Screening the fetal heart


  • L. Allan

    Corresponding author
    1. Harris Birthright Research Centre for Fetal Medicine, King's College Hospital, Denmark Hill, London SE5 9RS, UK
    • Harris Birthright Research Centre for Fetal Medicine, King's College Hospital, Denmark Hill, London SE5 9RS, UK
    Search for more papers by this author

Two papers in this month's Journal discuss the prenatal detection of congenital heart disease (CHD), albeit from very different approaches. Rasiah et al.1 reviewed the literature for publications about screening for CHD in the first trimester of pregnancy. The authors took a structured approach to the analysis of published data on the topic, in order to pool similar material. This type of analysis not only has the advantage of allowing evaluation of a larger patient group than is contained in any single study, but also demonstrates that the technique under scrutiny works in the hands of more than just a few experts in the field. The papers eventually selected from the literature search were published between 1993 and 2004, representing a study period of 1988 to 2003, which is of significance with respect to both technology and experience. All but two of the studies were targeted at high-risk fetuses. Inclusion criteria were strict and only patients scanned with the aim of cardiac evaluation in the fetus prior to 13 + 6 weeks of gestation were included. There were 1243 patients reported in the 10 studies finally selected, although four studies contributed almost a 1000 of the patients.

The authors defined adequate quality of follow-up as postmortem examination or postnatal echocardiography in all fetuses examined. I accept the criticism of mid-trimester fetal echocardiography as ‘inadequate’ follow-up in our study2, but over 20 years' experience and detailed follow-up data in more than 20 000 of our original series of fetuses have demonstrated that very little is missed by a targeted mid-trimester fetal echocardiogram in our hands. False negatives tend to be minor malformations of little or no consequence, or alternatively late-developing lesions, which, although important, are numerically rare.

Improved sensitivity rates were shown in later studies, which is not surprising, as ultrasound technology has continued to advance during the time period studied. In addition, techniques and operator expertise almost invariably improve with time and practice. Rather surprisingly, on the other hand, improved sensitivity rates were shown in the transabdominal rather than in the reported transvaginal studies. This may be because half of the transvaginal group were from publications of more than 10 years ago, and this result might be different were the studies more recent. Or this may be accounted for by the fact that, in general, the transabdominal studies were mainly performed by pediatric cardiologists, with perhaps ‘sharper eyes’ for cardiac defects, whereas the transvaginal studies were mainly conducted by obstetricians. There is no doubt in my mind that transvaginal ultrasound image quality can be much better when directly compared with transabdominal scans in the same patient but, in my opinion, even an experienced fetal cardiologist will never develop sufficient expertise at transvaginal scanning to become good at it. Even in a busy referral unit for nuchal translucency (NT), at the most he/she may do 2–6 high-risk 11–14-week scans a week. Compare this to the obstetrician or sonographer who is doing, perhaps, 10–20 nuchal scans per day transvaginally and therefore has the opportunity to learn and maintain the skills required for this approach. The ideal set-up would be for the pediatric cardiologist to use the skill of the transvaginal sonographer to demonstrate the heart in the 10% or so of patients where transabdominal views fail to provide adequate information for diagnosis. Because of the relatively small numbers, Rasiah et al. were unable to determine from the published material they analyzed whether the accuracy of cardiac diagnosis was greater if the fetus was closer to 14 weeks than 11. However, from personal experience, every day of fetal growth after 12 weeks makes an appreciable difference to the quality and clarity of the ultrasound image, and thus the security of diagnosis.

