Clinical application of chromosomal microarray analysis for fetuses with craniofacial malformations

Background The potential correlations between chromosomal abnormalities and craniofacial malformations (CFMs) remain a challenge in prenatal diagnosis. This study aimed to evaluate 118 fetuses with CFMs by applying chromosomal microarray analysis (CMA) and G-banded chromosome analysis. Results Of the 118 cases in this study, 39.8% were isolated CFMs (47/118) whereas 60.2% were non-isolated CFMs (71/118). The detection rate of chromosomal abnormalities in non-isolated CFM fetuses was significantly higher than that in isolated CFM fetuses (26/71 vs. 7/47, p = 0.01). Compared to the 16 fetuses (16/104; 15.4%) with pathogenic chromosomal abnormalities detected by karyotype analysis, CMA identified a total of 33 fetuses (33/118; 28.0%) with clinically significant findings. These 33 fetuses included cases with aneuploidy abnormalities (14/118; 11.9%), microdeletion/microduplication syndromes (9/118; 7.6%), and other pathogenic copy number variations (CNVs) only (10/118; 8.5%).We further explored the CNV/phenotype correlation and found a series of clear or suspected dosage-sensitive CFM genes including TBX1, MAPK1, PCYT1A, DLG1, LHX1, SHH, SF3B4, FOXC1, ZIC2, CREBBP, SNRPB, and CSNK2A1. Conclusion These findings enrich our understanding of the potential causative CNVs and genes in CFMs. Identification of the genetic basis of CFMs contributes to our understanding of their pathogenesis and allows detailed genetic counselling.


Background
Craniofacial malformations (CFMs) are among the most common congenital birth malformations in humans, with orofacial clefts accounting for approximately 13% of congenital malformations in all live births [1]. The majority of CFMs occur in fetuses without a family history. Thus, it is particularly important to evaluate the integrity of the craniofacial structures of fetuses using ultrasonographic screening. Generally, fetal CFMs include defects such as cranial malformations, ocular malformations, nasal dysplasia, and orofacial defects, among others. These defects may be isolated variations or may occur in combination with other congenital structural abnormalities such as central nervous system abnormalities, cardiac defects, abdominal wall defects, skeletal defects, and so on. Due to the similar phenotypes of craniofacial syndromes with and without multiple organ involvement, prenatal genetic counseling can be challenging. Although the causes of CFMs are currently unclear, genetic analysis can help to provide a genetic basis for prenatal diagnosis and can also contribute to our understanding of the pathogenesis of CFMs.
Conventional karyotyping is the classic method for detecting aneuploidy or chromosomal rearrangements [2,3]. However, this approach has gradually been replaced by chromosomal microarray analysis (CMA) due to its low resolution and low detection efficiency. Recently, CMA has been recommended for prenatal diagnosis when fetal abnormalities are detected by ultrasound [4]. Copy number variations (CNVs) in simple cranial or facial malformations have also been reported [5,6]; however, the comprehensive CMA assessment of fetuses with CFMs is limited. In the present study, we reviewed the clinical and molecular findings of 118 fetuses with CFMs to explore the clinical significance of CNVs in each case. This study aimed to provide useful information for prenatal diagnosis of CFMs and related genetic counseling.
All 33 fetuses identified to have P/LP CNVs were terminated in pregnancy or died in utero. Additionally, we found that 9 (9/118; 7.6%) fetuses that had variant of unknown significance CNVs. The remaining 76 cases (76/118; 64.4%) had no CMA abnormalities or only likely benign/benign CNVs.

Comparison of chromosomal abnormality detection rates
Overall, as shown in Table 4, the detection rate of chromosomal abnormalities in non-isolated CFM fetuses was significantly higher than in isolated CFM fetuses (26/71 vs. 7/47; p = 0.01). However, there was no significant difference in the chromosomal abnormality detection rate of complex CFMs compared to simple CFMs (6/17 vs. 27/101; p > 0.05).

