Aortic Valve Morphology Correlates With Left Ventricular Systolic Function and Outcome in Children With Congenital Aortic Stenosis Prior to Balloon Aortic Valvuloplasty

Abstract: Objectives. We sought to determine the relationship between aortic valve morphology and left ventricular (LV) systolic function in children with aortic stenosis (AS) prior to balloon aortic valvuloplasty (BAV). Background. Both aortic valve morphology and LV systolic function have been linked with outcomes in children with congenital AS undergoing BAV. The relationship between aortic valve morphology and LV function is poorly defined despite their importance in regard to outcomes. Methods. We performed a retrospective multicenter cohort study of 89 AS patients who underwent BAV between 2007-2013. Pre-BAV echocardiograms were analyzed for: aortic valve opening (AVO); aortic valve type (true bicuspid, functionally bicuspid, or unicuspid); maximal raphe length; aortic valve leaflet symmetry; and valve angle of excursion. The primary endpoint was low function, defined as LV shortening fraction (LVSF) <28%. Results. Median patient age was 0.17 years (interquartile range [IQR], 0.10-10.74 years) and the median aortic valve mean gradient was 47.00 mm Hg (IQR, 36.75-56.00 mm Hg). Multivariate analysis demonstrated that low AVO (P=.03) was associated with reduced LV function, independent of age or aortic valve gradient (R² = .652). Bicuspid aortic valve (P=.07) was associated with improved LV function compared with functionally unicuspid aortic valve. Low AVO <0.10 was associated with higher adverse outcome. Conclusion. LV systolic function is most significantly influenced by degree of valve stenosis. Qualitative aspects such as valve type may also affect LV systolic function. Further study may elucidate whether aortic valve morphology or LV function is the principal predictor of response to BAV and of late outcomes after BAV.

J INVASIVE CARDIOL 2016;28(9):381-388

Key words: balloon, valvuloplasty, bicuspid, unicuspid

Balloon aortic valvuloplasty (BAV) is the treatment of choice for children with congenital aortic stenosis (AS). Advancements in technique and balloon sizing have made BAV a safe and effective procedure that successfully reduces the gradient in most patients^1-6 and delays the need for valve replacement, with increase in aortic insufficiency (AI)^4-7 being the only common complication. Despite this, outcomes remain poor in neonates,^1,7-9 those with residual post-BAV gradients,⁷ and children with reduced left ventricular (LV) systolic function.^7,10,11 Specifically for the latter subgroup, multiple cohort studies have shown that the rates of reintervention on the aortic valve, development of significant AI, aortic valve surgical replacement, and death are greater in children with reduced LV systolic function.^1,7,8,10,12 No study has yet been able to determine the mechanism behind these findings, and an adequate model to risk stratify who would benefit the most from BAV is lacking.

A number of studies have demonstrated a relationship between aortic valve morphology and outcomes following BAV. Factors such as leaflet fusion pattern, valve leaflet thickness, and number of leaflets have been shown to have an impact on the achievable results following BAV.¹³ Other studies have indicated that valve morphology is associated with mid-term outcomes following BAV.^13-15 In fact, aortic valve morphology appears to influence outcomes of surgical valvotomy/valve repair as well.^15,16 A complex interplay among LV systolic function, aortic valve morphology, and early outcomes following BAV has been most notably evident among neonates with AS, in whom LV systolic function is most commonly diminished, valve stenosis is most severe, and who are at highest risk of repeat intervention post BAV.^9,10,17 Interestingly, while pre-BAV LV systolic function and aortic valve morphology both appear to influence outcomes of BAV and surgical valvotomy, the relationship between these two factors has been heretofore inadequately examined. Furthermore, previous studies have been limited by qualitative descriptors of valve morphology such as degree of thickening or aortic valve type in lieu of more precise measures reflecting degree of valve stenosis and overall obstructive pathology, such as aortic valve opening (AVO) area. The current study attempts to determine the relationship between aortic valve morphology and LV systolic function in children with congenital AS by using both previous metrics of aortic valve morphology and novel quantitative variables aimed at estimating the significance of aortic valve stenotic pathology.

