The Surgical Treatment for Ischemic Heart Failure (STICH) randomized trial was designed to identify an optimal management strategy for patients with ischemic cardiomyopathy. Baseline echocardiographic examinations were required for all patients. The primary aim of this report is to describe the baseline STICH Echocardiography Core Laboratory data. The secondary aim is to provide recommendations regarding how echocardiography should be used in clinical practice and research on the basis of the experience gained from echocardiography in STICH.
Between September 2002 and January 2006, 2,136 patients with ejection fractions (EFs) ≤ 35% and coronary artery disease amenable to coronary artery bypass grafting were enrolled. Echocardiography was acquired by 122 clinical enrolling sites, and measurements were performed by the Echocardiography Core Laboratory after a certification process for all clinical sites.
Echocardiography was available for analysis in 2,006 patients (93.9%); 1,734 (86.4%) were men, and the mean age was 60.9 ± 9.5 years. The mean left ventricular end-systolic volume index, measureable in 72.8%, was 84.0 ± 30.9 mL/m 2 , and the mean EF was 28.9 ± 8.3%, with 18.5% of patients having EFs > 35%. Single-plane measurements of left ventricular and left atrial volumes were similar to their volumes by biplane measurement ( r = 0.97 and r = 0.92, respectively). Mitral regurgitation severity by visual assessment was associated with a wide range of effective regurgitant orifice area, while effective regurgitant orifice area ≥ 0.2 cm 2 indicated at least moderate mitral regurgitation by visual assessment. Deceleration time of mitral inflow velocity had a weak correlation with EF ( r = 0.25) but was inversely related to estimated pulmonary artery systolic pressure ( r = −0.49).
In STICH patients with ischemic cardiomyopathy, Echocardiography Core Laboratory analysis of baseline echocardiographic findings demonstrated a wide spectrum of left ventricular shape, function, and hemodynamics, as well as the feasibility and limitations of obtaining essential echocardiographic measurements. It is critical that the use of echocardiographic parameters in clinical practice and research balance the strengths and weaknesses of the technique.
The Surgical Treatment for Ischemic Heart Failure (STICH) trial, supported by the National Heart, Lung, and Blood Institute of the National Institutes of Health (Bethesda, MD), is an international randomized trial designed to test two specific hypotheses in patients with left ventricular (LV) dysfunction and coronary artery disease. Hypothesis 1 (H 1 ) tested whether coronary artery bypass grafting (CABG) would result in improved long-term survival compared with intensive medical therapy alone. Hypothesis 2 (H 2 ) tested whether combining a surgical ventricular reconstruction (SVR) procedure with CABG would improve survival free from cardiac hospitalization in comparison with CABG alone in patients with reduced LV ejection fraction (EF) and dysfunctional anterior segments. The STICH protocol required that all patients undergo baseline, 4-month follow-up, and 2-year follow-up echocardiography and that measurements be performed by an echocardiography core laboratory. The primary outcomes data in H 2 patients (499 assigned to CABG vs 501 to CABG plus SVR) showed no overall benefit from the addition of SVR to CABG despite a more significant reduction in LV volumes and an increase in EF with SVR. The outcomes in H 1 patients (602 assigned to medical therapy vs 610 to CABG) showed no statistically significant benefit for CABG in the primary outcome of all-cause mortality. However, patients assigned to CABG compared with those assigned to medical therapy alone had lower rates of death from cardiovascular causes and of death from any cause or hospitalization for cardiovascular causes. Knowledge of LV structure, function (volumes, EF, and diastolic function), and hemodynamics in STICH patients would help us better understand the outcomes of tested treatment strategies in future subgroup analyses. Because the STICH trial was conducted at 122 clinical sites in 26 countries, we made a substantial effort to standardize and maintain the quality of echocardiograms of study patients. Our experience in operating the Echocardiography Core Laboratory in this large clinical trial provided insights into how echocardiography should be used in clinical trials and subsequent implementation of trial data in our clinical practice.
Therefore, the aims of this report are (1) to provide the feasibility of obtaining quality baseline echocardiographic data for the entire STICH trial cohort as well as for H 1 and H 2 separately, (2) to provide pertinent baseline echocardiographic data analyzed by the Echocardiography Core Laboratory in these patients, and (3) to provide recommendations for the use of echocardiography in clinical practice and trials.
Between September 2002 and January 2006, 2,136 patients with EFs ≤ 35% and coronary artery disease amenable to CABG were enrolled in STICH. The qualifying LV EF for enrollment was determined by clinical sites using any of available imaging modalities within 3 months of enrollment. More detailed inclusion and exclusion criteria have been published elsewhere.
