Reappraisal of European guidelines on hypertension management: a European Society of Hypertension Task Force document

Abstract

Abbreviations ACE: angiotensin-converting enzyme; BP: blood pressure; DBP: diastolic blood pressure; eGFR: estimated glomerular filtration rate; ESC: European Society of Cardiology; ESH: European Society of Hypertension; ET: endothelin; IMT: carotid intima-media thickness; JNC: Joint National Committee; LVH: left ventricular hypertrophy; LVM: left ventricular mass; PDE-5: phosphodiesterase-5; PPAR-γ: peroxisome proliferators-activated receptor-γ; PWV: pulse wave velocity; SBP: systolic blood pressure; WHO: World Health Organization. Introduction In the 2 years since the publication of the 2007 guidelines for the management of arterial hypertension of the European Society of Hypertension (ESH) and the European Society of Cardiology (ESC) [1], research on hypertension has actively been pursued and the results of new important studies (including several large randomized trials of antihypertensive therapy) have been published. Some of these studies have reinforced the evidence on which the recommendations of the 2007 ESH/ESC guidelines were based. However, other studies have widened the information available in 2007, modifying some of the previous concepts, and suggesting that new evidence-based recommendations could be appropriate. The aim of this document of the ESH is to address a number of studies on hypertension published in the last 2 years in order to assess their contribution to our expanding knowledge of hypertension. Furthermore, some critical appraisal of the current recommendations of the ESH/ESC, as well as of other guidelines, might be a useful step toward the preparation of a third version of the European guidelines in the future. The most important conclusions are summarized in boxes. The points that will be discussed are reported in Box 1.Box. 1Assessment of subclinical organ damage for stratification of total cardiovascular risk The 2007 ESH/ESC guidelines recommend total cardiovascular risk be evaluated in each patient to decide about important aspects of treatment: the blood pressure (BP) threshold at which to commence drug administration, the target BP to be reached by treatment, the use of two-drug combinations as the initial treatment step, and the possible addition to the antihypertensive treatment regimen of lipid-lowering and antiplatelet agents [1]. Among the criteria to assess total cardiovascular risk, the European guidelines consider subclinical organ damage to be a very important component, because asymptomatic alterations of the cardiovascular system and the kidney are crucial intermediate stages in the disease continuum that links risk factors such as hypertension to cardiovascular events and death. On the basis of a number of criteria (prognostic importance, prevalence in the population, availability and cost of the assessment procedures, etc.), the 2007 European guidelines considered detection of organ damage as important for the diagnostic and prognostic evaluation of hypertensive patients. They further subdivided the different types of organ damage into (1) those that can be identified by relatively simple and cheap procedures [electrocardiogram, serum creatinine, estimated glomerular filtration rate (eGFR), and measurement of urinary protein excretion in order to detect microalbuminuria or proteinuria], which were thus regarded as suitable for routine search in the whole hypertensive population, and (2) those that require more complex procedures or instrumentations (echocardiogram, carotid ultrasonography, pulse wave velocity), which were for this reason only recommended for a more in-depth characterization of the hypertensive patient. Since then, other studies have added useful information on the importance of detecting subclinical organ damage in the hypertensive population, strengthening the recommendation to use the most easily available and the least costly procedures in the routine examination of individuals with hypertension. Heart A few recent papers have revived interest in the power of the electrocardiogram to predict the risk of cardiovascular events. In a prospective survey including 7495 American adults, a new indicator of left ventricular hypertrophy (LVH), the Novacode estimate of left ventricular mass index that is based on both voltage and strain pattern criteria, has been reported to be significantly related to 10-year cardiovascular mortality [2]. The relation remained significant after adjusting for age, SBP, smoking, cholesterol, and diabetes. Furthermore, in the LIFE trial, the investigators have reported that in hypertensive patients with electrocardiographic LVH, left bundle branch block identifies individuals at increased risk of cardiovascular mortality (hazard ratio 1.6), sudden cardiovascular death (hazard ratio 3.5), and hospitalization for heart failure (hazard ratio 1.7) [3]. Finally, a very recent prospective study [4] focused on the