Methods

Patient selection. We queried institutional records for all patients undergoing BAV between January 1, 2007 and January 1, 2014 at Children’s Healthcare of Atlanta and Cincinnati Children’s Hospital Medical Center. Patients who underwent BAV for isolated valvular AS were included. Patients who did not have preoperative echocardiograms within 10 days of the procedure were excluded from the study. All pre-BAV echocardiograms within 10 days of BAV were reviewed, and those patients where poor echocardiogram quality precluded assessment of aortic valve morphology were also excluded. Poor-quality echocardiogram was defined as one in which no determination of aortic valve morphology was possible by either parasternal short-axis or parasternal long-axis imaging. 

For each patient included in the study, basic demographic information including age, neonate status, weight, height, and body surface area were recorded. A neonate was defined as any patient ≤30 days in age. The presence of other cardiac lesions (eg, coarctation of the aorta, mitral stenosis, etc) was noted. History of any previous aortic valve interventions, such as previous BAV or valve surgery, was also noted.

Echocardiogram. Pre-BAV echocardiograms within 10 days of valvuloplasty were read by at least two independent reviewers (KG, RS, SL, CP) and aortic valves were characterized according to basic morphologic pattern: (1) functionally unicuspid valves, with fusion of two commissures; (2) functionally bicuspid valves, with fusion of one commissure; and (3) true bicuspid valves, with no fusion of commissures, and with only two leaflets (Figure 1). The AVO area was measured in systole, and it was indexed to the total area within the aortic annulus in order to account for differences in aortic valve sizes. Additional morphologic characteristics were also noted (Figures 2A-2C). Raphe (fusion) presence, position (between right-left, right non-coronary, left non-coronary leaflets), and relative length (corrected for body surface area and aortic valve annulus diameter) were noted for each patient. All two-dimensional areas were measured through direct tracing in planimetry. Valve symmetry was determined by dividing the area of the smallest leaflet by the area of the largest leaflet (including leaflets created through fusion), as previously described.¹⁸ 

Angle of excursion of the anterior aortic valve leaflet was noted, and was defined as the angle between the plane of the aortic valve hingepoints and the line formed by the point of attachment for the anterior leaflet and its tip Figure 2B. Diameters of the aortic valve annulus, aortic root, sinotubular junction, and ascending aorta were also measured (Figure 2C). All linear measurements in centimeters were standardized by indexing to body surface area to correct for patient size. 

Left ventricular end-diastolic dimension (LVEDD) Z-scores, left ventricular shortening fraction (LVSF), and mean aortic valve gradient before BAV were recorded from the original echocardiogram report. 

Statistical analysis. Data are presented as number (%) or median with interquartile range (IQR) where appropriate. The primary endpoint for this study was low LV systolic function, defined as LVSF <28%. Patients with low LVSF (<28%) and normal LVSF (≥28%) were compared using univariate analysis of demographic, hemodynamic, and aortic valve morphologic factors. Student’s t-test was used to compare normally distributed continuous variables, while Kruskal-Wallis and Mann-Whitney U-tests were utilized to compare medians for non-normally distributed variables. Pearson’s χ² test was used for categorical variables. Variables with P<.10 were included in the multivariate binomial logistical regression after taking into account co-linearity. Survival and freedom from repeat intervention such as repeat valvuloplasty, surgical valvotomy, or aortic valve replacement were analyzed with Kaplan-Meier survival curves. A P-value <.05 was deemed statistically significant. The study protocol was approved by the institutional review boards of both participating institutions with appropriate data use agreements in place. 

Results

The study cohort consisted of 89 patients; 34 (38%) were neonates and 60 (68%) were male. Approximately 30% of prescreening candidates were excluded for poor-quality or missing echocardiographic data. The baseline characteristics are summarized in Table 1. The median age was 0.17 years (IQR, 0.10-10.74 years). Valve morphology type was bicuspid in 50 patients (56%); 41 of these patients had functionally bicuspid and 9 patients had true bicuspid aortic valves. The remaining 39 patients (44%) had functionally unicuspid valves. The majority of the patients in the study (72; 81%) had no previous interventions on their aortic valve at time of study. Among the 17 patients (19%) who had previous interventions on their aortic valves, 15 had 1 prior BAV, 3 patients had 1 prior surgical valvotomy, and 3 patients had >1 prior intervention. The pre-BAV mean echo gradient was 47.0 mm Hg (IQR, 36.7-56.0 mm Hg), which was reduced post BAV to a median of 24.0 mm Hg (IQR, 18.0-31.5 mm Hg) (P<.01). 

