Vasodilators, Aortic Elasticity, and Ventricular End-Systolic Stress in Nonanesthetized, Unrestrained Rats
Abstract We evaluated the effect of different vasodilators on ventricular end-systolic stress by investigating the impact of sodium nitroprusside, nifedipine, and hydralazine on blood pressure, aortic stiffness, and wave reflection during drug-induced hypotension (to 80 mm Hg mean blood pressure) in normotensive (central aortic mean blood pressure, 116 to 119 mm Hg; systolic pressure, 133 to 137 mm Hg), nonanesthetized, unrestrained rats. Aortic stiffness was evaluated from the slope of the linear regression relating pulse wave velocity (PWV) to central aortic mean or pulse pressure. The fall in central aortic systolic blood pressure was less than the fall in mean pressure, especially after hydralazine (122±4 mm Hg; sodium nitroprusside, 107±2; and nifedipine, 112±3 mm Hg; P<.05). The PWV/mean pressure slope was linear, positive, and similar in all three groups (hydralazine, 3.3±0.2; sodium nitroprusside, 3.8±0.3; and nifedipine, 3.9±0.3 cm · s−1 · mm Hg−1; P>.05). The PWV/pulse pressure slope was linear, negative, and less steep in the case of hydralazine (−4.9±0.6; sodium nitroprusside, −15.5±3.7; and nifedipine, −13.5±2.9 cm · s−1 · mm Hg−1; P<.05). The travel time and augmentation index of the reflected wave were similar in all groups. In conclusion, sodium nitroprusside and nifedipine had a more beneficial effect on end-systolic stress than did hydralazine. This does not appear to be related to any specific effect on wave reflection or the “static” relationship between PWV and aortic mean blood pressure; it may be related to the effects of these drugs on the “dynamic” relationship between PWV and pulse pressure.
As the aorta becomes stiffer with age,1 compliance decreases, and wave reflection occurs earlier.2 Thus, left ventricular end-systolic stress increases, leading to left ventricular hypertrophy and dysfunction.3 4 Drugs such as nitrates and dihydropyridine calcium channel blockers, which increase small peripheral arterial caliber and distensibility, can specifically reduce the effects of early wave reflection and lower central aortic systolic pressure.4 5 6 7 8 Other drugs, such as hydralazine, which reduces arteriolar peripheral resistance, may also indirectly modify wave reflection (by lowering aortic intraluminal pressure and hence pulse wave velocity [PWV]), but their effect on pulsatility is less pronounced. Thus, it has been shown in spontaneously hypertensive rats that chronic treatment with hydralazine increases characteristic impedance and the amplitude of the forward wave and is hence unable to regress left ventricular hypertrophy in spite of effective control of blood pressure.9
In this article, we examine the effects of sodium nitroprusside, nifedipine, and hydralazine on central aortic systolic pressure, wall stiffness, and the pulse pressure (PP) contour in normotensive, nonanesthetized, unrestrained rats. Static aortic isobaric elasticity was evaluated from the slope, relating thoraco-abdominal PWV to central aortic mean blood pressure,10 11 12 13 following a drug-induced decrease in blood pressure from normotension (120 mm Hg) to 80 mm Hg. Changes in the viscoelastic properties of the aortic wall were evaluated from changes in dynamic aortic isobaric elasticity (PWV versus central aortic pulse pressure) during drug-induced hypotension. We use the term “isobaric elasticity” in the same sense as “elastic modulus,” ie, an increase in isobaric elasticity denotes an increase in wall stiffness. The timing and amplitude of wave reflection were evaluated according to the method described by Murgo et al.14
Twenty-four adult, normotensive, male Wistar rats (451±5 g, Ico: Wi, IOPS AF/Han; Iffa Credo, 69210 l’Arbresle, France) were given a rodent diet (UAR, Villemoisson sur Orge, France) and water ad libitum for 1 week and then were separated into three groups (sodium nitroprusside, n=9; nifedipine, n=8; hydralazine, n=7; see below). Eight age- and body weight–matched controls received vehicle. Experiments were performed in accordance with the guidelines of the European Union and the French Ministry of Agriculture.
