Podgoreanu, Mihai V. MD*,; Djaiani, George N. MD†,; Davis, Elizabeth LPN, RDCS‡,; Phillips-Bute, Barbara PhD*, and; Mathew, Joseph P. MD*
Conventional real-time echocardiographic assessment of regional systolic left ventricular (LV) function is based on visual interpretation of the direction and amplitude of endocardial motion and wall-thickening characteristics, all highly subjective, qualitative, and experience dependent (1,2). The sensitivity and specificity of recognizing intraoperative ischemic episodes echocardiographically in real time are only moderate (76%), with substantial variability in evaluating the severity of regional dysfunction (3). Previous studies have demonstrated that alterations in the normal temporal pattern of the LV contraction/relaxation sequence (LV asynchrony) represent an early, more sensitive marker of myocardial ischemia, preceding changes in the amplitude of wall motion, thickening/thinning variables, and global LV ejection phase indices (4–6). Such temporal abnormalities of contraction are manifested either as a delayed onset of endocardial motion in the first third of ejection (tardokinesis) (5,7) or as persistent contraction after aortic valve closure (postsystolic shortening) (4). However, the visual assessment of LV asynchrony is limited and not improved by further review (8).
Color kinesis (CK), an echocardiographic technique based on automated border detection, compares tissue backscatter values between successive acoustic frames, detects pixel transitions between blood and myocardium, and color-encodes these transitions in real time. Segmental analysis of CK images thus provides an objective method for quantifying the spatiotemporal aspects of regional and global LV endocardial motion (9,10). Normal values for quantitative CK indices of LV wall motion have been recently established in a multicenter study (11), and the method has been validated to automatically detect both resting (10) and stress-induced wall motion abnormalities (12) in patients with coronary artery disease and in patients with dilated cardiomyopathy (13). However, intraoperative experience in diagnosing regional LV dysfunction has not been reported.
Accordingly, the purpose of this study was to compare the ability of segmental analysis of CK images and that of conventional visual assessment of two-dimensional echocardiography to diagnose regional wall motion abnormalities (RWMA) in cardiac surgical patients. We hypothesized that quantitative analysis of CK images, when compared with an experienced reviewer, would improve sensitivity, specificity, and diagnostic accuracy in detecting wall motion abnormalities over less experienced reviewers of conventional grayscale images. An additional aim of our study was to assess the temporal pattern of LV systolic endocardial motion by using quantitative analysis of CK images and to estimate the incidence of tardokinesis in a population of unselected patients undergoing surgical coronary revascularization. We postulated that such temporal analysis might help improve the identification of regional and global LV systolic dysfunction.
This study was performed with IRB approval. Midpapillary transgastric short-axis views were obtained during the precardiopulmonary bypass intraoperative transesophageal echocardiographic examination in 32 consecutive patients undergoing coronary artery bypass grafting surgery, using the Philips SONOS 2500 and 5500 ultrasound systems. The anesthetic regimen consisted of weight-based infusions of sufentanil/midazolam and pancuronium bromide. Exclusion criteria were suboptimal image quality (<75% of the endocardial border visualized in real time) (9–14), rhythm disturbances, and abnormal regional wall motion due to left bundle branch block, ventricular pacing, or previous sternotomy. After image quality and gain controls were optimized for endocardial tracking by the automated border detection system (15), CK software was activated to color-encode systolic endocardial excursion (at 30 frames per second) triggered by the R wave of the electrocardiogram. To minimize translation artifacts due to mechanical ventilation, CK data were acquired during end-expiratory apnea. Images were digitally stored as continuous loops on a magnetooptical disk for off-line analysis.
Conventional assessment of regional LV function was performed on images without CK overlays by an expert reviewer (director of perioperative echocardiography program) (16) using the American Society of Echocardiography guidelines for segmentation of standard views (1) and was graded on a segment-by-segment basis as normal (0) or abnormal (hypokinetic, 1; akinetic, 2; dyskinetic, 3). Two less experienced reviewers, cardiothoracic anesthesiology fellows who completed their basic training in perioperative echocardiography (16), also independently graded the same images. All reviewers were blinded to the results of CK analysis.
Digitized CK overlays were analyzed with custom software (Quantitative Color Kinesis; EchoSoft Co., Wilmington, DE) as previously described (10). End-systolic LV color-encoded images were automatically divided into 6 60° wedge-shaped segments based on the standard segmentation scheme (1), with the 0° line defined by the LV cavity centroid and a manually determined anatomic landmark at the anterior junction of the right ventricular endocardium with the interventricular septum (Fig. 1). The number and size of segments used in the automated CK analysis followed the same conventions used by reviewers of grayscale images. Dyskinetic segments were excluded from the quantitative analysis because wall dyskinesis results in paradoxical pixel transitions that are assigned the same red color by the CK software, regardless of the timing of transition, thus precluding the assessment of temporal patterns of endocardial motion.
