- Email: [email protected]
Predictors of Outcome of Arterial Switch Operation for Complex D-Transposition
PEDIATRIC CARDIAC
Predictors of Outcome of Arterial Switch Operation for Complex D-Transposition Danielle Gottlieb, MD, MPH, Marcy L. Schwartz, MD, Kara Bischoff, BA, Kimberlee Gauvreau, ScD, and John E. Mayer, Jr, MD Department of Cardiology and Cardiovascular Surgery, Children’s Hospital Boston, and Harvard Medical School, Boston, Massachusetts
Background. Overall mortality and reoperation risk for the arterial switch operation (ASO) for D-transposition of the great arteries (D-TGA) is low. D-TGA with ventricular septal defect (VSD) and aortic arch obstruction (AAO) is a higher risk subgroup in which we sought risk factors for mortality and reoperation after ASO. Methods. Echocardiograms of 74 patients who underwent ASO, VSD, and arch repair for D-TGA, VSD and AAO were reviewed; the reoperation analysis considered the 65 survivors. Pre-ASO clinical and anatomic characteristics were compared between survivors and nonsurvivors; patients who required (R) and did not require (NR) reoperation. Results. Distal transverse aortic arch (TrAo) z score equal to ⴚ2.5 or less, triscuspid valve z score less than 0, repaired muscular VSD, and circulatory arrest time were significant predictors of mortality. When stratified for circulatory arrest time below 60 minutes, small distal
transverse aortic arch and tricuspid valve remained significant predictors of mortality. Mean aortic annulus size was smaller in R than NR (p ⴝ 0.048). Left coronary artery arising posteriorly was associated with a reoperation hazard ratio of 5.2 (p ⴝ 0.022). Conclusions. Preoperative anatomy was associated with death and reoperation post-ASO. Small TrAo and TV were risk factors for mortality in univariate analysis, and remained significant in the subset of patients with short circulatory arrest times, suggesting that even when controlling for technical factors, anatomic risk factors predict mortality. Small aortic annulus and posterior left circumflex artery origin were associated with reoperation. Patients with D-TGA, VSD, and AAO constitute a higher risk group, which includes patients who may be marginal candidates for two-ventricle repair. (Ann Thorac Surg 2008;85:1698 –703) © 2008 by The Society of Thoracic Surgeons
A
s survival from D-transposition of the great arteries (D-TGA) has improved since the introduction of the arterial switch operation (ASO), indications for ASO have been extended to include more complex anatomic subsets [1]. Though ASO seems to be the operation of choice for patients with complex D-TGA, differential survival for patients with complex TGA has been demonstrated [2– 4]. Further, aortic arch obstruction (AAO) has been identified in multiple prior studies as a risk factor for less favorable outcome. However, the additional risk of AAO in the setting of D-TGA is incompletely understood as most analyses are limited by small numbers of patients with this additional anatomic complexity [2, 4 –7]. Previous risk factor analyses have identified “complex D-TGA” and ventricular septal defect (VSD) as risk factors for mortality, but additional detail is needed for accurate preoperative risk stratification, operative planning, and family counseling. Central to surgical planning decisions is the ability to distinguish between patients who have uncorrectable anatomy and those whose anatomy provides technical difficulty and consequent morbidity associated with longer cardiopul-
monary bypass and circulatory arrest times. We sought to better understand whether increased mortality in patients with D-TGA, VSD, and AAO can be predicted by anatomy, or whether anatomy is a risk factor for an intermediate variable, operative difficulty. We undertook this analysis in order to determine whether anatomic features contribute to the observed increased mortality in patients with D-TGA, VSD, and AAO. Multiple investigators have also considered whether preoperative anatomy is associated with the need for reoperation, but most studies have been performed in heterogeneous populations, with complex patients representing a small subgroup, and studies have therefore been underpowered to find a difference between simple and complex variants of D-TGA. Risk factors have varied between studies and have not been thoroughly examined in this complex subset of D-TGA, VSD, and AAO patients [1, 5]. We sought to define risk factors for right ventricular outflow tract obstruction (RVOTO) requiring reoperation and to determine whether risk is dictated by pre-ASO anatomy.
Accepted for publication Jan 23, 2008.
Patients and Methods
Presented at the Forty-third Annual Meeting of The Society of Thoracic Surgeons, San Diego, CA, Jan 29 –31, 2007.
Patients
Address correspondence to Dr Gottlieb, 300 Longwood Avenue, Boston, MA 02115; e-mail: [email protected].
Patients were identified by searching the Children’s Hospital cardiovascular program database for patients
© 2008 by The Society of Thoracic Surgeons Published by Elsevier Inc
0003-4975/08/$34.00 doi:10.1016/j.athoracsur.2008.01.075
who underwent ASO, VSD, and arch repair for D-TGA or D-TGA-like double outlet right ventricle (DORV) between 1983 and 2005, with additional diagnoses of VSD and AAO (hypoplasia, coarctation, or interruption). Patients were excluded if they did not meet anatomic criteria, underwent ASO at another institution, or did not undergo aortic arch intervention due to the finding of an insignificant gradient after closure of the ductus arteriosus. Patients were also excluded from analysis if they had no preoperative echocardiogram. This retrospective study was approved by the Institutional Review Board at Children’s Hospital, Boston on September 21, 2005 and individual informed consent was waived.
Outcomes We were primarily interested in whether preoperative anatomic variables were associated with the risk of either 30-day mortality or reoperation for RVOTO. Specifically, we hypothesized that threshold size of the tricuspid valve (TV), aortic valve, and distal transverse aortic arch (TrAo) would exist in this population, below which an anatomic repair would not reliably support compensated postoperative physiology and survival. Further, we were interested in understanding the importance of preoperative anatomy in postoperative RVOTO. Secondary analyses considered intensive care unit (ICU) and hospital length of stay and days of mechanical ventilation, as these are indicators of postoperative morbidity or slow recovery. We were able to confidently ascertain survival status on all patients in the sample. Long-term postoperative follow-up was collected through computerized medical records and archived correspondence. Where no records existed, we contacted the referring cardiologist for follow-up information.
Risk Factors We considered preoperative, operative, and postoperative variables. Anatomic detail was obtained from the preoperative echocardiogram; preoperative and postoperative morbidity from the hospital record and operative times from the anesthesia record. Coronary artery pattern was defined by the echocardiogram and the surgeon’s operative note. Where discrepancies existed, the operative findings were used.
