Author + information
- Received March 6, 2006
- Revision received September 22, 2006
- Accepted October 9, 2006
- Published online February 27, 2007.
- Phalla Ou, MD⁎,†,⁎ (, )
- David S. Celermajer, MBBS, DSc, FRACP‡,
- Elie Mousseaux, MD, PhD†,§,
- Alain Giron, PhD†,
- Yacine Aggoun, MD∥,
- Isabelle Szezepanski∥,
- Daniel Sidi, MD, PhD∥ and
- Damien Bonnet, MD, PhD∥,¶
- ↵⁎Reprint requests and correspondence:
Dr. Phalla Ou, Department of Pediatric Radiology, Hôpital Necker-Enfants Malades, 149, rue de Sèvres, 75743 Paris Cedex 15, France.
Objectives This study was designed to investigate the influence of aortic arch geometry on vascular remodeling after anatomically successful repair of coarctation of the aorta (CoA).
Background Abnormalities of the precoarctation vasculature are known to occur after CoA repair and appear related to adverse outcomes. The influence of aortic arch geometry on such abnormalities is unknown.
Methods Sixty-three postcoarctectomy subjects (age 15.9 ± 6.3 years) were compared with 63 control volunteers. Aortic arch shape was characterized on magnetic resonance imaging using both qualitative classification, identifying 3 subtypes of arch geometry (Gothic, Crenel, Romanesque), and a quantitative index, height/width ratio (H/W) of the aorta. Using ultrasound, we measured carotid artery intima-media thickness (IMT) and stiffness index and distensibility, as well as right brachial artery flow-mediated dilation (FMD) and glyceryl trinitate (GTN)-induced dilatation, to assess the precoarctation vasculature of these subjects.
Results Gothic arch type was associated with higher carotid IMT and stiffness index, lower carotid distensibility (p < 0.001 for all), and lower brachial reactivity (FMD, p < 0.01; GTN response, p < 0.001) compared with Crenel and Romanesque geometries and with control subjects. The height/width ratio was also significantly related to these vascular abnormalities. Even in CoA subjects with Romanesque arch geometry, arterial function and stiffness parameters were significantly impaired compared with control subjects.
Conclusions In young adult survivors of anatomically successful CoA repair, a gothic-type aortic arch with high H/W is associated with abnormal IMT, higher aortic stiffness index, and impaired arterial reactivity in the pre-CoA vasculature.
Despite operative repair, subjects who have had coarctation of the aorta (CoA) still remain at high risk of cardiovascular morbidity and mortality (1–3). Early vascular remodeling in upper limb conduit arteries has been detected in children and young adults, even after successful and early surgery (4). Vascular remodeling includes impaired vasoreactivity and increased intima-media thickness (IMT) of the pre-CoA arteries (4–6). Such vascular remodeling likely has clinical relevance, as these vascular changes have been shown to be related to cardiovascular risk (7,8).
The determinants of vascular remodeling in postcoarctectomy subjects are unclear. Although remodeling has a high prevalence in subjects studied late after CoA repair, there is a wide overlap of individual values of endothelial function and arterial stiffness parameters between postcoarctectomy subjects and control subjects (9,10). Therefore, the determinants of vascular remodeling in postcoarctectomy subjects have to be further investigated to identify the factors that predispose to the development of adverse vascular changes.
We have recently shown that aortic arch shape deformation with a peculiar “Gothic” geometry is an independent contributor to both resting (11) and exercise-induced hypertension (12) in subjects who have undergone successful CoA surgery. Our current aim was to assess whether Gothic arch geometry is also associated with adverse vascular remodeling in pre-CoA arteries. We therefore examined the relationship between aortic arch geometry and conduit artery structure and function in survivors of CoA repair and control subjects.
The study population (Table 1)included 63 subjects with successful CoA repair, of whom 10 were recruited from a previous study (11). All subjects had had classical postductal CoA repaired with simple or extended end-to-end anastomoses, according to the presence or absence of hypoplasia of the aortic arch. Standard end-to-end anastomosis was performed when CoA was simple without hypoplasia of the aortic arch. Extended end-to-end anastomosis was performed when CoA was associated with hypoplasia of the aortic arch as described previously (13). The proportion of subjects who underwent extended versus simple repairs was similar between groups (p = 0.2).
