Molecular and Cytogenetic Analysis of Romanian Patients with Differences in Sex Development

Differences in sex development (DSD) are often correlated with a genetic etiology. This study aimed to assess the etiology of DSD patients following a protocol of genetic testing. Materials and methods. This study prospectively investigated a total of 267 patients with DSD who presented to Clinical Emergency Hospital for Children Cluj-Napoca between January 2012 and December 2019. Each patient was clinically, biochemically, and morphologically evaluated. As a first intervention, the genetic test included karyotype + SRY testing. A high value of 17-hydroxyprogesterone was found in 39 patients, in whom strip assay analysis of the CYP21A2 gene was subsequently performed. A total of 35 patients were evaluated by chromosomal microarray technique, and 22 patients were evaluated by the NGS of a gene panel. Results. The karyotype analysis established the diagnosis in 15% of the patients, most of whom presented with sex chromosome abnormalities. Genetic testing of CYP21A2 established a confirmation of the diagnosis in 44% of patients tested. SNP array analysis was particularly useful in patients with syndromic DSD; 20% of patients tested presented with pathogenic CNVs or uniparental disomy. Gene panel sequencing established the diagnosis in 11 of the 22 tested patients (50%), and the androgen receptor gene was most often involved in these patients. The genes that presented as pathogenic or likely pathogenic variants or variants of uncertain significance were RSPO1, FGFR1, WT1, CHD7, AR, NIPBL, AMHR2, AR, EMX2, CYP17A1, NR0B1, GNRHR, GATA4, and ATM genes. Conclusion. An evaluation following a genetic testing protocol that included karyotype and SRY gene testing, CYP21A2 analysis, chromosomal analysis by microarray, and high-throughput sequencing were useful in establishing the diagnosis, with a spectrum of diagnostic yield depending on the technique (between 15 and 50%). Additionally, new genetic variants not previously described in DSD were observed.


Background
Differences in sex development (DSD), defined by atypical developments of chromosomal, gonadal, or phenotypic sex, are observed in 1:4000 newborns, although this incidence may differ according to the disorders included. Isolated hypospadias or cryptorchidism are more frequently observed, in as many as 1:200 newborns [1][2][3][4]. These disorders are at enzymatic block. If not, a gene panel or exome, or whole genome will be performed using next-generation sequencing (NGS) techniques [13]. If DSD is associated with other symptoms (e.g., syndromic DSD), chromosomal analysis by microarray is performed to observe the copy number variants (CNVs) that may have caused the disorder. Knowing the precise diagnosis will help in understanding the prognosis and designing better treatments that consider a precise etiopathogenetic mechanism. Therefore, the aim of this study was to assess the genetic etiology of patients with DSD presented in our service, who were clinically, biochemically, and anatomically evaluated.

Materials and Methods
Patients with DSD who presented to Clinical Emergency Hospital for Children Cluj-Napoca between January 2012 and December 2019 were investigated prospectively. The inclusion criteria for the study group were clitoral hypertrophy, posterior hypospadias, bilateral cryptorchidism or ectopia, unilateral cryptorchidism/testicular ectopia associated with hypospadias or micro-penis, puberty delay, and primary amenorrhea. A total of 267 patients were evaluated ( Figure 1) with karyotype and SRY testing (using either fluorescent in situ hybridization (FISH) or polymerase chain reaction (PCR) techniques).

Chromosomal Microarray Technique
The chromosomal microarray comprised an SNP array, which was performed using an Infinium OmniExpress-24 BeadChip array kit (Illumina, San Diego, CA, USA), and the platform iScan System (Illumina, San Diego, CA, USA). The bioinformatic instrument was Genome Studio software version 2.0 (Illumina, San Diego, CA, USA). This analysis was based on 700,000 markers, and the interpretation of the results was based on the American College of Medical Genetics (ACMG) recommendation [15].

Genes Panel Sequencing
The sequencing was performed using the TruSight One Kit (Illumina, San Diego, CA, USA), which targets 4800 genes associated with human pathology (12 Mb). This included around 150 genes or candidate genes associated with the clinical phenotype for DSD. The sequencing was performed with the MiSeq platform (Illumina, San Diego, CA, USA) [16] using the manufacturer's instructions. Bioinformatic analysis was performed using Galaxy bioinformatic platform, and variant interpretation was based on ACMG recommendations [17].

