2.1 Blood sampling and karyotyping

Blood samples were collected from all members of the studied family. Informed consents were obtained. The study was approved by the regional ethics committee at the Mashhad University of Medical Sciences, Mashhad, Iran. Chromosomal culture and karyotyping of both children and the parents were performed on routinely cultured peripheral blood lymphocytes. A total of 20 metaphase spreads were analysed in each case according to the GTG banding technique. The maximum banding resolution achieved was 500–600 bands.

2.2 Copy number variation analysis using aCGH

Whole genome oligo aCGH was performed using SurePrint G3 ISCA V2 8X60K whole genome oligo array version 2 and was analysed using Agilent Cytogenomic software v4. The array consists of 60,000 spots with overall median probe spacing of 60 kb and higher near to 500 targeted disease regions. The sample was hybridized against the male sample used as reference.

2.3 Bioinformatic analysis

For aligning the genes affected by detected CNVs to clinical and functional phenotypic features, we used PhenogramViz software. The Human Phenotype Ontology (HPO) terms intellectual disability, neurodevelopmental delay, cognitive impairment, delayed speech and language development, trigonocephaly, corpus callosum atrophy, low posterior hairline, highly arched eyebrow, wide intermammillary distance, wide nasal bridge, long philtrum, thin upper lip vermilion, open mouth, downturned corners of the mouth, cryptorchidism, and patient's brain MRI findings together with a list of two detected CNVs were used for phenotypic matching.

For finding similar CNVs, the DECIPHER database was searched separately for individuals who had deletions of chromosome 9p and duplication of 18p, which was overlapped with the deletion and duplication identified in our case. Our filtering strategy was based on: CNVs of the same size or smaller than the CNVs identified in this study that were entirely contained with the 9p22.3-p24.3 (204193–15260686) and 18p11.31-p11.32(148963–3458388); Pathogenic/likely pathogenic/uncertain CNVs; patients harbouring one genotype and having recorded phenotypes. We focused more on small CNVs affecting only one gene for finding specific gene–phenotype associations.

3.1 Clinical assessment

A 3.5-year-old Iranian female child (Patient 1) and his 10-month-old brother (Patient 2) both with dysmorphic features and congenital malformations were referred for chromosomal analysis and genetic counselling. She was the first child born to healthy, non-consanguineous parents after a full-term pregnancy and normal delivery. The mother was aged 27 years and the father 31 years at the time of the baby's birth. Obstetric history revealed a history of one spontaneous abortion occurring at 10 weeks of gestational age. It seemed that patient 2 has similar complaints, which was determined upon clinical investigation. The remaining family history was unremarkable.

At clinical examination, patient 1's speech, gross motor and social milestones were prominently delayed. She had a trigonocephalic head, low posterior hairline, highly arched eyebrows, wide intermammillary distance, wide nasal bridge, long philtrum, thin upper lip vermilion, open mouth with downturned corners, mild micrognathia, hand flapping and repetitive movements, tapering fingers, and pes planus (Figure S1). She also had a small ventricular septal defect (VSD) as demonstrated by echocardiography. Her transferrin and iron-binding capacity (TIBC) was high (500 μg/dl). Additional laboratory investigations, including brain and renal/urinary tract ultrasonography, blood urea, blood sugar, thyroid profile, creatinine, calcium, alkaline phosphatase, phosphorus, and serum electrolytes were all normal. Her brother had also relatively similar clinical manifestations along with cryptorchidism. Furthermore, his brain MRI showed a hypoplastic (thin) corpus callosum, cavum septum pellucidum, enlarged sylvian cistern, delayed myelination, and trigonocephaly (Figure 1).

3.2 Karyotyping and CNV analysis using aCGH

In all the metaphases from both children, a derivative chromosome 9 was detected, demonstrating a karyotype result of 46, XX, der(9),t(9;18)(p22;p11.31)mat and 46,XY,der(9),t(9;18)(p22;p11.31)mat. A normal karyotype was seen in the father. However, the mother's karyotype was 46, XX, t(9;18) (p22;p11.31) indicating that mother transmitted the detected abnormality to the children (Figure S2).

aCGH analysis of patient 1 showed the result of arr[GRCh37] 9p24.3p22.3 (204193–15260686)x1, arr[GRCh37] 18p11.32p11.31(148963–3458388)x3, indicating a 15.056 Mb deletion of 9p22.3-p24.3 encompassing 122 genes including 43 genes listed in Online Mendelian Inheritance in Man (OMIM) and a gain of 3.309 Mb on 18p11.31-p11.32 encompassing 42 genes including 16 genes listed in OMIM (Figure S3).

3.3 Phenotypic matching using PhenogramViz

For phenotypic matching, the patient's phenotype and CNV data were provided and entered to PhenogramViz. For the 9p deletion CNV, 53 affected genes were identified and 17 of them were matched to 24 HPO terms describing patient's phenotypes (Figure S4). Additionally, for the 18p duplication, 22 affected genes were identified and six of them were matched to 13 phenotypes (Figure S5).

