Association of Outcome with Left Ventricular Parameters Measured by Two-Dimensional and Three-Dimensional Echocardiography in Patients at High Cardiovascular Risk




Background


Left ventricular (LV) ejection fraction (EF) measured by two-dimensional echocardiographic (2DE) imaging is an important correlate of survival. Real-time three-dimensional echocardiographic (3DE) imaging has addressed some of the limitations of 2DE imaging. The aim of this study was to determine whether 3DE imaging is more predictive of outcomes than 2DE imaging.


Methods


A total of 529 patients undergoing LV assessment with 2DE and 3DE imaging in 2003 and 2004 at a tertiary referral cardiac center were studied. Patients had a high frequency of cardiovascular risk factors. Images were gathered over four cardiac cycles using a matrix-array transducer, with measurements performed offline. Follow-up (all-cause mortality or cardiac hospitalization) was obtained over 6.6 ± 3.4 years in 455 of 486 patients with images suitable for measurement (94%).


Results


There were 194 events (43%), including 75 deaths (16.4%). Larger LV volumes and lower EF were associated with worse outcomes independent of age, heart failure, or end-stage renal disease. In stepwise Cox regression analyses, the associations of cardiac hospitalization and survival with clinical variables (age, chronic kidney disease, and heart failure) were augmented by 3DE EF and end-systolic volume more than by 2DE parameters. The incremental model χ 2 value with 3DE EF was 14.67 ( P < .001), compared with 9.72 ( P = .002) for 2DE EF. Similarly, in Cox regression analyses of mortality, the effects of clinical variables (age, advanced renal disease, and heart failure) were augmented more by 3DE EF (incremental χ 2 = 14.04, P < .0001) than 2DE EF (incremental χ 2 = 5.13, P = .024).


Conclusions


In this outcome study, 3DE EF and volumes showed stronger associations with outcomes than those derived from 2DE imaging.


Left ventricular (LV) enlargement is an important step in the progression of heart failure (HF) and a major determinant of prognosis. Indeed, some studies suggest that the response of LV volumes to treatment may be a predictor of outcomes.


Two-dimensional echocardiographic (2DE) imaging has played a pivotal role in defining LV volume responses in major trials of HF treatments, including antagonists of the angiotensin and sympathetic nervous systems, exercise training, and cardiac resynchronization therapy. In these and other studies, LV ejection fraction (EF) calculated using the modified Simpson’s rule was used to predict clinical outcomes. However, the application of this information to individuals rather than populations poses significant limitations, in part because of the underestimation of LV volumes with 2DE imaging, especially if the left ventricle does not conform to a geometric shape. There are particular problems in the sequential measurement of these parameters, not only because of load alterations but also because of changes in LV shape that may not be adequately accounted for with measurements based on geometric assumptions and the influence of different imaging planes from one scan to the next.


Three-dimensional echocardiographic (3DE) imaging has overcome these limitations of 2DE imaging, and a number of investigators have shown that 3DE imaging has less test-retest variation and better reproducibility and accuracy in LV volume estimation compared with 2DE imaging. The recent publication of defined normal ranges for 3DE imaging should now improve clinical uptake of the technique. However, the effects of these differences in LV measurements between 2DE and 3DE imaging on the prediction of outcomes have only limited validation. Therefore, in this study, we sought to define whether 3DE imaging is more predictive of outcome than 2DE imaging.


Methods


Study Design


In 2003 and 2004, 8,791 patients were referred to our echocardiography laboratory (Princess Alexandra Hospital, a tertiary referral center in Brisbane, Australia). The majority attended for evaluation of LV function (including patients with hypertension, those commencing chemotherapy, those undergoing stress echocardiography, and those with syncope and dyspnea). Other referrals included patients undergoing evaluation of transient ischemia, those who had had myocardial infarctions, those with valvular disease, and those undergoing evaluation for the source of embolism ( Figure 1 ).




Figure 1


Patient indications for echocardiography. MI, Myocardial infarction; TIA, transient ischemic attack.


Two of our seven echocardiographic machines had 3DE capability. Among the echocardiographic studies performed, patients were excluded if there were insufficient 2DE data and if the studies were not performed by expert sonographers. Therefore, 529 consecutive and unselected patients were studied. In this longitudinal study, we compared their prospectively obtained LV volumes and EFs and patients’ LV measurements using 2DE and 3DE imaging, with follow-up over 6.6 ± 3.4 years. This follow-up was approved by the ethics committee of the Princess Alexandra Hospital, and all patients consented to undergo the studies.


