Objective
The purpose of this study was to determine the performance of Down syndrome screening in triplet pregnancy.
Study Design
Nuchal translucency (NT; n = 794), nasal bone (n = 219), and biochemistry (n = 198) were evaluated in triplet pregnancy. Screening performance was evaluated with the use of delta and Gaussian models.
Results
The median multiples of the median values for free beta human chorionic gonadotropin and pregnancy-associated plasma protein A were 2.86 and 3.48, respectively. A significant correlation in delta NT within pregnancy was observed (0.46-0.68). The modeled false-positive rates were 11.7%, 7.4%, and 8.9% with the delta model and 11.9%, 6.6%, and 12.0% with the Gaussian model for NT, NT + nasal bone, and NT + biochemistry. Based on simulation, the detection rate at 12 weeks’ gestation was 78%, 93%, and 80% for NT, NT + nasal bone, and NT + biochemistry at a 10% false-positive rate using either the delta or Gaussian models.
Conclusion
In triplet pregnancy, the addition of nasal bone lowers the false-positive rate of nuchal translucency screening. More data are required on the effectiveness of biochemistry.
The incidence of triplet and higher order multifetal gestation in the United States has risen several hundred percent since 1980, primarily because of the increasingly widespread availability of fertility therapies. The natural incidence of spontaneous triplet pregnancy is approximately 1 in 8000 births ; however, approximately 95% of triplet pregnancies now come from infertility therapies. By comparison, triplet births accounted for 1 in 2000 pregnancies in 2006. Because women who carry triplet pregnancies also tend to be older than those who carry singleton pregnancies, the questions of prenatal screening and diagnostic procedures have become more prominent. Most maternal serum markers that are used in prenatal screening are associated with the number of fetuses in utero with the median multiples of the median (MoM) values approximately equal to the number of fetuses, although some significant variation from those levels is well known. As a consequence, screening beyond twin pregnancies with second-trimester maternal serum has never been practical.
For Editors’ Commentary, see Table of Contents
First-trimester screening for chromosome abnormalities with the use of free beta human chorionic gonadotropin (β-hCG), pregnancy-associated plasma protein A (PAPP-A), and nuchal translucency (NT) now is offered routinely to patients in the United States, and first-trimester screening is replacing second-trimester screening as the primary method of evaluation. In addition, some centers can now provide an assessment of nasal bone during the ultrasound examination. The combination of biochemical and ultrasound screening initially was introduced primarily for singleton pregnancies. In multifetal pregnancies, the evaluation of ultrasound markers has advantages because an assessment of each fetus can be determined. However, the biochemical assessment applies to the pregnancy as a whole and cannot be applied directly to the individual fetus because there is no way to differentiate among the contributions of each fetus to the total. A number of reports have demonstrated that the addition of biochemical testing to NT assessment in twin pregnancy leads to improved screening performance, although in dizygotic twins the screening performance is still not as good as in singleton pregnancy.
The diminished screening performance in dizygotic twin pregnancy is because the biochemical markers will blend the contribution of the 2 fetuses. For example, for a marker that is 1 MoM in a normal singleton pregnancy and 2 MoM in an affected singleton pregnancy, the combination in a normal twin should model approximately 2 (1 + 1); in a twin pregnancy with 1 affected fetus in whom the analyte is elevated in Down syndrome pregnancies, the combination becomes 3 (1 + 2). Dividing the total by 2, the affected MoM is now 1.5 times that of the unaffected and therefore not as elevated as is observed in singleton (2.0) pregnancy. Similarly if the marker is 0.5 in a Down syndrome–affected singleton pregnancy, then the expected difference in twins discordant for Down syndrome would be just 0.75. Regardless, modeling and real patient data show that the addition of biochemical screening in twin pregnancies lowers the false-positive rate compared with ultrasound screening alone.
The difficulties that are involved with the use of biochemistry in twin pregnancies are further exacerbated in higher-order multiple pregnancies. With similar logic as described for twins, in triplet pregnancies the elevation that is derived from a single affected fetus is likely to be masked even further so that the elevation is only 1.33 (4/3) times that of the unaffected. However, with a large enough database on triplet pregnancies, this small differential may be enough to benefit the addition of biochemical screening to ultrasound screening. Recently, many centers have begun to evaluate triplet pregnancies using ultrasound markers alone. Here, we analyze a large set of triplet pregnancies for NT and nasal bone and evaluate the potential addition of biochemical screening.
Materials and Methods
We performed a retrospective analysis of all 794 sets of triplets between January 2003 and July 2010 in which NT measurement was performed and provided to the laboratory. Among these triplet pregnancies, nasal bone evaluation was performed in 219 patients, and dried blood was collected from 198 patients. NT was measured in all 3 fetuses for all cases, and nasal bone was measured in all 3 fetuses for 202 of the 219 pregnancies (92%) in which nasal bone was evaluated. All sonographers participated in either the Fetal Medicine Foundation or Nuchal Translucency Quality Review programs. For the 198 patients with dried blood specimens, 145 specimens were analyzed with in-house enzyme-linked immunosorbent assays for free β-hCG and PAPP-A, and 53 specimens were analyzed with in-house dual analyte DELFIA assays (Perkin Elmer Life Sciences, Boston, MA). The Institutional Review Board of Mount Sinai determined that the study did not constitute human research under Department of Health and Human Services and Food and Drug Administration regulations and was thus exempt.
