|
| HOME | SUBSCRIPTIONS | CURRENT ISSUE | PAST ISSUES | CARDIOSOURCE | SEARCH | HELP | FEEDBACK |
|
|
|
|||||||||
|
|||||||||||
|
J Am Coll Cardiol, 2006; 48:1698-1711, doi:10.1016/j.jacc.2006.06.074
(Published online 26 September 2006). © 2006 by the American College of Cardiology Foundation |
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Cardiovascular Sciences, University of Leicester School of Medicine, Leicester, United Kingdom.
Manuscript received June 20, 2006; accepted June 28, 2006.
* Reprint requests and correspondence: Dr. Bryan Williams, Clinical Sciences Building, Leicester Royal Infirmary, PO Box 65, Leicester, LE2 7LX, United Kingdom. (Email: www.bw17{at}le.ac.uk).
 |
The global burden of hypertension |
|---|
|
|
Hypertension awareness, detection, and control |
|---|
55%) but less progress in the percentage of those with hypertension whose BP is controlled to recommended goals on treatment, which continues to hover around 25% (Fig. 2). It should be noted, however, that the criteria for control have been tightened from <160/95 mm Hg in the earlier NHANES surveys to <140/90 mm Hg in the most recent. The Healthy People 2010 report (3) targets a control rate for the U.S. population of 50% in 5 years, which would only be reached if at least 80% of people with hypertension were aware of their condition, 90% were treated, and 70% of those treated had their condition controlled (4)—there is still much work to be done if these laudable aspirations are to be realized.
|
|
The hypertensive phenotype |
|---|
The risk factors for developing hypertension in adults were recently evaluated from the Strong Heart Study, a population-based longitudinal cohort study of cardiovascular risk factors in Native Americans from 3 communities in Arizona (6). Unsurprisingly, initial BP was a strong predictor of incident hypertension. However, incident hypertension could also be predicted by a person’s initial metabolic profile and unfavorable metabolic variations over time. Even in those with optimal initial BP levels (<120/80 mm Hg), increasing abdominal obesity and abnormal lipid profile were major predictors of the development of hypertension. The participants were examined at baseline, at 4 years, and at 8 years. In those with a normal baseline glucose tolerance, as well as an optimal BP, higher baseline waist circumference and systolic BP, as well as evidence of diabetes at 4 years, increases in systolic BP and waist circumference, and a decrease in high-density lipoprotein (HDL) cholesterol from baseline to the second examination predicted hypertension at 8 years. It was also of interest that the 4-year decrease in HDL cholesterol was as strong an independent predictor of 8-year incident hypertension as the baseline level of systolic BP. Similarly, a recent report examining factors predicting development of hypertension from the Physicians’ Health Study also reported a powerful association between disturbed lipid metabolism and the risk of developing hypertension (7). After following up this cohort for a mean of 14 years, using Cox proportional hazards modeling adjusted for lifestyle and clinical risk factors, men in the highest quintile of total cholesterol, non-HDL cholesterol, and the ratio of total to HDL cholesterol had increased risks of developing hypertension of 23%, 39%, and 54%, respectively, compared with participants in the lowest quintile. Moreover, men in the highest quintile of HDL cholesterol had a 32% decreased risk of developing hypertension compared with those in the lowest quintile.
Together, these observations support previous reports characterizing the prehypertensive phenotype (8–10) and highlight that this phenotype is a complex mix of dysregulated BP and metabolism that is potentially amenable to early lifestyle and/or drug interventions designed to prevent the development of hypertension.
The TROPHY study. With regard to early intervention aimed at preventing the development of hypertension, the TROPHY (Trial of Preventing Hypertension) study has provided some important insights (11). A total of 809 people with "prehypertension" (systolic BP 130 to 139 mm Hg; diastolic BP 89 mm Hg or lower) were all offered lifestyle advice for the duration of the study and randomized to treatment with either an angiotensin receptor blocker (ARB) (candesartan 16 mg once daily) or placebo for the first 2 years of follow-up, then placebo for both groups for a further 2 years of follow-up (Fig. 3). The objective was to determine whether drug treatment of prehypertension prevented or postponed the development of stage 1 hypertension. During the first 2 years, hypertension developed in 154 participants in the placebo group and in 53 of those in the candesartan group, a relative risk reduction of 66.3% (p < 0.001). The active treatment was then stopped and converted to placebo for a further 2 years. Thus, after a total follow-up of 4 years, hypertension had developed in 240 participants in the placebo group and 208 of those originally assigned to the candesartan group, a relative risk reduction of 15.6% (p < 0.007).