Clinical Findings


Clinical information at the time of the baseline studies included patient risk factors such as hypertension (blood pressure > 140/90 mm Hg), hypercholesterolemia (total cholesterol > 5.5 mmol/L), diabetes (fasting glucose ≥ 7 mmol/L), or treatment for these conditions; chronic kidney disease (CKD); diagnosed HF, either diastolic or systolic; previous treatment or operations for vascular disease; previous revascularization (percutaneous coronary intervention or coronary bypass grafting); and LV impairment, including previous myocardial infarction or cardiomyopathy.


Two-Dimensional Echocardiography


An experienced sonographer performed routine 2DE studies in accordance with a routine acquisition protocol using standard equipment (Sonos 7500 or iE33; Philips Medical Systems, Andover, MA) with a transthoracic 3-MHz phased-array transducer. The 2DE examinations included parasternal long-axis and short-axis views of the left ventricle and apical four-chamber and two-chamber views. Images were obtained with the patient in the left lateral decubitus position. LV volumes were obtained from the apical views from 2DE images and were calculated using the biplane Simpson’s rule. Measurements of LV end-diastolic volume (EDV), end-systolic volume (ESV), and EF were obtained in accordance with American Society of Echocardiography guidelines, with EDV measurements at the frame after mitral valve closure and ESV measured on the image with the smallest LV cavity. Volumes using biplane Simpson’s rule were obtained from the apical four-chamber and two-chamber views, with the papillary muscles being included in the LV cavity.


Three-Dimensional Echocardiography


Three-dimensional echocardiographic images were obtained by an expert sonographer from an apical four-chamber window with the patient in the same position as for 2DE. Images were gathered over four cardiac cycles using a standard commercial system (Sonos 7500 or iE33) with a matrix-array transducer.


Measurements were performed offline (QLAB 3DQ Advanced version 6; Philips Medical Systems). The 3DE volume data were displayed in three different cross-sections, which included conventional 2DE four-chamber and two-chamber views and a short-axis view. Frames for EDV and ESV measurements were identified by the same method as for 2DE imaging, and measurements were obtained with semiautomated LV border detection on the basis of fiducial marks on the annulus and apex. Manual adjustments to border tracing at EDV and ESV were obtained when required. The traced contours were then used to define an endocardial shell in three dimensions, and the enclosed three-dimensional volume was calculated. Patients in atrial fibrillation had measurements made from three cardiac cycles and then averaged. Both 2DE and 3DE volumes were indexed to body surface area (calculated using the method of DuBois and DuBois ).


Follow-Up


Follow-up data were collected from Queensland Hospitals Admitted Patient Data Collection for hospital admissions and the State Death Registry. We used a composite end point of all-cause mortality and cardiac hospitalization.


Reproducibility


Two experienced readers (T.S., C.J.) remeasured 2DE and 3DE LV volumes and EFs in 21 randomly selected patients. Each reader performed two separate readings in a random order on separate days, blinded to previous results. Readers were allowed to pick the cardiac cycle that in their opinion had the best image quality, as would occur in daily clinical practice.


Statistical Analysis


Continuous variables are expressed as mean ± SD or as median (interquartile range); categorical variables are expressed as percentages. Student’s t tests were used to compare differences between two groups for continuous variables. Chi-square tests were used to determine significant differences between two groups of categorical variables. In a random subgroup of 21 patients, the same set of three-dimensional and two-dimensional images measured by two separate observers was studied for interobserver variability. The same group was tested for intraobserver variability, with measurements repeated on the same data set by the same sonographer. Kaplan-Meier curves were constructed to illustrate the prognostic information provided by 2DE and 3DE measurements of EF. The log-rank test was used to test for statistical significance between strata.


The purpose of the models used in this study was to identify the association of LV measurements with outcomes, independent of other contributors. Not surprisingly, there was a high correlation between the 2DE and 3DE measurements. Attempts to add 2DE and 3DE variables in the same model led to significant instability (>50% increase in the standard error). Therefore, we undertook the approach of sequentially adding the 2DE and 3DE variables into a baseline clinical model. The independence and incremental value of each measure was assessed by comparing model χ 2 statistics. Model discrimination was further assessed using Harrell’s C-statistic. The first step used only the clinical variables available within this data set (excluding the echocardiography-measurement variables) in a forward, stepwise Cox proportional-hazards model, censoring data at the first event. The top three selected clinical variables with significance levels up to .25 were allowed into the multivariate baseline clinical model (keeping in mind that the baseline clinical model excluded echocardiographic measurements). Next, 2DE measurements were added sequentially to the baseline clinical model, and the process was repeated in a separate model of the baseline model and the 3DE variables, resulting in pairs of models for EF, ESV, and EDV.


