Prognostic value of myocardial deformation parameters for outcome prediction in tetralogy of Fallot

Abstract

Background

The prognostic value of myocardial deformation parameters in adults with repaired tetralogy of Fallot (rTOF) has not been well-elucidated. We therefore aimed to explore myocardial deformation parameters for outcome prediction in adults with rTOF using cardiovascular magnetic resonance imaging (CMR).

Methods

Adults with rTOF and at least moderate pulmonary regurgitation were identified from an institutional prospective CMR registry. Left ventricular (LV) and right ventricular (RV) global strains were recorded in longitudinal (GLS), circumferential (GCS), and radial (GRS) directions. Major adverse cardiovascular events (MACE) were defined as a composite of mortality, resuscitated sudden death, sustained ventricular tachycardia (>30 seconds), or heart failure (hospital admission >24 hours). In patients with pulmonary valve replacement (PVR), pre- and post-PVR CMR studies were analyzed to assess for predictors of complete RV reverse remodeling, defined as indexed RV end-diastolic volume (RVEDVi) <110 mL/m². Logistic regression models were used to estimate the odds ratio (OR) per unit change in absolute strain value associated with clinical outcomes and receiver operator characteristic curves were constructed with area under the curve (AUC) for select CMR variables.

Results

We included 307 patients (age 35 ± 13 years, 59% (180/307) male). During 6.1 years (3.3–8.8) of follow-up, PVR was performed in 142 (46%) and MACE occurred in 31 (10%). On univariate analysis, baseline biventricular ejection fraction (EF), mass, and all strain parameters were associated with MACE. After adjustment for LVEF, only LV-GLS remained independently predictive of MACE (OR 0.822 [0.693–0.976] p = 0.025). Receiver operator curves identified an absolute LV-GLS value less than 15 and LVEF less than 51% as thresholds for MACE prediction (AUC 0.759 [0.655–0.840] and 0.720 [0.608–0.810]). After adjusting for baseline RVEDVi, RV-GCS (OR 1.323 [1.094–1.600] p = 0.004), LV-GCS (OR 1.276 [1.029–1.582] p = 0.027) and LV-GRS (OR 1.101 [1.0210–1.200], p = 0.028) were independent predictors of complete remodeling post-PVR remodeling.

Conclusion

Biventricular strain parameters predict clinical outcomes and post-PVR remodeling in rTOF. Further study will be necessary to establish the role of myocardial deformation parameters in clinical practice.

Graphical abstract

1. Introduction

With remarkable advances in surgical technique and medical management of patients with tetralogy of Fallot (TOF), more than 95% of children survive into adulthood [1]. Because major adverse clinical events (MACE) escalate with advancing age in this growing population of adults [1], [2], risk prediction in this population is becoming increasingly important. Chronic pulmonary regurgitation (PR) contributes to right ventricular (RV) dilatation which may result in deterioration of ventricular systolic function. With the recognition that cardiovascular magnetic resonance imaging (CMR) is the reference standard for volumetric and functional assessment, routine surveillance CMR has become the standard of care for adults with repaired TOF (rTOF). It is widely recognized that systolic dysfunction, of one or both ventricles, is predictive of adverse events in rTOF [3]. Early identification of individuals at risk of functional decline is highly desirable. Feature tracking (FT) on CMR has sparked considerable interest as it allows for straightforward and reliable assessment of myocardial deformation using routinely prescribed and readily available cine acquisitions [4]. However, the incremental value of myocardial deformation measures, alongside or in place of, conventional CMR parameters, has not been well-elucidated.

The Comprehensive Outcomes Registry Late After Tetralogy of Fallot Repair (CORRELATE) is an established prospective CMR registry of children and adults identified as being at high risk of adverse events due to significant PR. The CORRELATE registry links clinical, imaging, and functional data with CMR measures to identify clinically relevant outcomes [5]. Utilizing this existing resource, the purpose of this study was twofold. First, we aimed to explore the relationship between CMR strain parameters and adverse clinical outcomes in adults with rTOF to determine whether strain provides incremental value in risk stratification of these patients. Second, with the ongoing debate regarding the optimal timing of pulmonary valve replacement (PVR), we investigated potential contributions from strain parameters for prediction of complete RV reverse remodeling following PVR.

2. Methods

2.1. Study population

We included eligible patients from the prospective CORRELATE database with imaging and clinical follow-up at our institution (2013–2023). The complete methodology as it pertains to the development of this international registry, including details of CMR analysis, has been previously described [5]. The larger CORRELATE study prospectively enrolled individuals $\ge 12$ years with surgically rTOF in childhood, CMR imaging within 18 months of enrollment, and moderate PR. Participants in the CORRELATE study provided written consent and institutional review board (IRB) approval was obtained from the respective sites. Inclusion criteria for the present study reflected those listed for the broader CORRELATE and with the addition that CMR was completed at our institution. Patients were excluded if PVR occurred before study enrollment or if CMR imaging was insufficient for functional analysis. For the present study, IRB approval was specifically obtained and the need for additional informed consent was waived.

