Turner syndrome: New insights from prenatal genomics and transcriptomics

Abstract

In some parts of the world, prenatal screening using analysis of circulating cell-free (cf) DNA in the plasma of pregnant women has become part of routine prenatal care with limited professional guidelines and without significant input from the Turner syndrome community. In contrast to the very high positive predictive values (PPVs) achieved with cfDNA analysis for trisomy 21 (91% for high-risk and 82% for low-risk cases), the PPVs for monosomy X are much lower (~26%). This is because the maternal plasma sample contains both maternal cfDNA and placental DNA, which is a proxy for the fetal genome. Underlying biological mechanisms for false positive monosomy X screening results include confined placental mosaicism, co-twin demise, and maternal mosaicism. Somatic loss of a single X chromosome in the mother is a natural phenomenon that occurs with aging; this could explain many of the false positive cfDNA results. There is also increased awareness of women who have constitutional mosaicism for 45, X who are fertile. It is important to recognize that a positive cfDNA screen for 45, X does not mean that the fetus has Turner syndrome. A follow-up diagnostic test, either amniocentesis or neonatal karyotype/chromosome microarray, is recommended. Research studies on cell-free mRNA in second trimester amniotic fluid, which is almost exclusively fetal, demonstrate consistent dysregulation of genes involved in the hematologic, immune, and neurologic systems. This suggests that some of the pathophysiology of Turner syndrome occurs early in fetal life and presents novel opportunities for consideration of antenatal treatments.

1 |. INTRODUCTION

In some parts of the world, prenatal genomic screening for Turner syndrome has quietly become incorporated into routine antenatal care with limited guidance from professional societies (Mennuti, Chandrasekaran, Khalek, & Dugoff, 2015) and without significant input from Turner syndrome support groups. In fact, a joint position article from the American and European Societies of Human Genetics recommended against routine screening for sex chromosome aneuploidies (SCAs; Dondorp et al., 2015; Howard-Bath, Poulton, Halliday, & Hui, 2018). How did screening for SCAs become routine or “optional”? Prior to 2013, a fetal diagnosis of Turner syndrome was primarily suggested by abnormal sonographic findings, including an increased nuchal translucency measurement in the first trimester, or cystic hygroma, nonimmune hydrops, cardiovascular and kidney anomalies, and short long bones in the second trimester (Liau et al., 2014). In one large French cytogenetics database, 84% of 975 cases diagnosed with Turner syndrome were initially detected due to the presence of fetal sonographic abnormalities (Gruchy et al., 2014).

In 2011, however, sequencing of cell-free (cf) DNA that circulates in maternal plasma during pregnancy became clinically and commercially available. Initially, the testing was provided as an alternate to serum biochemical analysis and nuchal translucency screening in pregnant women at high-risk for trisomy 21. By 2012, clinical trials began to additionally report relatively high sensitivity and specificity screening results for the other common autosomal aneuploidies (trisomies 13 and 18), as well as monosomy X (Bianchi et al., 2012). In parallel, the companies that offered testing recognized that pregnant women were very interested in learning the sex of their fetus, and that this could easily be added to the test menu at no extra charge. A consequence of this decision was the ability to screen for the SCAs, 45, X, 47, XXX, 47, XXY, and 47, XYY (Nicolaides, Musci, Struble, Syngelaki, & del Mar Gil, 2014; Samango-Sprouse et al., 2013). As of late 2012, cfDNA screening for monosomy X was first offered clinically to pregnant women whose fetuses had cystic hygromas detected via routine sonographic examinations (Bianchi et al., 2013).

2 |. CELL-FREE DNA SEQUENCING IN THE PLASMA OF PREGNANT WOMEN

To understand the benefits and limitations of using sequencing of maternal plasma cfDNA to detect fetal chromosome aneuploidies, it is helpful to briefly review both the biology and technology underlying the test. “Fetal” DNA is not sequenced. Instead, apoptotic DNA originating in the placenta serves as a proxy for the fetus (Taglauer, Wilkins-Haug, & Bianchi, 2014). Within the maternal plasma sample, maternal and placental DNA fragments are present; both are sequenced together. For routine aneuploidy screening, only DNA from the Y chromosome is uniquely fetal. In both the random (whole genome) and targeted approaches, the circulating fragmented DNA molecules are either sequenced or amplified, then aligned and mapped to specific human chromosomes (Bianchi & Chiu, 2018). The percentage of fragments that map to a chromosome of interest is determined and compared to a reference euploid sample. For the sex chromosomes, the reference DNA is from a 46, XX female. Euploid males have a reduced number of DNA fragments that map to the X chromosome, compared to the normalized chromosome values that are expected for females. In addition, in a male fetus, sequences from the Y chromosome are detected. If no Y chromosome DNA fragments are detected, and the number of X chromosome fragments is significantly lower or higher than the reference values, 45, X or 47, XXX, respectively, would be suspected (Bianchi et al., 2012; Bianchi et al., 2014). Occasionally, Y chromosome DNA is detected in the maternal plasma, and the fetus appears to have female genitalia on sonographic examination. The underlying mechanisms for this include a twin demise, a maternal disorder of sexual differentiation, such as Swyer syndrome, or that the mother has undergone a bone marrow or solid organ transplant from a male donor (Bianchi, 2018; Hartwig, Ambye, Sorensen, & Jorgensen, 2017).

