The prognostic value of deformation parameters of the systemic right ventricle in adults with D-transposition of the great arteries and prior atrial switch has not been reported.
Sixty-four adults with D-transposition of the great arteries and prior atrial switch (mean age, 29 ± 6 years; 22 women; mean right ventricular [RV] fractional area change, 22.9 ± 7.5%; 31 with pacemakers at baseline) and no histories of heart failure or ventricular tachycardia were prospectively evaluated. Global longitudinal strain (GS), global systolic strain rate (GSRs), and global early diastolic strain rate (GSRe) of the right ventricle were measured using speckle tracking from apical views and compared with standard parameters of RV function (fractional area change, tricuspid annular plane systolic excursion, tissue Doppler velocities, and isovolumic acceleration) for association with and potential prediction of clinical events, defined as incident stage C heart failure or ventricular tachycardia.
Baseline RV GS, GSRs, and GSRe were −12.5 ± 3.0%, −0.59 ± 0.14 sec −1 , and 0.68 ± 0.22 sec −1 , respectively. After a median of 2.4 years (interquartile range, 1.5–4.1 years), 12 patients (19%) presented with clinical events (heart failure in 11 patients, ventricular tachycardia in one patient). In Cox models, RV GS had the strongest association with clinical events (hazard ratio [HR] per 1%, 1.35; 95% confidence interval [CI], 1.14–1.58; P < .001), followed by GSRs (HR per 0.01 sec −1 , 1.06; 95% CI, 1.02–1.11; P = .006), GSRe (HR per −0.01 sec −1 , 1.04; 95% CI, 1.00–1.07; P = .031), and fractional area change (HR per −1%, 1.08; 95% CI, 1.00–1.17; P = .047). Other measures of RV function were not significantly associated with risk for events. In receiver operating characteristic analysis, RV GS ≥ −10% optimally predicted future events (C = 0.83; 95% CI, 0.71–0.91; P < .001).
Reduced longitudinal GS of the systemic right ventricle is associated with increased risk for clinical events among patients with D-transposition of the great arteries and prior atrial switch.
Surgical repair at experienced centers has substantially improved late outcomes in patients with D-transposition of the great arteries (D-TGA), with long-term (>20 years) survival ranging from 75% to 90% in large studies. Despite excellent surgical results, atrial switch (Mustard or Senning) operations are associated with increased late mortality, mainly because of heart failure (HF) and sustained ventricular tachycardia (VT) secondary to failure of the morphologic right ventricle in the systemic (subaortic) position. Although atrial switch has been gradually abandoned in favor of the physiology-restoring (but also more technically demanding) arterial switch in patients with suitable anatomy, a substantial number of adults with D-TGA seen in adult congenital heart disease programs today had atrial switch repair. The main concern in these patients is the function of the systemic right ventricle.
Various imaging modalities have been used to evaluate the systemic right ventricle, including angiography, radionuclide imaging, and magnetic resonance imaging. Nevertheless, the most commonly used modality in practice is echocardiography. Because of complex geometry, however, the assessment of systemic right ventricular (RV) function by echocardiography has remained mostly qualitative. This limits reliable detection of RV dysfunction and evaluation of the effects of medications and/or interventions on RV function. Advances in digital echocardiography allow for a more refined assessment of the right ventricle. Speckle-tracking echocardiography (STE), which is based on tracking of acoustic markers on high-resolution grayscale images, has been validated as a reliable method for myocardial deformation imaging. Considering the complex nature of the systemic right ventricle, the angle-independent nature of this modality may be advantageous in D-TGA. Recently, others and we have demonstrated the applicability of STE for assessment of RV function in patients with D-TGA and in patients with single right ventricles. Growing evidence suggests that deformation-based parameters of left ventricular (LV) function provide incremental prognostic information over standard parameters in populations at risk for or with LV failure. However, the prognostic value of deformation-based parameters of RV function in patients with congenital heart disease in general and with D-TGA in particular is currently unknown.
