Results of reparative surgery for tetralogy of Fallot: data from the European Association for Cardio-Thoracic Surgery Congenital Database

Transcription

1 European Journal of Cardio-Thoracic Surgery 42 (2012) doi: /ejcts/ezs478 ORIGINAL ARTICLE Results of reparative surgery for tetralogy of Fallot: data from the European Association for Cardio-Thoracic Surgery Congenital Database George E. Sarris a, *, Juan V. Comas b, Zdzislaw Tobota c and Bohdan Maruszewski c a b c Mitera Children s and Hygeia Hospital, Athens Heart Surgery Institute, Athens, Greece Pediatric Heart Institute, Hospital 12th of October, Madrid, Spain Children s Memorial Health Institute, Warsaw, Poland * Corresponding author. Mitera Children s and Hygeia Hospital, Athens Heart Surgery Institute, # 6 Erythrou Stavrou Street, Maroussi, Athens, Greece. Tel: ; fax: ; gsarris@mac.com, gsarris@mitera.gr (G.E. Sarris). Received 28 September 2011; received in revised form 12 July 2012; accepted 16 July 2012 Abstract OBJECTIVES: Right ventricular (RV) dysfunction and pulmonary insufficiency (PI) after tetralogy of Fallot (TOF) repair may contribute to early and late morbidity and mortality. RV dysfunction may be related to the ventriculotomy employed in the transventricular repair technique, particularly when it is combined with a transannular patch (TAP). Transatrial/transpulmonary (TA/TP) repair without ventriculotomy has been advocated as a method potentially diminishing such adverse events. However, the prevalence and early as well as the late results of these different surgical approaches in Europe have not been studied. To ascertain the current prevalence and associated early mortality of various surgical approaches for repair of TOF, relevant data in the European Association for Cardio-Thoracic Surgery Congenital Database were analysed. METHODS: The study population was all types of reparative operations (n = 6654) for patients with primary diagnosis of TOF reported between 1999 and RESULTS: Overall hospital mortality (HM) was 2.58% (172 of 6654). Repair via ventriculotomy with TAP was the most prevalent technique (n = 3827, 57.5%), with HM 3.11%. Repair via ventriculotomy with non-tap was performed in 1309 patients (19.7%, HM = 1.53%). Repair without ventriculotomy was performed in 1214 patients (18.2%, HM = 1.48%). Logistic regression analysis showed statistically significant association between HM risk and the type of surgery. In particular, ventriculotomy with TAP is associated with increased mortality risk significantly compared with ventriculotomy with non-tap (crude odds ratio [OR] 2.13 (95% confidence interval [CI]: ). Similar results were obtained by analysing for operative mortality risk (30-day mortality, OM). Operations that have been performed before 2005 have resulted in increased surgical risk compared with those performed after 2005 (ORs for OM 1.45, 95% CI: ). CONCLUSIONS: Overall HM for TOF repair is low. TOF repair by means of ventriculotomy with TAP is the most prevalent approach and is associated with higher mortality. Repair with ventriculotomy but no TAP and repair without ventriculotomy are both less prevalent and with lower mortality. Surgical risk appears to be decreasing over time. Keywords: Tetralogy of Fallot Surgical repair Transventricular repair Transatrial transpulmonary repair Ventriculotomy Transannular patch INTRODUCTION Although the surgical repair of tetralogy of Fallot (TOF) with generally low morbidity and mortality has been established for years [1 18], there is increasing concern regarding the incidence of late adverse events, including severe pulmonary insufficiency (PI), RV dilatation and dysfunction, ventricular arrhythmias and even sudden death. Reoperation rates, mostly for pulmonary Presented at the 25th Annual Meeting of the European Association for Cardio-Thoracic Surgery, Lisbon, Portugal, 1 5 October valve replacement, are reported to be as high as 30% with long enough follow-up [19 23]. The origin of various late adverse events as well as of early surgical morbidity and mortality is held by many [7 10, 13, 14, 16, 21] to be related to the deleterious effects on right ventricular (RV) function of the ventriculotomy utilized in the traditional transventricular repair technique particularly when combined with a generous transannular patch (TAP), which results in free pulmonary valve insufficiency. On the other hand, transatrial/ transpulmonary (TA/TP) repair, which was introduced in 1960 [7] and popularized in the 1980s and 1990s [8 14], does not use a ventriculotomy involving the body of the RV, and only a The Author Published by Oxford University Press on behalf of the European Association for Cardio-Thoracic Surgery. All rights reserved.