Among neonates, the aortic valve morphology differed from the rest of the cohort (Table 2). Neonates had more severely stenotic aortic valves represented by smaller AVO areas (P<.01) and longer raphe lengths (P<.01). Neonates with AS were found to have predominantly functionally unicuspid aortic valve morphology, while none had true bicuspid valves. This is in contrast to the non-neonate cohort, which was noted to have all three types of valve morphologic pattern (P <.01). Neonates likewise had lower LVEDD Z-scores (P=.02), more dilated aortic roots (P<.01), sinotubular junctions (P<.01), and ascending aorta diameters (P<.01). There was no difference between neonates and non-neonates with respect to aortic valve gradient before BAV or immediately post BAV (P=.83 and P=.50, respectively). 

Among the entire cohort, 25 patients met the primary endpoint of LVSF <28%. The remaining 64 had normal systolic function (≥28%). A univariate analysis demonstrated that low LV systolic function was associated with several demographic, morphological, and functional characteristics (Table 3). Patients with low LV systolic function were younger, with a median age of 0.05 years (IQR, 0.01-0.12 years) vs 0.79 years (IQR, 0.03-12.41 years) (P<.01). Indeed, the low LV systolic function group was comprised of 60.0% neonates, while only 29.7% of the normal LV systolic function group were neonates (P<.01). Across the entire cohort, patients with low LV systolic function also had lower mean gradients (P<.01). 

There were strong associations between aortic valve morphology and low LV systolic function. The low LV systolic function group had severely stenotic aortic valves, underscored by smaller AVOs (0.09 cm [IQR, 0.07-0.16  cm] compared to 0.22 cm [IQR, 0.14-0.31 cm]; P<.01) and longer raphe length (1.14 cm [IQR, 0.95-1.33 cm] compared to 0.87 cm [IQR, 0.65-1.12 cm]; P<.01). Regression demonstrates significant inverse linear correlations between severity of stenosis and LVSF (Figure 4). Valve type was also significant. Patients with low LV systolic function had predominantly functionally unicuspid valves (76%); in contrast, the normal LV systolic function group was comprised mostly of functionally bicuspid valves (56.3%), with only 31.2% functionally unicuspid valves (P<.01). Overall, patients with functionally unicuspid valves had lower LVSF than those who had bicuspid valves (Figure 3). Even the neonatal subset demonstrated this; neonates with functionally unicuspid valves had lower LVSF than non-neonates with unicuspid aortic valves (P=.05) (Figure 3). Non-valve cardiac lesions such as dilated LV and larger aortic root diameter, ascending aorta diameter, and sinotubular junction diameter were also associated with low LV systolic function (Table 3).

Other morphologic valve characteristics, including valve fusion pattern, aortic valve asymmetry, and angle of excursion failed to statistically associate with LV systolic function when function was taken as a continuous variable or as a binary (low vs normal LVSF). The presence of additional cardiac lesions also failed to correlate with low LVSF (presence of any other heart disease [P=.40], coarctation of the aorta [P=.89], mitral stenosis [P=.91], ventricular septal defect [P=.56], and Shone’s syndrome [P=.84]). 

A multivariate analysis was conducted to determine impact of valve morphologic characteristics, when controlled for patient age, on LV systolic function (Table 4). The overall model was statistically significant (P<.01; R²=0.652). It correctly classified 86.9% of all cases, identifying 13/20 patients with low LVSF (sensitivity, 65.0%) and 60/64 patients with normal LVSF (specificity, 93.7%). Larger AVO (P=.03) and mean echo gradient decreased the risk of low LV systolic function (P<.01), and larger LVEDD Z-score increased the risk of low LVSF (P<.01). Patients with bicuspid aortic valves had lower likelihood of low LV systolic function (odds ratio [OR], 0.21;  95% confidence interval [CI], 0.04-1.12; P=.07). Finally, neonate status was not significant (P=.14) in the multivariate analysis, suggesting that valve morphology, not patient age at BAV, is a determinant of preintervention LV systolic function. 