Design and Quality Assurance of the Echocardiography Core Laboratory
The Echocardiography Core Laboratory team that analyzed echocardiographic studies for STICH ( Appendix ) consisted of experienced physician echocardiographers (with level 3 training and >5 years of practice) and sonographers (with >3 years of clinical sonography). They were instructed in the goals of the STICH trial and measurement standards. A manual of operation for echocardiography was produced to standardize the sequence, duration, and technique of echocardiographic studies at all clinical sites in 26 countries. Each site was asked to submit one to three echocardiographic studies that demonstrated all of the required components of echocardiography to be certified before patient enrollment began. When the initial echocardiographic studies did not meet minimal criteria for measurements of LV volumes, LV EF, mitral regurgitation (MR) severity, diastolic function, and tricuspid regurgitation velocity, additional studies were requested until the requirements were met.
Once an echocardiographic study arrived at the Echocardiography Core Laboratory, it was transferred directly, if in digital format, to the Echocardiography Core Laboratory’s workstation (Digiview; Digisonics Inc., Houston, TX) for measurement and archiving; if the study was in an analog format, it was digitized first and then transferred. Echocardiographic measurements were performed by Echocardiography Core Laboratory sonographers and were approved by Echocardiography Core Laboratory physician staff members. The qualitative assessments, including MR severity, regional wall motion abnormalities, and grading of diastolic function severity, were performed primarily by a physician member.
All measurements and analyses were performed without knowledge of clinical or other laboratory data. An average of three cardiac cycles was used for sinus rhythm, and an average of three to five cardiac cycles was used for atrial fibrillation. If arrhythmia or poor image quality prevented quantitative measurements, LV EF was estimated visually. Interobserver variability in measuring LV volumes was determined in a subset of patients. The following parameters were measured mostly according to the recommendation of the American Society of Echocardiography.
LV dimensions were measured from the two-dimensional parasternal long-axis view of the left ventricle at the junction of the head of the papillary muscle and chordae. The long-axis dimension of the left ventricle was measured from the apical four-chamber view. The LV sphericity index was calculated as the ratio of the LV short-axis dimension and the maximum long-axis dimension.
LV Volume and LV EF Measurement
LV EF was measured primarily using Simpson’s volumetric method whenever possible. Either a combination of apical four-chamber and two-chamber views (preferentially) or a combination of apical four-chamber and long-axis views was used. If two apical views were not available, only one apical view was used for Simpson’s single-plane method. The LV endocardial border was traced contiguously from one side of the mitral annulus to the other, excluding the papillary muscles and trabeculations. LV end-diastolic volume (LVEDV) was measured at the time of QRS, when the LV cavity was largest, and LV end-systolic volume (LVESV) when the left ventricle was smallest. Both were indexed to body surface area. When the definition of the LV endocardial border was not satisfactory from any apical view, LV EF was determined by visual estimation.
LV Regional Wall Motion
LV regional wall motion was analyzed visually using the standard 16-segment model. On the basis of the contractility of each segment, a wall motion score was assigned: 1 = normal, 2 = hypokinesis, 3 = akinesis, and 4 = dyskinesis. The wall motion score index was calculated as an average of the individual wall motion scores of each visualized segment. If more than two segments were not visualized or wall motion abnormalities were global, wall motion analysis was not performed.
Left Atrial (LA) Volume
LA volume was measured using the area-length method ([ A 1 × A 2 ]/length) using the apical four-chamber view and the apical long-axis or two-chamber view. A 1 is LA area from the apical four-chamber view, A 2 is LA area from the apical two-chamber or long-axis view, and length is LA long-axis dimension of the line drawn from the center of the mitral annulus to the posterior wall of the left atrium from an apical view. LA volume was also calculated from the apical four-chamber view only using the following modified area-length method: ( A 1 × A 1 )/length.
Stroke Volume (SV) and Cardiac Output
SV was calculated using two methods: one from the LV outflow tract (LVOT) using the formula SV = LVOT area × LVOT time-velocity integral, and another from the LV volumes measured by the single-plane or biplane Simpson’s method as SV = LVEDV − LVESV. Cardiac output was calculated as the product of SV and heart rate.
The severity of MR was primarily determined by the physician’s visual assessment of the width, depth, and area of the MR jet. In addition, effective regurgitant orifice area (EROA) was determined using the proximal isovelocity surface area method, as previously described, whenever possible.