2 G.E. Sarris et al. / European Journal of Cardio-Thoracic Surgery 767 minimal, if any, extension of a pulmonary arteriotomy across the pulmonary valve annulus. This technique has been advocated as less damaging to the right ventricle and the pulmonary valve, is claimed to be associated with an improved both early and late outcome and has gained substantial popularity [8 14]. However, in the absence of definitive data from prospective randomized comparisons, the controversy persists regarding the optimal technique and the timing of repair. Clinical practices regarding tetralogy repair differ widely among institutions [4 17], and the overall prevalence and results associated with these different techniques have not been studied outside of North America [18]. To ascertain whether these issues and concerns regarding particularly late outcome may have had an effect on current practice, we sought to determine the prevalence and early results of various current surgical approaches for repair of TOF, utilizing the available data in the European Association for Cardio-Thoracic Surgery Congenital Database (EACTS CDB). MATERIALS AND METHODS The source of data for this retrospective study is the EACTS CDB ( The study material was the data on all types of reparative operations (n = 6654) for classical TOF reported to the EACTS CDB for years (At the time of writing of this paper, reporting by participating centres for operations performed during the year 2011 is partly completed). Reparative operations for tetralogy with absent pulmonary valve and for tetralogy with atrioventricular septal defect were not included. Palliative operations for TOF or late reoperations after prior repair of TOF were also not included. According to the EACTS STS nomenclature used in the EACTS Database [24], operations for TOF are classified as: (i) TOF repair, ventriculotomy, transanular patch (V-TAP). (ii) TOF repair, ventriculotomy, non-transanular patch (V-non-TAP). (iii) TOF repair, no ventriculotomy (No-V). (iv) TOF repair, RV to PA conduit (RV PA). (v) TOF repair, repair not otherwise specified (NOS). The database was queried to obtain the number of procedures in each of the above categories and the corresponding surgical mortality, measured as 30-day mortality (OM) and also as hospital mortality (HM). Statistical analysis Statistical analysis was performed in R software (open source software Instances and frequencies are expressed as absolute numbers and percentages for categorical variables (e.g. year of surgery), while mean values and standard deviations (SDs) are used to describe numerical variables (e.g. age). Operative mortality (defined as OM and separately as HM) is reported as a percentage. Comparisons between the cases of mortality and the rest of the population were made with t-test and χ 2 test where appropriate. For comparisons among the three primary operations of interest (i.e. V-TAP, V-non-TAP, No-V), analysis of variance (ANOVA) and χ 2 tests were used as appropriate. Standard statistical tests were used to calculate odds ratios (ORs) and 95% confidence intervals (CIs). A multivariate logistic regression analysis was implemented to identify potential risk factors for mortality. We included the type of operation in the model to calculate crude ORs and compared mortality risk in V-non-TAP and V-TAP with mortality risk in No-V (reference category). We also adjusted for other potential risk factors, specifically, age at reparative surgery, year of operation and history of previous palliative surgery. RESULTS A total of 6654 reparative operations for TOF from 119 participating centres were analysed. Twenty-seven (27) centres, contributing 2029 patients were from outside of Europe. Table 1 summarizes the number and frequency of each of the types of surgical repair of TOF reported in the database and the corresponding operative surgical mortality, recorded as both OM and HM. Figure 1 shows the distribution of the total number of operations by the type of operation. Overall, repair via ventriculotomy with TAP was the most prevalent technique (n = 3827, 57.5%), with HM 3.11%. Repair via ventriculotomy with non-tap was performed in 1309 patients (19.7%, HM = 1.53%). Repair without ventriculotomy was performed in 1214 patients (18.2%, HM = 1.48%), and repair with RV PA conduit in 216 patients (3.25%, HM = 6.02%). Patient characteristics Table 2 summarizes the important clinical characteristics of the patients who underwent one of the three reparative procedures of interest (V-TAP, V-non-TAP and No-V), listed by the type of surgical repair. These characteristics include year reparative surgery was performed, age and history of previous palliative surgery. The results of ANOVA for age reveal that the three subpopulations differ significantly in age. Patients who have undergone repair by ventriculotomy with non-tap are overall older (average age 34.4 months [SD 68.0]) than patients that have been treated with the other two operations (22.2 months, SD 47 for V-TAP, and 24.1 months, SD 63.4 for No-V). Moreover, the three populations differ with regard to the year of operation. Most recorded operations (>60%) have taken place after 2005 for all three types of operations, but particularly for V-non-TAP and V-TAP, this share is close to 70%. As for the incidence of prior palliative surgery, only marginal differences (P = 0.053) are noted between the groups, with slightly more patients treated with ventriculotomy-tap having had prior palliation. Figure 2 shows the evolution of the number of operations performed yearly, as well as the evolution of the percentage prevalence of each procedure type (V-TAP, V-non-TAP and No-V) between 1999 and Percentage prevalence was defined as the number of operations of each type over the total number of operations of any of the three types in each year. Reporting to the database for year 2011 was incomplete at the time of writing this paper. The relationship between the number of operations with the type of surgery and time was explored by fitting a linear model (R 2 = 0.83) with the number of operations as the dependent variable and the interaction between time and the type of surgery as the independent variable (Number of operations Year:Type CONGENITAL

