Pathogenic/likely pathogenic copy number variations and regions of homozygosity in fetal central nervous system malformations

To explore pathogenic/likely pathogenic copy number variations (P/LP CNVs) and regions of homozygosity (ROHs) in fetal central nervous system (CNS) malformations. A cohort of 539 fetuses with CNS malformations diagnosed by ultrasound/MRI was retrospectively analyzed between January 2016 and December 2019. All fetuses were analyzed by chromosomal microarray analysis (CMA). Three cases with ROHs detected by CMA were subjected to whole-exome sequencing (WES). The fetuses were divided into two groups according to whether they had other structural abnormalities. The CNS phenotypes of the two groups were further classified as simple (one type) or complicated (≥ 2 types). (1) A total of 35 cases with P/LP CNVs were found. The incidence of P/LP CNVs was higher in the extra-CNS group [18.00% (9/50)] than in the isolated group [5.32% (26/489)] (P < 0.01), while there was no significant difference between the simpletype and complicated-type groups. (2) In the simple-type group, the three most common P/LP CNV phenotypes were holoprosencephaly, Dandy–Walker syndrome, and exencephaly. There were no P/LP CNVs associated with anencephaly, microcephaly, arachnoid cysts, ependymal cysts, or intracranial hemorrhage. (3) Only four cases with ROHs were found, and there were no cases of uniparental disomy or autosomal diseases. The P/LP CNV detection rates varied significantly among the different phenotypes of CNS malformations, although simple CNS abnormalities may also be associated with genetic abnormalities.


Introduction
Fetal central nervous system (CNS) malformations are a common congenital anomaly, with an incidence of 0.14-0.16% among live births and 3-6% among stillbirths [1]. CNS malformations can lead to intellectual, motor and behavioral disorders and even death among children, causing distress to and placing a heavy burden on children, families and society [2]. CNS malformation is different from other system deformities. The fetal brain develops continuously throughout pregnancy, and the relationship between brain structure and function is uncertain. Children with CNS malformations who are carried to term and delivered can require long-term follow-up, even until school age. Therefore, fetal CNS malformations are the most challenging conditions addressed by prenatal diagnostics and counseling.
Recent data indicate that specific CNS anomalies relate to a series of chromosomal defects, such as trisomy 18, which is the most common aberration in the acrania/exencephaly/ anencephaly sequence [3]. The prognosis of patients with CNS abnormalities with definite chromosomal abnormalities or genetic syndromes is poor [4][5][6][7][8]. Chromosome microarray analysis (CMA) can detect variations in whole-genome chromosomal imbalances, and its resolution is higher than that of conventional chromosome karyotype analysis, 1 3 especially single-nucleotide polymorphism array (SNP array) technology, which can detect regions of homozygosity (ROHs). Additionally, due to the importance of timeliness in the analysis of prenatal biological samples, direct CMA detection is recommended in certain populations, such as those with structural abnormalities detected on ultrasound [9][10][11]. With the application of second-generation sequencing, the diagnostic yield of fetal CNS gene abnormalities has increased [11][12][13][14][15]. Although whole-genome sequencing (WGS) can identify aneuploidy, CNVs, single-nucleotide variants (SNVs/indels), and small CNVs [16], the medical expenses of the identified patients are increased significantly. Therefore, an appropriate prenatal diagnosis program is needed, but it is unclear which fetal CNS diseases are most important for CMA detection.
This study aimed to evaluate the rate of pathogenic/ likely pathogenic copy number variations (P/LP CNVs) in a population of fetuses affected by CNS anomalies by using the CMA methodology, with further investigation by whole-exome sequencing (WES) for cases in which ROHs are detected by CMA.