The magnitude of endocardial motion was quantified for each segment as the incremental regional fractional area change (RFAC), normalized to the regional end-diastolic area, and displayed as stacked histograms (Fig. 2A) (10). To account for the regional heterogeneity of normal LV wall motion (17), results for each segment were compared with the corresponding normal range, previously established in a multicenter CK study from 141 adult healthy subjects (11), and defined as ±1 sd around the mean of the normal group (9,10). Values ranging between 1 and 2 sd less than the normal mean indicated hypokinesis, whereas values <2 sd less than the normal mean indicated akinesis. Thresholds were defined on the basis of the narrow range (i.e., small sd) in the normal group, thus enhancing the sensitivity and accuracy of this method in diagnosing RWMA, and have been previously validated (9–12,18). Global fractional area change (FAC) was also automatically calculated by using combined pixel counts from all endocardial segments.
The temporal pattern of endocardial motion was quantified by computing the ejection rate as a function of time normalized to end-diastolic area, with segments displayed as stacked time histograms. This allows easy identification of both the peak ejection rate (PER) and the timing of the peak LV ejection (TPER). Global tardokinesis was defined as TPER more than one-third of the ejection time (Fig. 2B), on the basis of previous instantaneous frame-by-frame analyses of time to peak ejection in normal subjects and in patients with coronary artery disease (5).
Regional timing of ejection was quantified by use of two variables, as previously described (9,10). First, the regional mean time of ejection (RMTE) represents the average time required for pixels in each segment to change their blood/tissue attribute during ejection. Regional tardokinesis was defined as RMTE >1 sd above the normal mean for the respective segment (11). Second, the percentage of regional contraction completed at 50% ejection time (%RFAC50) was obtained from RFAC time curves normalized to reach 100% endocardial motion at 100% ejection time. This allows for intersegmental comparisons by eliminating the differences in magnitude of endocardial motion and the confounding effects of heart rate on the duration of systole (9–11). Both variables have been previously validated in experimental models of ischemia (7) and in clinical studies (9,10).
The intersegmental sd of the RMTE for each individual patient was used as an index of temporal heterogeneity (asynchrony) in regional systolic endocardial motion. CK images were analyzed by one observer. To assess intra- and interobserver variability, 10 randomly selected CK images (60 segments) were later analyzed by the same observer and by a second independent observer.
On the basis of a probability of disagreement between the expert grading of RWMA and quantitative CK of 17%(9), we calculated a sample size of 140 segments to estimate the level of intertechnique agreement with a 95% confidence interval. This sample size gives quantitative CK analysis a 95% power for detecting differences more than 1 sd in both RFAC and RMTE with a significance level of 0.05. The number of segments requiring duplicate assessments (n = 60) was calculated to estimate with 95% confidence the intra- and interobserver variability, on the basis of an anticipated probability of error of 11%(9,10).
Agreement between the expert grading, quantitative CK analysis, and each of the two less experienced reviewers was assessed by using the κ coefficient; chance agreement is reflected by a κ of 0, whereas a κ of 1.0 constitutes perfect agreement (19). The evaluation was performed separately on ratings of regional wall motion as “normal” or “abnormal” and after taking into account the severity of the abnormality. Sensitivity, specificity, and positive and negative predictive values were also calculated for the quantitative CK method and for the less experienced reviewers, with expert grading as the gold standard.
To analyze the magnitude and temporal pattern of endocardial motion, the standard ejection phase indices (FAC and PER) and the index of temporal regional asynchrony (sd of RMTE) were averaged in the groups of patients in whom global tardokinesis was present versus absent and were subjected to Student’s t-test. The RFAC and RMTE values were converted to z scores (units of sd from the mean of a normal reference population) (18) to allow intersegment and intersubject comparisons. The relationship between the degree of regional hypokinesis (RFAC z score) and the severity of regional tardokinesis (RMTE z score) was assessed by simple linear regression. %RFAC50 was compared between severely hypokinetic/akinetic segments and mildly hypokinetic segments by using the two-tailed Mann-Whitney U-test. The reproducibility of quantitative CK measurements (RFAC and RMTE) was expressed for both intra- and interobserver variability as the absolute difference between repeated measurements in the percentage of their mean, averaged for all analyzed segments. P < 0.05 was considered significant.
