Author + information
- Received July 5, 2007
- Accepted July 25, 2007
- Published online December 18, 2007.
- Ra K. Han, MD, FRCPC⁎,
- Rebecca C. Gurofsky, BSc⁎,
- Kyong-Jin Lee, MD, FRCPC⁎,
- Anne I. Dipchand, MD, FRCPC⁎,
- William G. Williams, MD, FRCSC†,
- Jeffrey F. Smallhorn, MD, FRCPC⁎ and
- Brian W. McCrindle, MD, MPH, FRCPC⁎,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. Brian W. McCrindle, Division of Cardiology, The Hospital for Sick Children, 555 University Avenue, Toronto, Ontario, Canada M5G 1X8.
Objectives The purpose of this study was to determine trends of growth of left heart structures after intervention for neonatal aortic valve stenosis.
Background The growth potential of left heart structures in neonatal aortic valve stenosis after relief of obstruction might influence risk for subsequent outcomes.
Methods From 1994 to 2004, 53 patients underwent neonatal (≤30 days old) balloon aortic valve dilation. Factors associated with time-related outcomes (death, reintervention, aortic valve replacement) and longitudinal changes in normalized left heart dimensions were sought.
Results The median age at intervention was 3.5 days (range 1 to 30 days). During a median follow-up of 3.2 years ranging up to 10.9 years, there were 31 reinterventions on the aortic valve in 21 (40%) patients and 7 deaths (13%). The presence of moderate or severe left ventricular (LV) endocardial fibroelastosis was the only independent predictor for time-related mortality (hazard ratio 22.1; p = 0.004), and a smaller initial aortic valve annulus z-score was a significant independent predictor for aortic valve replacement (hazard ratio 0.63 per 1-U change; p = 0.007). Aortic valve annulus, aortic sinus, and LV dimension z-scores significantly increased over time, whereas mitral valve z-scores remained below normal. The structure’s initial z-score and concomitant size of other left heart structures were significant independent factors associated with subsequent z-scores.
Conclusions There is potential catch-up growth of the aortic valve and LV over time for neonates after intervention for aortic valve stenosis. However, the continued hypoplasia of the mitral valve warrants further consideration in the long-term management of these patients.
Aortic valve stenosis in neonates is associated with varying degrees of hypoplasia of left heart structures. Fetal echocardiographic studies have demonstrated the progressive development of hypoplasia of left heart structures related to left-sided obstructive lesions (1,2). This hypoplasia adds to increased morbidity and mortality for this patient population (3). In severe cases, the hypoplasia of left heart structures associated with neonatal aortic valve stenosis necessitates univentricular palliation or transplantation (4). In less severe cases, ongoing or progressive hypoplasia or growth failure might contribute to subsequent morbidity, deficits in functional status, and need for further interventions. However, the pattern of growth in left heart structures after the relief of left-sided obstruction is unclear, even in the absence of hypoplasia. A recent study demonstrated potential catch-up growth of the aortic valve and left ventricle (LV) in those with smaller left heart structures (5). However, growth was not the focus of this study, the data provided were descriptive and limited to the aortic valve annulus and left ventricular end-diastolic (LVED) dimension, and multivariable longitudinal data analysis for trends and associated factors was not performed. Hence, we sought to determine the trends in longitudinal changes in left heart dimensions and aortic insufficiency and determine factors associated with clinical outcomes and growth of left heart structures for patients with neonatal balloon aortic valve dilation.
Patients with a diagnosis of aortic valve stenosis who underwent neonatal aortic valve dilation at ≤30 days of age at the Hospital for Sick Children in Toronto, Ontario, Canada, from January 1994 to December 2004 were identified from the cardiac catheterization database. All patients had situs solitus, levocardia, concordant atrioventricular and ventriculoarterial connections, a left aortic arch, and normal venous connections. Patients with initial univentricular palliation were excluded. Patients underwent initial percutaneous transcatheter balloon dilation of the aortic valve with previously described techniques (6). Patient demographic characteristics, clinical status before intervention and at last follow-up, procedural characteristics and hemodynamic status, and outcomes were abstracted from patient records. The study was approved by our institutional Research Ethics Board, and patient confidentiality was maintained.