Materials and methods
This was a retrospective study. From January 2016 to December 2019, at the Third Affiliated Hospital of Zhengzhou University, 573 fetuses were diagnosed with CNS malformations by prenatal ultrasound/MRI, and a prenatal CMA was performed. Eight experimental cases could not be analyzed due to maternal blood contamination or fetal DNA deficiency, 25 cases had aneuploidy, and one case had polyploidy; the aforementioned cases were excluded, and 539 cases were ultimately included. Three cases with ROHs detected by CMA were subjected to whole-exome sequencing (WES).
The cases were divided into two groups based on whether they had structural abnormalities outside of the CNS: the isolated group (isolated CNS structural abnormalities) and the extra-CNS group (CNS abnormalities combined with other structural abnormalities). The phenotypes of the CNS abnormalities were classified as simple or complicated according to the classification standard in the literature [17] (Fig. 1). Fetal lateral ventriculomegaly can be divided into mild (unilateral or bilateral lateral ventricle width is 10.0-12.0 mm), moderate (12.1-14.9 mm) and severe (≥ 15.0 mm). In the prenatal period, if the width of Grouping. CMA chromosomal microarray analysis, WES whole-exome sequencing *3 cases with regions of homozygosity the lateral ventricle is greater than 15 mm, it is classified as severe and is often called hydrocephalus [18].
This study was approved by the Medical Ethics Committee of the Third Affiliated Hospital of Zhengzhou University. All pregnant women and their families signed informed consent forms and received genetic counseling.

Fetal sample collection
Depending on the gestational age and clinical treatment administered, fetal samples were obtained by amniotic fluid, umbilical cord blood or tissue expelled by induced labor. Blood samples from the mothers were collected simultaneously to exclude the possibility of maternal blood contamination. The experimental results were received within 7 days.

CMA
A SNP array method was used. Exfoliated fetal cells in amniotic fluid were collected by centrifugation. Cell DNA was extracted using a QIAamp blood DNA purification kit (Qiagen, Germany). The short tandem repeat (STR) locus comparison method was used to identify maternal blood contamination. After confirming that there was no maternal blood contamination, CMA analysis was performed by an Affymatrix CytoScan 750 K chip (Thermo, USA), which contains 200,000 SNP probes and 550,000 nonpolymorphic probes. Enzymatic digestion, PCR amplification, labeling, hybridization, and chip scanning were performed according to the manufacturer's directions. The results were analyzed using public databases, including the Genome Variation Database (DGV, http:// dgv. tcag. ca/ dgv/ app/ home), the DECIPHER Database (http:// decip her. sanger. ac. uk/), cell genome array international standards (ISCA, https:// www. iscac onsor tium. org/), the Human Mendelian Genetics Online Database (OMIM, http:// www. omim. org), the University of Santa Cruz, California Genome Database (http:// genome. ucsc. edu) and the Cligen Database (https:// www. clini calge nome. org). The CNVs were divided into three categories: pathogenic, variant of unknown significance (VOUS) or benign. This study analyzed only CNVs with clinical significance (including P/LP CNVs) and homozygous regions.
For regions of homozygosity (ROHs) by CMA and for ROH ≥ 10 Mb found on only one chromosome, the possibility of uniparental disomy (UPD) should be given priority. Most chromosome UPDs have no clinical symptoms. However, UPD on chromosomes 6, 7, 11, 14, 15 and 20 can cause clinical symptoms. Therefore, when considering that the large-fragment ROHs detected on chromosomes 6, 7, 11, 14, 15 and 20 are possible cases of UPD, UPD verification should be performed. If ROH has been verified as a UPD and if UPD involves a related disease that can lead to a definite phenotype, it is reported to be a pathogenic UPD. If the possibility of UPD is excluded, it is classified as ROH with unclear clinical significance (VUS).

Whole-exome sequencing (WES)
WES was performed when homozygous regions were detected and when the fragment contained multiple protein-coding genes associated with a variety of autosomal recessive inherited diseases. First, the extracted DNA was prepared, and its quality was assessed. Then, 50 ng of DNA was enzymatically fragmented into approximately 200bp fragments for PCR amplification. The DNA fragments were hybridized and captured by NanoWES. The hybridization products were eluted, collected, amplified and purified by PCR. Next, the library was quantified by qPCR, and the size distribution was determined using an Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA). Finally, the genomic DNA of the family was sequenced using the NovaSeq6000 platform (Illumina, San Diego, USA) in the 150-bp end-to-end sequencing mode. The sequencing results were aligned with the human reference genome (hg19/grch37) using the Burrows-Wheeler Aligner tool and Picard v1.57 to delete PCR duplications (http:// picard. sourc eforge. net/). The Verita Trekker ® Variation Detection System from Berry Genomics and software GATK (https:// softw are. broad insti tute. org/ gatk/) were used for mutation detection. Variant annotation and interpretation were conducted using ANNOVAR and the Berry Genomics Enliven ® Variants Annotation Interpretation System, respectively. According to the American College of Medical Genetics and Genomics (ACMG) guidelines for the interpretation of genetic variation, these variations are divided into five categories: "pathogenic", "possible pathogenic", "uncertain significance", "possible benign" and "benign" [19][20][21][22]. In this study, a positive test was considered the detection of variants that were pathogenic or likely pathogenic. Sanger sequencing was used to verify the mutation sites of the fetus and parents. SPSS 22.0 software was used for statistical analyses. Count data are expressed as percentages. Nonnormally distributed data are expressed as the median M (min-max). The Pearson Chi-square test was used to compare the results between groups. P < 0.05 was considered statistically significant.

