To assess left ventricular remodeling patterns using cardiac computed tomography (CT) in children with congenital heart disease and correlate these patterns with their clinical course.
Left ventricular volume and myocardial mass were quantified in 17 children with congenital heart disease who underwent initial and follow-up end-systolic cardiac CT studies with a mean follow-up duration of 8.4 ± 9.7 months. Based on changes in the indexed left ventricular myocardial mass (LVMi) and left ventricular mass-volume ratio (LVMVR), left ventricular remodeling between the two serial cardiac CT examinations was categorized into one of four patterns: pattern 1, increased LVMi and increased LVMVR; pattern 2, decreased LVMi and decreased LVMVR; pattern 3, increased LVMi and decreased LVMVR; and pattern 4, decreased LVMi and increased LVMVR. Left ventricular remodeling patterns were correlated with unfavorable clinical courses.
Baseline LVMi and LVMVR were 65.1 ± 37.9 g/m2 and 4.0 ± 3.2 g/mL, respectively. LVMi increased in 10 patients and decreased in seven patients. LVMVR increased in seven patients and decreased in 10 patients. Pattern 1 was observed in seven patients, pattern 2 in seven, and pattern 3 in three patients. Unfavorable events were observed in 29% (2/7) of patients with pattern 1 and 67% (2/3) of patients with pattern 3, but no such events occurred in pattern 2 during the follow-up period (4.4 ± 2.7 years).
Left ventricular remodeling patterns can be characterized using cardiac CT in children with congenital heart disease and may be used to predict their clinical course.
Left ventricular hypertrophy (LVH) is a critical factor predicting adverse cardiovascular events such as arrhythmia, heart failure, and sudden cardiac death (1). However, LVH may be a beneficial compensatory reaction to left ventricular wall stress. To analyze the various patterns of left ventricular remodeling (LVR), it is important to consider changes in ventricular volume in addition to myocardial mass. LVR is largely divided into adaptive and maladaptive patterns. LVR caused by pressure and/or volume overload has been studied in patients with aortic valvular disease (234), but has rarely been reported in patients with congenital heart disease, except for congenitally corrected transposition of the great arteries before and after pulmonary artery banding to prepare the left ventricle in systemic circulation following subsequent anatomic repair (567). Cardiac magnetic resonance imaging (MRI) has been used to overcome inherent limitations of echocardiography with respect to geometric assumptions, acoustic window, and operator dependency, and to provide a better evaluation of LVR in patients with aortic stenosis (8). Accuracy and reproducibility of quantification are greatly influenced by the segmentation method for left ventricular volume and more greatly for myocardial mass (910). Generally, threshold-based segmentation, which includes the papillary muscles and trabeculations in the myocardial mass, is superior to the simplified contouring method (11). In addition, a three-dimensional (3D) version of threshold-based segmentation is desirable to reduce errors associated with a two-dimensional version (12). Recently, a semiautomatic 3D threshold-based segmentation method with high reproducibility was successfully applied to 3D cardiac computed tomography (CT) datasets, not only for ventricular volume measurement in patients with congenital heart disease (1314151617), but also for left ventricular myocardial mass quantification in patients with various degrees of LVH (18). However, LVR that considers both volume and mass has not been evaluated with cardiac CT and 3D threshold-based segmentation. Therefore, we aimed to assess LVR patterns using cardiac CT in children with congenital heart disease and correlate these patterns with their clinical course.
This retrospective study was approved by the local Institutional Review Board and the requirement for informed consent was waived.
Between June 2011 and August 2016, 17 patients with congenital heart disease (median age at the first cardiac CT, 22 days; range, 2 days to 16 years; male:female = 10:7) underwent initial and follow-up electrocardiography (ECG)-synchronized end-systolic cardiac CT examinations (time interval, 8.4 ± 9.7 months) using a second-generation dual-source scanner (SOMATOM Definition Flash; Siemens Healthineers, Forchheim, Germany). Therefore, 34 cardiac CT examinations were included in this study. Demographic characteristics, diagnosis, and treatment of the study population are described in Table 1. The median time interval between the most recent surgical or interventional treatment and the first CT study available in 47.1% (8/17) of the patients was 27 days (range, 1 day–8.8 years), and the median time between the most recent surgical or interventional treatment and the follow-up CT study available in all patients was 0.9 years (range, 7 days–9.0 years). The body surface area (m2) on cardiac CT was used to index volumetric measurements.