Discussion
Craniofacial anomalies are common in postnatal cases, especially in patients with nervous system disorders. The detection rate of prenatal cranial abnormalities is unclear; however, there is a high incidence of cranial abnormalities in neonates [7]. Nicolaides et al. (1993) reported a 7% incidence of facial defects in fetal malformations [8], of which orofacial clefts were the most common, with a prevalence of approximately 1 case per 700 deliveries [9]. In this study, it was observed that cranial abnormalities and orofacial clefts were most common, accounting for 40.6 and 43.6% of cases, respectively. This finding is consistent with the above reports.  CMA offers obvious advantages in improving the detection rate and identifying the pathogenicity of CNVs. Recently, there have also been several successive research publications on prenatal diagnosis of fetal cleft lip/palate and cranial anomalies with the use of CMA [5,6]. In our study, cytogenetic karyotyping revealed abnormal karyotypes in 15.4% of fetuses and the detection rate increased by 12.6% with CMA. The incidence of chromosomal aberrations and CNVs was significantly higher than the 4.6 and 6.3%, respectively, reported in a recent study of fetal structural abnormalities [10]. The reason for the discrepancies between previous reports [10,11] and our study is that we did not merely focus on isolated CFMs or cases with simple CFMs.  In this study, we attempted to determine the potential correlation between CNVs and CFMs in fetuses using CMA. This study identified 9 cases (cases [15][16][17][18][19][20][21][22][23] with CNVs relating to known chromosomal MMSs, of which 22q11 deletion/duplication syndrome was the most common, with an overall prevalence of 3.4% (4/118; 2 proximal deletion, 1 proximal duplication, and 1 distal deletion). 22q11 proximal deletion, also known as DiGeorge syndrome or velocardiofacial syndrome, involves more than 30 Mendelian genes; potential genes such as TBX1, COMT, UFD1L, GNB1L, TRXR2, MED15, and RANBP1 were researched to explore the phenotype/ CNV correlation. Cleft palate is among the most common problems in patients with this microdeletion, while simple cleft lip is only occasionally found [12]. According to previous reports, TBX1 is considered to be responsible for cleft lip/palate phenotypes in both 22q11 deletion and 22q11 duplication [13]. Fetus 17, with heart abnormalities and cleft lip, was found to carry 22q11 distal deletion syndrome. Heart problems are a usual finding, but cleft lip only, without cleft palate, has never been reported within the clinical spectrum of this syndrome. This suggests that simple cleft lip may need to be included in the phenotype spectrum of 22q11 distal deletion syndrome. Although researchers such as Spineli-Silva argue that the cause of CHDs and craniofacial anomalies in patients with distal 22q11 deletion may be haploinsufficient MAPK1 expression [14], the underlying mechanisms are still largely unknown.
In addition, we identified five distinct CNVs associated with rare MMSs, including two microdeletion syndromes (3q29 and 17q12) and three microduplication syndromes (7q11.23, 8p23.1 and 16p11.2). These syndromes are associated with a range of mental and physical disabilities as well as craniofacial abnormalities. We screened several candidate genes located in these regions that are involved in craniofacial development, such as PCYT1A (3q29), [15], DLG1 (3q29), [16]), and LHX1 (17q12), [17,18]. Although TBX6 is considered to be a key gene resulting in several major phenotypes in 16p11.2 duplication, potential genes associated with orofacial cleft in this region still require further exploration. Additionally, there is no reported correlation between 7q11.23 duplication and skull defects resulting from anencephaly, but this fragment has been confirmed as a pathogenic CNV of central nervous system development.
Other rare CNVs detected in the present study are also believed to contribute to the pathogenesis of CFMs. ZIC2 in 13q23.3 (cases 23-25) has been identified as a key gene associated with several major CFMs resulting from holoprosencephaly [19]. Deletion in the chromosome 7q34q36.3-encompassing gene SHH was identified in cases 26 and 27; SHH is involved in the organization and morphology of the developing embryo and is known to be a key gene in craniofacial abnormalities such as microcephaly, hypotelorism, midface hypoplasia, and cleft lip/palate [20]. In case 33 with isolated micrognathia, a 0.44 Mb deletion in region 1q21.1 was identified; haploinsufficiency of gene SF3B4 in 1q21.1 has been confirmed to be associated with micrognathia [21]. Additionally, there is also evidence of the pathogenicity of haploinsufficient FOXC1 expression. Heterozygous deletion of FOXC1 in 6p25.3 (cases 28 and 32) can lead to Axenfeld-Rieger syndrome (6p25 deletion syndrome); ocular hypertelorism and flat midface are prevalent in affected postnatal cases [22].
Case 31, which exhibited a maxillary protrusion, midface hypoplasia, and a flat nose, had a 12.49 Mb deletion in chromosome 11q24.1q25 and a 12.88 Mb duplication in chromosome 16p13.3p13.12 involving the gene CREBBP. There are several literature reports suggesting that duplication of the 16p13.3 region containing the CREBBP gene results in distinct similar facial dysmorphism [23], but, to date, no cases with duplication only encompassing the CREBBP gene have been reported. Case 29 had microcephaly < 2 SD and had a 21.21 Mb mosaic deletion in chromosome 20p13p11.21. Among 141 protein coding genes within this deletion region, mutations only in SNRPB and CSNK2A1 have been reported to be associated with autosomal dominant microcephaly [24,25]. However, to date, there is no evidence supporting their pathogenicity in haploinsufficiency. In case 30, we could not identify a gene specifically associated with the observed cleft palate; we only identified an autosomal recessive gene, HPGD, associated with a higharched palate and without dose-sensitive reports [26]. We suspect a single mutation on the other chromosome may explain the observed phenotype. In the future, the function of candidate genes within the identified CNVs should be further investigated.