Echocardiographic Measurements Echocardiograms were reviewed and measurements made offline from the original recording by a single cardiologist (M.L.S.) who was blinded to patient outcome. Measurements of TV, aortic valve and distal transverse aortic arch diameters were expressed as z scores based on our institutional normative database. The aortic annulus diameter was measured at the valve hinge points at end-systole from parasternal views. The lateral TV annulus diameter was measured at the hinge points at end-diastole from the apical four-chamber view. The distal TrAo diameter was measured between the left subclavian artery and the left common carotid artery in the long axis of the arch from the suprasternal notch. We categorized RV hypoplasia by both the echocardio-
GOTTLIEB ET AL OUTCOMES OF ARTERIAL SWITCH OPERATION
1699
grapher’s subjective determination as well as a ratio of right to left ventricular long axis measurements. The RV and LV long axis dimensions were measured in diastole from the apical four-chamber view and a RV:LV ratio was calculated. When echocardiographic images were inadequate to obtain accurate measurements, data were recorded as missing. Ventricular septal defect location was determined from the preoperative echocardiogram and confirmed by intraoperative findings.
Statistical Analysis Baseline patient characteristics were compared for subjects with and without preoperative echocardiograms using the Fisher exact test for categoric variables, and the Wilcoxon rank sum test for continuous variables. Mean z scores were compared for survivors (S) and nonsurvivors (NonS), and for those who did (R) and did not require (NR) reoperation using the two-sample t test. Univariate relationships between mortality and clinical and anatomic risk factors were evaluated using univariate logistic regression. For continuous variables (eg, z-score measurements), possible cut points or threshold values were explored. Because thresholds are clinically meaningful, we hypothesized that the ability of the ASO to restore normal physiology would be limited beyond a specific small size of critical anatomic structures. We therefore used an exploratory analysis to define this threshold. A stratified analysis was then performed in order to distinguish between anatomic and technical risk factors for mortality. The small number of events (nine deaths) precluded multivariate analysis [8]. For patients who survived the initial operation, relationships between patient risk factors and time to reoperation were explored using the Cox proportional hazards model.
Results Patients Our initial search yielded 102 patients who underwent ASO for D-TGA, or D-TGA-like DORV, with VSD and AAO including arch hypoplasia, coarctation, or interruption. Six patients underwent ASO at another institution and received only follow-up care at Children’s Hospital, Boston, and were therefore not reviewed. Thirteen patients had an initial diagnosis of aortic coarctation or hypoplasia, but had no aortic gradient on follow-up evaluation prior to surgery, and therefore did not meet our inclusion criteria of aortic arch obstruction requiring surgical repair. One patient had a missing chart and was therefore not reviewed. Eight patients who met anatomic inclusion criteria were not included in the analysis due to missing or inadequate preoperative echocardiograms. Echocardiograms of the remaining 74 patients were reviewed and included in the mortality analysis; 65 of these 74 patients survived ASO and were included in the reoperation analysis.
PEDIATRIC CARDIAC
Ann Thorac Surg 2008;85:1698 –703
1700
GOTTLIEB ET AL OUTCOMES OF ARTERIAL SWITCH OPERATION
Ann Thorac Surg 2008;85:1698 –703
Table 1. Summary of Patient and Operative Characteristics (n ⫽ 74) PEDIATRIC CARDIAC
Number (%), Median (range), or Mean ⫾ SD
Patient Characteristic Era of Repair 1983–1989 1990–1996 1997–2005 Age at surgery (days) Weight at surgery (grams) Male Aortic arch anatomy Diffuse aortic arch hypoplasia Localized aortic coarctation Interrupted aortic arch VSD type (86 total in 50 patients, missing n ⫽ 1) Malalignment Paramembranous Muscular Conal septal Atrioventricular canal Tricuspid valve annulus z-score Aortic annulus z-score Distal transverse aortic arch z-score Total perfusion time (minutes) Cross clamp time (minutes) Circulatory arrest time (minutes) Coronary artery pattern I (usual for D-TGA) II (Cx from RCA) III (single LCA) IV (single RCA) V (inverted origins) VI (inverted RCA and Cx) VII (intramural LAD)
16 (21.6%) 33 (44.6%) 25 (33.8%) 14 (2 to 3.2y) 3,600 (1,300 to 14,700) 55 (74.3%) 38 (51.4%) 51 (68.9%) 9 (12.2%)
Fig 1. Mortality associated with small transverse arch (TrAo) and tricuspid valve (TV) z score.
34 (39.5%) 14 (16.3%) 16 (18.6%) 17 (19.8%) 4 (4.7%) 0.71 ⫾ 1.32 ⫺0.14 ⫾ 1.35 ⫺2.51 ⫾ 1.13 197 (106 to 442) 119 (60 to 182) 50 (0 to 91)
Predictors of Mortality 38 (51.4%) 10 (13.5%) 2 (2.7%) 5 (6.8%) 6 (8.1%) 9 (12.2%) 2 (2.7%)
Cx ⫽ circumflex coronary artery; LAD ⫽ left anterior descending coronary artery; LCA ⫽ left coronary artery; RCA ⫽ right coronary artery.
Baseline Characteristics Baseline characteristics of the 74 patients with echocardiograms are shown in Table 1. All patients had aortic arch obstruction and VSD; VSDs shown were those Table 2. Univariate Predictors of Mortality Covariate
requiring surgical closure. To consider the possibility of bias introduced by the availability of an echocardiogram, we analyzed differences in baseline characteristics between patients with and without preoperative echocardiograms. Patients with echocardiograms were younger (median 14 days vs 0.4 years, p ⫽ 0.011), smaller (median weight at operation 3,600 g vs 5,350, p ⫽ 0.007), more likely to be receiving prostaglandins (51.3 vs 12.5%, p ⫽ 0.060), and to be mechanically ventilated preoperatively (56.8 vs 12.5%, p ⫽ 0.024) than the remainder of anatomically eligible patients (n ⫽ 8).
Died
Survived
p Value
Preoperative inotrope Preoperative ventilation Mean CAT (min) Muscular VSD Distal tranverse arch z score ⱕ ⫺2.5 Tricuspid valve z score ⬍0
6 (67%) 8 (89%) 62 ⫾ 16 3 (33%) 8 (89%)
19 (29%) 34 (52%) 47 ⫾ 20 3 (5%) 31 (49%)
0.054 0.069 0.026 0.021 0.033
6 (67%)
15 (23%)
0.013
CAT ⫽ circulatory arrest time;
VSD ⫽ ventricular septal defect.