|
It will be interesting to discover from subsequent analyses of the TROPHY data whether the pharmacologic treatment of the prehypertensive state also impacted on the metabolic disturbances mentioned earlier, or more intriguingly, looking to future studies, whether treatments targeting the metabolic disturbances per se that characterize the hypertensive phenotype might actually prevent the development of hypertension. The importance of the TROPHY study was not in the result, which was largely predictable, but that it served to highlight the potential of therapeutic strategies aimed at targeting the early hypertensive phenotype to prevent the development of more severe hypertension.
|
Lifestyle interventions for people with hypertension |
|---|
However, in light of the high transition rates from prehypertension to stage I hypertension and beyond, in my view, it is questionable whether lifestyle intervention alone will continue to be regarded as sufficient for people with uncomplicated stage 1 hypertension.
|
Ambulatory BP measurement |
|---|
After adjustment for gender, age, risk indices, and clinic BP, higher mean values of ambulatory BP were independent predictors for cardiovascular mortality. After correction for other risk factors, ambulatory BP was superior to clinic BP for prediction of cardiovascular mortality. Moreover, nighttime BP was the most potent predictor of outcome (Fig. 4). This finding is consistent with another recent report from Japan that examined the relationship between ABPM and patterns of stroke. In 1,430 people followed up for an average of 10 years, cerebral infarction risk was significantly higher in "non-dippers" (i.e., people with a <10% decline in night-time BP). On the other hand, an increased risk of cerebral hemorrhage was observed in people with a large morning pressor surge (i.e., >25 mm Hg in the 2 h after waking) (15).
|
So, when to use ABPM? The routine use for all patients is perhaps a step too far at this time, but ABPM is useful in the assessment of the untreated patient when there is uncertainty about the decision to treat and for assessing the quality of BP control in patients at high risk. Moreover, there needs to be greater use of ABPM and other strategies to record BP in clinical outcome trials to better characterize the effect of drugs on BP (16).
Ambulatory arterial stiffness index (AASI). In an endeavor to further refine risk stratification using ABPM, the data have been used to derive an AASI. The physiological principles underpinning the derivation of the AASI are that arterial stiffness varies nonlinearly with distending pressure throughout the day and that the relationship between systolic and diastolic BP will depend on the functional characteristics of large conduit arteries—in people with stiffer arteries, there will be a greater increase in systolic versus diastolic BP. It has therefore been proposed that the AASI is a surrogate measure of arterial stiffness. Using all of the readings from a 24-h ABPM recording, the AASI is derived by plotting diastolic versus systolic BP and the regression slope for this relationship is calculated. The AASI has been defined as 1 – this regression slope (17). In support of the concept that the AASI is a surrogate for arterial stiffness, the AASI correlates with the pulse wave velocity and pulse wave augmentation index and correlates more with the augmentation index than 24-h pulse pressure (17).
The use of AASI as a prognostic indicator of total and cause-specific cardiovascular mortality was evaluated utilizing the Dublin Outcomes Study cohort described earlier. Baseline AASI was derived from ABPM recordings from 11,291 patients (mean age 54.6 years). Over a median follow-up of 5.3 years, before and after adjustment for other cardiovascular risk factors, the AASI and pulse pressure significantly predicted total cardiovascular mortality.
The AASI was a stronger predictor than pulse pressure for stroke. The AASI seemed to be a stronger predictor of stroke mortality than pulse pressure, especially in normotensive patients (18). Clearly, further physiological and clinical studies will be required to validate this novel risk index and to confirm the predictive value of the AASI over and above conventional risk stratification in other cohorts. If confirmed and validated, the AASI could be readily computed from standard ABPM recordings and may in future provide additional data for risk stratification. However, it is unfortunate that the term AASI was applied to this index. There is much misuse of the term arterial stiffness, as expertly reviewed in the accompanying editorial (19). The concept that the AASI is a measure of arterial stiffness is simplistic, and truer markers of arterial stiffness, such as pulse wave velocity, have a stronger predictive value for mortality than reported for the AASI (20). Perhaps a better and suitably vague term would have been an ambulatory vascular index.
