Central Hemodynamics for Management of Arteriosclerotic Diseases

Abstract

Arteriosclerosis, particularly aortosclerosis, is the most critical risk factor associated with cardiovascular, cerebrovascular, and renal diseases. The pulsatile hemodynamics in the central aorta consists of blood pressure, flow, and stiffness and substantially differs from the peripheral hemodynamics in muscular arteries. Arteriosclerotic changes with age appear earlier in the elastic aorta, and age-dependent increases in central pulse pressure are more marked than those apparent from brachial pressure measurement. Central pressure can be affected by lifestyle habits, metabolic disorders, and endocrine and inflammatory diseases in a manner different from brachial pressure. Central pulse pressure widening due to aortic stiffening increases left ventricular afterload in systole and reduces coronary artery flow in diastole, predisposing aortosclerotic patients to myocardial hypertrophy and ischemia. The widened pulse pressure is also transmitted deep into low-impedance organs such as the brain and kidney, causing microvascular damage responsible for lacunar stroke and albuminuria. In addition, aortic stiffening increases aortic blood flow reversal, which can lead to retrograde embolic stroke and renal function deterioration. Central pressure has been shown to predict cardiovascular events in most previous studies and potentially serves as a surrogate marker for intervention. Quantitative and comprehensive evaluation of central hemodynamics is now available through various noninvasive pressure/flow measurement modalities. This review will focus on the clinical usefulness and mechanistic rationale of central hemodynamic measurements for cardiovascular risk management.

Introduction: Historical View

Arterial blood pressure (BP) has long been recognized as an important biomarker of arterial function. More than 280 years ago, Stephen Hales compared the arterial system to a fire extinguisher with an air chamber (the so-called “Windkessel”), showing that the pulse waveform of BP (e.g., systolic upstroke and diastolic exponential decay) depends on aortic distensibility¹). Until the invention of the brachial cuff ausculatory method by Riva-Rocci and Korotkoff, the BP waveform had been utilized to diagnose hypertension with arterial degeneration²⁾. In the early 20th century, it was realized that the pulsatile BP wave is transmitted along the arterial tree from the central aorta to peripheral arteries at a speed depending on arterial stiffness³⁾. In the mid-20th century, the reflection phenomenon of BP waves was recognized by experimental studies involving frequency domain analysis of aortic impedance^{4, 5)}. Human (invasive) studies involving time domain analysis of the intra-aortic BP waveform also confirmed that it is largely explicable by the transmission and reflection phenomena and closely associated with aortic wall stiffness^{6, 7)}. Clinical introduction of applanation tonometry for noninvasive BP recording revealed age-dependent (i.e., arteriosclerosis-induced) changes in the radial and carotid artery waveforms, leading to the proposal of the concept of the augmentation index (AIx)⁸⁾. A mathematical transformation method has also been devised to generate aortic waveforms from radial waveforms and to estimate aortic BP⁹⁾.

Similar to BP measurement, blood flow measurement has progressed from invasive intra-arterial recordings (e.g., through an electromagnetic flow-meter) to noninvasive duplex ultrasound or MRI recordings, which enable us to evaluate pulsatile hemodynamics in the central aorta more easily and comprehensively. With the increasing availability of these new pressure/flow measurement modalities, more attention has recently been paid to the clinical significance of central hemodynamics in cardiovascular diseases (CVDs). This review will summarize the current evidence on the potential for central hemodynamic indices to stratify cardiovascular risk and/or serve as a surrogate marker of clinical endpoints. This review will also focus on the pathophysiological mechanisms underlying the link between central hemodynamic abnormalities and end-organ damage/dysfunction and discuss future directions of clinical research.

What is Central BP?

Central BP is generally defined as BP in the ascending aorta. In a broader sense, it can refer to BP in the thoracic or abdominal aorta, representing BP exerted at the level of the heart, brain, and kidney¹⁰⁾. Due to the phenomenon of pulse pressure (PP) amplification¹¹⁾, the central (systolic and pulse) pressures essentially differ from the brachial pressures. The central– brachial BP difference considerably varies among individuals¹²⁾; for instance, there is an approximate difference of < 5 mmHg in some people (e.g., elderly people with a very stiff aorta), whereas it occasionally exceeds 30 mmHg in others (e.g., young people with a distensible aorta¹³⁾ or during exercise¹⁴⁾). This difference cannot be estimated directly from conventional cuff BP measurement alone (i.e., solely from the two extremes of brachial BPs), but it can be actually estimated by adding information on the BP pulse waveform (see “How is central BP measured?” below).