2.2. Data collection

Clinical data pertaining to the general demographics of this subpopulation were abstracted from existing medical records. This included review of all available medical reports and detailed recording of relevant information from electrocardiograms, echocardiograms, Holter monitors, exercise studies, and cardiovascular procedures. In keeping with the larger CORRELATE registry, MACE was a composite of all-cause cardiovascular mortality, resuscitated sudden death, sustained ventricular tachycardia (VT >30 seconds) or heart failure (with hospital admission >24 hours); additionally, time to first pulmonary valve intervention (surgical or percutaneous) was recorded. All study outcomes were independently adjudicated by two adult congenital heart disease clinicians with 12 and 24 years of clinical experience, respectively. Post-PVR CMR studies were completed 6–9 months after PVR to ensure that remodeling of the ventricles was captured. We defined complete RV reverse remodeling in the post-PVR group as RV end-diastolic volume indexed (RVEDVi) <110 mL/m² [6]. A sensitivity analysis was performed to explore a less stringent threshold for reverse remodeling of RVEDVi <120 mL/m² as defined more recently by some authors [7].

2.3. Cardiovascular magnetic resonance imaging

All CMR studies were completed on a 1.5T Siemens scanner (Magnetom Avanto Fit, Healthineers, Erlangen, Germany) with commercially available surface coils. Multiplanar breath-held steady-state free precession cine imaging with the use of generalized autocalibrating partially parallel acquisitions with acceleration factor of 2 at a slice thickness of 8 mm with 2 mm inter-slice gap and retrospective cardiac gating was performed (in 2/3/4-chamber views, axial and short-axis stack, and sagittal oblique views across the RV outflow tract). Measurements of RV-free wall strain and global strain for the left ventricle (LV) were completed using commercially available FT software (CVI 42 version 5.17, Circle,Clgary, Alberta, Canada). For FT strain analysis, LV and RV-free wall endocardial and epicardial borders were contoured at end-diastole on the complete short-axis stack (8–12 slices), excluding only the surgical patch, and on long-axis (2/3/4-chamber) cine views (Fig. 1). A single experienced cardiothoracic radiologist blinded to clinical outcomes completed strain analysis with peak global longitudinal strain (GLS) calculated by averaging all peak longitudinal myocardial values, while peak global circumferential strain (GCS) and peak global radial strain (GRS) were obtained by averaging peak myocardial strain values across the short-axis stack. Of note, although better GLS and GCS are conventionally expressed with greater negative values and GRS with greater positive values, for simplicity of presentation of data comparisons, measurements are generally shown as absolute values without assignment of positivity or negativity. The reliability of CMR measurements within and between observers was evaluated in a random subset of 30 studies. Intra-observer measurements occurred 1 month apart and inter-observer measurements were completed independently. Normal RV and LV strain values for GLS, GCS, and GRS based on institutional normal values are included for purposes of comparison (Supplemental Table S1).

Fig. 1.

Biventricular strain analysis with manual contouring of the left ventricular (LV) epicardium (green), LV endocardium (red), right ventricular (RV) epicardium (blue), and RV endocardium (yellow) as shown. All three long-axis views, four-chamber (a), two-chamber (b), and three-chamber (c) were used for global longitudinal strain analysis. Complete short-axis cine stack, 8–12 slices (d), eliminating only the akinetic/dyskinetic RVOT patch (h) were used for global circumferential and global radial strain analysis. Biventricular global strain curves were obtained in all three directions (e-g). GLS global longitudinal strain, GCS global circumferential strain, GRS global radial strain, RVOT right ventricular outflow tract.

2.4. Statistical analysis

Continuous data were evaluated for normal distribution using the Shapiro-Wilk test. Continuous variables were described using mean (standard deviation) or median (and interquartile ranges) as appropriate. Categorical variables were shown using numbers (percentages). Logistic regression models were used to estimate the odds ratio (OR) for MACE and RV remodeling (shown with 95% confidence intervals) expressed as an incremental per unit change in the absolute strain value irrespective of strain direction. For the assessment of incremental diagnostic value, the likelihood ratio test was used to compare nested models with RVEDVi alone versus the sequential addition of strain parameters. Receive operator characteristic (ROC) curves were constructed and the area under the curve (AUC) was examined for sensitivity and specificity thresholds. Inter-observer and intra-observer agreement was evaluated using the intraclass correlation coefficient (ICC) with a two-way random effect model. A two-tailed p value of <0.05 was considered statistically significant. Statistical analysis was performed using a commercially available software package, STATA v14.1 (StataCorp, College Station, Texas, ). Datasets will be made available upon reasonable request.

3. Results

3.1. Study population

In total, 317 prospectively enrolled patients with rTOF were reviewed and 307 were included in this study (mean age 35 ± 13 years, 59% (180/307) male). Ten patients were excluded due to insufficient imaging for FT analysis. The median duration of follow-up was 6.1 years (range 3.3–8.8). Demographic and clinical characteristics of the study population are shown (Table 1). Definitive surgical repair (mean age 4.4 ± 3.6 years) was transannular patch repair in the majority (57%, n = 175) and valve sparing in the minority (23%, n = 72). In total, 35 non-mutually exclusive MACE events were observed in 31 patients (10%): death (n = 7), resuscitated cardiac arrest (n = 3), sustained VT (n = 14), and heart failure requiring hospital admission (n = 11) .

Table 1.

Baseline characteristics of study population (n = 307). Data are presented as numbers (%) or mean ± standard deviation (SD).