Results
A total of 267 patients were evaluated with karyotype and SRY testing. Of these patients, 39 (14.6%) were diagnosed with different chromosomal abnormalities, most of them involving the sex chromosomes (36 of 39 patients; Table 1). The 46,XX karyotype was observed in 104 patients (39%), while the 46,XY karyotype was observed in 122 patients (46%). Two patients with the 46,XX karyotype were SRY positive. Patients were evaluated by imaging studies (ultrasound, pelvic MRI), with additional studies (such as exploratory laparotomy and gonadal biopsy) performed depending on the clinical context. Hormonal testing (for 17-hydroxyprogesterone, DHEAS, delta4 androstenedione, testosterone, DHT, and AMH) was also performed for some patients depending on the clinical context.
In 46,XX patients with high 17-hydroxyprogesterone values (greater than 2 ng/mL), testing for 21-hydroxylase deficiency was performed. For some patients, genetic testing was performed if 17-hydroxyprogesterone was over 10 ng/mL after a stimulation test with synthetic ACTH. Forty-one 46,XX patients were genetically tested using strip assay analysis for the 11 most common mutations of CYP21A2 ( Figure 1).
For patients for whom karyotype, SRY, and 21-hydroxylase deficiency testing did not establish a diagnosis, the SNP array technique was used to evaluate CNVs (copy number variants) if the patients continued to be evaluated. Some patients declined follow-up investigations precluding obtaining further clinical data or biological samples. These patients had no clinical, hormonal, or morphological differences from the group further tested. Therefore, given limited financial funds for genomic testing (SNP array or high throughput sequencing) provided by our healthcare system, it was decided to perform these tests only for the patients who continued the follow-up for their disorders.
SNP array testing was completed for 35 patients (Figure 1), of whom 22 patients had negative results and were consequently evaluated with gene panel sequencing (TruSight One panel, Illumina) ( Figure 1). This panel included known genes associated with human pathology, particularly those associated with DSD.
The research was approved by the ethics committee of "Iuliu Hatieganu" University of Medicine and Pharmacy of Cluj-Napoca. Written informed consent was obtained from the parents of all patients in the study.

Chromosomal Microarray Technique
The chromosomal microarray comprised an SNP array, which was performed using an Infinium OmniExpress-24 BeadChip array kit (Illumina, San Diego, CA, USA), and the platform iScan System (Illumina, San Diego, CA, USA). The bioinformatic instrument was Genome Studio software version 2.0 (Illumina, San Diego, CA, USA). This analysis was based on 700,000 markers, and the interpretation of the results was based on the American College of Medical Genetics (ACMG) recommendation [15].

Genes Panel Sequencing
The sequencing was performed using the TruSight One Kit (Illumina, San Diego, CA, USA), which targets 4800 genes associated with human pathology (12 Mb). This included around 150 genes or candidate genes associated with the clinical phenotype for DSD. The sequencing was performed with the MiSeq platform (Illumina, San Diego, CA, USA) [16] using the manufacturer's instructions. Bioinformatic analysis was performed using Galaxy bioinformatic platform, and variant interpretation was based on ACMG recommendations [17].