|
Trials of BP-lowering therapy |
|---|
The second question addressed by ASCOT-BPLA was whether conventional BP-lowering therapy (atenolol ± bendroflumethiazide-K, as required) was sufficient for optimal cardiovascular disease prevention when compared with a more contemporary regimen of newer drugs (amlodipine ± perindopril, as required). The patients were predominantly male (77%), with a mean age of 63 years. Importantly, people with a prior history of myocardial infarction or treated angina were excluded; thus, this was a primary prevention trial. The ASCOT-BPLA was stopped earlier than anticipated (follow-up median 5.5 years) on the recommendation of the data safety monitoring board because of a clear benefit of the amlodipine-based therapy on most cardiovascular end points and total mortality (reduced by 11%) and in particular on cardiovascular mortality (reduced by 24%), even though the primary end point (fatal coronary heart disease or nonfatal myocardial infarction) was nonsignificantly different at the time the study was terminated (21).
Some have argued that because most of the power of a clinical trial is invested in its primary end point, if this is not significantly different, then subsequent analysis of secondary end points should be viewed with caution. This seems unreasonable when a study is terminated because of safety concerns due to substantial and significant differences in key end points in favor of one of the treatment strategies. In ASCOT, when compared with atenolol-based treatment, amlodipine-based treatment significantly reduced stroke by 23%, all-cause death by 11%, cardiovascular death by 24%, and all cardiovascular events and procedures by 16%. There was also a significant 30% reduction in new-onset diabetes in favor of amlodipine-based therapy (21).
The BP control was better throughout ASCOT-BPLA with amlodipine-based therapy, especially in the first few months. Moreover, the differences persisted, albeit to a lesser extent throughout the trial, despite greater use of add-on therapy with the atenolol-based therapy. These differences in brachial BP between the treatment arms are likely to have been an important driver of the differences in clinical outcomes in ASCOT, but further analyses adjusting for BP and other variables throughout the study suggested that other factors were also important (24). The CAFE (Conduit Artery Function Evaluation) study (25,26), a substudy of ASCOT, suggested that the differential effects of BP-lowering therapy on central aortic pressures may also have been an important determinant of the beneficial effect of amlodipine-based therapy in ASCOT. This is further discussed later.
The ASCOT study confirmed the importance of initial drug selection in defining the subsequent quality of BP control. The ASCOT study also showed that those patients randomized to the most effective BP-lowering regimen (amlodipine-based therapy) and a statin had their risk of stroke and myocardial infarction reduced by half when compared with those randomized to conventional BP-lowering therapy without a statin. This is a very important message that reinforces a shift in thinking regarding the optimal management of risk in people with hypertension. The cardiovascular disease risk of the typical ASCOT patient was <20% over 10 years, thereby suggesting that the typical hypertensive (i.e., European or American men over 50 years old) would have sufficient cardiovascular disease risk to benefit from statin therapy, even if their cholesterol levels appear normal. This is a very significant development that is already impacting treatment guidelines (27).
Finally, ASCOT also reinforced data from previous studies suggesting that beta-blocker–based therapy, especially in older patients (i.e., >55 years) is less effective than alternative classes of therapy at reducing cardiovascular events, notably stroke, and is more likely, especially in combination with thiazide diuretics, to increase the risk of developing diabetes. The impact of these findings on the fate of beta-blockers for the routine treatment of hypertension is discussed later.
The CAFE study. The CAFE study was a large substudy of ASCOT that examined the hypothesis that the different BP-lowering drug regimens used in ASCOT could have had different effects on central aortic pressures despite similar effects on brachial BP (25,26). The CAFE study recruited 2,199 patients, and radial artery applanation tonometry and pulse wave analysis were used to derive central aortic pressures and hemodynamic indices. Despite a similar brachial systolic BP between treatment groups, there were significantly greater reductions in central aortic systolic and pulse pressures with the amlodipine-based treatment regimen when compared with the atenolol-based regimen (Fig. 5). Thus, BP-lowering drugs have the potential to differentially impact on central aortic pressures, despite similar effects on brachial BP. This important finding suggests that brachial BP does not always faithfully reflect the impact of different BP-lowering drugs on central aortic pressures and hemodynamics. A similar conclusion has been reached recently by other investigators using different drug treatments in much smaller, shorter-term studies (28,29).