3 768 G.E. Sarris et al. / European Journal of Cardio-Thoracic Surgery Table 1: Summary of cumulative results of reparative operations for TOF from the EACTS CDB Operation abbreviation Operation name Total number of operations OM Hospital mortality n % n % n % V-TAP TOF repair, ventriculotomy, transanular patch V-non-TAP TOF repair, ventriculotomy, non-transanular patch No-V TOF repair, no ventriculotomy Total (V-TAP, V-non-TAP and No-V) RV PA TOF repair, RV PA conduit NOS TOF repair, repair not otherwise specified Total (RV PA, NOS) Total (all) V-TAP: ventriculotomy with TAP; V-non-TAP: ventriculotomy with a non-tap; No-V: no ventriculotomy; RV-PA: RV-PA conduit; NOS: repair not otherwise specified; TOF: tetralogy of Fallot; OM: 30-day mortality. All operations: n = Figure 1: Distribution of reparative operations for tetralogy of Fallot according to the procedure type. V-TAP: ventriculotomy with TAP; V-non-TAP: ventriculotomy with a non-tap; No-V: no ventriculotomy; RV PA: RV to PA conduit; NOS: repair not otherwise specified. of surgery). Results reveal statistically significant relationship between the interaction of the two variables with the number of observations. While coefficients (estimate) for Year:V-non-TAP and Year:No-V are almost equal (β = 0.05), the coefficient of the term Year:V-TAP is almost three times larger than the coefficients of the previous two (β = 0.15). Hospital characteristics Information about the distribution of the operations among the three types of surgeries in the total of 119 hospitals has also been analysed to identify the extent of departures from the average distribution of the three main procedure types of interest (V-TAP, V-non-TAP and No-V) in the total number of operations of these three types. The average distribution of the total reparative procedures into the three types is 6:2:2 (V-TAP, V-non-TAP and No-V, Table 1). Analysis revealed that hospitals differ in a statistically significant way with regards to the distribution of operations into the three types of surgeries (X 2 = , df = 194, P < 2.2e-16; we have omitted 21 hospitals from this analysis due to the very small number (<5) of total reparative procedures (V-TAP, V-non-TAP and NO-V) in order to have more robust statistical conclusions). Therefore, the parameter hospital played a significant role in the selection of the type of reparative surgery performed. Due to the large number of hospitals, it is difficult to examine the degree of departure from the standard distribution for each hospital separately. We have therefore produced box plots (Fig. 3), which summarize the prevalence (%) of each type of surgery (V-TAP, V-non-TAP and No-V) in the total of 98 hospitals with more than five operations of any of the three surgeries. Median values (i.e. horizontal line inside each box) are close to the anticipated theoretical distribution (60% for V-TAP, 16.8% for V-non-TAP and 14.6% for No-V). Each box represents 50% of the observations, and in particular, it extends from the 25th to the 75th percentile. The box for V-TAP is somewhat wider than the other two boxes, implying larger inter-quartile range. Distributions for V-non-TAP and No-V are skewed right (i.e. to small prevalences), while the distribution for V-TAP is slightly skewed left. In 25% of the hospitals, the prevalence of V-TAP is <43.9%, with two outliers with zero prevalences. For V-non-TAP, in 25% of the hospitals, the prevalence of that surgery is >32.8% with three outliers (75, 87.5 and 100%). Finally, in 25% of

4 G.E. Sarris et al. / European Journal of Cardio-Thoracic Surgery 769 Table 2: Characteristics of the different groups of patients that have undergone reparative operations for TOF: V-TAP, V-non-TAP and No-V V-TAP V-non-TAP No-V n % n % n % Year of surgery X 2 = , df = 2, P < History of previous palliative surgery Yes X 2 = , df = 2, P = No Age (days: mean ± SE) ± ± ± F = 23.92, P < V-TAP: ventriculotomy with TAP; V-non-TAP: ventriculotomy with a non-tap; No-V: no ventriculotomy; SE: standard error. Figure 2: Yearly evolution of the number of reparative operations for TOF and of the yearly percentage prevalence of the three types of procedures. V-TAP: ventriculotomy with TAP; V-non-TAP: ventriculotomy with a non-tap; No-V: no ventriculotomy ( ). Note: Reporting for 2011 is incomplete at the time of writing this paper. CONGENITAL hospitals, the prevalence of No-V is >27.7% with three outliers (73.8, 95.2 and 97%). Analysis of mortality Overall HM was 2.58% (172 of 6654; Table 1). If reparative operations utilizing an RV PA conduit (which may have been performed on many patients of TOF with pulmonary atresia) and operations whose type was not defined (coded as NOS) are excluded, overall HM is 2.47% (157 of 6350). Mortality according to the type of operation is also listed. In further statistical analyses, operations utilizing RV PA conduits (N = 216) and those coded as NOS (N = 88) were excluded. Table 3 summarizes the statistics of surgical mortality assessed by OM and HM, respectively, and only for the three main types (V-non-TAP, V-TAP and No-V). Surgical mortality has been analysed with respect to the year of surgery, age (days) and history of previous palliative surgery. Comparison has been made on the distribution of these characteristics between the