R-wave voltage in lead aVL as being rather closely associated with left ventricular mass (LVM), and additionally predictive of incident cardiovascular events even when hypertension is not accompanied by electrocardiographic LVH (9% higher risk for each 0.1 mV higher R-wave). Additional evidence is also available on the predictive power of cardiac abnormalities, as detected by echocardiography, an approach of continuing interest because of its ability to more directly and precisely quantify LVM and geometric LVH patterns. A retrospective study has recently updated information from more than 35 000 normotensive and hypertensive participants with normal left ventricular ejection fraction [5]. Despite normal left ventricular function, an abnormal left ventricular geometric pattern was found in 46% of the patients (35% left ventricular concentric remodeling and 11% LVH), and the associated risk of all-cause mortality was twice as large as that of patients with normal left ventricular geometry. Although in another study on an African–American population, the relationship between left ventricular geometric patterns and all-cause mortality was markedly attenuated after adjusting for baseline variables, and remained significant only in men [6], the increased risk associated with LVH has been confirmed by other observations. In a prospective study on a cohort of 1652 Greek hypertensive patients followed up for 6 years, echocardiographic LVH was significantly associated with either a composite of all-cause mortality and cardiovascular events (hazard ratio 1.53) and with stroke (hazard ratio 2.01), after adjustment for major cardiovascular risk factors [7]. Furthermore, a retrospective analysis of 1447 Japanese hypertensive patients who participated in the CASE-J trial showed that cardiovascular events occurred about 2.6 times more frequently in patients with a LVM index 125 g/m2 or more compared with those with a LVM index below this value [8]. Finally, in the PAMELA population, echocardiographic LVH was associated with a four-fold to five-fold significant increase in cardiovascular morbidity and mortality when data were adjusted for a large number of potential confounders, including office, home, and ambulatory BP values. A 10% increase in LVM increased the risk more markedly when baseline LVM was already abnormal, but an increasing risk was evident also when calculated from LVM values within the normal range [9]. Blood vessels The relationship of carotid intima–media thickness (IMT) and plaques with subsequent cardiovascular events, already discussed in the 2007 guidelines, has been further strengthened by data from ELSA [10], which have shown that baseline carotid IMT predicts cardiovascular events independent of BP (clinic and ambulatory) and this occurs both for the IMT value at the carotid bifurcations and for the IMT value at the level of the common carotid artery. This suggests that both atherosclerosis (reflected by the IMT value at the bifurcations) and vascular hypertrophy (reflected by the common carotid IMT) exert an adverse prognostic effect in addition to that of high BP. An adverse prognostic significance of carotid plaques (hazard ratio 2.3) has also been reported in a sample of residents of the Copenhagen County free of overt cardiovascular disease, which was prospectively followed for about 13 years [11]. Evidence has also accrued on the adverse prognostic value of arterial stiffening. In the Copenhagen County population, an increased pulse wave velocity (PWV >12 m/s) was associated with a 50% increase in the risk of a cardiovascular event [11]. Furthermore, an independent predictive value of PWV for cardiovascular events has been shown in Japanese men followed for 8.2 years [12]. Finally, indirect indices of aortic stiffness and wave reflection, such as central BP and augmentation index, have been confirmed as independent predictors of cardiovascular events in two recent studies [13,14]. In particular, in one of these studies of 1272 normotensive and untreated hypertensive patients, only central SBP consistently and independently predicted cardiovascular mortality after adjustment for various cardiovascular risk factors, including LVM and carotid IMT [14]. However, it should be emphasized that in most available studies, the additive predictive value of central BP beyond brachial pressure appears limited, which leaves the question whether central BP measurements should be regularly considered in the clinical profiling of hypertensive patients in need of further investigation. Kidney Several new data [15] reinforce the already solid evidence on the prognostic value of eGFR that was available at the time of the 2007 guidelines [1]. In the population of Gubbio (Italy), an eGFR in the lowest decile was associated with a significantly higher incidence of cardiovascular events (hazard ratio 2.14) [16], and in the above-mentioned Greek study [7], an eGFR between 15 and 59 ml/min per 1.73 m2 was associated with a 66% increase in