Nonlinear associations were sought, and the proportional-hazards assumption was checked using the time-varying covariates method. Because renal failure and myocardial infarction may have a synergistic effect with LV dysfunction and outcomes, a multiplicative interaction term was added, and P values for interactions were sought. Results are presented as hazard ratios (HRs) with 95% confidence intervals (CIs). Analysis was carried out using SAS version 9.2 (SAS Institute Inc, Cary, NC). Significance was measured as P < .05 unless otherwise stated.




Results


Patient Characteristics


Of the 529 patients who underwent 2DE and 3DE imaging for LV evaluation in 2003 and 2004, images were unsuitable for measurement in 43 patients (17 with poor 2DE and 3DE image quality and 26 with just poor 3DE image quality, for a total of 8%). A further 31 patients (6.4%) had complete baseline data but were lost to follow-up. Of individuals who were lost to follow-up, 3DE volumes were significantly larger than 2DE volumes for both EDV (145.4 ± 55.2 vs 120.4 ± 38.3 mL, P < .01) and ESV (68.4 ± 39.8 vs 57.3 ± 28.2 mL, P = .01), with EF nonsignificantly lower on 3DE imaging (51.3 ± 12.0% vs 53.6 ± 11.4%, P = .16). Patients who were lost to follow-up had a lower incidence of diabetes and CKD but had significantly higher volumes and lower EFs with both 3DE and 2DE imaging, probably because of the higher incidence of coronary artery disease ( Table 1 ). In the remaining patients, there was a high prevalence of risk factors as well as valvular disease, including two patients who had already undergone aortic valve replacement, 14 with mitral valve replacement, and three for mitral valve repair. Nine patients had moderate to severe mitral regurgitation, two with mild aortic stenosis, and there were varying degrees of pulmonary hypertension in eight patients (39–66 mm Hg). Diastolic function was assessed in 412 of the 455 patients, with exclusions due to valvular pathology and insufficient or poor Doppler tracings. Only 21% had normal function; the majority (53%) had grade 1 diastolic dysfunction, with 20% having grade 2 dysfunction and 6% having grade 3 dysfunction.



Table 1

Patient characteristics




























































































































Patients with follow-up
(n = 455)
Patients lost to follow-up
(n = 31)
P
Age (y) 66.0 ± 11.4 66.4 ± 13.0 .874
Men 315 (69.2%) 23 (74.2%) .561
Height (cm) 170 ± 9 170 ± 7 .935
Weight (kg) 85 ± 17 88 ± 21 .460
Hypertension 280 (62%) 16 (52%) .273
Hypercholesterolemia 291 (64%) 21 (68%) .671
Diabetes 268 (59%) 5 (16%) <.001
Vascular disease 70 (15%) 0 .018
Coronary artery disease 190 (42%) 21 (78%) .005
PCI/angioplasty 48 (11%) 0 .057
CABG 68 (15%) 3 (10%) .422
HF 74 (16%) 4 (13%) .342
CKD 112 (25%) 1 (3%) .006
2DE EF (%) 57.7 ± 11.2 53.6 ± 11.4 .051
3DE EF (%) 56.8 ± 10.9 51.3 ± 12.0 .007
2DE EDV (mL) 96.7 ± 37.2 120.4 ± 38.3 .001
2DE EDV index (mL/m 2 ) 48.7 ± 18.4 59.6 ± 17.9
3DE EDV 119.3 ± 47.4 145.4 ± 55.2 .003
3DE EDV index (mL/m 2 ) 59.9 ± 23.5 71.5 ± 23.9
2DE ESV (mL) 43.5 ± 26.8 57.3 ± 28.2 .006
2DE ESV index (mL/m 2 ) 21.8 ± 13.6 28.5 ± 13.9
3DE ESV 54.0 ± 33.4 68.5 ± 39.8 .021
3DE ESV index (mL/m 2 ) 27.2 ± 17.2 33.9 ± 18.6

CABG , Coronary artery bypass graft; PCI , percutaneous coronary intervention.