In this prospective study, our primary objective was to investigate (1) the association of standard and deformation-based parameters of systemic RV function with clinical events related to RV dysfunction, defined as incident stage C (clinically manifest) HF or sustained (lasting >30 sec or leading to syncope) VT, and (2) the value of these parameters for clinical event prediction in adults with D-TGA and prior atrial switch. Our secondary objective was to correlate RV deformation parameters with clinical characteristics.
We prospectively collected clinical and echocardiographic data on consecutive adult patients with D-TGA and prior atrial switch surgery who underwent initial evaluation in the adult congenital heart disease program of the Center for Heart Failure Therapy and Transplantation at Emory University (Atlanta, GA) between July 2005 and December 2010. Inclusion criteria were (1) age >18 years at inception, (2) D-TGA repaired with a Mustard or Senning operation, and (3) clinically stable for ≥1 month before baseline evaluation. We excluded patients with (1) definite diagnosis of HF at baseline or possible diagnosis of HF (i.e., under active investigation for HF symptoms) or (2) histories of documented (by electrocardiography, 24-hour Holter recording, or device interrogation) sustained (>30 sec or leading to syncope) VT or possible VT (syncope of undefined origin at the time of evaluation). The institutional review board approved the protocol.
Overall, we evaluated 97 patients who met the inclusion criteria. Of these, 21 fulfilled the criteria for prevalent stage C HF at baseline (definite HF), and two were under active evaluation for HF (probable HF); one patient had a documented history of VT with an implanted cardioverter-defibrillator for secondary prevention; and one patient had clinically manifest failure of the subpulmonic ventricle secondary to pulmonary hypertension. In addition, eight patients had not returned for follow-up by the time of the current database lock and therefore were excluded from this analysis. A total of 64 patients with D-TGA fulfilled all the criteria and were included in the study.
Definition, Assessment, and Rationale of Clinical Events
The event of interest was defined as the composite of (1) incident HF and (2) incident VT. Information on clinical events was collected during (1) regular follow-up visits, as part of standard clinical care (usually annual visits unless a change in therapy or a recent condition required shorter follow-up), and (2) unplanned office visits for new symptoms. Diagnosis of VT was based on electrocardiographic (rest or stress) documentation, 24-hour Holter recordings, or pacemaker interrogation. Our practice is to interrogate pacemakers at every encounter; these patients undergo also telephone monitoring every 3 months. Monitoring for those without pacemakers is performed for symptoms potentially related to dysrhythmias. All data elements for these visits were captured by physicians of the adult congenital heart disease program in a central database maintained by the Center for Heart Failure Therapy and Transplantation. In addition, we reviewed medical records for hospitalizations or evaluations by noncardiovascular specialists using the electronic medical record system of Emory Healthcare.
Because functional capacity in D-TGA may be compromised as a result of chronotropic incompetence and other problems beyond RV dysfunction, diagnosis of stage C HF was based on (1) the presence of congestive symptoms (pulmonary or peripheral edema), (2) need for HF treatment with diuretics and vasodilators or β-blockers, and (3) symptoms of exercise intolerance improving after diuretic administration. These criteria have been used for stage C HF identification in other cohort studies ; the criterion of symptomatic improvement was added in our study because of competing sources of exercise intolerance in D-TGA. We opted to include only sustained VT in event definition, because supraventricular tachyarrhythmias in patients with D-TGA (1) are multifactorial, albeit more frequent among those with impaired systemic right ventricles ; (2) are not directly associated with mortality ; and (3) become a common characteristic of all patients with D-TGA after a few years of follow-up, questioning thus their prognostic relevance.