The BP pulse waveform can be explained on the basis of the wave transmission and reflection phenomena along the arterial tree. In cardiac systole, the blood ejected from the left ventricle generates an incident (forward) BP wave. The incident wave then travels antegradely from the central aorta to the peripheral arteries at a speed termed “pulse wave velocity (PWV).” When the incident wave encounters high-impedance sites (e.g., high-resistance arterioles and arterial bifurcations), it is reflected and then travels retrogradely toward the central aorta at a speed similar to PWV. PWV is high (normally around 5–10 m/s) enough for the pulse to return up to the central aorta within the same cardiac cycle so that the reflected wave interferes with the incident wave to augment BP. The degree of augmentation is generally evaluated as AIx, which is the ratio of augmented pressure to PP (Fig. 1¹⁵⁾).

Fig. 1.

Pressure pulse waveforms of radial artery (A) and central aorta (B)

AIx_A indicates aortic augmentation index; AIx_R, radial augmentation index; AMP, pulse pressure amplification; AMP_A-R, aorta-to-radial AMP ratio; AP, augmented pressure; BP, blood pressure; P1, early systolic peak pressure; P2, late systolic peak pressure; P_d, diastolic pressure; P_i, pressure at inflection point; PP, pulse pressure; and T_r, round-trip travel time. Reproduced from Hashimoto¹⁵⁾.

How is Central BP Measured?

Until a quarter-century ago, aortic BP was only available in an invasive manner by inserting an intraarterial catheter-tip manometer or a fluid-filled catheter connected to a pressure transducer. Noninvasive estimation of aortic BP has become feasible since the technological development of applanation tonometry^{16, 17)} performed on arteries running near the body surface (such as the carotid and radial arteries)⁸⁾ and mathematical transformation of tonometric BP waveforms^{9, 18)}. Applanation (flattening) of arteries can be manually performed using a pencil-type probe (i.e., tonometer) or (semi)automatically performed using a hands-free tonometric system. To date, there are three major methods available for central BP estimation.

1). Estimation from Carotid BP Waveform

This method is based on the premise that the carotid waveform is deemed to be identical to the aortic waveform¹⁹⁾. The carotid waveform is usually recorded with applanation tonometry and then calibrated by the brachial cuff pressures [i.e., by the mean arterial pressure (MAP) and diastolic BP] to derive absolute values of central BP. The rationale for calibration by brachial BPs is the observation that MAP and diastolic BP are almost constant in the elastic and conduit arteries (e.g., between the central aorta and brachial artery)²⁰⁾. MAP is either calculated directly from the area under the curve of the brachial (or radial) pulse waveform or estimated indirectly from an equation such as MAP = diastolic BP + 0.4 × PP²¹⁾. This method has the advantage of not requiring any mathematical waveform transformation to estimate absolute central BP values. Nevertheless, it may be technically difficult in some cases to record (flatten) the carotid waveforms appropriately or reproducibly because of the anatomical characteristics of the carotid artery surrounded by soft tissues. The carotid pulse wave recordings can occasionally be affected by respiratory variation. Carotid AIx correlates well with invasively determined aortic AIx, although the former is reported to be half of the latter¹⁹⁾.

2). Mathematical Conversion from Radial to Central BP Waveform with Transfer Function

In this method, the radial waveform is recorded with applanation tonometry and then converted to the corresponding aortic waveform using a frequency domain transfer function. There are two types of transfer function: generalized^{9, 18)} and individualized. The former is commercially available and more widely used. Though still debated²²⁾, a generalized transfer function is considered valid for normal subjects, for patients with various diseases^{18, 23, 24)}, during exercise²⁵⁾, during pharmacological intervention²⁶⁾, and during the Valsalva maneuver¹⁸⁾. The synthesized aortic waveform is normally calibrated with brachial diastolic BP and MAP to obtain central BP; MAP is calculated from the area under the radial waveform on the basis that the radial and brachial waveforms are deemed to be identical⁸⁾. Aortic BP estimated using this method somewhat underestimates systolic BP and overestimates diastolic BP as measured directly using an invasive method²⁷⁾, but these errors almost entirely depend on the inaccurate calibration by the brachial cuff measurement²⁸⁾. The estimated aortic BP closely correlates with invasive aortic BP²⁹⁾. Reproducibility has been repeatedly verified in terms of aortic BP and AIx estimation^30–32).