Data are expressed as mean ± SD or as number (percentage).


Predictors of Events


Of 455 patients with follow-up (mean, 6.6 ± 3.4 years), 194 (43%) had events; 75 (16.4%) died (the majority with primarily cardiac causes of death and others as a result of diseases such as end-stage renal disease and sepsis), and 119 (26.2%) had cardiac admissions. Of these cardiac admissions, 55 (46.2% total; 23 [19.3%] with acute coronary syndromes, 13 [10.9%] with arrhythmias, and 19 [16%] with HF) were unplanned, and 64 (53.8% total; 27 [22.7%] with coronary angiography and or percutaneous coronary intervention and 37 [31.1%] with cardiac surgery) were planned.


Among the patients with events, 2DE EF (53.3 ± 11.8%) and 3DE EF (52.5 ± 11.9%) were similar ( P = .14), but EDV was less by 2DE than 3DE imaging (107.8 ± 42.1 vs 133.3 ± 53.3 mL, P < .001), as was ESV (52.7 ± 30.4 vs 65.9 ± 38.7 mL, P < .001). In addition, although in the overall population, there was a nonsignificant difference between mean 2DE EF and 3DE EF, there were greater differences between 2DE EF and 3DE EF in those who had events (−0.8 ± 8.3% vs −1.0 ± 7.9%) or died (−0.7 ± 8.2% vs −1.8 ± 7.9%) compared with those who did not.


The clinical features associated with events are listed in Table 2 . Clinical predictors of outcome were CKD (HR, 1.83; 95% CI, 1.35–2.47), HF (HR, 4.14; 95% CI, 3.03–5.65), and age (HR, 1.02; 95% CI, 1.01–1.04). Clinical predictors of coronary artery disease, LV impairment, and the presence of regional wall motion abnormalities were not used in the model, because of their close correlations with HF. In separate Cox regression models combining clinical data with 3DE and 2DE imaging, 3DE EF (HR for each 1% increment, 0.95; 95% CI, 0.95–0.97) and 2DE EF (HR, 0.95; 95% CI, 0.95–0.97), 3DE EDV (HR for each 1-mL increment, 1.02; 95% CI, 1.01–1.02) and 2DE EDV (HR, 1.02; 95% CI, 1.01–1.03), 3DE ESV (HR for each 1-mL increment, 1.03; 95% CI, 1.02–1.04) and 2DE ESV (HR, 1.03; 95% CI, 1.02–1.04) were all found to be independent predictors of outcomes.



Table 2

Clinical features associated with total events































































































































Variable Univariate Multivariate (χ 2 = 129.7)
HR (95% CI) P HR (95% CI) P
Demographics
Age 1.03 (1.01–1.04) <.0001 1.02 (1.01–1.04) <.01
Gender 0.61 (0.43–0.86) <.01
Risk factors
Hypertension 1.62 (1.18–2.21) .03
Diabetes mellitus 0.62 (0.46–0.82) <.01
Hypercholesterolemia 1.52 (1.11–2.08) .01
CKD 1.83 (1.35–2.47) <.0001 1.71 (1.24–2.36) <.01
Obstructive sleep apnea 0.71 (0.38–1.30) .26
Clinical
HF 4.14 (3.03–5.65) <.0001 3.35 (2.34–4.8) <.0001
Peripheral vascular disease 1.80 (1.28–2.54) <.01
Known CAD 3.65 (2.70–4.95) <.0001
Myocardial infarction 3.08 (2.31–4.11) <.0001
Past PCI 2.69 (1.86–3.87) <.0001
Past CABG 3.42 (2.48–4.72) <.0001
ICD or pacemaker 3.42 (2.04–5.74) <.0001
RWMA 1.31 (1.2–1.43) <.0001 1.17 (1.05–1.31) .01
Diastolic dysfunction 1.52 (1.26–1.83) <.0001

CABG , Coronary artery bypass graft; CAD , coronary artery disease; ICD , implantable cardioverter-defibrillator; PCI , percutaneous coronary intervention; RWMA , regional wall motion abnormalities.