General information on fetal CNS malformations
This study included a total of 539 cases, with a median maternal age of 28 (17-45) years. The median gestational age at testing was 26.4  weeks. The proportions of amniocentesis performed in the isolated group and extra-CNS group were 88.5% (433/489) and 74% (37/50), respectively. No patients underwent chorionic villus sampling, and only four patients with an isolated fetal CNS anomaly underwent cordocentesis.
Fetal CNS malformations were divided into 17 simple types depending on the phenotype. The most common was lateral ventriculomegaly (62.34%), followed by hydrocephalus (4.27%), agenesis of the corpus callosum (3.15%), ependymal cyst (2.78%), cerebellar hypoplasia (2.04%), and arachnoid cyst (2.04%), and their were less than ten remaining phenotypic cases (Table 1). There were 44 cases of fetal CNS malformations combined with extra-CNS structural abnormalities, including 13 cases of multisystem structural abnormalities, 9 cases of cardiovascular system abnormalities, 8 cases of skeletal system abnormalities, 10 cases of urinary system abnormalities, and 4 cases of thoracic abnormalities.

P/LP CNVs in fetal CNS malformations
The overall CMA detection rate for P/LP CNVs was 6.49% (35/539). The P/LP CNV diagnostic yields of the isolated group and extra-CNS group were 5.32% (26/489) and 18.00% (9/50), respectively, and there was a significant difference between the two groups (χ 2 = 10.019, P = 0.002). Further comparison of the P/LP CNV diagnostic yield for simple and complicated types in the isolated and extra-CNS groups showed that there was no significant difference (P = 0.971, P = 0.560) ( Table 1).

Homozygous regions in fetal CNS malformations
Only four patients had homozygous regions; after genetic counseling, one termination of pregnancy was performed, and samples from the other three were analyzed by WES. No cases of uniparental disomy or autosomal diseases were found. Two fetuses were successfully followed up to 1 year of age and are now healthy (Table 3).

