|Home | About | Journals | Submit | Contact Us | Français|
The influence of atrioventricular septal defect (AVSD) subtype on outcomes after repair is poorly understood.
Demographic, procedural, and outcome data were obtained 1 and 6 months after AVSD repair in an observational study conducted at 7 North American centers.
The 215 AVSD patients were subtyped as 60 partial, 27 transitional, 120 complete, and 8 with canal-type VSD. Preoperatively, transitional patients had the highest prevalence of moderate or severe left atrioventricular valve regurgitation (LAVVR, p = 0.01). At repair, complete AVSD and canal-type VSD patients, both with the highest prevalence of trisomy 21 (p < 0.001), were younger (p < 0.001), had lower weight-for-age z scores (p = 0.005), and had more associated cardiac defects (p < 0.001). Annuloplasty was similar among subtypes (p = 0.91), with longer duration of ventilation and hospitalization for complete AVSD (p < 0.001). Independent predictors of moderate or severe LAVVR at the 6-month follow-up were older log(age) at repair (p = 0.02) but not annuloplasty, subtype, or center (p > 0.4). Weight-for-age z scores improved in all subtypes at the 6-month follow-up, and improvement was similar among subtypes (p = 0.17).
AVSD subtype was significantly associated with patient characteristics and clinical status before repair and influenced age at repair. Significant postoperative LAVVR is the most common sequela, with a similar prevalence across centers 6 months after the intervention. Annuloplasty failed to decrease the postoperative prevalence of moderate or severe LAVVR at 6 months. After accounting for age at repair, AVSD subtype was not associated with postoperative LAVVR severity or growth failure at 6 months. Further investigation is needed to determine if interventional strategies specific to AVSD subtype improve surgical outcomes.
Atrioventricular septal defects (AVSD) include a spectrum of abnormalities involving the atrial and ventricular septa and AV valves . Techniques used for repairing AVSD subtypes vary among surgeons . AVSD surgical outcomes have been the focus of benchmarking efforts and are used as a surrogate for institutional expertise in the management of congenital heart operations [3, 4, 5], but differences among the subtypes have not been systematically studied in the modern era. Using a multicenter observational database, we sought to compare patient characteristics, resource utilization, center differences, and outcomes among AVSD subtypes and to identify predictors of postoperative left AV valve regurgitation (LAVVR).
Between June 2004 and February 2006, all patients with AVSD undergoing biventricular repair at 7 North American centers were screened for potential enrollment into a drug trial designed to evaluate the effect of angiotensin-converting enzyme inhibitors on LAVVR and supported by the National Heart, Lung and Blood Institutes (NHLBI)–funded Pediatric Heart Network. Before the drug trial was initiated, echocardiographic and clinical data were collected 1 and 6 months postoperatively in an observational phase, as previously described [4, 5]. Briefly, data were collected prospectively collected for 178 patients who were participants in the planned drug trial and retrospectively for 37 screened patients from the same period with similar baseline characteristics who were not enrolled. Informed consent was obtained from parents/guardians for prospectively enrolled patients. A waiver of consent was obtained for retrospectively enrolled patients. The information in the database was obtained under the supervision of an independent Data and Safety Monitoring Board and approved by the center’s Institutional Review Board (Clinical Trials.gov Identifier: NCT00113698).
Demographic data, including race/ethnicity, sex, presence of trisomy 21, and age and weight at operation and at the 6-month follow-up echocardiogram were recorded. Weight-for-age z scores were calculated using sex-specific reference values for trisomy 21  and normal children , as appropriate. Days of mechanical ventilation, intensive care, and total hospitalization were recorded to estimate resource utilization.