the composite endpoint of all cause mortality and cardiovascular events after adjustment for baseline cardiovascular risk and independent of LVH [7]. Likewise, in a post hoc analysis of data from the VALUE trial [17], eGFR according to the MDRD formula was significantly predictive of all outcomes except stroke (with hazard ratios between 1.23 and 1.70 according to the different outcomes) and was more sensitive than calculation of the creatinine clearance value according to the Cockroft–Gault formula, which was only predictive of all-cause mortality. The baseline eGFR by the MDRD formula turned out to be importantly predictive of both renal and cardiovascular events also in the large number (n = 11 140) of type 2 diabetic patients included in the ADVANCE trial, even when data were adjusted for many potential confounders, including the concomitant urinary protein excretion value. For every 50% reduction of baseline eGFR the risk of cardiovascular events significantly increased 2.2-fold, the concomitant increase in the risk of cardiovascular death and renal events being 3.6-fold and 63.6-fold, respectively [18]. New evidence is also available to support the already large amount of data in favor of the prognostic value of the moderate increase in urinary protein excretion, defined as microalbuminuria [19,20]. In two population studies, the Gubbio study [16] and the Copenhagen County study [11], microalbuminuria was confirmed as an important predictor of cardiovascular outcome, the adjusted hazard ratio being, respectively, 2.15-fold and 3.10-fold greater in patients with microalbuminuria compared with those without. In the Gubbio study, the association of microalbuminuria with low eGFR had a multiplicative effect (hazard ratio 5.93). In the ADVANCE trial [18], a change from one clinical stage of albuminuria to the next was associated with a 1.6-fold, 2.0-fold, and 3.3-fold increase in the multivariate-adjusted risk of cardiovascular events, cardiovascular death, and renal events, respectively, this being the case also when the change from normoalbuminuria to microalbuminuria was involved. The effects of higher baseline urinary protein excretion and reduced eGFR were independent of each other and the association of microalbuminuria and an eGFR value less than 60 ml/min per 1.73 m2 brought about an additional increase in risk: 3.2-fold for cardiovascular events, 5.9-fold for cardiovascular mortality, and 22.2-fold for renal events. Additional measures of organ damage The 2007 European guidelines mention a number of additional measures of organ damage for which evidence of prognostic relevance was available, but no use in the clinical practice could be foreseen because of drawbacks of practical relevance, such as the high cost and low availability of the devices involved, the complexity and time consumption inherent in the procedures, and in several instances the lack of standardization of the values obtained between laboratories and across countries. Based on the evidence available in the last 2 years, no addition to the measures of organ damage included in the 2007 guidelines can be supported, although the growing availability of more sophisticated techniques and the reduced cost of their use brought about by technological progress, makes future additions likely. In this context, the use of nuclear magnetic resonance deserves special mention. Although not prospective in nature, a very recent study systematically employing nuclear magnetic resonance imaging in a group of 142 hypertensive patients without overt cardiovascular disease has provided the interesting information that silent cerebrovascular lesions are even more prevalent (44%) than cardiac (21%) and renal (26%) subclinical damage, and do frequently occur in the absence of other signs of organ damage [21]. Increasing evidence also relates these lesions to cognitive dysfunction [22,23], a problem of primary importance because of the senescence of the population [24]. With magnetic resonance imaging becoming more and more frequently employed in diagnostic procedures, silent cerebrovascular disease is likely to become more frequently investigated in prognostic and therapeutic studies in hypertension. The prognostic value of structural alterations in small subcutaneous arteries has recently been confirmed by two independent studies [25,26]. However, the invasive nature of this measurement prevents larger scale application of this method. A new noninvasive method for assessing the media–lumen ratio of small retinal arteries seems promising for large-scale evaluation [27], although its predictive value remains to be investigated. Evidence remains inconclusive on a marker of a vascular alteration that has been actively investigated in the past decade, namely endothelial dysfunction. In a population sample of individuals without overt cardiovascular disease (67% with hypertension and 22% with diabetes mellitus) from the Northern Manhattan study, measures of flow-mediated vasodilatation predicted the incidence of cardiovascular events, but this effect was not independent of traditional cardiovascular risk factors [28]. Likewise, in the large cohort of elderly patients of the Cardiovascular Health Study, flow-mediated vasodilatation added very little to the prognostic accuracy of traditional risk factors [29]. On the contrary, Muiesan et al.