It is no exaggeration that echocardiography, because of the comprehensive information it generates and its noninvasive, real-time format, is indispensible in modern medicine. These features have made it the most frequently performed cardiac imaging test, whose images, by the physical principles underlying their construction, are accurate representations of cardiac anatomy and absolutely quantifiable. However, the freehand nature of measurement techniques, the susceptibility of images to technical error and degradation, and the non-Euclidean shapes of cardiac structures (among other things) may lead to erroneous estimation of cardiac size. Volumetric algorithms incorporate larger samples of anatomy in computing cardiac size than single-dimensional linear measurements and are thus superior to single dimensions. However, volumetric measurements have diffused only slowly into practice despite their historic contributions to the development of the cardiology canon through pioneering volumetric methods such as quantitative angiography. A leading reason that echocardiographers avoid volumetric methods is the notion that they are “too time consuming.” However, as large clinical trials have proliferated, considerations of cost and patient safety and tolerance, in conjunction with deploying reference standard techniques such as magnetic resonance imaging, computed tomography, or scintigraphy, have resulted in a growing interest in relying on volumetric data from echocardiography. This shift has been facilitated by published guidelines for both the content and conduct of clinical trials that emphatically endorse volumetric echocardiography.
Recognizing this trend and seeking to improve the application echocardiography in clinical trials, the members of the Echocardiography Core Laboratory (ECL) of the Surgical Treatment for Ischemic Heart Failure (STICH) trial have analyzed their experience with baseline volumetric and Doppler data that were gathered at the onset of the trial and, on the basis of their impressions, present suggestions for improvement in clinical trials and clinical practice.
STICH is a large multicenter study of treatment strategies in advanced ischemic cardiomyopathy. Entry criteria include an ejection fraction (EF) < 35% and coronary obstruction amenable to bypass grafting. Between 2002 and 2006, 2,136 patients were enrolled and studied from 122 sites worldwide. The two hypotheses tested in this study were that surgery is associated with superior long-term survival compared with intensive medical therapy and that surgical resection of anterior dysfunctional myocardial segments combined with bypass has superior long-term survival to bypass alone. Echocardiographic data were collected as adjunctive to understanding the outcome end points. In their analysis of initial baseline data reported in this issue of JASE , Oh et al. aim to (1) define the feasibility of obtaining quality baseline data for the STICH cohort, (2) document baseline data from this cohort, and (3) generate recommendations for using echocardiography clinically.
Left Ventricular Volumetric Measurements in the Surgical Treatment for Ischemic Heart Failure Trial
Arguably, the most striking finding in the STICH baseline study was the disconcertingly low yield (43.5%) of paired apical images (two-chamber and four-chamber views), where both were of adequate quality to permit planimetric areas to be combined for calculating ventricular volume by the biplane method of discs. The potential for an adverse impact of this shortfall on achieving an adequately powered sample size for subgroup analysis is probably large. The problem may amplify as the study continues, because the yield of patients returning for second or third studies is usually no better than 80%. In my group’s research and clinical practice, we are able to obtain measureable biplane images in close to 90% of subjects. Factors that might account for the low yield in STICH include the nature of the enrolled population. Severe left ventricular (LV) dilation was universal among enrollees; the mean population LV end-diastolic volume index was >115 mL/m 2 , compared with the established upper normal value of 74 mL/m 2 . In these large, spherically dilated ventricles, the displaced lateral wall in apical views is often obscured by rib shadow; lung; or poor lateral ultrasound beam formation, focus, and power. Although the 122 sites that contributed patients were given a manual of technical guidelines and were vetted by the ECL through prestudy sample images and feedback, most of the technical requirements contained in this manual are not described in the report. One important detail that is described is the method for choosing end-systolic and end-diastolic frames for measurement as the “time of the QRS.” In the STICH population, in which conduction delays and asynchronous contraction are common, precise selection of end-systolic and end-diastolic frames is required to avoid introducing considerable nonbiologic variability. To circumvent this problem, one should choose the frame before the onset of mitral opening as end-diastole and the frame before mitral closure as end-systole. If the STICH ECL had used this method, a potential source of variability could have been eliminated. Regardless of methodologic considerations, the low yield of measureable biplane images is an eloquent argument for the routine use of left-sided contrast in clinical trials, because this practice ensures a high yield of volumetrically accurate and reproducible images. Increased yield from contrast use would have compensated for the additional cost of contrast agent.
Volumes and EF
In the ECL, when one of the two apical views was not of sufficient quality, the data from the single-plane method of discs were substituted. Among patients with both two-chamber and four-chamber views of good quality, there was a good correlation between the two views. On the basis of this correlation, the investigators adopted the practice of using a single plane as a biplane substitute. Although there is justification for this practice in symmetric hearts, the high prevalence of major segmental disease and irregular geometry in STICH may introduce a substantial number of data points that are unrepresentative of global ventricular geometry and function. Again, left-sided border detection by contrast would have made this compromise unnecessary.
EF
When there were neither two-chamber nor four-chamber views of sufficient quality to perform planimetry, the readers at the ECL used the visual method of estimating EF. Contrast opacification of the left ventricle would have made this controversial practice unnecessary. In our experience, the visual method of estimating EF is inaccurate and variable among readers and should not be used in a large clinical study. Accuracy of visual EF is particularly compromised in ventricles with heterogeneous contraction patterns, because it is difficult to mentally integrate inhomogeneous wall motion imposed by infarcted segments and conduction delays. In a study in which a degree of certainty is sought, contrast enhancement and objective measurement are recommended, and visual estimation is proscribed. The finding that 18.5% of patients exceeded the entry criteria of EF < 35% when measured by the ECL is difficult to interpret because of uncertainty about how many of the 18.5% visually estimated data points were used in this calculation. Regardless, the use among the 122 contributing sites of a variety of methods to measure EF was also contributory to variation in the estimation of EF.
End-Systolic Volume (ESV)
End-diastolic LV volume is dependent on preload and varies with intravascular volume shifts. ESV depends on the contractile state and when displayed in combination with LV end-systolic pressure (i.e., as a pressure volume loop) is termed elastance. On the basis of research conducted by our group using data from the Heart and Soul Study and in clinical practice, ESV indexed to body surface area (ESVI) is one of the most informative and reliable values considered in interpreting an echocardiogram. In Heart and Soul Study, a longitudinal study of outcomes in 1,024 patients with coronary artery disease, ESVI was a more powerful predictor of heart failure hospitalization than either EF or end-diastolic volume index. Figure 1 is a locally weighted smoothed scatterplot of the proportion of patients hospitalized with heart failure during follow-up of 5.5 years and graphically depicts the relationship between ESVI and heart failure outcomes in the overall cohort over the 5.5 years of follow-up. Note that only a modest increase in ESVI to >25 mL/m 2 is accompanied by a sharp rise in congestive heart failure events during 5-years follow-up.