3). Direct Estimation from Radial BP Waveform without Using Transfer Function

It is empirically recognized that aortic systolic (maximum) BP is equal to the late systolic (second) peak (or “shoulder”) of radial BP³³⁾. Based on this recognition, the radial waveform is calibrated with brachial systolic and diastolic BPs to determine radial late systolic BP, which corresponds to aortic systolic BP. Although this method cannot generate the aortic waveform itself, it can offer the advantage of directly estimating aortic systolic BP and PP without applying a complicated waveform transformation. Radial AIx is usually calculated as the ratio of late systolic to early systolic BP peak amplitude (Fig. 1)³⁴⁾. This method has been validated using invasive studies^{35, 36)}, and the estimated radial AIx and late systolic pressure are reported to be reproducible³⁷⁾. Though only rare (< 5%), aortic BP may be unavailable because of an undetectable radial second peak (e.g., under conditions of marked tachycardia or vasodilation).

In addition, several devices have been newly developed and are currently available for noninvasive central BP recordings. Some of them adopt brachial cuff-based pulse volume plethysmography in place of applanation tonometry^38–41); therefore, they are portable and capable of measuring ambulatory 24-h central BP. The N-point moving average method has also been proposed for the estimation of the aortic BP peak from the radial BP waveform⁴²⁾. These novel methodologies can provide more detailed information on central BP (e.g., circadian variation); further validation studies are required for general use in terms of accuracy, reproducibility, and prognostic significance.

Factors Affecting Central BP

1). Age (Arteriosclerosis)

Age is the most potent accelerating factor associated with arteriosclerosis, and it strongly influences aortic BP waveforms (Table 1). With advancing age, elastic fibers in the aortic medial layer gradually undergo degeneration and are replaced with stiffer collagenous fibers, and the resultant decrease in the medial elastin/collagen ratio causes progressive stiffening of the aortic wall and reduction in aortic compliance^{43, 44)}. Such a pathological process is usually referred to as arteriosclerosis. Hypertension accelerates this process because of the fatiguing effects of excessive cyclic stress on elastic fibers⁴³⁾. Aortic stiffening increases aortic PWV and shortens the return time of the reflected BP wave and thus increases the augmented pressure. Moreover, as aortic distensibility decreases with age, systolic upstroke and diastolic decay of aortic BP become steeper. All these changes are responsible for a marked widening of aortic PP with age (Table 1). It should be noted here that compared with central elastic arteries, peripheral muscular arteries undergo little arteriosclerotic change⁴⁵⁾; therefore, age-dependent PP change is more remarkable in the former⁴⁶⁾.

Table 1. Comparisons of aortic pulse waveform characteristics between subjects with and without aortosclerosis.

Aortosclerosis	(−)	(+)
Aortic pressure waveform*
Incident pressure wave	Low	High
Augmented pressure	Low	High
Pulse pressure	Narrow	Wide
Pulse wave velocity	Slow	Fast
Round-trip travel time	Long	Short
Augmentation index	Normal	High
Diastolic pressure decay	Slow	Fast
Pulse pressure amplification	Large	Small

^*Arrows indicate the arrival of reflected waves.

Epidemiological studies have shown progressive linear increases in aortic augmented pressure and reflected wave magnitude with age^{47, 48)}. Aortic AIx increases until the sixth decade but reaches a plateau or even decreases thereafter^{47, 49)}, probably because both the numerator (augmented pressure) and denominator (PP) increase linearly with age but with different intercepts on the pressure axis⁵⁰⁾.

2). Gender and Height

In general, women show a higher AIx than men⁵¹⁾. This is mainly attributed to their difference in body height⁵²⁾, which relates to the path length between the central aorta and peripheral reflecting sites and thus determines the timing of wave reflection. However, the association between gender and AIx can be observed even independently of body height, suggesting an intrinsic gender difference in arterial function⁵³⁾. Given a higher AIx in women, gender difference in aortic PP is less obvious than that in brachial PP⁵⁴⁾.

3). Heart Rate and Ejection Duration

AIx increases as the heart rate decreases⁵⁵⁾. This is because a longer ejection duration resulting from a lower heart rate causes a wider incident wave to overlap with a reflected wave, with the round-trip travel time of these two waves being little affected by the heart rate. It can also be a reason why nonvasodilating beta-blocker administration increases AIx⁵⁶⁾.