Four patients (12.5%) had inadequate automated border detection and were excluded from the study. CK files were corrupted during the digital storage and retrieval process in two additional patients and could not be used in further analysis. In the remaining 26 patients, 6 dyskinetic segments (3.8%) were excluded, resulting in 150 evaluable segments.
The concordance with the expert grader of each of the less experienced reviewers and of the quantitative CK method in diagnosing abnormalities in the magnitude of regional wall motion is presented in Table 1. Detailed statistics of diagnostic accuracy for the two methods are presented in Table 2.
Global tardokinesis was identified in 9 (35%) of the 26 evaluable patients. Of these, two had normal and seven had moderately reduced global FAC. The group of patients in whom global tardokinesis was present had no statistically significant differences in FAC but showed a trend toward a decreased PER (P = 0.06) when compared with patients without global tardokinesis (Table 3).
Regional tardokinesis was identified in 48 (32%) of the 150 analyzed segments. Of these, on the basis of their RFAC, 27 (56%) segments had a normal magnitude of wall motion, 18 (38%) were hypokinetic, and 3 (6%) were severely hypokinetic to akinetic. The distribution of tardokinetic segments by location was as follows: inferior, n = 3; septal, n = 7; anteroseptal, n = 4; anterior, n = 9; lateral, n = 16; and posterior, n = 9. We found a poor correlation (r = −0.13) between the severity of hypokinesis and the degree of tardokinesis, as assessed by the respective z scores. However, the %RFAC50, an index of regional tardokinesis, was larger in the severely hypokinetic/akinetic segments (median, 68; Q1, 61; Q3, 83) than in the mildly hypokinetic segments (median, 46; Q1, 32; Q3, 55) (P < 0.0001;Fig. 3). This reflects an increased dependency on late ejection in segments with mild abnormalities in the amplitude of endocardial motion, whereas the residual motion in severely dysfunctional segments occurs earlier in systole.
The sd of the intersegmental RMTE (normal, 12 ms) (10), an index of temporal heterogeneity of endocardial motion, ranged from 14 to 54 ms, with a median of 29 ms. The sd of RMTE was higher in the group that displayed global tardokinesis (P = 0.02) (Table 3).
The interobserver variability for the RFAC and RMTE measurements was 17.1% and 5.4%, respectively. Intraobserver variability was 16.2% for RFAC and 8.5% for RMTE.
In this study, we demonstrated that a quantitative echocardiographic method for evaluating regional LV systolic function on the basis of segmental analysis of CK images in patients undergoing cardiac surgery compares favorably with the off-line interpretation of RWMA by an expert grader. Diagnostic agreement and accuracy of RWMA can be improved by CK. Our findings are consistent with the results of previous studies assessing the diagnostic accuracy of CK for detection of both resting and stress-induced RWMA in patients with coronary artery disease (20,21). We have also shown that global/regional tardokinesis may be present even when standard ejection phase indices are normal. Quantitative CK analysis revealed a biphasic temporal pattern of regional contraction: mildly hypokinetic segments showed delayed systolic motion, whereas residual motion of severely hypokinetic/akinetic segments occurred in early systole, reflecting passive effects produced by adjacent myocardial contraction.
Nonuniformity of mechanical performance is inherent to the multicellular nature and geometry of the ventricular myocardium and has been described in terms of myoarchitecture, electrical excitation, and activation-contraction coupling. In the ischemic heart, inappropriately increased spatial and temporal nonuniformity results in an imbalance of forces and distribution of load; this reduces mechanical efficiency during contraction and, especially, during relaxation (22). One manifestation of this increased temporal nonuniformity, the regional delay in the onset of endocardial motion (tardokinesis), is an important early marker of ischemia and precedes changes in either amplitude of motion or wall thickening (5,6). Derumeaux et al. (23) demonstrated that an ischemia-induced delay of 30 ms in a myocardial segment is associated with a >70% stenosis in the coronary vessel supplying that segment. It is important to note that such abnormalities in endocardial motion during the early phase of systole are missed by any of the standard global ejection phase indices (ejection fraction, mean/PER, mean time of ejection, velocity of circumferential fiber shortening), which provide only an average of all the events occurring during systole (24,25). Furthermore, the accuracy of experienced observers in identifying LV myocardial asynchrony in a two-dimensional echocardiographic image is limited, because regional delays of <89 ms are not recognized in a moving image. Accuracy is not improved even with additional review because of the temporal resolution of both the ocular system and the data acquisition and display methodology (8). In this study, by using segmental analysis of CK images, we quantified the asynchrony in the temporal pattern of LV contraction in patients undergoing coronary artery bypass grafting surgery. Delayed systolic endocardial motion (global and regional tardokinesis), common among cardiac surgery patients, was found even when standard ejection phase indices (i.e., FAC, RFAC) were within the normal range. Previous studies have demonstrated that such a slow global ejection rate in early systole relative to mid or late systole is equivalent to a decreased acceleration of blood ejection from the LV—a particularly sensitive indicator of ventricular performance (5,24,25). Experimental studies on myocardial ischemia have shown that a decrease in maximum acceleration can occur before any changes in other contractile indices (26). From this perspective, the delayed TPER found in 35% of our patients (i.e., global tardokinesis) suggests an initial worsening of ventricular performance. Furthermore, increased intersegmental heterogeneity in the temporal pattern of LV contraction (as quantified by the sd of RMTE) is associated with and may be responsible for decreased acceleration of blood ejection because of asynchronous, inefficient contraction. This is consistent with previous findings that asynchronous LV contraction may result in decreased initial LV impulse (26).
