Screening for trisomy 21 in twins using first trimester ultrasound and maternal serum biochemistry in a one-stop clinic: a review of three years experience


* Dr K. Spencer, Department of Clinical Biochemistry, Harold Wood Hospital, Gubbins Lane, Romford, Essex RM3 0BE, UK.


Objective To evaluate the performance of screening for fetal trisomy 21 in the first trimester of twin pregnancies by a combination of maternal serum biochemistry and ultrasonography.

Design Three year retrospective review of screening performance.

Setting District General Hospital maternity unit.

Population All women booked to receive routine antenatal care at Harold Wood Hospital between 1 June 1998 and 30th September 2001. The population included 13,940 women of all ages presenting with pregnancies between 10 weeks 3 days and 13 weeks 6 days gestation. Of these, 230 had a twin pregnancy.

Methods Women booked into the clinic were offered screening using a combination of maternal serum free β-hCG and pregnancy-associated plasma protein-A (PAPP-A) and fetal nuchal translucency thickness. Women at increased risk of carrying a fetus with trisomy 21 or trisomy 13/18 (≥1 in 300 at sampling) were offered counselling and an invasive diagnostic procedure. Follow up of the outcome of all pregnancies was carried out. For women who on examination were at 14 weeks of gestation or greater, or for women presenting as late bookers beyond 14 weeks, screening was performed in the same time frame using only maternal serum free β-hCG and α-fetoprotein.

Main outcome measures The first trimester detection rate for trisomy 21 and all aneuploides, false positive rate, uptake of screening, uptake of invasive testing in women identified at increased risk and fetal loss rates after invasive testing.

Results Overall, 97.4% of the women with twins (224/230) accepted first trimester screening. The rate of detection of trisomy 21 was 75% (3/4). Fetal death at presentation was found in 3.4% of fetuses (16/460). Of women who accepted screening, 4.3% (10/230) presented too late for fetal nuchal translucency measurement and 10.0% of women (23/230) presented too early. A risk for trisomy 21 was calculated for each fetus based on the individual fetal nuchal translucency thickness and the maternal biochemistry. The false positive rate among those eligible for first trimester screening was 9.0% (19/206) of pregnancies and 6.9% of fetuses (28/412). Uptake of invasive testing was 59% (10/17) with chorionic villus sampling in eight cases and amniocentesis in two. No fetal loss occurred within 28 days of chorionic villus sampling and no loss occurred after amniocentesis. One case of trisomy 21 was identified for every three invasive procedures.

Conclusion First trimester screening for trisomy 21 in twin pregnancies is both theoretically possible and practically achievable using a combination of nuchal translucency thickness and maternal serum biochemistry. However, dilemmas for the mother and health professionals when both nuchal translucency thickness measurements are normal might suggest that greater reliance be placed on the nuchal translucency thickness risk alone when counselling women about invasive testing.


In the first trimester of pregnancy, screening by a combination of fetal nuchal translucency thickness and maternal serum free β-human chorionic gonadotrophin (β-hCG) and pregnancy-associated plasma protein-A (PAPP-A) has been shown in retrospective1 and prospective studies2,3 to identify approximately 90% of cases of trisomy 21 in singleton pregnancies for a 5% false positive rate.

The incidence of twin pregnancy is increasing in the UK, partly as a result of the increased use of assisted reproductive techniques. Furthermore, the pregnancy population is getting older and the rate of twinning increases with maternal age. In 1979, the incidence of multiple pregnancies was 0.97%, this has increased to 1.43% by 19994. In the second trimester, biochemical screening of twin pregnancies was shown to be possible using a pseudo risk approach5,6. The detection rate in twins discordant for trisomy 21 was expected to be lower than in singleton pregnancies. Nevertheless, this approach identified twins concordant for trisomy 21 in prospective screenings7.

In the first trimester, both fetal nuchal translucency thickness8 and maternal serum biochemistry can be combined together9 to provide detection rates approaching those achieved in singleton pregnancies. Such algorithms do not seem to be dependent upon chorionicity10 and an example case of twins discordant for trisomy 21 that had been identified by prospective screening has been described11. Screening of twin pregnancies is, however, still considered by many to be problematical because of the significant clinical, technical and ethical challenges posed for the diagnosis and clinical management of such pregnancies6,12,13. Here we examine the outcome of screening for trisomy 21 in the first trimester of twin pregnancies over a three-year period in our OSCAR clinic2,3.