The United Kingdom National Screening Committee published a meta-analysis of the performance of antenatal cfDNA sequencing in pregnancies at both high and low-risk for the common autosomal aneuploidies (Taylor-Phillips et al., 2016). Within 5 years of the test’s introduction into clinical care, the positive predictive values (PPVs) in high-risk gestations were 91% (trisomy 21), 84% (trisomy 18), and 87% (trisomy 13). Due to the lower prevalence of fetal aneuploidy in a low-risk maternal population, the PPVs for trisomies 21, 18, and 13 were 82%, 37%, and 49%, respectively (Taylor-Phillips et al., 2016). In pregnant women at low-risk for fetal autosomal aneuploidies, studies that directly compared cfDNA sequencing to serum biochemical and nuchal translucency measurements in the same individual showed that the PPVs were significantly higher and the false positive rates were significantly lower with DNA analysis (Bianchi et al., 2014; Norton et al., 2015; Zhang et al., 2015). Regardless of the PPVs, it is universally recommended that a positive cfDNA aneuploidy screen should be confirmed by a diagnostic procedure such as amniocentesis, chorionic villus sampling (CVS), or neonatal karyotype or chromosome microarray (CMA).

3 |. CELL-FREE DNA SCREENING AND MONOSOMY X

Given the high PPVs achieved with cfDNA screening for the common autosomal aneuploidies, it was expected that the test performance for SCAs would be equally good. As of 2013, reports began to surface of false positive test results (Mennuti, Cherry, Morrissette, & Dugoff, 2013). To address the question of accuracy, in collaboration with a commercial clinical laboratory, we undertook an analysis of the performance of noninvasive prenatal testing (NIPT) for fetal SCAs (Bianchi et al., 2015). Utilizing a database of 18, 161 samples with sex chromosome results, 204 (1.1%) of samples were identified with one of four SCAs (monosomy X, XXX, XXY, or XYY). Clinical outcomes were obtained to the extent possible. Cases were coded as “concordant” if the NIPT result matched the fetal or neonatal karyotypes or the birth outcomes. Similarly, the cases were coded as “discordant” if they did not match. “No discordance” was reported if no follow-up information was available to the testing laboratory. The analysis showed that 198 of 204 (91%) of the SCA cases identified involved the X chromosome, and that 47 of 48 known discordant cases involved the X. Monosomy X was the most common SCA identified, accounting for 148 (72.5%) out of the total of 204 test results. Outcome information was only available on 44 of 148 cases; 35 of 44 cases were discordant, meaning that the fetus had a karyotype or CMA result of 46, XX (Table 1).

TABLE 1.

Positive predictive values for 45, X in clinical studies with outcome data

Study	Total samples	45, X results	Outcome data	True positives	False positives	Positive predictive values
Yao et al. (2014)	5,950	7	5	2	3	40%
Bianchi et al. (2015)	18,161	148	44	9	35	20%
Reiss, Discenza, Foster, Dobson, and Wilkins-Haug (2017)	2,851	11	11	1	10	9%
Zhang et al. (2017) (same data in Yu et al. (2017))	10, 275	17	17	5	12	29.4%
Petersen et al. (2017)	712	89	89	24	65	27%
Suo et al. (2018)	8, 384	37	37	12	25	32.4%
Totals	46, 333	309	203	53	150	26.1%

Over time, it has become apparent that with multiple data sets reporting on outcome information for SCAs, the PPVs for cfDNA screening for monosomy X are consistently lower than for the autosomal aneuploidies (Table 1). On average, the PPVs are around 26%. There are two important take away messages from this finding: (a) a positive cfDNA screening result for monosomy X does not mean that the fetus has Turner syndrome; and (b) the test is not inaccurate. There are multiple, mainly biological, underlying reasons for a false positive result.