Initial pre-procedural and follow-up post-procedural echocardiograms and echocardiographic reports were reviewed. Measurements of left heart structures were performed offline with electronic calipers and included the standard published (4) morphologic and functional characteristics. Measurements were normalized for body surface area as z-scores on the basis of published normative data (7). Published normative data was used rather than institution-specific normative values to facilitate comparison with different institutions. Degree of aortic valve insufficiency was assessed both qualitatively with a global assessment based on aortic insufficiency jet width ratio, LV dilation and function, and diastolic flow reversal in the descending aorta and quantitatively with aortic insufficiency jet width to aortic valve annulus ratio in diastole measured offline by a single reviewer. The initial pre-procedural echocardiographic measurements were used to determine the Congenital Heart Surgeons Society (CHSS) score with the published regression equation for critical aortic stenosis (4) and the Rhodes score (3).
Data are presented as frequencies, means with standard deviations (SDs), or median with minimum and maximum, as appropriate. Time-related survival, time to reintervention, and time to aortic valve replacement were modeled with nonparametric Kaplan-Meier estimates. Factors associated with these outcomes were sought from demographic characteristics, medical status at presentation, and echocardiography measurements with Cox proportionate hazard regression. Variables were initially included in a stepwise model (p < 0.05 to enter), and those selected were included in a multivariable model through backward selection to obtain a final model for each event.
Changes in the z-scores of left heart structure dimensions over time after initial intervention but before reintervention were modeled with linear regression adjusted for repeated measures through an autoregressive covariance structure. Associations between changes in z-scores over time and demographic characteristics, echocardiography measurements at presentation, and at follow-up and the degree of aortic and mitral valve stenosis and insufficiency were initially tested in a stepwise regression model (p < 0.25 to enter); backward selection was used to obtain a final multivariable model. Results from these models are given as parameter estimates (PE) with standard error that correspond to the average increase in outcome measure for each increase of 1 measurement unit (or otherwise indicated) in predictor measurement. All data analyses were performed with SAS statistical software (version 9.1, SAS Institute Inc., Cary, North Carolina).
Between January 1994 and December 2004, 53 patients underwent neonatal balloon aortic valve dilation. Fetal diagnosis of isolated aortic valve stenosis was made in 6 (11%) patients. The majority (81%) of patients were male. There were 2 (4%) pre-term neonates born at 34 and 35 weeks’ gestational age.
The median age at presentation was 2 days (range 0 to 30 days). Median weight was 3.4 kg (range 2.0 to 4.5 kg), median length was 50 cm (range 44 to 57 cm), and median body surface area was 0.22 m2(range 0.16 to 0.26 m2). Although 26 (49%) patients presented with a fetal diagnosis or an asymptomatic murmur, 27 (51%) presented with symptoms of low cardiac output or heart failure. Before the initial intervention, prostaglandin E1 infusion was initiated in 33 (62%) patients, 26 (49%) were mechanically ventilated for symptoms or for hospital transfer, and 15 (28%) received inotropic medications.
Initial anatomic features on echocardiogram before the initial intervention are listed in Table 1.Decreased LV systolic function (ejection fraction <55%) was noted in 25 (47%) patients. Table 2lists the z-scores of left heart structures at presentation. The initial mitral valve dimensions, LV dimension and length, aortic valve annulus, aortic root at the sinus, and sinotubular junction were generally smaller than normal, whereas the ascending aorta was larger than normal. The median CHSS score was 0 (range −46 to 55). Only 28 (53%) patients had a negative CHSS score, which predicts a survival benefit with biventricular repair, whereas 25 (47%) patients had a positive CHSS score indicating that they would have had better predicted survival with univentricular palliation rather than a biventricular repair. The median Rhodes score was 0.28 (range −0.98 to 1.39). With a score of <−0.35 being predictive of death after a 2-ventricle repair, the Rhodes score predicted death in 8 (15%) patients. Patients in the more recent era (years 2000 to 2004) had more favorable median CHSS and Rhodes scores at −2 and 0.15, respectively, compared with patients in the earlier era (years 1994 to 1999) whose median CHSS and Rhodes scores were 6 and −1.80, respectively. However, this difference was not statistically significant.
