Chromosomal mosaicism detected by karyotyping and chromosomal microarray analysis in prenatal diagnosis

Abstract To investigate the incidence and clinical significance of chromosomal mosaicism (CM) in prenatal diagnosis by G‐banding karyotyping and chromosomal microarray analysis (CMA). This is a single‐centre retrospective study of invasive prenatal diagnosis for CM. From 5758 karyotyping results and 6066 CMA results, 104 foetal cases with CM were selected and analysed further. In total, 50% (52/104) of foetal cases with CM were affected by ultrasound‐detectable phenotypes. Regardless of whether they were singleton or twin pregnancies, isolated structural defects in one system (51.35%, 19/37 in singletons; 86.67%, 13/15 in twins) and a single soft marker (18.92%, 7/37 in singletons; 13.33%, 2/15 in twins) were the most common ultrasound anomalies. Mosaic autosomal trisomy (19.23%, 20/104) was the most frequent type, and its rate was higher in phenotypic foetuses (28.85%, 15/52) than in non‐phenotypic foetuses (9.62%, 5/52). There was no difference in mosaic fractions between phenotypic and non‐phenotypic foetuses based on specimen sources or overall classification. Discordant mosaic results were observed in 16 cases (15.38%, 16/104) from different specimens or different testing methods. Genetic counselling and clinical management regarding CM in prenatal diagnosis remain challenging due to the variable phenotypes and unclear significance. Greater caution should be used in prenatal counselling, and more comprehensive assays involving serial ultrasound examinations, different specimens or testing methods verifications and follow‐up should be applied.

time, stage and location of these mechanisms occurring during embryonic development. 3 In clinical practice, the phenotypic effects of CM are generally considered to be highly associated with this mosaic distribution pattern.
Due to the variable and unpredictable distribution patterns of abnormal cell lineages, CM is a challenging issue in prenatal diagnosis. In particular, low-level mosaicism (<15%-20%), foetal-placental discrepancies and the UPD generated by embryo rescue have been reported to be responsible for the increased risk of erroneous diagnoses. 4,5 Recently, improvements have been made in the field of molecular genetics technologies such as quantitative fluorescent polymerase chain reaction (QF-PCR), fluorescence in situ hybridization analysis (FISH), CMA and next-generation sequencing so that supplementary analyses on amniotic fluid or cord blood are often performed to confirm the true foetal involvement and its clinical significance. [6][7][8][9][10] Nevertheless, even CM thought to be confined to the placenta may reflect a cryptic foetal mosaicism that may or may not give rise to phenotypic consequences, or lead to placental dysfunction related to foetal growth restriction. 11,12 Thus, based on the unclear clinical significance of CM and limited data about CM in prenatal diagnosis, it is necessary to understand more comprehensively the potential correlation between CM and its phenotypic effects. In this study, we explored the incidence and characteristics of CM detected by G-banding karyotyping and/or CMA in 104 foetal cases, analysing their phenotypic features, and further comparing the discordant CM results identified from different specimens or different testing technologies.

| Case selection
We retrospectively reviewed all pregnant women who underwent Pre-test counselling was given to all parents, including informing them about the benefits and disadvantages of invasive prenatal diagnosis, the testing process, possible results and limitations of the karyotyping and CMA. Written informed consent to participate and receive invasive prenatal diagnoses was provided by all parents. This study was approved by the Research Ethics Committee of Nanfang hospital and was conducted according to the Declaration of Helsinki principles.
F I G U R E 1 Flow diagram of the chromosomal results in 104 foetal cases with mosaicism. A total of 104 foetal cases were recruited, including 85 singleton pregnancies and 19 pairs of twin pregnancies. Among, forty-three cases only underwent G-banding karyotyping, 33 cases only underwent CMA, and 28 cases underwent both karyotyping and CMA. In twin pregnancies, only one foetus was affected by ultrasound anomalies in each pair. In addition, the chromosomal results of the affected foetuses were also reported. Other mosaic abnormal karyotypes refer to abnormal karyotypes detected by G-banding karyotyping such as isochromosomes and marker chromosomes, not including aneuploidies. CMA, chromosomal microarray analysis; US, ultrasound; No. of, the number of; T, Trisomy; CNVs, copy number variants Indications for invasive prenatal diagnosis included advanced maternal age, high risk detected by maternal serum screening and non-invasive prenatal testing, an adverse pregnancy or family history, parental thalassaemia or chromosomal abnormalities, and abnormal ultrasound findings ( Figure 2).
Gestational age was determined based on the last menstrual period and an ultrasonogram conducted at 11-13 +6 weeks. Foetal growth and anatomy were confirmed by performing an ultrasound examination. Abnormal ultrasound findings included structural defects and soft markers in singleton or twin pregnancies, and foetal growth restriction (FGR) in singleton pregnancies. Soft markers included the absence of a nasal bone, a thickened nuchal translucency or neck fold, ventriculomegaly, echogenic bowel and a shortened long bone. 13 FGR in a singleton pregnancy is diagnosed when the estimated foetal weight (EFW) is less than the 10th percentile for the gestational age. 14 The twin pregnancies were classified as monochorionic diamniotic (MCDA) or dichorionic diamniotic (DCDA) according to the ultrasound images at 11-13 +6 weeks. 15 In addition, we also reviewed 80 miscarriage cases with CM detecting by CMA to obtain more information about the potential correlation between CM and its phenotypic effects ( Figure S1).