|
Using Cox proportional hazards modeling, the CAFE investigators went on to show that on-treatment central pulse pressure was significantly associated with clinical outcomes (26). These findings suggest that differential drug effects on central aortic pressures could be an important determinant of drug-related differences in clinical outcomes in clinical trials such as ASCOT and others that have often referred to drug benefits as "beyond blood pressure." The CAFE findings also suggest that drug development programs that are predicated on regulatory requirements for effective reductions in brachial pressure may be underestimating the potential benefit of drugs that favorably impact pulse wave reflection.
In another study, baseline pulse wave analysis was performed in small a subset of participants in the ANBP-2 (Australian National Blood Pressure Study 2) study to derive central aortic pressures (30). In this analysis, there were only 53 events (in contrast to CAFE’s 305 events) during a median of 4.1 years of follow-up in 484 women. The ANBP-2 study did not show that higher baseline central aortic pressures were a significant predictor of clinical outcomes. However, the study was clearly underpowered to test this hypothesis and thus prone to a type 2 statistical error, and differs from the CAFE study in that it did not examine on-treatment pressures.
Further analyses from ALLHAT. The ALLHAT (Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial) investigators continue to generate important data from various subanalyses. The ALLHAT compared treatment with amlodipine or lisinopril with the reference drug chlorthalidone. The primary end point was fatal coronary heart disease or nonfatal myocardial infarction. Three reports have emerged in the past year looking at clinical outcomes from prespecified subgroup analyses: 1) black and nonblack patients (31), 2) people with diabetes (32), and 3) renal outcomes (33).
ALLHAT outcomes in black and nonblack patients. This was an important prespecified analysis because of the paucity of clinical outcome data in variety of ethnic groups with hypertension. In ALLHAT, 11,792 patients were black (35%). No significant difference was found between treatment groups for the primary coronary heart disease (CHD) outcome in either racial subgroup (31). For lisinopril versus chlorthalidone, systolic BP was lowered more effectively with the diuretic, and this coincided with more effective stroke prevention in black patients (relative risk [RR] 1.40 [95% confidence interval 1.17 to 1.68]) with the diuretic. The difference in stroke rate is likely to be caused by the differences in BP control. There was less difference in BP control between treatments in nonblack patients, and no significant difference in stroke rate (Table 1).
|
The main conclusions to be drawn from this important report are: 1) achieved BP is a powerful determinant of outcome for black patients and nonblack patients, especially for stroke, and 2) a thiazide-type diuretic or calcium channel blocker (CCB) will deliver the most effective initial BP control in black patients.
ALLHAT outcomes in people with diabetes. The ALLHAT investigators have also reported a prespecified subgroup analysis of people with type 2 diabetes (n = 13,101) or normoglycemia (17,012) at baseline (32). This is by far the largest study of the treatment of hypertension in people with diabetes.
When comparing treatment arms, there was no significant difference in relative risk for the primary outcome of fatal CHD or non-fatal myocardial infarction in people with or without diabetes at baseline. These findings are important because: 1) ALLHAT does not support the superiority of angiotensin-converting enzyme (ACE) inhibition for the prevention of CHD in people with diabetes; 2) ALLHAT does not support the ill-founded concerns about the use of CCBs for people with diabetes generated in previous retrospective case-control studies, some meta-analyses, and much smaller prospective studies (34–37); 3) ALLHAT provides confirmation that thiazide-type diuretics are very effective and often an essential component of hypertension treatment in people with type 2 diabetes.
These findings support the conclusions of a recent meta-analysis of patients with diabetes and hypertension randomized into major clinical outcome trials (38). This meta-analysis also analyzed the outcome in those with "more versus less intensive" (
BP 5/3 mm Hg) BP lowering in people with or without diabetes at baseline. There was clear benefit, reducing the risk of major cardiovascular events and total mortality, with the more intensive BP-lowering strategy in people with diabetes. The benefit of this strategy was less clear for those without diabetes at baseline. These findings provide retrospective support for the more intensive BP-lowering strategy advocated for people with diabetes in treatment guidelines.