5 770 G.E. Sarris et al. / European Journal of Cardio-Thoracic Surgery subpopulation of cases of mortality (either OM or hospital mortality ) and the remaining patients who survived. Results reveal that the two populations (mortalities vs survivals) do not differ significantly for any of these characteristics. Only for the year of surgery differences are marginally significant (P = 0.053), with somewhat higher mortality rates for operations performed before 2005 (2.88%) compared with those performed after 2005 (2.06%). When HM is analysed, findings are similar to those for OM. No significant differences are observed in the age distribution and previous surgery between incidents of death and incidents of no death. The year of surgery is again only marginally important (P = 0.059) with slightly increased mortality rates for years before 2005 (3%) compared with years after 2005 (2.2%). The results of logistic regression (Table 4) reveal statistically significant association between mortality risk (OM) and the type Figure 3: Summary of hospitals according to the prevalence (%) of each type of surgery. V-TAP: ventriculotomy with TAP; V-non-TAP: ventriculotomy with a non-tap; No-V: no ventriculotomy. Table 3: Characteristics of the study population (V-TAP, V-non-TAP and No-V) 30-day mortalities (n = 148) Survivals (n = 6202) n % n % of surgical repair. In particular, V-TAP is associated with significantly increased risk of death compared with No-V (crude OR 2.10 [95% CI: ]. Allowing for confounders (year reparative surgery was performed, age and history of previous palliative surgery), no important changes in ORs for V-TAP are observed. The odds of dying for patients that have been operated with V-non-TAP were 1.09 times more than the odds for those operated with No-V; however, this finding is not statistically significant (95% CI: ). When HM is considered, findings are similar: V-TAP is associated with significantly increased risk of HM, with crude OR 2.13 (95% CI: ). When adjusted for year reparative surgery was performed, age and history of previous palliative surgery, OR increases to 2.21 (95% CI: ). Multivariate analysis showed a significant impact on the risk of both OM and HM of the year of reparative surgery: in the multivariate models in Table 4 where the year of surgery is included, there is statistically significant evidence that patients operated before 2005 have increased mortality risk compared with those operated from year 2005 onwards (1.45, 95% CI: for OM and, for HM, 1.42, 95% CI: ). The analysis presented herein has been performed in a dataset extraction of the EACTS CDB. This dataset of 6654 records has not been verified according to the source data verification protocol, which has been employed until now only in a small subset of the total data in the database. Thus, the verified data subset of the overall data in the 6654 records analysed above covers a shorter period ( ) and is considerably smaller (n = 678), precluding sound statistical analysis. For this verified data subset, Table 5 presents a summary of the numbers of reparative operations for TOF (V-TAP, V-non-TAP, No-V, RV PA and NOS), along with corresponding mortality rates (OM and HM). Operative (30 days) mortality for V-non-TAP and No-V is the same (1.05%), while for V-TAP, OM is larger (1.87%). HM for V-non-TAP is largest (2.11%), followed by V-TAP (1.87%) and Year of surgery X 2 = , df = 1, P = History of previous palliative surgery a Yes Fisher s exact test: P = No Age (days: mean ± SE) ± ± t = , df = , P = Hospital mortalities (n = 157) Survivals (n = 6193) n % n % Year of surgery X 2 = , df = 1, P = History of previous palliative surgery Yes Fisher s exact test: P = 1 No Age (days: mean ± SE) ± ± t = 0.06, df = , P = a Previous palliative surgical procedure. SE: standard error.

6 G.E. Sarris et al. / European Journal of Cardio-Thoracic Surgery 771 Table 4: Crude and adjusted ORs Table 5: No-V (1.58%). However, these figures should be interpreted with caution due to the very small number of incidents of death. DISCUSSION Summary of cumulative results of reparative operations for TOF from the EACTS CDB: verified data Operation abbreviation Operation name Total number of operations Many studies have documented excellent early results for the surgical repair of TOF [1 18]. However, there is increasing awareness of and concern regarding the incidence of significant late complications, including pulmonary valve insufficiency, RV dilatation and dysfunction, ventricular arrhythmias and sudden death [19 23]. Management of these complications requires major reoperations in an increasing number of patients. With surgical mortality for tetralogy repair generally reduced to low levels in most reports, emphasis is shifting towards strategies to minimize the incidence of late complications, thus fuelling a continued debate on the optimal management strategy, including timing and technique of repair. Many publications support the complete primary repair of TOF in infancy, even in neonates 30-day mortality Crude OR 95% CI Adjusted OR a 95% CI Adjusted OR b 95% CI No-V V-non-TAP V-TAP Age Year of surgery: Previous palliative surgery: Yes Hospital mortality Crude OR 95% CI Adjusted OR a 95% CI Adjusted OR b 95% CI No-V V-non-TAP V-TAP Age Year of surgery: Previous palliative surgery: Yes OR: odds ratios; CI: confidence interval; No-V: no ventriculotomy; V-non TAP: ventriculotomy with a non-tap; V-TAP: ventriculotomy with TAP. a Adjusted for year of surgery and age. b Adjusted for year of surgery, age and previous palliative surgery. OM Hospital mortality n % n % n % V-TAP TOF repair, ventriculotomy, transanular patch V-non-TAP TOF repair, ventriculotomy, non-transanular patch No-V TOF repair, no ventriculotomy Total (V-TAP, V-non-TAP and No-V) RV PA TOF repair, RV PA conduit NOS TOF repair, NOS Total (RV PA, NOS) Total (all) OM: 30-day mortality; V-TAP: ventriculotomy with