Measuring Ejection Fraction
Measuring Ejection Fraction
Abstracts & Commentary
Synopsis: Ejection fraction by gated single photon emission computed tomography is unpredictably inaccurate. Mitral annular motion on echocardiography can estimate ejection fraction, but they are not linearly related.
Sources: Vallejo E, et al. J Nucl Cardiol 2000;7:461-470; Emission K, et al. J Am Soc Echocardiogr 2000;13:896-901.
Left ventricular ejection fraction (ef) is a powerful predictor of mortality and an important guide to many therapies in heart disease. Thus, techniques for easily measuring or estimating EF are of interest. The latest generation of quantitative ECG gated single photon emission computed tomography (GSPECT-QGS) software is widely used for computation of EF from SPECT images. However, its accuracy in clinical situations remains undefined. Vallejo et al selected 990 patients from 1694 SPECT myocardial perfusion studies who also had first-pass radionuclide angiograms (FPRNA), but 279 FPRNAs had to be excluded for technical reasons. Also, 311 SPECT studies were excluded for technical problems. Therefore, the final study population was 400 patients. The overall correlation between GSPECT-QGS and FPRNA EFs was r = 0.66, SEE = 11.85%, and the mean QGS EF was lower (52 vs 58%; P < 0.001). Also, at low EFs the QGS EF was lower, but at high EFs it tended to be higher than the FPRNA EF. However, if only EFs less than 50% were considered, the correlation was better: r = 0.77, SE = 6.4%, and mean EF was similar by both methods (37 vs 34%; P = 0.07). In addition, agreement between QGS and FPRNA was better in hearts with large end-diastolic volumes. Vallejo and colleagues concluded that EF by GSPECT-QGS is often lower than with FPRNA and accuracy is adversely affected by technical and anatomic factors.
Comment by Michael H. Crawford, MD
In a time where hype and exaggeration seem to be the norm, it is nice to see the nuclear cardiologists publicly admitting they have a problem. However, the group from Yale has always been appropriately critical. Due to the underestimation of EF and the unpredictable inaccuracies caused by high extra cardiac activity, low count densities, and small left ventricles, which could not be reliably corrected with the automated program, Vallejo et al caution against using GSPECT-QGS for serial EF measurements to guide clinical decisions such as chemotherapy doses. On the other hand, they believe it is accurate enough for prognostication in coronary disease patients if one considers 0.45 the lower limit of normal.
The reason they provide such strong conclusions is the perceived strength of the study. They studied a large unselected group with a wide range of EFs (12-84%). Some of their patients had perfusion defects and ventricular enlargement. I was amazed at the technical difficulties they had—590 patients (60%) were excluded due to various technical factors that they thought would adversely affect the results. If a less experienced laboratory did not recognize some of these difficulties, their results could be even more inaccurate. The major limitation of this study was using FPRNA as the gold standard since it has its own problems. However, in their laboratory, FPRNA has a 5% variance and their results with QGS agree with experimental studies. Thus, GSPECT-QGS to estimate EF should be used with these limitations in mind.
In patients with technically inadequate echocardiograms for analyzing endocardial motion ordered for evaluation of LV function, inspection of mitral annular excursion in the apical four-chamber view has been suggested as a surrogate for estimating EF. Although originally the relationship between M-mode excursion of the mintral annulus and EF was thought to be linear, recent studies have cast doubt on this construct. Thus, Emisson et al studied 182 patients with excellent echocardiograms and no conduction abnormalities or arrhythmias. They performed a meta-analysis with their data and 252 patients from the literature, resulting in a total of 434 patients. EF was determined by the biplane Simpson’s rule technique. Use of a linear model showed an r2 of 0.74 for the correlation between mitral annular motion (MAM) and EF. Use of a curvilinear model had a higher r2 of 0.79. Also, at lower EF levels MAM overestimates EF using the linear model. As an example; a MAM of 4 mm would correspond to an EF of 23% with the linear model and 14% with a curvilinear model. In addition, the relationship between MAM and EF is negatively influenced by heart size. Thus, the larger heart sizes of the patients with lower EFs may contribute to the overestimation of EF in this range by the linear model. Vallejo et al present a new regression equation that takes LV size into consideration so one can calculate EF from a MAM measure, but they suggest that MAM be used alone as a measure of LV function.
Although the use of MAM alone as a measure of LV function makes sense theoretically, it is a tough sell to physicians used to the EF measure. Other indices of LV function, such as fractional shortening and fractional area change, have been suggested over the years, but none have caught on and supplanted EF. Despite the observation that MAM is not linearly related to EF especially at lower EFs, the commonly used clinical cut point of a MAM of 10 mm corresponds to an EF of 50% is validated by this study. Thus, a MAM less than 10 mm indicates an abnormally low EF in general. However, keep in mind that a MAM of 5 mm does not indicate an EF of 25% (linear) but rather an EF of 15%.
Subscribe Now for Access
You have reached your article limit for the month. We hope you found our articles both enjoyable and insightful. For information on new subscriptions, product trials, alternative billing arrangements or group and site discounts please call 800-688-2421. We look forward to having you as a long-term member of the Relias Media community.