Heart and Metabolism

Critical evaluation of diagnostic techniques to assess left ventricular hypertrophy

Scott Takeda, John Chambers
Cardiothoracic Centre, St Thomas’s Hospital, London, UK
Correspondence: Dr John Chambers, Cardiothoracic Centre, St Thomas’s Hospital, London SE1 7EH, UK. Tel: +44-20 7928 9292, fax+44-20 7969 5680, e-mail: johnchambers@dial.pipex.com

Introduction
Left ventricular hypertrophy is an independent risk factor for stroke, myocardial infarction, congestive heart failure and sudden death[1] in patients with hypertension or aortic stenosis,[2] but also in apparently normal people with no evidence of pressure overload.[3] This article assesses the techniques available for estimating left ventricular mass in clinical practice.

Echocardiography
M-mode echocardiography is still the most widely used method in clinical practice, despite several limitations. Dimensions measured at the base of the heart can only be used to generalize to the whole heart in the presence of symmetrical geometry. Non-uniformity of any wall as a result of myocardial infarction or localized hypertrophy invalidates the technique. The M-mode cursor must be placed perpendicular to the posterior wall and septum. This was not possible in up to 34% of patients[4,5] until the relatively recent introduction of movable or ‘anatomical’ M-mode, although measurements from two-dimensional echocardiography could be substituted. In at least 10% of patients, adequate images used to be unobtainable because of a poor ‘window’. Image quality has now improved as a result of new technologies, particularly second harmonic imaging, although preliminary experience suggests that wall thickness may be slightly larger using second harmonic than fundamental imaging.
The widely used Devereux formula[6] treats the left ventricle as a cube so that the myocardial volume is the difference between the outer margin of the left ventricle and the cavity. The cavity diameter is the left ventricular diastolic dimension (LVID), and the outer diameter is the LVID plus the septal width (IVS) plus the posterior wall width (PWT). Mass is derived from volume after multiplying by the density of cardiac muscle, 1.04 g/cm3. By correlating calculated left ventricular mass with angiographic and postmortem findings, Devereux et al.[6] derived a correction factor of 13.6 g, giving the formula:
Left ventricular mass = 1.04 x [(LVID + IVS + PWT)³ - (LVID)³] - 13.6 g

The ‘Devereux formula’ requires the measurements to be made using the Penn convention in which the endocardial echoes are excluded from the septal and posterior walls. However, dimensions are usually measured using the convention of the American Society of Echocardiography (ASE) from ‘leading edge to leading edge’. Use of the ASE convention in the cube-function formula consistently overestimates left ventricular mass by about 15–25%6,[7] and in a recent comparison with magnetic resonance, by up to 37%.[8] Furthermore, the formula was derived in only 34 patients and comparison with necropsy findings was by regression equations. It is well known that strong correlations are possible even in the absence of adequate agreement.[9]
There is potentially major intra- and interobserver variability in M-mode measurements which can be minimized by averaging at least 3–5 cycles and by meticulous care in obtaining optimal image quality and orientation. In the PRESERVE trial[10] using M-mode or linear two-dimensional measurements according to ASE recommendations, the between-study standard deviation in left ventricular mass was only 6 g/m2. However, standard deviations may sometimes be as high as 30 g,[11,12] which is of the same magnitude as expected changes in left ventricular mass. This means that M-mode measurements should rarely be used for quantifying a change in left ventricular mass over serial studies except in a large population of several hundred patients.[13]

Two- and three-dimensional echocardiography
There are two frequently used methods for quantifying myocardial mass by two-dimensional echocardiography, the area/length and the truncated ellipsoid methods. Both correlate well with anatomic left ventricular mass,[14,15] provided that care is taken to avoid foreshortening the left ventricular cavity. Correlations are better (r = 0.93, standard error of estimate error [SEE] 31 g) than by M-mode (r = 0.86, SEE = 59 g), especially in hearts with abnormal geometry,[16] and reproducibility is also better.[17] Despite this, these two-dimensional methods are seldom used, mainly because they are more time-consuming and partly because so much experience with M-mode already exists.
Three-dimensional techniques are expected to describe the heart better than M-mode or two-dimensional echocardiography. Early comparisons with magnetic resonance imaging have shown good correlations in volume estimates
(r = 0.91, SEE = 28 ml)[18] and it is possible that a transpulmonary contrast agent may further improve volume estimates using transoesophageal three-dimensional echocardiography.[19] Early commercial transthoracic three-dimensional systems are now available, but are not yet widely used.

