Left ventricular hypertrophy: beyond the image and defining the human cardiac phenotype in hypertension

Introduction Left ventricular hypertrophy (LVH) has proven to be a very useful risk marker in hypertension. It indicates a higher risk of cardiovascular events and particularly sudden death [1,2]. It defines the effects of hypertension on an important target organ, the heart, and is one step along the way to heart failure, a major complication of hypertension. The relationship between hypertension and subsequent heart failure may be underestimated by the clinical community, given data from the Framingham population indicating that over 90% of subjects who developed heart failure during follow-up had been identified as hypertensive at some time earlier in the study, and calculations of attributable risk for heart failure indicate that hypertension is quantitatively the most important risk factor for heart failure [3]. The cardiac phenotype in essential hypertension LVH is universal in animal models of hypertension, but, according to the many definitions applied in clinical assessment, it is far less common in human essential hypertension. There are probably many reasons for this. Perhaps the most important is ‘signal to noise’ ratio. There is great variation in cardiac dimensions in the free-living human population; hypertension is generally a mild haemodynamic stimulus in the majority of patients in contemporary practice, and many of the measurement techniques are technically demanding [4]. These considerations also underlie the fact that identification of LVH is a major determinant of the need for drug therapy in recent international guidelines on the management of hypertension. However, because the effects of therapy on LVH are small and occur slowly, they are difficult to follow in an individual patient. Guidelines have placed little emphasis on regular follow-up to monitor changes in LVH over time. Although there has been intense interest into whether certain drug classes reverse LVH better than others, by far the most important determinant of progression or regression of LVH is the long-term blood pressure [5,6]. In clinical practice, we generally assume on this basis that if the patient's blood pressure is well controlled then other benefits will follow, including removal of the propensity for LVH to progress to decompensation and manifest clinically as heart failure. In this sense, LVH can be considered as an intermediate phenotype between the healthy heart and the large dilated ventricle seen in heart failure. However, cardiac hypertrophy (if defined as a large heart due to cardiomyocyte hypertrophy) is only part of the cardiac phenotype in hypertension. Ventricular cardiomyocytes make up approximately one-third of the total cell number but account for 70–80% of the heart's mass [7]. Hypertrophy may not even be the most important part of the hypertensive cardiac phenotype in a pathophysiological sense, but a clinical marker based on estimation of LV mass will inevitably be dominated by the effects of disease on cardiomyocyte size. Alongside cardiomyocyte hypertrophy, changes in systolic and diastolic function occur in hypertension because of effects on the contractility of the heart and its filling properties. The latter in part reflect interstitial changes, particularly myocardial fibrosis, as well as the increasing bulk of the myocardium. The balance in perfusion between endothelium and epithelium is altered as are the myocardial cell: capillary ratio, electrophysiological properties, and the relative density of sympathetic innervation. Coronary artery disease may supervene with its own effects on cardiac structure and function [5,8]. At the molecular level, there is reversion in cardiomyocytes to genes that were previously expressed during fetal life including atrial and B-type natriuretic peptides (ANP, BNP), fetal forms of contractile proteins such as skeletal α-actin and β-myosin heavy chain, as well as down-regulation of other genes normally seen in the adult rather than the fetal heart [9,10]. There is evidence that functional changes such as alteration in cardiac contractility, and reduced filling properties of the ventricle occur early in the development of the cardiac hypertension phenotype and before changes in LV mass can be detected with standard clinical measures such as echocardiography [4]. The evidence for sequential changes in cardiac structure and function in human hypertension is circumstantial, as detailed sequential studies have not been performed. Nevertheless, the animal literature and human data available (generally assuming that borderline hypertension is an early manifestation of established hypertension) suggest a relatively stereotyped and evolving sequence over years and decades. In the adult heart, growth is generally well matched to functional load. Thus, if there is volume loading either from a physiological stimulus, such as in the athlete or in patients with valvular regurgitation, the heart dilates to compensate for increased diastolic wall stress. It should be noted, incidentally, that cardiomyocyte signaling cascades are quite different in physiological and pathological hypertrophy so that hearts that appear similar at the gross anatomical level may be very different functionally and consequentially over the longer term [11]. In hypertension, the load is increased during systole and the heart responds with a hypertrophic response in order to counterbalance wall stress. Mild concentric hypertrophy is characterized by increased relative wall thickness (ventricular wall thickness: LV cavity diameter). LV contractility at this stage is increased, contributing to higher than normal cardiac output often seen in early hypertension [12]. Reduced LV filling properties are also