In recent years, the focus of antenatal screening for Down syndrome has shifted from the second to the first trimester of pregnancy in most developed countries. This has occurred largely because the single most discriminatory marker, sonographic nuchal translucency thickness (NT), is only reliable at 11–13 weeks' gestation. Screening protocols which use NT together with maternal serum markers at 10–13 weeks can achieve a much better screening performance than do second-trimester serum-only protocols. For example, a ‘combined test’ using NT at 11 weeks with serum pregnancy-associated plasma protein-A and free β-human chorionic gonadotropin (β-hCG) at 10 weeks has a model-predicted 82% detection rate for a 3% false-positive rate (multivariate Gaussian model with marker parameters derived by meta-analysis[1, 2] and a standardized maternal age distribution). In comparison, a second-trimester maternal serum biochemistry ‘quadruple test’, using serum free β-hCG, α-fetoprotein (AFP), unconjugated estriol and inhibin-A, has a detection rate of only 64% for the same false-positive rate. A combined NT and biochemistry test with the routine addition of nasal bone (NB) determination has a predicted detection rate of 91% and with the routine addition of second-trimester quadruple markers the rate is 93%. A more efficient, ‘contingent test’ approach is to restrict the use of additional markers to those women with borderline risks based on the combined markers, resulting in detection rates of 89% and 90%, respectively. The borderline risk group for contingent testing would typically encompass about 15% of the population screened: two-thirds of positive results and about one-eighth of negative results.
Where does that leave second-trimester ultrasound markers of Down syndrome? These have four potential roles: in women with a high risk following first-trimester screening; in those with a low risk following such screening; in twins; and in all pregnancies in which first-trimester screening was not performed.
It is not uncommon for a woman who has had a positive combined test result to delay a decision over invasive prenatal diagnosis until ultrasound evidence for or against Down syndrome can be found at 14–24 weeks. This is more likely to be done if the risk is borderline positive. Under these circumstances, considerable care is needed in the interpretation of ultrasound findings and in this regard the review by Agathokleous et al. in this issue of the Journal will be of great value. Given this evidence, a simplistic interpretation, in which the presence of one or more markers is taken to be sufficient to tip the balance in favor of invasive testing and the absence of any markers is sufficient to contraindicate testing, is no longer acceptable. Instead, information from the scan should be used to revise the risk from the combined test: when a marker is present the risk will be increased and when it is absent the risk will be reduced, but the magnitude of these changes differs from marker to marker. Agathokleous and colleagues have quantified these likelihood ratios (LRs) by carrying out a meta-analysis of the published literature. Using all available data, combined with powerful statistical techniques, they have derived the most reliable LR estimates to date. On the basis of their analysis, a woman with an a priori Down syndrome risk of 1 in 100 and subsequent finding of intracardiac echogenic focus, echogenic bowel and no other markers would have a revised risk of 1 in 10; if the echogenic focus was an isolated finding it would be 1 in 105 and if there were no markers it would be 1 in 760.
When making such a revision, it is important to ascertain how the reported combined test risk was calculated. If the software was based on either the maternal age-specific birth prevalence or the estimated incidence of the disorder in the mid-trimester, no further action is needed. However, if it was based on the estimated incidence at the time of the test, an adjustment will be needed to account for intrauterine loss of Down syndrome fetuses between the time of the test and that of the scan. From prenatal diagnosis studies, an estimated 46% of these fetuses are lost between the late first trimester and term and 24% are lost between the mid-trimester and term.
Some women whose combined test result is negative may also benefit from second-trimester ultrasound marker determination. Those whose screening results are borderline negative and who are consequently anxious could, in most cases, be reassured by the absence of any markers, although the remainder will have further cause for concern. For such women, the LRs from the meta-analysis should be used to quantify the individual risk rather than interpreting the scan simplistically.
However, most women with negative combined test results are not anxious and although in many centers they will have a routine second-trimester anomaly scan, this is principally to detect or exclude fetal structural abnormalities rather than Down syndrome. Nevertheless, an anomaly that is associated with aneuploidy is occasionally found, generating considerable anxiety despite the first-trimester result. This may sometimes be resolved only by invasive testing. Again, rather than making a decision solely on the basis of the scan, it is best to incorporate the first-trimester risk by using the appropriate LRs.