TAP; V-non-TAP: ventriculotomy with a non-tap; No-V: no ventriculotomy; RV-PA: RV-PA conduit; NOS: repair not otherwise specified; TOF: tetralogy of Fallot. [3 5, 12, 15], but concerns regarding possible increased morbidity and mortality for very early repair has led others to advocate a selective, non-neonatal approach [10, 13, 14, 16]. Furthermore, the increased use of a large TAP during neonatal and very early repair may be associated with greater degrees of PI and, ultimately, may lead to greater reoperation rates for pulmonary valve replacements. Attempts to minimize this possibility have included the implantation of various monocusp valves at repair ( pericardial, allograft or polytetrafluoroethylene), but the longterm effects of these efforts on the ultimate rates of reoperation are not proven. On the other hand, there has been increasing enthusiasm for adoption of the TA TP repair technique, which was initially introduced by Hudspeth et al. [7] and Kawashima et al. [8, 9] and largely popularized by Mee [10]. This technique obviates the use of a ventriculotomy, focusing on the maximal preservation of the RV and pulmonary valve. Its proponents have reported low-mortality rates, even in neonates (although most prefer non-neonatal repair), and suggest that the incidence of late RV dysfunction, arrhythmias and reoperation for pulmonary valve CONGENITAL

7 772 G.E. Sarris et al. / European Journal of Cardio-Thoracic Surgery replacement may be reduced [9, 10, 12 14, 16]. While these controversies persist, with proponents of alternative approaches focusing on future largely theoretical benefits, there is a paucity of data regarding the actual prevalence of the various surgical techniques currently available for surgical repair of TOF outside of North America [18], and the corresponding early results of these approaches, let alone truly long-term outcome. Our data demonstrated that repair of TOF as reported in the ECDB over the last decade is accomplished with low mortality, which appears to be decreasing. Regarding the types of procedures employed and their associated surgical mortality, ventriculotomy with TAP remains the most prevalent technique (57.5%), and it is associated with higher mortality (2.9 and 3.11%) than the other techniques. Transventricular repair (with or without TAP) accounts for more than three quarters of all repair procedures (77%). Perhaps somewhat surprisingly, given rising recent concerns regarding the potential long-term adverse effects of a ventriculotomy, repair without ventriculotomy was less prevalent overall and with decreasing prevalence in the more recent data, despite lower mortality. Clearly, patients who had an RV PA conduit represent a different subgroup and have been excluded from more detailed analyses. Of course, one must be careful not to over-interpret these data, which clearly do not permit an accurate direct comparison of the relative efficacy and safety of the various procedures reported, since the underlying patient populations subjected to different procedures are not necessarily comparable. The ECDB is set-up as a registry and not as a research database, and therefore, it does not include data fields for important anatomical and pathophysiological parameters relevant to detailed research investigations of the management of TOF. Despite this significant limitation, analysis of these data raises some important points, which suggest that differences in observed mortalities associated with different types of repair cannot be dismissed by ascribing them exclusively to the clinical and anatomical patient substrate. First, the choice of the type of procedure employed does not appear to be dictated only by patient factors, as there is a highly statistically significant influence of the hospital on the distribution of the reparative procedures among the tree main types. Indeed, these data show that there are hospitals which perform almost exclusively a V-TAP type of repair, and others which perform almost exclusively a No-V type of repair. This finding is in accordance with the well known fact that individual centres strongly favour or even exclusively use one or the other surgical technique for all patients with classical TOF. Furthermore, no differences were found between the patients subjected to the three main types of reparative procedures with regard to certain preoperative factors, which may have had an important influence on mortality, such as year of surgery or history of prior palliative surgery. An interesting exception was age at surgery, with patients who had undergone repair by ventriculotomy non-tap being significantly older. As it is known that transventricular neonatal or early infant repair of TOF is associated with increased use of a TAP, this finding may indicate that patients who underwent V-non-TAP may have represented a more anatomically favourable subgroup, possibly explaining the associated lower mortality vs that of the V-TAP group. Similarly, one could also argue that the No-V group, that is, the TA TP repair group, is less prevalent and has better mortality possibly because it includes anatomically and pathophysiologically more favourable surgical substrates. These speculative arguments, however, cannot be further corroborated in the absence of detailed anatomical data (such as pulmonary valve annulus size) recorded in the ECDB. Interestingly, in an important similar analysis of data from the Society of Thoracic Surgeons Database, the findings of Al Habib et al. [18] were similar to those of the present study. These authors showed that, in North American congenital heart surgery centres, overall discharge mortality was 1.3% for primary TOF repair, and 0.9% for repair following earlier palliation. These overall mortality figures for TOF repair are lower than those reported in the ECDB