Characteristic	% (n) or mean ± SD
Male	58.6% (180)
Age at study entry (years)	34.5 ± 12.6
Diagnosis
Tetralogy of Fallot with pulmonary stenosis	92.8% (285)
Tetralogy of Fallot with pulmonary atresia	3.3% (10)
Tetralogy of Fallot with absent pulmonary valve	1.3% (4)
Tetralogy of Fallot with atrioventricular septal defect	0.6% (2)
Other	1.9% (6)
Race/ethnicity*
White	66.4% (204)
Asian	17.3% (53)
Black/African	1.9% (6)
Hispanic/Latino	1.6% (5)
First Nation	1.3% (4)
Multiple	4.2% (13)
Others	2.6% (8)
Not reported	4.5% (14)
Body mass index (kg/m2)
<18	3.0% (9)
18–25	43.0% (131)
>25	54.0% (166)
Age at definitive surgical repair (years)	4.4 ± 3.6
Type of definitive surgical repair
Transannular patch	57.0% (175)
Non-transannular patch (valve sparing)	23.0% (72)
Right ventricle to pulmonary artery conduit	4.0% (13)
Unknown	15.0% (47)
Syndrome	16.0% (50)
22q11 deletion	7.0% (21)
Trisomy 21	3.0% (10)
Holt-Oram syndrome	0.6% (2)
Charge syndrome	0.6% (2)
Other	5.0% (15)
Cardiovascular risk factors
Hypertension	8.0% (24)
Hypercholesterolemia	5.0% (16)
Diabetes	2.6% (8)
Smoking history	21.0% (65)
Specific activity scale (classification I–IV)
I	65.0% (198)
II	11.0% (33)
III and IV	6.0% (20)
Not reported	18.0% (56)
ECG characteristic
QRS interval duration (msec)	151.3 ± 25.5
Echocardiographic parameter
Tricuspid regurgitation, moderate or severe	9.0% (27)
Cardiopulmonary study
Peak heart rate, absolute value (bpm)	155 ± 24
Peak heart rate, percent of maximum predicted (%)	83 ± 12
Peak aerobic capacity (VO2), absolute value (mL/kg/min)	23 ± 7
Peak aerobic capacity (VO2), % of maximum predicted	68 ± 16
Anaerobic threshold, absolute (mL/kg/min)	14 ± 5
Anaerobic threshold, % of maximum predicted	44 ± 12
Peak O2 pulse, absolute (mL/beat)	11 ± 3
Peak O2 pulse, % of maximum predicted	85 ± 18

ECG electrocardiogram, SD standard deviation.

^*Since the Comprehensive Outcomes Registry Late After Tetralogy of Fallot Repair (CORRELATE) study originated in Canada, the breakdown of ethnicity is according to categories provided by the government of Canada for the collection of census data which can be viewed using the link https://www.statcan.gc.ca/eng/statistical-programs/instrument/5178_Q1_V1-eng.pdf (please see section 19, page 10 for race/ethnicity categories).

3.2. Ventricular strain and prediction of MACE in rTOF

The CMR characteristics of the population stratified by clinical outcomes and PVR status are demonstrated (Table 2, Fig. 2). Briefly, the subgroup with MACE+ had larger baseline ventricular volumes, lower biventricular ejection fraction (EF), and lower biventricular global strain values as compared with the MACE− subgroup (Table 2).

Table 2.

Comparison of baseline demographic and cardiovascular magnetic resonance imaging parameters in subjects with and without major adverse cardiovascular events (MACE) and those referred for pulmonary valve replacement (PVR). Data are presented as numbers (%) or mean ± standard deviation.