ALLHAT renal outcomes in high-risk hypertensive patients.
This post-hoc analysis of the ALLHAT database examined whether treatment with either a CCB or an ACE inhibitor influenced the likelihood of developing end-stage renal disease (ESRD) or a reduction in glomerular filtration rate (GFR) of at least 50%, when compared with treatment with a thiazide-type diuretic, over a mean follow-up of 4.9 years (33). Patients were stratified according to their baseline GFR (estimated by the simplified Modification of Diet in Renal Disease equation), into 3 categories: normal or increased GFR (
90 ml/min per 1.73 m2, n = 8,126), mild reduction in GFR (60 to 89 ml/min per 1.73 m2, n = 18,109), or moderate-severe reduction in GFR (<60 ml/min per 1.73 m2, n = 5,662). In 448 participants ESRD developed, and in 1,049 the composite end point developed, defined as a 50% or greater decline in GFR or ESRD. There was no difference in treatment effects for either end point for patients taking amlodipine or lisinopril compared with those taking chlorthalidone across the 3 GFR subgroups, either for the total group or for participants with diabetes at baseline.
These are important and intriguing findings. They challenge the view that renin angiotensin aldosterone system (RAAS) blockade is especially renoprotective in people with different categories of renal disease. The findings cannot be dismissed lightly because the number of patients in each category of mild and moderate renal insufficiency in ALLHAT exceeded the total number of similar patients in many renal outcome studies. Moreover, the number of patients who reached ESRD in ALLHAT was greater than in any previous renal trial. Furthermore, in previous studies examining renoprotection in which the benefits of RAAS blockade have been established, there has invariably been a lower on-treatment BP favoring the RAAS blockade group. In ALLHAT, the converse was true. This supports the view that BP lowering is a key driver of renoprotection. Nevertheless, there are caveats. The ALLHAT did not measure urinary albumin excretion, and it is conceivable that the renoprotective benefits of RAAS blockade are more pronounced in people with significant albuminuria. Also, the ALLHAT investigators concede that it was likely that atherosclerotic or ischemic kidney disease (i.e., tubulointerstitial disease) accounted for much of the renal insufficiency in their patients and that this type of renal injury may be less amenable to the protective effect of RAAS blockade (33).
The MOSES study. The MOSES (Morbidity and Mortality After Stroke, Eprosartan Compared With Nitrendipine for Secondary Prevention) study evaluated BP lowering for secondary prevention in patients who had suffered a computed tomography– or magnetic resonance imaging–documented cerebral event in the previous 24 months (39). The study compared BP lowering with either an ARB-based (eprosartan), or CCB-based (nitrendipine) therapy. Blood pressure control was similar with both treatment strategies. The primary end point of the MOSES study was the composite of all-cause mortality and the number of cardiovascular and cerebrovascular events, including all recurrent events. This was significantly reduced by eprosartan-based therapy (incidence density ratio 0.79, 95% confidence interval 0.66 to 0.96, p < 0.014). It is unusual not to focus the primary end point of an outcome trial on first event, rather than also counting additional events in the same patient. Nevertheless, when only first events were included in the analysis, a similar trend of benefit for the ARB was observed. Further studies of optimal timing and treatment strategies for BP lowering in the secondary prevention of stroke are needed.
|
Atrial fibrillation in hypertension |
|---|
The 4 main factors predicting the risk of developing AF in the LIFE study in order of effect size were: 1) age (each year of age associated with a 9% higher rate of new-onset AF), 2) male gender (56% increase in risk compared with women), 3) systolic BP (6% increase per 10 mm Hg), and 4) the magnitude of ECG-LVH by Cornell product (4% increase per 100 mV · ms) (Table 2). In addition, randomization to losartan rather than to atenolol was associated with a 33% lower rate of new-onset AF independent of other risk factors (p < 0.001). Furthermore, patients receiving losartan-based therapy tended to stay in sinus rhythm longer.
|
|
Previous studies have suggested that blockade of the RAAS with either ACE inhibition or an ARB can reduce the risk of developing AF (46–48), but these studies usually compared active drug with placebo, thus it was unclear whether the reduction in incident AF was related to BP lowering or to a specific drug effect. These new data from the studies cited earlier imply a drug treatment effect because there was no difference in BP reduction between the treatment arms. Moreover, the findings from the LIFE study will come as a surprise to many who have considered beta-blockade a preferred therapy for preventing AF as well as the preferred treatment for rate control in established AF (49).