What constitutes left ventricular hypertrophy?
Left ventricular mass is related to body habitus. It is therefore conventional to index mass to body surface area taken from Dubois nomograms, although in some circumstances, especially in the morbidly obese, it may be more appropriate to normalize to height.[20] Left ventricular mass is a continuous variable in terms of cardiovascular risk, but it is still useful to have a bipartite division into normal or abnormal for clinical characterization. The specific criteria proposed by Hammond et al.[21] are widely used: >134 g/m2 for men and >110 g/m2 for women. However, left ventricular mass is dependent on several factors other than body habitus, including age, gender, level of physical activity, race and possibly angiotensin-converting enzyme genotype.[22,23] For research studies, particularly with small population sizes, a carefully matched control population may be needed to determine thresholds of abnormality for left ventricular mass.

Left ventricular geometry
According to the law of Laplace,[24] left ventricular wall stress (S) is directly proportional to intracavitary pressure (P) and chamber radius (R) and is inversely proportional to wall thickness (Th):
S = P . R/Th

Left ventricular hypertrophy tends to reduce wall stress by increasing wall thickness and reducing cavity size. However, the relationship between left ventricular mass, cavity size and pressure is not fixed. Three different geometric responses can be defined by the relative wall thickness ratio, which is the thickness of the posterior wall divided by the left ventricular radius in diastole: (1) concentric hypertrophy defined by increased left ventricular mass and a high relative wall thickness ratio (>0.45); (2) eccentric hypertrophy defined by increased left ventricular mass and normal relative wall thickness ratio (<0.45); and (3) concentric remodelling defined by normal left ventricular mass with a high wall thickness ratio. Concentric remodelling is associated with an increased risk of morbid events (2.39 per 100 patient-years) compared with normal geometry (1.12 per 100 patient-years)[25] (Table 1).

Table 1: Left ventricular geometry.

Other techniques for estimating left ventricular mass
Ultrafast computed tomography provides higher image resolution than echocardiography and allows truly tomographic assessment of the left ventricle.[26] It does not rely on geometric assumptions and has been anatomically verified in dogs.[27] It has also been shown to have excellent reproducibility in serial studies,[28] but is limited by the use of ionizing radiation and the need for intravenous contrast.
Nuclear magnetic resonance imaging has superior image quality compared with echocardiography, and truly tomographic images with little reliance on geometric assumptions. Magnetic resonance imaging has been validated against anatomic mass in cadaver hearts (r = 0.99, SEE = 6.8 g)[29,30] and in dogs’ hearts compared with subsequent postmortem assessment (r = 0.98, SEE = 6.1 g).[31] It is accurate in assessing asymmetric hearts, for example in dogs with experimentally induced myocardial infarction[32] or in humans with hypertrophic cardiomyopathy.[33] The standard error in estimating left ventricular mass is about one-third that of echocardiography, and so magnetic resonance imaging is more accurate for serial estimates of left ventricular mass.[34,35]
Magnetic resonance imaging also has advantages over ultrafast computerized tomography in that it does not require ionizing radiation and is therefore better suited for serial studies.35 A number of patients cannot be scanned because of claustrophobia or the presence of implanted metallic devices such as pacemakers or some mechanical heart valves. A previous drawback was the unacceptably long scan times of up to 45–60 min, but newer techniques using single-phase cardiac magnetic resonance imaging have reduced scan times to about 12 min.[#36

Conclusion
Left ventricular hypertrophy is one criterion for beginning treatment in patients with borderline hypertension, and there is evidence that regression of left ventricular hypertrophy is associated with a reduction of risk in hypertension. For routine clinical work the simple and quick M-mode methods are probably adequate. However, more than one method of assessing the ventricular response should be quoted, including septal width, indexed left ventricular mass, diastolic function and, possibly, geometry. For research studies addressing regression of left ventricular hypertrophy, it is necessary to use either large population sizes or a technique with a lower variability such as two-dimensional echocardiography or magnetic resonance imaging.

REFERENCES

1: Am J Med 1983 Sep 26;75(3A):4-11