Parameter	Overall n = 307	MACE+ n = 31	MACE− n = 276	p value	PVR+ n = 142	PVR− n = 165	p value
Age (years)	34 ± 13	45 ± 13	33 ± 12	<0.001	36 ± 12	33 ± 12	0.019
Sex (males)	180 (59%)	24 (77%)	156 (57%)	0.025	92 (65%)	88 (58%)	0.042
Complex TOF*	22 (7%)	1 (3%)	21(8%)	0.370	11 (77%)	11 (67%)	0.716
Ethnicity (non-Whites)	103 (33%)	11 (65%)	91 (67%)	0.779	93 (65%)	110 (66%)	0.842
Body mass index
<18	9 (3%)	3 (10%)	6 (2%)	0.008	76 (54%)	90 (54%)	0.713
18–25	131 (43%)	7 (22%)	124 (45%)		63 (44%)	69 (42%)
>25	167 (54%)	21 (68%)	146 (53%)		3 (2%)	6 (4%)
BSA (kg/m2)	1.8 ± 0.2	1.9 ± 0.3	1.8 ± 0.2	0.260	1.8 ± 0.2	1.8 ± 0.2	0.777
Cardiovascular risk factors**	92 (30%)	11 (35%)	81 (29%)	0.480	43 (30%)	49 (30%)	0.911
Syndrome	50 (16%)	8 (26%)	42 (15%)	0.130	21(15%)	29 (18%)	0.510
Specific activity scale
Class I	198 (65%)	14 (45%)	183 (67%)	0.043	88 (62%)	109 (66%)	0.293
Classes II–IV	53 (17%)	8 (26%)	45 (16%)		27 (19%)	26 (16%)
Unknown	56 (18%)	9 (29%)	48 (17%)		27 (19%)	30 (18%)
ECG QRS duration (ms)	147 ± 22	162 ± 30	150 ± 55	0.009	155 ± 28	147 ± 22	0.003
Echocardiogram TR (moderate/severe)	27 (9%)	6 (19%)	21 (8%)	0.256	17 (12%)	10 (6%)	0.377
Cardiopulmonary study
Peak HR (bpm)	155 ± 24	143 ± 25	156 ± 23	0.016	154 ± 22	156 ± 25	0.403
Peak HR (%)	83 ± 12	82 ± 12	82 ± 12	0.739	83 ± 10	82 ± 14	0.796
Peak VO2 (mL/kg/min)	23 ± 7	19 ± 6	24 ± 7	0.001	23 ± 7	23 ± 7	0.840
Peak VO2 (%)	68 ± 16	62 ± 19	69 ± 16	0.065	69 ± 16	67 ± 16	0.465
Anaerobic threshold (mL/ kg/min)	15 ± 5	12 ± 3	15 ± 5	0.053	14 ± 5	15 ± 5	0.114
Anaerobic threshold (%)	43 ± 12	42 ± 15	43 ± 12	0.146	43 ± 11	44 ± 13	0.236
Peak O2 pulse (mL/beat)	11 ± 3	10 ± 3	11 ± 3	0.723	11 ± 3	11 ± 3	0.832
Peak O2 pulse (%)	85 ± 18	77 ± 20	86 ± 18	0.105	87 ± 20	83 ± 17	0.345
CMR
RVEDV (mL)	300 ± 86	336 ± 82	296 ± 86	0.017	346 ± 89	261 ± 59	<0.001
RVEDVi (mL/m2)	166 ± 42	182 ± 47	164 ± 41	0.028	191 ± 41	145 ± 29	<0.001
RVESV (mL)	172 ± 62	214 ± 67	168 ± 60	<0.001	204 ± 66	145 ± 42	<0.001
RVESVi (mL/m2)	95 ± 31	116 ± 37	93 ± 30	<0.001	113 ± 31	80 ± 22	<0.001
RVEF (%)	43 ± 7	36 ± 7	44 ± 7	<0.001	41 ± 7	44 ± 7	<0.001
RV mass (g)	63 ± 17	73 ± 15	62 ± 17	<0.001	70 ± 18	57 ± 13	<0.001
RV mass indexed (g/m2)	35 ± 7	39 ± 7	34 ± 7	<0.001	38 ± 7	31 ± 5	<0.001
LVEDV (mL)	157 ± 48	180 ± 73	155 ± 44	0.068	165 ± 55	149 ± 40	0.003
LVEDVi (mL/m2)	86 ± 20	94 ± 28	85 ± 19	0.076	91 ± 23	82 ± 17	<0.001
LVESV (mL)	73 ± 30	99 ± 54	71 ± 25	0.008	80 ± 35	67 ± 24	<0.001
LVESVi (mL/m2)	40 ± 13	51 ± 22	39 ± 11	0.005	43 ± 15	37 ± 11	<0.001
LVEF (%)	54 ± 7	47 ± 11	55 ± 6	0.001	52 ± 8	55 ± 6	0.008
LV mass (g)	94 ± 30	120 ± 38	92 ± 28	0.005	98 ± 32	91 ± 28	0.047
LV mass indexed (g/m2)	51 ± 12	60 ± 15	50 ± 11	0.001	53 ± 12	49 ± 11	0.040
RV-GLS	−19 ± 4	−17 ± 5	−20 ± 4	<0.001	−19 ± 4	−20 ± 4	0.052
RV-GCS	−15 ± 3	−12 ± 3	−15 ± 3	<0.001	−14 ± 3	−16 ± 3	<0.001
RV-GRS	25 ± 8	23 ± 17	26 ± 7	0.407	24 ± 10	26 ± 6	0.028
LV-GLS	−16 ± 3	−13 ± 4	−16 ± 3	<0.001	−16 ± 3	−16 ± 3	0.093
LV-GCS	−17 ± 3	−14 ± 4	−17 ± 3	0.001	−16 ± 3	−17 ± 2	0.002
LV-GRS	28 ± 7	23 ± 9	29 ± 7	0.001	27 ± 8	29 ± 6	0.024
Right atrial area (cm2)	23 ± 7	28 ± 10	22 ± 7	0.004	28 ± 10	25 ± 8	<0.001
Left atrial area (cm2)	18 ± 6	22 ± 9	18 ± 5	0.032	21 ± 7	20 ± 6	0.130

BSA body surface area, CMR cardiovascular magnetic resonance imaging, ECG Electrocardiogram, HR heart rate, LV left ventricular, LV-GCS left ventricular global circumferential strain, LV-GLS left ventricular global longitudinal strain, LV-GRS left ventricular global radial strain, LVEDVi indexed left ventricular end-diastolic volume, LVEF left ventricular ejection fraction, LVESVi indexed left ventricular end-systolic volume, RV right ventricular, RV-GLS right ventricular global longitudinal strain, RV-GRS right ventricular global radial strain, RVEDVi indexed right ventricular end-diastolic volume, RVEF, right ventricular ejection fraction, RVESVi indexed right ventricular end-systolic volume, RVGCS right ventricular global circumferential strain, TOF tetralogy of Fallot, TR tricuspid, regurgitation, VO2 volume of oxygen uptake.

Statistically significant p values are shown in bold.

^*Complex TOF includes pulmonary atresia with confluent pulmonary arteries (n = 10), coronary artery anomalies (n = 5), absent pulmonary valve (n = 4), atrioventricular septal defect (n = 2), and pulmonary atresia and major aortopulmonary collaterals (n = 1).

^**Cardiovascular aggregate risk includes the presence of any of the following risk factors: diabetes, hypertension, dyslipidemia, and/or smoking history.

Fig. 2.