Baseline Aortic Blood Pressure, PWV, and Wave Reflection in Nonanesthetized, Unrestrained Rats
This technique has been described in detail elsewhere.10 11 12 13 Polyethylene cannulas were introduced with rats under halo-thane/oxygen anesthesia into the descending thoracic aorta (central aortic blood pressure) and abdominal aorta (peripheral aortic blood pressure) (systolic, diastolic, mean, and pulse; mm Hg). A femoral venous cannula was implanted for drug infusion. Animals were allowed 24 hours to recover, during which time food and water consumption were not significantly different than before instrumentation.
Twenty-four hours later, the aortic cannulas were filled with heparinized, gas-free 0.15 mol · L−1 NaCl and connected to low-volume pressure transducers (Baxter, Bentley Laboratories Europe). The dynamic frequency response15 of the whole system (cannula filled with gas-free saline solution+transducer+amplifier) was flat up to 15 Hz and then slightly underdamped from 15 to 30 Hz (maximal value+15±2% at 30 Hz). Phase lag was slightly different from 0 at 30 Hz (−6.2±0.6%). In a separate experiment (n=60 observations) following recording of the pressure waveforms with rats under anesthesia with the cannula systems, the cannula systems were removed and replaced by Millar Mikro-tip transducers (0.67 mm outside diameter, SPR 407, Millar Instruments), and pressure recording was repeated. In one half of the animals, the order was reversed (Millar then cannula). Pressure waveforms were very similar (Fig 1⇓), and wave reflection parameters were not significantly different (results not shown).
Signals were amplified, converted into digital form, and recorded at a sampling rate of 256 Hz (approximately 40 sampling points per pressure wave) by a computer (PC 4-486-66/L, Dell Computer). An algorithm detected the minimum (diastolic) and maximum (systolic) of each waveform and calculated mean pressure on the basis of the area under the waveform. Pulse amplification (peripheral/central PP) was calculated. Heart rate (beats per minute) was determined by counting the number of pressure cycles. Transit times between the two pressure signals were determined by a second algorithm that shifted the peripheral signal in time with respect to the central signal and determined the value of the time shift, giving the highest correlation between the two signals.10 13 Measurements were performed for 4-second periods (approximately 26 heartbeats and 3 respiratory cycles), repeated once per minute during 15 minutes, and averaged. PWV (cm · s−1) was calculated as the distance between the two cannula tips divided by the transit time. The distance between the two cannula tips was determined in situ after postmortem dissection. A damp cotton thread was stuck onto the aorta between the tips of the two cannulas, which were marked on the thread. The thread was then removed and laid straight for measurement of the distance between the two marks (8.8±0.1 cm or 19.5±0.2 cm · kg−1, n=24).
Wave reflections were measured on the central pressure signal only. All pressure waveforms were of type A as defined by Murgo et al.14 The inflection point of the reflected wave was detected by simple inspection, and the following parameters were calculated (Fig 1⇑): (1) the height from the shoulder of the reflected wave to the systolic peak (ΔP, mm Hg) and (2) the augmentation index is the ratio (%) of ΔP to PP. The latter is an estimate of the intensity of wave reflection. The travel time of the reflected wave (time from the foot of the pressure wave to the shoulder [Δtp, milliseconds]) and the left ventricular ejection time (time from the foot of the pressure wave to the diastolic incisura; left ventricular ejection time [LVET], milliseconds) were also measured. The effective reflection site distance (Lp, cm) was calculated as PWV×tp/2.14 At baseline, the PWV measurement was used; at 100 and 80 mm Hg, PWV was calculated from the linear regression of PWV versus central aortic mean blood pressure (see below). The time resolution of the wave reflection parameters was equal to the speed of data acquisition (3.9 milliseconds). Because Δtp varied between 14 and 19 milliseconds, the error was 21% to 28%.
Aortic Isobaric Elasticity and Wave Reflection in Nonanesthetized, Unrestrained Rats After Drug-Induced Hypotension
After baseline measurements had been made, rats were infused with sodium nitroprusside (2.3 mmol · L−1), nifedipine (1.5 mmol · L−1), or hydralazine (152 mmol · L−1). Central aortic mean blood pressure was lowered from baseline (120 mm Hg) to 80 mm Hg by stepwise (six steps) increases in infusion rate. After each increase in volume, aortic blood pressure fell rapidly and then stabilized. The total duration and mean dose of infusion were for sodium nitroprusside 24±2 and 107, nifedipine 29±2 and 107, and hydralazine 23±2 minutes and 8090 nmol · kg−1 · min−1, respectively. Infusion rates were chosen on the basis of preliminary experiments in which at least three rates (n=6 animals per rate) for each drug were tested (results not shown). Animals received a total blood volume of 4%. At each step, 15 measurements (≈400 heartbeats) were performed. PWV was expressed as a function of central aortic mean blood pressure or central aortic PP (Fig 2⇓). Slopes andintercepts were treated as independent parametric variables and averaged (technique based on Gosling et al16 17 and Kawasaki et al18 ). Wave reflections were measured at central mean aortic pressures of 100 and 80 mm Hg.