Pulmonary Artery Systolic Pressure (PASP)
PASP was estimated from the peak tricuspid regurgitation velocity, obtained by continuous-wave Doppler echocardiography, and estimated right atrial pressure, as previously described.
Determination of Diastolic Function
Mitral inflow velocities were recorded by placing a small sample volume at the tip of the mitral valve during diastole. Early diastolic velocity (E), late diastolic velocity with atrial contraction (A), and deceleration time (DT) of E velocity were measured from the inflow velocity recording. A velocities were not available for patients with atrial fibrillation. Mitral annular velocities were measured using Doppler tissue imaging by placing a sample volume over the medial and/or lateral annulus to determine early diastolic velocity (e′) and late diastolic velocity with atrial contraction (a′). In patients with sinus rhythm, diastolic function was graded as follows: grade 1 = relaxation abnormality (no elevation of filling pressure), E/A < 0.8, and DT > 240 msec; grade 2 = pseudonormalized filling (relaxation abnormality and mild elevation of filling pressure), E/A of 0.8 to 1.5, and DT of 160 to 240 msec; and grade 3 = restrictive filling (relaxation abnormality and marked elevation of filling pressure), E/A > 1.5 and DT < 160 msec. Diastolic function was regarded normal if medial or lateral e′ velocity was >8 or >10 cm/sec, respectively. If there was discrepancy among diastolic parameters in grading, function was classified as “indeterminate.” Diastolic function was not graded in patients with atrial fibrillation.
The LV Tei index, or LV index of myocardial performance, was derived from the mitral inflow and LVOT velocity time-intervals, as previously described.
Continuous variables are summarized as mean ± SD and categorical variables as percentages of the group total. Two-sample t tests were used to compare echocardiographic continuous variables, and χ 2 tests were used to compare categorical data. Because there was some overlap between the H 1 and H 2 patient groups and the statistical tests we used require independent samples, 74 patients included in both groups were excluded from the statistical analysis. Pearson’s correlation coefficients are presented when describing relationships among echocardiographic parameters.
Of the total of 2,136 patients, baseline echocardiograms were available for analysis in 2,006 patients (93.9%). The mean age was 60.9 ± 9.5 years, and 86.4% were men. Atrial fibrillation was present in 85 patients (5%). Table 1 shows echocardiographic variables and their values measured by the Echocardiography Core Laboratory as well as the number of patients in whom each variable could be measured.
|Measurement||Overall ( n = 2,006)||H 1 ( n = 1,144)||H 2 ( n = 936)||P ∗|
|Number of patients||Value||Number of patients||Value||Number of patients||Value|
|LVEDD (cm)||1,432||6.3 ± 0.8||804||6.3 ± 0.8||680||6.4 ± 0.8||.027|
|LVESD (cm)||1,352||5.4 ± 0.9||767||5.3 ± 0.9||635||5.3 ± 0.9||.842|
|LV long-axis dimension (cm)||1,506||9.2 ± 1.0||846||9.2 ± 1.0||721||9.3 ± 1.0||.007|
|Sphericity index (diastole)||1,154||0.69 ± 0.09||648||0.69 ± 0.09||551||0.68 ± 0.09||.264|
|LVEDV (mL)||1,460||222.4 ± 68.8||806||220.4 ± 67.3||710||225.0 ± 69.4||.167|
|LVESV (mL)||1,460||160.7 ± 60.4||806||160.2 ± 60.1||710||161.0 ± 60.3||.766|
|LVEDV index (mL/m 2 )||1,460||116.3 ± 34.6||806||115.6 ± 33.8||710||117.0 ± 35.5||.417|
|LVESV index (mL/m 2 )||1,460||84.0 ± 30.9||806||84.1 ± 30.7||710||83.8 ± 31.2||.854|