4). Lifestyle (Smoking, Alcohol, Caffeine, and Sodium Intake)

Smoking of cigarettes or cigars increases aortic AIx and PWV acutely and chronically^{57, 58)}. Chronic smokers had a higher aortic (but a similar brachial) systolic BP than nonsmokers⁵⁷⁾. Smoking-induced AIx increase correlates with the change in the plasma cotinine level⁵⁹⁾. Even short-term smoking cessation reduces AIx⁶⁰⁾.

Alcohol intake has divergent effects on AIx; acute ingestion reduces aortic systolic BP and AIx, while chronic excessive consumption elevates these in men⁶¹⁾. The effect of chronic alcohol consumption on aortic BP may differ by gender; heavier alcohol intake is associated with a higher AIx in healthy young men⁶²⁾ but not in postmenopausal women⁶³⁾.

Caffeine affects peripheral wave reflection to elevate aortic BP to a greater extent than that apparent from brachial BP⁶⁴⁾. Caffeinated but not decaffeinated coffee ingestion raises aortic AIx and PWV⁶⁵⁾. Black and green tea can also elevate AIx⁶⁶⁾. The combination of smoking and caffeine intake produces a synergistic detrimental effect on aortic PWV and AIx⁶⁷⁾.

Excessive sodium intake increases aortic stiffness and wave reflection, while sodium restriction improves them effectively^{68, 69)}.

5). Diabetes, Hypercholesterolemia, and Obesity

Inconsistent results have been reported with regard to AIx in diabetes. Pediatric patients with type 1 diabetes are reported to have an increased aortic AIx⁷⁰⁾, while adult patients with type 2 diabetes can have an increased⁷¹⁾, normal⁷²⁾, or even reduced⁷³⁾ AIx despite widened PP. The mechanism underlying these apparently conflicting observations is still debated, but some patients with type 2 diabetes may have compensatory hyperinsulinemia (in response to insulin resistance), which could induce peripheral vasodilation and diminish wave reflection, even with a stiffened aorta. It may also involve central (abdominal) obesity, which is known to reduce AIx⁷³⁾. Aortic AIx in patients with hypercholesterolemia is reported to be higher than⁷⁴⁾ or similar to⁷⁵⁾ that in subjects with normocholesterolemia. Metabolic syndrome may not be associated with AIx in treated hypertension⁷⁶⁾. Taking these reports together, metabolic disorders can have inconsistent effects on AIx and thus peripheral wave reflection, in contrast to a relatively consistent (increasing) effect on PWV and arterial stiffness⁷⁷⁾. It should be of note, however, that aortic BP elevation can often be observed in these disorders even without any increase in wave reflection, namely through an increase in the incident wave height.

Arterial stiffness and wave reflection may be influenced not only by glucose or lipid metabolism but also by mineral metabolism abnormalities. For instance, higher serum phosphorus levels are shown to be associated with increases in PWV and AIx even among the general population with normal kidney function⁷⁸⁾.

6). Other Related Diseases

An elevation in AIx is observed in hypothyroidism⁷⁹⁾ and hyperparathyroidism⁸⁰⁾, both of which represent the endocrine disorders considered responsible for hypertension. In obstructive sleep apnea syndrome, aortic (but not peripheral) BP and AIx can be elevated, particularly in the early morning⁸¹⁾. Aortic BP may decrease in response to nasal continuous positive airway pressure treatment even without parallel changes in peripheral BP⁸²⁾. In patients with Kawasaki disease who develop coronary artery lesions, wave reflection indices are elevated and the aortic waveforms resemble those observed in the elderly⁸³⁾. Primary inflammatory diseases, including rheumatoid arthritis, may be independently associated with increased central BP and AIx⁸⁴⁾, although it remains controversial whether systemic inflammation in primary hypertension and chronic kidney disease (CKD) is a cause or consequence of aortic stiffness⁸⁵⁾. Systemic sclerosis, which often involves microvascular lesions, is associated with enhanced peripheral wave reflection despite normal aortic stiffness⁸⁶⁾. Sickle cell disease with the hemoglobin SS genotype is associated with lower arterial stiffness and wave reflections owing to higher nitric oxide availability⁸⁷⁾.

Is Central BP Useful for Risk Stratification of CVD?