Echocardiographic studies were performed using commercially available cardiovascular ultrasound systems (Vivid 7; GE Healthcare, Milwaukee, WI) by a sonographer with extensive experience in congenital heart disease (M.A.P.). In addition to the standard institutional protocol, dedicated apical RV views for postprocessing were acquired in grayscale and in tissue Doppler (color-coded) mode for investigational purposes, ensuring (1) an adequate field of view to image the entire right ventricle throughout the cardiac cycle and (2) a grayscale frame rate >40 frames/sec for optimal tracking of the myocardium during postprocessing. For offline measurements, the images were transferred in uncompressed format on a dedicated workstation; all studies were analyzed with the same version of EchoPAC (EchoPAC ’09; GE Healthcare).
The following standard echocardiographic parameters were measured offline: (1) chamber quantification measurements, including basal and mid RV diameters, end-diastolic and end-systolic RV area, and RV fractional area change ; (2) tricuspid annular plane systolic excursion (TAPSE) ; (3) s′, e′, and a′ velocities at the tricuspid annulus using pulsed-wave tissue Doppler; (4) tissue Doppler myocardial performance index (MPI) of the right ventricle ; and (5) LV ejection fraction. Because of distorted LV morphology, acquisition of apical two-chamber views is inconsistent in patients with D-TGA. Therefore, LV ejection fraction was calculated from a single plane (apical four-chamber view). Because isovolumic acceleration of the RV free wall has been previously reported to reflect contractile RV function in D-TGA after atrial switch, we also measured isovolumic acceleration offline from color-coded tissue Doppler images. A single investigator blinded to clinical data and events performed all measurements.
The principles of STE have been described in detail else where. Briefly, to assess RV longitudinal deformation, we adopted a six-segment RV model as previously described. In the apical RV view, the endocardial border is manually outlined; the myocardium is then tracked by the algorithm and divided into six segments ( Figures 1 and 2 ). The speckle-tracking algorithm detects the QRS onset from the electrocardiographic signal to define the point of zero strain, amenable to correction, and calculates segmental and global strain curves. By temporal derivation of strain, the corresponding strain rate is obtained. Global parameters refer to the total deformation of the chamber during the cardiac cycle in the selected view; the entire length of the tracked myocardium is considered as baseline length. We recorded global peak systolic strain (GS), global systolic strain rate (GSRs), global early diastolic strain rate (GSRe), and global late diastolic strain rate of the systemic right ventricle in the longitudinal direction. We have previously reported on the reproducibility of RV deformation parameters with STE in a sample of 27 adults with D-TGA in our laboratory. The mean absolute percentage error for measurements performed in random order by two observers was 6.9%, 8.9%, and 12.3% for GS, GSRs, and GSRe, respectively.
Descriptive statistics are presented as mean ± SD for continuous variables and as number (percentage) for categorical variables. The following standard parameters of RV function were considered as potential predictors of events: RV fractional area change, end-diastolic RV area (both raw and indexed for body surface area), TAPSE, tricuspid annular s′ and e′ velocities, RV isovolumic velocity and acceleration, and RV MPI. Among deformation-based parameters of RV function, GS, GSRs, and GSRe were considered. The association of clinical and echocardio graphic parameters with risk for clinical events was assessed using Cox proportional-hazards models. Proportionality of hazards was tested using the Schoenfeld residuals. We evaluated also for nonlinear associations of RV parameters with risk using multivariate fractional polynomials. In multivariate models, we examined the association of echocardiographic parameters with clinical events, controlling for clinical predictors of events. The correlation of parameters of RV function with clinical characteristics was assessed with linear regression models. The predictive value and optimal cutoff point for variables significantly associated with risk for events was assessed using receiver operating characteristic curve analysis; the confidence intervals (CIs) for the area under the curve were obtained using the exact binomial method. Analyses were performed using Stata version 11 (StataCorp LP, College Station, TX).