seen in early hypertension. With prolonged hypertension, especially when coronary artery disease supervenes, the thickened but well functioning left ventricle gives way with decompensation to progressive dilatation, reduced systolic function, myocardial fibrosis and diastolic dysfunction. Thus, it seems there is a ‘tipping point’ at which the full hypertension cardiac phenotype is present with concentric hypertrophy, myocardial fibrosis, and reduction in systolic and diastolic function. Beyond this, the picture is one of a failing heart with LV dilatation, systolic and diastolic failure, leading to a very different haemodynamic profile and ominous short-medium term outcome. In clinical practice, we would very much like to be able to accurately identify the ‘tipping point’ and target therapies to the individual patient phenotype at any given time. What tools are available to us? The electrocardiogram pattern of LVH and strain identifies a group of patients at high risk of cardiovascular events and of sudden cardiac death [1]. Present in only a small proportion of hypertensive patients, it is unable to distinguish increased cardiac dimensions due to wall thickening or to cavity dilatation. The R wave, used to assess LVH semi-quantitatively from the electrocardiogram, is susceptible to changes in volume status, preload, or afterload of the heart within seconds and cannot be considered a measure of anatomical size of the myocardium. Nevertheless, the presence of electrocardiographic signs of LVH in a patient with hypertension is a very important prognostic indicator [1]. Likewise, the echocardiogram, which can detect milder forms of LVH, predicts outcome in patients with hypertension [2]. Comprehensive echocardiographic study allows assessment of the bulk of the myocardium and measurements of wall thickness and dimensions of the LV and other cardiac chambers, systolic function can be determined from fractional shortening of the LV, and Doppler recording provides additional information on systolic and diastolic function. In short, most elements of the cardiac hypertensive phenotype can be identified with a careful echocardiogram. These elements are usually identified individually and qualitatively. Research use demands both quantification of important variables, such as LV mass, and those related to systolic and diastolic function. Furthermore, it is desirable to combine variables to characterize the hypertensive heart from the healthy heart, and from that subject to other pathologies. One way to do this from echocardiographic data is to use a multivariate approach and discriminant analysis to separate normotensive, and hypertensive subjects from a mixed population. Without knowledge of blood pressure variables, this was accurate in categorizing over three-quarters of those into hypertensive or normotensive groups when a combination of structural (e.g. wall thickness, relative wall thickness, LV mass) and functional variables (e.g. fractional shortening, deceleration time or E/A ratio) was used. However, accurate discrimination required both structural and functional variables to be included in the model and could not be obtained from LV structure alone [4]. Echocardiographic recording is becoming increasingly sophisticated as techniques including tissue Doppler and three-dimensional reconstruction have become commonplace in the clinic. Benchmark technologies for assessment of LVH now involve imaging modalities with better spatial resolution than echocardiography. Magnetic resonance imaging has been used in a number of studies and rapid computed tomography scanners in others [13,14]. These techniques provide estimates of LV mass that are systematically lower than those obtained with the standard echocardiographic measurements and will demand another look at normal values for LV structural variables in human populations. On the positive side, better resolution means the number of patients required for adequate power to detect a 5% change in LV mass with treatment is considerably lower; scores of patients are required rather than hundreds of patients when using echocardiography. ‘Inappropriate left ventricular mass’ Another approach explored in a paper by Chinali et al. [15] is to make an assessment of the appropriateness of LV mass estimated by echocardiogram in relation to the haemodynamic load and to define a group in whom the LV mass is excessive for the apparent demands put on the LV by cardiac afterload. ‘Appropriate LV mass’ was calculated by correcting LV mass for gender, height and stroke volume. Values exceeding 28% of the predicted value, corresponding to the 95th percentile of distribution in the reference population, were defined ‘inappropriate LV mass’. This work is the culmination of a huge amount of careful preliminary work which established values for cut-offs for ‘inappropriate LV mass,’ and showed that these cut-offs defined a high risk group by clinical outcomes [13]. In short, inappropriate mass cut-offs helped capture the patients with a more fully developed cardiac hypertension phenotype of concentric hypertrophy, reduced LV systolic and diastolic function. This ‘cluster’ was much better defined by appropriateness or inappropriateness than by other markers of the hypertensive heart (e.g. presence of LVH defined by high LV mass). A number of questions arise from this analysis. Has the application of a stricter criterion for LVH narrowed the group down to a smaller group with more severe manifestation of hypertensive heart disease? Here, the answer seems to be clearly ‘yes’, but would the same groups have been identified with a higher cut-off for pathological elevation of LV mass? Here, the answer is probably ‘no’ because a significant minority of those with ‘inappropriate LV mass’ did not have LV mass outside