F I G U R E 2
Indications of invasive prenatal diagnosis in foetuses with mosaicism. The sum of the number of cases for different indications exceeds the total number of whole cases (104 cases) because more than one indication may be identified in each case. NIPT, non-invasive prenatal testing; IUFD, intrauterine foetal death; FGR, foetal growth restriction, n, the number of cases. Parental chromosomal abnormalities include maternal and/or paternal chromosomal abnormalities such as balanced translocation. Ultrasound anomalies include foetal structural defects, soft markers and FGR

| Invasive prenatal diagnosis by karyotyping and CMA
Samples from the chorionic villus, amniotic fluid or cord blood of the foetuses were collected by ultrasound-guided transabdominal chorionic villus sampling, amniocentesis or cordocentesis. Cells were cultured and prepared for G-banding karyotyping using standard protocols. Genomic DNA was extracted from the uncultured samples, and then CMA was carried out by using a whole genome-wide Affymetrix CytoScan HD array (Affymetrix Inc, Santa Clara, CA, USA) according to the manufacturer's operating procedures. The data were ana-

| Statistical analysis
Statistical analyses were performed with SPSS software v22.0 (SPSS Inc, Chicago, IL, USA). The chi-square test and two independent sample t test were applied to analyse the statistical data. Differences were considered statistically significant when P < 0.05.

| Clinical characteristics
A total of 5758 G-banding karyotyping results and 6066 CMA results from invasive prenatal diagnosis were reviewed, of which 104 foetal cases were identified with CM. Among them, 43 cases underwent karyotyping, 33 cases underwent CMA and 28 cases underwent both karyotyping and CMA ( Figure 1). The indications for invasive prenatal diagnosis of the 104 prenatal cases with CM are listed in Figure 2. Ultrasound anomalies were found in 50% (52/104) cases, including 37 singleton pregnancies and 15 twin pregnancies.
Advanced maternal age was the most common indication for patients with CM (35.58%, 37/104) after excluding ultrasound anomalies.
The clinical characteristics of the patients were further investigated based on cases complicated with or without ultrasound anomalies as shown in Table 1. The mean maternal ages were 30.38 ± 4.86 and 33.63 ± 5.03 years in patients with and without ultrasound anomalies, respectively. The number of patients older than 35 years was higher in cases without ultrasound anomalies (46.15%, 24/52) than those in cases with ultrasound anomalies (23.08%, 12/52). The mean gestational ages for undergoing invasive prenatal diagnosis were 19.48 ± 4.32 and 22.83 ± 6.05 weeks in patients without and with ultrasound anomalies, respectively.

| Ultrasound anomalies
Fifty-two foetal cases were affected by ultrasound anomalies among 104 cases, including 37 singleton pregnancies and 15 twin pregnancies. In the twin pregnancies, only one foetus was affected in each pair. The detailed ultrasound anomalies are shown in Table 2.
In singleton pregnancies, FGR was identified in 5 foetuses (ratio = 13.51%, 5/37), structural defects were found in 25 foetuses (ratio = 67.57%, 25/37), and soft markers were found in 9 foetuses (ratio = 24.32%, 9/37). Among these, most of the structural defects and soft markers were isolated defects in one system (ratio = 51.35%,   Note: The sum of different ultrasound abnormalities exceeds the total number of singleton or twin pregnancies due to more than one ultrasound anomaly detected in one case. US anomalies* include foetal structural defects, soft markers and FGR. (% a ): the percentages listed in Table 3 were used to calculate the detailed ratio of different types of ultrasound anomalies in cases with ultrasound phenotypes, but not used to calculate the detection rates of ultrasound anomalies in all 104 cases. other b refer to other structural defects such as isomerism syndrome, excluding structural defects listed in Table 3.  Table S1 and Table S2.

