Noninvasive prenatal testing creates an opportunity for antenatal treatment of Down syndrome


  • Funding sources: None

  • Conflicts of interest: Dr. Bianchi is the Chair of the Clinical Advisory Board of Verinata Health, Inc., an Illumina company. She receives an honorarium for this role. Her laboratory also receives sponsored research funding from Verinata that is administered by Tufts Medical Center.


Trisomy 21 (T21) is the most common autosomal aneuploidy that is associated with intellectual disability. It is the focus of many prenatal screening programs across the globe. Pregnant women who receive a prenatal diagnosis of T21 in their fetus currently have the option of continuing or terminating their pregnancy, but no fetal treatment is available. In this paper, we review compelling morphogenetic, cellular, and molecular studies that, taken together, suggest that there is an important window of opportunity during fetal life to positively impact brain development to improve postnatal neurocognition and behavior. Although substantial progress has been made in understanding the basic neurobiology of Down syndrome (DS), the majority of pre-clinical trials is currently focused on adults. There are a number of challenges in the identification and development of novel antenatal therapies for DS, including the lack of toxicity and teratogenicity for the pregnant woman and the fetus, evidence that the compounds can cross the placenta and achieve therapeutic levels, and the demonstration of clinical improvement. Preliminary experiments in mouse models suggest that prenatal treatment of DS is an achievable goal. © 2013 John Wiley & Sons, Ltd.

In 2013, if a pregnant woman is told that her fetus has DS, she has two options: continuing or terminating her pregnancy. If she selects the former, she can meet with the subspecialists who will care for her child, and she can plan her delivery at a medical center that is capable of providing sophisticated care to a baby who has DS.[1] It is currently unknown how the integration of noninvasive prenatal testing (NIPT) into antenatal care, and in particular, its high sensitivity and specificity for the detection of T21, will affect the decision-making of pregnant women. In this paper, we will explore a third option, antenatal treatment of affected fetuses to improve postnatal cognition. Given that T21 can be detected as early as 11 weeks of gestation by using NIPT and 12 weeks of gestation by chorionic villus sampling, there is a potential 28 to 29-week time frame to therapeutically intervene. Here, we will review current information on the prevalence of live births with DS in the United States and selected other countries, fetal phenotypic abnormalities in affected human fetuses and different mouse models, and the limited experience with fetal treatment to date. Our goal is to emphasize the potential window of opportunity that exists to advance fetal brain development, as well as to highlight how the dynamics of first trimester prenatal screening and diagnosis might change if a therapeutic option becomes available in the future. Our vision is that prenatal intervention is just a beginning. Therapy would potentially continue postnatally, during infancy and childhood, when the brain is still actively developing.


There is an opportunity to treat fetuses with Down syndrome (DS) because not all pregnant women terminate affected pregnancies. International registries suggest that there are large regional variations in both the acceptance of prenatal screening for aneuploidy and termination of an affected fetus.[2] The International Clearinghouse for Birth Defects Surveillance and Research reported on 20 registries in 14 countries [nine in the European Union (EU) and five outside of the EU] over the period of 1993 through 2004.[2]

During this time, mean maternal age at conception increased, which significantly increased the mean prevalence of fetuses affected with DS from 13.1 to 18.2/10 000 births. There was significant regional variation in the termination of affected pregnancies. For example, in France, Italy, and the Czech Republic, termination rates increased, but in the United States (Atlanta region), Canada (province of Alberta), and Mexico, termination rates decreased. Over the 11-year period, the mean live born prevalence of infants with DS remained stable at 8.3/10 000 births, but certain countries, such as Norway and Israel, actually reported an increase in affected live births. Data from the National DS Cytogenetic Register for England and Wales suggest that 92% of English and Welsh women terminate their affected pregnancies, which is one of the highest rates in the world.[3] In The Netherlands, data from the national Perinatal Registry (1997–2007), which include trends in births that occurred before and after neonatal screening became available, show an unchanged prevalence in DS births.[4]

No national registry of births or cytogenetic diagnoses exists for the United States. To address this deficit, Egan and colleagues[5] used National Center for Health Statistics birth certificate data to develop a mathematical model in which they compared the expected with the actual number of DS births. They tracked US trends in DS live births over the years 1989 to 2006 and specifically examined the effects of maternal age, geographic region, marital status, maternal education and race, and Hispanic ethnicity.[5] They showed that the percentage of expected DS births varied by demographics, with the highest percentage occurring in the Midwest (67% of expected), and in women with less than or equal to 12 years of education (60% of expected), White women (56% of expected), and women of Hispanic ethnicity (55% of expected). Overall, the expected number of DS births declined between the years 1989 and 1998, before increasing on a steady trajectory. The limitation of this study was that these numbers did not distinguish between pregnant women who either were not offered or declined prenatal screening, and pregnant women who accepted screening and diagnosis and terminated their pregnancies.