Discussion
The study of 539 cases included 17 types of CNS malformations, all of which were analyzed by CMA. The overall detection rate for P/LP CNVs was 6.49% (35/539). The diagnostic rate of CNVs was different for different genotypes of fetal central nervous system abnormalities. In the extra-CNS group, the P/LP CNV diagnostic yield was 18.00% (9/50), which was significantly higher than that in the isolated group Thus, it is necessary to monitor the development of the fetal CNS, including the neural tube, lateral ventricle, midbrain, posterior fossa, and especially cortical development, which is frequently missed at the ultrasound examination. Although only 2 cases of fetal cortical abnormalities were included in our study, both were associated with other brain malformations. A recent study included 332 pregnant women and found that sulcal development and cortical maturation could be evaluated prenatally with specific transabdominal ultrasonography at different gestational ages [23]. MRI can be used as a supplementary method and is good for evaluating the overall development of the central nervous system [24].
Among these anomalies, the incidences of lateral ventriculomegaly and hydrocephalus were the highest, followed by posterior fossa cistern malformations and midbrain development abnormalities. CNS malformation commonly occurs in conjunction with extracranial structural abnormalities and an increased risk of aneuploidy [25]. When the chromosome karyotype results are normal and the fetus has major structural abnormalities, prenatal chromosome microarray analysis can detect approximately 8.2% of the pathogenic CNVs [26]. However, Sun [27] et al. found that there was no significant difference in the incidence of pathogenic CNVs between fetuses with isolated nervous system malformations and those with combined extracranial malformations. In contrast to our findings, the diagnostic yield of P/LP CNVs in fetuses with extra-CNS structural abnormalities was higher than that in fetuses with isolated CNS abnormalities. However, the incidence of P/LP CNVs in the complicated type was not significant compared with that in the simple type. Therefore, when CNS malformations are combined with extracranial structural malformations, the risk for fetal chromosomal abnormalities increases, and simple fetal CNS abnormalities also need to be investigated by chromosomal analysis.
Within the brain of an affected individual, the detection rates of pathogenic CNVs differ depending on the CNS malformations and phenotype [28][29][30][31]. Shaffer [32] et al. found various structural abnormalities in 2,858 fetuses with a normal karyotype, and pathogenic CNVs were found most frequently in malformations of the fetal posterior fossa and whole forebrain, at 14.6 and 10.6%, respectively. In our study, there were 35 cases of P/LP CNVs, and the five most frequent phenotypes were holoprosencephaly, Dandy-Walker syndrome, exencephaly, cerebellar hypoplasia and Blake's cyst. We compared and evaluated the P/ LP CNVs identified in this study with those identified in the central nervous system in recent years [27,[33][34][35], and it seems that the probability of P/LP CNVs in anencephaly, microcephaly, arachnoid cysts, ependymal cysts or intracranial hemorrhage was low. In our opinion, on the one hand, the above abnormalities often occur in the third trimester of pregnancy, and the risk of amniocentesis or cordocentesis increases. For arachnoid cysts and ependymal cysts, if prenatal screening is normal and no other structural abnormalities are combined, we could give priority to recommending ultrasonic follow-up for parents. On the other hand, for some brain abnormalities, such as microcephaly and intracranial hemorrhage, we need to pay attention to the importance of gene detection [14,15]. Therefore, it is more conducive for prenatal genetic counseling and helpful for clinicians to  There are no high-frequency genetic syndromes related to CNS-specific fetal malformations. Several cases of syndromic conditions, such as Miller-Dieker syndrome, Williams-Beuren syndrome, 5p syndrome, and Pallister-Killian syndrome, were detected in our study. Miller-Dieker syndrome had the highest incidence, but there were only 3 cases. It is caused by deletion or mutation of the PAFAH1B1 and/or YWHAE gene in the region and is characterized by CNS abnormalities, including the absence of gyri, microcephaly and the absence of the corpus callosum or dysplasia, accompanied by congenital abnormalities [36]. Combined with the published literature [27,[33][34][35], these observations indicate that there are various pathogenic causes that highlight the value of CMA to identify the genes causing fetal CNS malformations (Table 4).
In addition, CMA can detect homozygous regions. Regions of homozygosity have been detected by CMA in all chromosomes except chromosomes 6, 7, 11, 14, 15 and 20, which contain imprinted pathogenic genes [37]. These are considered pathogenic when uniparental disomy (UPD) occurs, and the clinical significance of UPD in other chromosomes is not clear. While larger LOH fragments are not pathogenic, we should pay attention to their relationships to the risk for recessive genetic diseases. In this study, only 4 cases involved homozygous regions. Three cases were classified as negative by WES analysis, and two cases had good results. Song [34] et al. reported 356 cases with CNS abnormalities, and 7 cases of LOH were detected, including 3 cases of induced labor, 3 cases that resulted in normal children and 1 case of developmental delay, but there was no specific description. We consider that ROH plays an unimportant role in fetal central nervous system abnormalities [38]. For ROHs, we should combine the maternal medical history and fetal ultrasound results, a comprehensive analysis should be made on the possibility of the disease and the severity of the disease, and a comprehensive evaluation should be made on whether full exon sequencing is required to exclude potential recessive genetic diseases.
Some limitations of our study are that, as a large retrospective study, it was impossible to classify ventriculomegaly according to its severity due to a lack of data. Additionally, the vast majority of this case series involved ventriculomegaly; therefore, generalization should be performed cautiously based on the current results.
CNS malformations present with a greater variety of phenotypes and chromosomal abnormalities than malformations of other systems. When a simple and isolated fetal brain structural abnormality is present, genetic testing should still be considered. In summary, CMA has substantial clinical application value for the prenatal diagnosis of CNS malformations, and we should use it reasonably and in an orderly manner.