It has been known since the 1980s that the maternal serum levels of certain pregnancy-derived proteins are shifted away from the median in pregnancies affected by Down syndrome. In the second trimester (14–20 weeks), for example, hCG values in a pregnancy affected by T21 are approximately twice those of a normal pregnancy, and the maternal serum AFP is approximately half. The higher the hCG, and the lower the AFP, the higher the risk for Down syndrome becomes. Low hCG and high AFP levels are conversely associated with a lower risk of Down syndrome. Computer algorithms combine maternal serum levels with the maternal age at conception and compute an adjusted risk value for T21, which is usually expressed as the chance of a livebirth of a baby with Down syndrome. In the 1980s and early 1990s in the UK, if this risk was higher than 1 in 250, the result was described as “screen positive” and an amniocentesis was offered. This was called the double test, and it carried a false positive rate (FPR) of 5 %, with a sensitivity of only approximately 60–65 %. The later addition of oestriol, and then inhibin A, described as the triple and quadruple tests, respectively, has helped to improve the sensitivity. The quadruple test remains the second trimester screening test recommended in the UK by FASP [1], where the standard to be reached is a DR of 80 % for a FPR of between 2.5 and 3.5 %. Only women with a risk of 1 in 150, or greater, are now offered invasive testing with amniocentesis. This detection rate of 80 % means that 1 in 5 affected pregnancies will be deemed “low risk” by the screening test (so the diagnosis will be missed) and approximately 1 in 30 women will have a false positive screening result. Furthermore, not all screening programmes using the quadruple test have been able to reach these standards.

There is an understandable desire for Down syndrome screening to occur as early in pregnancy as possible. The levels of pregnancy associated plasma protein A (PAPP-A) tend to be lower in Down syndrome pregnancies in the first trimester (9–13 weeks), and those of human chorionic gonadotrophin (hCG) and free β-hCG tend to be higher. Even when used together, these first trimester biochemical markers have insufficient sensitivity to constitute a viable screening test, but they have been very successfully combined with nuchal translucency scanning (see below – the combined test) resulting in a much higher sensitivity despite lower false positive rates. The addition of first trimester maternal serum levels of AFP and placental growth factor (PlGF) may further benefit the performance of first trimester Down syndrome screening protocols in the future.

There are also additional first trimester ultrasound features that can be used to further refine the risk for Down syndrome. Tricuspid regurgitation, reversal of the “a” wave in the ductus venosus, and absence or hypoplasia of the nasal bone are all more common in the fetus with trisomy 21. These are perhaps even more demanding to perform than a nuchal translucency measurement and their use tends to be confined to private services, and fetal medicine units.

Second trimester scanning can also be used to adjust the risk for Down syndrome [2], although opinions vary significantly with regard to the value of the “genetic sonogram” in this regard [3]. Finding a congenital heart defect will significantly increase the risk of Down syndrome, with approximately 40 % of fetuses with major septal defects having aneuploidy, very commonly T21 [4]. However, short femur, increased nuchal fold, echogenic cardiac foci, mild renal pelvic dilatation, echogenic bowel, mild cerebral ventriculomegaly and absent or hypoplastic nasal bone have all been found more commonly in pregnancies affected by T21. Some of these ultrasound features carry a greater likelihood ratio of T21 than do others, and the more of these features that are present, the higher the risk becomes. Absence of any of these features on ultrasound at 18–23 weeks probably does reduce the prior screening or age related risk for Down syndrome and Agathokleous et al [5] have calculated that the combined negative likelihood ratio is 0.13.

The UKNSC issued a Programme Statement in 2009 following review of the available evidence. This stated that women who were found to be at low risk of DS following a formal first or second trimester screening test should not have this chance value recalculated in the presence of the following findings on the second trimester scan:

These “soft markers” were considered to have too weak an association with T21 to be of value. The statement went on to say that there are other findings which should be reported and should prompt referral for further assessment. These are;

It is not clear from the statement what effect, if any, these should have on the quoted Down syndrome risk. Most fetal medicine specialists will offer amniocentesis if there is cerebral ventriculomegaly, a significantly small baby or an enlarged nuchal fold. Some still also offer invasive testing for a finding of echogenic bowel.

It is clear then that there are a plethora of variables, accessible to prenatal scrutiny, which can be used to refine a risk for Down syndrome. Because these variables are independent of one another, each can be used to separately adjust the maternal age related risk, either up or down. The more variables are included in the screening protocol, the higher the sensitivity becomes for a fixed false positive rate. Worldwide, the protocols available through state funded care, or private care, are determined by resources, moral and ethical values.