We included consecutive subjects with isolated CoA who were normotensive at rest, defined as having systolic and diastolic blood pressure (BP) within the normal range according to the Task Force Report on High Blood Pressure in Children and Adolescents (14) and who were not receiving any antihypertensive medication. We excluded subjects with bicuspid aortic valve, aberrant subclavian or other similar vessel abnormalities, or any associated aortic wall abnormalities that may independently influence arterial stiffness (15). Subjects with an interposition graft or subclavian flap repair, those with any previous treatment for re-CoA (balloon angioplasty, stenting, or patch aortoplasty), and those with evidence of aortic arch narrowing and/or re-CoA with any of the following criteria were also excluded: 1) right arm-leg systolic BP gradient at rest >15 mm Hg, 2) systolic peak flow velocity >3m/s with evidence of diastolic runoff on echocardiography with Doppler (using a General Electric Vivid 7 echo machine [Milwaukee, Wisconsin] with 3- and 5-MHz multifrequency probes), and 3) percentage of residual stenosis >30% at the isthmus on magnetic resonance imaging (MRI). Aortic diameters were measured at the narrowest diameter at the site repair (Φm) and descending aorta 10 cm distal to the site repair (ΦD). The percentage of residual stenosis was calculated as 100 × (1 − (Φm/ΦD) (Fig. 1).
We investigated, using an identical protocol, 63 comparable healthy control volunteers. These were either outpatients (n = 38) found to have innocent murmurs or children of hospital employees (n = 25). There were no significant differences between these 2 subgroups of control subjects in any of the parameters measured. The hospital ethics committee approved this study, and all subjects or their guardians gave informed consent.
Resting BP in the right arm and right leg was measured simultaneously by an automatic oscillometric cuff device (Dinamap, Critikon Inc., Tampa, Florida) following at least 10 min of rest in the supine position. The mean of the second and third measures of systolic BP, diastolic BP, pulse pressure, and arm-leg systolic BP gradient were used.
Arterial noninvasive investigations were performed by the same investigator (I.S.), who was not aware of the status of the subjects or control patients.
All subjects and control patients underwent assessment of endothelium-dependent flow-mediated dilation (FMD) and endothelium-independent dilatation to glyceryl trinitrate (GTN) of the right brachial artery using previously described methods (16). After the subject had rested for at least 30 min, FMD was defined as change in arterial diameter in response to reactive hyperemia (increased flow producing endothelium-dependent vasodilation), and GTN response was defined as changes in arterial diameter in response to the endothelium-independent vasodilator GTN, 200 μg given sublingually.
B-mode and M-mode echo measurements were performed on the right common carotid artery. We used a real-time B-mode ultrasound imager (Acuson Sequoia, Mountain View, California). Intima-media thickness and lumen diastolic (Dd) and systolic (Ds) diameters were measured as previously described and validated (16).
One author (Y.A.), blinded to the disease state of the subject, performed image analyses of brachial and carotid arteries off-line with semiautomated video image analysis software (16). Three parameters for functional arterial wall alterations were calculated: the distensibility (distensibility = 2[Ds − Dd/Dd]/ΔP) were the pulse pressure ΔP = systolic BP − diastolic BP, the stiffness index (β stiffness index = [1n(systolic BP/diastolic BP)]/[Ds − Dd/Dd]), and the incremental elastic modulus of elasticity (Einc = 3[1 + (LCSA/IMCSA)]/distensibility). Lumen cross-sectional area (LCSA) equals πDd2/4, and carotid intima-media cross-sectional area (IMCSA) equals π (Dd/2 + IMT)2− π (Dd/2)2.
All MR studies were performed on a 1.5-T magnet (Signa LX,GE Medical Systems, Milwaukee, Wisconsin) equipped with a 23-mT/m/ms gradient system and commercially available software using the body phased-array coil. Image acquisition was triggered on the electrocardiogram. The entire thoracic aorta was obtained according to standard technique as previously described (11). Briefly, a series of T1-weighted images were obtained using acquisition of a 3-dimensional breath hold gadolinium-enhanced MR angiography in an oblique sagittal plane to obtain images along the long axis of both the ascending and descending aorta parallel to blood flow (repetition time/echo time = 4.5/2 ms; image matrix size from 128 to 190 × 256; field of view of 180 to 400 mm and 1 to 3 mm for slice thickness, peripheral injection of 0.1 mmol/kg of gadolinium). Analyses were carried out by an experienced observer (E.M.), who was blinded to subject status and unaware of BP and arterial measurements.