1). Prospective Studies

So far, a number of prospective observational cohort studies have been conducted to investigate the predictive ability of central BP indices for CVD events. The results are summarized in Table 2^88–106) and have been discussed elsewhere in our previous review¹⁰⁷⁾. These study cohorts can be classified according to the underlying diseases, e.g., patients with renal disease, patients with coronary heart disease, and the general population. As a whole, most (but not all) of currently available data indicate that central BP indices (such as central PP, AIx, and PP amplification) can predict CVD events more precisely than and/or independently of brachial BP, particularly in high-risk populations (Table 2; for more details, please refer to the previous review¹⁰⁷⁾).

Table 2. Prospective observational studies on predictive value of central blood pressure parameters for cardiovascular events.

First Author	Subjects	N	Male (%)	Age (years)*	Brachial BP (mmHg)†	Follow-up (years)*	Recording/Estimation	Endpoint	Event rate‡	Central BP parameter	Adjusted RR	Lower 95%CI	Upper 95%CI	Central vs. brachial
London88)	ESRD	180	60	54	156/83	4.3	Carotid	All-cause mortality CV mortality	89.7 51.3	AIx (/10%) AIx (/10%)	1.51 1.48	1.23 1.16	1.86 1.90	Sup, Ind Ind
Safar89)	ESRD	180	59	54	156/83	4.3	Carotid	All-cause mortality	89.7	PP (/25 mmHg) AMP (/16%)	1.40 0.50	1.10 0.30	1.80 0.80	Sup Sup
Covic90)	ESRD	92	54	43	129/84	5.1	Radial	All-cause mortality	32.1	AIx (> 24.6 vs. < 12.0%)	0.64	0.13	3.15	n/a
Briet91)	CKD	180	74	60	134/74	3.1	Carotid	Hemodialysis	73.5	PP (/10 mmHg)	1.24	1.03	1.49	n/a
Verbeke92)	Renal Tx	512	59	53	135/80	5.0	Radial	CV events	37.1	AugP (/8.6 mmHg)	1.49	1.22	1.81	Ind
Ueda93)	CAD	103	78	63	(139/72)	0.5	Intraaortic	Coronary restenosis	699.0	AIx (/10%) Tr (/-10 ms)	1.70 1.34	1.16 1.10	2.48 1.63	n/a n/a
Weber94)	CAD	262	71	66	132/78	2.0	Radial	CV events	116.0	AIx@Hr75 (/1%)	1.04	1.01	1.07	Sup
Chirinos95)	CAD	297	100	64	136/70	3.2	Intraaortic	CV events	132.3	AugP (/10 mmHg) AIx (/10%)	1.19 1.28	1.06 1.09	1.34 1.50	Sup Sup
Jankowski96)	CAD	1109	74	58	(135/72)	4.4	Intraaortic	CV events	50.5	PP (/18.2 mmHg)	1.28	1.10	1.49	Sup
Weber97)	CAD	520	100	64	133/80	4.1	Radial	CV events	80.1	AIx (/10%) AIx@Hr75 (/10%)	1.15 1.25	1.01 1.06	1.32 1.47	Ind Ind
Roman98)	General	2403	35	63	132/75	4.8	Radial	CV events	27.7	PP (/10 mmHg)	1.15	1.07	1.24	Sup
Roman99)	General	2405	35	63	132/75	5.6	Radial	CV events	25.5	PP (≥ 50 vs. ≤ 31 mmHg)	1.69	1.20	2.39	Sup
Pini100)	General	398	45	73	145/84	8.0	Carotid	CV events	38.3	PP (/10 mmHg)	1.23	1.10	1.37	Sup
Leone101)	General	3337	39	73	145/82	3.6	Nomogram	Coronary events	10.6	PP (/14.6 mmHg)	1.39	1.17	1.60	n/a
Benetos102)	General	125151	58	40	132/81	12.1	Nomogram	All-cause mortality CV mortality	2.6 0.5	PP (/7.8 mmHg) PP (/7.8 mmHg)	1.18 1.21	1.12 1.15	1.25 1.28	Sup Sup
Wang103)	General	1272	53	52	139/88	10.8	Carotid	CV mortality	2.7	PP (/10 mmHg)	1.26	1.02	1.56	Sup
Wang104)	General	1272	53	52	139/88	15.0	Carotid	CV mortality	3.3	Pb (/6 mmHg)	1.59	1.21	2.09	Sup, Ind
Huang105)	General	1014	54	52	139/89	15.0	Carotid	All-cause mortality	13.2	PP (/1 mmHg)	1.16	1.01	1.32	Sup
Mitchell106)	General	2232	42	63	127/74	7.8	Carotid	CV events	8.7	PP (/1 mmHg) AIx (/1%)	1.00 0.91	0.99 0.77	1.01 1.07	n/a n/a