and noted in our study, but the reasons for this difference are not clear. Mortality for palliation overall was 7.5%, while mortality for palliation in neonates was 6.2%, and mortality for primary repair in neonates was 7.8%. With regard to the types of reparative procedures used, primary repair in the first year of life was the most prevalent strategy, while ventriculotomy with TAP was the most prevalent technique (52%), with ventriculotomy without TAP being as frequent as the no-ventriculotomy technique. As in our study, repair with ventriculotomy represented three quarters of repairs. It is important to emphasize that this study did not aim and could not provide information regarding differences in the outcome of surgery for TOF between institutions or countries. Also, the study did not seek to and made no attempt to limit its analysis to data from European participating centres, or to make comparisons between European and non-european centres, which contributed 30% of the patients. However, as shown, the verified data subset, which is derived exclusively from European centres listed on the web site of the ECDB, shows similar surgical mortality patterns as the entire dataset. Furthermore, this study did not intend and cannot provide conclusions or recommendations regarding the relative value with regards to early outcome of the different surgical options studied, despite the observed differences in mortality, which may be suggestive but actually only provide a description of current practice. The study does not attribute observed mortality differences solely or primarily to the surgical approaches employed, and it does not advocate any one of the available types of repairs, which are all suitable for the treatment of a wide spectrum of patients with TOF. Furthermore, the study includes no data on long-term outcome and makes no statements pertaining to the probability of late phase events, such as reoperation. Clearly, to resolve persisting controversies regarding optimal therapy for patients with TOF and to provide evidence-based answers to important questions regarding the effect of various management strategies and surgical techniques on both early and, most importantly, late outcome, it seems necessary to organize large, multicenter prospective studies, preferably in collaboration of congenital cardiac surgical and paediatric cardiology scientific societies, as truly long-term follow-up measured in decades will probably be required. LIMITATIONS The limitations of this study are direct consequences of the fact that it is a retrospective observational study with the source of data (the EACTS CDB) being a multicenter registry with predefined data fields. In this setting, it is not possible to determine the precise extent to which the various operations performed to repair TOF

8 G.E. Sarris et al. / European Journal of Cardio-Thoracic Surgery 773 represent the preference of the repair technique by individual surgeons and congenital heart surgery centres, or imposed by individual patient anatomy, pathophysiology and overall clinical status. Furthermore, operations for repair of TOF are categorized in the database according to the use (or non-use) of a right ventriculotomy and the use of a TAP. Importantly, the length of a transannular incision onto the anterior wall of the RV is not specified and therefore, not coded. It is possible that many operations categorized as no ventriculotomy may have actually employed a limited incision onto the RV infundibulum (though most probably, not onto the body of the RV itself). Some may argue that these cases should be coded as ventriculotomy with TAP cases, and that many centres may have coded such cases precisely in this way. However, to the extent that this is indeed the case, and that such cases (which could be considered anatomically favourable by virtue of the fact that only a minimal transannular incision was required) may have actually been coded as ventriculotomy with TAP cases, this should have had the effect of dilution of the apparently more risky ventriculotomy procedures with the lower risk ones, artificially lowering mortality in the ventriculotomy-tap group. Importantly, the data subjected to detailed analysis is derived from the entire dataset of the database, and not from the subset of verified data. However, previous comparison of verified and unverified data in the EACTS CDB has not revealed the presence of any statistically significant differences [25]. In addition, the distribution of types of reparative procedures for TOF and associated mortalities is similar in the verified data subset and the entire TOF repair dataset. CONCLUSIONS This study demonstrated that, as reported in the ECDB over the last decade, repair of TOF is accomplished with low mortality, which appears to be decreasing. Of the various types of TOF repair employed, ventriculotomy with TAP remains the most prevalent technique reported by participating EACTS CDB Centres, and it is associated with higher early mortality. TOF repair without ventriculotomy is both less prevalent and is associated with lower mortality. Clearly, only large multicenter prospective studies focusing on many parameters of early as well as late outcome and with follow-up extended to decades could confirm any possible early and, most importantly, late benefits of a no ventriculotomy approach. ACKNOWLEDGEMENTS This study would not have been possible without the numerous congenital heart surgeons and centres who have contributed their patient data voluntarily to the EACTS CDB. We like to thank Tonia Ieromnimon (ieromnia@gmail.com) for expert biostatistical support. Conflict of interest: none declared. REFERENCES [1] Kirklin JW, Ellis FH Jr, McGoon DC, DuShane JW, Swan HFC. Surgical treatment for the tetralogy of Fallot by open intracardiac repair. J