Freedom from death and repeat intervention through repeat valvuloplasty, surgical valvotomy, or aortic valve replacement were analyzed using Kaplan-Meier curves from single institution data (Figures 5 and 6). The results demonstrated that those with more severe aortic valve stenosis had less freedom from adverse events. For those with AVO <0.10 cm, the mean time free from adverse events was 18.49 months (95% CI, 0.00-37.84) compared to those with less severe stenosis at 51.82 months (95% CI, 37.56-67.09). Furthermore, the relationship between freedom from adverse events and LVSF <28% was also examined. Those with reduced LVSF also had lower mean time free from adverse events of 27.63 months (95% CI, 9.49-45.77) compared to those with normal LVSF at 56.96 months (95% CI, 40.35-73.56). 

Discussion

The impact of aortic valve morphology on outcomes following BAV in children has long been recognized. Several studies have shown that substrate is a contributing factor to procedural success of both BAV and surgical valvotomy. Sholler et al recognized that thickened valve leaflets and valves with fewer individual mobile leaflets were less likely to respond positively to BAV.¹³ Solymar et al showed that bicuspid valves are more likely to tear among commissures and are less likely to develop AI post BAV.¹⁹ Other studies have shown that valve morphology is associated with long-term outcomes, with functionally bicuspid valves performing better.^1,11,14 These studies have not attempted to quantify degree of aortic valve stenosis or opening area in their analysis.

The current study demonstrates a significant relationship between aortic valve morphology and LV systolic function, a metric already shown in numerous studies to be associated with long-term outcomes. Patients in our study with functionally bicuspid valves were more likely to have normal LV systolic function compared with patients with unicuspid valves; this may partly explain the findings in the previous articles demonstrating improved mid-term outcomes in patients with preserved LV systolic function.^10,12 Perhaps more importantly, our study has shown that aortic valve morphology can be described through quantitative variables such as degree of aortic valve stenosis, a variable in our study that has most strongly correlated with intermediate outcomes and LV function, suggesting that degree of valve stenosis may be the underlying pathology responsible for the outcomes above, and that valve type – which has previously been the focus of studies on aortic valve morphology – may play a secondary role.

The importance of LV systolic function in outcomes of children with AS has been demonstrated in multiple studies. Yet, the interaction of factors such as systolic function and valve morphology are complex, and their respective impact on outcomes following balloon or surgical aortic valvuloplasty have been incompletely explained. A possible explanation suggests that severely stenotic aortic valves provoke pathologic myocardial remodeling through increased afterload,^20,21 thus impairing ventricular contractility and worsening outcomes after valvuloplasty as a result. This relationship is consistent with reports that find worse outcomes following BAV in patients with reduced LV systolic function.^10,12 Our study’s survival analysis is consistent with previous articles, but we have also revealed worse intermediate outcomes in those with low AVO, suggesting that perhaps aortic valve pathology is ultimately responsible for poor outcomes. Future study may clarify whether ultimately morphology or systolic function is the more definitive predictor of outcomes following BAV. 

The current study lends weight to the theory that valve morphology is deterministic for outcomes following valve intervention. We found a significant link between valve morphology and preintervention LV systolic function. Importantly, we found that comprehensive morphologic characterization may be useful in predicting functional valve characteristics as well, which will have clinical significance.

Neonates have also been shown in previous studies to have worse outcomes following BAV, and are more likely to have unicuspid or dysplastic morphology. Our study showed similar findings, with neonates more commonly having functionally unicuspid valves, smaller valve openings, and longer raphes. In the multivariate analysis, however, it was aortic valve morphology exemplified by degree of valve stenosis and unicuspid valve type, not neonatal status, which was associated with low LV systolic function. Given the more severe valve pathology seen in neonates, our study suggests that children with severe stenosis should undergo BAV early as neonates instead of undergoing the procedure later in childhood.

