Prognostic Significance of Ultrasound Myocardial Tissue Characterization in Patients With Cardiac Amyloidosis
Background— Cycle-dependent variation of myocardial integrated backscatter (CV-IB) is an objective measurement that may detect myocardial abnormalities. However, no data exist about the prognostic value of CV-IB in primary cardiac amyloidosis.
Methods and Results— We prospectively examined 208 consecutive biopsy-proven patients with primary amyloidosis. The magnitude of CV-IB was analyzed at the interventricular septum and left ventricular (LV) posterior wall and its prognostic value was compared with standard Doppler measurements with to the Tei index (isovolumic contraction time plus isovolumic relaxation time divided by ejection time). One hundred thirty-three patients had cardiac involvement (mean LV thickness > 12 mm). Forty-one patients (20%) (32 cardiac deaths) died during a mean follow-up of period of 307±156 days. Univariate analysis showed that the CV-IB at the LV posterior wall was the best predictor of cardiac death (P<0.0001) and all-cause death (P< 0.0001). The Tei index did not identify patients at risk of death. Multivariate analysis showed that CV-IB at the LV posterior wall was the only independent predictor of both cardiac and overall deaths.
Conclusions— Among patients with cardiac amyloidosis, CV-IB at the LV posterior wall is a powerful predictor of clinical outcome and is superior to standard echocardiographic/Doppler flow indexes.
Received February 20, 2002; revision received May 13, 2002; accepted May 13, 2002.
Cardiac amyloidosis is characterized by increased left ventricular (LV) wall thickness, normal or decreased LV cavity size, and congestive heart failure (CHF) with normal or near-normal fractional shortening.1– 7⇓⇓⇓⇓⇓⇓ A marked increase in wall thickness, reduced LV systolic function,2 shortened deceleration time, and increased early diastolic filling velocity to atrial filling ratio7 have all been proposed as predictors of cardiac death, but it has been suggested that a Doppler-derived index of combined systolic and diastolic myocardial performance (known as the Tei index) is a more useful predictor of clinical outcome.8 Because most of these studies were retrospective or contained very few patients, we sought to examine prospectively a large group of patients to reevaluate these parameters. Furthermore, because cardiac amyloidosis is associated with abnormal echogenicity, we investigated the prognostic value of acoustic quantification with cyclic variation (CV) of integrated backscatter (IB) from the myocardium during the cardiac cycle.9–12⇓⇓⇓
Acoustic quantification has been performed with IB ultrasonic imaging under a variety of conditions such as the aging heart,13 dilated cardiomyopathy,14,15⇓ hypertrophic cardiomyopathy,16–18⇓⇓ myocardial ischemia,19–21⇓⇓ diabetes mellitus,22,23⇓ and cardiac allograft rejection,24,25⇓ and various abnormalities have been described. We sought to clarify the prognostic value of cycle-dependent variation of myocardial integrated backscatter (CV-IB) analysis in primary (AL) amyloidosis and to compare its prognostic value with that of the previously described 2-dimensional and Doppler flow data.2,5–8⇓⇓⇓⇓
Two hundred thirty-four consecutive biopsy-proven patients with AL amyloidosis entered the study. All patients were evaluated at the Boston University Amyloidosis Treatment and Research Center between March 21, 2000, and April 11, 2001. The diagnosis of amyloidosis was made when the biopsy specimen of an involved organ, viewed under polarized light, demonstrated typical Congo red birefringence. A monoclonal protein in the serum or urine and/or a monoclonal population of plasma cells in the bone marrow (evaluated by immunohistochemistry) confirmed AL amyloidosis.4 Patients with familial, secondary (AA), or senile amyloidosis were excluded. Patients with AL amyloid and poor echocardiographic quality (n=16), atrial fibrillation (n=3), hypertension (n=5), or significant valvular heart disease (n=2) were also excluded, leaving 208 patients. Cardiac involvement in AL amyloidosis is generally considered to be present when increased wall thickness is present in the absence of hypertension, valvular disease, or criteria for LV hypertrophy on the ECG.3 Thus, cardiac involvement was defined as a mean value of LV thickness (half of the sum of the thickness of interventricular septum and posterior walls) ≥12 mm.
All echocardiograms were reviewed by two readers who concurred on the presence or absence of cardiac involvement. Of the 208 patients, 133 had cardiac involvement and 75 had no echocardiographic features of amyloidosis. The latter group was defined as group 1 (noncardiac amyloid). CHF was defined as dyspnea on exertion, associated with orthopnea, paroxysmal nocturnal dyspnea, or a chest radiographic appearance of heart failure and/or the presence of elevated jugular venous pressure. The diagnosis of CHF was made by one of the investigators, who obtained a detailed history and examined each patient without knowledge of the results of the IB.