| Chromosomal mosaicism
To explore the correlation of CM and its phenotypic effects, we further compared the differences of mosaic types and mosaic fractions between cases with and without ultrasound anomalies.
First, the incidence of autosomal trisomy in cases with ultrasound anomalies (28.85%, 15/52) was much higher than that in cases without (9.62%, 5/52) (P = 0.023 < 0.05) (Figure 3). In addition, 89.04% (65/73) of miscarriage cases were diagnosed as mosaic autosomal trisomy, excluding one case with 48, XXY, +9 ( Figure S1). The highest detection rate of mosaic autosomal trisomy was found in miscarriage cases, followed by foetuses with ultrasound anomalies, and finally cases without phenotypes (P = 0.000 < 0.05). In cases of mosaic unbalanced structural abnormalities and other chromosomal mosaicisms, no significant differences were identified between foetuses with and without ultrasound phenotypes (P > 0.05) ( Figure 3). Second, in mosaic fractions, there were also no differences detected in the mosaic fractions whether it was based on the specimen sources or the overall classification (Table S3). We then investigated mosaic fractions of aneuploidies and pathogenic CNVs both detected in foetuses with and without phenotypes as shown in Table S4. The most common mosaic aneuploidy in the two groups was 45, X/46, XN, while no difference was found in its mosaic fraction (P = 0.090 > 0.05). Similar results were also detected in other aneuploidies and for pathogenic CNVs (Table S4).

| Discordant chromosomal mosaicism
In our cohort, discordant mosaic results were observed in 16   Ultrasound anomalies in foetuses with CM seem variable, and no evidence was found that a CM may account for some ultrasound anomalies. Approximately 50% of our CM carriers had ultrasound anomalies, of which isolated structural defects in one system and single soft markers were the most common findings in both singleton and twin pregnancies. In a study with 100 CVS samples, 5 mosaic chromosomal aberrations were detected, of which two cases both showed an isolated but complex structural defect in one system, including a hypoplastic right ventricle of the heart in one foetus and corpus callosum and hypoplasia of the cerebellar vermis in another foetus. 18 This suggested that CM foetuses with phenotypes are probably not associated with multiple structural defects or soft markers, but instead are often detected as isolated ultrasound

| D ISCUSS I ON
anomalies.
In addition, 5 foetuses with CM were diagnosed as FGR and their samples were all from cord blood, excluding the possibility of CPM.
A previous study suggested that mosaicisms from CVS should be further verified for TFM by AF or CB due to the presence of fetoplacental discrepancies. 19 However, although most pregnancies diagnosed with CPM are deemed to have good postnatal outcomes but an impaired placenta may provide insufficient support for the pregnancy, leading to foetal complications such as FGR or other adverse outcomes. 20 during pregnancy is important to prevent adverse complications for foetuses with TFM and CPM.
The incidences of CM were compared between foetuses with and without phenotypes. Among these, the incidence of mosaic autosomal trisomy was higher in phenotypic foetuses than that in no-phenotypic foetuses. Interestingly, gradually increased detection rates of mosaic autosomal trisomy were found from data in foetuses without ultrasound phenotypes to cases with phenotypes and to miscarriage samples. We speculated that most of the mosaic autosomal trisomy induced by meiotic or mitotic non-disjunction error remains confined to the placenta because they are almost universally lethal or pathogenic for foetuses. For this reason, the incidence of mosaic autosomal trisomy in miscarriage samples was the highest among all prenatal cases. 22 Another critical factor possibly impacting the phenotypic effect of CM is mosaic fractions. In our study, even for the same CM detected in two groups, mosaic fractions displayed no difference, which is not in that different mosaic tetrasomy 12p was identified from amniotic fluid, skin, placenta and cord blood by karyotyping, FISH, QF-PCR and array-CGH. 28 Their results showed that CB and placentas were prone to a negative result when compared with AF, and array-CGH on uncultured CB or FISH on cultured CB may be better to use for prenatal diagnosis. Others hold the opposite view that mosaicism in AF requires further cordocentesis for confirmation, and that the pregnancy is safe when a normal result is identified in CB. 29 These findings demonstrate that increased risk in genetic counselling is due to discordant CM from different specimens or testing methods. It is highly recommended to use more comprehensive assays such as a combination of CMA, FISH and karyotyping to detect mosaicism in AF and CB before any irreversible decision is made in regard to the pregnancy.
There are a few drawbacks in our study. The first is our uncompleted long-term follow-up may lead to information loss for postnatal phenotypes. The second is the presence of selection bias in this retrospective analysis. The incidence or detection rate of CM may be higher in our data than in the general population. The third is the confirmation in AF or CB was performed in only a small proportion of cases from CVS, directly causing failure to distinguish CPM or TFM in the rest of the cases.
In summary, the incidence and clinical characteristics of CM were investigated in foetuses with and without ultrasound phenotypes by karyotyping and CMA, providing valuable information for genetic counselling and management of prenatal mosaic cases.
Greater caution should be used in prenatal counselling, and more comprehensive assays involving serial ultrasound examinations, different specimens or testing methods verifications and effective follow-up should be applied.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets analysed in this study are available from the corresponding author on reasonable request.