Overall 30-day all-cause mortality was 12.2% in our cohort. There were no deaths after 30 days. Univariate analysis of mortality is given in Table 2. We hypothesized that small tricuspid valve and distal transverse aortic arch z scores would be associated with excess mortality in this patient population. We defined a small distal TrAo to be z equal to ⫺2.5 or less, and a small TV as z less than 0. In our sample, mortality rate with neither a small TrAo nor a small TV was 0%; with small TrAo alone, 11.1%; with small TV alone, 11.1%. With both small TV and TrAo, mortality rate was 41.7% (2 test for trend, p ⫽ 0.002 for comparison of 0 vs 1 vs 2 risk factors). Figure 1 provides a graphic representation of significant anatomic risk factor data. We searched for a relationship between TV z score and TrAo z score, and found none (p ⫽ 0.80). We found no differences in mortality between patients with small right ventricles by either our subjective or measured definition (data not shown). There were no differences in mortality found between operating surgeons and no differences in mortality associated with era of operation. By univariate analysis, the presence of an isolated muscular VSD was associated with increased mortality, with three of the nine deaths occurring in patients with isolated muscular VSDs (Table 2). The presence of multiple VSDs was not associated with increased mortality. Patients were stratified by circulatory arrest time greater and less than 60 minutes. Anatomic risk factors were tested in univariate analysis in the subset of pa-
GOTTLIEB ET AL OUTCOMES OF ARTERIAL SWITCH OPERATION
1701
interruption, or a combination) was associated with an increased risk of reoperation for RVOTO (data not shown). Neither era of operation nor surgeon had an associated increase in risk of reoperation (data not shown). With regard to recurrent AAO in our series, two of 74 patients (2.7%) had recurrent or newly symptomatic aortic coarctation requiring operation or balloon dilation following ASO, and one patient (1.4%) had supravalvar aortic stenosis and underwent operative repair.
Secondary Outcomes
Fig 2. Kaplan-Meier curve for patients with left coronary artery (LCA) with posterior versus anterior origin.
tients with circulatory arrest times less than 60 minutes. Distal transverse aortic arch and tricuspid valve below the threshold remained significant predictors in this group; patients with small TrAo and circulatory arrest time less than 60 minutes were significantly more likely to die (p ⫽ 0.043), as were patients with TV less than 0 and CAT less than 60 minutes (p ⫽ 0.049). These risk factors were nonsignificant in patients with circulatory arrest times equal to or above 60 minutes.
Predictors of Reoperation Reintervention (balloon angioplasty or operation) for RVOTO was performed in 15.4% (ten patients). One patient had additional aortic regurgitation; no patients were found to have recurrent AAO. Two variables, initial preoperative aortic annulus (neopulmonary) z score and coronary artery pattern predicted risk of first reoperation for RVOTO. Initial aortic annulus size was smaller in patients who eventually required reoperation for RVOTO (z score ⫺0.69 ⫾ 0.83 in R vs 0.03 ⫾ 1.43 in NR, p ⫽ 0.048). In our cohort, 49.2% had the usual coronary pattern for D-TGA. We found that patients with a coronary pattern in which the LCA arose posteriorly (LCA-post ⫽ any left coronary vessel, including left anterior descending, left circumflex, or both, arising from the posterior facing sinus) required reoperation for RVOTO sooner than patients with the LCA arising anteriorly (LCA-ant). The hazard ratio for reoperation in children with LCA-post was 5.2 (confidence interval [CI 1.1, 24.5], p ⫽ 0.022) (Fig 2). The LCA origin was related to great vessel orientation: patients with LCA arising posteriorly had a higher prevalence of side-by-side great vessels (p ⬍ 0.001). We found no increased risk of reoperation in patients who underwent any specific type of original aortic arch repair (end-to-end vs patch plasty, data not shown). Similarly, timing of aortic repair did not influence the risk of reoperation for RVOTO; patients who had aortic arch repair before or after ASO (two stages), or at the time of ASO (single stage), showed no differences in their risk of reoperation for RVOTO (data not shown). Finally, no specific type of AAO (hypoplasia, coarctation or
Longer length of post-ASO ICU stay was associated with the requirement for preoperative inotropic and ventilatory support, and with a TV z score less than 0. Increases in CAT were found to be associated with increased ICU length of stay, though the correlation was weak (Spearman rank correlation coefficient ⫽ 0.34, p ⫽ 0.008). Preoperative mechanical ventilation, preoperative inotropic support, and longer duration of CAT were also associated with longer postoperative ventilatory support and overall hospital length of stay.
Comment We have identified preoperative anatomic risk factors for both mortality and RVOTO requiring reoperation in the largest cohort of patients with D-TGA, VSD, and AAO described. Our findings reinforce and add detail to several previously described risk factors, and additionally identify previously unknown anatomic risk factors to be predictors of mortality in this complex population: small distal transverse aortic arch z score, small tricuspid valve lateral diameter z score, and surgically corrected muscular VSD. Though we expected surgeon and era of operation to influence outcome, CAT was the only technical factor found to be associated with mortality. When we considered the subset of patients who died but had short circulatory arrest times, we found that anatomic predictors remained strongly associated with the risk of mortality. We conclude that even with a technically satisfactory operation, these anatomic factors remain associated with death after ASO. Our data show that in a subset of patients with D-TGA, VSD, and AAO who have small TV and TrAo, mortality could be as high as 41.7%. Further investigation is warranted to determine whether these patients would be better served by alternative procedures such as transplantation or single-ventricle management. Though our sample of patients with D-TGA, VSD, and AAO had an increased prevalence of the less common coronary artery patterns for D-TGA, we did not find coronary artery pattern type to be associated with mortality in this analysis [1, 11]. As our sample was predominantly derived from patients undergoing operation in the years 1990 to 1994, technical improvements in coronary reimplantation may have neutralized the mortality risk of unusual coronary artery patterns [9, 10]. Hutter and colleagues [5] found in a sample of 177 patients from
PEDIATRIC CARDIAC
Ann Thorac Surg 2008;85:1698 –703
1702
GOTTLIEB ET AL OUTCOMES OF ARTERIAL SWITCH OPERATION
PEDIATRIC CARDIAC
1977 to 1999 that coronary artery anatomy was not a predictor of mortality in all patients with D-TGA, but surgical coronary obstruction predicted mortality, independent of original coronary anatomy. Right ventricular (RV) hypoplasia and RV dysfunction have previously been described as risk factors for mortality in arterial switch procedures [1, 12]. We did not confirm this finding in our sample. We attempted to quantify RV size indirectly by measuring several dimensions related to the RV, because we did not have reliable volumetric measurements. However, there was an important association between smaller tricuspid annular diameter and outcome. The finding that the presence of an isolated muscular VSD was associated with increased mortality was unexpected. Despite the potential for a residual muscular VSD, which could contribute to increased morbidity or mortality, none of these patients had a significant residual VSD. Therefore, the association between increased mortality risk and the presence of an isolated muscular VSD is unexplained. With regard to reoperation, we found native aortic annulus (neopulmonary) size to be associated with subsequent RVOTO. This finding suggests that a small aortic annulus size was not reliably addressed by ASO, and may indicate the need for RVOT enlargement in such patients at the time of ASO. Second, a posterior origin of the left coronary artery was a predictor of reoperation for RVOTO. Great vessel orientation was associated with coronary anatomy; our sample was too small to discern whether these risk factors were independent. Our study was limited by sample size. Multivariate models for mortality could not be tested due to the relatively small number of deaths in this population. Small sample sizes raise the concern for lack of generalizability and idiosyncrasies of a particular data set. These results could be validated in a larger, national data set in order to address these limitations. Despite the sample size limitation, our results suggest that anatomic risk factors are strongly associated with operative mortality in this population. The potential for bias in our sample was considered through analysis of patients with and without preoperative echocardiograms. Patients with echocardiograms accessible for review were younger and more likely to be mechanically ventilated than patients without echocardiograms at our institution. This difference may be explained by apnea due to prostaglandin administration in young patients, though may nonetheless generate a source of bias in our analysis. However, this concern relates to a small number of patients who had no echocardiogram for review (n ⫽ 8).
