Which Components of the Arterial Pressure Wave Best Predict Risk?

A continuing debate in the field of hypertension is the relative importance of the different components of the arterial pressure wave in determining cardiovascular risk. There are four candidates: systolic, diastolic, pulse, and mean pressure. An additional issue is whether the traditional brachial artery pressure should be used or the central aortic pressure. For many years, the diastolic pressure reigned supreme, and most of the early hypertension treatment trials used a high diastolic pressure as an entry criterion. This was reinforced by the publication of an analysis by MacMahon and colleagues,1 based on pooled data from 420,000 subjects, which showed a log-linear relationship between diastolic pressure and the risk of stroke and myocardial infarction. Although the importance of systolic pressure was never in doubt, it gained precedence over diastolic pressure with the publication of a series of epidemiologic studies showing not only that a high systolic pressure was the best predictor of risk in the elderly, but that a low diastolic pressure was associated with increased risk.2 An analysis by Franklin and associates3 examined the Framingham Heart Study data that provided an elegant solution to this apparent paradox. In subjects younger than age 50, the best predictor of risk was a high diastolic pressure, but in those older than 60, systolic pressure was the best predictor, and the relationship between diastolic pressure and risk was now negative, so that a low diastolic pressure was related to higher risk. It is well known that systolic pressure increases steadily with age, but after the age of 50, diastolic pressure starts to fall. These findings have led to the idea that pulse pressure may be the best predictor of risk in the elderly. In the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC 7),4 however, which is still the official guideline for the evaluation and management of hypertension, there is no mention of pulse pressure as a predictor. Two of the figures in the JNC 7 publication show the relationships of both systolic and diastolic pressure with ischemic heart disease events and strokes based on a meta-analysis of more than one million adults in 61 studies.5 For both measures of pressure, there are log-linear relationships, which, even in the oldest age group (80–89 years), show positive slopes that become shallower with increasing age. So how can this finding be compatible with the apparently contradictory Framingham analysis showing that in older people, a low diastolic pressure can be harmful? First, the linear relationship between diastolic pressure and risk shown in the JNC 7 figures only applies to a diastolic pressure of 70 mm Hg and higher (there is a suggestion that diastolic pressures below this level are not associated with lower risk). Second, and more important, the explanation of the apparent paradox is that the relationship depends not just on diastolic pressure, but on its interaction with systolic pressure. This was clearly shown by an analysis of the Multiple Risk Factor Intervention Trial (MRFIT)6 data, which found that the highest risk group included patients who had a systolic pressure above 160 mm Hg and a diastolic pressure below 80 mm Hg. Thus, if the systolic pressure is normal, a low diastolic pressure in an elderly patient is quite harmless.7 This raises the issue of whether pulse pressure should be the target in older patients. Not all studies that have compared the predictive powers of systolic vs. pulse pressure in older patients have found that pulse pressure is superior,8, 9 and some of its apparent superiority may be a statistical artifact. Thus, many of the studies expressed the risk in terms of a 10-mm Hg change of pressure; since systolic pressure is always numerically greater than pulse pressure, this will tend to attribute a greater predictive value to the latter. In an analysis of the National Health and Nutrition Examination Survey (NHANES) data,10 it was shown that the relative importance of pulse pressure diminishes if the standard deviations for the two measures are used. In practical terms, high-risk patients can be identified by a very high systolic pressure (e.g., >160 mm Hg) and a very low diastolic pressure (e.g., <70 mm Hg), without any specific reference to pulse pressure. There are no evidence-based guidelines relating to the treatment of a high pulse pressure. The trials examining the effects of treating systolic hypertension recruited patients largely on the basis of a high systolic pressure alone11 or in combination with a diastolic pressure <90 or 95 mm Hg.12, 13 In the Systolic Hypertension in China (Syst-China) trial,11 the pretreatment blood pressure was 170/86 mm Hg, and the final pressure in the active treatment group was 150/81 mm Hg. Thus, the treatment had a greater effect on systolic than diastolic pressure, with a reduction of pulse pressure, and none of these trials lowered systolic pressure below 145 mm Hg.14 The best we have relating to treatment based on pulse pressure is an analysis of the Systolic Hypertension in the Elderly Program (SHEP)15 results, which found that the baseline pulse pressure did not influence the benefits of treatment. However, it was also found that if the diastolic pressure on treatment was <70 mm Hg, there was a significantly increased risk of cardiovascular events. These findings are thus consistent with the idea that a very low diastolic pressure in combination with a very