AIx: augmentation index; AIx@Hr75: AIx adjusted for heart rate of 75 bpm; AMP: pulse amplification; AugP: augmented pressure; BP: blood pressure; CAD: coronary artery disease (or suspect); CI: confidence interval; CKD: chronic kidney disease; CV: cardiovascular; ESRD: end-stage renal disease; Ind: predictive independency of central BP parameters from brachial BP; n/a: data not available; Pb: reflected wave amplitude; PP: pulse pressure; RR: relative risk; Sup: predictive superiority of central BP parameters over brachial BP; Tr: travel time; Tx: transplantation.

^*Mean or median.

^†Except for aortic BP in parentheses.

^‡Per 1000 person-year (by estimate). Reproduced from Hashimoto et al.¹⁰⁷⁾.

2). Cut-Off Value

The definition of the cut-off values for central BP has not yet been established. While the normal upper limits of brachial (casual) BP are usually defined as 140/90 mmHg, it is difficult to set corresponding thresholds for central BP because of the PP amplification, which considerably varies among individuals according to age and BP levels¹²⁾. Based on a distribution approach from an epidemiological cross-sectional database, brachial BP values of 140/90 mmHg correspond to central BP values of ∼125/90 mmHg¹⁰⁸⁾. Based on a prognostic outcome-driven approach from a prospective database, central BP of 130/90 mmHg has been proposed as a diagnostic threshold value¹⁰⁹⁾. Clearly, the validity of these cut-off values needs to be confirmed by further investigation.

Does Central BP Serve as a Surrogate Marker for Intervention?

Brachial BP has long been recognized as a useful marker for the management of CVD as well as hypertension. For introducing central BP measurement into routine clinical practice, the most critical issue is to determine whether central BP is superior to brachial BP in evaluating therapeutic effects and predicting their consequences.

It is known that nitroglycerine has a stronger hypotensive effect on central BP than that apparent from brachial BP measurement¹¹⁰⁾. This is attributable to a reduction in wave reflection resulting from the peripheral vasodilating action of nitroglycerine. The opposite effects are observed for nonvasodilating β-blockers, which lower central BP to a lesser extent than brachial BP¹¹¹⁾. So far, different effects of various antihypertensive drug classes on central BP have been demonstrated by several randomized multicenter clinical trials. For instance, the CAFÉ substudy⁵⁶⁾ of the ASCOT trial¹¹²⁾ showed that individuals assigned to the amlodipine/perindopril group had a 4.3 mmHg lower central systolic BP than those assigned to the atenolol/thiazide group, despite a similar brachial BP. The EXPLOR study¹¹³⁾ demonstrated that an amlodipine/valsartan combination decreases central BP and AIx more effectively than an amlodipine/atenolol combination. In addition, the PARAMETER study has recently shown that sacubitril/valsartan (LCZ696), an angiotensin receptor neprilysin inhibitor, is more potent than olmesartan in reducing central BP¹¹⁴⁾.

Central BP can serve as a therapeutic surrogate marker for the prevention of major CVD events. Data from the abovementioned CAFÉ study indicate that the modest intergroup difference in central BP, which is undetectable by conventional brachial BP measurement, contributed to the better cardiovascular outcome in the amlodipine group observed in the ASCOT study¹¹²⁾. In addition, compared with brachial BP, central BP can better predict the regression of target organ damage (e.g., left ventricular hypertrophy¹¹⁵⁾ and carotid artery wall hypertrophy¹¹⁶⁾) during antihypertensive treatment. Such regression, which likely results from vasolidator-induced reduction in peripheral wave reflection^{117, 118)}, could be closely linked to a reduction in CVD events. These lines of evidence suggest a potential benefit of central BP-guided antihypertensive therapy for CVD management.

