SEARCH

SEARCH BY CITATION

Keywords:

  • atrioventricular canal;
  • atrioventricular septal defect;
  • cardiac defect;
  • fetal biometry;
  • fetal echocardiography;
  • prenatal diagnosis

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Objectives

Atrioventricular septal defects (AVSDs) are the most common cardiac abnormality in fetuses with numerical chromosomal aberrations, in particular trisomy 21. The majority of AVSDs are not detected by routine ultrasound examination in pregnancy. We report two simple cardiac measurements that may substantially improve antenatal detection of AVSDs.

Methods

Cross-sectional ultrasound images through the fetal thorax demonstrating the four-chamber plane of the heart were obtained in 123 normal fetuses between 10 and 38 weeks of gestation. Heart length was measured at the level of interventricular septum by placing the calipers on the epicardium at the apex of the heart and on the endocardium at the top of the atrium. Ventricular length was measured by shifting the atrial caliper to the crossing point of the ventricular septum and mitral valve. Atrial length was calculated as the difference between the heart length and ventricular length. Based on these measurements, the atrial-to-ventricular length (AVL) ratio was calculated. Data were compared to measurements from 29 consecutive fetuses with AVSD between 13 and 39 weeks of gestation.

Results

In normal fetuses, the AVL ratio did not change with gestation and the mean AVL ratio was 0.47 (95% prediction interval 0.35 to 0.63). In the AVSD group, the mean AVL ratio was 0.77 (range, 0.59–0.99). If a cut-off value for the AVL ratio of 0.6 was chosen, the detection rate of AVSD was 86.2% at a 5.7% false-positive rate. For a 100% detection rate, the false-positive rate was 7.3%.

Conclusions

The AVL ratio can accurately discriminate between hearts with AVSDs and normal cardiac anatomy. Incorporation of the AVL ratio measurement into routine antenatal ultrasonography may substantially improve the ability to diagnose AVSDs antenatally. Copyright © 2004 ISUOG. Published by John Wiley & Sons, Ltd.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Atrioventricular septal defects (AVSDs) are a common anomaly, occurring in 7.5% of liveborns with congenital heart defects1. Furthermore, they are the most common cardiac malformation associated with aneuploidy, in particular trisomy 212. In our experience, AVSD is usually only diagnosed if either extracardiac anomalies and soft markers of trisomy 21 are present or if amniocentesis has revealed an abnormal karyotype. The detection rate for isolated AVSD is likely to be even lower.

In the fetus, AVSD may potentially be detected by visualizing the four-chamber view3, 4. Compared to hypoplastic left heart or univentricular heart defects, major distortions are not necessarily apparent in AVSD5. In a pediatric cardiology study in the UK, only 38% of 579 children with AVSD had the anomaly diagnosed during fetal life, whereas detection rates for hypoplastic left heart and univentricular heart defects were much higher at 66% and 60%, respectively6. Difficulties in analyzing the four-chamber view correctly may be the cause of these differences5.

The aim of our study was to investigate whether measurements of fetal heart length, ventricular length, atrial length or corresponding ratios in the four-chamber view will discriminate between hearts with AVSD and normal cardiac anatomy.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Patients

The control group comprised 123 prospectively examined normal singleton pregnancies with confirmed gestational age between 10 and 38 weeks. The patients were referred either for routine antenatal ultrasonography and follow-up or for targeted cardiac examination on account of a positive family history of cardiac abnormalities. Exclusion criteria were evidence of major anomalies, in particular cardiac abnormalities and arrhythmia, fetal growth restriction and chromosomal anomalies. In all infants, normal anatomy and development were confirmed by postnatal clinical examination. All patients gave verbal consent to participate in the study. The procedure was part of routine antenatal ultrasonography and did not increase the usual examination time. Clinical management of pregnancies was not influenced by the results of measurements. Ultrasound examinations were performed by three experienced examiners (A.M., R.C. and K.S.H.).

The study group comprised 29 consecutive pregnancies with fetal AVSD between 13 and 39 weeks of gestation. Prenatal diagnosis was confirmed either by postnatal echocardiography or autopsy. A total of 17/29 measurements were prospectively performed, whereas the remaining 12 cases were analyzed retrospectively in digitally stored images or images exported from videotape to a digital workstation.