A “combined” test describes a screening protocol where a number of ultrasound and biochemical variables are tested at approximately the same gestation, and their effects on the Down syndrome risk are superimposed. This term is used in the UK to describe the current NSC “gold standard” of Down syndrome screening which is an NT scan at 12 weeks gestation with first trimester PAPP-A and free β-hCG values between 10 and 13 weeks gestation. The performance of this test varies, but a recent position statement by the International Society for Prenatal Diagnosis quoted an 80 % sensitivity for a 3 % FPR. The FASP standard for the combined test is a sensitivity rate of 85 % for an FPR of between 1.8 and 2.5 %. Adding in examination for the nasal bone to the combined test increases the sensitivity to 91 %. Failure to achieve a technically satisfactory NT measurement remains a problem, especially in women with a raised BMI, and some women continue to book later than the NT window. Current UK recommendations are for these women to be offered the quadruple test as an alternative.

Integrated testing describes the biochemical testing of the pregnancy in both the first and second trimesters, with or without scan variables, and only giving a risk estimate after the second trimester component. Sensitivity rates comfortably exceed 90 % for a 3 % FPR; however, the risk estimate is provided relatively late in gestation and the protocols are resource heavy. Contingency screening is a compromise between combined and integrated testing. Following the first round of the screening protocol in the first trimester, only the women with borderline risk estimates go forward to the second stage. Women with a very high risk are offered invasive testing immediately and those with a very low risk are not offered the second round. Approximately 1 in 5 women will go forward to the second stage, and this significantly saves on resources without seriously compromising detection rates.

As testament to the efforts of research teams, and the Fetal Anomaly Screening Programme, detection rates for T21 have increased significantly, hand in hand with a reduction in the false positive rate. Fewer women are labelled “high risk” and have to face the dilemma of invasive testing, and yet the detection rate for T21 has continued to climb. However, the PPV of even the best screening protocols remains only 3–4 %, meaning that approximately 30 invasive tests are performed for every affected pregnancy detected. With a miscarriage risk of 0.5–1.0 % associated with amniocentesis and CVS, the potential for iatrogenic loss of an unaffected pregnancy remains evident.

“Cell free” DNA (cfDNA) refers to fragments of DNA which circulate in plasma having been released from the nuclei of damaged or dying cells. They have a short life span but because they are continually escaping from cells there is a relatively steady state. The fragments are random in size and chromosomal origin but in totality cover the entire genome and give an indication of gene dosage, i.e., how many copies of a particular part of the genome exist in that individual. Tumour biologists recognised some time ago that cfDNA arising from tumour cells could provide non-invasive genetic information regarding a malignancy by the analysis of a simple blood test from an affected individual. Even though cfDNA from cancerous cells could not be physically isolated from the cfDNA of normal cells, somatic genetic mutations contributing to the malignant potential of a tumour could be identified by studying the cfDNA because their DNA sequences differed from those of the host DNA. In 1997, Lo [6] showed that a pregnancy also contributes to the pool of cfDNA circulating in a pregnant woman. As the “host” she would not be expected to have cfDNA derived from the Y chromosome in her circulation, and identification of such Y-chromosomal DNA fragments using sensitive PCR techniques applied to her plasma, from a simple blood draw, would indicate she was carrying a male fetus. This was the first application of this knowledge and technique to prenatal testing and was soon joined by the non-invasive testing of fetal Rhesus D blood group in pregnancies complicated by Rhesus D isoimmunisation. Rhesus D negativity is usually caused by a deletion of the entire RhD gene. A RhD negative woman with anti-D antibodies would not be expected therefore to have cfDNA fragments from the Rhesus D gene in her circulation if she was carrying a RhD negative fetus. A RhD positive fetus would contribute RhD DNA fragments to the total cfDNA pool, and these would then be detectable using PCR techniques. Prior to this non-invasive prenatal testing (NIPT), an amniocentesis would have been required to ascertain the RhD status where the father of the baby was RhD heterozygous. Both these applications are now common practice in clinical genetics and fetal medicine clinics, and there are a number of other single gene disorders that can be tested for non-invasively in this way. Until recently, however, the clinical applications in a prenatal setting have been limited to this relatively small group of specific indications. A shift in thinking and rapid progress in DNA sequencing technology have now moved NIPT into the realm of screening for the common trisomies and testing for some of the more common deletion syndromes.