The patients in the study had an average gestational age of 87.85 days (SD, ± 4.18 days) and an average maternal age of 33.5 ± 4.8 years. The subgroup that had nasal bone evaluation performed had an average gestational age of 86.45 ± 4.23 days and an average maternal age of 33.95 ± 5.0 years. The subgroup that had biochemistry performed had an average gestational age of 87.0 ± 4.36 days and an average maternal age of 33.48 ± 4.9 years.
Down syndrome risks were calculated for NT, NT + nasal bone, and NT + biochemistry. Only 11 patients had NT, nasal bone, and biochemistry so there was not enough power to evaluate the combination. Risks were calculated with the delta model. Namely, NT likelihood ratios were determined on the basis of delta values with the use of the algorithm for singleton pregnancies, and this likelihood ratio was applied to each fetus. Nasal bone likelihood ratios were determined with a published formula ; if any of the 3 fetuses did not have nasal bone assessment, then a likelihood ratio of 1 was used for that fetus. Biochemistry was factored in to the risk algorithm with the approach that Spencer and Nicolaides reported for twin pregnancies. The free β-hCG and PAPP-A MoM values were divided by the median MoM value that was observed in triplet pregnancies. The adjusted MoM values were input into the singleton risk algorithm for biochemistry to determine the biochemical likelihood ratio. For the combined NT and nasal bone protocol, the likelihood ratio for NT and nasal bone in each fetus were multiplied together to determine the overall likelihood ratio. For the combined NT and biochemistry protocol, the NT likelihood ratio for each fetus was multiplied by the biochemistry likelihood ratio to determine the overall likelihood ratio for each fetus. The overall likelihood ratio was multiplied by the previous risk of Down syndrome to get the posterior Down syndrome risk.
Risk for triplet pregnancies was also evaluated with a Gaussian model. The parameters for the NT distribution were based on previously published parameters. The parameters for the biochemical distribution were the same as in singleton pregnancies, except for the means. In the unaffected distribution, the mean log (MoM) of each analyte was set equal to the log of the median MoM that was observed in triplet pregnancies. For the affected distribution, an estimate of the median for each analyte was determined by the estimation of the median to be the sum of 2 plus the affected median in singleton pregnancy and then normalization of this value by multiplication by the ratio of the unaffected median in triplet pregnancies to 3.0. The affected mean was then set equal to the log of the estimated median.
For comparative purposes, 275,690 singleton pregnancies and 7455 dichorionic twin pregnancies in which both dried blood biochemistry using the DELPHIA assay and NT were performed between November 2009 and September 2010 were evaluated.
A risk cutoff of 1 in 300 pregnancies was used; a pregnancy was considered to be at risk if the risks for any of the fetuses in the pregnancy were above the cutoff point. Age-adjusted false-positive rates were modeled with the observed likelihood ratios and the age distribution of live births for singleton, twin, and triplet pregnancies as appropriate. Age-adjusted false-positive rates for triplet pregnancies were also modeled by application of the age distribution for singleton pregnancies to correct for the maternal age effect because triplet pregnancies were older than singleton pregnancies. In modeling for twin and triplet pregnancies, 1 year was subtracted from the age to account for use of egg donors. The 1-year adjustment factor was based on a review of our most recent 6-month data on multiple pregnancies in which the average age (based on the patient’s date of birth) was 1 year greater than the average age based on the patient’s eggs.
Detection rates for 1 affected triplet were estimated for testing at 12 weeks’ gestation by simulation of 100,000 sets of MOM values from the affected distributions that were described earlier and then application of the likehood ratio calculations described earlier. The simulated NT MoM values were adjusted by multiplication by the observed unaffected median.
Results
The median MoM values for free β-hCG and PAPP-A were 2.86 (95% confidence interval [CI], 2.69–3.08) and 3.48 (95% CI, 3.24–3.91), respectively. The median delta NT values for triplet pregnancies A, B, and C were −.12 (95% CI, −0.14 to −0.07), −.09 (95% CI, −0.11 to −0.06), and −0.09 (95% CI, −0.11 to −0.06) respectively. The median delta NT of all measurements was −0.10 (95% CI, −0.12 to −0.08). The Spearman rank correlation coefficient between the minimum and median, the minimum and maximum, and the median and maximum delta NT in each triplet pregnancy was 0.65, 0.46, and 0.68, respectively. With the use of MoMs, the median NT MoM was 0.93 (95% CI, 0.91–0.95), 0.94 (95% CI, 0.93–0.96), 0.93 (95% CI, 0.91–0.95) in triplets A, B, and C, respectively. The overall median NT MoM was 0.93 (95% CI, 0.92–0.95). The Spearman rank correlation coefficient between the minimum and median, the minimum and maximum, and the median and maximum NT MoM in each triplet pregnancy was 0.8476, 0.6704, and 0.7918, respectively. In all cases in which nasal bone was assessed, the nasal bone was present.
Screening with NT alone, 103 of 794 triplet pregnancies (13.0%) were at risk for Down syndrome, compared with 5.8% and 9.0% in singleton and twin pregnancies, respectively ( Table 1 ). Among the triplet pregnancies at increased risk, 63 pregnancies had 1 fetus at risk; 19 pregnancies had 2 fetuses at risk, and 21 pregnancies had 3 fetuses at risk. Overall, 164 of 2382 triplet fetuses (6.9%) were at risk for Down syndrome. When nasal bone was added to NT, the screen positive rate dropped to 11.9% of pregnancies (26/219; Table 2 ), compared with 2.8% and 4.7% in singleton and twin pregnancies, respectively. The addition of the biochemical markers to NT resulted in a 12.1% false-positive rate ( Table 3 ), compared with 5.3% and 6.4% in singleton and twin pregnancies, respectively.