4 |. MECHANISMS FOR FALSE POSITIVE NIPT RESULTS OF MONOSOMY X

The biological mechanisms for false positive results include confined placental mosaicism (CPM), demise of a co-twin with monosomy X, or maternal constitutional or somatic mosaicism (Bianchi et al., 2015). As with the autosomal aneuploidies, a fetal diagnostic procedure is recommended to confirm the cfDNA test result. An amniocentesis is preferred over CVS due to the high incidence of mosaicism for monosomy X in chorionic villi. This was demonstrated by an Italian study that examined cytogenetic results from 52,673 women who had both amniocentesis and CVS performed during the same pregnancy. Monosomy X was associated with a very high incidence of CPM (59% of cases; Grati et al., 2015). Thus, if the NIPT results (from the apoptotic placental cells) show monosomy X and a CVS result also shows monosomy X, the mechanism is most likely CPM. The true fetal karyotype will be best confirmed by an amniocentesis.

Given that the blood sample contains maternal DNA, the mother must also be considered as a source of the false positive result. Unfortunately, in many clinics, pretest counseling does not generally focus on the possibility of detecting abnormal maternal findings, yet maternal secondary genomic results are being increasingly documented (Bianchi, 2018). In a Chinese study of 187 pregnant women whose plasma NIPT results showed a SCA, maternal white blood cells were also sequenced. One hundred twenty four of 187 women had a result consistent with X chromosome loss; in 10 cases (8.06%) it was shown that, the mother’s cells were the origin of the monosomy X sequence data (Wang et al., 2014). Thus, this group recommended that the maternal karyotype should be determined for any NIPT results involving the X chromosome. These data suggest that there are probably more women in the general population who have mosaic Turner syndrome, and that they are fertile. What we do not know, however, is whether subsequent medical management of the woman who is determined to have constitutional 45, X mosaicism should change. Some of the potential considerations are given in Table 2.

TABLE 2.

Medical management considerations for the pregnant woman who is determined to have 45, X mosaicism as a secondary genomic finding following cfDNA sequencing

• Should she undergo investigation for cardiovascular disease, including coarctation of the aorta?

• Is she at increased risk for the development of hypertension and/or pre-eclampsia?

• Should she deliver in a tertiary medical center?

• Is she at risk for secondary infertility?

• Postdelivery, should she be monitored for lipid, thyroid, and/or bone mineral density abnormalities?

In their report, Wang et al. (2014), did not describe any clinical findings in the 10 women with monosomy X mosaicism. Therefore, it is impossible to know if these women had constitutional or somatic mosaicism. A woman who is significantly shorter than her same sex parent or siblings, or who may have had infertility issues, may have constitutional mosaicism. To make cfDNA sequencing results even more difficult to interpret, the possibility of maternal age associated somatic loss of the X chromosome must be considered. In one study designed to aid in the interpretation of the detection of a low level 45, X cell line during routine cytogenetic analysis, 19, 650 lymphocytes from 655 females (neonates to 80 years old) were studied (Russell, Strike, Browne, & Jacobs, 2007). A clear relationship between chronologic aging and X chromosome monosomy was demonstrated. With these results in mind, we reanalyzed the data from our commercial database study. The mean maternal age for the discordant cases was 36.7 years compared with 31.7 years for the nondiscordant cases (p < 0.001) (Bianchi et al., 2015). The significantly older maternal ages in the discordant group would suggest that somatic mosaicism due to maternal aging explains many of the false positive results. In all likelihood, somatic mosaicism that is detected as a secondary finding of NIPT has no medical or management implications for the mother or fetus.

5 |. THE TURNER SYNDROME TRANSCRIPTOME AND NOVEL APPROACHES TO ANTENATAL THERAPY

To better understand early pathophysiology in Turner syndrome, we performed a pilot study of global gene expression analysis of cell-free RNA fragments present in the amniotic fluid supernatant of five second trimester fetuses with confirmed 45, X, compared with five 46, XX female fetuses matched for gestational age in weeks (Massingham et al., 2014). The amniocenteses were performed for routine clinical indications (such as sonographic abnormalities or advanced maternal age), and the karyotypes were determined in a CLIA-approved diagnostic laboratory. Amniotic fluid was used because it is nearly 100% fetal in origin. We have previously shown that cell-free RNA in amniotic fluid provides simultaneous information on multiple developing organ systems, and furthermore, that each fetal aneuploidy has a different, yet characteristic set of dysregulated genes in the second trimester (Zwemer & Bianchi, 2015).