Comparison of the differences in biventricular strain values (y axis) in MACE+ and MACE− subgroups. LV-GCS left ventricular global circumferential strain, LV-GLS left ventricular global longitudinal strain, LV-GRS left ventricular global radial strain, MACE major adverse cardiac events; RV-GCS right ventricular global circumferential strain, RV-GLS right ventricular global longitudinal strain, RV-GRS right ventricular global radial strain. Note: The magnitude of GLS and GCS values are shown as positive values to facilitate comparisons with GRS.

Associations between CMR measurements and MACE are demonstrated (Table 3a). Univariate analysis revealed that baseline indexed LV/RV volumes, indexed LV/RV mass, LV/RVEF, and all LV/RV global strain values were significant predictors of MACE. A sensitivity analysis restricted to non-PVR patients confirmed the association between LV/RV-GLS and LV-GCS and MACE (Supplemental Table S2). In multivariable models with adjustment for LVEF, LV-GLS was the only strain measure independently predictive of MACE (OR 0.82, [0.69–0.97], p = 0.025). Examining ROC curves for optimal sensitivity and specificity thresholds, LV-GLS absolute value <15 and LVEF <51% were identified as values with the strongest discriminatory capacity for MACE prediction (AUC 0.76 [0.655–0.840] and 0.72 [0.608–0.810], respectively) (Fig. 3). However, in multivariable analysis with adjustment for RVEF, no strain parameters retained statistical significance.

Table 3a.

Baseline CMR predictors of major adverse clinical events (MACE).

	Univariate			Multivariable analysis* Model adjusted for baseline RVEDVi
	OR	95% CI	p value	OR	95% CI	p value
RVEDVi (mL/m2)	1.009	1.001–1.016	0.013	-	-	-
RVESVi (mL/m2)	1.019	1.009–1.029	<0.001
RVEF (%)	0.873	0.828–0.921	<0.001	-	-	-
RV mass (g/m2)	1.007	1.032–1.128	0.01	-	-	-
LVEDVi (mL/m2)	1.009	1.003–1.016	0.007	-	-	-
LVESVi (mL/m2)	1.054	1.029–1.080	<0.001
LVEF (%)	0.890	0.848–0.935	<0.001	-	-	-
LV mass (g/m2)	1.068	1.036–1.101	<0.001	-	-	-
RV-GLS	0.845	0.766–0.932	0.001	0.852	0.769–0.944	0.002
RV-GCS	0.758	0.671–0.856	<0.001	0.771	0.679–0.875	<0.001
RV-GRS	0.938	0.903–0.976	0.001	0.942	0.904–0.980	0.003
LV-GLS	0.749	0.658–0.851	<0.001	0.763	0.668–0.872	<0.001
LV-GCS	0.761	0.674–0.859	<0.001	0.775	0.681–0.881	<0.000
LV-GRS	0.892	0.843–0.944	<0.001	0.902	0.850–0.957	0.001

CI confidence interval, LVEDVi indexed left ventricular end-diastolic volume, LVEF left ventricular ejection fraction, LVESVi indexed left ventricular end-systolic volume, LV-GCS left ventricular global circumferential strain, LV-GLS left ventricular global longitudinal strain, LV-GRS left ventricular global radial strain, OR odds ratio, RVEDVi indexed right ventricular end-diastolic volume, RVEF right ventricular ejection fraction, RVESVi indexed right ventricular end-systolic volume, RV-GCS right ventricular global circumferential strain, RV-GLS right ventricular global longitudinal strain, RV-GRS right ventricular global radial strain, LV left ventricular.

Of note, when adjusted for RVEF instead of RVEDVi, no strain variables remain significant; when adjusted for LVEF only LV longitudinal strain remains significant OR 0.822 (95% CI 0.692–0.976), p = 0.025 (see text for further details).

Statistically significant p values are shown in bold.

^*Adjustment for RVEDVi shown.

Fig. 3.

Receiver operator curve (ROC) characteristics demonstrate sensitivity and specificity thresholds for left ventricular ejection fraction (LVEF) and left ventricular global longitudinal strain (LV-GLS) for major adverse cardiovascular event prediction in rTOF. AUC area under the curve, rTOF repaired tetralogy of Fallot.

Further analyses were conducted to explore the association between RVEDVi and strain for prediction of MACE. In a multivariable model with strain parameters and adjustment for baseline RVEDVi, biventricular strain measurements in all three directions remained independent predictors of adverse cardiac events (Table 3a). Furthermore, in nested Cox proportional hazard models with RVEDVi, model fit was improved by the addition of each of the baseline LV and RV strain parameters, respectively, suggesting that strain parameters have an incremental value for predicting risk of MACE beyond RVEDVi. Specifically, as compared with baseline model performance χ² = 5.42, model fit was incrementally improved by sequential addition of each of the six strain values: LV-GCS (χ² = 16.3, p < 0.001), LV-GRS (χ² = 12.8, p < 0.001), LV-GLS (χ² = 23.92, p < 0.001), RV-GCS (χ² = 22.94, p < 0.001), RV-GRS (χ² = 15.49, p < 0.001), and RV-GLS (χ² = 15.72, p < 0.001) (df = 1 for each model, respectively).

3.3. Ventricular strain and prediction of reverse remodeling after PVR

During the study follow-up, a total of 142 patients (46%) underwent PVR and there were 19 MACE events post-PVR. There was no statistically significant difference in MACE between those with and without PVR (p = 0.086). The baseline CMR parameters of the study population stratified by receipt of PVR are shown (Table 2). Specifically, baseline RVEF and LVEF were lower in the PVR+ group (p < 0.001 and p = 0.008). Baseline biventricular GCS and GRS differed statistically between PVR+ and PVR− groups; LV-GLS did not meet statistical significance (p = 0.093) and RV-GLS had borderline statistical significance (p = 0.052). Out of the 142 patients who underwent PVR, 77 patients had a post-PVR CMR available for analysis (comparison of strain values between baseline and post-PVR studies are shown in Supplemental Table S3).

