Identification of fetal cardiac anatomy and hemodynamics: a novel enhanced screening protocol

Congenital heart disease (CHD), accounting for about 0.4–1.3 % of all live births [1‐4], is the most common congenital malformation leading to perinatal morbidity and mortality, and is considered the leading cause of death in newborns with congenital anomalies [5, 6]. Fetal echocardiography is undoubtedly the best method currently available to diagnose congenital cardiac anomalies prenatally [7]. However, it is not practical for fetal echocardiographers to make a diagnosis for every fetus during routine obstetric scans. Thus a detailed cardiac screening performed by obstetric sonographers may be an alternative to solving the problem [8‐10]. When CHD is suspected, patients should be referred for comprehensive cardiac examination by a fetal echocardiographer. To this end, highly efficient screening might then lead to a high detection rate for fetal CHD.

A symmetric four-chamber view (4CV) has been used to confirm a normal fetal heart since the early 1990s [11, 12]. However, it is well-documented that 4CV alone is inadequate in ruling out conotruncal anomalies [13‐15]. The ISUOG guideline [16] has been revised and updated to include more cardiac views that improve the efficiency of prenatal screening for CHD. In our center, 4CV, combined with double outflow tract views and three-vessel and trachea view (3VT), has been used in routine obstetric screening since 2005. It is confirmed that this produces more effective referrals than ever before. However, we found that there were still many serious CHDs that were missed during screening. We therefore propose an improved screening protocol to enhance current cardiac screening methods.

A total of 300 fetuses were included in the current study. All cases were confirmed by autopsy findings, operative findings, fetal/postnatal echocardiography, or by phone calls in order to acquire the information from routine infant physical examination reports. In our study, patent foramen ovale and patent ductus arteriosus were both considered normal cardiac structures during the follow-up.

In total, 52 cases of fetal CHD were included in the current study, of which five cases were from the routine screening pregnancies. Prenatal cardiac screenings for CHDs by the two protocols and fetal outcomes are summarized in Table 3. Sixteen families made a choice to terminate the pregnancy and four fetuses died within the uterus. For the living neonates, seven died and the other 25 survived the neonatal period. The detection rate for CHD under Protocol B was significantly higher than under Protocol A (96.2 % vs. 61.5 %, P < 0.01).

For fetuses with cardiac malpositions, only 50 % cases were detected under Protocol A, while all cases were visualized under Protocol B. We also found that sonographers easily detected malformations associated with systemic and pulmonary venous connections under Protocol B (100 %), while these anomalies were almost not noted under Protocol A. A movie file shows confirmation of situs in a case of heterotaxy in detail [See Additional file 1: Video]. Figure 1 and three additional movie files (See Additional files 2, 3 and 4: Video) show a case of suspected interrupted IVC and azygos continuation. Figure 2 and two additional movie files (See Additional files 5 and 6: Video) show a case of total anomalous pulmonary venous connection (TAPVC), in which the pulmonary veins were suspected to drain into the coronary sinus. Moreover, Protocol B was superior in detecting some lesions at the arterial level, such as right aortic arch (Fig. 3), transposition of the great arteries (TGA) (Fig. 4), congenitally corrected transposition of the great arteries (ccTGA) (Fig. 5), etc. Several additional movie files showed the detection of these cardiac anomalies in detail (See Additional file 7 for right aortic arch; Additional files 8, 9 and 10 for TGA; and Additional files 11, 12 and 13 for ccTGA: Video).

Protocol B was efficient at detecting VSD, valvular stenosis/atresia/dysplasia, in which color Doppler might play an important role. Figure 6 and two additional movie files (See Additional files 14 and 15: Video) show a case of pulmonary valve atresia in which the lesion was barely visualized by 2D ultrasound as the PA was normal in diameter; but was easily detected by color Doppler due to the retrograde flow in DA and main PA. Figure 7 and two additional movie files (See Additional files 16 and 17: Video) show a case of tricuspid dysplasia with regurgitation, which was unobserved during routine screenings by 2D sonography. In fact, the thickened valves could only be visualized in certain views, while the regurgitation was very easily seen. For the CHD fetuses, the mean scanning and imaging interpretation times for cardiac screening were 8.7 ± 2.3 min and 17.2 ± 5.8 min for sonographers under Protocol A and B, respectively. Protocol B also required much more time than did Protocol A. For Protocol A, the interobserver and intraobserver variability was 6.5 and 9.8 % for detecting CHDs, respectively. The interobserver and intraobserver variability was 4.1 and 6.2 % for Protocol B, respectively.

Two hundred forty-eight fetuses with normal cardiac structures were also included in the current study. For these normal fetuses, the mean scanning and imaging interpretation times for cardiac screening were 5.6 ± 2.9 min and 8.3 ± 3.8 min for sonographers under Protocol A and B, respectively. Protocol B was again more time-consuming than Protocol A. Most of the normal fetuses were considered normal cardiac structures under both protocols. Table 4 summarizes the cardiac findings for those were suspected abnormal cardiac structures during screening but were confirmed to be normal by fetal/neonatal echocardiography.