The median age at initial intervention was 3.5 days (range 1 to 30 days). The median aortic valve annulus size by angiography and echocardiogram was 6.8 mm (range 4.5 to 8.6 mm). The median size of the largest balloon used was 7 mm (range 5 to 8 mm), giving a median maximum balloon-to-annulus ratio of 1.0 (range 0.9 to 1.3). The approach used for balloon dilation was antegrade in 40 (75%) and retrograde in 13 (25%) patients.
Before initial balloon aortic valve dilation, the mean LV and aorta systolic pressures by pullback or simultaneous recording measured 115 ± 27 mm Hg and 57 ± 11 mm Hg, respectively, giving a mean peak-to-peak gradient of 59 ± 27 mm Hg. After the procedure, the mean LV and aortic systolic pressures by pullback or simultaneous recording measured 78 ± 16 mm Hg and 61 ± 15 mm Hg, respectively, giving a mean peak-to-peak gradient of 17 ± 10 mm Hg. The overall mean reduction in peak-to-peak gradient was 69 ± 20%.
Aortic valve insufficiency
Follow-up echocardiograms demonstrated aortic valve insufficiency in the majority of patients. Although 9 (19%) patients did not have any aortic valve insufficiency immediately after intervention, there was qualitatively trivial-to-mild insufficiency in 24 (50%), moderate in 14 (29%), and severe in 1 (2%) patient. The mean aortic valve insufficiency to aortic valve annulus jet width ratio measured 33 ± 15%. However, there was no significant change in the degree of aortic valve insufficiency over time (Fig. 1),as analyzed with general linear modeling for longitudinal data.
There were 7 (13%) early deaths during the initial hospital stay with no late deaths. Hence, survival was 86% at 1 year (95% confidence interval [CI] 76% to 96%), which was unchanged at 5 years and 10 years after intervention (Fig. 2).The causes of death were: severe aortic valve regurgitation with poor LV function in 4 patients, 2 of whom underwent surgical reintervention with urgent Ross-Konno procedures before death, and 1 patient each who had residual aortic valve stenosis and underwent reintervention with the Ross-Konno procedure followed by aortic valve replacement with homograft before death; poor ventricular function with high end-diastolic pressure and no residual aortic valve stenosis resulting in multiorgan failure; and fatal intracranial hemorrhage during thrombolytic therapy for arterial thrombosis.
The median duration of follow-up was 3.2 years (range 5 days to 10.9 years). There were 31 subsequent reinterventions in 21 (40%) patients during the follow-up period, as shown in Figure 3.The indications for reintervention were aortic valve stenosis for 20 procedures, aortic valve regurgitation for 8 procedures, and combined aortic valve stenosis and regurgitation for 3 procedures. The median age at the first reintervention was 0.7 years (range 1 day to 7.6 years). The time-related freedom from reintervention was 68% at 1 year (95% CI 52% to 79%), 56% at 5 years (95% CI 39% to 66%), and 33% at 10 years (95% CI 12% to 55%) (Fig. 4).
There were 11 aortic valve replacement procedures in 10 (19%) patients at a median age of 0.6 years (range 1 day to 9.0 years). The indications for aortic valve replacement were aortic valve stenosis for 3 procedures, aortic valve regurgitation for 6 procedures, and combined aortic valve stenosis and regurgitation for 2 procedures. All 3 patients who underwent aortic valve replacement in the neonatal period died before hospital discharge. The freedom from aortic valve replacement was 87% at 1 year (95% CI 81% to 93%), 75% at 5 years (95% CI 61% to 89%), and 59% at 10 years (95% CI 36% to 81%) (Fig. 5).