Thorac Surg 1959;37: [2] Shumway NE. Total surgical correction of tetralogy of Fallot. Arizona Med 1966;S106. [3] Castaneda AR, Freed MD, Williams RG, Norwood WI. Repair of tetralogy of Fallot in infancy. Early and late results. J Thorac Cardiovasc Surg 1977; 74: [4] Reddy VM, Liddicoat JR, McElhinney DB, Brook MM, Stanger P, Hanley FL. Routine primary repair of tetralogy of Fallot in neonates and infants less than three months of age. Ann Thorac Surg 1995;60:S [5] Pigula FA, Khalil PN, Mayer JE, del Nido PJ, Jonas RA. Repair of tetralogy of Fallot in neonates and young infants. Circulation 1999;100: II [6] Hirsch JC, Mosca RS, Bove EL. Complete repair of tetralogy of Fallot in the neonate: results in the modern era. Ann Surg 2000;232: [7] Hudspeth AS, Cordall AR, Johnston FR. Transatrial approach to total correction of Tetralogy of Fallot. Circulation 1963;27: [8] Kawashima Y, Kitamura S, Nakano S, Yagihara T. Corrective surgery for tetralogy of Fallot without or with minimal right ventriculotomy and with repair of the pulmonary valve. Circulation 1981;64(Suppl. 12): [9] Kawashima Y, Kobayashi J, Matsudo A. Long term evaluation after correction of tetralogy of Fallot. Kyobu Geka 1990;43: [10] Karl TR, Sano S, Pornviliwan S, Mee RB. Tetralogy of Fallot: favorable outcome of nonneonatal transatrial, transpulmonary repair. Ann Thorac Surg 1992;54: [11] Pacifico AD, Sand ME, Bargeron LM, Calvin EC. Transatrial transpulmonary repair of tetralogy of Fallot. J Thorac Cardiovasc Surg 1987;93: [12] Stellin G, Milanesi O, Rubino M, Michielon G, Moreolo GS, Boneva R et al. Repair of tetralogy of Fallot in the first six months of life: transatrial versus transventricular approach. Ann Thorac Surg 1995;60:S [13] Fraser CD Jr, McKenzie ED, Cooley DA. Tetralogy of Fallot: surgical management individualized to the patient. Ann Thorac Surg 2001;71: [14] Giannopoulos NM, Chatzis AK, Karros P, Zavaropoulos P, Papagiannis J, Rammos S et al. Early results after transatrial/transpulmonary repair of tetralogy of Fallot. Eur J Cardiothorac Surg 2002;22: [15] Kanter KR, Kogon BE, Kirshbom PM, Carlock PR. Symptomatic neonatal tetralogy of Fallot: repair or shunt? Ann Thorac Surg 2010;89: [16] Sfyridis PG, Kirvassilis GV, Papagiannis JK, Avramidis DP, Ieromonachos CG, Zavaropoulos P et al. Preservation of right ventricular structure and function following transatrial-transpulmonary repair of tetralogy of Fallot. Eur J Cardiothorac Surg 2012; doi: /ejcts/ezs221. [17] Boni L, Garcia E, Galletti L, Perez A, Herrera D, Ramos V et al. Current strategies in tetralogy of Fallot repair: pulmonary valve sparing and evolution of right ventricle/left ventricle pressures ratio. Eur J Cardiothorac Surg 2009;35: [18] Al Habib HF, Jacobs JF, Mavroudis C, Tchervenkov CI, O Brien SM, Mohammadi S et al. Contemporary patterns of management of Tetralogy of Fallot: data from the Society of Thoracic Surgeons Database. Ann Thorac Surg 2010;90: [19] Fuster V, McGoon DC, Kennedy MA, Rotter DG, Kirklin JW. Long-term evaluation (12 to 22 years) of open heart surgery for tetralogy of Fallot. Am J Cardiol 1980;46: [20] Hickey EJ, Veldman G, Bradley TJ, Aungkana G, Manlhiot C, Williams WG et al. Late risk of outcomes for adults with repaired tetralogy of Fallot from an inception cohort spanning four decades. Eur J Cardiothorac Surg 2009;35: [21] d Udekem Y, Ovaert C, Grandjean F, Gerin V, Cailteux M, Shango-Lody P et al. Tetralogy of Fallot: transannular and right ventricular patching equally affect late functional status. Circulation 2000;102(19 Suppl. 3): III [22] Bacha EA, Scheule AM, Zurakowski D, Erickson LC, Judy Hung J, Lang P et al. Long-term results after early primary repair of tetralogy of Fallot. J Thorac Cardiovasc Surg 2001;122: [23] Lindberg HL, Saatvedt K, Seem E, Hoel T, Birkeland S. Single-center 50 years experience with surgical management of tetralogy of Fallot. Eur J Cardiothorac Surg 2011;40: [24] Jacobs ML. Congenital Heart Surgery Nomenclature and Database Project: tetralogy of Fallot. Ann Thorac Surg 2000;69:S CONGENITAL

9 774 G.E. Sarris et al. / European Journal of Cardio-Thoracic Surgery [25] Maruszewski B, Lacour-Gayet F, Monro JL, Keogh BE, Tobota Z, Kansy A. An attempt at data verification in the EACTS Congenital Database. Eur J Cardiothorac Surg 2005;28: APPENDIX. CONFERENCE DISCUSSION Dr J. Jacobs (St. Petersburg, FL, USA): This presentation and your manuscript are important. They represent another example of the tremendous partnership between the STS and the EACTS. It is notable that both the STS congenital heart surgery database and the EACTS congenital heart surgery database have transitioned from publishing papers about database methodology, as was common five to ten years ago, to publishing papers about the actual analysis of surgical outcomes. Our sister databases have published data about outcomes in multiple diagnostic and procedural cohorts including AV septal defects, heterotaxy, transposition, Norwood, Fontan, Down s syndrome, and now tetralogy of Fallot. As you know, the STS published a similar manuscript in the Annals entitled Contemporary patterns of management of tetralogy of Fallot: data from the STS Database. In this analysis, we examined patients undergoing 3059 operations for tetralogy of Fallot. We also found that, despite contemporary awareness of the late consequences of pulmonary insufficiency, ventriculotomy with transannular patch remains the most prevalent technique. Based on your excellent analysis, I would like to ask you two questions. First, based on these data, can you truly recommend which type of operation to perform for the patient with tetralogy of Fallot who is suitable for either a transatrial or transventricular repair? Dr Sarris: I take this opportunity to first of all thank all the contributors to the database because it