Aortic arch geometry was assessed using either native or maximal intensity projection images of the thoracic aorta. Morphology of the aortic arch was assessed in left anterior oblique projection. We have previously described different types of aortic shape, based both on geometric patterns observed as well as on an objective measure of height and width of the aortic arch (12). First, we qualitatively classified aortic arch shape as “Gothic,” “Crenel,” or “Romanesque” (Fig. 2).Briefly, Gothic arch had a triangular form, Crenel arch had a rectangular form, and Romanesque arch had a semicircular rounded form. This classification was made by two independent experienced MRI radiologists (E.M. and P.O.) with complete concordance between them as regards classification into arch type. Next, we quantitatively characterized aortic arch shape by measuring its height (H) and width (W) and calculating a H/W ratio (Fig. 1). Width corresponded to the maximal transverse length between the midpoints of the ascending and descending aorta, close to the axial plane going through the right pulmonary artery. Height of the aortic arch was defined as the maximal vertical distance between the line W and the highest midpoint of the aortic arch.
All data were analyzed using the JMP software version 5.0.1 (SAS Institute Inc., Cary, North Carolina) and the R software version 1.7.0 (R Foundation for Statistical Computing, Vienna, Austria). Data are presented as the mean value ± SD. The Kolmogorov-Smirnov test for normality was performed to determine whether continuous variables were normally distributed, which was the case for each such parameter. The CoA subjects (as a group) and the control subjects were compared by 2-sample ttests. For assessment of differences between the subtypes of arch geometry and control subjects, we performed ANOVA followed by post-hoc pairwise testing with Scheffe’s test (excluding the small group of subjects with Crenel-type geometry, n = 9).
Univariate and multiple linear regression analyses (with simultaneous entry) were performed in the CoA subjects and on the different subtypes of aortic arch geometry with IMT, distensibility, β stiffness index, Einc, FMD, and GTN as the dependent variables. The following covariables were included in the models: age at operation, age at time of the study or duration of follow-up after surgery, height, weight, resting systolic BP, residual stenosis, H/W ratio, or aortic arch geometry category (Gothic, Crenel, or Romanesque).
A 2-tailed p value ≤0.05 was considered to be statistically significant.
Pre-CoA vascular studies: CoA versus control subjects
Subject characteristics and vascular results from the study are summarized in Table 1. The CoA subjects and control subjects were comparable in age, gender, weight, height, and resting BP. We noted that CoA subjects had significantly higher IMT, lower carotid distensibility, higher β stiffness index and Einc, and reduced brachial FMD and GTN compared with control subjects. No correlation was found between arterial stiffness or endothelial function parameters and age at surgery or duration of follow-up.
Aortic arch geometry types
Twenty-seven subjects (43%) had Gothic geometry, 9 (14%) had Crenel geometry, and 27 (43%) had Romanesque geometry. As anticipated, H/W was significantly higher in subjects with Gothic geometry when compared to the 2 other morphologies (Gothic = 0.82 ± 0.11, Crenel = 0.65 ± 0.16, Romanesque = 0.67 ± 0.1; p < 0.001). The percentage of residual stenosis was comparable among the 3 geometries (Gothic = 16.3 ± 10.4%, Crenel = 19.1 ± 9.9%, Romanesque = 18.4 ± 7.9%; p = 0.1).
Correlations between aortic arch geometry and vascular studies
Vascular parameters for each aortic arch type are shown in Table 2.The 3 subgroups of aortic arch geometry were comparable in age, weight, height, and BP. Carotid IMT was significantly increased in subjects with Gothic arch geometry compared with control subjects. Subjects with a Romanesque geometry did not have increased IMT compared with control subjects. In addition, IMT was significantly higher in Gothic geometry compared with the Romanesque geometry subgroup. Arterial stiffness was significantly higher in Gothic geometry compared with control subjects and the Romanesque geometry subgroup. Finally, both brachial FMD and GTN were lower in the Gothic geometry group compared with control patients and those with Romanesque geometry (Fig. 3).
When using the continuous variable H/W ratio to describe the aortic geometry, univariate analysis showed positive correlations between H/W ratio and carotid IMT as well as with β stiffness index and Einc. Negative correlations were found between H/W ratio and carotid distensibility, brachial FMD and GTN (Fig. 4).