Three investigators independently reviewed all operative reports for AVSD subtype, cleft closure, annuloplasty, and additional operations for associated defects. The Society of Thoracic Surgeons nomenclature was used to define AVSD subtypes [4, 5, 8]: partial AVSD (PAVSD) had a primum atrial septal defect (ASD) and cleft left AV valve (LAVV); transitional AVSD (TAVSD) had a primum ASD, restrictive inlet VSD and 2-orifice AVV; complete AVSD (CAVSD) had a primum ASD, inlet VSD, and single orifice AVV; and canal-type VSD had an inlet VSD, common AV junction, cleft LAVV, and no primum ASD.
Because the operative notes often contained insufficient detail regarding annuloplasty technique, annuloplasty was defined as any additional operation on the LAVV beyond cleft closure, such as annuloplasty along the lateral leaflet with absorbable suture, sutures at the superior and inferior bridging-lateral leaflet commissures. Total cardiopulmonary bypass and aortic cross-clamp times and return to bypass were recorded. Reoperations performed during the 6-month follow-up were noted.
Echocardiographic data were collected as previously reported [4, 5]. Local investigators reviewed all preoperative echocardiograms. Echocardiograms performed 1 month postoperatively were reviewed locally for the participants retrospectively enrolled and in the core laboratory for those prospectively enrolled. Only prospectively enrolled patients had 6-month echocardiograms and all were reviewed in the core laboratory. The degree of LAVVR was graded as none/trace, mild, moderate, or severe (moderate or severe was considered significant). Color Doppler imaging was also used to demonstrate a residual ASD or VSD (diameters > 3 mm were considered significant). Pulsed Doppler interrogation was used to evaluate LAVV inflow and left ventricular (LV) outflow. A LAVV mean inflow gradient of 7.5 mm Hg or higher and a LV outflow peak gradient of 20 mm Hg or higher were considered significant.
Descriptive statistics are presented as mean ± standard deviation, median and interquartile range, or frequencies and percentages. The distributions of categoric variables by AVSD subtype were compared using the Fisher exact or χ2 test. Ordinal variables were analyzed using the Mantel-Haenszel test for linear trend. Analysis of variance and the Kruskal-Wallis test were used to compare normally distributed and skewed variables by subtype, respectively. The one-sample t test was used to determine whether change in the weight-for-age z score differed from 0 for each subtype. Univariate and multivariable logistic regressions were used to identify patient and surgical factors associated with the outcome of moderate or severe LAVVR at the 6-month follow-up. With all participants combined, interactions between defect subtype and prespecified factors (age at operation, annuloplasty, and preoperative and 6-month postoperative LAVVR) were evaluated. Linear regression was used to identify predictors of log-transformed hospital days in patients who underwent operations age 1 month or older.
All variables significant at the 0.20 level in univariable analysis were evaluated for inclusion in multivariable regression models. A test of interaction was used to assess whether the association of age at repair with hospital outcomes depended on the number of associated cardiac anomalies. Exploratory data analyses were conducted using generalized additive models. Stepwise multivariable linear regressions were performed for each defect subtype to explore factors associated with the 6-month change in weight-for-age z score. The R2 values reported are adjusted R2 and maximum rescaled R2 for linear and logistic regression, respectively. All analyses were performed using SAS 9 software (SAS Institute, Inc, Cary, NC).
The study enrolled 215 patients (Table 1) with the following AVSD subtypes: PAVSD, 60 (28%); TAVSD, 27 (13%); CAVSD, 120 (56%), and canal-type VSD, 8 (4%). Compared with other subtypes, PAVSD participants were significantly older at the time of repair. The CAVSD and canal-type VSD groups had a higher percentage of children with trisomy 21, significantly lower average weight-for-age z scores at operation, and significantly more participants underwent repair of additional cardiac lesions. Subgroups differed in mean cardiopulmonary bypass and cross-clamp times (p < 0.001).