the normal limits as defined for the present study (Fig. 1). Thus, an echocardiographic study with measurements subject to a series of mathematical manipulations has defined a high-risk group with a pattern of abnormalities in cardiac structure and function that typify advanced hypertensive cardiac changes at or just before the previously discussed ‘tipping point’.Fig. 1Methodology in contemporary quantitative echocardiography It is salutary at this point to consider these mathematical transformations and aspects of the source data, and to question whether this is as far as we can go with the use of echocardiograms to quantify aspects of the hypertensive heart. The hypertension community is very aware of balancing the practical advantages (and theoretical disadvantages) of defining cut-offs that break near-normally distributed blood pressure variables into categorical variables. Perhaps the same considerations apply to LV mass. The standard cube method developed by Devereux and Reichek [16] and subsequently modified slightly by others begins with recording of two-dimensional and M-mode echocardiograms. M-mode traces used for quantification need to be precisely on axis (i.e. symmetrical cuts through the left ventricle at a predefined level). Routine clinical echocardiograms rarely meet the standard required. It is then conventional to apply the Penn convention. This excludes endocardial echoes when they are visible (often they are not), even though the endocardium is clearly part of the LV and contributes to its mass. The reason for doing so is pragmatic as application of the standard equation for calculating LV mass in early validation studies [16] overestimated the value according to previous validation studies comparing LV mass measured directly at autopsy with that estimated by echocardiography. The standard equation of Devereux and Reichek [16] makes geometrical assumptions that the left ventricle is symmetrical and transverse linear dimensions can be cubed to obtain myocardial volume. The volume estimate is multiplied by the average specific density of human myocardial tissue (1.04) to obtain mass and a correction factor (13.6) is subtracted, again based on the results of previous validation experiments. Of course larger people have larger body parts, including hearts, so the vexing issue at this point, having obtained an estimate of LV mass in grams is to adjust for body size. There is also a systematic gender difference. The latter can be accounted for by using different cut-off values for males and females. The former has been adjusted for in many ways. Simply dividing by body mass index (LV mass index) works relatively poorly and there are no a priori reasons for assuming that the relationship between LV mass and body size is a linear function of body mass index [4]. Other approaches have included adjustment for lean body mass but the index that empirically has worked best [17] has been to adjust LV mass to height to the power2.7. There are also systematic differences between ethnic groups. However, in a recent study using cardiac resonance imaging to estimate LV mass, we that LV also well for these apparent differences or not LV hypertrophy is present was determined by Chinali et al. [15] from values determined from the 95th percentile of a different normotensive population for for go to the with blood we do not normally cut-offs on such as the 95th percentile of a normal population. to use for a for clinical such as a level of blood pressure at which risk of complication seems and the of a major determinant of the number of subjects with disease that outside a normal determined by the 95th percentile in a healthy population is the of the in The latter is by variation and by the and of the Thus, if a has a large in the reference population, it is more difficult to the limits that and to be as than if there is a small a large this the literature on the with which hypertension is by LVH with estimates from the high variables used to obtain to approximately three-quarters when variables or multivariate analysis was used [4]. The important question is whether or not the of LVH were those from the haemodynamic load or if it was into account the relationship between LV mass and load seen in the population. The an LV mass’ by correcting LV mass for gender and for in an equation stroke work in volume systolic blood pressure This equation was previously to define a high risk group in a up for clinical outcomes A of LV mass of was in the 95th percentile in a reference population and patients with a higher had outcomes than the in an earlier it was not to have high LV mass to within the of ‘inappropriate LV mass’. Of of a normal in this way also defines a group with lower than LV mass (i.e. values the This group patients in the study by Chinali et al. function was by dividing stroke volume by diastolic again calculated by the LV using the LV be adjusted for body size but, in this to by Another indicator of contractile was shortening as this is afterload this was for systolic stress. shortening defined the group with reduced systolic function, again using values obtained from another reference population. filling can be from the of the early measured by Doppler the deceleration time of the However, E/A are by and by heart leading to another mathematical to normal from and E/A are and A adjusted and and is heart Values of the adjusted were as indicating LV [13]. Another of the cardiac hypertension phenotype is concentric This is generally from relative wall thickness, the of the LV wall and the measurements by the LV axis and obtained from M-mode Chinali et al. [15] adjust this to again using data from a reference population and the and are relative wall thickness with and Values were considered based on values from a