[30] have recently reported that in a small cohort (n = 172) of uncomplicated hypertensive persons followed for about 8 years, flow-mediated vasodilatation of the brachial artery below the median value was significantly associated with a 2.7-fold increase in incident cardiovascular events even after adjusting for all major cardiovascular risk factors. However, the same group of investigators also have reported that endothelial dysfunction in the subcutaneous vessels of hypertensive patients was not predictive of cardiovascular events [31], possibly because endothelial dysfunction in different vascular beds may have a different prognostic significance. Clearly, the prognostic value of endothelial dysfunction in hypertension remains to be further elucidated. It should be emphasized that the addition of new measures of organ damage to the assessment of total cardiovascular risk requires not only the demonstration of their prognostic importance, but it has to improve the power to predict the incidence of cardiovascular events. This is by no means easy to be documented, and indeed data are available that in some instances new risk factors of individual prognostic significance do not improve, when added to the others, the accuracy by which cardiovascular risk can be quantified, thus only making the diagnostic procedures more complex, time consuming, and costly. This is exemplified by the recent results of the Framingham study, which showed that inclusion of inflammatory markers did not lead to any substantial improvement in the accuracy (sensitivity and specificity) by which total cardiovascular risk was assessed [32]. Subclinical organ damage as a marker of high cardiovascular risk Although subclinical organ damage undoubtedly increases the level of cardiovascular risk, the question arises whether it always brings the patient into the high-risk category, that is, an absolute risk of at least 20 cardiovascular events in 10 years per 100 patients. The 2007 European guidelines classify hypertensive patients with subclinical organ damage among those with a high total cardiovascular risk. This is further supported by more recent evidence on the contribution of subclinical cardiac, vascular, and renal damage to the total cardiovascular risk. As regards to subclinical cardiac damage, analysis of the data provided by some of the major prospective studies indicates that in hypertensive patients, echocardiographic LVH, particularly if of the concentric variety, is associated with an incidence of cardiovascular events equal to or above 20% in 10 years [5,7,33]. An incidence greater than 20% in 10 years has also been reported for men, but not for women, with echocardiographic LVH in the Framingham population study [34]. Finally, in the hypertensive patients of the CASE-J trial, echocardiographic LVH was associated with a 10-year incidence of cardiovascular events of 24% compared with the 10% incidence seen in patients without LVH [8]. Similar evidence exists for vascular damage. In the elderly patients of the Cardiovascular Health Study [35], the 10-year incidence of major cardiovascular events was higher than 20% when the common carotid IMT was 1.06 mm or more (fourth and fifth quintiles) and below 10% in those with an IMT in the first quintile (<0.87 mm). In the hypertensive patients of the ELSA study [10], the incidence of all (major and minor) cardiovascular events was greater than 20% in 10 years when IMT (common carotid plus bifurcation) was in the third and fourth quartiles (≥1.16 mm) or when at least one plaque had been detected. In contrast, patients with IMT in the first or the smallest IMT quartile (<0.98 mm) had incident cardiovascular events below 10% in 10 years. In hypertensive patients, the 10-year incidence of major cardiovascular events was higher than 20% when carotid-femoral PWV (aortic stiffness) was 16.3 m/s or more (fifth quintile) and below 10% in those with an aortic stiffness in the first and second quintiles [36]. Furthermore, even asymptomatic peripheral vascular disease as detected by a positive ankle-brachial index has prospectively been found to be associated in men with an incidence of cardiovascular events approaching 20% in 10 years [37,38]. Finally, old and recent evidence leaves little doubt that in hypertensive individuals, renal subclinical organ damage is associated with a 10-year risk of cardiovascular events of 20% or more. It has already been reported some years ago that reduced renal function, defined by a serum creatinine more than 1.5 mg/dl is associated with a 10-year incidence of cardiovascular events 20% or more [39,40]. In the recent prospective cohort of Greek hypertensive patients [7], a low eGFR was associated with incident cardiovascular events of about 20% in 10 years, an even higher incidence being observed when low eGFR occurred together with LVH. Furthermore, in the hypertensive patients prospectively studied by Jensen et al.