Surgical approaches used in the CAVSD group included double-patch (72%), single-patch (18%), and the Australian technique (10%), and their effects on outcomes have been reported . When pairwise comparisons of the four group means were performed, PAVSD participants had the lowest bypass and cross-clamp times (for both: p = 0.007 vs TAVSD; p < 0.001 vs CAVSD and vs canal-type VSD). Mean times were lower in TAVSD participants (p = 0.02) than in CAVSD participants (p = 0.002), but were similar in CAVSD (p = 0.49) and canal-type VSD (p = 0.96). Cleft closure was performed in 75% with a canal-type VSD and in 93% or more of all other subtypes.
Six patients died, 1 with PAVSD and 5 with CAVSD, all with associated pathology [4, 5]. Median ventilation, intensive care unit, and hospital days were significantly longer in the CAVSD group compared with PAVSD and TAVSD groups (Table 1). At the 6-month follow-up, 1 TAVSD and 1 CAVSD patient had a residual ASD (> 3 mm) and 1 CAVSD patient had a residual VSD (> 3 mm). At the 6-month follow-up, 1 PAVSD patient had subaortic stenosis (peak gradient, 39 mm Hg), 1 TAVSD and 3 CAVSD patients had LAVV stenosis (mean gradient range, 7.7 to 10.9 mm Hg), and 1 CAVSD patient required a pacemaker for complete heart block.
Perioperative complications have been reported previously [4, 5]. Sepsis and pleural effusions were the most frequent complications for PAVSD (7%) and TAVSD (5%). For the CAVSD group, gastrointestinal complications (15%) and pleural effusions (14%) were the most common. Pleural effusion (2 of 8) was the most frequently reported complication for canal-type VSD.
Preoperatively, TAVSD subjects had the highest prevalence (44%) of moderate or severe LAVVR (Table 2) and canal-type VSD participants had the lowest (0 of 8). Reoperation for residual LAVVR within 6 months of initial repair was performed in 7 patients (Table 3). The 1-month postoperative echocardiogram revealed a decrease in the prevalence of moderate or severe LAVVR for PAVSD and TAVSD subtypes but an increase in CAVSD and canal-type VSD subtypes. At the 6-month follow-up, 39 of 150 (26%) overall had moderate or severe LAVVR. The prevalence of LAVVR was similar among centers. CAVSD Rastelli classifications (A = 63; B = 2; C = 27; unclassified = 28) were not associated with the prevalence of moderate or severe LAVVR at the 1-month or 6-month follow-up.
The rate of annuloplasty use was similar among AVSD subtypes (p = 0.91; Table 1), but annuloplasty use varied widely, from 3% to 49%, among the 7 centers (p < 0.001). Patients with moderate or severe LAVVR preoperatively were more likely to undergo annuloplasty than those with mild or none LAVVR (38% vs 20%; Table 4), with an odds ratio of 2.5 for annuloplasty (95% confidence interval [CI], 1.27 to 4.83, p = 0.008). Despite this, the use of annuloplasty was not associated with a significant reduction in the prevalence of moderate or severe LAVVR 6 months postoperatively (24% in the annuloplasty group vs 26% in the no annuloplasty group, p = 0.83).
Multivariable analysis showed that the strongest predictors of LAVVR at the 6-month follow-up varied among subtypes (Table 5). For PAVSDs, moderate or severe LAVVR on the preoperative and 1-month postoperative echocardiograms independently predicted moderate or severe LAVVR at 6 months. Older age at repair was the only independent predictor of significant LAVVR for TAVSD, and LAVVR at the 1-month postoperative echocardiography was the only independent predictor of LAVVR for CAVSD at the 6-month follow-up. To maximize the power of LAVVR predictors across all subtypes, modeling using all participants was conducted, with interaction terms evaluated for subtype by age at repair, preoperative LAVVR, 1-month postoperative LAVVR, and weight-for-age z score at operation. No interactions were significant.
The multivariable model based on all patients identified two independent predictors associated with moderate or severe LAVVR at the 6-month follow-up for all subtypes: older log(age) at repair (p = 0.02) and moderate or severe LAVVR on the 1-month postoperative echocardiography study (p < 0.001). Annuloplasty, center, return to bypass, degree of preoperative LAVVR, and subtype were not independent predictors of moderate or severe LAVVR at 6 months.