Also, in the United States, Natoli et al.[6] recently reviewed 24 full-text publications that reported on termination rates following prenatal diagnosis of DS over the years 1989 to 2007. These studies included population (state)-based registries, hospital-based reports, and outcome investigations following the detection of fetal sonographic anomalies. Similar to the Egan et al. paper, this group documented significant variation in termination rates as a function of maternal and gestational ages, ethnic background, and geographic region. Pacific Islanders, Filipinos, and Hispanic women had the lowest termination rates. Over the period of the study, there was a significant decline in termination rates, with a current mean of 67% of US women terminating their pregnancies if affected by T21. These authors concluded, however, ‘…a single summary termination rate does not adequately address regional and demographic differences’.[6]


The fetal phenotypic abnormalities in DS are of great importance because they represent primary targets through which brain development could be positively affected by treatment. Current DS research focuses mostly on improving neurocognition, because many of the other problems associated with DS are amenable to surgical treatment (e.g. repair of congenital heart defects) or pharmacotherapy (e.g. treatment of hypothyroidism). However, current pharmacologic treatment is mainly being tested in adults. Here, we will explore the evidence for developmental differences that are already present in the human fetal brain that could potentially be treated in utero.

Anatomical differences

Anatomical differences between fetuses affected with DS and euploid fetuses are significant enough to be detected in vivo using prenatal sonography. The areas of the brain that are primarily affected and are visible by ultrasound examination are the cerebellum and the frontal lobe.[7] In a postmortem analysis of fetuses at 15 to 38 weeks of gestation, Guihard-Costa and her colleagues[8] examined the biometry of 355 fetuses with DS and 922 controls. They showed that skull brachycephaly and microcephaly were present in the trisomic fetuses as early as 15 weeks. Overall brain weight was significantly smaller in the affected fetuses, starting in the second trimester with region-specific alterations affecting the frontal lobe, the hippocampus, and the cerebellum. Similarly, the transverse cerebellar diameter was also significantly decreased. This very large study concluded that marked restriction of brain growth in the occipitofrontal dimension, and specifically in the supratentorial region, was evident by the midtrimester. A much smaller study of 17 fetuses with DS and 10 controls at 15 to 22 weeks of gestation demonstrated no morphological differences before 22 weeks.[9] After this time, there appeared to be a 25% to 50% reduction of fetal neurons, mainly in the granular layers of the cerebral cortex. Taken together, these studies suggest that commencing treatment in the second trimester should still be early enough to have a positive impact on brain morphogenesis.

Histological differences

Most stereological studies have demonstrated that neurogenesis and synaptogenesis are already abnormal during fetal life. For example, Guidi et al.[10] performed a postmortem study of hippocampal development in six fetuses with DS at 17 to 21 weeks of gestation compared with age-matched controls. They showed that all fetuses had a decreased number of neurons in the hippocampus because of decreased proliferation and increased apoptosis. They suggested that defective neurogenesis during a critical phase of brain building may be a critical component of the characteristic reduction in size of the cerebral hemispheres, frontal lobe, temporal cortex, hippocampus, and cerebellum in individuals with DS. Significantly, these investigators found numerous proliferating cells in the ventricular zone of the hippocampus and the parahippocampal gyrus of the affected fetuses, indicating that these regions still have the capacity to proliferate up to at least 21 weeks of gestation, the latest age examined in their study.[10] This same group examined the cerebellum in fetuses with DS at 17 to 21 weeks and showed that the DS cerebellar volume was 79% of control fetuses.[11] This was due to impaired histogenesis and cytogenesis that began in early developmental stages and resulted in a severe hypocellularity of all cerebellar layers. The problem was a defect in proliferation rather than increased apoptosis.

A variety of other cellular abnormalities have been found in fetuses with DS. At approximately 19 weeks of gestation, Weitzdoerfer and colleagues[12] demonstrated a significant reduction in drebrin, a marker for dendritic spines-associated and synaptosomal-associated proteins. Abnormally delayed synaptic development in fetuses with DS was also confirmed by another group that showed a high percentage of primitive and a low percentage of intermediate synaptic contacts.[13]

Differences in gene expression

Despite abundant information on the morphometric and cellular abnormalities in DS, there is a clear gap in knowledge regarding the molecular alterations responsible for these changes. Mao and collaborators analyzed RNA transcripts in heart and brain from three fetuses with T21 and three age and gender-matched controls.[14] Their results showed a statistically significant overall increase in the expression of genes that map to human chromosome 21 (HSA21). These authors also suggested that secondary downstream effects on disomic genes would be likely to play a major role in the pathophysiology of DS. The most statistically significantly regulated functions that differed between euploid and DS cerebellar tissues included ‘integral to plasma membrane’, ‘RNA binding’, ‘structural constituent of ribosome‘, ‘G-protein coupled receptor activity’, and ‘transmission of nerve impulse’. The limitation of this study was the small number of brains analyzed.