Sodium nitroprusside, nifedipine, and hydralazine were purchased from Sigma Chemical Co. Nitroprusside and hydralazine were dissolved in phosphate-buffered distilled water (10 mmol · L−1, pH 7.4, at 25°C; Sigma). Nifedipine was dissolved in the same phosphate buffer (50%) plus 40% polyoxyethylene glycol 400 (Coopération Pharmaceutique Française) and 10% absolute ethanol (Farmitalia Carlo Erba). Controls received the nifedipine solvent.
Each variable showed a continuous unimodal distribution (results not shown). Values are given as mean±SEM. Linear ANOVA was performed using standard techniques, and results are expressed as slope and intercept. Differences were evaluated by ANOVA plus the Scheffé test. A value of P<.05 was chosen as indicative of statistical significance.
Changes in Aortic Blood Pressure, Isobaric Elasticity, and Wave Reflection After Drug-Induced Hypotension
There were no differences in baseline aortic blood pressures, PWV, or wave reflection between the groups (Tables 1⇓ and 4⇓). Vehicle did not modify central mean aortic blood pressure (initial value, 118±3 mm Hg; final value, 119±3 mm Hg), PWV (initial value, 700±2 cm · s−1; final value, 699±22 cm · s−1), or wave reflection (results not shown).
Each drug was administered at a dose producing a fall in central and peripheral aortic mean blood pressures of −28% to −34% (Table 2⇓ compared with Table 1⇑). Falls in diastolic blood pressure were of the same order and similar in the three groups. Drugs (especially hydralazine) induced a proportionally smaller fall in central aortic systolic pressure, and hence pulse amplification decreased.
Slopes and intercepts relating PWV to central aortic mean blood pressure were similar in all three groups; the slope relating PWV to central aortic pulse pressure was less steep after hydralazine (Table 3⇓, Fig 2⇑).
The reflected wave inflection point was undetectable in 8%, 4%, and 12% of the pressure signals studied at 120, 100, and 80 mm Hg, respectively. None of the three vasodilators modified the timing (Δtp) of wave reflection (Table 4⇓). LVET tended to decrease as reflex tachycardia developed (Table 2⇑). Pulse pressure increased markedly after hydralazine. Augmentation index increased after sodium nitroprusside. The effective reflection site distance was similar in all groups and at all pressures.
Our results show that in rats, as in humans, acute administration of different vasodilators has different effects on central aortic systolic blood pressure. Sodium nitroprusside and nifedipine lower central aortic systolic blood pressure (and hence presumably left ventricular end-systolic stress) to a greater extent than does hydralazine. However, the mechanisms behind such differences are different in the two species. In humans, vasodilators such as nitrates lower central aortic systolic blood pressure because they diminish the effect of wave reflection; for instance, the augmentation index is halved after following nitroglycerine administration.7 8 In rats, we found that the augmentation index increased after sodium nitroprusside (and did not decrease with the two other vasodilators).