high systolic pressure is associated with high risk. There are at least two explanations for this, which we may call direct and indirect. Taking the direct explanation first, it is possible that a very low diastolic pressure might impair coronary artery perfusion, which takes place during diastole. If this were the case, we should expect the adverse outcomes resulting from a low diastolic pressure to be ischemic cardiac events, but not strokes. The indirect explanation is that the low diastolic pressure is not harmful per se, but is a marker for generalized cardiovascular disease or, more particularly, increased arterial stiffness. It is generally accepted that the age-related increase of systolic and decrease of diastolic pressure that is seen in the brachial artery is largely the result of accelerated wave reflection in stiff arteries.16 There are some studies that have examined the relationship of pulse and diastolic pressure with specific outcomes. Millar and Lever17 analyzed the data from the MRC trial and found that at low levels of diastolic pressure (<90 mm Hg), there was an increased risk of myocardial infarction (the J-curve phenomenon), whereas for strokes, the lowest risk was associated with the lowest diastolic pressure. Madhavan et al.7 also found that pulse pressure predicted myocardial infarction. If the further reduction of an initially low diastolic pressure in patients with systolic hypertension is harmful, it would be desirable to have antihypertensive drugs that selectively reduce pulse pressure. Unfortunately, most of the currently available drugs have relatively little effect, which is reflected by the fact that the single most important factor associated with inability to reduce blood pressure to target levels is old age.18 There are differences in the extent to which antihypertensive drugs lower pulse pressure relative to their effects on the other components: in elderly patients, the greatest reductions of pulse pressure were seen with clonidine and hydrochlorothiazide, whereas atenolol had very little effect.19 Mean pressure is another potential candidate as a risk marker. It is customarily defined as diastolic plus one third of the pulse pressure, but it is measured directly by oscillometric devices, although the numbers are rarely displayed. In an analysis of three intervention studies in elderly patients, Blacher and associates2 found that pulse pressure, but not mean pressure, predicted cardiovascular outcomes. However, in younger patients, mean pressure is more important. In two studies of younger patients, pulse pressure was the strongest predictor of coronary events,7, 17 while mean pressure was the best predictor of stroke.20 This finding was confirmed by Verdecchia and colleagues21 in a study using ambulatory blood pressure monitoring. One explanation for this finding may be that the patient with the high pulse pressure has increased arterial stiffness and is effectively being hit with a “double whammy”: the high systolic pressure in the central aorta increases the afterload on the heart and the low diastolic pressure impairs coronary perfusion. The reason the mean pressure is a better predictor of stroke may be that it is closer to diastolic than to systolic pressure, and many studies have shown that the relationship between diastolic pressure and stroke is steeper than for coronary events.1 No studies have shown that a very low diastolic pressure predicts stroke. The arterial pressure wave in the brachial artery looks very different from the waves recorded at more proximal sites, where the damage to target organs such as the heart and brain occurs, and it is now possible to estimate the central aortic pressure indirectly from measurements made noninvasively from peripheral sites such as the radial artery. In older patients with stiff arteries, the central aortic systolic pressure is similar to the brachial pressure, whereas in younger subjects with compliant arteries, it is substantially lower.16 As seen above, different drugs have different effects on pulse pressure, and recent studies have shown that similar differences occur in their effects on central pressure.22 There are, however, many different methods for measuring arterial stiffness,23 and there is disagreement as to which is the most reliable. Some have been shown to predict cardiovascular events,24, 25 and it is likely that such measurements will become part of routine clinical practice in the near future. At the present time, most official guidelines emphasize the importance of systolic pressure, particularly in older patients, and there is no consensus about pulse pressure. However, the European Society of Hypertension26 has suggested that in patients older than 50, the goal of treatment should be a mean pressure below 100 mm Hg and a pulse pressure below 50 mm Hg. While this can be justified scientifically, in practice it would probably just add confusion if widely adopted. The superiority of pulse pressure over systolic pressure implies that a low diastolic pressure is harmful and that the ideal treatment for patients with very high pulse pressures should be to simultaneously reduce systolic and increase diastolic pressure. Unfortunately, we have no good way of doing this with our currently available drugs, so the concept remains unproven. Until that time, it would seem appropriate to continue to focus on systolic pressure. The measurement of central aortic pressure has great promise, but is not yet at the stage where it can be applied in clinical practice.