Whilst there appears to be clinical utility in the ad hoc use of second-trimester ultrasound for individual women with borderline combined test risks, it may not be an effective public health policy if adopted routinely. Using data from the First and Second Trimester Evaluation of Risk (FaSTER) trial, Aagaard-Tillery et al. found that replacing the quadruple markers in a contingent test by ultrasound examination did not alter substantially the Down syndrome detection rate for a given false-positive rate. However, they also showed that second-trimester ultrasound substantially increased the Down syndrome detection rates for the first-trimester combined, second-trimester quadruple and sequential tests.
Biochemical markers have much poorer discriminatory power in dichorionic twins than they do in singleton pregnancies or monochorionic twins. This is because in twins discordant for Down syndrome, fetoplacental products from the unaffected fetus can mask abnormal levels produced by the affected fetus. Consequently, in twins shown to have a first-trimester ultrasound placental ‘lambda’ sign, many centers use only NT for screening, ideally allowing for the correlation between NT measurements between even dichorionic fetuses[7, 8]. In dichorionic twins, whether or not a full combined test is performed, the predictive value of a negative result is relatively low, so there is much to be gained by reassessing the risk using additional first- and second-trimester ultrasound markers. The reliability of such risk revision in twins is not known, but data on false-positive rates suggest that for some first-trimester markers it should not differ markedly from that in singletons. Findings in large series of twins have been reported showing similar false-positive rates compared with singletons for first-trimester absent NB and reversed a-wave in the ductus venosus. To our knowledge, there are no published twin series for second-trimester ultrasound markers. However, in the absence of evidence to the contrary, it is reasonable to assume that twin risks revised using the LRs derived by Agathokleous et al. from singletons will be reliable.
In many parts of the world, first-trimester screening is not generally available and in those countries with established first-trimester programs, some women are not screened because they present too late in pregnancy. Second-trimester screening with an anomaly scan may be an option for them, although in the UK, the National Fetal Anomaly Screening Program has stated that unscreened women should ‘have counselling based on maternal age and/or family history not on whether normal variants are found during scanning’.
Routine use of the second-trimester anomaly scan to calculate Down syndrome risk in women who were not screened earlier is likely to have a low detection rate. In the FaSTER trial, using the anomaly scan in this way, albeit using their own ultrasound marker LRs, yielded a 59% detection rate for a 3% false-positive rate. However, the results could be improved by incorporating concurrent biochemical testing. Modeling with these results predicts that combining the quadruple test with an anomaly scan routinely would increase detection to 80%, and for a contingent policy of using the anomaly scan in those with borderline quadruple tests it would be 77%. In countries already providing second-trimester biochemical screening, either as the main policy or just for those women not tested earlier, this might be attractive. However, caution is required because of quality control considerations, particularly in the context of mass screening rather than specialist fetal medicine.
The success of the combined test in yielding a high Down syndrome detection rate has been achieved by standardization of NT measurement, training, optimal use of the information and external quality assessment. The Fetal Medicine Foundation (FMF) was instrumental in developing and promoting a standard way of performing the measurement. Training is provided by The FMF, the Nuchal Translucency Education and Quality Review program and others, and methods have been published to assess the quality of each examination. Initially, NT results were expressed in mm units and compared with a fixed cut-off. Subsequently, it was realized that because NT changes with gestation, even within the 11–13-week window, it is better to express results as multiples of the gestation-specific median (MoM) or difference from the median (delta). The overlapping distribution of these transformed measurements fit Gaussian distributions, just like the biochemical markers, so that the Down syndrome risk for an individual woman can be calculated readily from her maternal age, family history and marker profile, including NT. This also facilitates external quality assessment programs designed to monitor the performance of each sonographer. Simple and robust performance indicators are used, including the median NT-MoM, the SD of log10 MoM and the slope (rate of change) in NT against gestation and, more recently, the cumulative sum chart (CUSUM)[15, 16].