Complete RV remodeling to RVEDVi <110 mL/m² was predicted by baseline RVEF (OR 1.084 [1.017–1.156], p < 0.013) but not LVEF (OR 1.066 [0.990–1.147], p = 0.089). On univariate analysis, RV-GCS, LV-GCS, and LV-GRS were predictive of reverse remodeling (Table 3b). The same parameters remained predictive when adjusted for baseline RVEDVi, specifically RV-GCS (p = 0.010), LV-GCS (p = 0.043), and LV-GRS (p = 0.045). Based on a sensitivity analysis using a less stringent remodeling threshold of 120 mL/m² [7], the findings remained unchanged (Table 3b). In completely remodeled ventricles, post-intervention absolute LV-GLS and LV-GCS strain parameters were significantly higher than the incompletely remodeled ventricles and LV-GRS had borderline significance. In contrast, there were no statistically significant differences in any of the RV strain parameters in those with versus without normalization post-PVR (Table 4).

Table 3b.

Baseline strain parameters and remodeling after pulmonary valve replacement (PVR).

	OR	95% CI	p value	OR	95% CI	p value
Univariate analysis: Post-PVR RVEDVi ≤120 mL/m2
RV-GLS	0.947	0.817–1.098	0.072	-	-	-
RV-GCS	1.323	1.094–1.600	0.004	-	-	-
RV-GRS	0.993	0.948–1.040	0.758	-	-	-
LV-GLS	1.153	0.965–1.379	0.135	-	-	-
LV-GCS	1.276	1.029–1.582	0.027	-	-	-
LV-GRS	1.101	1.010–1.200	0.028	-	-	-
Univariate analysis: Post-PVR RVEDVi ≤110 mL/m2			Multivariable analysis model adjusted for baseline RVEDVi
RV-GLS	0.985	0.852–1.138	0.836	0.915	0.781–1.073	0.275
RV-GCS	1.225	1.025–1.463	0.025	1.230	1.064–1.587	0.010
RV-GRS	0.990	0.944–1.037	0.662	0.979	0.930–1.029	0.405
LV-GLS	1. 085	0.929–1.275	0.327	1.145	0.959–1.367	0.135
LV-GCS	1.289	1.034–1.067	0.024	1.256	1.007–1.566	0.043
LV-GRS	1.098	1.009–1.196	0.031	1.094	1.002–1.195	0.045

LV-GCS left ventricular global circumferential strain, LV-GLS left ventricular global longitudinal strain, LV-GRS left ventricular global radial strain, RVEDVi indexed right ventricular end-diastolic volume, RV-GCS right ventricular global circumferential strain, RV-GLS right ventricular global longitudinal strain, RV-GRS right ventricular global radial strain, PVR pulmonary.

Statistically significant p values are shown in bold.

Table 4.

Comparison of post-pulmonary valve replacement (PVR) strain values in subjects according to right ventricular size post-PVR.

Post-PVR strain parameters
RVEDVi with a normalized cut-off ≤110 mL/m2	Normalized+ n = 30	Normalized− n = 47	p value
LV-GCS	−19 ± 2	−17 ± 3	0.047
LV-GRS	33 ± 6	27 ± 8	0.057
LV-GLS	−18 ± 2	−15 ± 3	0.008
RV-GCS	−13 ± 5	−11 ± 3	0.371
RV-GRS	34 ± 15	31 ± 15	0.613
RV-GLS	−18 ± 5	−16 ± 5	0.386
LVEF	56 ± 7	53 ± 7	0.085
RVEF	43 ± 9	37 ± 8	0.008
RVEDVi with a normalized cut-off ≤120 mL/m2	Normalized+ n = 38	Normalized− n = 39	p value
LV-GCS	−19 ± 2	−16 ± 3	0.047
LV-GRS	33 ± 6	27 ± 9	0.055
LV-GLS	−18 ± 2	−15 ± 3	0.001
RV-GCS	−13 ± 4	−10 ± 2	0.050
RV-GRS	35 ± 17	27 ± 9	0.153
RV-GLS	−18 ± 6	−16 ± 4	0.204
LVEF	57 ± 7	52 ± 7	0.016
RVEF	43 ± 9	36 ± 7	0.002

LVEF left ventricular ejection fraction, LV-GCS left ventricular global circumferential strain, LV-GLS left ventricular global longitudinal strain, LV-GRS left ventricular global radial strain, RVEDVi right ventricular end-diastolic volume indexed, RVEF right ventricular ejection fraction, RV-GCS right ventricular global circumferential strain, RV-GLS right ventricular global longitudinal strain, RV-GRS right ventricular global radial strain, RVEDVi indexed right ventricular end-diastolic volume.

Statistically significant p values are shovn in bold.

3.4. Intra- and inter-observer reliability for strain measurements

Inter and intra-observer reliability assessments for the CMR measurements in our larger CORRELATE registry (specifically biventricular volumes, function, and mass) have been previously published [8]. For strain measurements, reliability was good to excellent for all inter-observer strain measurements (ICC range 0.778–0.994) and intra-observer strain measurements (ICC range 0.818–0.997). While LV strain measurements were associated with excellent reliability (all ICC values >0.9), the RV strain measurements, particularly RV circumferential and longitudinal strain, were associated with good to excellent reliability (ICC range 0.75–0.9).