Quantitative CK analysis of the relationships between the magnitude and timing of myocardial motion on a regional basis by using normalized time curves (%RFAC50) revealed a biphasic pattern of regional systolic endocardial motion. Although this study was not designed to follow the progression of RWMA, our findings suggest that regional delays in contraction (tardokinesis) are usually associated with mild abnormalities in the magnitude of endocardial motion, whereas residual motion in severely dysfunctional (akinetic) segments occurs earlier in systole. This is also consistent with results from previous studies (7,10) and likely reflects passive motion in the akinetic segments caused by contraction of adjacent myocardial tissue. Therefore, simultaneous analysis of the amplitude and timing of regional endocardial motion by CK may help exclude such tethering effects that cloud the visual interpretation of regional ventricular function. We found that more than half of the tardokinetic segments had a normal amplitude of motion, thus confirming that regional delay in the onset of endocardial motion may occur independently of the changes in the magnitude of wall motion (5–7).
Some study limitations must be considered when interpreting our findings. First, similar to other echocardiographic techniques, the success of CK depends on image quality. Consistent with previous reports (9,26), accurate tracking of endocardial motion by CK was obtained in most segments in 87% of consecutive patients. Second, the reference for automated quantification of RWMA was represented by CK patterns of contraction obtained by transthoracic echocardiography in healthy subjects. Although such normal CK patterns were found to be reproducible and highly consistent, this technique may be limited when anesthetic regimens with negative inotropic effects are administered and may necessitate the redefinition of normal ranges in anesthetized patients. Third, the relatively low temporal resolution (33 ms) imposed by the fixed frame rate (30 Hz) at which CK operates may restrict the definition of endocardial motion at increased heart rates and may result in inaccurate temporal indices of regional motion. The newer version of high-frame-rate CK has corrected this deficiency. Fourth, we limited the analysis of RWMA to a single short-axis echocardiographic plane. However, the technique allows similar evaluations to be performed with long-axis views.
An additional limitation of our study is the use of a fixed reference system to analyze the images. Because CK adopts a fixed reference system for real-time tracking of endocardial motion, we minimized any potential translation artifacts by acquiring data during apnea. Finally, despite the real-time color encoding of endocardial motion, the quantitative analysis of the CK images remained an off-line process. Nevertheless, the short average time required to complete the analysis (approximately two minutes) and the user-friendly software package create the possibility for intraoperative on-line availability of such analyses.
The results of this study indicate that quantitative CK analysis may be an important supplement to visual assessment of regional LV dysfunction. Quantitative CK provides a fast, objective, sensitive, reproducible, and automated measurement of the magnitude and temporal nature of intraoperative regional ischemic changes. By diagnosing tardokinesis, common among cardiac surgical patients even with normal standard ejection phase indices, quantitative CK may improve the intraoperative detection of regional ischemic changes. Simultaneous acquisition of CK and myocardial perfusion images may facilitate the distinction between stunned, hibernating, and ischemic myocardium, with important therapeutic consequences.
1. Schiller NB, Shah PM, Crawford M, et al. Recommendations for quantitation of the left ventricle by two-dimensional echocardiography: American Society of Echocardiography Committee on Standards, Subcommittee on Quantitation of Two-Dimensional Echocardiograms. J Am Soc Echocardiogr 1989; 2: 358–67.
2. Picano E, Lattanzi F, Orlandini A, et al. Stress echocardiography and the human factor: the importance of being expert. J Am Coll Cardiol 1991; 17: 666–9.
3. Bergquist BD, Leung JM, Bellows WH. Transesophageal echocardiography in myocardial revascularization. I. Accuracy of intraoperative real-time interpretation. Anesth Analg 1996; 82: 1132–8.
4. Ehring T, Heusch G. Left ventricular asynchrony: an indicator of regional myocardial dysfunction. Am Heart J 1990; 120: 1047–57.