Echocardiographic measurements

The study was performed using high-resolution, two-dimensional echocardiography equipment (Voluson 730 Expert, GE-Kretztechnik, Zipf, Austria; Logic G9, GE, Solingen, Germany and HDI-5000, ATL Philips Medical Systems, Solingen, Germany) with zoom and cine-loop technique facilities. Convex transducers with a broadband range between 2 and 8 MHz were used. Assessment of the fetal heart was carried out using narrow image sector and high frame rate presets. The apical four-chamber planes were visualized with perpendicular insonation of the atrioventricular (AV) valves and the image was magnified using the zoom function until the fetal thorax filled the whole screen. For measurement purposes the image of the four-chamber view was stored at end diastole with closed AV valves, just prior to the onset of the systole. This was achieved using the cine-loop function and scrolling through the heart cycle frame by frame. All stored images were analyzed offline by a single operator (A.M.). An additional nine images (6.9%) with lateral orientation of the apex cordis were excluded because of an insufficient angle of insonation between the ultrasound beam and the AV valves. All measured distances refer to a line along the ventricular and atrial septum (Figure 1a). In order to obtain reproducible measurements in the four-chamber view, calipers were set at structures that provide maximum contrast. Therefore, heart length was measured by placing apical calipers at the outer border of the epicardium and basal calipers at the inner border of the atrial endocardium. Ventricular length was measured by keeping the apical caliper unchanged and replacing the atrial caliper to the crux cordis at the level of mitral valve attachment. To measure ventricular length in AVSD, the second caliper was placed at the level of the closed common AV valve (Figure 1b). Atrial length was calculated as the difference between heart length and ventricular length. Consequently, an atrial-to-ventricular length (AVL) ratio was calculated as follows: AVL ratio = (heart length − ventricular length)/ventricular length.

thumbnail image

Figure 1. (a) Apical four-chamber view in a 24-week fetus demonstrating measurement of heart length (solid arrows) and ventricular length (open arrows) along the interventricular septum. (b) Corresponding image of the four-chamber view in a 26-week fetus with atrioventricular septal defect.

Download figure to PowerPoint

Statistical analysis

Statistical analysis was performed using SPSS for Windows, version 11.0 (SPSS Inc., Chicago, IL, USA). A P-value of 0.05 was considered significant. In cases of non-normal distribution, logarithmic transformation of data was performed. Repeatability analyses were performed for both heart and ventricular lengths according to the method described by Bland and Altman7. Intraobserver variation was tested in 10 fetuses (18–23 weeks of gestation). Coefficients of repeatability and 95% coverage intervals of differences (CID) for heart length and for ventricular length were 1.3 mm (95% CID, 0.920–1.023) and 1.7 mm (95% CID, 0.87–1.09), respectively. Interobserver variation was tested in 10 fetuses (20–36 weeks of gestation). Coefficients of repeatability for heart length and ventricular length were 3.2 mm (95% CID, 0.925–1.152) and 3.5 mm (95% CID, 0.892–1.230), respectively.

Backward stepwise regression analysis was performed in order to check for dependence from gestational age and to construct reference ranges in the controls8. After normalization for gestational age using Z-scores, controls and cases were compared by t-test. Subgroup analyses with respect to fetal orientation revealed no differences in heart length, ventricular length, atrial length values and in AVL ratio (Chi-square test). Receiver–operator analysis was performed to predict the best cut-off.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Heart length, ventricular length and atrial length increase with advancing gestation. Results of regression analysis in the controls are shown in Table 1. In all three distances a quadratic term was found to be significant.

Table 1. Regression equations for measured distances in the four-chamber view (y) with gestational age (x), residual SD and R2 in the control group (n = 123)
Parametery =SDR2n
Heart length−18.79 + 2.34 x − 0.017 x22.410.96123
Ventricular length−12.53 + 1.58 x − 0.011 x21.860.94123
Atrial length−6.24 + 0.77 x − 0.006 x21.240.90123

When normalized for gestational age, comparison between the study group and controls showed no differences for heart length; however, in fetuses with AVSD, ventricular lengths were shorter (P < 0.001) and atrial lengths were longer (P < 0.001) than those in controls (Figure 2).

thumbnail image

Figure 2. Box plots representing heart length (equation image), ventricular length (equation image) and atrial length (equation image) in the study group (atrioventricular septal defect, AVSD) and controls after normalization for gestational age using Z-scores. Z-scores represent the deviation of the value from the predicted mean value for gestational age, expressed in multiples of standard deviation.

Download figure to PowerPoint

The AVL ratio was found to be independent of gestational age. The mean AVL ratio in controls was 0.47 (95%CI, 0.46–0.48) and the 95% prediction interval (in which 95% of the expected measurements will lie) ranged from 0.35 to 0.63.

In the study group with AVSD, measured values of the AVL ratio ranged from 0.59 to 0.99. In 24/29 fetuses the AVL ratio was above the 95% prediction interval. None of the cases of AVSD in our series showed an AVL ratio below the normal mean (Figure 3). Selecting a cut-off for normal AVL ratio at 0.6 would detect 86.2% of AVSDs at a 5.7% false-positive rate, and a cut-off of 0.58 would include all AVSDs and 7.3% of the controls.

thumbnail image

Figure 3. Values of the atrial-to-ventricular length (AVL) ratio in the study group (▪) and controls (○). The solid line represents the mean AVL ratio in controls and the dashed lines mark the 95% reference range.

Download figure to PowerPoint

In the study group there were 13 fetuses with trisomy 21, two with trisomy 18 and one case with Robertsonian translocation. Twelve fetuses had a normal karyotype and in one fetus the karyotype remained unknown. In the cases with AVSD, the AVL ratio was higher in euploid fetuses than in aneuploid fetuses (P = 0.008). The mean value of the AVL ratio was 0.73 (95% CI, 0.67–0.78) if trisomy 21 was present and 0.83 (95% CI, 0.75–0.91) in euploid fetuses, respectively.