Conclusion
The current findings enrich our understanding of the potential causative CNVs and genes in CFMs. We detected several CNVs, including nine MMS regions associated with CFMs, and found a series of clear or suspected dosage-sensitive CFM genes including TBX1, MAPK1, PCYT1A, DLG1, LHX1, SHH, SF3B4, FOXC1, ZIC2, CREBBP, SNRPB, and CSNK2A1.

Study subjects
The present study was approved by the institutional research ethics committee of our unit. All parents agreed to participate in the study and provided written informed consent. We retrospectively analyzed a cohort of 118 fetuses with CFMs that were diagnosed at the Wenzhou Prenatal Diagnosis Center between 2012 and 2019. All the pregnant women underwent prenatal ultrasound examination or magnetic resonance imaging performed by experienced maternal fetal medicine specialists and ultrasound technicians. The pregnant women ranged in age from 21 to 43 years, with their gestational week ranging between 13 to 32 weeks. The eligibility conditions for inclusion in this study were isolated CFMs and non-isolated CFMs (with other structural abnormalities or sonographic soft markers). According to the International Society of Ultrasound in Obstetrics and Gynecology (ISUOG) guidelines [27] and the recent detailed ultrasonographic report of prenatal CFMs [28], the CFMs included cranial malformations (such as abnormal size, shape, and integrity of the skull) and various facial structure abnormalities in coronal, transverse, and sagittal planes (such as abnormal number, size, shape, mass, or location of orbits, lens, palpebral fissure, mandible, maxilla, forehead, nose/nostrils, lips, tongue, soft palate, or ears). Of note, nasal bone absence or dysplasia as sonographic soft markers were excluded from the facial malformations. Fetal samples were obtained from aborted fetuses (14 cases) or were collected by amniocentesis (40 cases) or cordocentesis (64 cases) based on the gestational week at the time of invasive prenatal testing.

Karyotype analysis
A total of 104 fetal samples (14 fetal tissues from abortions were excluded) were analyzed using standard Gbanded karyotyping at 320-450 bands resolution to diagnose overall chromosomal abnormalities. At least 20 metaphase cells from each sample were carefully examined by an experienced technician to detect chromosomal structural abnormalities and numerical abnormalities.
According to the American College of Medical Genetics Standards and Guidelines, the CNVs were classified as pathogenic, likely pathogenic, benign (B), likely benign (LB), or variant of unknown significance (VOUS) [29,30]. Given the reliability of the minimum detection range, the reporting threshold for P or LP CNVs was 100 Kb; LB/B CNVs and VOUS CNVs smaller than 500 Kb deletion or 1000 Kb duplication were not reported, consistent with the Canadian College of Medical Geneticists (CCMG)-Society of Obstetricians and Gynaecologists of Canada (SOGC) recommendations [31]. All reported CNVs were according to the National Center for Biotechnology Information human genome build 37 (hg 19). CMA or quantitative real-time polymerase chain reaction was also performed on parental DNA samples, if DNA were available, to determine whether the CNVs detected were inherited or de novo.
To identify CFM-associated CNVs and candidate genes, we further examined and analyzed the genes within identified CNVs by retrieving related literature and examining related databases.

Statistical analysis
Statistical analysis was performed with SPSS 23.0 (IBM Corporation, USA). The CMA detection rates of P/LP variants were compared between various isolated and non-isolated CFM fetuses, and simple and complex CFM fetuses; p < 0.05 was considered statistically significant.