At present, only limited data are available regarding the therapeutic impact of medication other than antihypertensive drugs (e.g., statin and antidiabetic drugs). In the CAFÉ-LLA substudy¹¹⁹⁾, atorvastatin therapy sufficient to reduce CVD events in the ASCOT trial¹¹²⁾ had no influence on central BP or AIx. In contrast, in a placebo-controlled study¹²⁰⁾, atorvastatin but not the placebo induced significant reductions in central BP, AIx and PWV. Linagliptin (a dipeptidyl peptidase-4 inhibitor) did not affect central BP and AIx in a placebo-controlled study of patients with diabetes¹²¹⁾. Metformin (but not the placebo) reduces central BP and AIx in patients with nonalcoholic fatty liver disease¹²²⁾ and in young women with polycystic ovary syndrome¹²³⁾. Tetrahydrobiopterin, a nitric oxide synthase cofactor, can lower central AIx but not PWV in patients with CKD¹²⁴⁾. Anti-inflammatory treatment may reduce PWV and/or AIx in some patients with primary inflammatory disorders¹²⁵⁾. Obviously, further investigation is required.

Pathophysiological Relations of Central Hemodynamics to Arteriosclerotic Diseases: Hypothetical View and Future Directions

As the central aorta stiffens, peripheral endorgans undergo progressive degeneration in the vascular structure and gradual deterioration in the physiological function, which can lead eventually to the sudden occurrence of symptomatic CVD. There is substantial evidence showing that aortic PWV predicts major cardiovascular events, including myocardial infarction, heart failure, stroke, and renal failure^{126, 127)}. Until quite recently, however, relatively little was known about the underlying mechanisms by which aortic stiffening causes target organ damage or dysfunction. Results of the latest studies indicate that the central hemodynamics, including aortic BP and flow, plays a pivotal role in the causal mechanisms.

Aortic pulsatile flow is an important constituent of central hemodynamics. The flow waveform differs from the pressure waveform and greatly varies among different aortic sites¹²⁸⁾. According to previous studies based on Doppler ultrasound, the flow waveform in the proximal descending aorta is biphasic (to-and-fro) comprising systolic forward and diastolic reverse components, and the reverse/forward flow ratio linearly increases with aortic stiffening¹²⁹⁾. Such an increase in aortic flow reversal could alter the blood flow distribution to various organs and thus directly relate to the pathophysiological mechanisms of arteriosclerotic CVD (see below). The bidirectional (i.e., antegrade and retrograde) nature of aortic flow seems quite intriguing from a physiological view, given its potential relevance to aortic BP composed of the two waves going in opposite directions (i.e., forward and reflected waves). The PP amplification (i.e., pressure gradient) and aortic/peripheral PWV ratio (i.e., stiffness gradient) also seem to be significant determinants of pulsatile flow production^130–132).

1). Stroke

Cerebrovascular disease can be broadly divided into two categories: microvascular and macrovascular stroke (Table 3). Cerebral microvascular diseases include lacunar infarction, white matter hyper-intensity lesions, and intracerebral hemorrhage (including microbleeds), which are often observed in the regions perfused by deep perforating arteries. Several previous studies have associated these microvascular lesions with elevations in aortic PWV and BP^133–137). The underlying mechanisms are likely that 1) the small perforating arteries are directly branched from the large cerebral arteries and 2) BP and flow pulsations are prone to deeply transmitting into the cerebral microvasculature on account of its low impedance properties^{132, 138)}. It is thought that with aortic stiffening, BP pulsation in large arteries becomes greater; thus, higher pulsatile stress is imposed on the fragile microvascular walls, which results in microvascular injury and eventually causes ischemic or hemorrhagic stroke.

Table 3. Different characteristics of micro- and macrovascular stroke.

Stroke types	Microvascular	Macrovascular
Designations	Lacunar infarction Hemorrhagic stroke (including microbleeds)	Cortical infarction
Main causes	Lipohyalinosis	Atherosclerosis Thromboembolism
Vascular territories	Small deep perforating arteries	Cerebral/carotid/vertebral arteries
Association with aortic stiffness	+	+
Suggested central hemodynamic mechanisms	High aortic pressure	High aortic pressure Aortic flow reversal