Ann Thorac Surg 2008;85:1698 –703
Our findings raise questions about the optimal operation for patients with D-TGA, VSD, and AAO who have small tricuspid valve and transverse aortic arches, because these additional anatomic features, either individually or in combination, are associated with significantly increased mortality risk. The arterial switch operation has become the standard of care in patients with D-TGA, and represents a major advance in the history of congenital cardiac surgery. With improvement in our ability to predict and monitor outcomes, we may find that within this complex set of patients there is a subset that is best served by an alternate surgical strategy.
References 1. Blume ED, Altmann K, Mayer JE, Colan SD, Gauvreau K, Geva T. Evolution of risk factors influencing early mortality of the arterial switch operation. J Am Coll Cardiol 1999;33: 1702–9. 2. Brown JW, Park HJ, Turrentine MW. Arterial switch operation: factors impacting survival in the current era. Ann Thorac Surg 2001;71:1978 – 84. 3. Daebritz SH, Nollert G, Sachweh JS, Engelhardt W, von Bernuth G, Messmer BJ. Anatomical risk factors for mortality and cardiac morbidity after arterial switch operation. Ann Thorac Surg 2000;69:1880 – 6. 4. Hawkins J, Clark E, Doty D. Morphological characteristics of aortic atresia: implications for fetal hemodynamics. Int J Cardiol 1986;10:127–32. 5. Hutter PA, Bennink GB, Ay L, Raes IB, Hitchcock JF, Meijboom EJ. Influence of coronary anatomy and reimplantation on the long-term outcome of the arterial switch. Eur J Cardiothorac Surg 2000;18:207–13. 6. Karl T, Cochrane A, Brizard C. Arterial switch operation: surgical solutions to complex problems. Texas Heart Inst J 1997;24:322–33. 7. Kołcz J, Januszewska K, Mroczek T, Malec E. Anatomical correction of complex forms of transposition of the great arteries in neonates. Scand Cardiovasc J 2004;38:164 –71. 8. Peduzzi PN, Concato J, Kemper E, Holford TR, Feinstein A. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 1996;99:1373–9. 9. Losay J, Touchot A, Capderou A, et al. Aortic valve regurgitation after arterial switch operation for transposition of the great arteries: incidence, risk factors, and outcome. 2006;47: 2057– 62. 10. Pasquali SK, Hasselblad V, Li JS, Kong DF, Sanders SP. Coronary artery pattern and outcome of arterial switch operation for transposition of the great arteries: a metaanalysis. Circulation 2002;106:2575– 80. 11. Planche C, Serraf A, Comas JV, Lacour-Gayet F, Bruniaux J, Touchot A. Anatomic repair of transposition of the great arteries with ventricular septal defect and aortic arch obstruction: one stage versus two stage procedure. J Thorac Cardiovasc Surg 1993;105:925–33. 12. Pocar M, Villa E, Degandt A, Mauriat P, Pouard P, Vouhé PR. Long-term results after primary one-stage repair of transposition of the great arteries and aortic arch obstruction. J Am Coll Cardiol 2005;46:1331– 8.
DISCUSSION DR JEFFREY P. JACOBS (St. Petersburg, FL): Thank you for that nice presentation. On Sunday at the postgraduate course, Francois Lacour-Gayet discussed this problem a little bit and talked about addressing
right ventricular outflow tract obstruction at the time of the switch. Was that done in any of the patients in the Boston series or does the team in Boston have any experience with that type of an approach?
DR GOTTLIEB: There were no transannular patches at the time of original switch. I think Dr Mayer may want to add some more. DR JOHN E. MAYER (Boston, MA): As Dr Gottlieb said, there were no direct operations on the neopulmonary annulus. We frequently employed infundibular patches, particularly for the double-outlet right ventricle type patients. So there was no primary transannular patching as Dr Gottlieb said. DR CHRISTOPHER A. CALDARONE (Toronto, Ontario, Canada): Would you comment on the use of the bivariate analysis technique that you used. It was very illustrative to see the way a small tricuspid valve plus a small transverse arch worked in a synergistic way to really increase mortality. It was very illustrative. Why not utilize a more commonly used multivariable analysis technique to identify important risk factors and examine the interplay of risk factors rather than limiting yourself to a bivariate analysis? DR GOTTLIEB: In statistics there’s a common practice that’s used based on a more complex mathematical determination of the number of covariates allowable for a model to be statistically stable. Basically a model is required to have ten outcomes for each predictor that is entered into a model. So with the sample size of 74 in our mortality analysis, we had a total of nine deaths, which means that our largest model should include one, at most two, predictors. Taking the bivariate model approaches assures that our results are as reliable as possible. DR CALDARONE: And would you speculate about why the posterior left coronary artery was so important in terms of RV outflow tract obstruction?
GOTTLIEB ET AL OUTCOMES OF ARTERIAL SWITCH OPERATION
1703
DR GOTTLIEB: We found an association between the relationship of the great vessels and coronary anatomy. Side-by-side anatomy was associated with a posterior origin of the left coronary artery. And we again did a similar bivariate analysis in which we entered great vessel relationship and left coronary artery origin anatomy, looking at the reoperation risk. We found that with both variables in the model, great vessel orientation loses its significance, meaning that the stronger predictor of RV outflow tract obstruction was coronary artery position and not relationship of the great vessels. DR RALPH S. MOSCA (New York, NY): It’s interesting that tricuspid valve size, small neoaortic outflow, and hypoplasia of the transverse arch all seem to be risk factors. Aren’t they all just surrogates for reduced flow through that side of the heart prior to the switch? Since repair of the arch should remove it as a hemodynamic factor it is interesting that it still appears as a significant risk factor for mortality. Do you have any insight into why this is so? DR GOTTLIEB: It’s a really good question, and we had the same one. We tried to look at whether there was a relationship between tricuspid valve size and a small distal transverse aortic arch size and found no statistical relationship. The p value was 0.8. So, even though there may be a developmental relationship, there was no statistical relationship in our study. Tricuspid valve size and aortic arch size appear to have independent influences on mortality and a synergistic effect on mortality when in combination.