Clinical Characteristics and Events
Sixty-four patients with D-TGA fulfilled all enrollment criteria and were included in the study. The mean age was 29 ± 6 years, 22 patients were women, and 31 had pacemakers. The type of atrial switch procedure was Mustard in 43 patients (67%) and Senning in 21 patients (33%). The mean time from atrial repair in our population was 27 ± 5 years (median, 26 years; interquartile range, 23–30 years). Beta-blockers were used in 41% of patients, 31% were receiving angiotensin-converting enzyme inhibitors, and 28% were receiving digoxin. Seven patients (11%) were in atrial flutter or fibrillation at baseline. Median follow-up was 2.4 years (interquartile range, 1.5–4.1 years). During this period, 12 patients (19%) presented with clinical events; stage C HF developed in 11 patients and VT in one patient (detected on 24-hour Holter recording for clinical symptoms). The baseline clinical characteristics of the cohort and their associations with clinical events are presented in Table 1 . In univariate Cox proportional-hazards models, only time from atrial switch repair (hazard ratio [HR] per year since repair, 1.12; 95% CI, 1.01–1.24; P = .037) and use of β-blockers (HR, 4.31; 95% CI, 1.16–16.0; P = .029) were associated with risk for clinical events among clinical characteristics.
|Parameter||Value||HR (95% CI)||P|
|Age (y)||29 ± 6||1.06 (0.98–1.16)||.15|
|Women||22 (34%)||0.40 (0.09–1.82)||.24|
|Nonwhite||14 (22%)||1.69 (0.45–6.30)||.43|
|Type of atrial switch|
|Senning||21 (33%)||0.96 (0.25–3.62)||.95|
|Time from repair (y)||27 ± 5||1.12 (1.01–1.24)||.037|
|Body mass index (kg/m 2 )||25.8 ± 4.9||1.07 (0.96–1.20)||.24|
|Systolic blood pressure (mm Hg)||118 ± 13||1.03 (0.98–1.08)||.21|
|Diastolic blood pressure (mm Hg)||73 ± 11||1.05 (0.99–1.12)||.13|
|Heart rate (beats/min)||63 ± 10||1.02 (0.97–1.09)||.42|
|QRS duration (msec)||111 ± 27||1.01 (1.00–1.03)||.11|
|Corrected QT interval (msec)||438 ± 34||1.00 (0.99–1.02)||.49|
|Pacemaker||31 (48%)||1.01 (0.32–3.14)||.99|
|Atrial flutter/fibrillation||7 (11%)||0.59 (0.08–4.58)||.61|
|β-blockers||26 (41%)||4.31 (1.16–16.0)||.029|
|Angiotensin-converting enzyme inhibitors||20 (31%)||1.47 (0.47–4.66)||.51|
|Digoxin||18 (28%)||1.00 (0.30–3.33)||.99|
|Antiarrhythmic agents (all classes)||5 (8%)||0.78 (0.10–6.05)||.56|
Echocardiographic Parameters and Association with Clinical Events
Mean RV fractional area change was 22.9 ± 7.5%, mean indexed RV area at diastole was 21.8 ± 3.7 cm 2 /m 2 , mean tricuspid s′ was 5.1 ± 1.8 cm/sec, mean TAPSE was 9.8 ± 2.8 mm, and mean RV MPI was 0.63 ± 0.13. The mean LV ejection fraction was 61.9 ± 9.0%. Table 2 summarizes the standard echocardiographic parameters at baseline and their associations with clinical events. Among standard echocardio graphic parameters, only RV fractional area change was significantly associated with risk for clinical events (HR per 1% decrease, 1.08; 95% CI, 1.00–1.17; P = .047) in univariate Cox models. The base diameter of the right ventricle was only marginally associated with risk for events (HR per millimeter, 1.11; 95% CI, 0.99–1.23; P = .072).