Using multivariate analysis, the only independent predictor of carotid IMT was the H/W ratio (p < 0.001) when the variables used in the model were age at operation, age at time of the study, or duration of follow-up; weight; height; systolic BP; H/W ratio; and residual stenosis. When the geometry category was used in the model instead of the continuous H/W ratio, Gothic geometry was the only independent predictor of the carotid IMT level (p < 0.001). Using the same models, we found that distensibility, β stiffness index, and Einc were also independently associated with both high H/W ratio and Gothic geometry.
Despite successful repair of CoA in infancy, concerns remain about the long-term follow-up results in these subjects (1–11). There is now a large body of literature that describes a consistent vascular phenotype of the conduit arteries proximal to the CoA repair site (17–20). This phenotype includes increased IMT, increased stiffness, and reduced vasoreactivity of the precoarctation arteries. The determinants of such vascular changes are unclear. Our study may contribute, at least in part, to understanding the complex mechanisms with regard to vascular abnormalities reported late after successful CoA repair. Indeed, we have identified a previously unrecognized factor contributing to the variability in vascular abnormalities seen in post-CoA subjects. The present study is the first to report the role of postoperative aortic arch geometry in vascular remodeling in subjects having undergone successful CoA repair. These novel findings have clinical relevance, as aortic arch geometry appears to identify subjects at high risk of abnormal BP response and adverse vascular remodeling.
Intima-media thickness and stiffness after CoA repair are associated with hypertension during daily life and with increased left ventricular mass, both of which are important predictors for late morbidity and mortality (21). Isolated systolic hypertension, influenced by the stiffness of large arteries, is the most common subtype of hypertension after CoA repair (22) and is a major risk factor for stroke, coronary artery disease, and total cardiovascular mortality (23). Hence, there is a rationale for understanding the mechanisms of increased IMT and stiffness after CoA repair. Abnormal findings in the vascular reactivity measures (FMD and GTN) are not consistently found in subjects who had CoA repair (4,9,10,24,25) and, indeed, there was a wide range of values in these parameters in the current study. Similarly, although long-term follow-up studies show a high prevalence of hypertension, a significant proportion of CoA survivors remain normotensive after 30 years (1,2,22). These findings suggest the existence of variable factors contributing to hypertension and vascular dysfunction, which (to date) are incompletely understood.
Older age at surgery is associated with late hypertension (26), and Heger et al. (27) have also shown that persistent impairment of FMD and GTN is more likely to be present in subjects corrected after 9 years of age. Conversely, de Divitiis et al. (4) found that both endothelial and smooth muscle functional abnormalities may persist despite early repair, and Vogt et al. (28) have documented that vascular abnormalities may even remain unchanged after successful CoA surgery in the neonatal period. In the current study, we found no correlation between age at surgery (median 2 months) and arterial reactivity or stiffness parameters.
Gothic arch, hypertension, and vascular remodeling
We have recently shown that Gothic geometry of the aortic arch, characterized by an elevated H/W ratio, was an independent predictor of resting hypertension after CoA repair (11). In the present study, Gothic arch geometry was independently associated with important alterations of arterial wall thickness (increased IMT) and function (impaired vasoreactivity and increased stiffness) of the pre-CoA arteries. Although not as marked as with Gothic arch structure, there were persistent abnormalities in certain vascular parameters noted in the post-CoA subjects with Romanesque arch geometry compared with control subjects. This suggests that aortic arch geometry does not completely explain the existence of vascular abnormalities after late repair.
Arterial stiffness after CoA repair
To establish the extent of arterial wall rigidity, we assessed 3 separate parameters: arterial distensibility, β stiffness index, and incremental modulus of elasticity. The distensibility reflects the direct relation between pulse pressure and change in dimension and provides information on the elasticity of the artery. However, it is directly dependent on the actual BP and therefore may be a less reliable indicator to establish arterial stiffness, especially in the presence of hypertension. The β stiffness index combines information on intrinsic wall behavior and changes in wall dimensions under a given relative pressure. It is independent of BP in normotensive subjects (29). Finally, the Einc provides direct information about arterial wall rigidity independent of its geometry (30). All 3 parameters suggested enhanced rigidity in association with Gothic arch geometry, and this (like increased IMT in subjects with Gothic arch) may be the consequence of structural abnormalities of the arterial wall. An increase in collagen and a decrease in smooth-muscle cell content as well as elastic fiber degeneration in the aortic segment above the CoA repair site has been documented previously (31) and could account for our observations.