[41], the incidence of ischemic heart disease was 20% in 10 years in the presence of microalbuminuria and of only 5% in its absence. Also, in the Gubbio population study, the incidence of cardiovascular events was greater than 20% in 10 years, but only in those individuals in whom microalbuminuria in the highest decile was associated with eGFR in the lowest decile [16]. Over 78% of these patients had hypertension. The 2007 European guidelines classify patients with subclinical organ damage as being at high risk also when BP is in the high normal range, but admittedly evidence that this is invariably the case is less clear. In the general population of the Framingham study, no information was made available on the prognostic value of echographic LVH, separately in the normotensive and hypertensive population [34]. Furthermore, in the same population, the association of renal dysfunction with cardiovascular events was lost after adjustment for cardiovascular risk factors, including BP [42]. In the PREVEND population study [43], microalbuminuria (20–200 mg/l) was associated with only a 4.7% cardiovascular mortality in 10 years, that is, a moderate absolute risk according to the SCORE classification [44], and in the nonhypertensive, nondiabetic individuals of the Framingham study, a microalbuminuria above the median value was associated with a rate of incident cardiovascular events of only 8.8% in 10 years compared with a 2.9% rate in individuals with microalbuminuria below the median value [45]. Prognostic value of treatment-induced modifications of subclinical organ damage The 2007 European guidelines have emphasized that treatment-induced changes of organ damage affect the incidence of cardiovascular events, thereby recommending that organ damage be measured also during treatment. Reference was made to the data obtained in the LIFE study [46], in which hypertensive patients in whom treatment was accompanied by regression of echocardiographic LVH or a delayed increase in LVM had less incident cardiovascular events, including sudden death, than those in whom no regression from or earlier progression to LVH occurred. It was also mentioned that both in LIFE [47] and in other studies [48], a similar relationship was found between treatment-induced changes in proteinuria and renal or cardiovascular events. This means that, compared with patients in whom treatment had little or no antiproteinuric effect, reduction in proteinuria was associated with a reduced incidence of cardiovascular events and less progression to end-stage renal disease. Since 2007, data on the relationship between treatment-induced changes in cardiac damage and cardiovascular protection have been enriched by further analyses of the LIFE study, which have shown that also treatment-induced changes in left atrial dimension [49], left ventricular geometry [50], and in electrocardiographic signs of LVH correlate with incident cardiovascular event rate [51]. Furthermore, there have been reports that in hypertension, inappropriate changes in LVM during treatment adversely affect cardiovascular prognosis [52]. Finally, the predictive power of treatment-induced IMT changes in the carotid arteries has for the first time been investigated in a recent analysis of ELSA trial data. This analysis failed to show a predictive role of treatment-dependent IMT changes, but the smallness of these changes compared with the large individual differences in baseline IMT makes it difficult to draw definitive conclusions [10]. The correlation of treatment-induced changes in proteinuria with cardiovascular event incidence has been challenged by some findings of the ONTARGET trial. In this trial on a large number of high or very high cardiovascular risk patients, the group treated with a combination of an angiotensin-converting enzyme (ACE) inhibitor and an angiotensin receptor antagonist showed, throughout the study duration, less increase in proteinuria than the group on monotherapy with one or the other drug, but this relative antiproteinuric effect was not accompanied by a reduction in cardiovascular events and was even associated with an increase in renal events [53]. However, these results do not necessarily undermine the important concept that treatment-induced changes in proteinuria can be a marker of the more or less pronounced beneficial effects of treatment because alternative explanations for the ONTARGET results are possible. For example, in ONTARGET, most patients had a normal renal function and few (4%) exhibited overt proteinuria, which resulted in a very limited number of the endpoint that matters for renal protection, that is, chronic renal failure. Furthermore, in the very high