Of the 133 patients with cardiac amyloidosis, 61, defined as group 3, had prior or current evidence of CHF. The 72 with cardiac amyloid but no CHF were defined as group 2. Follow-up data were obtained from correspondence with patients’ families or from the attending physician. Written informed consent was obtained from each patient.
Standard Ultrasound Examination and Measurements
Ultrasound examinations were performed with a commercially available echocardiographic machine (Sonos 5500, Hewlett-Packard). Standard M-mode measurements of the LV were made. Transmitral flow velocity was recorded from the apical 4-chamber view, with the sample volume at the mitral valve tips. The sample volume was also placed in the area of the anterior mitral valve leaflet to record simultaneous LV outflow tract and transmitral flow profiles (Figure 1). Pulsed Doppler of pulmonary venous and LV outflow tract velocity patterns was recorded, positioning the sample volume at the right upper pulmonary vein 1cm below the ostium and just below the aortic valve, respectively, in the apical 4-chamber view.
Analysis of Doppler flow was performed with computerized planimetry (Color Cineview Plus, TomTec Imaging Systems). Three or more consecutive beats were averaged for each measurement. The peak velocity of early (E) and late filling (A) waves, E/A ratio, deceleration time of E wave and peak velocities of systolic (S), diastolic (D), A waves, and the D/S ratio of pulmonary venous flow were measured. (Figure 1). Isovolumic relaxation time was measured as the time between the end of LV outflow tract flow and the onset of transmitral flow (Figure 1). Isovolumic contraction time was measured as the time between the R wave and the onset of LV outflow. Ejection time was measured as the duration of LV outflow velocity profile. The Tei index was calculated as the sum of isovolumic contraction time and isovolumic relaxation time divided by ejection time.8
Acquisition and Analysis of IB Data
A software package (Acoustic Densitometry) incorporated into the echocardiography machine was used with a wide-band sector transducer (S4) (Figure 2). This provides an IB image in which the gray level is displayed proportional to the IB power. Sixty frames from consecutive cardiac cycles (30 Hz) are displayed, captured into cine-loop memory, and digitally stored on magneto-optical disk. The dynamic range of the IB image is ∼64 dB. The parasternal short-axis view of the LV was obtained at the level of the chordae. The time-gain compensation was adjusted to give equal gain at each depth of the LV in each patient. For IB analysis, the backscatter images were retrieved from the optical disk into the system memory. An 11×11- or 21×21-pixel elliptical region of interest was placed in the interventricular septum and posterior wall, and the region of interest was manually adjusted to avoid specular reflections from interfaces between endocardium and blood. From the time-intensity waveforms, the magnitude of CV-IB was calculated as the difference between the minimal and maximal values in a cardiac cycle, averaged over 3 consecutive beats (Figure 2). Two cine loops were acquired consecutively to analyze 3 cardiac cycles. All data from the ultrasound examination were analyzed by one investigator, who was unaware of patients’ clinical information.
Data are expressed as mean value±SD. Statistical analyses were done with a commercially available software program (StatView 5.0, SAS Institute Inc). Comparisons of the values of CV-IB between the interventricular septum and LV posterior wall were made by means of the 2-tailed paired t test. Differences among 3 groups were assessed with the χ 2 test for categorical variables, and comparisons of continuous variables among 3 groups were made by means of 1-way factorial analysis of variance, followed by the Scheffé test. A probability value <0.05 was considered significant.
The median values of variables were used to divide patients into two groups when survival free of cardiac death and overall survival were estimated by use of the Kaplan-Meier method.26 We also evaluated values defined in previous studies to divide patients into two groups.2,7,8⇓⇓ To determine suitable cutoff points of CV-IB, we constructed receiver operating characteristic curves to determine values in which sensitivity was as close as possible to specificity. The area under the curve was measured to determine a summary measure of performance. Univariate analyses were followed by a log-rank test. Multivariate analyses to determine the relative contribution of variables were examined by Cox proportional hazards regression model.27 A 0.05 level of significance was applied to determine whether variables were added or removed from the model.
The clinical characteristics are shown in Table 1. There were more men in the groups with cardiac involvement (groups 2 and 3) and the heart rate was significantly greater in group 3 than that in other 2 groups.
Two-Dimensional Echocardiographic and Doppler Measurements
No echocardiographic or Doppler features differentiated group 2 from group 1, other than the predefined wall thickness (Table 1). Patients in group 3 had more abnormalities than either of the other 2 groups; LV systolic dimension and left atrial diameter were significantly greater in group 3 than in groups 1 and 2, the LV diastolic dimension in group 3 was smaller than in group 1, and the LV fractional shortening was smaller in group 3 than in the other 2 groups. Transmitral flow, pulmonary venous flow, and Doppler time intervals are shown in Table 1. Patients in group 3 showed greater values in peak E-wave velocity, peak E/A ratio of transmitral flow, peak D-wave velocity, and peak D/S ratio of pulmonary venous flow. The Tei index in group 3 was greater than that in group 1, and patients in group 3 showed smaller values in peak transmitral A wave, peak S, and A velocity of pulmonary venous flow, isovolumic relaxation time, and ejection time compared with the other groups. The E-wave deceleration time in group 3 was significantly smaller than in group 1, but the isovolumic contraction time was not different among the 3 groups.