Wave Reflection and Reflection Site Distance
Any conclusion on a species-specific drug effect on wave reflection depends on an accurate determination of the inflection point of the reflected wave. Although the dynamic frequency response of our system was reasonably good up to the 4th or the 5th harmonic, one could wonder whether such a system can accurately record the rising limb of the waveform that has high harmonic components. We think that this is the case because recordings made using Millar Mikro-tip transducers showed a very similar waveform with a clear inflection on the rising limb. This is at odds with curves shown by others19 in rats of similar size and species, where there was little or no evidence of systolic inflection. This point merits further investigation. In our results, the calculated effective reflection site distance places the site of wave reflection at 4 to 5 cm from the recorded site, ie, at the level of the renal arteries. This is interesting because some authors have suggested that in humans, the aorta tapers at this point, thereby producing a reflection site.20 We examined this possibility in the rat using in situ fixation under pressure. Although there was a decrease in both internal diameter (1.79±0.04 to 0.89±0.02 mm, n=5, P<.05) and medial thickness (88±2 to 57±1 μm, P<.05, unpublished results) from the 1st (just below the left carotid artery ostium) to the 8th (iliac bifurcation site) cm of the aorta, changes were gradual, and there was no evidence of any abrupt change in either parameter at the 5th or 6th cm. There was, however, an abrupt fall in the amount of the elastin-specific amino acids, desmosine and isodesmosine, at this level (from 503±51 in the thoracic aorta to 341±27 μg · g−1 wet wt at the 6th cm, P<.05; unpublished results, Marque et al, 1996). This can probably be interpreted as an abrupt change in the wall elastin content and in wall stiffness at this point. Such a change could constitute a reflection site.
Speed of the Wave Reflection
It is paradoxical that the reflected wave was not slowed down despite the decrease in PWV. It is possible that the precision of the method (3.9 milliseconds for a time of 14 to 19 milliseconds) was not sufficient to allow small changes in the timing of wave reflection to be detected. When timing (Δtp) was expressed as a percentage of LVET, it tended to increase (from 27±3% to 36±4% and 34±4%, from 25±2% to 26±2% and 27±1%, and from 26±2% to 32±3% and 28±2%, in sodium nitroprusside, nifedipine, and hydralazine groups, from baseline to 100 and 80 mm Hg, respectively). This result is consistent with a reduction in PWV. However, it could also be explained by a reduction in LVET following reflex tachycardia. Whatever the mechanism, the increase in Δtp/LVET does not prevent the reflected wave adding to the incident wave in the systolic part of the pressure waveform, thereby increasing left ventricular end-systolic pressure.
It is possible that the structure of the arterial network in the rat, especially the short effective reflection site distance, is such that the wave is reflected extremely quickly under baseline conditions; therefore, it is impossible to significantly reduce the timing of wave reflection. The rat, even at a very young age when presumably aortic wall elasticity is at its highest,2 has a central pressure waveform of type A as defined by Murgo et al14 (reflected wave in the early part of the systole and/or augmentation index >12%).
Lack of a Specific Drug Effect on Static Aortic Isobaric Elasticity
Over the mean blood pressure range of 120 to 80 mm Hg, static aortic isobaric elasticity was similar whatever the vasodilator used, suggesting that in the rat at least, all three drugs increased static aortic elasticity by a simple reduction of intraluminal distending pressure. These results are in contrast to those of Safar et al,4 who showed in the brachial artery of humans that after a similar reduction in systemic mean arterial blood pressure elicited by infusion of nitroglycerin, nifedipine, or dihydralazine, arterial diameter decreased in parallel with pressure in the case of dihydralazine, whereas it increased as pressure fell with nitroglycerin and nifedipine. This was associated with greater improvement of arterial elasticity after nitroglycerin and nifedipine infusion than after dihydralazine infusion. They concluded that nitrates and calcium entry blockers increase elasticity by a local decrease in smooth muscle cell tone of the arterial wall6 in addition to a reduction in intraluminal pressure. Similar results have been obtained in a large-order branch artery of the coronary bed in conscious dogs.21 Coronary arterial diameter increased despite a concomitant reduction in mean blood pressure after nitroglycerin administration, implying once again that localized vasodilation can occur and that this decrease in smooth muscle cell tone could lead to a pressure-independent decrease in wall stiffness. These studies and our own were acute; therefore, possible differences in structural effects can be ruled out. Measurements were performed at stable aortic mean blood pressure values; thus, in all studies wall elastic conditions were presumably stable. The doses we used (sodium nitroprusside, 0.8; nifedipine, 1.1; and hydralazine, 16.5 mg · kg−1) were greater than those used by Safar et al4 (nitrates, 0.3; calcium entry-blockers, 0.4; and dihydralazine, 0.3 mg · kg−1) or Vatner et al21 (nitroglycerin, 25 μg · kg−1). Thus, the use of doses that were too low in our experiments can also be ruled out.