PEDIATRIC CARDIAC
Ann Thorac Surg 2008;85:1698 –703
Predictors of Outcome of Arterial Switch Operation for Complex D-Transposition Danielle Gottlieb, MD, MPH, Marcy L. Schwartz, MD, Kara Bischoff, BA, Kimberlee Gauvreau, ScD, and John E. Mayer, Jr, MD Department of Cardiology and Cardiovascular Surgery, Children’s Hospital Boston, and Harvard Medical School, Boston, Massachusetts
Background. Overall mortality and reoperation risk for the arterial switch operation (ASO) for D-transposition of the great arteries (D-TGA) is low. D-TGA with ventricular septal defect (VSD) and aortic arch obstruction (AAO) is a higher risk subgroup in which we sought risk factors for mortality and reoperation after ASO. Methods. Echocardiograms of 74 patients who underwent ASO, VSD, and arch repair for D-TGA, VSD and AAO were reviewed; the reoperation analysis considered the 65 survivors. Pre-ASO clinical and anatomic characteristics were compared between survivors and nonsurvivors; patients who required (R) and did not require (NR) reoperation. Results. Distal transverse aortic arch (TrAo) z score equal to ⴚ2.5 or less, triscuspid valve z score less than 0, repaired muscular VSD, and circulatory arrest time were significant predictors of mortality. When stratified for circulatory arrest time below 60 minutes, small distal
transverse aortic arch and tricuspid valve remained significant predictors of mortality. Mean aortic annulus size was smaller in R than NR (p ⴝ 0.048). Left coronary artery arising posteriorly was associated with a reoperation hazard ratio of 5.2 (p ⴝ 0.022). Conclusions. Preoperative anatomy was associated with death and reoperation post-ASO. Small TrAo and TV were risk factors for mortality in univariate analysis, and remained significant in the subset of patients with short circulatory arrest times, suggesting that even when controlling for technical factors, anatomic risk factors predict mortality. Small aortic annulus and posterior left circumflex artery origin were associated with reoperation. Patients with D-TGA, VSD, and AAO constitute a higher risk group, which includes patients who may be marginal candidates for two-ventricle repair. (Ann Thorac Surg 2008;85:1698 –703) © 2008 by The Society of Thoracic Surgeons
A
s survival from D-transposition of the great arteries (D-TGA) has improved since the introduction of the arterial switch operation (ASO), indications for ASO have been extended to include more complex anatomic subsets [1]. Though ASO seems to be the operation of choice for patients with complex D-TGA, differential survival for patients with complex TGA has been demonstrated [2– 4]. Further, aortic arch obstruction (AAO) has been identified in multiple prior studies as a risk factor for less favorable outcome. However, the additional risk of AAO in the setting of D-TGA is incompletely understood as most analyses are limited by small numbers of patients with this additional anatomic complexity [2, 4 –7]. Previous risk factor analyses have identified “complex D-TGA” and ventricular septal defect (VSD) as risk factors for mortality, but additional detail is needed for accurate preoperative risk stratification, operative planning, and family counseling. Central to surgical planning decisions is the ability to distinguish between patients who have uncorrectable anatomy and those whose anatomy provides technical difficulty and consequent morbidity associated with longer cardiopul-
monary bypass and circulatory arrest times. We sought to better understand whether increased mortality in patients with D-TGA, VSD, and AAO can be predicted by anatomy, or whether anatomy is a risk factor for an intermediate variable, operative difficulty. We undertook this analysis in order to determine whether anatomic features contribute to the observed increased mortality in patients with D-TGA, VSD, and AAO. Multiple investigators have also considered whether preoperative anatomy is associated with the need for reoperation, but most studies have been performed in heterogeneous populations, with complex patients representing a small subgroup, and studies have therefore been underpowered to find a difference between simple and complex variants of D-TGA. Risk factors have varied between studies and have not been thoroughly examined in this complex subset of D-TGA, VSD, and AAO patients [1, 5]. We sought to define risk factors for right ventricular outflow tract obstruction (RVOTO) requiring reoperation and to determine whether risk is dictated by pre-ASO anatomy.
Accepted for publication Jan 23, 2008.
Patients and Methods
Presented at the Forty-third Annual Meeting of The Society of Thoracic Surgeons, San Diego, CA, Jan 29 –31, 2007.
Patients
Address correspondence to Dr Gottlieb, 300 Longwood Avenue, Boston, MA 02115; e-mail: [email protected].
Patients were identified by searching the Children’s Hospital cardiovascular program database for patients
© 2008 by The Society of Thoracic Surgeons Published by Elsevier Inc
0003-4975/08/$34.00 doi:10.1016/j.athoracsur.2008.01.075
who underwent ASO, VSD, and arch repair for D-TGA or D-TGA-like double outlet right ventricle (DORV) between 1983 and 2005, with additional diagnoses of VSD and AAO (hypoplasia, coarctation, or interruption). Patients were excluded if they did not meet anatomic criteria, underwent ASO at another institution, or did not undergo aortic arch intervention due to the finding of an insignificant gradient after closure of the ductus arteriosus. Patients were also excluded from analysis if they had no preoperative echocardiogram. This retrospective study was approved by the Institutional Review Board at Children’s Hospital, Boston on September 21, 2005 and individual informed consent was waived.
Outcomes We were primarily interested in whether preoperative anatomic variables were associated with the risk of either 30-day mortality or reoperation for RVOTO. Specifically, we hypothesized that threshold size of the tricuspid valve (TV), aortic valve, and distal transverse aortic arch (TrAo) would exist in this population, below which an anatomic repair would not reliably support compensated postoperative physiology and survival. Further, we were interested in understanding the importance of preoperative anatomy in postoperative RVOTO. Secondary analyses considered intensive care unit (ICU) and hospital length of stay and days of mechanical ventilation, as these are indicators of postoperative morbidity or slow recovery. We were able to confidently ascertain survival status on all patients in the sample. Long-term postoperative follow-up was collected through computerized medical records and archived correspondence. Where no records existed, we contacted the referring cardiologist for follow-up information.
Risk Factors We considered preoperative, operative, and postoperative variables. Anatomic detail was obtained from the preoperative echocardiogram; preoperative and postoperative morbidity from the hospital record and operative times from the anesthesia record. Coronary artery pattern was defined by the echocardiogram and the surgeon’s operative note. Where discrepancies existed, the operative findings were used.