If a second-trimester test combining ultrasound and serum markers is to be successful, it too should be encouraged to follow this paradigm. Many of the markers in the anomaly scan are qualitative; for example, intracardiac echogenic focus and echogenic bowel. These are not so amenable to external quality assessment. In contrast, other markers are quantitative, so could be expressed in gestational age-specific terms and monitored by performance indicators such as median, SD, slope and CUSUM. Nuchal skinfold thickness (NF), femur length (FL) and humerus length (HL) are often expressed in MoM, whilst hydronephrosis and ventriculomegaly are not. Nasal bone length (NBL) is usually expressed as a ratio of the biparietal diameter, which is a less precise method than is using MoM to allow for gestation. Absent NB is qualitative but could be classified as very low NBL below, say, 0.1 MoM. Generally, all of the qualitative markers are dichotomized into large or small using a fixed mm cut-off rather than making use of them as continuous variables; this is wasteful of information. FL and HL are highly correlated and it is usual to choose only one of these in any risk calculation; this is also wasteful. In biochemical screening, the multivariate Gaussian distributions used to calculate risk can incorporate correlated markers by having suitable correlation coefficients in the parameters.
Benn et al. investigated the idea of a quadruple test enhanced by simply measuring NF, FL and HL. All markers, biochemical and ultrasound, were expressed in MoM with suitable Gaussian parameters. The model predicted that the detection rate for a 5% false-positive rate was increased by 15% compared with that of the quadruple test alone. For the quadruple test with just NF, the increase was 11%. Cuckle and Benn also modeled the addition of just NF to a quadruple test and predicted a 9% increase in detection rate. Borrell et al. did the same analysis for second-trimester free β-hCG and AFP (‘double’ test) and predicted a 12% increase in detection.
In the same way as sonographic NT is a central component of the first-trimester combined test, NF could be central to a second-trimester combined serum and ultrasound screening protocol. Rather than using FL and/or HL as further ultrasound markers, Maymon et al. proposed using quantitative facial profile markers determined in the same plane as NF. One such marker is prenasal thickness (PT), which has been confirmed in a total of seven studies[20-26], and tables have been published for ‘bedside’ estimation of Down syndrome risk. In five of the studies combined, the median PT in 105 cases of Down syndrome was 1.33 MoM, with a very narrow SD of log MoM. Modeling predicted that a quadruple test combined with sonographic NF, NBL and PT would have a 90% detection rate for a 3% false-positive rate. Additional facial markers, that could increase detection even further, have been investigated. These relate to the smaller size and dorsal displacement of the maxilla in Down syndrome, such as the frontomaxillary facial angle, and the prefrontal space ratio[28, 29].
Hence, we now have second-trimester combined tests with the potential to achieve the same screening performance as that of the first-trimester combined test. In principle, there is no reason why we should not introduce these tests immediately. However, in practice, there are two prerequisites before this is likely to happen. First, experience with NT shows that standardization of the measurement techniques and external quality assessment will be needed, perhaps as an extension of existing first-trimester schemes. Second, a paradigm shift in thinking is required. The anomaly scan is an established part of late second-trimester antenatal care and there is an accepted list of ‘soft’ markers relating to aneuploidy; in the USA, the scan is even referred to as the ‘genetic sonogram’. Moreover, in some countries, this scan is the domain of maternal–fetal medicine specialists. The proposal to disengage NF from the anomaly scan, combine it with facial profile markers and have them measured routinely by technicians sounds revolutionary, but if the public-health will is sufficiently great as to provide an equable Down syndrome screening service for women who book too late for the combined test, these organizational difficulties can be readily overcome.
The advent of non-invasive prenatal testing, by analyzing fetal cell-free (cf)DNA in maternal blood, will radically change Down syndrome screening protocols in the near future. Local cost considerations will dictate how this is done, but in most localities it is likely that conventional screening markers will be retained and used to select women ‘contingently’ for cfDNA testing. If so, second-trimester markers will continue for some time to play a role in this field.