The interaction between age at repair and subtype was in part not significant because most patients (CAVSD) had repairs only in infancy. Therefore, we conducted nonparametric regression analyses of age at repair vs 6-month LAVVR separately for each subtype to qualitatively depict associations with age (Fig 1). The relationship between the highest prevalence of LAVVR at 6 months postoperatively and age at repair varied by subtype (Fig 1). For PAVSD, the peak prevalence clearly occurred in those undergoing repair after 4 years of age. For TAVSD and CAVSD, however, the prevalence of LAVVR increased linearly with increasing age at repair. These two subgroups accounted for 104 of 142 patients with 6-month LAVVR data and explain why model 4 (Table 5) found log(age) was significantly associated with outcome in logistic regression using the combined cohort.
All subtypes showed a similar increase in weight-for-age z scores after repair, with a mean change of 0.5 to 1.3 (p = 0.17, Fig 2), and the overall 95% CI for the improvement in weight-for-age z score was 0.94 to 1.17. Separate stepwise multivariable linear regression analyses were performed by subtype (Table 6) to explore potential factors associated with the change in weight-for-age z score (z score at 6 months minus z score at operation). Stratified analyses showed factors associated with weight gain varied among subtypes. For PAVSD, weight improved the most in those who had longer intensive care stays (p = 0.004). For TAVSD, the presence of trisomy 21 and no annuloplasty at repair were independently associated with a greater improvement in weight-for-age z score (p < 0.001 for both) and, along with center adjustment, accounted for 96% of the variation in outcome. For CAVSD, shorter hospital stay (p = 0.02) and lower pre-operative weight-for-age z score (p < 0.001) were independently associated with greater postoperative weight gain, accounting for 39% of the variation in outcome.
Median days of ventilation, intensive care, and hospitalization varied by subtype (p < 0.001, Table 1). Predictors of hospitalization days were examined after excluding 5 patients who required urgent repair because of associated defects before they were 1 month old (Table 7). No significant interaction between subtype and potential predictors was found. Analysis using data from all subtypes showed no cleft closure, longer bypass times, moderate or severe LAVVR on 1-month postoperative echocardiogram, and younger log(age) at repair were independently associated with longer hospitalization (R2 = 0.31). Annuloplasty was associated with more hospital days only for CAVSD (17.9 ± 24.9 vs 11.8 ± 16.5, p = 0.003).
Atrioventricular septal defects are one of the congenital heart defects for which outcomes are used to assess the pediatric heart surgeon’s skill as well as institutional expertise [3–5]. This multicenter observational study reports contemporary outcomes for all AVSD subtypes and analyzes patient characteristics, resource utilization, and practice variation as potential predictors of these outcomes.
After adjusting for subtype, centers were similar in age at repair and days of ventilation, intensive care, and hospitalization. The outcomes at 6 months compared favorably with those reported from single-center studies. Specifically, overall mortality was low, with only 6 deaths (3%) compared with the 7% to 15% mortality rate reported from other centers [9–13]. Septal defects (> 3 mm) occurred in 2% at the 6-month follow-up, similar to previous reports where residual defects occurred in 5% or fewer [12, 13]. Subaortic stenosis and LAVV stenosis were uncommon and lower than the 5% previously reported for these lesions , but this low prevalence may be attributed to the short duration of follow-up.