The maternity unit under study is located in the borough of Havering at the eastern boundary of Greater London. The screening population is predominantly white Caucasian from the London Borough of Havering (84.2% with an RM postcode) and the Local Authority of Brentwood (13.2% with a CM postcode), with a small proportion of women from the London Borough of Barking and Dagenham (2.3% with an IG postcode) and 0.3% from other adjacent areas. The two major areas comprise a relatively affluent population with approximately half the population in social class I or II. The 2000 Index of Multiple Deprivation provided by the Department of Environment, Transport and the Regions14 showed Havering with a rank of 233/354 districts in England and Wales, while that of Brentwood was 311/354. Barking and Dagenham on the other hand has a higher proportion of manual skilled, partly skilled and unskilled workforce. The Index of Multiple Deprivation in this Borough was ranked 24/354, showing considerably higher deprivation.

All women booked for maternity care at Harold Wood Hospital, Essex are given an appointment to attend the antenatal clinic for the first time at around 12 weeks. Screening using the first trimester one-stop approach began in June 199815. In the first year of screening, the qualifying gestational age was between 10 weeks 3 days and 13 weeks 6 days by ultrasound dating. During the second and third year, the minimum gestation was increased to 11 weeks 0 days to allow for better identification of structural anomalies at the slightly later gestations. Women are initially seen in the community by a member of the community midwifery team (65% of cases) and they receive an information leaflet about the one-stop clinic service and the screening tests that are available with their initial booking appointment letter. On attending the clinic at a specified appointment time, women have a consultation with a midwife concerning the available tests and, if they elect to have prenatal screening for chromosomal anomalies, all relevant clinical information is recorded on the request forms. After pre-test counselling, women opting into the screening programme go to the phlebotomy room where blood samples are taken for serum biochemical screening, in addition to samples for other routine antenatal investigations. Prenatal screening blood samples are passed through to the adjacent clinic laboratory and the women then move on to the ultrasound suite where fetal nuchal translucency thickness and crown–rump length are measured using standard procedures16 ( by sonographers certified by the Fetal Medicine Foundation. At the same time, a mini anomaly scan is also performed. The whole ultrasound process can be completed within a 20-minute period in 99% of the cases. All relevant clinical information is recorded on a networked fetal database (ViewPoint, Webling, Germany). If the ultrasound dating reveals the fetus has a gestation prior to 11 weeks (45 mm crown–rump length—38 mm in year one), then the mother is sent back to the reception to be rebooked at the appropriate period. If the gestation is beyond 13 weeks 6 days (84 mm by crown–rump length), then the laboratory is informed and α-fetoprotein is measured instead of PAPP-A.

During the time the mother is in the ultrasound suite, the serum from the blood sample is separated and analysed for free β-hCG and PAPP-A using the Kryptor analyser (Brahms Diagnostica, Berlin—formerly CIS) in the clinic laboratory. The quality performance of this system has previously been described1,17. When the results are available (within 20 minutes) these are logged onto the fetal database and a composite risk report is produced by the time the woman has returned from the ultrasound suite. This risk report is then available for the midwife/counsellor to discuss with the woman. Appropriate, further follow up and management is arranged. Patient-specific risks were calculated by a multivariate approach using populations parameters established in our retrospective study1, after correction of the maternal serum biochemistry for maternal weight18 and for twin pregnancies10, and the age-related risk of the trisomy in singleton pregnancies at the time of sampling 19,20. Although we have previously shown that in the first trimester free β-hCG levels are significantly increased (21%) in singleton pregnancies achieved by assisted reproduction21, no correction was made for this as no data have been published in twin pregnancies. Similarly, although the incidence of raised nuchal translucency thickness is higher in monochorionic twins than in dichorionic twins8, an appropriate correction algorithm has yet to be developed.

Women with an increased risk in either fetus (≥1 in 300) for trisomy 21 are referred to the Harris Birthright Centre for Fetal Medicine for chorionic villus sampling or amniocentesis and fetal karyotyping. Provisional results from quantitative PCR were available within 48 hours and a confirmed diagnosis by conventional karyotyping within seven days.

Outcome of all pregnancies was ascertained from delivery room records, hospital PAS system and Child Health records and was cross checked with the fetal database. Cytogenetics records were obtained from the laboratory, the Child Health computer records and the National Down's Syndrome register.