Results
A total of 267 patients were evaluated with karyotype and SRY testing. Of these patients, 39 (14.6%) were diagnosed with different chromosomal abnormalities, most of them involving the sex chromosomes (36 of 39 patients; Table 1). The 46,XX karyotype was observed in 104 patients (39%), while the 46,XY karyotype was observed in 122 patients (46%). Two patients with the 46,XX karyotype were SRY positive. Of patients with the 46,XX karyotype, 39 were tested for the main CYP21A2 mutations (using strip assay). This testing was performed if their basal 17-hydroxyprogesterone was greater than 2 ng/mL or if stimulated 17-hydroxyprogesterone (after synthetic ACTH administration) was greater than 10 ng/mL. In 17 (44%) of the 39 patients, the diagnosis was confirmed.
Of 85 patients with the 46,XX karyotype (without an established etiologic diagnosis, after CYP21A2 testing) and of 122 patients with the 46,XY karyotype, 8 and 27 patients, respectively, were further analyzed using chromosomal microarray ( Table 2). The other undiagnosed patients were not tested further for different reasons (i.e., absent biological sample, absent clinical follow-up, limited possibility of genomic testing). Three of the eight 46,XX patients (38%) presented pathogenic CNVs (two patients) or VUS (variant of unknown significance)(one patient); two of these patients presented syndromic 46,XX DSD, while one presented an isolated form ( Table 2).
Patient p10, 46,XY, presented a recurrent 16p11.2 deletion of 597kb, which has classically been associated with intellectual disability, developmental delay, autism spectrum disorders, and obesity. Cryptorchidism has also been mentioned in patients with this CNV. However, the precise gene responsible for cryptorchidism is not known. Patient p4 presented with disomy of chromosome 7, and his phenotype was concordant with the Russell-Silver phenotype. This syndrome has been associated with cryptorchidism, micropenis, and hypospadias in other patients, and thus the genital phenotype in this patient was considered a result of this syndrome. Two patients, p14 and p21, presented 15 chromosome isodisomy, and further MLPA identified a maternal origin for these chromosomes. The clinical phenotype superposed that of Prader-Willi syndrome, and in this clinical context, the cryptorchidism was due to hypogonadotropic hypogonadism. Patient p18, who had an exon 4 deletion in the OTC gene, presented cryptorchidism. OTC mutations have not been previously described in association with this phenotype; however, the metabolic defect associated with hypotonia at a younger age may have had some association with cryptorchidism. Patients p3 and p6 presented uncertain CNVs in the 15q11.2 region, but previous data have not supported the involvement of these CNVs in DSD. Patients p24, p26, and p35 presented variants of uncertain significance, with no clear argument for their consideration in a final diagnosis.
P30, a 46,XX patient, presented 1q21.1 deletion, or BP2-BP3 (thrombocytopenia absent radius [TAR] syndrome), and the clinical phenotype of the patient superposed with this syndrome. Genitourinary abnormalities and Mayer-Rokitansky-Küster-Hauser syndrome have been described in TAR syndrome, and this diagnosis was etiologically appropriate for the uterus agenesia described in patient p30. The gene RBM8A is suggested to be responsible for the Mayer-Rokitansky-Küster-Hauser phenotype in TAR syndrome.
Patient p31 was diagnosed with 18p11.32-18p11.31 deletion, an etiological diagnosis for genitourinary abnormality, cryptorchidism, and micro-penis that have been described in other patients with this deletion. The CYP21A2 homozygous deletion in 46,XX patient p32 identified by SNP array technique was also an etiologic diagnosis for this patient. Chromosomal microarray led to a clear etiologic diagnosis for two of eight 46,XX patients (25%) and for five of 27 46,XY patients (19%). Thus, of the 35 patients tested (46,XX and 46,XY), 7 (20%) had a final etiologic diagnosis.
Patients with negative results after SNP array testing were analyzed by gene panel sequencing that included the known and candidate genes involved in DSD (Table 3)        Four 46,XY patients (p5, p8, p13, and p16) presented variants in the AR gene. In patient p5, the variant c.2071_2073del was classified as likely pathogenic (an in-frame deletion, not described in gnomAD, found in the hot-spot region of the AR gene, predicted as pathogenic by phyloP). Patient p8 presented two VUS variants, c.167T>A and c.170_171insGCAGCAGCA (not found in gnomAD, predicted as pathogenic by phyloP). In patient p13, the variant c.1097A>C was classified as VUS (missense variant in exon 1, not found in gnomAD, 96% of missense variants in the AR gene are pathogenic, prediction as pathogenic by the prediction platforms