To address the drawbacks of obtaining fetal brain tissue, our laboratory has studied cell-free (cf) nucleic acids in amniotic fluid supernatant, which is usually discarded after measurement of alpha fetoprotein levels. These cf nucleic acids are mainly fetal (and not placental or maternal) in origin.[15-17] In prior work, we performed a bioinformatics analysis of 476 transcripts that are ubiquitously present in the amniotic fluid of euploid midtrimester fetuses.[15] Twenty-three of these transcripts were organ specific, and seven were highly expressed by the fetal brain and spinal cord. In later studies, we performed additional tissue analyses; these showed that the genes in the amniotic fluid transcriptome are derived from fetal brain cortex and Cajal–Retzius cells.[17] These data suggested that amniotic fluid provides novel information about fetal brain development in vivo.

Of specific interest for DS research, gene expression differences in second trimester fetuses with and without DS matched for gestational age and sex were characterized.[16] We extracted cfRNA from amniotic fluid and identified 311 statistically significant differentially regulated genes (p < 0.05). Only five of these genes were physically located on chromosome 21, suggesting that a systems biology approach to identifying endpoints for therapy is warranted. Subsequent functional genomic analysis of the differentially regulated genes highlighted the importance of oxidative stress occurring in the fetuses with DS.[16]


Any attempts to offer prenatal treatment in pregnant women carrying fetuses with DS will need to be preceded by animal studies. The mouse is the preferred model because large segments of the murine genome are syntenic with the human. The three main strains that have been studied with regard to fetal phenotype are the trisomy 16 (T16) mouse, the Ts65Dn mouse, and the Ts1Cje mouse. T16 mice have a triplication of a subset of genes that are orthologous to HSA21 as well as other genes orthologous to other chromosomes. Unfortunately, however, T16 mice have high perinatal mortality. Both Ts65Dn and Ts1Cje are trisomic for genes that are exclusively orthologous to HSA21. Both strains have brain morphometric and cellular abnormalities that are similar to those observed in people with DS. Several studies have highlighted altered dendritic spine morphology, behavioral deficits, and abnormal dentate gyral electrophysiology in these mice (reviewed in Haydar and Reeves[18]).

Anatomical differences

In contrast to the brains of humans with DS, total brain weight and volume are unchanged in Ts1Cje and Ts65Dn mice compared with euploid controls. A close examination of different brain regions, however, revealed region-specific hypoplasia of the hippocampus and the cerebellum in both strains.[19, 20] A detailed time-course study in Ts65Dn mice at embryonic stages E13.5 to E18.5 showed that these brain morphometric abnormalities originate during fetal life.[21] The authors demonstrated that total fetal brain weight and cortical layer thickness were significantly decreased at days E13.5 to 16.5 but that the delayed growth recovered by day E18.5 and after birth. Most importantly, hippocampal size was consistently reduced at all embryonic stages and after birth. Similar to Ts65Dn embryos, Ts1Cje embryos at day E14.5 have brains that are significantly smaller with reduced cortical thickness and enlarged lateral ventricles,[22] although no information on the morphometry of the hippocampus was provided.

Histological differences

Stereological analysis of Ts65Dn and Ts1Cje embryo brain sections that were immunohistochemically stained with the cell proliferation marker Ki67 or obtained from pregnant females injected with 5-bromo-21-deoxyuridine (BrdU) at midgestation revealed a significant deficit in neurogenesis. Neuronal production was dramatically decreased in both the developing cerebral cortex and the hippocampus. This continued during the early postnatal life in both brain regions as well as in the cerebellum.[21-24] Defective neurogenesis is thought to trigger a long lasting postnatal alteration of excitatory and inhibitory neuronal balance,[21] which may actively contribute to the development of other histological abnormalities observed in DS, including abnormal synaptogenesis and axon myelination.

Differences in gene expression

At present, there is little information available regarding the fetal molecular changes responsible for the abnormal early brain phenotype in the Ts1Cje and Ts65Dn mice. Very few studies have investigated gene expression changes in the cerebellum or whole brain in the immediate postnatal period. The most comprehensive study focused on the postnatal cerebellar development (P0, P3, and P7, and P10) in Ts1Cje pups. The authors identified 12 genes that were consistently differentially expressed, but no further pathway analyses were performed.[24] This study provided limited molecular information on only the 15 574 genes (out of 23 158 in the mouse genome) that were represented on the mouse pangenomic microarray used. To more deeply explore the fetal gene expression changes and cellular processes altered in day E15 Ts1Cje embryos, our laboratory is currently implementing a systems biology approach using whole transcriptome mouse genome arrays. This work is ongoing.