|LV EF (%)||1,460||28.9 ± 8.3||806||28.5 ± 8.5||710||29.5 ± 8.1||.016|
|LA volume index (mL/m 2 )||1,237||41.9 ± 15.2||696||41.7 ± 14.7||596||42.1 ± 15.6||.500|
|Global hypokinesis||1,985||227 (11%)||1,130||149 (13%)||929||88 (9%)||.006|
|Wall motion score index||1,758||2.2 ± 0.3||981||2.3 ± 0.3||841||2.2 ± 0.3||.020|
|DT (msec)||1,492||186.2 ± 56.2||842||189± 58.2||708||183.3 ± 53.4||.028|
|MV E velocity (m/sec)||1,635||0.73 ± 0.25||920||0.72 ± 0.26||778||0.73 ± 0.25||.198|
|MV A velocity (m/sec)||1,535||0.67 ± 0.24||860||0.67 ± 0.24||736||0.68 ± 0.24||.562|
|E/A ratio||1,532||1.3 ± 1.1||859||1.4 ± 1.3||734||1.3 ± 0.9||.473|
|e′ or Ea septal velocity (m/sec)||1,002||0.05 ± 0.02||592||0.04 ± 0.02||450||0.05 ± 0.02||<.001|
|e′ or Ea lateral velocity (m/sec)||971||0.06 ± 0.03||573||0.06 ± 0.03||436||0.06 ± 0.03||.934|
|Septal E/Ea or E/e′ ratio||920||17.6 ± 9.6||544||18.1 ± 9.7||413||16.8 ± 9.3||.041|
|Normal||5 (0.3%)||3 (0.3%)||2 (0.2%)|
|Grade 1||604 (30%)||362 (32%)||268 (29%)|
|Grade 2||590 (30%)||304 (27%)||311 (33%)|
|Grade 3||433 (22%)||253 (22%)||194 (21%)|
|Grade 4||2 (0.1%)||1 (0.1%)||1 (0.1%)|
|Indeterminate||358 (18%)||209 (18%)||158 (17%)|
|0||514 (26%)||316 (28%)||227 (24%)|
|1||871 (44%)||465 (41%)||437 (47%)|
|2||306 (15%) ∗||174 (15%) ∗||142 (15%) ∗|
|3||110 (6%) ∗||58 (5%) ∗||52 (5%) ∗|
|4||51 (3%) ∗||30 (3%) ∗||24 (3%) ∗|
|Indeterminate||138 (7%) ∗||95 (8%) ∗||44 (5%) ∗|
|TR velocity (m/sec)||596||2.9 ± 0.5||342||2.9 ± 0.5||274||2.8 ± 0.5||.186|
|PASP (mm Hg)||430||42.8 ± 15.5||241||43.4 ± 15.8||205||41.7 ± 15.0||.241|
LV Dimensions and Sphericity Index
The mean LV end-diastolic dimension and LV long-axis dimension were significantly longer in H 2 patients than the dimensions in H 1 patients ( P = .03 for end-diastolic dimension and P = .007 for long-axis dimension). The sphericity index was similar in H 1 and H 2 patients ( P = .26). There was no significant difference between H 1 and H 2 patients in LV end-systolic dimension ( P = .84).
LV Volumes and EF
In 873 of 2,006 patients (43.5%), reliable delineations of the LV endocardial border were feasible from two apical views; in 587 patients (29.3%), border detection was possible from a single apical view only. Therefore, in 1,460 patients (72.8%), LVEDV and LVESV (and hence EF) were measured using Simpson’s method. When those 1,460 patients with volume measurement were compared with 546 patients without volume measurement, the latter group was older and heavier, with more patients with hypertension or diabetes ( Table 2 ).
|Variable||Have volume measurements |
( n = 1,460)
|Do not have volume measurements |
( n = 546)
|Age (y)||60.6 ± 9.51||61.8 ± 9.54||.0104|
|Women||13.6% (198)||13.6% (74)||.9960|
|Weight (kg)||78.3 ± 14.0||83.4 ± 19.3||<.0001|
|Body mass index (kg/m 2 )||27.0 ± 4.19||28.6 ± 5.56||<.0001|
|Myocardial infarction||82.7% (1208)||78.4% (428)||.0253|
|Stroke||7.3% (106)||5.3% (29)||.1210|
|Hypertension||58.2% (849)||65.0% (355)||.0052|
|Atrial flutter or fibrillation||11.9% (174)||13.2% (72)||.4406|
|Diabetes||34.5% (504)||43.8% (239)||.0001|
|Previous CABG||2.5% (37)||3.7% (20)||.1757|
|Previous PCI||15.0% (219)||16.8% (92)||.3083|
|I||10.4% (152)||9.7% (53)||.6469|
|II||46.2% (675)||49.5% (270)|
|III||39.2% (573)||37.0% (202)|
|IV||4.1% (60)||3.8% (21)|
|Visual EF ∗||0.28 ± 0.08||0.29 ± 0.08||.1415|
|0||33.6% (490)||41.9% (229)||.0083|
|1||46.7% (682)||42.7% (233)|
|2||15.8% (230)||12.5% (68)|
|3||3.5% (51)||2.4% (13)|
|4||0.5% (7)||0.5% (3)|