Background
Atrioventricular septal defects (AVSDs) account for 7% of all congenital cardiovascular malformations. The atrioventricular septum (AVS) is the portion of the septal tissue that separates the right atrium from the left ventricle; deficiency of the AVS contributes to the AVSD phenotype. A study of case and control families was performed to identify whether an intermediate phenotype consisting of a shortened AVS existed in relatives of children with AVSDs.
Methods
AVS length (AVSL) was measured on the echocardiograms of clinically unaffected parents and siblings from families that were identified through children with nonsyndromic AVSDs and in families with no histories of congenital heart disease.
Results
No significant differences were seen between case and control family members in terms of gender, age, weight, and height. AVSLs were significantly shorter in case parents compared with control parents. Similar findings were noted within the sibling groups. There was significant evidence for two-component distributions in the case parent, case sibling, and control sibling groups after standardizing AVSL for age and body surface area. Heritability of AVSL standardized for age and body surface area was 0.82 and 0.71 in nonsyndromic case and control families, respectively.
Conclusions
Evidence for two-component distributions from the analysis of AVSL standardized for age and body surface area for case parents and case siblings suggests the presence of an intermediate phenotype for nonsyndromic AVSD. The high heritability in the control families suggests that there may be polygenic involvement in the determination of AVSL. Broadening the definition of AVSD to include those with shortened AVSL may increase the power of genetic association and mapping studies to identify susceptibility genes for AVSD.
Congenital heart defects constitute a major proportion of clinically significant birth defects and are an important component of pediatric cardiovascular disease, with an estimated prevalence of six to nine per 1,000 live births. Atrioventricular septal defects (AVSDs), also known as atrioventricular canal defects or endocardial cushion defects, include a spectrum of anomalies characterized by involvement of the atrial and/or ventricular septa and one or both of the atrioventricular valves; they account for approximately 7% of all congenital heart defects.
With normal cardiac development, the septal leaflet of the tricuspid valve inserts into the septum slightly more inferior than the septal leaflet of the mitral valve. A small portion of septal tissue superior to the tricuspid septal leaflet insertion separates the right atrium from the left ventricle; this is the atrioventricular septum (AVS) ( Figure 1 ).
There is a paucity of information regarding the AVS, and details regarding the normal development of the AVS are relatively unknown. Failure of fusion of the atrioventricular endocardial cushions has long been suggested as the mechanism for AVSD formation. However, more recent studies have demonstrated that the development of the atrioventricular septal area is highly complex, involving multiple primordial structures, including the endocardial cushions. Results from these investigations suggest a possible role of the endocardial cushions, mesenchyme from the primary atrial septum, and spina vestibuli in the development of the AVS. Atrioventricular septal length (AVSL) is not routinely measured on echocardiography, unless there is concern for Ebstein’s malformation. In prior investigations, however, it was noted that all of the patients with normally structured tricuspid valves had body surface area (BSA)–standardized AVSL (sAVSL) < 8 mm/m 2 , whereas those with Ebstein’s malformation had sAVSL > 8 mm/m 2 , providing the basis for the displacement index, which can assist in the diagnosis of Ebstein’s malformation. It has also been demonstrated that the absence of the AVS results in the AVSD phenotype, implying that those patients with complete AVSDs have AVS lengths that measure 0 mm.
Although AVSDs commonly occur in the setting of Down syndrome, they also occur in infants without diagnosed syndromes. Numerous investigations have been performed in the attempt to identify causative genes in syndromic and nonsyndromic AVSDs. Although such causative genes have not yet been identified, it has been demonstrated that the developmental origins of syndromic and nonsyndromic AVSDs differs. Nonsyndromic AVSDs are estimated to occur in approximately one per 10,000 live births. Most nonsyndromic AVSDs are considered to be sporadic and/or the result of multifactorial inheritance. However, there are numerous reports of nonsyndromic AVSDs transmitted within families, suggesting that the defect segregates with a Mendelian pattern. The pattern of recurrence has most often suggested an autosomal dominant model with monogenic or oligogenic inheritance. Although AVSDs appear to be transmitted in an autosomal dominant fashion, there are parents in these pedigrees who do not demonstrate the phenotype of an AVSD or a defect along its spectrum, and yet they may have multiple affected offspring. These parents may have an intermediate phenotype (e.g., a shortened AVSL), but no intermediate phenotype has yet been sought.
The major goal of this investigation was to measure AVSLs in a study of case and control families to determine whether the parents and siblings of children with nonsyndromic AVSDs demonstrate shorter AVSLs, possibly indicating an intermediate phenotype. We hypothesized that a subset of the “unaffected” parents and siblings of case children would have shorter AVSLs than the remainder of the case parents and siblings, whose AVSLs, in turn, would not be different from the AVSLs of parents and siblings in control families.