Most DS treatments that are currently being developed focus on adults. We believe that treating DS in adulthood will only have a limited clinical impact (Table 1). For example, the number of neurons will never be normalized because adult neurogenesis generates fewer neurons compared with embryonic neurogenesis. Similarly, brain morphogenesis is largely fixed by the time of fontanelle closure.

Table 1. Comparison of treatment effects in prenatal and adult time frames
Phenotypic alterationsPrenatal treatmentAdulthood treatment
Brain morphogenesis  
MicrocephalyCan be fully treatedNo treatment possible
Cerebellar hypoplasiaCan be fully treatedNo treatment possible
Hippocampal hypoplasiaCan be fully treatedNo treatment possible
Cellular abnormalities  
NeurogenesisCan be fully recoveredOnly partially improved
SynaptogenesisCan be fully recoveredOnly partially improved
Axon myelinationCan be fully recoveredNo or only partial recovery
Gene expression data  
Genes crucial for brain developmentCan be fully correctedNo correction possible
Genes crucial for neuronal transmissionCan be fully correctedCan be partially treated
NeurocognitionCan be fully normalizedCan be partially normalized

Recently, research has started in the area of antenatal treatment of T21. The first approach has used vasoactive intestinal peptides (VIP). Human neonates with DS have increased levels of VIP in their umbilical cord blood.[25] Stimulation of astrocytes by VIP results in the release of neuroprotective neurotrophic factors such as activity dependent neuroprotective protein (ADNP) and activity dependent neurotrophic factor (ADNF). Active fragments of ADNP and ADNF, known as NAP and SAL, respectively, have been previously demonstrated in mouse models to have therapeutic potential for both fetal alcohol exposure and DS.[26] In one study, pregnant Ts65Dn females received NAP and SAL or saline placebo by intraperitoneal injection at gestational days 8 through 12. Neurocognitive outcome was assessed blindly at 8 through 10 months of postnatal age by using the spatial memory Morris water maze test.[26, 27] Prenatal treatment of Ts65Dn pups with NAP and SAL significantly improved spatial memory compared with the untreated pups. Prenatal administration of NAP and SAL had long-lasting cognitive effects by specifically increasing NR2A, NR2B, and GABAAβ3 gene expression to wild-type mice levels.

In our laboratory, we have used a systems biology approach to analyze gene expression data from the second trimester amniotic fluid of human fetuses with T21. We identified oxidative stress as the major functional abnormality in these fetuses.[16] We hypothesize that treating oxidative stress in utero will improve fetal neurogenesis and postnatal cognitive outcome. We used the Connectivity map database, which contains a large number of gene expression profiles from cell lines treated with Food and Drug Administration approved drugs, to identify candidate antioxidant molecules that can reverse the DS transcriptome. In a preliminary study, we used apigenin, one of the antioxidants suggested by the Connectivity map. Our preliminary results show that this molecule is able to statistically significantly reduce oxidative stress in amniocytes from fetuses with DS without causing significant toxicity.[28] Importantly, when apigenin was given antenatally to pregnant female mice, it significantly improved the postnatal exploratory behavior of Ts1Cje pups in the open field test.[28]

Long-term considerations for antenatal treatment in humans include the following: The proposed drug must be safe during pregnancy and have no teratogenic effects; it must cross the placenta and have an easy (preferably oral) route of administration. Pharmacokinetic experiments will need to be performed to determine the optimal timing for therapeutic effects.


The studies described here demonstrate conclusively that fetal brain development is abnormal in both humans with DS and mice with a model form of the disease. There is therefore a real opportunity to intervene prenatally to improve neurogenesis in affected fetuses. The challenges will be to identify compounds that (1) are safe for pregnant women and their fetuses, (2) can cross the placenta to achieve therapeutic levels in the fetus, and (3) show evidence of clinical improvement after birth. It would be very exciting if prenatal screening for T21 could create an opportunity to provide fetal treatment and ultimately improve neurocognition in DS. Preliminary experiments in mouse models suggest that prenatal treatment of SD is an achievable goal.


  • DS is the most common autosomal aneuploidy associated with intellectual disability.
  • Worldwide, most screening programs focus on prenatal detection of DS.
  • Research to improve neurocognition in people with DS is almost exclusively focused on adults.
  • Pregnant women carrying affected fetuses with DS can choose to continue or terminate their pregnancies, but there is no fetal treatment available.


  • We present data to show that many pregnant women continue their pregnancies when their fetus is affected with DS.
  • We review the published literature on brain pathology in human fetuses with DS and embryonic mice affected with a model form of the disease.
  • We summarize the limited available information on prenatal treatment approaches for DS and make the case that there is an important window of opportunity to positively impact neurogenesis and brain morphogenesis by providing treatment during fetal life.