It is tempting to speculate on the mechanisms underpinning these observations. At the cellular level, studies in animals have suggested the potential for blockade of the action of angiotensin II to prolong the action potential and prevent the acute electrical remodeling response to short-term rapid atrial pacing (50,51). Others have shown no effect on electrical remodeling to more chronic atrial pacing and no effect of angiotensin II infusions on electrical physiology in humans (52). Other possibilities relate to the beneficial effects of blockade of RAAS, or more specifically the angiotensin II receptor, on atrial fibrosis (53), LVH regression, remodeling, and left atrial size (54,55). Whatever the mechanism, these findings should prompt formal studies of the role of RAAS blockade (ACE inhibition, ARB or renin inhibition) in the prevention of new and recurrent AF in people with and without hypertension and in the role of routine RAAS blockade in the prevention of cardiovascular events in people with established refractory AF.
|
Beta-blockers and hypertension |
|---|
|
20% (59). These data, taken together with the observations that beta-blockers are less effective than alternatives at regressing cardiovascular structural changes and in preventing the development of AF than anticipated (discussed earlier), are likely to condemn beta-blockers when considering the preferred initial therapy options for the routine treatment of hypertension in future guidelines. What are the potential mechanisms for less effective stroke prevention with beta-blockade? One possibility is that they are less effective at BP lowering than the alternatives. This is supported by the early months of ASCOT and may be part of the explanation in older patients (21). Nevertheless, it is not sufficient to explain ASCOT or the results of other studies such as LIFE. Another possibility highlighted by the CAFE study (see earlier) is that beta-blocker–based therapy may be less efficient at lowering central aortic BP, despite similar effects on brachial BP (26). Another consideration is that the adverse effects of beta-blocker therapy on the metabolic milieu, notably increased triglycerides, reduced HDL cholesterol, and impaired glycemic control, may all conspire to reduce their effectiveness at optimally reducing cardiovascular disease risk.
There are a number of important caveats to the anticipated demise of beta-blockers for the routine initial treatment of hypertension: 1) Most of the data for their role in the treatment of hypertension have been acquired with atenolol, the most popular beta-blocker used for the treatment of hypertension worldwide. It is unclear whether the outcomes would be similar with other beta-blockers, especially those with different pharmacologic properties. Nevertheless, in the absence of outcome data with other beta-blockers in people with hypertension, it would be difficult to support their routine use as a preferred initial therapy for hypertension. 2) There are specific indications for continued use of beta-blockers in people with hypertension, irrespective of their BP effect, notably, in patients with symptomatic angina, chronic stable heart failure and after myocardial infarction. 3) Most of the trials of BP-lowering therapy have been conducted in older patients, and it is unclear whether the same concerns would apply to younger people with hypertension, in whom their BP-lowering efficacy may on occasion exceed that of other classes of therapy. 4) There are also circumstances in which alternative treatments may be poorly tolerated or contraindicated, or ineffective at controlling BP, circumstances in which it would be perfectly reasonable to select a beta-blocker to lower BP, the benefits of which exceed no treatment at all. This is a debate of great importance and one that will continue to generate much interest.
|
Summary and reflections |
|---|
However, let us not lose sight of how little we know about how current BP therapies may or may not work. The CAFE study and others have highlighted the importance of studies of human vascular function that will be further enhanced by the emergence of ever-improving imaging modalities. No doubt, this will cause us to lament the lost tribes of cardiovascular physiologists and will emphasize how studies of human physiology and its response to modern drug therapy need to be reinvigorated to better define how existing drugs work and how future novel therapies might be developed.
Around the corner is the launch of a new class of BP-lowering drugs—the renin inhibitors, and the late-stage evaluation in treatment-resistant hypertension of drugs that inhibit endothelin, new combinations of BP-lowering drugs, further analyses from recent clinical outcome trials, continuing debate about beta-blockers and the clinical significance of drug-induced diabetes, and ultimately the assimilation of these new data into hypertension treatment guidelines. Who said hypertension was sorted!
|
Footnotes |
|---|
|
References |
|---|
Top
The global burden of...
Hypertension awareness,...