Methods
Subject Population
Case families were those who participated in the Family Study of Endocardial Cushion Defects, which was conducted between 1994 and 2004 at the University of Iowa. The nonsyndromic AVSD cases were identified through cardiac catheterization, echocardiographic, and surgical records at the University of Iowa Hospitals and Clinics and recruited for the study. If the family agreed to participate, a three-generation pedigree was constructed, and a health history questionnaire was administered over the phone. The families (parents and siblings) were then scheduled for echocardiographic examinations. Seventy-two families of children with nonsyndromic AVSDs were recruited and examined.
Children free of congenital heart defects and their parents and siblings from Muscatine, Iowa, were also recruited to serve as control families. Echocardiography was performed in a similar fashion for the families who agreed to participate. Seventy-four control families were recruited and examined. Case and control family members underwent echocardiographic examination by separate sonographers. The study was approved by the University of Iowa Institutional Review Board.
Echocardiographic Analysis
The 427 available echocardiograms from the case and control family members were reviewed to measure AVSLs in an attempt to define and describe an intermediate phenotype of AVSDs. Because of the time frame of the conduct of the original study, echocardiograms were stored on VHS tapes and were not digitized. A Philips Sonos 5500 echocardiographic machine (Philips Medical Systems, Andover, MA) was used for detailed measurements.
AVSL was measured using the caliper tool as part of the installed software package on the machine. AVSL was defined as the length from the hinge point of the mitral valve to the hinge point of the tricuspid valve along the septum in an apical four-chamber view. The three sharpest apical four-chamber views during systole were chosen for measurements. Three repeat measurements were in each view made by the primary investigator (S.S.P.), for a total of nine measurements of AVSL.
For the purposes of assessing interrater reliability, repeat measurements (10%) were made by a blinded independent investigator (L.T.M.) with substantial echocardiographic experience. Two repeat measurements of AVSL were made in each of two views, chosen independently by the independent investigator, for a total of four measurements.
Statistical Analysis
Descriptive Analysis
Using the set of measurements obtained by the primary investigator, the mean of the three AVSL measurements from each of the three separate views (i.e., the mean of nine measurements) was determined. Descriptive statistics for gender, age, weight, height, BSA, and AVSL measurements were estimated for case and control parents and siblings. Means and standard deviations were estimated for continuous variables, and frequencies were determined for categorical variables. Case and control subgroups were compared for differences using Student’s t and χ 2 tests.
Reliability
An intraclass correlation coefficient was estimated using the mean of the three AVSL measurements from each of the three separate views to determine intrarater reliability. Similarly, the intrarater reliability among the independent investigator’s measurements was assessed using the mean of the two AVSL measurements from each of the two separate views. Because two different sonographers obtained the echocardiograms for the case and control families, two separate intraclass correlation coefficients were calculated to account for sonographer differences. Interrater reliability was assessed by performing paired t tests using the overall mean measurement (of nine and four measurements, respectively) from each investigator and calculating percentage error and a coefficient of variation to estimate the degree of variability between the two investigators.
AVSL Standardization
Pediatric echocardiographic measurements of cardiovascular structures are routinely adjusted to account for the effects of body size. BSA appears to be a better parameter of growth than height or weight alone.
The AVSL measurements of the case parents were examined using univariate and multivariate linear regression analysis models to determine if there was an association with BSA, age, or gender. Similar analyses were performed for the case sibling, control parent, and control sibling groups. An evaluation of regression diagnostics was performed; influential points were identified by examination of scatterplots and were reevaluated for measurement and/or data entry error. Measurements were standardized for age and BSA on the basis of these results (asAVSL).
Admixture Analysis
Admixture analysis was used to test the hypothesis that the overall observed distribution of asAVSL measurements actually reflected the sum of two or more separate component distributions, each of which might represent a different etiology (e.g., a gender effect or a major genetic effect or a smaller vs a larger number of AVSL-shortening alleles).
The distribution of the asAVSL measurements was formally tested for evidence of admixture using the program NOCOM. Case parents, case siblings, control parents, and control siblings were separately analyzed.
Heritability Analysis
For a quantitative trait, familial aggregation (the tendency for a trait to cluster in families) can be examined by estimating correlation coefficients between pairs of relatives (e.g., between parents and children). Significant evidence for familial aggregation suggests that shared genetic and/or shared environmental factors are likely to be involved in the etiology of the trait.
The heritability of asAVSL was estimated on the basis of measurements from unaffected siblings and parents of cases using the Polygenic and QTL Mapping option in MENDEL. The heritability of asAVSL was also estimated for the control families.
Results
Participants
Four hundred twenty-seven echocardiograms were reviewed using the measurement protocol. Eighteen family members were excluded because of insufficient information: either the echocardiographic windows were inadequate for reliable measurements of AVS or height and/or weight data were not available for the calculation of BSA. Of the remaining participants, 212 were members of nonsyndromic case families and 199 were members of control families.