The cyclic variation of integrated backscatter (CV-IB) in the interventricular septum and LV posterior wall are shown in Table 2. In contrast to the traditional echo-Doppler features (Table 1), which could not differentiate between groups 1 and 2, there was a significant and progressive decrease in CV-IB in the posterior wall among all 3 groups, whereas the septal CV-IB differed only between groups 1 and 3.
Prognostic Value of Echo/Doppler Indexes and CV-IB
Of the 209 patients, 41 died during a mean follow-up period of 307±156 (162 to 547) days. Thirty-two of 41 deaths were cardiac related (either sudden or caused by CHF). Group 3 showed significantly poorer prognosis than that of the other 2 groups (Figure 3).
Univariate analysis showed that CV-IB in the LV posterior wall, E/A ratio of transmitral flow, left atrial diameter, deceleration time of transmitral E wave, D/S ratio of pulmonary venous flow, and LV fractional shortening were predictors of all death and cardiac death (Table 3 and Figure 4). Univariate analyses that used published cutoff values showed that E/A ratio of transmitral flow (2.1) was only the predictor of all-cause and cardiac death, whereas fractional shortening (0.30), mean LV thickness (15 mm), and Tei index (0.77) were not predictors.2,7,8⇓⇓ Multivariate analysis showed that the CV-IB at the LV posterior wall and the deceleration time of transmitral E wave were independent predictors of overall death, and the CV-IB at the LV posterior wall was the only independent predictor of cardiac death (Table 3).
The receiver-operating characteristic analysis (Figure 4) indicated that 4.7 dB was the best cutoff value to predict patient outcome within 5 months (the minimum follow-up period of this study). This cutoff value (4.7 dB) was a stronger predictor of adverse outcome than the median value (5.35 dB, Figure 4).
Because CHF itself was a strong variable (discontinuous) to predict adverse outcome, a subgroup analysis was made in group 3 (n=61). The median value of CV-IB (4.3 dB) still significantly distinguished patients with all-cause and cardiac death (Figure 5). The receiver-operating analysis indicated that 4.0 dB was the best cutoff value, predicting patient outcome within 5 months in group 3 (Figure 5), and this value (4.0 dB) was a stronger predictor of all-cause and cardiac death than median value (4.3 dB, Figure 5).
In a previous 2-dimensional echocardiographic study of patients, it was suggested that an increased mean LV wall thickness and decreased fractional shortening were the most important echocardiographic predictors of poor outcome in AL amyloidosis.2 In the present, larger study, univariate analysis confirmed that a decreased fractional shortening was a predictor of both overall and cardiac death, although the mean LV thickness was not helpful to predict cardiac death. Previously published cutoff values of fractional shortening (0.30) and mean LV thickness (15 mm) were not predictors of death in our series of patients.2
Left ventricular inflow filling variables have been described as independent predictors of cardiac death in cardiac amyloidosis, and, consistent with this, we found that univariate analysis demonstrated that the E/A ratio and deceleration time of transmitral flow were significant predictor of cardiac death7 (Table 3).
The Tei index, a Doppler-derived index of combined systolic and diastolic myocardial performance, has been retrospectively studied in 45 patients with amyloidosis and was reported as being the only continuous independent variable to predict survival.8 Contrary to this finding, this index was not a predictor of outcome in our prospective study. Isovolumic contraction and relaxation times (components of the Tei index) were significantly prolonged in the previous study,8 whereas we found no significant difference in isovolumic contraction time among the 3 groups (although the isovolumic relaxation time in group 3 was significantly shorter than the other 2 groups). Our results are consistent with the other investigators who found that advanced cardiac amyloidosis showed a normal isovolumic relaxation time.28 We therefore conclude that the Tei index has no prognostic value in predicting clinical outcome of cardiac amyloidosis.29
Advantages of the Present Study
We prospectively examined the largest published consecutive number of patients with AL amyloidosis2,6–8⇓⇓⇓ and confirmed that previously published values of transmitral Doppler E/A ratio and E-wave deceleration time are strong predictors of adverse outcome, 7 whereas mean LV thickness and fractional shortening were not predictors of death.2 We also demonstrated that the CV-IB at the LV posterior wall is the most sensitive marker among multiple ultrasonic variables for predicting poor outcome. The CV-IB in myocardium is believed to be independent of heart rate, preload, and afterload and to follow changes in intrinsic myocardial contractility.30 This may partly explain why it is a more sensitive prognostic indicator than load-dependent LV inflow filling variables.31
Our study confirmed that the existence of CHF itself is a strong predictor of adverse outcome (Figure 3).7,8⇓ However, even when we restricted the analysis to the subgroup of patients with CHF (Figure 5), CV-IB still had prognostic significance.