These genetic disorders result in a quantitative change in DNA, rather than a qualitative change. Individuals with T21 do not have genetic mutations per se, they have an extra copy of the otherwise normal genes on chromosome 21. This initially excluded NIPT from this field of prenatal testing because the early techniques relied on a difference in sequence between the mother and her unborn baby. However, as our ability to sequence DNA fragments has improved exponentially, it is now possible to sequence literally millions of DNA fragments in a matter of hours. The chromosomal origin of these fragments can be identified because we have mapped the entire human genome. Although the feto-placental unit makes only a small contribution of its cfDNA to the total pool of cfDNA in the maternal circulation, this is sufficient to mean that if the fetus has an imbalance in its chromosomal make-up, this will be detectable in the maternal pool of cfDNA. So, a fetus with trisomy 21 will contribute more chromosome 21 fragments to the cfDNA pool, meaning that a greater proportion of the total pool of cfDNA fragments will come from chromosome 21, even though the fetal cfDNA cannot be identified separately from the maternal. The technology behind this is remarkable, and it is being improved and becoming cheaper all the time.

A number of different techniques have been developed, the details of which will not be described here. They each rely on meticulous laboratory standards and complex statistical algorithms to maximise the sensitivity and positive predictive value of the testing process.

The use of cfDNA in the testing of pregnancies for Down syndrome and other trisomies has been developed and promoted, until very recently, by the commercial sector, with biotechnology companies in North America and Hong Kong dominating the market. There are now a plethora of published studies attesting to the power of these new techniques (summarised in a meta-analysis by Taylor –Phillips [7]). However, the picture emerging is that NIPT cannot be considered a diagnostic test. The headlines quoting detection rates of 99 % and false positive rates of <0.1 % are eye-catching; however, the test may not perform quite as well when confined to the end of the first trimester, or when used in a general obstetric population, rather than the high risk groups common to many of these studies. A recent meta-analysis of 41 studies [7] also found publication bias, which will further overestimate test accuracy. The International Society for Prenatal Diagnosis [8] have recently calculated a positive predictive value of 56 % for cfDNA testing for Down syndrome, although the Taylor-Phillips meta-analysis put this at 91 %. The latter figure would mean that when a cfDNA test gave a high risk for Down syndrome, in 9 out of 10 cases a subsequent invasive test would confirm the diagnosis. All commercial providers firmly recommend invasive testing for women with a cfDNA test showing a “high likelihood of an affected pregnancy.” Of course, when compared with the PPV of 3–4 % of current screening methods for Down syndrome, this is a major advance. Far fewer women are given a “high risk” result, and of those that are, nearly all will be carrying an affected pregnancy. Decision making will be easier, and the number of test-related miscarriages will fall dramatically if cfDNA is introduced widely.

The RAPID study [9] is the only one to date to use cfDNA testing for Down syndrome in a publicly funded “real life” healthcare system and, as such, gives the most valuable insight into how cfDNA testing would perform if it was introduced as part of a national screening programme. RAPID used cfDNA testing in a contingent manner, i.e., all women were offered the combined test or quadruple test as per current NSC guidelines. Those with a risk >1 in 1000 for Down syndrome were then offered NIPT. If this test indicated a high likelihood of Down syndrome, they were then offered invasive testing. In the final analysis, the RAPID team were able to set the threshold for the offer of NIPT at 1 in 150, 1 in 500 or 1 in 1000. At all thresholds, the overall detection for Down syndrome was increased above that of the current screening protocol, and there was a huge reduction in the number of invasive tests performed, and the number of test-related miscarriages. However, with current costs of cfDNA technology, only the use of a threshold of 1 in 150 kept costs for the screening programme neutral (in fact it had a small associated saving). As the costs of testing fall, the threshold of the primary screening test at which NIPT can be offered will also fall, with an associated increase in the proportion of Down syndrome pregnancies detected. This leads to the question as to whether cfDNA testing should be offered to all women, as the primary screening test. Costs are prohibitive currently for this option, and there are realistic concerns that test accuracy will fall, and the number of invasive tests performed will climb again, despite only a relatively small increase in the additional cases of Down syndrome detected. Concerns have been raised about losing the nuchal translucency scan, and the use of first trimester biochemical markers, because of the other pregnancy complications these test components can be a marker of.