Incremental Value of LV Parameters to Predict Events


Clinical predictors were placed in separate stepwise regression models with 2DE and 3DE volumes and EF ( Table 3 ). The addition of 2DE EF increased the power of the model (χ 2 = 9.72, P = .002), but 3DE EF was associated with greater power (χ 2 = 14.67, P < .001). Similarly, the incremental value of 2DE ESV index (χ 2 = 6.19, P = .013) was exceeded by 3DE ESV index (χ 2 = 7.06, P = .008).



Table 3

Clinical and 2DE and 3DE features associated with total events


































































































Variable Univariate Multivariate C-statistic (95% CI)
HR (95% CI) P HR (95% CI) P
Clinical
Age 1.03 (1.01–1.04) <.0001 1.02 (1.00–1.03) .01 0.64 (0.60–0.69)
HF 4.14 (3.03–5.65) <.0001 3.60 (2.61–4.07) <.0001
CKD 1.83 (1.35–2.47) <.0001 1.66 (1.23–2.26) .001
2DE imaging
EF 0.96 (0.95–0.97) <.0001 0.98 (0.96–1.00) .04 0.68 (0.64–0.72)
EDV index (10 mL/m 2 ) 1.02 (1.01–1.03) .001 1.01 (1.00–1.02) .001 0.68 (0.64–0.72)
ESV index (10 mL/m 2 ) 1.03 (1.02–1.04) <.0001 1.01 (1.00–1.02) .01 0.67 (0.64–0.71)
3DE imaging
EF 0.96 (0.95–0.97) <.0001 0.97 (0.96–0.99) <.0001 0.68 (0.64–0.72)
EDV index (10 mL/m 2 ) 1.02 (1.01–1.02) <.0001 1.01 (1.00–1.01) .013 0.67 (0.62–0.71)
ESV index (10 mL/m 2 ) 1.02 (1.02–1.03) <.0001 1.01 (1.00–1.02) .005 0.67 (0.63–0.71)


Predictors of Mortality


The clinical features associated with mortality are listed in Table 4 . Clinical predictors of outcomes were similar to those for events: CKD (HR, 3.44; 95% CI, 2.18–5.42), HF (HR, 5.14; 95% CI, 3.1–8.64), and age (HR, 1.05; 95% CI, 1.02–1.08). In separate Cox regression models combining clinical data with 3DE and 2DE imaging, 3DE EF (HR for each 1% increment, 0.97; 95% CI, 0.96–0.99) and 2DE EF (HR, 0.98; 95% CI, 0.96–0.99), 3DE EDV (HR for each 1-mL increment, 1.01; 95% CI, 1.0–1.01) and 2DE EDV (HR, 1.01; 95% CI, 1.0–1.02), 3DE ESV (HR for each 1-mL increment, 1.01; 95% CI, 1.0–1.02), and 2DE ESV (HR, 1.01; 95% CI, 1.0–1.02) were all found to be independent predictors of outcomes.



Table 4

Clinical features associated with mortality































































































































Variable Univariate Multivariate (χ 2 = 65.3)
HR (95% CI) P HR (95% CI) P
Demographics
Age 1.06 (1.03–1.08) <.0001 1.05 (1.02–1.08) <.0001
Gender 0.95 (0.58–1.54) .84
Risk factors
Hypertension 1.35 (0.84–2.18) .22
Diabetes mellitus 0.61 (0.39–0.95) .03
Hypercholesterolemia 0.99 (0.62–1.57) .96
CKD 3.44 (2.18–5.42) <.0001 2.93 (1.81–4.76) <.0001
Obstructive sleep apnea 0.63 (0.23–1.71) .36
Clinical
HF 6.59 (4.12–10.54) <.0001 5.14 (3.1–8.64) <.0001
Peripheral vascular disease 2.82 (1.72–4.62) <.0001
Known CAD 2.34 (1.48–3.69) <.0001
Myocardial infarction 2.97 (1.88–4.68) <.0001
Past PCI 1.19 (0.55–2.6) .66
Past CABG 2.29 (1.3–4.0) <.01
ICD or pacemaker 4.08 (1.86–8.93) <.0001
RWMA 1.29 (1.11–1.5) <.01
Diastolic dysfunction 1.78 (1.34–2.35) <.0001

Only gold members can continue reading. Log In or Register to continue

May 31, 2018 | Posted by in CARDIOLOGY | Comments Off on Association of Outcome with Left Ventricular Parameters Measured by Two-Dimensional and Three-Dimensional Echocardiography in Patients at High Cardiovascular Risk
Premium Wordpress Themes by UFO Themes