reference population. of the patients with hypertension had ‘inappropriate LV mass’ but not within the of the group who had LVH defined on LV mass (Fig. 1). The former were much more to have systolic and diastolic and concentric (i.e. they had more advanced hypertensive cardiac disease and were no at higher were also more given the at which were for related to body the question as to whether this is a factor in the risk with echocardiographic Chinali et al. as well as who have to the development of echocardiographic to at this point is a and to the with which these measurements have been This is particularly given the mathematical transformations and for a of patient These often so standard of the source data are It is also a of given many assumptions were along the of these such as the of the left the Penn adjustment of LV mass for make very little or are at best very have into common because they work in and have helped better define the of cardiac changes in hypertension. have been in because echocardiography is a well relatively for the heart. However, it the resolution of other techniques such as resonance imaging Left ventricular hypertrophy in the community a major factor in the quantitative assessment of LVH by echocardiography is variation in the healthy population leading to standard of variables. LV mass can individual of a healthy population. measurement and between study this is a relatively in the population. It can be due to a of abnormalities and have been in genes for contractile the same often considerably in phenotype and the same may be with normal or normal hearts, and with severe cardiac hypertrophy in This indicates a propensity for relatively and for to have a on cardiac structure and probably for a significant amount of variation in approximately of the in LV mass is also have an important on cardiac structure and function. Of the most important is there is evidence that particularly high may to cardiac hypertrophy The heart and the heart with pathological hypertrophy the common of an in LV mass However, there are and In physiological hypertrophy due to not into dilated and heart failure In healthy hearts or those with physiological hypertrophy, a of and In pressure hypertrophy, which in hypertension, this is There is of cardiac proteins and relative to the number and size of cardiac has an in early hypertrophy and related to but interstitial fibrosis the factor with and hypertension. changes with a reversion to fetal forms is seen in pathological but not physiological hypertrophy On the other the pattern seen in the athlete to the concentric of hypertrophy seen in hypertension, the pattern of for the in clinical outcomes seen between the There are models of hypertrophy that progress to decompensation and failure of the genes expressed in pathological hypertrophy, such as and are and probably changes signaling do to be in pathological and physiological cardiac is the hypertrophy It is well established in the clinical and literature that including and are of pathological cardiac hypertrophy by pressure These in of one of the and of of the and signaling (Fig. of the or in cardiac hypertrophy that is with cardiac cardiac and death On the other signaling in the heart LVH in response to pressure Furthermore, pathological but not physiological hypertrophy Thus, the seems in the development of pathological to the present of the of with pathological hypertrophy in human hypertension is that the that define pathological hypertrophy (e.g. hypertrophy, fibrosis, can be at the molecular in a model of pressure hypertrophy was with of fibrosis but not a reduction in heart mass This for the of the hypertensive cardiac phenotype especially in clinical studies of the of LVH or the effects of specific are hypertrophy The growth factor is particularly important in the of physiological hypertrophy in response to with cardiac hypertrophy in response to pressure but not to Furthermore, of the in a of pathological hypertrophy is with the that this is important in physiological hypertrophy, cardiac of but not or was higher in than in subjects healthy group of subjects with hypertension will with of of the group by Chinali et al. often of such as or and the of the subjects will with to common of genes in the cardiac contractile In a of the between these on cardiac structure and function, with hypertension and concentric hypertrophy on a long-term of there was no change in LV mass but the relative wall thickness and the LV cavity increased in a change from the concentric to of hypertrophy One could that the pathological hypertrophy by hypertension is into a more physiological hypertrophy by this also changes in the relative of of the and hypertrophy due to hypertension is part of the of cardiac hypertensive disease but is not its It is to measurements of LV mass to the full particularly as the of the cardiac response to hypertension is not a progressive and in LV mass. Other of the hypertensive heart, such as fibrosis and functional also change cardiac dimensions in the population are normally distributed and are by and has a way in to define the changes in cardiac structure and function in hypertension through and progress has been given the in spatial resolution (and an increasingly hypertensive the limits may be when mathematical of measurements of to define cut-offs in mass variables expressed in dimensions in and adjusted and variables to in the of variables into categorical variables. The in in technologies that have better spatial resolution and are able to make measurements that do not tomography and resonance imaging this and there are relative advantages and of knowledge of the of LVH and the to heart failure may also to that reflect the differences in and at in the clinical of by the and Research of and the of