Echocardiographic Measurements Echocardiograms were reviewed and measurements made offline from the original recording by a single cardiologist (M.L.S.) who was blinded to patient outcome. Measurements of TV, aortic valve and distal transverse aortic arch diameters were expressed as z scores based on our institutional normative database. The aortic annulus diameter was measured at the valve hinge points at end-systole from parasternal views. The lateral TV annulus diameter was measured at the hinge points at end-diastole from the apical four-chamber view. The distal TrAo diameter was measured between the left subclavian artery and the left common carotid artery in the long axis of the arch from the suprasternal notch. We categorized RV hypoplasia by both the echocardio-
GOTTLIEB ET AL OUTCOMES OF ARTERIAL SWITCH OPERATION
1699
grapher’s subjective determination as well as a ratio of right to left ventricular long axis measurements. The RV and LV long axis dimensions were measured in diastole from the apical four-chamber view and a RV:LV ratio was calculated. When echocardiographic images were inadequate to obtain accurate measurements, data were recorded as missing. Ventricular septal defect location was determined from the preoperative echocardiogram and confirmed by intraoperative findings.
Statistical Analysis Baseline patient characteristics were compared for subjects with and without preoperative echocardiograms using the Fisher exact test for categoric variables, and the Wilcoxon rank sum test for continuous variables. Mean z scores were compared for survivors (S) and nonsurvivors (NonS), and for those who did (R) and did not require (NR) reoperation using the two-sample t test. Univariate relationships between mortality and clinical and anatomic risk factors were evaluated using univariate logistic regression. For continuous variables (eg, z-score measurements), possible cut points or threshold values were explored. Because thresholds are clinically meaningful, we hypothesized that the ability of the ASO to restore normal physiology would be limited beyond a specific small size of critical anatomic structures. We therefore used an exploratory analysis to define this threshold. A stratified analysis was then performed in order to distinguish between anatomic and technical risk factors for mortality. The small number of events (nine deaths) precluded multivariate analysis [8]. For patients who survived the initial operation, relationships between patient risk factors and time to reoperation were explored using the Cox proportional hazards model.
Results Patients Our initial search yielded 102 patients who underwent ASO for D-TGA, or D-TGA-like DORV, with VSD and AAO including arch hypoplasia, coarctation, or interruption. Six patients underwent ASO at another institution and received only follow-up care at Children’s Hospital, Boston, and were therefore not reviewed. Thirteen patients had an initial diagnosis of aortic coarctation or hypoplasia, but had no aortic gradient on follow-up evaluation prior to surgery, and therefore did not meet our inclusion criteria of aortic arch obstruction requiring surgical repair. One patient had a missing chart and was therefore not reviewed. Eight patients who met anatomic inclusion criteria were not included in the analysis due to missing or inadequate preoperative echocardiograms. Echocardiograms of the remaining 74 patients were reviewed and included in the mortality analysis; 65 of these 74 patients survived ASO and were included in the reoperation analysis.
PEDIATRIC CARDIAC
Ann Thorac Surg 2008;85:1698 –703
1700
GOTTLIEB ET AL OUTCOMES OF ARTERIAL SWITCH OPERATION
Ann Thorac Surg 2008;85:1698 –703
Table 1. Summary of Patient and Operative Characteristics (n ⫽ 74) PEDIATRIC CARDIAC
Number (%), Median (range), or Mean ⫾ SD
Patient Characteristic Era of Repair 1983–1989 1990–1996 1997–2005 Age at surgery (days) Weight at surgery (grams) Male Aortic arch anatomy Diffuse aortic arch hypoplasia Localized aortic coarctation Interrupted aortic arch VSD type (86 total in 50 patients, missing n ⫽ 1) Malalignment Paramembranous Muscular Conal septal Atrioventricular canal Tricuspid valve annulus z-score Aortic annulus z-score Distal transverse aortic arch z-score Total perfusion time (minutes) Cross clamp time (minutes) Circulatory arrest time (minutes) Coronary artery pattern I (usual for D-TGA) II (Cx from RCA) III (single LCA) IV (single RCA) V (inverted origins) VI (inverted RCA and Cx) VII (intramural LAD)
16 (21.6%) 33 (44.6%) 25 (33.8%) 14 (2 to 3.2y) 3,600 (1,300 to 14,700) 55 (74.3%) 38 (51.4%) 51 (68.9%) 9 (12.2%)
Fig 1. Mortality associated with small transverse arch (TrAo) and tricuspid valve (TV) z score.
34 (39.5%) 14 (16.3%) 16 (18.6%) 17 (19.8%) 4 (4.7%) 0.71 ⫾ 1.32 ⫺0.14 ⫾ 1.35 ⫺2.51 ⫾ 1.13 197 (106 to 442) 119 (60 to 182) 50 (0 to 91)
Predictors of Mortality 38 (51.4%) 10 (13.5%) 2 (2.7%) 5 (6.8%) 6 (8.1%) 9 (12.2%) 2 (2.7%)
Cx ⫽ circumflex coronary artery; LAD ⫽ left anterior descending coronary artery; LCA ⫽ left coronary artery; RCA ⫽ right coronary artery.
Baseline Characteristics Baseline characteristics of the 74 patients with echocardiograms are shown in Table 1. All patients had aortic arch obstruction and VSD; VSDs shown were those Table 2. Univariate Predictors of Mortality Covariate
requiring surgical closure. To consider the possibility of bias introduced by the availability of an echocardiogram, we analyzed differences in baseline characteristics between patients with and without preoperative echocardiograms. Patients with echocardiograms were younger (median 14 days vs 0.4 years, p ⫽ 0.011), smaller (median weight at operation 3,600 g vs 5,350, p ⫽ 0.007), more likely to be receiving prostaglandins (51.3 vs 12.5%, p ⫽ 0.060), and to be mechanically ventilated preoperatively (56.8 vs 12.5%, p ⫽ 0.024) than the remainder of anatomically eligible patients (n ⫽ 8).
Died
Survived
p Value
Preoperative inotrope Preoperative ventilation Mean CAT (min) Muscular VSD Distal tranverse arch z score ⱕ ⫺2.5 Tricuspid valve z score ⬍0
6 (67%) 8 (89%) 62 ⫾ 16 3 (33%) 8 (89%)
19 (29%) 34 (52%) 47 ⫾ 20 3 (5%) 31 (49%)
0.054 0.069 0.026 0.021 0.033
6 (67%)
15 (23%)
0.013
CAT ⫽ circulatory arrest time;
VSD ⫽ ventricular septal defect.