The fundamental scan parameters of the cardiac CT were as follows: 2 × 64 × 0.6-mm slices with the z-flying focal spot technique; gantry rotation time, 0.28 seconds; temporal resolution, 75 ms; slice width, 0.75 mm; and reconstruction interval, 0.4 mm. Scan protocols were further optimized to minimize radiation dose and maintain diagnostic image quality (1920212223242526). For ECG-synchronized data acquisition, a breath-hold retrospectively ECG-gated spiral scan was used in 8 examinations; free-breathing combined prospectively ECG- and respiratory-triggered sequential scan in 13, and free-breathing prospectively ECG-triggered sequential scan in 13. Of the free-breathing children, respiratory triggering was added to ECG-triggered sequential scanning in patients weighing > 5 kg in whom respiratory motion was large enough to degrade the image quality of the cardiac CT; a pressure-sensing belt from a respiratory gating system (AZ-733V; Anzai Medical Co, Tokyo, Japan) was used to obtain respiratory triggering as previously described (27). For sedation, children (usually younger than 6 years of age) were initially given oral chloral hydrate (50 mg/kg) followed by administration of intravenous midazolam (0.1 mg/kg) or ketamine (1 mg/kg) as required. The lowest possible tube voltage was selected for greater radiation dose efficiency and iodine contrast-to-noise ratio; 70 kV was used in 17 examinations, 80 kV in 13, and 100 kV in four.
Iodinated contrast agent (Iomeron 400, iomeprol 400 mg I/mL; Bracco Imaging SpA, Milan, Italy; 1.5–2.0 mL/kg) was administered intravenously using a dual-head power injector at an injection rate of 0.3–3.0 mL/s. A tri-phasic or quadri-phasic injection protocol was utilized to achieve uniform cardiovascular enhancement and minimal peri-venous streak artifacts from undiluted contrast agent. The scan delay time was determined using a bolus tracking technique with a trigger threshold of 150 HU in the left ventricular cavity.
Volume CT dose index and dose-length product values based on a 32-cm phantom were 3.5 ± 4.7 mGy and 60.6 ± 98.8 mGy·cm, respectively. The effective dose value of cardiac CT calculated by multiplying the dose-length product by the age, sex, and tube voltage-specific conversion factors was 1.2 ± 1.4 mSv (2829).
Left ventricular end-systolic cavity volume and myocardial mass were quantified by using a commercially available workstation (Advantage Workstation 4.6; GE Healthcare, Milwaukee, WI, USA). First, the epicardial border of the left ventricle was delineated consecutively using one-click identification of the left ventricle, a 3D region-growing method, and manual adjustment. The mitral and aortic valve planes were then defined manually. For the total volume of the left ventricle, the left ventricular cavity volume was separated from the left ventricular myocardial volume by using 3D threshold-based segmentation. The left ventricular myocardial mass (in grams) was calculated by multiplying the specific gravity of the myocardium (1.05 g/mL or g/cm3) by the quantified myocardial volume. Left ventricular mass-volume ratio (LVMVR) was calculated by dividing the left ventricular myocardial mass by the left ventricular volume. The left ventricular volume and myocardial mass were indexed to body surface area. Based on a previous article using the same segmentation method, the indexed left ventricular myocardial mass (LVMi) was categorized as either no hypertrophy (< 60.2 g/m2), borderline hypertrophy (60.2–74.0 g/m2), or definite hypertrophy (≥ 74.1 g/m2 ) (18).
Left ventricular overloaded conditions in each patient were characterized based on the following echocardiographic and cardiac CT findings: pressure overload was regarded present when substantial obstruction was observed in the aortic arch (pressure gradient ≥ 20 mm Hg), aortic valve (≥ 36 mm Hg), or left ventricular outflow tract (≥ 30 mm Hg), or when the aortic valve annulus was small (Z score < −2). Volume overload was considered present when aortic or mitral regurgitation greater than grade 2 (or moderate) was observed, or when right-to-left or bidirectional shunt flow through a ventricular septal defect was identified.
Based on changes in LVMi and LVMVR between the two serial cardiac CT examinations, LVR was categorized into one of four patterns: pattern 1, increased LVMi and increased LVMVR (Fig. 1); pattern 2, decreased LVMi and decreased LVMVR (Fig. 2); pattern 3, increased LVMi and decreased LVMVR (Fig. 3); and pattern 4, decreased LVMi and increased LVMVR.
All patients were followed up (4.4 ± 2.7 years) to determine whether an unfavorable event had occurred, such as re-operation or re-intervention for a residual lesion causing left ventricular pressure or volume overload, application of extracorporeal membrane oxygenation, left ventricular dysfunction, myocardial infarction, arrhythmia, heart failure, or cardiac death.
Continuous variables are presented as mean ± standard deviation or median with range, and categorical variables are expressed as frequency with percentage. To explore their clinical significance, LVR patterns were correlated with unfavorable clinical courses. Descriptive data analyses were performed using Excel (Microsoft Corp., Redmond, WA, USA).