is the data from the surgeons and centres that make this study possible. To answer your question, the data demonstrate that repair without ventriculotomy is associated with lower mortality. Of course, this supports the hypothesis that if repair can be achieved without any incision on the right ventricle, results will be better. I personally believe that this supports efforts for the transatrial or the no-ventriculotomy approach, and we certainly use this technique in our unit. We believe it is feasible to do in practically all patients with classical tetralogy, and outcomes in our unit and in many others support that early results are very good with very low mortality. However, to be fair, these data cannot be used to absolutely recommend transatrialtranspulmonary repair: in other words, they do not prove that this method is better than the ventriculotomy approach. As I already mentioned, the reason is that the dataset is a retrospective dataset with very defined data fields. The patient groups subjected to the different operations are not necessarily comparable, and differences in mortality could reflect differences not only in surgeon preferences and centre preferences, but also differences in the underlying pathophysiology of the patients. In short, I believe that the data support the idea of a transatrial-transpulmonary, no-ventriculotomy approach, but they do not prove that such an approach is superior. Dr Jacobs: I certainly agree with that interpretation of the data, and that leads to my follow-up final question. If you could add fields of data to our sister databases to provide more information about tetralogy of Fallot, what fields of data would you add? Dr Sarris: This is a critically important question. If one wanted to compare the different types of operations with regard to outcome, one would need to have information that should be encoded in the database on whether the operations performed were elective or mandated by symptoms, the types of symptoms, and the degree of saturation, arterial oxygen saturation of the patient. Also included should be important anatomical variables such as the size of pulmonary arteries, the size of the annulus, the features of the pulmonary valve, and some very important immediate postoperative variables such as RV-LV pressure ratio, residual RVOT gradient or degree of pulmonary valve insufficiency. And these data, again, would only be sufficient to compare early outcome, but the perhaps the more important question in comparing these different approaches is whether the late outcome is affected by the differences in early approach. And, therefore, one would need, I think, to make provisions for including data fields and prospectively longitudinally follow these patients and encode late events such as reintervention. Dr P. Vouhé (Paris, France): I m sorry, George. You are taking up a lot of time with the questions. We have to be strict. We are over time already by three minutes. Another question, please. Dr V. Hraska (Sankt Augustin, Germany): Frankly, I have a problem with the interpretation of this study. I strongly believe that the surgical approach is determined by underlying morphology of the right ventricular outflow tract and not by whether we do it transatrial or transannular. If the annulus is hypoplastic, George, what would you do? You have to transect the annulus, yes? And whether you call it mini transannular patch or mini ventriculotomy patch, mini-transannular approach, or mini ventriculotomy approach, it is a transannular approach. So to say that the transannular approach is more dangerous I think you do not have data for that. Dr Sarris: Although this argument could be used semantically, I think there is a real difference. A transatrial-transpulmonary approach, in contrast to the transventricular approach, would only violate the annulus (and therefore the fibres of the right ventricle) to a very minimal extent. It would be entirely insufficient to perform VSD closures through such mini trans-annular incisions, so I think these two are very different operations. Dr F. Lacour-Gayet (New York, NY, USA): I have one comment and one question. My comment is this database study can answer the question do we do things right?, but will not answer the question do we do the right thing? To know whether we do the right thing with a transannular-transpulmonary versus ventriculotomy approach, we need a long-term follow-up study that we do not have today. On the other hand, my question relates to what the database can do because I think the EACTS database particularly, included morbidity. What would be interesting to know is what the difference in morbidity is between the procedure of the transannular-tricuspid approach and the transventricular approach. In the morbidity difference is there some evidence that manipulating the tricuspid valve, pulling on the valve, is generating more arrhythmia, particularly atrial arrhythmia? Dr Sarris: I think it is a good comment. I agree that long-term information will be necessary. In general, reports of long-term results of the transatrial approach are remarkable for the absence of ventricular arrhythmias, but this ought to be studied prospectively. Dr T. Spray (Philadelphia, PA, USA): Only one quick comment, and that is about the database. The problem with this study is that the database does not allow you to code a diagnosis of a transatrial-transpulmonary repair with a minimal transannular patch, because if I code that, it requires coding a ventriculotomy. As soon as you cross the annulus, it is a ventriculotomy. So if you code that way, then you will have an increased number of ventriculotomy patients in the database, which is what you have shown in the later years. So it is problem. I think there needs to be a new code in the database to address this problem.