Residual LAVVR remains the Achilles’ heel of AVSD repair, with 3% (7 of 215) requiring reoperation and 26% with moderate or severe LAVVR by the 6-month follow-up. AVSD subtype and center were not independent predictors of the prevalence of moderate or severe LAVVR at 6 months. Despite the nearly uniform adoption of complete LAVV cleft closure, the prevalence of significant LAVVR has remained unchanged during the past decade [9, 15, 16]. To decrease LAVVR, some surgeons have advocated the addition of an annuloplasty [16, 17], but this remains controversial and its use varied widely among the 7 centers. For all subtypes, some surgeons performed annuloplasty prophylactically, even in those with mild or no preoperative LAVVR, some used it only if the preoperative LAVVR was graded as moderate or severe, and others did not use it at all. Regardless of the reason for its use, however, we were unable to demonstrate that annuloplasty decreased the prevalence of moderate or severe LAVVR at 6 months postoperatively. In fact, for reasons that are unclear, annuloplasty appeared to have some negative aspects in this study, where its use was associated with poorer weight gain in TAVSD patients and increased resource utilization in the CAVSD group.
Older log(age) at repair and the presence of moderate or severe LAVVR within 1 month of the operation predicted moderate or severe LAVVR at the 6-month follow-up. The prevalence of moderate or severe LAVV regurgitation clearly peaked after later operations for PAVSD but increased linearly for TAVSD/CAVSD. Other investigators have noted this trend and advocated earlier repair [9, 10]. Despite earlier repair during the past decade, however, the prevalence of postoperative LAVVR has remained unchanged [9, 15, 16]. Our data do not permit us to determine the ideal time of repair to minimize postoperative LAVVR.
Patient characteristics differed by subtype. CAVSD and canal-type VSD had a significantly more patients with trisomy 21, more associated cardiac defects, and longer cardiopulmonary bypass and cross-clamp times compared with PAVSD or TAVSD. We confirmed the commonly held belief that CAVSDs were more likely to have lower weight-for-age z scores at the time of repair. More surprising is the finding that weight gain significantly improved 6 months after repair in all subtypes, including those with PAVSD, and that all subtypes showed a similar increase in weight-for-age z scores. Although the predictors of improvement in weight varied with subtype, all subtypes likely share the common characteristic of being more nutritionally depleted before the operation and thus are able to display better catch-up growth after repair.
This study has several limitations: First, the small numbers of each subtype limited the ability to detect some associations and interactions of outcome predictors and subtype.
Second, because our data were collected both prospectively and retrospectively, all measurements could not be centrally interpreted, and 6-month postoperative echocardiographic data were not available for some participants. In addition, reliable and validated echocardiographic methods for quantitative evaluation of LAVVR grade are not available for children, particularly in the setting of multiple or eccentric jets characteristic of the repaired AVSD valve. We used qualitative assessment of the color Doppler jet because this was the standard for clinical decision making in all 7 centers. This has also been the most commonly used method for grading LAVVR in other AVSD studies, allowing for comparisons to be made.
Finally, we were unable to explore the effect of variations in annuloplasty technique on valve function because annuloplasty details were not uniformly available from the operative reports.
In conclusion, survival in the current era is excellent, with few residual defects after surgical repair for all AVSD subtypes at a similar age of repair among the 7 centers. After accounting for age at repair, we found subtype and center were not important prognostic indicators of resource utilization, postoperative LAVVR, or improvement in growth. Significant postoperative LAVVR is the most common sequela, with a similar prevalence across the centers at 6 months. Annuloplasty use varied among centers and failed to decrease the prevalence of moderate or severe LAVVR at the 6-month follow-up in this cohort. Older age at repair was an independent predictor of moderate or severe LAVVR. Further investigation is needed to determine if intervening in center practice variation or earlier referral of subtypes traditionally repaired at an older age can affect the relatively high prevalence of postoperative LAVVR.
This work was supported by National Heart, Lung and Blood Institute U01 grants HL068269, HL068270, HL068279, HL068281, HL068285, HL068292, HL068290, and HL068288. The contents of this publication are solely the responsibility of the authors and do not necessarily represent National Heart, Lung and Blood Institute official views.
Presented at the Forty-seventh Annual Meeting of The Society of Thoracic Surgeons, San Diego, CA, Jan 31–Feb 2, 2011.