Factors associated with clinical outcomes
Factors associated with time-related mortality, reintervention, and aortic valve replacement are shown in Table 3.
Patients who died had a mean CHSS score of 31 ± 20 and a mean Rhodes score of −0.24 ± 0.43, compared with the patients who survived, who had a mean CHSS score of −2 ± 19 and a mean Rhodes score of 0.31 ± 0.56. However, moderate or severe LV endocardial fibroelastosis was the only significant independent factor associated with time-related mortality from multivariable analysis, and after controlling for this variable, no other variable was significant.
Trends in longitudinal change in left heart dimensions
Follow-up echocardiograms after initial intervention were available for 48 patients for review. The length of time from initial intervention to an end-state (i.e., death, reintervention, or last follow-up) ranged from 4 days to 9.8 years. There was a median of 2 echocardiograms/patient (range 1 to 14) from this time period. The z-scores of measurements of left heart structures after the initial intervention until an end-state was reached were plotted over time for each individual patient, and a fitted line representing the overall trend of all patients was derived. The trends in longitudinal change in left heart dimension are shown in Figure 6.
Aortic valve annulus, aortic sinus, and LVED diameter z-scores were significantly associated with time (p < 0.001), with a rapid increase during the first year after initial intervention followed by a plateau or slowing in the rate of increase. There was no significant trend of change with time in the z-score of LV endocardial length. The mitral valve lateral dimensions were also significantly related to time (p < 0.05). However, the mitral valve dimension z-scores decreased in the first year after intervention before becoming fairly constant below normal, although remaining within 2 SDs of predicted normal and without clinically important gradients.
Independent factors associated with z-scores of left heart structures during follow-up
Independent factors associated with z-scores of left heart structures during follow-up were determined with general linear modeling for repeated measures, and factors that influenced time trends (interactions) were also sought. Results are reported as PE from the multivariable model, which represent the amount by which a unit change in the factor influences the z-score during follow-up, adjusted for all other factors significant in the model. Independent factors significantly associated with a larger aortic annulus z-score during follow-up were a larger initial aortic valve annulus z-score (PE 0.33 ± 0.07, p < 0.001), a larger aortic sinus z-score during follow-up (PE 0.49 ± 0.07, p < 0.001), and longer time from initial intervention (PE 0.10 ± 0.03/year, p = 0.005). Independent factors associated with a larger aortic sinus z-score during follow-up were a larger initial aortic sinus z-score (PE 0.31 ± 0.10, p = 0.005) and a lower CHSS score (favoring a biventricular repair, −0.02 ± 0.01, p < 0.05) as well as a larger LVED diameter z-score (PE 0.10 ± 0.05, p = 0.04) and a larger aortic valve annulus z-score (PE 0.39 ± 0.06, p < 0.001) during follow-up and longer time from initial intervention (PE 0.15 ± 0.03/year, p < 0.001).
Independent factors significantly associated with a larger LV endocardial length z-score during follow-up were a larger initial LV endocardial length z-score (PE 0.30 ± 0.07, p = 0.001), a larger mitral valve lateral dimension z-score during follow-up (PE 0.20 ± 0.06, p = 0.001), and a higher echocardiographic peak instantaneous LV outflow tract gradient during follow-up (PE 0.01 ± 0.003, p = 0.04). Independent factors associated with a larger LVED diameter z-score during follow-up were a larger initial LVED diameter z-score (PE 0.23 ± 0.08, p = 0.006), a larger mitral valve lateral dimension z-score during follow-up (PE 0.31 ± 0.09, p = 0.001), a larger aortic sinus z-score during follow-up (PE 0.23 ± 0.09, p = 0.008), and a longer time from initial intervention (PE 0.12 ± 0.05/year, p = 0.02).