4. Discussion

We examined the prognostic value of myocardial deformation parameters in adults with rTOF previously enrolled in a prospective CMR registry of patients with significant PR. Using CMR-FT to complete a comprehensive evaluation of biventricular strain in three directions, incorporating all available short-axis and long-axis slices into our analysis, several novel findings emerged as follows: (1) LV-GLS was identified as a robust independent predictor of MACE beyond LVEF, (2) MACE risk increased when LV-GLS absolute value was <15 and LVEF was <51%, respectively, (3) incremental value was provided by biventricular strain parameters, beyond RVEDVi, for MACE prediction, (4) biventricular GCS and LV-GRS were predictors of remodeling post-PVR while RV-GLS and LV-GLS were not. Taken together, these data suggest that CMR-derived strain parameters have value in the prediction of both adverse clinical events and post-PVR imaging outcomes in rTOF.

4.1. Myocardial deformation parameters and MACE

Research efforts are increasingly being focused on delineation of robust predictors of risk in rTOF to optimize late clinical outcomes. The identification of early markers of deterioration is highly desirable. In this regard, the measurement of strain values may have a particular value. Unlike biventricular volumes and EF, there is a relative paucity of literature describing the utility of myocardial deformation using CMR in rTOF. In a small study of 16 rTOF patients, Moon et al identified GLS in the RV and LV as significant predictors of ventricular tachycardia (VT) and sudden cardiac death [9]. In a larger study of 372 rTOF patients, Orwat et al identified LV-GCS and RV-GLS as outcome predictors on univariate analysis [10]. Although their cohort was similar in size to ours, noteworthy differences included a younger cohort (mean age 18 years at entry), a relatively broad primary outcome (non-sustained VT included along with death and sustained VT) with relatively few events (precluding multivariable analysis), and strain analysis was simplified to a single mid-ventricular short-axis slice. We observed predictive value of strain in all three directions, which may reflect a more extensive analysis incorporating all available longitudinal views and short-axis slices from base to apex as well as a higher adverse event rate (10% at 6 years of follow-up).

The central role of LV function in late outcomes prediction in rTOF is well-established. Systolic dysfunction of the LV is not infrequent, occurring in 20%–25% of adults with rTOF [11], [12]. Diller et al identified LV-GLS, albeit on echocardiography, as a predictor of life-threatening ventricular arrhythmia and death in rTOF [13]. Our findings align with published literature providing confirmation that CMR measures of global and longitudinal LV function relate to adverse outcomes in rTOF. In the normal heart, LV contraction is predominantly concentric; however, this pattern may be compromised in the setting of diastolic septal flattening induced by RV volume overload secondary to PR, a manifestation of ventricular-ventricular interactions. Indeed, the relative contribution of ventricular strain parameters is expected to dynamically adapt according to altered loading conditions with consequent geometric alterations [14]. Our data suggest that, in the setting of chronic PR with septal flattening, LV-GCS can be compromised resulting in a compensatory increase in LV-GLS. This postulate is supported by our multivariable analysis which identified LV-GLS as an independent predictor of MACE after adjustment for LVEF (OR 0.822 [0.692–0.976], p = 0.025) with strong discriminatory capacity at an LV-GLS threshold of −15 (AUC 0.760 [0.655–0.840]). Unlike LV strain, none of the RV strain parameters were independent predictors of MACE in multivariable analysis adjusted for RVEF. This may reflect the limited range of RV systolic function values in both MACE+ and MACE− subgroups in our cohort (moderate and mild RV dysfunction, respectively, Table 2) such that RV strain cannot predict early deterioration given already widely established ventricular dysfunction.

In rTOF, RVEDVi is a critical driver in the cascade of pathophysiologic deterioration. Knauth et al observed that RVEDVi was an independent predictor of outcomes in rTOF [15]. We therefore adjusted for RVEDVi in our multivariable models and in doing so observed that all RV and LV strain values retained significance as independent predictors of MACE (Table 3a). Some investigators have identified RV mass:volume ratio as an important independent predictor of death and sustained VT in rTOF [16]. According to LaPlace’s law, chamber enlargement initially leads to an increase in systolic stress that would subsequently normalize by thickening of the ventricular wall with preservation of mass-to-volume ratio [17]. It follows that, in the context of ventricular enlargement, strain parameters will also be altered. Further confirmation of the importance of myocardial deformational parameters beyond RV size is substantiated by the observation that there is incremental value of biventricular strain measures in all directions over RVEDVi for prediction of MACE as demonstrated by nested models in this study.

4.2. Myocardial deformation parameters and reverse remodeling after PVR

The deleterious effect of severe RV dilatation, morphologically and clinically, is well-recognized and there are multiple reports corroborating threshold values for RV volumes, both end-diastolic and end-systolic, beyond which reverse remodeling may not occur consistently despite successful PVR [6], [18], [19], [20], [21], [22]. Although baseline RV size is undeniably important, other functional factors are presumed to have an additional role in reverse remodeling [21], [23]. It is notable that despite adequate ventricular remodeling, RVEF may not recover and late adverse outcomes are still reported [18], [24]. Therefore, our observations that baseline global strain parameters can predict post-PVR reverse remodeling offer an additional dimension that may contribute to our growing understanding of post-PVR outcomes.