Some fetuses were referred for further echocardiography as pericardial effusion was suspected by sonographers under both Protocols A and B. In fact, this phenomenon has been mentioned in the ISUOG guidelines [16] as the small hypoechogenic rim around the fetal heart usually represents a normal variation. In one case, coarctation of the aorta (COA) was suspected mainly because of the angle of the ultrasound beam. In another case, a space-occupying lesion was suspected while in actuality it was the hypertrophic papillary muscle of the right ventricle. VSDs were also suspected falsely because of acoustic drop-out artifacts (Fig. 8). In another case, PAVSD was falsely suspected because the coronary sinus was regarded as a deficiency of the atrial septem primum (Fig. 9). It was interesting to note that Protocol B was better than Protocol A in reducing the false positives in detecting septal defects. Color Doppler also played a significant role in this phenomenon.

Of all the congenital anomalies, CHD is the most common and has the worst prognosis compared with other disorders due to its high morbidity and mortality [5, 6]. From the early 1990s, fetal echocardiography made it possible to detect CHDs prenatally [11, 12]. Initially, indications for fetal cardiac scans were only based upon some parental risk factors such as maternal diabetes or a positive family history of cardiac anomalies [22]. In the early 2000s, people realized that cardiac examinations should be extended to all fetuses undergoing routine obstetric scans [8, 23‐25], and this challenged most obstetric sonographers as fetal cardiac scans and interpretations required a unique set of advanced expertise and knowledge.

Initially, obstetric sonographers recognized that 4CV was easy to visualize during routine scans, and that it was relatively easy to obtain the standard 4CV if the spinal acoustic shadow was avoided. The optimal view of the 4CV is usually obtained when the cardiac apex is directed toward the maternal wall. The introduction of 4CV in routine screenings has proven its value as it could detect some cardiac anomalies at the atrial and ventricular levels, such as large VSD, atrioventricular septal defect (AVSD), Ebstein’s anomaly, hypoplastic left heart syndrome (HLHS), hypoplastic right heart syndrome (HRHS) and space-occupying lesions, etc [9, 26]. In fact, it was relatively easy for us to visualize four asymmetric chambers during prenatal screenings, which provided evidence for further comprehensive examinations. Unfortunately, 4CV alone could not provide the anatomic information at the arterial level, which leads to the low detection rate for conotruncal anomalies [9, 27].

As more concerns have been raised about the low detection rate of 4CV screening, ISUGG revised their previously published guidelines for cardiac screening in mid-gestation. In the new version of the guidelines, transverse scans were recommended to involve both the 4CV and outflow tract views. Also, AIUM has proposed similar guidelines [28, 29]. It is therefore a very important step in improving detection of CHD by complementing the 4CV with the outflow tract views under these guidelines, as documented by several teams and regional studies [13, 15]. Theoretically, most cardiac anomalies can be detected if all the planes could be acquired and the image data interpretations are made correctly. However, our results were not as satisfactory as we had hoped. According to our study, the detection of CHD was only 61.5 % under the protocol using transverse scans. Though this detection rate might be underestimated, as we included more CHD fetuses in our study than in routine screenings, the results implicated the possibility during routine screenings of missed diagnoses for special CHDs. In our experience, we found many CHD neonates manifesting TAPVC, TGA, pulmonary atresia, etc., whose mothers had undergone prenatal screenings at our center but had received no positive fetal diagnosis. Lacking accurate prenatal diagnosis made it impossible to provide appropriate operative schemes and multidisciplinary care. Therefore, many of these neonates died and their families suffered great pain. In fact, as the largest prenatal screening and diagnosis center in northeast China, we have both the best ultrasound equipment and most experienced sonographers. We expect that there are possibly more CHDs misdiagnosed in rural areas, and this is why we decided to propose a new protocol to strengthen the overall efficiency of fetal cardiac screenings.

When the study was completed, we summarized the reasons for the low detection rate under the protocol using transverse scans (Protocol A). An echocardiographer could recognize and diagnose almost all of the CHDs in our study when carefully reviewing the recorded image data acquired from transverse scans. For example, in the case of TAPVC, the tubular-like echo between the descending aorta (DAO) and LA was an important clue to drawing attention to potential cardiac anomalies. However, sonographer under Protocol A did not visualize this abnormal structure because they were not aware that there were no special anatomical structures in the areas between LA and DAO. Also, in the case of left superior vena cava, no vessel-like structures should be detected at the left side of main pulmonary artery (MPA) for normal fetuses. Had the screening sonographers been made aware of this, they would not have made any mistakes. In our proposed new protocol (Protocol B), we solved this problem by introducing into our training program many ultrasonic characters that reflect special cardiac anomalies according to the pathologic anatomy. In fact, Jeanty et al. published a series of reviews [18‐21] in which they summarized many of these useful ultrasonic characters for fetal CHDs, and some of these were referenced in our training program in Protocol B (mentioned in the Methods Section). The method proved its value in that the screening sonographers underwent relevant training and could then detect more CHDs than with other techniques.

