tracking correlation for C₁ and C₂ over a period of 3–10 months was computed to be 0.67 for C₁ and 0.62 for C₂ (16). Therefore pulse wave velocity, C₁, and C₂ are stably measured over a short time interval, but only C₂ seems to qualify as a “stable characteristic” over a period of 7 years in middle age.

Chen et al. noted several limitations, including a relatively small sample size and a lack of physical activity and dietary data for studying confounding by these factors (3). They studied only continuous BP and omitted persons who were treated for hypertension at baseline or during follow-up. This decision truncated the range of BP change, which could potentially have limited the ability to find that baseline arterial stiffness predicted BP increases and future hypertension. Their data set would support addressing this last point.

Chen et al. concluded that “the hemodynamic and vascular functional changes underlying hypertension differ during younger versus older age periods in that the arterial wall may not be stiff enough in youth to alter BP levels during young adulthood” (3, p. 606). We agree that this is one possible conclusion. We also consider the possibility that the meanings of the different measures of arterial stiffness are not fully understood. Pulse wave velocity is a straightforward measure computationally and is usually regarded as a marker of large-vessel stiffness. However, the stiffness may be the result of higher BP rather than structural changes in the larger vessel per se. Elevated BP may result from changes in the distal arteries (resistance arteries); thus, afPWV might indicate small-vessel changes. Bogalusa Heart Study investigators should report the correlation between afPWV and C₂ or C₂R, which we speculate is at least as high as the correlation between afPWV and C₁ or C₁R.

Vascular alterations that reflect changes in C₁ and C₂ have been debated. It is of interest that C₂ and C₂R were highly predictive in MESA, not only of future hypertension but also of incident cardiovascular disease (17) and renal function decline (18). C₁ and C₁R were much less predictive in these respects. As was noted above, C₁ and C₂ were initially derived to represent elasticity in larger and smaller arteries, analogous to capacitive and oscillatory compliance in an electrical circuit. Some previous studies have supported this interpretation (19, 20). However, even if this analogy were not apt, the windkessel-derived fit to the diastolic waveform (a 6-parameter model: a decaying exponential function plus a cosine function dampened by a second decaying exponential function) is a good intuitive mathematical fit. C₂ is defined as the inverse of the sum of the primary rate of exponential decay (decrease in pressure during diastole) and twice the secondary rate of exponential decay of oscillations (decrease in magnitude of oscillations during diastole). An alternative or complementary interpretation of C₂R is that it expresses the extent to which arterial wave reflections from the periphery to the heart, which originate during systole, persist past aortic valve closure (beginning of dicrotic notch), thereby altering the shape of the diastolic waveform. Furthermore, a 6-parameter, nonlinear regression fit is mathematically demanding and not completely understood. It is possible that the windkessel function model parameters behave differently in waveforms derived from a younger person with robust arteries than in waveforms derived from an older person who has suffered various forms of arterial damage, and particularly that estimation of the 2 exponential decay parameters is not mathematically robust. Given the strong predictive power of C₂ in MESA, its properties and meaning should be further studied.

The report by Chen et al. (3) is a valuable contribution to the literature, as it presents data pertaining to the origins of essential hypertension and the meaning of arterial stiffness. However, the study raises more questions than it answers. Arterial BP should be studied longitudinally starting in childhood, together with markers of arterial function, to learn whether long-term BP level or arterial stiffness is a determinant of future hypertension and vascular disease. Preventing the development of hypertension is one of the challenges of the future, needed to reduce target organ damage and cardiovascular disease burden. A focus of cardiovascular preventive therapy should be the design of a BP-lowering regimen which would not only prevent the development of hypertension but also delay or attenuate the trajectory of arterial stiffening at an early age. This would be very beneficial in terms of healthy cardiovascular aging.