Descriptive Analysis
Table 1 describes group characteristics (percentage male and mean age, height, weight, and AVSL) for each study group. There were no significant differences between the case and control parent groups in terms of gender distribution, age, weight, height, and BSA. However, mean AVSL was significantly different between the two groups of parents. More specifically, the mean AVSL measurement in the case parent group was significantly shorter than in the control parent group.
| Variable | Nonsyndromic AVSD parents ( n = 118) | Control parents ( n = 109) | P | Nonsyndromic AVSD siblings ( n = 92) | Control siblings ( n = 90) | P |
|---|---|---|---|---|---|---|
| Male gender ∗ | 45.76% | 46.79% | .88 | 52.17% | 52.22% | .99 |
| Age (y) † | 38.36 ± 9.32 | 37.83 ± 6.24 | .61 | 15.28 ± 10.26 | 10.81 ± 4.69 | .0002 ‡ |
| Weight (kg) † | 80.20 ± 18.54 | 77.55 ± 16.92 | .26 | 56.08 ± 32.63 | 43.00 ± 22.14 | .0018 ‡ |
| Height (cm) † | 171.05 ± 10.17 | 170.82 ± 9.13 | .86 | 149.60 ± 33.66 | 144.55 ± 25.18 | .25 |
| BSA (m 2 ) † | 1.96 ± 0.27 | 1.92 ± 0.24 | .30 | 1.50 ± 0.61 | 1.29 ± 0.44 | .0089 ‡ |
| AVSL (mm) † | 6.42 ± 1.98 (2.8–13.0) | 8.90 ± 2.75 (3.1–17.4) | <.0001 ‡ | 5.88 ± 2.12 (1.8–10.5) | 7.82 ± 2.34 (2.9–15.2) | <.0001 ‡ |
| sAVSL (mm/m 2 ) † | 3.36 ± 1.25 | 4.67 ± 1.44 | <.0001 ‡ | 4.41 ± 1.84 | 6.43 ± 2.08 | <.0001 ‡ |
∗ Chi-square tests were performed to test for differences among groups.
† Student’s t test was performed to test for differences among groups.
No significant differences in gender distribution were noted among the sibling groups. Significant differences were noted, however, in terms of age, weight, and BSA. Because case families were ascertained using historical records, many siblings were adults when they underwent echocardiography. Because of their significantly older age, the case siblings also had a higher mean weight and BSA. However, the case sibling mean AVSL was significantly shorter than the control sibling group.
Reliability
Interrater reliability of these AVSL measurements was first assessed using a paired t test to test the hypothesis of no difference. The t statistic was not significant ( P = .57), suggesting that the measurements made by the two investigators were similar. The coefficient of variation between the two investigators was 13.7%. In addition, the percentage error was 10.9%. These two measures suggest that the variation of the measurements was small. Intrarater reliability for each investigator was also evaluated by the estimation of an intraclass correlation coefficient. The overall intraclass correlation coefficient of the three means for the primary investigator (S.S.P.) was 0.979, and the intraclass correlation coefficient of the two means for the independent investigator (L.T.M.) was 0.983. When the sonographer who obtained the echocardiogram was taken into account, the intraclass correlation coefficients for the case group for the primary and independent investigators were 0.968 and 0.992, respectively. Similarly, the respective intraclass correlation coefficients were 0.955 and 0.993 for the control group.
Standardization
Figure 2 is a scatterplot of unadjusted AVSL measurements versus BSA by relationship. The regression line is also plotted for each relationship group. Univariate analysis was performed to evaluate the association between AVSL and BSA ( P < .0001). On the basis of the two regression lines, it appears that AVSL increases as BSA increases during childhood, and once adult size is reached, increase in AVSL plateaus. Given the significant association, AVSL was standardized using the BSA estimate to account for body growth differences.
A scatterplot was constructed depicting the relationship between unadjusted AVSL measurements and age by gender (data not shown). Male and female subjects appeared to be distributed equally within the plot. Student’s t test was conducted to evaluate for differences in unadjusted AVSL by gender, which was not significant, suggesting the AVSL measurements were similar with respect to gender ( P = .26). Univariate analysis did not identify a significant relationship between sAVSL measurements and gender ( P = .14).
The unadjusted AVSL measurements were examined for association with age for all siblings and parents ( Figure 3 ). Similar to the relationship with BSA, Figure 3 suggests that there is rapid growth of AVS throughout childhood and adolescence, which plateaus in adulthood. Univariate analysis revealed a significant association between sAVSL measurements and age ( P < .0001). For the remainder of the analyses, measurements were adjusted to account for both age and BSA differences (asAVSL).
Admixture Analysis
Histograms of the asAVSL measurements were plotted and examined for evidence of admixture (i.e., multiple components) in each group: nonsyndromic AVSD case parents and siblings and control parents and siblings ( Figure 4 ). By examination, it appears that there are two underlying distributions for asAVSL in the case parents and siblings.
To formally test the hypothesis that there is more than one distribution within the subgroup distributions, likelihood ratio tests were conducted in the each case and control subgroup ( Table 2 ). Significant evidence was found for two-component distributions in the case parent and case sibling groups. The control parent group did not yield evidence for admixture. The control sibling group demonstrated evidence for two-component distributions; however, the second component contained only 2% of the measurements in the extreme upper tail.