Our newly proposed protocol emphasized confirming whether the connections of the two adjacent cardiac segments were correct or not. Sequential segmental analysis was recommended to evaluate the fetal cardiac structure and hemodynamcis. Although such analysis is a little difficult for screening sonographers, it justified its value by identifying all the TGA and ccTGA anomalies in the current study. As the 4CV appeared symmetric, these cardiac malformations were easily missed without a careful scan, even if the outflow tract views were scanned during routine screenings. On the contrary, identifying each anatomic structure at each segment, together with confirming the connections between each segment, ensured the successful detection of these complex anomalies.

In addition, the newly proposed protocol emphasized the application of color Doppler during cardiac scans and it justified its value by detecting more cardiac anomalies. For example, in the case of tricuspid dysplasia and tricuspid regurgitation, the thickened valves went undetected in the hands of a screening sonographer using only 2D, although they could be identified by an echocardiography specialist. In fact, color Doppler aids in the detection of flow disturbance, including stenosis and regurgitation, which may not be obvious from interrogation via 2D imaging alone. Furthermore, color Doppler discloses the flow direction in vessels, which is very useful in the detection of cardiac anomalies by assessing cardiac hemodynamics, and Jeanty et al [18‐21] reported their experiences in their publications. In our study, we proved this technique’s value in the case of pulmonary atresia in detecting retrograde flow in MPA and DA. As the valvular movement was difficult to observe via 2D, detection of this cardiac abnormality was missed by sonographer under Protocol A, in which color Doppler application was not mandatory.

We must state that there are some limitations to our proposed protocol. First, systematic training was needed to carry out this protocol, which was expensive and time-consuming. However, a 6-month training program has proven to be of great value in improving the efficiency of detecting CHDs during screening. Government investment and funding support may solve this problem. In addition, scans and imaging interpretations under the new protocol, based on sequential segmental analysis, require more time and expertise than under the previous screening protocol. However, we believe that the time commitment might be reduced gradually with the increase in proficiency. In our hands, the time spent scanning and interpreting images was about 15 min for a fetus with normal cardiac structures under the new protocol when the program began, which was then reduced to about 5 to 10 min by the time the study was terminated.

The current study was limited in that it was a relatively small study and did not represent a normal population, i.e., more CHD fetuses were intentionally entered into the study, not reflecting a non-selected population that underwent prenatal screenings. This overall design may bias the results so as to overestimate the difference between the two protocols. We emphasized to summarize the reasons for the cases being missed or detected by the two protocols, as the overall aim of the study was to improve cardiac screening efficiency by proposing a new protocol. As a mono-center based, short-term study, it might be impossible to observe various kinds of CHDs during routine screenings, and the spectrum of CHDs might vary greatly. The current design, then, was intended to include more CHD fetuses to evaluate the ability of the two protocols in screening CHDs, which might compromise the study. However, these limitations, had they been addressed in a multi-center study in a normal screening population, would likely have led to improvements in the completeness of the studies, strengthening our conclusion that the newly proposed protocol was more efficient in detecting CHDs.

The study was also limited in that all fetuses were scanned by only one sonographer under each protocol. The design limited the scanning time for fetal ultrasound examinations under the principle of exposure “as low as reasonably achievable.” Alternatively, we made inter- and intro- variability analysis by having another two sonographers with similar experience to re-analyze the imaging data and made interpretations, respectively. However, this may bias the results as one sonographer only read the videos recorded by the other. If the videos were of poor quality (i.e., a non-standard view, or critical information not presented well), the detection of CHD by the sonographer who did not perform the scan was inevitably low and underestimated.

The current study introduces an enhanced protocol for fetal cardiac screening, under which the obstetric screening sonographers identify the fetal cardiac anatomy and hemodynamics systematically. Some specific ultrasonic signs in certain views help to screen for specific cardiac anomalies. A short-term training program makes it possible for the screening sonographers to become familiar with the new protocol; this has confirmed its value in improving screening efficiency.

2D, two-dimensional; 3VT, three-vessel and trachea view; 4CV, four-chamber view; AVSD, atrioventricular septal defect; ccTGA, congenitally corrected transposition of the great arteries; CHD, congenital heart disease; COA, coarctation of the aorta; DA, ductus arteriosus; DAO, descending aorta; HLHS, hypoplastic left heart syndrome; HRHS, hypoplastic right heart syndrome; IVC, inferior vena cava; LA, left atrium; LV, left ventricle; MPA, main pulmonary artery; PA, pulmonary artery; PAVSD, partial atrioventricular septal defect; PV, pulmonary vein; RA, right atrium; RV, right ventricle; SVC, superior vena cava; TAPVC, total anomalous pulmonary vein connection; TGA, transposition of the great arteries; VSD, ventricular septal defect