|Parameter||Value||HR (95% CI)||P|
|RV fractional area change (%) ∗||22.9 ± 7.5||1.08 (1.00–1.17)||.047|
|RV diameter (base) (mm)||45.6 ± 5.6||1.11 (0.99–1.23)||.072|
|RV wall thickness (mm)||10.0 ± 1.8||1.17 (0.86–1.58)||.32|
|RV area at diastole (cm 2 )||41.4 ± 7.9||1.03 (0.96–1.10)||.37|
|RV area at diastole (cm 2 /m 2 )||21.8 ± 3.7||0.93 (0.80–1.09)||.36|
|Tricuspid s′ velocity (cm/sec) ∗||5.1 ± 1.8||1.16 (0.83–1.63)||.39|
|Tricuspid e′ velocity (cm/sec) ∗||5.4 ± 2.3||1.04 (0.83–1.30)||.76|
|RV isovolumic velocity (cm/sec) ∗||4.0 ± 1.6||1.12 (0.78–1.61)||.53|
|RV isovolumic acceleration (m/sec 2 ) ∗||1.33 ± 0.68||1.33 (0.48–3.68)||.58|
|RV MPI †||0.63 ± 0.13||1.01 (0.97–1.06)||.51|
|TAPSE (mm) ∗||9.8 ± 2.8||1.00 (0.81–1.23)||.97|
|Tricuspid regurgitation severity ‡||1.5 ± 1.3||0.94 (0.56–1.59)||.82|
|Left (subpulmonic) ventricular ejection fraction (%) ∗||61.6 ± 9.0||1.00 (0.95–1.06)||.93|
|RV global strain (%)||−12.5 ± 3.0||1.35 (1.14–1.58)||<.001|
|RV global systolic SR (sec −1 ) †||−0.59 ± 0.14||1.06 (1.02–1.11)||.006|
|RV global early diastolic SR (sec −1 ) §||0.68 ± 0.22||1.04 (1.00–1.07)||.031|
|RV global late diastolic SR (sec −1 ) §,¶||0.32 ± 0.17||1.02 (0.98–1.06)||.42|
The mean frame rate during acquisition for speckle tracking was 68 ± 17 frames/sec (median, 71 frames/sec; interquartile range, 63–79 frames/sec). Longitudinal GS, GSRs, and GSRe of the right ventricle were −12.5 ± 3.0%, −0.59 ± 0.14 sec −1 , and 0.68 ± 0.22 sec −1 , respectively ( Table 2 ). The longitudinal GS of the right ventricle was most strongly associated with risk for events, with an HR of 1.35 per 1% worsening (95% CI, 1.14–1.58; P < .001), followed by GSRs (HR per 0.01 sec −1 , 1.06; 95% CI, 1.02–1.11; P = .006), and GSRe (HR per −0.01 sec −1 , 1.04; 95% CI, 1.00–1.07; P = .031).
The association of GS (HR per 1%, 1.28; 95% CI, 1.05–1.56; P = .014) and GSRs (HR per 0.01 sec −1 , 1.05; 95% CI, 1.01–1.11; P = .030) with clinical events remained significant after adjusting for clinical predictors of events (time since repair and use of β-blockers), whereas the association of GSRe (HR per −0.01 sec −1 , 1.02; 95% CI, 0.99–1.06; P = .146) and RV fractional area change (HR per 1% decrease, 1.04; 95% CI, 0.96–1.13; P = .36) were attenuated to nonsignificant levels.
In receiver operating characteristic curve analysis ( Figure 3 ), we observed that RV GS ≥ −10% at baseline optimally discriminated risk for clinical events with 75% sensitivity and 90% specificity and a clinically relevant area under the curve (C = 0.83; 95% CI, 0.71–0.91; P < .001). Using this cutoff point, patients with GS ≥ −10% at baseline had significantly higher risk for clinical events compared with those with GS < −10% (HR, 12.6; 95% CI, 3.4–46.6; P < .001), defining thus two groups with distinctly different prognoses ( Figure 4 ). The corresponding areas under the curve for other RV echocardiographic parameters significantly associated with clinical events were 0.78 (95% CI, 0.66–0.87) for GSRs, 0.75 (95% CI, 0.62–0.85) for fractional area shortening, and 0.73 (95% CI, 0.61–0.84) for GSRe.