The mean baseline LVMi and LVMVR in 17 patients were 65.1 ± 37.9 g/m2 and 4.0 ± 3.2 g/mL, respectively. Of 34 examinations, LVMi showed no hypertrophy in 20, borderline hypertrophy in eight, and definite hypertrophy in six. There was no change in the degree of LVH between serial examinations in 14 patients, but 3 patients showed a change in category. LVMi increased from 52.2 ± 39.0 g/m2 at the initial CT to 70.4 ± 33.1 g/m2 at the follow-up CT in 10 patients during the interval of 0.9 ± 1.0 years, whereas LVMi decreased from 77.9 ± 42.8 g/m2 at the initial CT to 63.1 ± 40.5 g/m2 at the follow-up CT in 7 patients during the interval of 0.5 ± 0.4 years. On the other hand, LVMVR increased from 2.1 ± 1.1 g/mL at the initial CT to 3.4 ± 1.1 g/mL at the follow-up CT in 7 patients during the interval of 0.9 ± 1.1 years, while LVMVR decreased from 6.6 ± 4.4 g/mL at the initial CT to 3.0 ± 2.6 g/mL at the follow-up CT in 10 patients during the interval of 0.6 ± 0.4 years.
Of 17 patients, pressure overload was present in 11 patients (aortic arch obstruction in six, a small aortic valve annulus in three, left ventricular outflow obstruction in two, aortic valve stenosis in one, and pulmonary artery banding in one), and two kinds of pressure overload were identified in two patients. Of 17 patients, volume overload was present in four patients (mitral regurgitation in two, aortic regurgitation in one, and shunt through a ventricular septal defect in one). Two patients had a combination of pressure and volume overload.
Patterns of LVR are summarized in Table 2. Pattern 1 was observed in seven patients (Fig. 1), pattern 2 in seven (Fig. 2), and pattern 3 in three patients (Fig. 3). No patient was categorized into pattern 4. Unfavorable events were identified in four patients (left ventricular dysfunction in three, application of extracorporeal membrane oxygenation in two, cardiac death in two, myocardial infarction in one, and arrhythmia in one) with multiple events in three patients. Unfavorable events were observed in 29% (2/7) of patients with pattern 1 and 67% (2/3) of patients with pattern 3, but no such events occurred in patients with pattern 2 during the follow-up period.
This study demonstrated that LVR patterns could be characterized using cardiac CT in children with congenital heart disease, and the patterns were used to predict clinical course. Based on the preliminary results of this study, the most favorable clinical course was observed in pattern 2 (decreased LVMi and LVMVR), the most unfavorable clinical course in pattern 3 (increased LVMi and decreased LVMVR), and the clinical course in pattern 1 (increased LVMi and LVMVR) was intermediate between the two extremes.
LVH is an adaptive reaction to pressure overloaded conditions that increase left ventricular wall stress to maintain cardiac performance despite the elevated left ventricular systolic pressure, but is associated with excess cardiac mortality and morbidity (3). Notably, incomplete regression of LVH is associated with decreased survival and regarded as a maladaptive response leading to left ventricular diastolic dysfunction (30). To analyze LVR patterns to pressure and/or volume overloaded conditions, LVMi and LVMVR were utilized to consider left ventricular volume in addition to myocardial mass in this study. For example, an increase in left ventricular volume and maintenance of relatively normal LVMVR is generally regarded a hallmark of compensatory hypertrophy. In fact, LVMVR has commonly been used to evaluate LVR and increased LVMVR (i.e., concentric LVH) is associated with a hypertensive response to exercise in patients with repaired coarctation of the aorta (31).
Favorable LVR occurs as a regression process after the relief of overloaded conditions. In a previous study using cardiac MRI (2), it was characterized by a decrease in both LVMi and LVMVR, with an improvement in diastolic filling in patients with aortic valve stenosis early (approximately 9 months) after aortic valve replacement (2). This LVR is comparable to pattern 2 demonstrated in this study.
As previously mentioned, LVH may develop as a compensatory remodeling process (5). After pulmonary artery banding in transposition of the great arteries, serial increases in both LVMi and LVMVR as measured using echocardiography or cardiac MRI indicate that the increment of myocardial mass is disproportionately greater than that of ventricular volume (56). This remodeling pattern is analogous to pattern 1 in this study, showing an intermediate clinical outcome between patterns 2 and 3.