Interestingly, we found that smaller AVOs were more commonly seen with dilated aortic roots and ascending aortas. Previous studies likewise have shown a correlation between AS and aortic root dilation;^22,23 however, none demonstrated the relationship between AVO area and the extent of dilation. The prevailing hypothesis is that a jet of turbulent high-speed flow from the stenotic aortic valve striking the aorta is responsible for the dilation. The results in this study are consistent with this hypothesis, as smaller valve openings should in theory generate higher velocity streams. This is quite an important incidental finding as it sheds light on the importance of aortic valve morphology in yet another significant cardiovascular complication.

One cannot rule out the possibility that a common in utero developmental abnormality affects both the aortic valve and the LV. A follow-up study showing increased worsening of systolic function in valves with smaller openings may distinguish between the two hypotheses.

Study limitations. Limitations of this study include its retrospective nature and its limited patient inclusion criteria to only children undergoing BAV. A broader study including echocardiograms of children with congenital AS who do not require valvuloplasty would strengthen the conclusions of the paper. The study also does not examine the extent of pathological LV fibrosis or remodeling that may occur as a result of aortic valve stenosis. Echocardiography is able to accurately determine the amount of ventricular hypertrophy, which would be interesting to look at in a follow-up study. Although LV dilation was examined in this study, myocardial remodeling, inflammation, and fibrosis could not be accurately visualized by echocardiogram. Magnetic resonance imaging as a modality has been shown to accurately visualize inflammation and scarring, as well as being able to differentiate ischemic and non-ischemic myocardial damage.²⁴ When magnetic resonance imaging becomes more widespread in the pediatric population, fibrosis, inflammation, and cardiomyopathy can be studied in further detail in children with differing valve morphologies. 

Conclusion

In this study of patients with congenital AS undergoing BAV, the LV systolic function is strongly influenced by aortic valve morphology, most significantly by degree of valve stenosis. Valves with smaller opening areas relative to the aortic annulus tend to have worse systolic function, and functionally unicuspid valves have worse function than bicuspid valves. The multivariate analysis has demonstrated that morphology, and not age, primarily determines outcome. Further study may elucidate whether aortic valve morphology or LV function is the principal predictor of response to BAV and of late outcomes after BAV.

References

1.    Crespo D, Miro J, Vobecky SJ, et al. Experience in a single centre with percutaneous aortic valvoplasty in children, including those with associated cardiovascular lesions. Cardiol Young. 2009;19:372-382.

2.    Demkow M, Ruzyllo W, Ksiezycka E, et al. Long-term follow-up results of balloon valvuloplasty for congenital aortic stenosis: predictors of late outcome. J Invasive Cardiol. 1999;11:220-226.

3.    Moore P, Egito E, Mowrey H, Perry SB, Lock JE, Keane JF. Midterm results of balloon dilation of congenital aortic stenosis: predictors of success. J Am Coll Cardiol. 1996;27:1257-1263.

4.    Sandhu SK, Lloyd TR, Crowley DC, Beekman RH. Effectiveness of balloon valvuloplasty in the young adult with congenital aortic stenosis. Cathet Cardiovasc Diagn. 1995;36:122-127.

5.    Shaddy RE, Boucek MM, Sturtevant JE, Ruttenberg HD, Orsmond GS. Gradient reduction, aortic valve regurgitation and prolapse after balloon aortic valvuloplasty in 32 consecutive patients with congenital aortic stenosis. J Am Coll Cardiol. 1990;16:451-456.

6.    Sandhu SK, Silka MJ, Reller MD. Balloon aortic valvuloplasty for aortic stenosis in neonates, children, and young adults. J Interv Cardiol. 1995;8:477-486.

7.    Maskatia SA, Ing FF, Justino H, et al. Twenty-five year experience with balloon aortic valvuloplasty for congenital aortic stenosis. Am J Cardiol. 2011;108:1024-1028.

8.    Fratz S, Gildein HP, Balling G, et al. Aortic valvuloplasty in pediatric patients substantially postpones the need for aortic valve surgery: a single-center experience of 188 patients after up to 17.5 years of follow-up. Circulation. 2008;117:1201-1206.

9.    McElhinney DB, Lock JE, Keane JF, Moran AM, Colan SD. Left heart growth, function, and reintervention after balloon aortic valvuloplasty for neonatal aortic stenosis. Circulation. 2005;111:451-458.