Independent factors associated with a larger anteroposterior dimension z-score during follow-up were a larger initial mitral valve anteroposterior dimension z-score (PE 0.15 ± 0.07, p < 0.05), a larger LVED diameter z-score during follow-up (PE 0.26 ± 0.05, p < 0.001), and less time from initial intervention (PE −0.09 ± 0.03/year, p = 0.001). Independent factors associated with a larger lateral dimension z-score during follow-up were a larger LV endocardial length z-score (PE 0.25 ± 0.09, p = 0.004), a larger LVED diameter z-score during follow-up (PE 0.20 ± 0.05, p < 0.001), and less time from initial intervention (PE −0.09 ± 0.03/year, p = 0.001).
There was significant interaction between the initial z-scores and the trend of change with time. Patients with lower initial z-scores had lower z-scores during follow-up compared with those with higher initial z-scores. However, those with lower initial z-scores demonstrated greater potential catch-up growth in the first year, as illustrated in Figure 7.
Age at initial intervention, gender, antenatal diagnosis, Rhodes score, mitral or aortic valve morphology, and severity of endocardial fibroelastosis were not found to be significantly associated with longitudinal change in any of the left heart dimensions. The degree of aortic valve insufficiency immediately after initial intervention and at follow-up by qualitative assessment or quantitative jet width ratio was also not significantly associated with longitudinal change in left heart dimension. Although LVED dimension z-scores exceeded predicted normal values during follow-up, the degree of LV dilation was similar regardless of the degree of aortic valve insufficiency immediately after intervention, as illustrated in Figure 8.
Hypoplasia of left heart structures associated with neonatal aortic valve stenosis contributes importantly to increased morbidity and mortality in this patient population. We examined a contemporary group of patients undergoing neonatal intervention for aortic valve stenosis and evaluated time-related clinical outcomes and longitudinal trends in left heart dimensions and factors associated with these outcomes and measurements over time.
Neonatal intervention for aortic valve stenosis is associated with considerable morbidity and mortality (6,8). Associated hypoplasia of left heart structures is a well-established risk factor for mortality (3,4) in this population. Consistent with previous studies, our findings highlight the important impact of associated left heart hypoplasia and endocardial fibroelastosis on survival after biventricular repair. These anatomic factors also contribute to the high rate of reintervention and aortic valve replacement in this patient population.
Growth of left heart structures
There is an early phase in the first year after initial intervention where there are rapid increases in aortic valve and LV diameter z-scores. This is consistent with previous studies of neonates with coarctation of the aorta or aortic valve stenosis and hypoplastic LV (5,9–12). This might be affected by changes in the geometry of the LV after the relief of obstruction but might also reflect volume load in the presence of aortic valve regurgitation. In fact, we found that LVED dimension z-scores exceeded normal predicted values in the first year after initial intervention, although we could not find an association with the degree of aortic valve stenosis. After the initial period of catch-up growth or dilation in the first year, however, the aortic valve and LV seem to have a normal rate of growth, resulting in a plateau in the mean z-scores near or above normal.
In contrast, the mitral valve dimension z-score in our study decreased in the first year after initial intervention and remained below normal thereafter although within 2 SDs of predicted normal. There were no interventions in our series for mitral valve stenosis during the intermediate follow-up period; hence, the functional impact and long-term implications of this continued mitral valve hypoplasia remain unclear.
The highest rate of growth in heart structures occurs in the first year of life (13) and might represent a time for potential catch-up growth. Humpl et al. (14) in their study of pulmonary atresia and intact ventricular septum found that the growth of hypoplastic right heart structures was proportional to the initial values. This was also our finding, with higher initial z-scores being predictive of higher z-scores during follow-up. This implies that patients with normal or near normal size structure maintain their normality during follow-up. There was also size concordance between different left heart structures, so that if a patient had a higher z-score for one left heart structure, they were more likely to have higher z-scores for the other left heart structures as well. Although the degree of aortic valve insufficiency by both qualitative and quantitative assessment was not statistically significantly associated with higher LVED z-scores over time in our study, McElhinney et al. (5) have found that patients with post-dilation aortic valve insufficiency had a more rapid increase in LVED z-scores and higher LVED z-scores in follow-up. In contrast, the peak instantaneous gradient across the LV outflow tract was a significant independent predictor for higher LV length and aortic valve annulus z-scores in our study. This then raises the question as to how residual lesions, either regurgitation or stenosis, influence the growth of heart structures.