In contrast to microvascular stroke, macrovascular stroke usually occurs in the regions perfused by large cortical arteries and could be fatal in some cases. Studies have demonstrated that aortic PWV is an independent predictor of fatal stroke¹³⁹⁾. The underlying hemodynamic mechanisms linking aortic stiffness to macrovascular stroke may include 1) cerebral thromboembolism due to plaque rupture in the carotid/cerebral arteries and 2) cardiogenic embolism due to atrial fibrillation, both of which are triggered by widened central PP and/or enhanced peripheral wave reflection^{140, 141)}. In addition, another mechanism has recently been proposed: 3) retrograde plaque embolism due to the abovementioned aortic flow reversal¹²⁹⁾. In particular, the term “retrograde embolism” means that atherosclerotic mobile plaques are detached from the aortic arch or descending aortic walls and then retrogradely delivered through the supra-aortic (i.e., carotid and subclavian) arteries up to cerebral arteries to cause “cryptogenic” stroke. This type of stroke can indeed occur¹⁴²⁾ because diastolic reverse flow in the descending aorta directly contributes to forward flow into the supra-aortic arteries. In addition, the risk of the retrograde embolism may be heightened by aortic stiffening because the stiffer the aorta is, the greater the aortic flow reversal becomes¹²⁹⁾. The rationale for this retrograde embolism also includes the facts that 1) cryptogenic stroke can occur without any embolic sources in the heart or ascending aorta and 2) mobile plaques are more often seen in the descending aorta than in the ascending aorta¹⁴³⁾.

2). CKD

CKD is defined as the presence of (micro)albuminuria and/or a reduced glomerular filtration rate (GFR). Both albuminuria and GFR reduction are often preceded by aortic stiffening, but they can predict future CVD events independently of each other¹⁴⁴⁾. This indicates the existence of at least partially different etiologies underlying these distinct entities. Recent investigation does suggest different mediating influences of central hemodynamic (i.e., BP and flow) abnormalities on these aorto-renal associations.

The kidney has a unique structure and function. The renal arteries branch off from the abdominal aorta into intrarenal smaller arteries, leading to tiny afferent arterioles and fragile glomerular capillaries. Intraglomerular pressure is as high as 60/40 mmHg even under normal conditions, reflecting the low impedance properties of the renal vasculature. In case of aortosclerosis, the central PP widens, and then, the high BP pulsation is transmitted down to the glomerular capillaries. This most likely results in intraglomerular hypertension and endothelial injury, leading to albuminuria¹³²⁾. In fact, central PP and PWV have been shown to be independent predictors of the urinary albumin/creatinine ratio^{131, 145)}.

Aortic stiffness is also known to be associated with GFR decline^{146, 147)}. The GFR decline may result from such glomerular injury, as mentioned above, but another possibility has recently been raised that aortic flow abnormality is a primary cause of renal dysfunction accompanying aortosclerosis. As mentioned earlier, aortic stiffening increases flow reversal in the thoracic descending aorta and thus reduces forward flow toward the suprarenal abdominal aorta¹²⁹⁾. This reduction in suprarenal aortic flow (also caused by an impaired Windkessel function) could reduce blood inflow into the renal arteries (i.e., renal blood flow) and thereby deteriorate GFR¹⁴⁸⁾. This flow mechanism could also explain why renal function decreases with age in the general population (i.e., even in the absence of albuminuria).

3). Heart Disease

For over a half century, numerous studies have shown an important etiological role of central hemodynamics in cardiac diseases. The details have been extensively reviewed previously^{43, 149)}. In brief, an elevation in central BP, which directly reflects left ventricular afterload, is primarily responsible for left ventricular hypertrophy^{150, 151)}, causing various cardiac dysfunctions (e.g., heart failure and arrhythmia). Hastening of exponential decay of central BP during diastole, which is accelerated by aortic stiffening, reduces the diastolic inflow from the ascending aorta to the coronary arteries and thereby predisposes hypertensive patients to myocardial ischemia¹⁵²⁾. The risk of myocardial ischemia could be further heightened by the presence of atherosclerotic coronary artery stenosis¹⁵³⁾ and/or left ventricular hypertrophy with increased oxygen demand¹⁵⁴⁾. Obviously, more studies are needed to investigate the crucial link between aortic and coronary pulsatile flow.

Conclusions

The quantitative assessment of central pulsatile hemodynamics has now been enabled by noninvasive tonometry/ultrasound and pulse waveform analysis. Central hemodynamic indices have some advantages in terms of 1) the technical facility for repeated measurements, 2) the immediate response to antihypertensive treatment, and 3) the predictive ability of therapeutic consequences (e.g., preventive effects on CVD). Recent international guidelines or recommendations for vascular biomarkers have also endorsed their usefulness^{155, 156)}. In the future, for wider application to clinical practice, further investigation is needed to determine the normal reference values, develop handy apparatuses with high precision and dedicated analytical programs, and verify the epidemiological evidence and mechanistic rationale for general use in CVD risk management.

Conflict of Interest

The author belonged to an endowment department supported with unrestricted grants from Omron Healthcare and MSD.