Within a three-year and four-month period (1st June 1998 to 30th September 2001), a total of 13,940 pregnant women were offered first trimester screening in the OSCAR clinic. Of these 230 cases were twin pregnancies (1.65%) of which 21% were conceived by assisted conception. The uptake of first trimester screening among women with a twin pregnancy was 97.4% (224/230). The population consisted of 84% Caucasian, 2% Afro-Caribbean, 3% Asian and 11% of either unknown or other ethnic origin. The median gestational age was 12 weeks 1 day (range 10 weeks 4 days to 13 weeks 6 days) and median crown–rump length was 59 mm (range 38–84). The median maternal age was 31.5 years (range 19.1–42.7). The median maternal weight of the population was 66 kg (range 43–185). Cigarette smoking was self-reported in 18.3%, with 80.2% reporting themselves as non-smokers and the status was unknown in 1.5% of cases.

Fetal death was noted on ultrasound examination in 3.4% of fetuses (16/460). According to the measurement of fetal crown–rump length, the gestation was prior to the minimum acceptable period in 10% of cases (23/230); these women were rebooked for repeat examination at the appropriate gestation. In addition, 4.3% of women (10/230) had a gestation too late for nuchal translucency thickness measurement and these women had conventional second trimester screening22 with α-fetoprotein and free β-hCG. In total, after exclusion of those women with fetal demise at presentation, those declining screening and those with crown–rump length beyond 84 mm, some 206 with a twin pregnancy had first trimester screening performed.

Of the twin fetuses screened in the OSCAR clinic, 6.8% (28/412) had risks greater than the 1 in 300 cutoff and 9.2% of pregnancies (19/206) had at least one fetus with an increased risk. After counselling, 37% of women (7/19) declined the offer of an invasive diagnostic test, while 63% (12/19) accepted invasive testing. Chorionic villus sampling was the procedure chosen by 83% of women (10/12) requesting invasive testing while 17% (2/12) decided to wait until 14 weeks for amniocentesis. In all cases of invasive testing when the fetus was shown to be of normal karyotype, the pregnancy continued with a viable fetus at least beyond 28 days after the procedure.

The individual combined risks and the fetal nuchal translucency thickness in those women declining invasive testing are shown in Table 1, indicating a predominance of women with risks very close to the cutoff and with normal nuchal translucency thickness in both fetuses.

Table 1.  Risk and NT MoM in twins when the mother declined invasive testing.
CaseRisk fetus 1Risk fetus 2NT fetus 1NT fetus 2

Among the twin study population, four cases with trisomy 21 were ascertained (Table 2). A combination of maternal age, fetal nuchal translucency thickness, maternal serum free β-hCG and PAPP-A identified 75% (3/4) cases with trisomy 21. There was one case of trisomy 21 detected per three invasive procedures. All women with an identified fetus with trisomy 21 elected to undergo embryo reduction, which was carried out by ultrasound-guided injection of potassium chloride in the chest of the affected fetus. In each of the three cases, the surviving twin progressed to a normal healthy delivery.

Table 2.  Cases of twins discordant for trisomy 21.
CaseMaternal age (years)Gestation (week + day)ChorionicityNT MoM 1NT MoM 2NT risk 1NT risk 2Free β-hCG MoMPAPP-A MoMBiochemical riskCombined risk 1Combined risk 2Outcome 1Outcome 2
13511 + 5Dichorionic–diamniotic1.431.86212461.860.6412012227NT21
23912 + 2Dichorionic–diamniotic1.032.5850330.800.461187584NT21
33312 + 5Dichorionic–diamniotic0.690.80298829881.570.5521417761776NT21
43512 + 1Dichorionic–diamniotic1.061.981162320.971.0615932230225NT21

Table 3 provides a brief summary of the obstetric history to each case.

Table 3.  Obstetric history of twin pregnancies discordant for trisomy 21. ICSI = intracytoplasmic sperm injection.
CaseGravidityParityIUD <15 weeksConceptionSmoker

Among the 19 women identified ‘at increased risk’, in addition to the three cases with trisomy 21, fetal death occurred in two sets of twins at 24 weeks. In one of these, twin-to-twin transfusion had occurred. In one further case, preterm delivery occurred at 26 weeks with neither twin surviving, and in yet a further case, one of the twins had a missing hand.