The vascular abnormality we observed with Gothic geometry in this study could be the result of an enhanced early-wave reflection at the site of the acute angle in the aortic arch. This may shorten the timing of the arrival of the reflected wave at the proximal aorta and increase the proportion of central pulse pressure that results from arterial wave reflection. We are currently performing applanation tonometry studies and MRI velocities measurements to investigate this possibility. Recently, Vriend et al. (32) showed that even a mild residual narrowing at the site of CoA repair was a strong and independent predictor of daytime systolic BP and common carotid IMT, which may also be due to enhanced and early-wave reflection at the site of restenosis. Interestingly, these authors raised the question of lowering the threshold for re-intervention on mild re-CoA to improve long-term outcome in these subjects. Our results suggest that abnormal aortic arch shape after repair, especially the peculiar Gothic arch, may be an aggravating factor that accelerates vascular remodeling and dysfunction after CoA repair, even in subjects who are considered successfully repaired on the basis of traditional anatomical criteria.
Although the classification of arch by shape into Gothic, Crenel, and Romanesque forms is relatively simple in many cases, this is a qualitative categorization that may not always be easy to assign. For this reason, we also calculated a more objective and continuous variable, aortic arch H/W ratio, which is greater in subjects with Gothic arch. Furthermore, only a few subjects (n = 9) had Crenel geometry, making it difficult to draw firm conclusions about this group. Whichever classification for arch geometry is used to describe postoperative aortic shape, our findings highlight an important relationship between deformation of the repaired aortic arch and vascular abnormality in the pre-CoA arteries. Further study will be required to see whether Gothic shape predates surgery in any or all cases, or if surgical modifications might be possible to preserve a more Romanesque, rounded shape after CoA repair. Even if surgical modifications cannot be devised successfully, our data identify arch geometry as a risk factor for late arterial abnormalities.
Gothic arch geometry observed after successful CoA repair is a strong predictor of large-artery abnormality in the pre-CoA arterial bed, including vascular thickening, stiffening, and dysfunction. This arterial abnormality is present in normotensive survivors of CoA repair. As arterial dysfunction is also present in some subjects with Romanesque aortic arch shape, the geometry of the aortic arch is not the only contributor to the onset of hypertension after CoA repair. Nevertheless, assessment of aortic arch geometry using MRI and studying vascular function in these subjects may be important in predicting the risk of late hypertension, a major cause of cardiovascular morbidity and mortality late after CoA repair. We would suggest that aortic imaging in the future might therefore consider arch geometry, as well as being a screening tool for detection of discrete re-CoA or late aneurysm formation.
This study was supported by a grant from the Fondation Wyeth pour la Santé de l’Enfant et de l’Adolescent (France) and from the Fédération Française de Cardiologie (France).
- Abbreviations and Acronyms
- blood pressure
- incremental elastic modulus of elasticity
- flow-mediated dilation
- height/width ratio
- intima-media thickness
- magnetic resonance imaging
- Received March 6, 2006.
- Revision received September 22, 2006.
- Accepted October 9, 2006.
- American College of Cardiology Foundation
- Cohen M.,
- Fuster V.,
- Steele P.M.,
- Driscoll D.,
- McGoon D.C.
- Meyer A.A.,
- Joharchi M.S.,
- Kundt G.,
- Schuff-Werner P.,
- Steinhoff G.,
- Kienast W.
- Vriend J.W.,
- de Groot E.,
- Bouma B.J.,
- et al.
- Teragawa H.,
- Kato M.,
- Kurokawa J.,
- Yamagata T.,
- Matsuura H.,
- Chayama K.
- de Divitiis M.,
- Pilla C.,
- Kattenhorn M.,
- et al.
- Ou P.,
- Bonnet D.,
- Auriacombe L.,
- et al.
- Ou P.,
- Mousseaux E.,
- Celermajer D.S.,
- et al.
- National High Blood Pressure Education Program Working Group on Hypertension Control in Children and Adolescents
- Ong C.M.,
- Canter C.E.,
- Gutierrez F.R.,
- Sekarski D.R.,
- Goldring D.R.
- Gardiner H.M.,
- Celermajer D.S.,
- Sorensen K.E.,
- et al.
- Vogt M.,
- Kuhn A.,
- Baumgartner D.,
- et al.
- Laurent S.,
- Girerd X.,
- Mourad J.J.,
- et al.
- Sehested J.,
- Baandrup U.,
- Mikkelsen E.
- Vriend J.W.,
- Zwinderman A.H.,
- de Groot E.,
- Kastelein J.J.,
- Bouma B.J.,
- Mulder B.J.