5. Cucchini F, Baldi G, Barilli AL, et al. Tardokinesis in coronary artery disease: evidence with instantaneous analysis of left ventricular ejection. Eur J Cardiol 1981; 12: 153–66.
6. Guth BD, Schulz R, Heusch G. Time course and mechanisms of contractile dysfunction during acute myocardial ischemia. Circulation 1993; 87: IV35–42.
7. Mor-Avi V, Collins KA, Korcarz CE, et al. Detection of regional temporal abnormalities in left ventricular function during acute myocardial ischemia. Am J Physiol Heart Circ Physiol 2001; 280: H1770–81.
8. Kvitting JP, Wigstrom L, Strotmann JM, Sutherland GR. How accurate is visual assessment of synchronicity in myocardial motion? An in vitro study with computer-simulated regional delay in myocardial motion: clinical implications for rest and stress echocardiography studies. J Am Soc Echocardiogr 1999; 12: 698–705.
9. Lang RM, Vignon P, Weinert L, et al. Echocardiographic quantification of regional left ventricular wall motion with color kinesis. Circulation 1996; 93: 1877–85.
10. Mor-Avi V, Vignon P, Koch R, et al. Segmental analysis of color kinesis images: new method for quantification of the magnitude and timing of endocardial motion during left ventricular systole and diastole. Circulation 1997; 95: 2082–97.
11. Mor-Avi V, Spencer K, Gorcsan J, et al. Normal values of regional left ventricular endocardial motion: multicenter color kinesis study. Am J Physiol Heart Circ Physiol 2000; 279: H2464–76.
12. Koch R, Lang RM, Garcia MJ, et al. Objective evaluation of regional left ventricular wall motion during dobutamine stress echocardiographic studies using segmental analysis of color kinesis images. J Am Coll Cardiol 1999; 34: 409–19.
13. Vandenberg BF, Oren RM, Lewis J, et al. Evaluation of color kinesis, a new echocardiographic method for analyzing regional wall motion in patients with dilated left ventricles. Am J Cardiol 1997; 79: 645–50.
14. Lindower PD, Rath L, Preslar J, et al. Quantification of left ventricular function with an automated border detection system and comparison with radionuclide ventriculography. Am J Cardiol 1994; 73: 195–9.
15. Bednarz JE, Marcus RH, Lang RM. Technical guidelines for performing automated border detection studies. J Am Soc Echocardiogr 1995; 8: 293–305.
16. Cahalan MK, Abel M, Goldman M, et al. American Society of Echocardiography and Society of Cardiovascular Anesthesiologists task force guidelines for training in perioperative echocardiography. Anesth Analg 2002; 94: 1384–8.
17. Pandian NG, Skorton DJ, Collins SM, et al. Heterogeneity of left ventricular segmental wall thickening and excursion in 2-dimensional echocardiograms of normal human subjects. Am J Cardiol 1983; 51: 1667–73.
18. Gelberg HJ, Brundage BH, Glantz S, Parmley WW. Quantitative left ventricular wall motion analysis: a comparison of area, chord and radial methods. Circulation 1979; 59: 991–1000.
19. Fleiss JL. Statistical methods for rates and proportions. 2d ed. New York: Wiley, 1981.
20. Hartmann T, Kolev N, Blaicher A, et al. Validity of acoustic quantification colour kinesis for detection of left ventricular regional wall motion abnormalities: a transoesophageal echocardiographic study. Br J Anaesth 1997; 79: 482–7.
21. Vitarelli A, Sciomer S, Schina M, et al. Detection of left ventricular systolic and diastolic abnormalities in patients with coronary artery disease by color kinesis. Clin Cardiol 1997; 20: 927–33.
22. Brutsaert DL. Nonuniformity: a physiologic modulator of contraction and relaxation of the normal heart. J Am Coll Cardiol 1987; 9: 341–8.
23. Derumeaux G, Ovize M, Loufoua J, et al. Doppler tissue imaging quantitates regional wall motion during myocardial ischemia and reperfusion. Circulation 1998; 97: 1970–7.
24. Johnson LL, Ellis K, Schmidt D, et al. Volume ejected in early systole: a sensitive index of left ventricular performance in coronary artery disease. Circulation 1975; 52: 378–89.
25. Slutsky R, Battler A, Karliner JS, et al. First-third ejection fraction at rest compared with exercise radionuclide angiography in assessing patients with coronary artery disease. Radiology 1980; 136: 197–201.
26. Rushmer RF. Initial ventricular impulse: a potential key to cardiac evaluation. Circulation 1964; 29: 268–83.