The median MoM free β-hCG after weight correction and before correction for twin pregnancy was 2.15 with a mean log10 of 0.3306 and a log10 standard deviation of 0.2544. The median MoM PAPP-A after weight correction and before correction for twin pregnancy was 1.93 with a mean log10 of 0.2628 and a log10 standard deviation of 0.2178. These values are very similar to those published in a previous series9.


The prevalence of twin pregnancies increases with maternal age. In addition to being twice at risk of structural defects, one would expect the risk for chromosomal abnormalities to be higher than in singletons23. However, when congenital malformation rates in twins have been studied, it has been difficult to demonstrate this for trisomy 21 and some studies have reported a lower rate of trisomy 21 in twins compared with singletons24. The difficulty with such studies is related to population size, as the incidence of twins is only 15 per 1000 births, even the largest study only included 42 cases with trisomy 21.

In the second trimester, maternal serum biochemical markers on average appear twice as high in twins compared with that of a singleton pregnancy at the same gestational age5,6,25. Using the ‘pseudo risk’ approach correcting for twins5, the provision of risks in twin pregnancies leads to detection rates in twins some 15% lower than in singleton pregnancies6,25. Although such approaches have been successful in routine practice7, many centres still consider the ethical and technical difficulties too problematical in routine obstetric units13. As a consequence, many do not offer screening in twin pregnancies.

In the first trimester, the use of individual nuchal translucency thickness has allowed the calculation of specific risks for each fetus. This physical marker can specifically identify the fetus at increased risk8 and thus be used as a guide when undertaking chorionic villus sampling and selective fetocide in twins discordant for trisomy 21. In addition, it has been suggested that the risk based on nuchal translucency thickness and maternal age can be used as the basis for making decisions regarding the appropriate diagnostic procedure to be followed in such circumstances26. The projected detection rate using fetal nuchal translucency thickness and maternal age alone was 75.2%, although this did not take into account chorionicity, in which the incidence of increased nuchal translucency thickness is 1.5 times greater in monochorionic than in dichorionic twins8.

Levels of first trimester maternal serum biochemical makers are similarly on average twice as high in twins than in singleton pregnancies of the same gestation9. Using a ‘pseudo risk’ approach, combining fetal nuchal translucency thickness and maternal serum biochemistry was predicted to identify 80% of cases of twins discordant for trisomy 21 (at a 5% false positive rate) and 81.5% in concordant twins. Thus, it was argued that while maternal serum biochemistry alone could not specifically identify the fetus at risk in the presence of twins discordant for trisomy 21, it would be possible to enhance the detection rate in twins by some 5–6% yet still retain the benefits of nuchal translucency thickness in identifying the specific affected twin9. Furthermore, it has been shown that chorionicity has no impact on the maternal serum biochemical marker levels in twin pregnancy10. The first practical example of this new first trimester twin algorithm was presented in a report11 of the first case in Table 2.

In this paper, we have outlined our first 3.3 years experience in screening twin pregnancies. Our incidence of twin pregnancies is a little above the national average but not greatly so. However, our incidence of twins discordant for trisomy 21 is far greater than one might have expected (1 in 58); based on the maternal age distribution of our twin pregnancies and the risk of trisomy 21 in singleton pregnancies at 12 weeks of gestation, we would have expected to see 1.5 fetuses with trisomy 21 rather than 4. Our detection of 75% of cases (3/4) is in line with our theoretical projection. The rate of acceptance of first trimester screening among twin pregnancies was no different from that observed in singleton pregnancies in which a very high uptake was achieved3. Fetal death was twice as common among twin pregnancies than in our singleton series, reflecting the increased perinatal mortality in multiple pregnancy. Uptake of invasive testing was lower than in the case with singleton pregnancies3 when 80% of women accept invasive testing. This reflects both the added risk and complexity of the invasive procedure in twins and the increased potential fetal loss of the normal twin as a result of selective fetocide. Clearly, instances when nuchal translucency thickness is normal in both twins, yet the biochemistry is abnormal, present a further dilemma in that even if chorionic villus sampling was performed on both twins—fetal reduction would be inadvisable as it would be difficult to identify the affected twin. Although adding maternal serum biochemistry does improve detection, clearly, nuchal translucency thickness risk alone is the predominant factor by which women should be counselled regarding invasive testing.


The OSCAR clinic is a multidisciplinary clinic which functions because of the dedication of all contributors to the team. The authors would like to thank the support and contributions from all those in midwifery, obstetrics, clerical, sonography, pathology and support workers.