cardiovascular risk population studied, the powerful blockade of the renin–angiotensin system provided by the ACE inhibitor and angiotensin receptor antagonist combination might have exhibited an adverse effect of its own that superseded and masked the beneficial influence associated with a reduction in proteinuria. In favor of this beneficial influence are some recent analyses of the ADVANCE study in patients with type 2 diabetes. In these patients, on-treatment values of proteinuria showed a close independent association with both renal and cardiovascular events, the contribution of proteinuria being unrelated to the concomitant values of eGFR [18]. Conclusion Evidence on the important prognostic role of subclinical organ damage continues to grow. In both hypertensive patients and the general population, the presence of electrocardiographic and echocardiographic LVH, a carotid plaque or thickening, an increased arterial stiffness, a reduced eGFR (assessed by the MDRD formula), or microalbuminuria or proteinuria substantially increases the total cardiovascular risk, usually moving hypertensive patients into the high absolute risk range. The changes in electrocardiographically or echocardiographically detected LVH induced by treatment reflect the effects on cardiovascular events, thereby offering valuable information on whether patients are more or less effectively protected by the adopted treatment strategy. Despite some recent inconsistent results [53], solid evidence suggests that this is the case also for treatment-induced changes in urinary protein excretion, although the problem remains open for treatment-induced vascular changes. Thus, assessing the presence of subclinical organ damage is of crucial importance in the hypertensive population. This assessment can make use of simple and cheap procedures that can provide routine information before and at various times during treatment. It can also rely on more sophisticated approaches that can further characterize patients' cardiac and vascular status. In all instances, multiple organ damage assessment is useful because of the evidence that in the presence of two signs of organ damage (even when inherent to the same organ), cardiovascular risk may be more markedly increased, with an almost inevitable upgrading to the high cardiovascular risk category [7,16]. It is not clear from published data whether subclinical organ damage can bring total cardiovascular risk to the high range also in patients with high normal BP. However, organ damage when it is particularly pronounced, or affects multiple organs, or is accompanied by metabolic risk factors, is associated with a two-fold or three-fold increase in relative risk also in normotensive individuals [11,54–56], and the 2007 guidelines recommend considering relative risk as a guide for the need of treatment in young and middle-aged patients. In this context, it is also important to emphasize that the occurrence of undetected organ damage in patients that doctors decide to treat probably explains the apparently paradoxical findings of several observational studies that the incidence of cardiovascular events is higher in treated than in untreated hypertensive patients even after adjustment for usual cardiovascular risk factors and past clinical history [57–62]. This is consistent with the concept that antihypertensive treatment even if beneficial cannot usually take a high total risk back to a low-risk category [63]. These findings presumably reflect the fact that in medical practice, BP-lowering treatment is often deferred until organ damage occurs, when complete reversibility is not achievable [63,64]. More extensive use of organ damage assessment may thus help to reach a more timely decision about the initiation of treatment and thus favor its greater success. Some of the issues discussed in Assessment of subclinical organ damage for stratification of total cardiovascular risk section are summarized in Box 2.Box. 2Treatment approach Major guidelines [1,65–70] on the management of hypertension recommend the initiation of antihypertensive drugs in all patients with a SBP 140 mmHg or more and/or a DBP 90 mmHg or more, and to adjust the treatment strategy in order for the patients to be below these values. They further recommend drug treatment to be initiated within a lower BP range, that is, a SBP between 130 and 139 mmHg and a DBP between 85 and 89 mmHg in patients with diabetes or a history of cardiovascular or renal disease, aiming at achieving SBP/DBP values <130/80 mmHg. The 2007 ESH/ESC guidelines [1] have accompanied these recommendations with information on the evidence they are based upon, and a critical reappraisal of this issue has recently been undertaken by members of the present Task Force [71], in the light of further information provided by recent trials. The purpose of the present ESH document is to clarify the size and the type of evidence on which these recommendations are based, and thus help the planning and conduc