In summary, then, using modern technology, with practice and experience, an accurate diagnosis of CHD can be made by the expert in the first trimester, particularly in the high-risk fetus. Because of the high rate of association of CHD with increased NT—at least 25% of fetuses with major CHD have an increased measurement3, with an increasing incidence the greater the nuchal measurement—this may be a more important high-risk group than any other traditionally considered at increased risk and referred for fetal echocardiography. However, an extremely important caveat must be borne in mind. We have learned from detailed follow-up of second-trimester scans that a few cases of major CHD develop or progress late in gestation and that even with hindsight, on review, the earlier scan was either completely normal or showed only minor abnormalities. This has been found to apply particularly to aortic4 or pulmonary stenosis and coarctation of the aorta. Dramatic progression of severity between 20 weeks' gestation and term is unusual; my estimate would be of the order of 1/5000 high-risk scans. At the present time, however, we do not yet know how frequently CHD progresses between 12 weeks and 20 weeks. We have had personal experience of a typical case of pulmonary atresia with intact septum at 20 weeks, which showed a normally sized right ventricle with normal flow into it, and into the pulmonary artery, at 12 weeks, and I have seen a similar case of aortic atresia presented at a meeting (Heling K-S, Chaoui R. unpubl. observ.). In another of our cases, mild ventricular disproportion was noted in a fetus at 12 weeks but the arch was specifically noted to be of good size compared to the arterial duct. A classic picture of coarctation, including arch hypoplasia, was subsequently found at 20 weeks. Thus, it is fair to reassure the patient that, where good quality images have been obtained at 12 weeks, the heart appears to be normal but a follow-up scan must be recommended, ideally at around 20 weeks. On the evidence we have at present, it is likely that progression from a passably normal scan at 12 weeks to an important defect at 20 weeks is as uncommon as it is between 20 and 40 weeks, but greater experience with larger numbers is needed to identify the correct incidence of this phenomenon.

In the second paper on cardiac screening, Tegnander and Eik-Nes compared the impact of the examiner's experience in detecting CHD in a low-risk population5. The authors had the important aim of identifying reasons for failure and of understanding ways to improve training. The setting for the study was perfect, in that the data were strictly based on a local population, prospectively registered, with an admirable acceptance rate of ultrasound screening at 18 weeks in 97% of the pregnant population in the region. It involved almost 30 000 patients over a 10-year period between 1991 and 2001, with complete follow-up. The expected rate of ascertainment should be approximately 2–3 cases of major CHD per thousand live births, so the identification of 82 cases in their population seems about right.

As in other studies over the years, the results are rather disappointing in that major CHD was still missed even by the experienced sonographer, despite a structured program that aimed at four-chamber view and great artery analysis in all patients, during a 30-min time slot for each anomaly scan. However, those of us who work in referral centers should not underestimate the difference between their practice and that of routine scanning, where the vast majority of patients will have an absolutely normal scan in all its parameters. This is exemplified in this study by the increase in detection rate of CHD when there were also extracardiac findings. After 1995, because of the low rate of failure to obtain the cardiac views—the proportion of non-detection of the four-chamber view fell from 3% to 1%—it became practical to offer a follow-up scan if the heart was not adequately seen at the first scan.

I would have some minor differences of opinion on the authors' assignment of cases to detectable on four-chamber view scanning or not (their Table 3). For example, complex transposition and pulmonary stenosis, even when severe, are quite compatible with a normal four-chamber view, but corrected transposition and simple coarctation are not. In tetralogy of Fallot, the four-chamber view is usually normal and in aortic stenosis, depending on severity, it may be normal. It is disappointing that the pediatric cardiologist to whom the authors turned for an expert cardiac opinion missed CHD which had been suspected by the midwife–sonographer. It is interesting that inadequate imaging on the routine scan was attributed to fetal position in 40% of those cases where the cardiac structures were not fully identified. But this can usually be overcome by manual manipulation of the fetus or by turning the patient on one side or the other or, alternatively, by making them walk about for 10 min or so. There is little we can do about the increasing proportion of our pregnant patients who are obese, but I often think we are a little lazy in accepting a poor fetal position as an excuse, and I include myself in this criticism.