10.    Petit CJ, Ing FF, Mattamal R, Pignatelli RH, Mullins CE, Justino H. Diminished left ventricular function is associated with poor mid-term outcomes in neonates after balloon aortic valvuloplasty. Catheter Cardiovasc Interv. 2012;80:1190-1199.

11.    Reich O, Tax P, Marek J, et al. Long term results of percutaneous balloon valvoplasty of congenital aortic stenosis: independent predictors of outcome. Heart. 2004;90:70-76.

12.    Agnoletti G, Raisky O, Boudjemline Y, et al. Neonatal surgical aortic commissurotomy: predictors of outcome and long-term results. Ann Thorac Surg. 2006;82:1585-1592.

13.    Sholler GF, Keane JF, Perry SB, Sanders SP, Lock JE. Balloon dilation of congenital aortic valve stenosis. Results and influence of technical and morphological features on outcome. Circulation. 1988;78:351-360.

14.    Maskatia SA, Justino H, Ing FF, Crystal MA, Mattamal RJ, Petit CJ. Aortic valve morphology is associated with outcomes following balloon valvuloplasty for congenital aortic stenosis. Catheter Cardiovasc Interv. 2013;81:90-95.

15.    Bhabra MS, Dhillon R, Bhudia S, et al. Surgical aortic valvotomy in infancy: impact of leaflet morphology on long-term outcomes. Ann Thorac Surg. 2003;76:1412-1416.

16.    Miyamoto T, Sinzobahamvya N, Wetter J, et al. Twenty years experience of surgical aortic valvotomy for critical aortic stenosis in early infancy. Eur J Cardiothorac Surg. 2006;30:35-40.

17.    Han RK, Gurofsky RC, Lee KJ, et al. Outcome and growth potential of left heart structures after neonatal intervention for aortic valve stenosis. J Am Coll Cardiol. 2007;50:2406-2414.

18.    Cognet T, Seguela PE, Thomson E, et al. Assessment of valvular surfaces in children with a congenital bicuspid aortic valve: preliminary three-dimensional echocardiographic study. Arch Cardiovasc Dis. 2013;106:295-302.

19.    Solymar L, Sudow G, Berggren H, Eriksson B. Balloon dilation of stenotic aortic valve in children. An intraoperative study. J Thorac Cardiovasc Surg. 1992;104:1709-1713.

20.    Hirt MN, Sorensen NA, Bartholdt LM, et al. Increased afterload induces pathological cardiac hypertrophy: a new in vitro model. Basic Res Cardiol. 2012;107:307.

21.    Pacileo G, Calabro P, Limongelli G, et al. Left ventricular remodeling, mechanics, and tissue characterization in congenital aortic stenosis. J Am Soc Echocardiogr. 2003;16:214-220.

22.    Keane MG, Wiegers SE, Plappert T, Pochettino A, Bavaria JE, Sutton MG. Bicuspid aortic valves are associated with aortic dilatation out of proportion to coexistent valvular lesions. Circulation. 2000;102:III35-III39.

23.    McElhinney DB, Lacro RV, Gauvreau K, et al. Dilation of the ascending aorta after balloon valvuloplasty for aortic stenosis during infancy and childhood. Am J Cardiol. 2012;110:702-708.

24.    Pattanayak P, Bleumke DA. Tissue characterization of the myocardium: state of the art characterization by magnetic resonance and computed tomography imaging. Radiol Clin North Am. 2015;53:413-423.

From the Sibley Heart Center, Department of Pediatrics, Emory University School of Medicine, Children’s Healthcare of Atlanta, Atlanta, Georgia; and the Heart Institute, Department of Pediatrics, University of Cincinnati School of Medicine, Cincinnati Children’s Hospital Medical Center, Cincinnati, Ohio.

Disclosure: The authors have completed and returned the ICMJE Form for Disclosure of Potential Conflicts of Interest. The authors report no conflicts of interest regarding the content herein.

Manuscript submitted November 23, 2015, provisional acceptance given January 19, 2016, final version accepted April 22, 2016.

Address for correspondence: Christopher J. Petit, MD, Associate Professor of Pediatrics, Emory University School of Medicine, 1405 Clifton Road, Atlanta, GA 30322. Email: cjpetit@emory.edu