The AVSD was isolated in 12/29 cases in the study group. In the remaining 17 cases there were five with coarctation, three with tetralogy of Fallot, three with right atrial isomerism, three with additional ASD or VSD and one each with hypoplastic aortic arch and pulmonary hypoplasia. Partial AVSD was found in one fetus with an AVL ratio of 0.71. A total of 9/13 fetuses with trisomy 21 had isolated AVSDs, whereas only 3/12 euploid fetuses showed isolated defects. Partial AVSD was found in one fetus with an AVL ratio of 0.71.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

The AVL ratio determined in the four-chamber view discriminates with great accuracy between normal cardiac anatomy and the presence of an AVSD. If 0.6 is used as cut-off value, 86.2% of AVSDs are detected at the expense of 5.7% false-positives. A 100% detection rate is achieved at a rate of only 7.3% false-positives. These results suggest that integration of the AVL ratio into routine ultrasound screening may increase the detection rate of AVSD.

To our knowledge, there is no prenatal ultrasound study that quantifies the dimensions of atrial and ventricular septum lengths in relation to complete heart length. In AVSD, precise data describing the position of the common atrioventricular valve in the four-chamber view are lacking. Our data confirm that the ventricular length is smaller when there is an AVSD. This is supported by previous studies in children and adolescents with AVSD, which found that the defect resulted from a greater deficiency of ventricular than atrial septal tissue9, 10. In these studies it was shown that in all cases of complete and partial AVSD, septal deficiency extends the level of the common AV junction in the ventricular direction. In addition to a smaller ventricular length, we found a greater atrial length in our study group, which is in contrast to previous postnatal studies9, 10. This discrepancy can be explained by the fact that atrial length determined by our method was the distance from the atrial border to the common AV valve. It thus comprises both atrial septal length and part of the AVSD.

Fetal cardiac measurement studies have been reported since the 1980s using M-mode and two-dimensional ultrasound11–13. In our study, heart length measurement is comparable to that recorded in previous studies13, whereas owing to different measurement techniques, ventricular length and atrial length are not. In our study, ventricular length corresponded to the distance between the crux cordis and apex cordis in the controls and between the closed AV valve and apex cordis in hearts with AVSD. Therefore, studies that determined internal ventricular length in the left and right ventricles separately provided shorter measurements14–16.

In the last 25 years, biometry of fetal heart dimensions has not become part of routine ultrasound screening because it is rather time consuming and, with few exceptions17, ineffective in detecting heart anomalies. Measurements of internal chamber dimensions are suitable to compare corresponding chambers in cases of suspected hypoplasia or dilatation. In the present study we have introduced the AVL ratio as a new parameter to be measured during routine screening ultrasonography. Measurement of the AVL ratio will enable the detection of at least one major cardiac anomaly with good reliability. Measurement of the AVL ratio is both simple and effective when standardized as proposed in our protocol.

As we have learned from nuchal translucency (NT) studies, standardized protocols that obtain, document and quantify a certain cross-sectional plane have numerous positive side effects: they urge the examiner's attention to crucial regions. Abnormal measurements may identify a group at high risk for certain anomalies and which require further examination. The application of ultrasound equipment functions such as zoom and cine-loop may improve the quality of examination in general, and in particular in fetal echocardiography5. With adequate training and audit, even distances of a few millimeters may be obtained in the fetus with great accuracy as shown in NT and nasal bone studies18, 19.

AVSDs are frequently associated with chromosomal abnormalities and heterotaxy syndromes20. In subgroup analysis we found a high rate of aneuploidy when the AVL ratio was in the upper normal range (0.59–0.63). Very high values of the AVL ratio were more likely to be associated with complex heart defects. These results suggest that in cases with isolated AVSD, which are commonly found in trisomy 21, distortion of the four-chamber view is minimal. In contrast to hypoplastic left or right heart defects, the four-chamber view in hearts with an AVSD may on first sight appear normal5. This may explain the low antenatal detection rates in the case of this particular heart defect. We assume that a more careful cardiac examination in cases with an AVL ratio of 0.6 and above will permit detection of additional cases with AVSD.

In the study group with AVSD, the lowest value for the AVL ratio was found to be 0.59. These data suggest that AVSDs are unlikely to be present if the AVL ratio is below the mean value of 0.47. Since this preliminary study included a small number of pregnancies, conclusions about the efficacy of incorporating this echocardiographic measurement into screening should be drawn with caution until data from larger prospective studies are available. Our study was performed by examiners experienced in fetal echocardiography who may have been biased by the fact that they had identified this condition.

Our findings imply limitations regarding applicability in the first trimester of pregnancy. We report on normal values from 10 gestational weeks onwards. However, in our study only one AVSD was found before 16 weeks. In this 13 + 5-week fetus, the AVL ratio was 0.61, which is in the upper normal range. Due to the limited number of cases, we cannot exclude a possible need for a lower cut-off in early pregnancy.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References