Overall 30-day all-cause mortality was 12.2% in our cohort. There were no deaths after 30 days. Univariate analysis of mortality is given in Table 2. We hypothesized that small tricuspid valve and distal transverse aortic arch z scores would be associated with excess mortality in this patient population. We defined a small distal TrAo to be z equal to ⫺2.5 or less, and a small TV as z less than 0. In our sample, mortality rate with neither a small TrAo nor a small TV was 0%; with small TrAo alone, 11.1%; with small TV alone, 11.1%. With both small TV and TrAo, mortality rate was 41.7% (2 test for trend, p ⫽ 0.002 for comparison of 0 vs 1 vs 2 risk factors). Figure 1 provides a graphic representation of significant anatomic risk factor data. We searched for a relationship between TV z score and TrAo z score, and found none (p ⫽ 0.80). We found no differences in mortality between patients with small right ventricles by either our subjective or measured definition (data not shown). There were no differences in mortality found between operating surgeons and no differences in mortality associated with era of operation. By univariate analysis, the presence of an isolated muscular VSD was associated with increased mortality, with three of the nine deaths occurring in patients with isolated muscular VSDs (Table 2). The presence of multiple VSDs was not associated with increased mortality. Patients were stratified by circulatory arrest time greater and less than 60 minutes. Anatomic risk factors were tested in univariate analysis in the subset of pa-
GOTTLIEB ET AL OUTCOMES OF ARTERIAL SWITCH OPERATION
1701
interruption, or a combination) was associated with an increased risk of reoperation for RVOTO (data not shown). Neither era of operation nor surgeon had an associated increase in risk of reoperation (data not shown). With regard to recurrent AAO in our series, two of 74 patients (2.7%) had recurrent or newly symptomatic aortic coarctation requiring operation or balloon dilation following ASO, and one patient (1.4%) had supravalvar aortic stenosis and underwent operative repair.
Secondary Outcomes
Fig 2. Kaplan-Meier curve for patients with left coronary artery (LCA) with posterior versus anterior origin.
tients with circulatory arrest times less than 60 minutes. Distal transverse aortic arch and tricuspid valve below the threshold remained significant predictors in this group; patients with small TrAo and circulatory arrest time less than 60 minutes were significantly more likely to die (p ⫽ 0.043), as were patients with TV less than 0 and CAT less than 60 minutes (p ⫽ 0.049). These risk factors were nonsignificant in patients with circulatory arrest times equal to or above 60 minutes.
Predictors of Reoperation Reintervention (balloon angioplasty or operation) for RVOTO was performed in 15.4% (ten patients). One patient had additional aortic regurgitation; no patients were found to have recurrent AAO. Two variables, initial preoperative aortic annulus (neopulmonary) z score and coronary artery pattern predicted risk of first reoperation for RVOTO. Initial aortic annulus size was smaller in patients who eventually required reoperation for RVOTO (z score ⫺0.69 ⫾ 0.83 in R vs 0.03 ⫾ 1.43 in NR, p ⫽ 0.048). In our cohort, 49.2% had the usual coronary pattern for D-TGA. We found that patients with a coronary pattern in which the LCA arose posteriorly (LCA-post ⫽ any left coronary vessel, including left anterior descending, left circumflex, or both, arising from the posterior facing sinus) required reoperation for RVOTO sooner than patients with the LCA arising anteriorly (LCA-ant). The hazard ratio for reoperation in children with LCA-post was 5.2 (confidence interval [CI 1.1, 24.5], p ⫽ 0.022) (Fig 2). The LCA origin was related to great vessel orientation: patients with LCA arising posteriorly had a higher prevalence of side-by-side great vessels (p ⬍ 0.001). We found no increased risk of reoperation in patients who underwent any specific type of original aortic arch repair (end-to-end vs patch plasty, data not shown). Similarly, timing of aortic repair did not influence the risk of reoperation for RVOTO; patients who had aortic arch repair before or after ASO (two stages), or at the time of ASO (single stage), showed no differences in their risk of reoperation for RVOTO (data not shown). Finally, no specific type of AAO (hypoplasia, coarctation or
Longer length of post-ASO ICU stay was associated with the requirement for preoperative inotropic and ventilatory support, and with a TV z score less than 0. Increases in CAT were found to be associated with increased ICU length of stay, though the correlation was weak (Spearman rank correlation coefficient ⫽ 0.34, p ⫽ 0.008). Preoperative mechanical ventilation, preoperative inotropic support, and longer duration of CAT were also associated with longer postoperative ventilatory support and overall hospital length of stay.
Comment We have identified preoperative anatomic risk factors for both mortality and RVOTO requiring reoperation in the largest cohort of patients with D-TGA, VSD, and AAO described. Our findings reinforce and add detail to several previously described risk factors, and additionally identify previously unknown anatomic risk factors to be predictors of mortality in this complex population: small distal transverse aortic arch z score, small tricuspid valve lateral diameter z score, and surgically corrected muscular VSD. Though we expected surgeon and era of operation to influence outcome, CAT was the only technical factor found to be associated with mortality. When we considered the subset of patients who died but had short circulatory arrest times, we found that anatomic predictors remained strongly associated with the risk of mortality. We conclude that even with a technically satisfactory operation, these anatomic factors remain associated with death after ASO. Our data show that in a subset of patients with D-TGA, VSD, and AAO who have small TV and TrAo, mortality could be as high as 41.7%. Further investigation is warranted to determine whether these patients would be better served by alternative procedures such as transplantation or single-ventricle management. Though our sample of patients with D-TGA, VSD, and AAO had an increased prevalence of the less common coronary artery patterns for D-TGA, we did not find coronary artery pattern type to be associated with mortality in this analysis [1, 11]. As our sample was predominantly derived from patients undergoing operation in the years 1990 to 1994, technical improvements in coronary reimplantation may have neutralized the mortality risk of unusual coronary artery patterns [9, 10]. Hutter and colleagues [5] found in a sample of 177 patients from
PEDIATRIC CARDIAC
Ann Thorac Surg 2008;85:1698 –703
1702
GOTTLIEB ET AL OUTCOMES OF ARTERIAL SWITCH OPERATION
PEDIATRIC CARDIAC
1977 to 1999 that coronary artery anatomy was not a predictor of mortality in all patients with D-TGA, but surgical coronary obstruction predicted mortality, independent of original coronary anatomy. Right ventricular (RV) hypoplasia and RV dysfunction have previously been described as risk factors for mortality in arterial switch procedures [1, 12]. We did not confirm this finding in our sample. We attempted to quantify RV size indirectly by measuring several dimensions related to the RV, because we did not have reliable volumetric measurements. However, there was an important association between smaller tricuspid annular diameter and outcome. The finding that the presence of an isolated muscular VSD was associated with increased mortality was unexpected. Despite the potential for a residual muscular VSD, which could contribute to increased morbidity or mortality, none of these patients had a significant residual VSD. Therefore, the association between increased mortality risk and the presence of an isolated muscular VSD is unexplained. With regard to reoperation, we found native aortic annulus (neopulmonary) size to be associated with subsequent RVOTO. This finding suggests that a small aortic annulus size was not reliably addressed by ASO, and may indicate the need for RVOT enlargement in such patients at the time of ASO. Second, a posterior origin of the left coronary artery was a predictor of reoperation for RVOTO. Great vessel orientation was associated with coronary anatomy; our sample was too small to discern whether these risk factors were independent. Our study was limited by sample size. Multivariate models for mortality could not be tested due to the relatively small number of deaths in this population. Small sample sizes raise the concern for lack of generalizability and idiosyncrasies of a particular data set. These results could be validated in a larger, national data set in order to address these limitations. Despite the sample size limitation, our results suggest that anatomic risk factors are strongly associated with operative mortality in this population. The potential for bias in our sample was considered through analysis of patients with and without preoperative echocardiograms. Patients with echocardiograms accessible for review were younger and more likely to be mechanically ventilated than patients without echocardiograms at our institution. This difference may be explained by apnea due to prostaglandin administration in young patients, though may nonetheless generate a source of bias in our analysis. However, this concern relates to a small number of patients who had no echocardiogram for review (n ⫽ 8).