A more plausible explanation may reside in the type of artery studied. We measured PWV along the aorta (thoracic and abdominal), which is a less muscular, more elastic artery than either the brachial or the coronary. The elastin/smooth muscle ratio in the thoracic aorta (0.87) is approximately twice that found in the coronary artery (0.44).22 23 Our results confirm those of van Gorp et al.24 In the thoracic aorta, they showed that infusion of methoxamine produced parallel increases in diastolic aortic blood pressure (from 105 to 145 mm Hg) and end-diastolic aortic diameter and a decrease in absolute wall distension, suggesting that changes in local smooth muscle tone do not modify aortic wall elasticity. Other reports also suggest that smooth muscle cells do not participate in determining aortic wall elasticity under normotensive or hypotensive conditions.25 26 In the latter studies, activation of smooth muscle cells did modify aortic elastic properties but only at aortic pressures >130 mm Hg.
In the above paragraph, it is assumed that in situations where arterial diameter does not passively follow intraluminal pressure, local smooth muscle tone changes, and this modifies elasticity. However, if diameter were to increase (after acute administration of vasodilator), then wall stress would increase, and this would increase wall stiffness.27 It could also be argued that if a specific vasodilator were to relax the aortic smooth muscle cell,5 this event would not necessarily change diameter2 but would transfer wall strain to elastin fibers.28
Intraluminal Pressure as the Main Determinant of Aortic Stiffness
In light of these remarks, we suggest that the three vasodilators used had similar effects on aortic elasticity and that this is linked to their effects on intraluminal pressure. The hypothesis that this is caused by peripheral vasodilation and that the drugs have no direct effect on aortic smooth muscle tone can only be tested by direct measurement of aortic diameter by echo-tracking. However, other observations suggest that intraluminal pressure is the main determinant of aortic elasticity, as suggested by Nichols and O’Rourke.2 7 8 Thus, in this experiment, the values for static aortic isobaric elasticity obtained in nonanesthetized, unrestrained rats after a decrease in aortic mean blood pressure elicited by infusion of vasodilators were similar to those obtained in the pithed rat following a phenylephrine-evoked increase in aortic blood pressure (3.3 to 4.110 11 12 13 ), although the mechanisms used to vary aortic blood pressure were very different. The values for static aortic isobaric elasticity obtained in these latter studies (and in the present one) were slightly lower than those that we reported in our first publication on this method (5.7±0.7 cm · s−1 · mm Hg−1).11 This difference may be explained by the fact that carotid-femoral PWV was measured in the study by Makki et al11 instead of thoraco-abdominal aortic PWV, as was done in the present study and others. The carotid and femoral arteries have higher pulse wave velocities than the aorta.1 Furthermore, in awake sheep in which mean aortic blood pressure was modified by infusion of norepinephrine, angiotensin II, or sodium nitroprusside,29 the calculated aortic isobaric elasticity was 4 cm · s−1 · mm Hg−1, a value very similar to that reported here.
In conclusion, in rats as in humans, sodium nitroprusside and nifedipine had a more beneficial effect on central systolic aortic blood pressure (and presumably on end-systolic stress) than did hydralazine. The mechanism involved is probably different in the two species. In humans, it has been proposed that certain drugs (such as nitrates) attenuate the effects of wave reflection with a reduction in augmentation index. In rats, none of the drugs reduced augmentation index. The differences between the effects of the vasodilators on central systolic aortic blood pressure in the rat are not related to their effect on static isobaric elasticity but may be related to their specific effects on stroke volume19 30 31 and/or dynamic isobaric elasticity. These latter possibilities remain to be further investigated.
We acknowledge a grant from the French Ministry of Education and Research (JE 250 DRED) and financial support from LIPHA Pharmaceutical Company, Lyon, France. The Nancy group is part of the BioMed network “EureCa.”
- Received December 16, 1996.
- Revision received January 15, 1997.
- Accepted April 24, 1997.
Milnor WR. Hemodynamics. Baltimore: Williams & Wilkins; 1989;235:77-79.
Nichols WW, O’Rourke MF. McDonald’s Blood Flow in Arteries. London: Edward Arnold; 1990:398-437.
Safar ME, Toto-Moukouo JJ, Bouthier JA, Asmar RE, Levenson JA, Simon AC, London GM. Arterial dynamics, cardiac hypertrophy, and antihypertensive treatment. Circulation. 1987;75(suppl 1):I-156-I-161.