As the authors point out, obtaining the view or views is not enough. They must be analyzed correctly, particularly the four-chamber view, which is the key to cardiac evaluation. In 1986 we predicted from pediatric experience that the four-chamber view would be abnormal in about two-thirds of major cardiac malformations presenting in infants6—the figure in this paper is 71%. Not only is the four-chamber view vital in the analytical sense, but it is the reference point from which to initiate the great artery sweep in a technical sense. In my experience of teaching sonographers, it is easy to add the great artery views, once the operator has gained the skill to competently assess the four-chamber view every time.

I recently reviewed a paper which showed that a teaching program aimed at adding great artery views to routine four-chamber view scanning had the unexpected side-effect of improving the detection rate of four-chamber view anomalies, presumably just by dint of reinforcing methods of analysis and focusing attention on the heart.

I could not agree more with the authors in their conclusion that adding extra views or fancy technologies is not the main answer to improving the detection rate of CHD at the routine screening level. We must concentrate on teaching new personnel the basic steps which, once confidently mastered, can lead on to the addition of further skills. This teaching effort must be continually renewed and reinforced. For the experienced sonographer, regular exposure to the appearance of possible malformations that will be rarely encountered in their setting is vital. Fortunately in this day and age, with the ease of showing clips during teaching sessions, this is not difficult.

So where are we with screening for CHD during the routine obstetric scan at the present time, in this country and elsewhere? Much of our data unfortunately are now a little dated. In Europe, from data collected during the 1990s7, the rate of detection of CHD varied from 11 to 48%, depending on the local policy on routine ultrasound screening. In a paper by Bull8 covering the period 1993–1995 in the UK, 23% of CHD was detected nationwide but there was a marked geographical variation in detection rates. It would be very unusual nowadays for a case of hypoplastic left heart syndrome, for example, to present as a neonate to a pediatric cardiology center in the southeast of England, as nearly all cases are diagnosed in the fetus. In contrast, only 25% of neonates with an atrioventricular septal defect had been detected prenatally in a recent series reported from Leeds (north of England)9. Although some small atrioventricular septal defects can be difficult to detect, the majority of cases should be detectable on four-chamber view screening. So it is likely that success is still geographically variable and dependent on attitudes, motivation, training and, as the paper by Tegnander and Eik-Nes clearly shows, experience in the local obstetric screening unit. The conclusion of the RADIUS study10, published in 1993, should have been that prenatal ultrasonography in the hands of untrained personnel was not able to detect major fetal anomalies and therefore ultrasound screening had no impact on perinatal outcome. This should have been a cause for shame at the low standards practiced in a developed country and for effective recommendations for improvement, rather than the implication that it was somehow the technology that was at fault. However, there is some scope for optimism. A study from Atlanta, Georgia, covering the period 1990–1994, showed an increase in prenatal detection of CHD during the time of the study from 2.6 to 12.7%11. A more recent paper from the US showed an increase in prenatal diagnosis in cases presenting to a specialist cardiology center from 8% to over 50% in the period 1992–200212.

In conclusion, therefore, the detection of CHD during routine obstetric scanning is inexorably improving, albeit slowly and rather variably and with much effort on the part of both the ultrasonographer and the educator. It is likely that the addition of increased NT as a targeted high-risk group for referral for detailed fetal echocardiography will increase the detection rate of CHD. However, training for the pediatric cardiologist involved in fetal scanning and diagnosis also needs to improve. The purpose of prenatal diagnosis is that it can allow for termination of pregnancy in the most severe cases of cardiac malformation, where the prognosis for the child is poor. As a result of NT screening, cardiac diagnosis can be accomplished at an increasingly early gestational age, which makes pregnancy interruption easier emotionally for the parents and safer physically for the mother. In continuing pregnancies, delivery at the site of surgery in those cases requiring immediate cardiological care has been shown to improve mortality13. In the future, it is likely that long-term morbidity, especially in terms of neurological deficit, will also be shown to be improved by prenatal diagnosis. It is therefore of great importance that further efforts are made to detect CHD in the fetus.