Ann Thorac Surg 2008;85:1698 –703
Our findings raise questions about the optimal operation for patients with D-TGA, VSD, and AAO who have small tricuspid valve and transverse aortic arches, because these additional anatomic features, either individually or in combination, are associated with significantly increased mortality risk. The arterial switch operation has become the standard of care in patients with D-TGA, and represents a major advance in the history of congenital cardiac surgery. With improvement in our ability to predict and monitor outcomes, we may find that within this complex set of patients there is a subset that is best served by an alternate surgical strategy.
References 1. Blume ED, Altmann K, Mayer JE, Colan SD, Gauvreau K, Geva T. Evolution of risk factors influencing early mortality of the arterial switch operation. J Am Coll Cardiol 1999;33: 1702–9. 2. Brown JW, Park HJ, Turrentine MW. Arterial switch operation: factors impacting survival in the current era. Ann Thorac Surg 2001;71:1978 – 84. 3. Daebritz SH, Nollert G, Sachweh JS, Engelhardt W, von Bernuth G, Messmer BJ. Anatomical risk factors for mortality and cardiac morbidity after arterial switch operation. Ann Thorac Surg 2000;69:1880 – 6. 4. Hawkins J, Clark E, Doty D. Morphological characteristics of aortic atresia: implications for fetal hemodynamics. Int J Cardiol 1986;10:127–32. 5. Hutter PA, Bennink GB, Ay L, Raes IB, Hitchcock JF, Meijboom EJ. Influence of coronary anatomy and reimplantation on the long-term outcome of the arterial switch. Eur J Cardiothorac Surg 2000;18:207–13. 6. Karl T, Cochrane A, Brizard C. Arterial switch operation: surgical solutions to complex problems. Texas Heart Inst J 1997;24:322–33. 7. Kołcz J, Januszewska K, Mroczek T, Malec E. Anatomical correction of complex forms of transposition of the great arteries in neonates. Scand Cardiovasc J 2004;38:164 –71. 8. Peduzzi PN, Concato J, Kemper E, Holford TR, Feinstein A. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 1996;99:1373–9. 9. Losay J, Touchot A, Capderou A, et al. Aortic valve regurgitation after arterial switch operation for transposition of the great arteries: incidence, risk factors, and outcome. 2006;47: 2057– 62. 10. Pasquali SK, Hasselblad V, Li JS, Kong DF, Sanders SP. Coronary artery pattern and outcome of arterial switch operation for transposition of the great arteries: a metaanalysis. Circulation 2002;106:2575– 80. 11. Planche C, Serraf A, Comas JV, Lacour-Gayet F, Bruniaux J, Touchot A. Anatomic repair of transposition of the great arteries with ventricular septal defect and aortic arch obstruction: one stage versus two stage procedure. J Thorac Cardiovasc Surg 1993;105:925–33. 12. Pocar M, Villa E, Degandt A, Mauriat P, Pouard P, Vouhé PR. Long-term results after primary one-stage repair of transposition of the great arteries and aortic arch obstruction. J Am Coll Cardiol 2005;46:1331– 8.
DISCUSSION DR JEFFREY P. JACOBS (St. Petersburg, FL): Thank you for that nice presentation. On Sunday at the postgraduate course, Francois Lacour-Gayet discussed this problem a little bit and talked about addressing
right ventricular outflow tract obstruction at the time of the switch. Was that done in any of the patients in the Boston series or does the team in Boston have any experience with that type of an approach?
DR GOTTLIEB: There were no transannular patches at the time of original switch. I think Dr Mayer may want to add some more. DR JOHN E. MAYER (Boston, MA): As Dr Gottlieb said, there were no direct operations on the neopulmonary annulus. We frequently employed infundibular patches, particularly for the double-outlet right ventricle type patients. So there was no primary transannular patching as Dr Gottlieb said. DR CHRISTOPHER A. CALDARONE (Toronto, Ontario, Canada): Would you comment on the use of the bivariate analysis technique that you used. It was very illustrative to see the way a small tricuspid valve plus a small transverse arch worked in a synergistic way to really increase mortality. It was very illustrative. Why not utilize a more commonly used multivariable analysis technique to identify important risk factors and examine the interplay of risk factors rather than limiting yourself to a bivariate analysis? DR GOTTLIEB: In statistics there’s a common practice that’s used based on a more complex mathematical determination of the number of covariates allowable for a model to be statistically stable. Basically a model is required to have ten outcomes for each predictor that is entered into a model. So with the sample size of 74 in our mortality analysis, we had a total of nine deaths, which means that our largest model should include one, at most two, predictors. Taking the bivariate model approaches assures that our results are as reliable as possible. DR CALDARONE: And would you speculate about why the posterior left coronary artery was so important in terms of RV outflow tract obstruction?
GOTTLIEB ET AL OUTCOMES OF ARTERIAL SWITCH OPERATION
1703
DR GOTTLIEB: We found an association between the relationship of the great vessels and coronary anatomy. Side-by-side anatomy was associated with a posterior origin of the left coronary artery. And we again did a similar bivariate analysis in which we entered great vessel relationship and left coronary artery origin anatomy, looking at the reoperation risk. We found that with both variables in the model, great vessel orientation loses its significance, meaning that the stronger predictor of RV outflow tract obstruction was coronary artery position and not relationship of the great vessels. DR RALPH S. MOSCA (New York, NY): It’s interesting that tricuspid valve size, small neoaortic outflow, and hypoplasia of the transverse arch all seem to be risk factors. Aren’t they all just surrogates for reduced flow through that side of the heart prior to the switch? Since repair of the arch should remove it as a hemodynamic factor it is interesting that it still appears as a significant risk factor for mortality. Do you have any insight into why this is so? DR GOTTLIEB: It’s a really good question, and we had the same one. We tried to look at whether there was a relationship between tricuspid valve size and a small distal transverse aortic arch size and found no statistical relationship. The p value was 0.8. So, even though there may be a developmental relationship, there was no statistical relationship in our study. Tricuspid valve size and aortic arch size appear to have independent influences on mortality and a synergistic effect on mortality when in combination.
PEDIATRIC CARDIAC
Ann Thorac Surg 2008;85:1698 –703