The cell-free RNA was extracted, reverse transcribed, amplified, purified, labeled, and hybridized onto Human Genome U133 Plus 2.0 arrays (Massingham et al., 2014). Data were log-transformed, and individual differentially expressed probe sets were identified via paired t tests. Each pair consisted of a 45, X sample Each pair consisted of a 45, X sample and a 46, XX samples matched for gestational age. Four hundred seventy differentially expressed annotated genes were identified. Only 16 (3.4%) mapped to the X chromosome. The remainder were widely distributed across the genome. Not surprisingly, the X-inactive specific transcript gene, XIST, was significantly down-regulated (p < 0.0001). The short stature homeobox (SHOX) gene was not differentially expressed in the amniotic fluid supernatant sample set. This was unexpected because decreased expression of SHOX is postulated to be one reason for short stature in girls and women with Turner syndrome.

Ingenuity pathway analysis of the differentially expressed genes revealed that the hematologic and immune systems were more significantly represented in comparison to other fetal aneuploidies that we had previously studied using similar methodology. The overexpression of immune system transcripts suggests that autoimmune dysregulation occurring in fetal life may play a role in early pathology (such as ovarian degeneration), as well as later-onset autoimmune disease.

Manual curation of statistically significantly up-regulated genes identified three of potential clinical significance. These include NFATC3, which is associated with perivascular tissue remodeling and is activated by intermittent hypoxia. This could be an important gene in the pathophysiology of coarctation of the aorta and development of hypertension in females with Turner syndrome. Similarly, the low-density lipoprotein receptor LDLR is up-regulated in fetal life; this may be associated with the hyperlipidemia and atherosclerosis observed in women with Turner syndrome. Lastly, the overexpression of IGFBP5 in the 45, X samples may have a role in the development of short stature. IGFBP5, when over-expressed, impairs osteoblast function, increases formation of osteoclasts and increases cellular senescence. The results of the second trimester amniotic fluid transcriptome analysis are publicly available in the Gene Expression Omnibus (GEO #48521; https:www.ncbi.nlm.nih.gov/geo.). Additional manual curation of the data set may be helpful to further understand the mechanisms underlying organ and tissue development and dysfunction in Turner syndrome.

Lastly, the BioGPS Gene Expression Atlas (http://biogps.org) is also a publicly available, internet-based resource, which is useful for annotating gene and protein expression. We utilized it to identify the tissue origin of transcripts that were consistently dysregulated in all five samples from the fetuses with 45, X. Tissues that were overrepresented in 45, X, compared with euploid, trisomy 21 and trisomy 18 amniotic fluid data sets, were the hematologic, immune, and neurologic systems.

Our results also suggest that the dysregulated gene set in the Turner syndrome transcriptome can be used to identify targets for antenatal therapy, such as the prevention of potential autoimmune-based destruction of the ovaries. We are using a similar approach to identify potential therapies that could be given antenatally to a pregnant woman carrying a fetus with trisomy 21 to improve brain development and neurocognition (Guedj & Bianchi, 2014).

6 |. SUMMARY

Noninvasive prenatal screening for 45, X is already widely clinically available. In pregnant women at general risk for fetal aneuploidy, the PPVs for Turner syndrome are significantly lower than for trisomy 21 (~26%) (Table 1). All positive screening results should be confirmed with a diagnostic procedure, preferably amniocentesis. If the fetal karyotype is 46, XX, the mother or the placenta are the likely sources of the 45, X cell line.

Transcriptomic analysis of cell-free RNA in the second trimester amniotic fluid of fetuses with Turner syndrome suggests that many of the later-onset pathologies have their origins in fetal life, such as atherosclerosis, hypertension, and short stature.

The new insights that have been garnered from the large-scale clinical screening for Turner syndrome as well as the transcriptome research suggest multiple areas for future research. First of all, there has been very little input from affected women and their families. How have members of Turner syndrome support groups been personally affected by prenatal screening? What do they think of it? How does the recognition of many fertile women with mosaic Turner syndrome change pre- and post-natal counseling? As shown in Table 2, how should the management of pregnant women who are detected, via cfDNA screening, to have mosaic Turner syndrome, change? Finally, what opportunities exist for antenatal treatment of fetuses diagnosed with 45, X? It is our laboratory’s hope that extended knowledge of the fetal transcriptome will identify dysregulated pathways that can serve as targets to develop earlier, novel therapies.

Biography

Diana W. Bianchi, MD, is the Director of the Eunice Kennedy Shriver National Institute of Child Health and Human Development at the National Institutes of Health. She is a medical geneticist with research interests in prenatal genomics and fetal therapy. Her laboratory has published on the detection of full and mosaic Turner syndrome in both the mother and baby via noninvasive prenatal testing, as well as gene expression in living fetuses with Turner syndrome.