Safar ME, Bouthier JA, Levenson JA, Simon AC. Peripheral large arteries and the response to antihypertensive treatment. Hypertension. 1983;5(suppl 3):63-68.
O’Rourke MF, Kelly R, Avolio A. The Arterial Pulse. Philadelphia, Pa: Lea & Febiger; 1992.
O’Rourke M. Arterial stiffness, systolic blood pressure, and logical treatment of arterial hypertension. Hypertension.. 1990;15:339-347.
Mitchell GF, Pfeffer MA, Pfeffer JM. Equipotent antihypertensive agents variously affect pulsatile hemodynamics and regression of cardiac hypertrophy in spontaneously hypertensive rats. Hypertension.. 1991;18:417. Abstract.
Niederhoffer N, Lartaud-Idjouadiene I, Giummelly P, Duvivier C, Peslin R, Atkinson J. Calcification of medial elastic fibers and aortic elasticity. Hypertension.. 1997;29:999-1006.
Tatchum-Talom R, Niederhoffer N, Amin F, Makki T, Tankosic P, Atkinson J. Aortic stiffness and left ventricular mass in a rat model of isolated systolic hypertension. Hypertension.. 1995;26:963-970.
Amin F, Niederhoffer N, Tatchum-Talom R, Makki T, Guillou J, Tankosic P, Atkinson J. A new technique for study of impact of arterial elasticity on left ventricular mass in rats. Am J Physiol.. 1996;270:H981-H987.
Murgo JP, Westerhof N, Giolma JP, Altobelli SA. Aortic input impedance in normal man: relationship to pressure wave forms. Circulation.. 1980;62:105-116.
Peslin R, Fredberg JJ. Oscillation mechanics of the respiratory system. In: Macklem PT, Meads J, eds. Handbook of Physiology, Section 3, Vol III: Mechanics of breathing. Bethesda: American Physiological Society; 1986:145-177.
Mitchell GF, Pfeffer MA, Westerhof N, Pfeffer JM. Measurement of aortic input impedance in rats. Am J Physiol.. 1994;267:H1907-H1915.
Latham RD, Westerhof N, Sipkema P, Rubal BJ, Reuderink P, Murgo JP. Regional wave travel and reflections along the human aorta: a study with six simultaneous micromanometric pressures. Circulation.. 1985;72:1257-1269.
Vatner SF, Pagani M, Manders T, Pasipoularides AD. Alpha adrenergic vasoconstriction and nitroglycerin vasodilation of large coronary arteries in the conscious dog. J Clin Invest.. 1980;65:5-14.
Apter J. Correlation of visco-elastic properties of large arteries with microscopic structure. Circ Res.. 1966;19:104-121.
Fischer GM, Llaurado JG. Collagen and elastin content in canine arteries selected from functionally different vascular beds. Circ Res.. 1966;19:394-399.
Van Gorp AD, van Ingen-Schenau DS, Willigers J, Hoeks APG, De Mey JGR, Struyker-Boudier HAJ, Reneman RS. A technique to assess aortic distensibility and compliance in anesthetized and awake rats. Am J Physiol.. 1996;270:H780-H786.
Barra JG, Armentano RL, Levenson J, Cabrera-Fischer EI, Pichel RH, Simon A. Assessment of smooth muscle contribution to descending thoracic aortic elastic mechanics in conscious dogs. Circ Res.. 1993;73:1040-1050.
Armentano RL, Barra JG, Levenson J, Simon A, Pichel RH. Arterial wall mechanics in conscious dogs: assessment of viscous, inertial, and elastic moduli to characterize aortic wall behavior. Circ Res.. 1995;76:468-478.
Hayoz D, Nussberger J, Waeber B, Brunner HR. The renin-angiotensin system and arterial wall behavior. J Cardiovasc Pharmacol. 1993;22(suppl 5):S48-S52.
Dobrin PB, Rovick AA. Influence of vascular smooth muscle on contractile mechanics and elasticity of arteries. Am J Physiol.. 1969;217:1644-1651.
Adamson SL, Whiteley KJ, Langille BL. Pressure wave propagation and input impedance in thoracic aorta of conscious newborn sheep. Am J Physiol.. 1995;268:H558-H568.
Saxena PR, Bolt GE. Haemodynamic profiles of vasodilators in experimental hypertension. Trends Pharmacol Sci. 1986;501-506.