Discussion
In this study of 267 patients with DSD, of the 93 patients that completed all applicable tests, a diagnosis was obtained for 87 (94%) of the patients. A total of 174 patients did not pursue genetic testing and remained etiology unknown following karyotyping, SRY testing, and CYP21A2 strip testing. Karyotype testing established the diagnosis in 15% of patients, most of whom presented abnormalities of the sex chromosomes. A value of basal 17-hydroxyprogesterone above the threshold of 2 ng/mL or of stimulated 17-hydroxyprogesterone above 10 ng/mL indicated a deficiency of 21-hydroxylase, and genetic testing of CYP21A2 confirmed the diagnosis in 17 out of 39 patients (44% of CYP21A2 tested patients).
SNP array analysis was particularly useful in DSD patients who presented associated signs (syndromic DSD). Of patients tested by SNP array, 20% received a final etiologic diagnosis, and almost all of them, with one exception, presented syndromic DSD (this percentage was similar between 46,XX and 46,XY DSD). Gene panel sequencing, which included DSD-associated genes (known and candidate) established or strongly suggested the etiologic diagnosis in 11 of 22 patients tested (50%). The androgen receptor gene mutations were most observed in the study group, but variants in some genes less commonly associated with DSD were also observed, such as those in the GATA4 gene.
Regarding the percentage of chromosomal abnormalities observed in patients with DSD, a similar result (around 15%) has been found by other studies, and most of these abnormalities involve the sex chromosomes [18,19]. Thus, karyotype has a role not only in establishing the first step in a pathogenetic algorithm (in 46,XX or 46,XY DSD) but also in observing an etiology in a number of cases (in sex chromosome DSD). The karyotype also provides advantages for identifying mosaicisms, translocations, and X chromosome structural variants, which in some situations are not easily suggested using chromosomal microarray [13].
Concerning the CNVs observed in the study group, the percentage of pathogenic CNVs and uniparental disomy was 20%, similar to that found in other studies, which ranged from 15-20% [18,[20][21][22][23]. One study found that chromosomal microarray is similar to classical cytogenetics when identifying Turner syndrome [24]. However, chromosomal microarray analysis has advantages over karyotype regarding the identification of cryptic Y chromosome material in patients with Turner syndrome [13]. For example, 1q21 deletion (involving RBM8A gene), identified in a 46,XX patient, is a described etiology in Mayer-Rokitansky-Küster-Hauser syndrome, which is present in some patients with this BP2-BP3 deletion [25].
Another patient was diagnosed with 21-hydroxylase deficiency due to homozygous deletion identified by SNP array testing. The presence of maternal uniparental disomy of 15 and uniparental disomy of chromosome 7 in three of the 35 patients studied by chromosomal microarray provided useful information, as the phenotype of these disorders (Prader-Willi or Russell-Silver syndromes) includes signs of DSD usually associated with other findings that are less obvious. Concerning the other pathogenic CNVs or VUS described in the study, a clear association was not found between genes included in these regions and DSD; these CNVs or genes have been more frequently associated with neurologic development than that of other systems or organs.
High-throughput sequencing was found to be the most effective in cases of DSD patients due to its high percentages of positive results. The findings of high-throughput sequencing indicated etiologic and pathogenetic mechanisms, which could be very important for designing optimal therapy and achieving a better prognosis. Of the study's patients, 50% had a pathogenic or likely pathogenic variant responsible for their clinical development, and a similar percentage was observed in a large international cohort of DSD patients [26]. However, this percentage was higher than that observed in other studies on DSD patients (around 30%) [27][28][29][30][31][32]. This higher percentage may have been due to the fact that the patients tested by sequencing comprised those with a severe DSD phenotype [28]. A similar percentage of diagnosis was found for both 46,XY and 46,XX patients. The genes that presented pathogenic or likely pathogenic variants or VUS were RSPO1, FGFR1, WT1, CHD7, AR, NIPBL, AMHR2, AR, EMX2, CYP17A1, NR0B1, GNRHR, GATA4, and ATM genes. As in other studies, the frequency of AR gene mutations was higher than that of other genes [28,[33][34][35].
The variant described in the GATA4 gene, c.698C>A (p.Thr233Lys), fell within the N-terminal zinc finger region, one that was recently proved by van der Bergen to have a pathogenic effect in 46,XY DSD, unlike variants in other regions of this gene, which usually have a benign contribution [36]. Another pathogenic variant identified in the present study was in the NIPBL gene; this variant is associated with Cornelia de Lange syndrome. The genetic diagnosis for this patient was due to genital abnormalities seen and less to the dysmorphic and neuropsychiatric signs classically associated with this syndrome.
A limitation of the present study was the impossibility of evaluating the parents to establish the de novo or inherited characteristics of the unknown variants, as well as the inability to perform functional studies for these variants. However, an important contribution of this study was that it was the first in the authors' country to perform a well-defined algorithm for genetic testing in DSD.
In conclusion, an evaluation following a genetic testing algorithm including karyotype and SRY gene testing, CYP21A2 analysis, chromosomal analysis by microarray, and nextgeneration sequencing provided a diagnosis for 87 patients with a spectrum of diagnostic yield between 15 and 50%, depending on the technique. Additionally, new genetic variants not previously described in DSD patients were observed.