This potential for growth of left heart structures might have implications for patients with aortic valve stenosis and “borderline left ventricles.” Although the left heart appears small and inadequate on initial assessment, there might be immediate expansion or growth after the relief of obstruction and change in loading conditions, followed by potential catch-up growth sufficient to support the systemic circulation. Many studies have attempted to define the lower limits of LV hypoplasia that are adequate to support a systemic circulation (3,15), but these lower limits continue to be challenged (16,17). The patterns in longitudinal changes in left heart structures might suggest that if patients survive the potentially difficult early post-intervention period, they might have sufficient dilation or growth to support a biventricular circulation. However, the cost of achieving a biventricular circulation must be weighed against the potentially higher early mortality and reintervention rates that might make single ventricle palliation preferable in selected cases.
Controversy still remains as to whether fetal intervention for aortic valve stenosis would allow for sufficient catch-up growth and hence prevention or reversal of the development of hypoplastic left syndrome in utero. Tworetzky et al. (18) presented their early results for balloon dilation of severe aortic stenosis in the fetus and found that there was an increase in mitral valve, aortic valve, and ascending aorta diameters as well as LV length, for the fetuses that had a successful intervention compared with those that did not. Although still in the early stages of development and application, the potential catch-up growth in the neonatal and fetal hearts after the relief of obstruction presents a hopeful area for intervention and prevention.
The limitations of this study include its retrospective design, with the lack of clear pre-defined indications for reintervention or aortic valve replacement, and the variable length and frequency of monitoring, including patients undergoing part of their follow-up assessments outside of our institution. Hence, patients requiring reintervention and aortic valve replacement might be over-represented in our sample, because these patients would be more likely to continue their follow-up assessments at our tertiary care institution. In addition, with our contemporary patient population, we have intermediate follow-up data available, but further long-term longitudinal research is required particularly as it relates to any potential areas of concern, such as the implications for the continued hypoplasia of the mitral valve.
Although neonates with aortic valve stenosis might have associated small left heart structures, there is potential catch-up growth of the aortic valve and growth or dilation of the LV over time, with no evidence for growth failure. This potential for catch-up growth after relief of left-sided obstruction has implications for the management of patients with borderline LVs. However, persistent hypoplasia of the mitral valve, when present, warrants further consideration in the long-term follow-up and management for this patient population.
- Abbreviations and Acronyms
- Congenital Heart Surgeons Society
- confidence interval
- left ventricle/ventricular
- left ventricular end-diastolic
- parameter estimates
- Received July 5, 2007.
- Accepted July 25, 2007.
- American College of Cardiology Foundation
- Hornberger L.K.,
- Sanders S.P.,
- Rein A.J.J.T.,
- Spevak P.J.,
- Parness I.A.,
- Colan S.D.
- Makikallio K.,
- McElhinney D.B.,
- Levine J.C.,
- et al.
- Rhodes L.A.,
- Colan S.D.,
- Perry S.B.,
- Jonas R.A.,
- Sanders S.P.
- McElhinney D.B.,
- Lock J.E.,
- Keane J.F.,
- Moran A.M.,
- Colan S.D.
- Serraf A.,
- Piot J.D.,
- Bonnet N.,
- et al.
- Humpl T.,
- Soderberg B.,
- McCrindle B.W.,
- et al.
- Parsons M.K.,
- Moreau G.A.,
- Graham T.P. Jr..,
- Johns J.A.,
- Boucek R.J. Jr.
- Tworetzky W.,
- Wilkins-Haug L.,
- Jennings R.W.,
- et al.