In our cohort, normalization of RV volumes was predicted by baseline RVEF (OR 1.084 [1.017–1.156], p < 0.013) but not LVEF. Furthermore, while RV-GCS, LV-GCS, and LV-GRS proved to be predictive of remodeling, GLS measures of both ventricles were not. It is known that in a structurally normal heart, the RV has predominantly longitudinal contraction, in a peristaltic pattern from base to apex, a large contribution of which is from septal shortening [25]. In contrast, due to altered RV geometry and loading conditions in rTOF, circumferential shortening predominates over longitudinal shortening due to myocyte rearrangement rendering the contraction pattern similar to LV architecture [25], [26]. Additionally, significant PR is also known to decrease longitudinal displacement due to loss of atrioventricular coupling [27]. Our results highlight the contribution of circumferential shortening in ventricular remodeling in rTOF. Specifically, our data suggest that when biventricular circumferential strain is preserved, the RV retains remodeling capacity.

Finally, additional insights can be derived from a review of the myocardial deformation characteristics of the subset of ventricles where adequate post-PVR reverse remodeling was achieved. It is notable that in the subgroup of patients with complete versus incomplete reverse remodeling, LV strain values were significantly higher in the former group, with no difference in RV strain parameters between the two subgroups. This observation may be explained, at least in part, by RV-LV interactions with the restoration of a more favorable septal position post-PVR. With recent studies documenting improved LV strain post-PVR [28] and pre-PVR LV strain being identified as an outcome predictor post-PVR [29], larger studies are needed to evaluate the clinical relevance of LV strain values for longer-term risk stratification post-PVR. Of note, RV strain did not differ significantly between those with and without optimal remodeling suggesting that restoration of RV size post-PVR may not equate with amelioration of RV deformational integrity, perhaps due to persistent dysfunction at the myocyte level.

4.3. Potential clinical applications of our study findings and future directions for research

With respect to the prediction of adverse events, our study provides confirmation of the importance of LV-GLS and additionally suggests threshold measurements for LV global function and longitudinal strain, which can further refine our identification of high-risk individuals. While our study used FT-CMR, LV-GLS and LVEF can be derived from routine echocardiography and further validation of their clinical utility should ideally occur using complimentary imaging modalities to facilitate incorporation into clinical practice. Following PVR and complete reverse remodeling, LV-GLS can be expected to improve; however, the implications of incomplete recovery of longitudinal contraction on long-term outcomes remain undefined. Finally, biventricular baseline GCS and LV-GRS are identified as promising novel predictors of post-PVR remodeling reflecting the unique myocardial fiber arrangement and contraction pattern in rTOF.

4.4. Study limitations

Although comparable to reported literature [1] only 10% of our patients experienced MACE thereby limiting the scope of multivariable modeling. Our findings should be further validated in additional centers with longer follow-up to increase study power and to enhance generalizability, with a specific focus on variables of interest that have emerged from our study of this cohort. The external validity of our results may be limited given the evolution of surgical strategies and the younger age at definitive repair in contemporary cohorts TOF patients. It should be noted that ventricular strain, such as EF, incorporates myocardial deformation but omits consideration of preload/afterload and is, therefore, a load-dependent parameter; the impact of varying degrees of ventricular loading from valve regurgitation was beyond the scope of this study. The potential value of RV strain as an early predictor of systolic dysfunction could not be realized in our population as there was pervasive RV systolic dysfunction in our cohort; future studies with greater variability in RV systolic function might allow for further evaluation of RV strain as an early predictor of systolic dysfunction. Finally, the small number of MACE events post-PVR did not allow for exploration of the impact of baseline deformation parameters or post-PVR strain measurements on post-PVR clinical outcomes.

5. Conclusion

In adults, late after TOF repair, biventricular strain can be reliably assessed by FT-CMR using sequences acquired during routine clinical studies. We confirm the robust contribution of LV-GLS and demonstrate the independent and incremental role of biventricular strain in all directions over RVEDVi for the prediction of MACE. Post-PVR, baseline RVEF, biventricular GCS, and LV-GRS predict reverse remodeling. Further studies in larger populations will be necessary to establish the clinical value of myocardial deformation parameters relative to conventional CMR measurements.

Funding

This research was funded by a Canadian Institutes of Health Research Operation Grant (MOP 119353).

Author contributions

Subin K. Thomas: Writing – review and editing, Writing – original draft, Validation, Project administration, Methodology, Investigation, Formal analysis, Data curation. Romina DSouza: Writing - review and editing, Writing - original draft, Validation, Project administration, Methodology, Investigation, Formal analysis, Data curation. Kate Hanneman: Writing – review and editing, Writing – original draft, Methodology, Formal analysis. Gauri R. Karur: Writing – review and editing, Writing – original draft. Christian Hubois: Writing – review and editing, Writing – original draft, Formal analysis, Conceptualization. Ayako Ishikita: Writing – review and editing, Writing – original draft, Formal analysis, Data curation, Conceptualization. Luigia D'Errico: Writing - review and editing, Writing - original draft, Formal analysis, Data curation, Conceptualization. Isaac S. Begun: Writing – review and editing, Writing – original draft, Visualization, Software. Ming-Yen Ng: Writing – review and editing, Writing – original draft, Data curation, Conceptualization. Rachel M. Wald: Writing – review and editing, Writing – original draft, Visualization, Validation, Supervision, Software, Resources, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization.

Ethics approval and consent

This study was approved by the University Health Network research ethics board and a waiver of consent was provided (study number 20-5873).

Consent for publication

Not applicable.

Declaration of competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary material

Supplementary material.

.

Supplementary material.

.