Maladaptive LVR may produce patterns 3 or 4. Pattern 3 is characterized by a disproportionally greater increase in ventricular volume than that in myocardial mass, resulting in a reduction in LVMVR. In contrast, pattern 4 is characterized by a disproportionally greater decrease in ventricular volume than that in myocardial mass, leading to increased LVMVR. In this study, a poor clinical outcome frequently occurred in pattern 3. Pattern 4 was observed in patients with aortic regurgitation after aortic valve replacement (2).
The same strategy as that used in this study may be applied to characterize right ventricular remodeling patterns in response to severe pulmonary regurgitation in patients with repaired tetralogy of Fallot and other congenital heart diseases, mainly involving the right ventricle (32). Myocardial fibrosis is responsible for late diastolic dysfunction that slowly normalizes or persists after treatment in patients with LVH (34). Therefore, the evaluation of myocardial fibrosis using late gadolinium enhancement MRI (33), T1 mapping with extracellular volume map (34), or myocardial delayed-enhancement CT (35), has an additional prognostic value in analyzing LVR.
Retrospective study design and the small study population are major limitations of this study. Its retrospective nature might lead to selection bias in which more severe cases were enrolled in the study population. Moreover, statistical and subgroup analyses were limited due to the small number of patients with variable patient age and pathology. The time intervals between serial cardiac CT examinations were also highly variable. Pathology-based verification of left ventricular mass quantification using 3D CT data is practically impossible. Nonetheless, the same 3D threshold-based approach as used in this study previously demonstrated high reproducibility with a mean difference between the two evaluation sessions of 2.3 ± 1.1% (18). To validate the clinical utility of the CT-based pattern analysis of LVR demonstrated in this study, further studies are necessary. Prospective studies will be particularly valuable to prove its clinical value. Optimal timing of the evaluation and further divisions of LVR patterns remain to be determined. Nevertheless, the results of this study could unveil the great potential of CT-based pattern analysis of LVR in predicting clinical outcomes.
In conclusion, LVR patterns can be characterized using cardiac CT in children with congenital heart disease and may be used to predict their clinical course.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
| Case Number | Sex | Age at 1st Cardiac CT | Time Interval between Serial CT Examinations | Diagnosis and Previous Treatment | Interval Treatment |
|---|---|---|---|---|---|
| 1 | M | 2 days | 19 days | TGA, VSD, S/P BAS | Arterial switch operation |
| 2 | M | 2 days | 82 days | HLHS variant, CoA | Bilateral PAB |
| 3 | F | 4 days | 895 days | CoA | Arch repair |
| 4 | M | 6 days | 378 days | CoA | Arch repair |
| 5 | M | 7 days | 361 days | CoA | Arch repair |
| 6 | M | 15 days | 357 days | IAA, VSD, S/P arch repair | N/A |
| 7 | F | 16 days | 349 days | CoA | Arch repair |
| 8 | F | 20 days | 436 days | Critical aortic stenosis, S/P Sano modified Norwood operation | Pulmonary artery angioplasty, right modified Blalock-Taussig shunt |
| 9 | F | 22 days | 77 days | IAA, VSD, S/P arch repair | N/A |
| 10 | M | 1 month | 344 days | CoA | Arch repair |
| 11 | M | 8 months | 98 days | Congenially-corrected TGA with intact ventricular septum, S/P PAB | N/A |
| 12 | M | 8 months | 161 days | Large ASD, BPD, pulmonary hypertension | Sutureless repair, ASD closure |
| 13 | F | 17 months | 99 days | Williams syndrome | Modified sliding aortoplasty |
| 14 | M | 9 years | 82 days | Williams syndrome | Aortoplasty, bypass graft |
| 15 | F | 10 years | 31 days | TGA, VSD, tricuspid atresia, S/P Fontan operation | VSD extension, mitral valvuloplasty |
| 16 | F | 10 years | 929 days | CoA | N/A |
| 17 | M | 16 years | 19 days | Bicuspid aortic valve, aortic regurgitation | Bentall operation |
| Left Ventricular Remodeling Patterns | LVMi | LVMVR | No. of Cases | Pressure Overload | Volume Overload | Unfavorable Events | Follow-Up Period (Years) |
|---|---|---|---|---|---|---|---|
| Pattern 1 | Increased | Increased | 7 | 57% (4/7) | 29% (2/7) | 29% (2/7) | 2.5–8.2 |
| Pattern 2 | Decreased | Decreased | 7 | 57% (4/7) | 14% (1/7) | 0% (0/5) | 1.1–8.2 |
| Pattern 3 | Increased | Decreased | 3 | 100% (3/3) | 33% (1/3) | 67% (2/3) | 0.8–8.3 |
Tweets
Pattern Analysis of Left Ventricular Remodeling Using Cardiac Computed Tomography in Children with Congenital Heart Disease: Preliminary Results
Twitter
Facebook
WhatsApp
LinkedIn