Although CV-IB is affected by age,13 there was no significant difference in the mean ages among groups in our study. Cardiac involvement was defined as a mean value of LV thickness >12 mm in patients with AL amyloid confirmed by a biopsy from any site. Thus, many patients were diagnosed as having cardiac amyloidosis without endomyocardial biopsy. However, on the basis of autopsy studies, it generally is accepted that the finding of LV thickening by echocardiography in the absence of diseases associated with LV hypertrophy is highly specific for the finding of cardiac amyloid deposition at biopsy or autopsy.32
Cyclic variation of integrated backscatter at the LV posterior wall is a useful and powerful predictor of clinical outcome in patients with AL cardiac amyloidosis, compared with the traditional Doppler flow indexes. This finding may help to select patients for treatment, particularly when high-dose intravenous melphalan with autologous stem cell transplantation is considered.7,33⇓
This study was supported by a research grant from the Japan Health Sciences Foundation (Dr Koyama), the Sue Sellors Finley Amyloid Fund (Drs Koyama and Falk), and the General Clinical Research Center M01RR00533. The authors gratefully acknowledge the advice and helpful comments of Professor J.G. Miller.
- ↵Siqueira-Filho AG, Cunha CLP, Tajik AJ, et al. M-mode and two-dimensional echocardiographic features in cardiac amyloidosis. Circulation. 1981; 63: 188–196.
- ↵Comenzo RL. Vosburgh E. Falk RH et al. Dose-intensive melphalan with blood stem-cell support for the treatment of AL (amyloid light-chain) amyloidosis: survival and responses in 25 patients. Blood. 1998; 91: 3662–3670.
- ↵Klein AL, Hatle LK, Taliercio CP, et al. Prognostic significance of Doppler measures of diastolic function in cardiac amyloidosis. Circulation. 1991; 83: 808–816.
- ↵O’Brien PD, O’Brien WD, Rhyne TL, et al. Relation of ultrasonic backscatter and acoustic propagation properties to myofibrillar length and myocardial thickness. Circulation. 1995; 91: 171–175.
- ↵Vered Z, Barzilai B, Mohr GA, et al. Quantitative ultrasonic tissue characterization with real-time integrated backscatter imaging in normal human subjects and in patients with dilated cardiomyopathy. Circulation. 1987; 76: 1067–1073.
- ↵Wong AK, Verdonk ED, Hoffmeister BK, et al. Detection of unique transmural architecture of human idiopathic cardiomyopathy by ultrasonic tissue characterization. Circulation. 1992; 86: 1108–1115.
- ↵Masuyama T, St Goar FG, Tye TL, et al. Ultrasonic tissue characterization of human hypertrophied hearts in vivo with cardiac cyclic-dependent variation in integrated backscatter. Circulation. 1989; 80: 925–934.
- ↵Sagar KB, Rhyne TL, Warltier DC, et al. Intramyocardial variability in integrated backscatter: effects of coronary occlusion and reperfusion. Circulation. 1987; 75: 436–442.
- ↵Vitale DF, Bonow RO, Gerundo G, et al. Alterations in ultrasonic backscatter during exercise-induced myocardial ischemia in humans. Circulation. 1995; 92: 1452–1457.
- ↵Colonna P, Montisci R, Galiuto L, et al. Effects of acute myocardial ischemia on intramyocardial contraction heterogeneity: A study performed with ultrasound integrated backscatter during transesophageal atrial pacing. Circulation. 1999; 100: 1770–1776.
- ↵Masuyama T, Valantine HA, Gibbons R, et al. Serial measurement of integrated ultrasonic backscatter in human cardiac allografts for the recognition of acute rejection. Circulation. 1990; 81: 829–839.
- ↵Angermann CE, Nassau K, Stempfle H-U, et al. Recognition of acute cardiac allograft rejection from serial integrated backscatter analyses in human orthotropic heart transplant recipients: comparison with conventional echocardiography. Circulation. 1997; 95: 140–150.
- ↵Cox DR. Regression models and life-tables (with discussion). J R Stat Soc B. 1972; 34: 187–220.
- ↵Sagar KB, Pelc LE, Rhyne TL, et al. Influence of heart rate, preload, afterload, and inotropic state on myocardial ultrasonic backscatter. Circulation. 1988; 77: 478–483.
- ↵Arbustini E, Verga L, Concardi M, et al. Electron and immuno-electron microscopy of abdominal fat identifies and characterizes amyloid fibrils in suspected cardiac amyloidosis. Amyloid. In press.