3.4 Overlapped CNVs in the DECIPHER

We summarized the comparable cases of deleted CNVs of 9p22.3-p24.3 (204193–15260686) and duplicated CNVs of 18p11.31-p11.32 (148963–3458388) in the DECIPHER database. In the 9p22.3-p24.3 region, approximately 80 pathogenic/likely pathogenic/uncertain CNVs were identified in DECIPHER based on our filtration strategy. The size ranged from 12.01 kb to 18.45 Mb and 52 CNVs were smaller than 1 Mb in size affecting 10 OMIM morbid genes, NFIB, FREM1, DOCK8, GLDC, KANK1, KCNV2, GLIS3, TYRP1, JAK2 and MPDZ (Tables 1 and S1). Of note, 14 disease-associated genes exist in the 9p22.3-p24.3 region and four genes at 18p11.31-p11.32, which are associated with diverse recessive and dominant Mendelian diseases (Tables 1 and S3). The remaining 28 CNVs with a size larger than 1 Mb harbour four other morbid genes SMARCA2, VLDLR, SLC1A1 and RIC1.

Based on including criteria for overlapping CNVs, the 18p11.31-p11.32 region had 19 overlapping CNVs in the DECIPHER with the size ranging from 23.42 kb to 1.82 Mb (Table S2) These CNVs affect six intolerant genes (high pLI genes) TGIF1, COLEC12, SMCHD1, THOC1, TYMS, and USP14 (Tables 1 and S3). Figure 2 represents an illustration of the affected genes within 9p22.3p24.3 and 18p11.31-p11.32 for which pLI score have been reported.

4.1 Neurodevelopmental and neuroimaging phenotypes

For matching the neurodevelopmental features of our patients to affected genes within detected CNVs, we used the HPO terms, intellectual disability, neurodevelopmental delay, cognitive impairment, delayed speech and language development. Seven genes on the 9p22.3-p24.3, SMARCA2, DOCK8, KANK1, VLDLR, MPDZ, SLC1A1 and GLDC were matched to neurodevelopmental phenotypes, but no matching was seen for genes located in 18p11.31-p11.32 (Figure 3).

SMARCA2 is an intolerant gene to loss-of-function variants (pLI = 1, Figure 4) and also to haploinsufficiency mechanisms (score = 1.60), was directly or indirectly matched to 18 phenotypes of our cases indicating the critical role of this gene in clinical manifestation of 9p deletion syndrome (Figure 4). Mutations and deletions in this gene have been reported to be responsible for the Nicolaides–Baraitser syndrome (OMIM 601358) that is characterized by ID, seizures, microcephaly, sparse hair, short stature, typical face, brachydactyly, and behavioural problems.¹³

Heterozygous mutations and deletions in the DOCK8 gene have been reported in individuals affected with autosomal dominant ID¹⁴ and deletion of the KANK1 gene has been associated with congenital cerebral palsy, hypotonia and ID.¹⁵ VLDLR is another critical gene in the 9p24.3 region, mutations in which can cause cerebellar ataxia, ID and disequilibrium syndrome 1.¹⁶ Above mentioned information together with gene–phenotype matching using PhenogramViz can support that haploinsufficiency of the DOCK8, KANK1 and VLDLR may be responsible for neurological manifestation in individuals with 9p deletion syndrome (Figure S6).

For brain MRI findings, seven genes on 9p, and four genes on 18p were matched to hypoplasia of corpus callosum (SMARCA2, KIAA0020, NFIB, GLDC and FREM1 located at 9p and TGIF1, TYMS, ENOSF1 located at 18p, Figure 5A,C), cavum septum pellucidum (TGIF1, Figure 5C), enlarged sylvian cistern (KANK1 and MPZ located at 9p and TGIF1, Figure 5B,C), delayed myelination (no matching was found) and trigonocephaly (FREM1, TGIF1, ENOSF1 and NDC80, Figure 5A,C).

In support of the role of the KANK1 gene, in a study by Lerer et al.,¹⁷ neuroimaging for nine children belonging to a four-generation family with deletion of the KANK1 gene showed brain atrophy and ventriculomegaly. Hypothetically, ventriculomegaly may be act as a common ancestor phenotype between KANK1 and enlarged sylvian cistern. Accumulation of CSF in the ventricles due to mutations in the MPDZ gene may be also cause ventriculomegaly.¹⁸ The other potential disease-associated gene for the patient 2's observed neuroimaging phenotypes is TGIF1 located at 18p11.3. As a dosage-sensitive gene, TGIF1 was matched to four neuroimaging phenotypes including hypoplasia of corpus callosum, cavum septum pellucidum, enlarged sylvian cistern, and trigonocephaly.

Trigonocephaly as a notable craniofacial feature in our patients (Figures 1 and S1) was matched also to FREM1 gene, which resides in the 300 kb interval on 9p24.3 region. This segment for the first time was defined by Swinkels et al.¹⁹ as a candidate for metopic craniosynostosis, the early fusion of the frontal bones that results in trigonocephaly as a prominent feature of 9p deletion syndrome.

4.2 Facial phenotypes

Six different facial phenotypes including hypertelorism, wide nasal bridge, anteverted nostrils, long philtrum, thin upper vermilion and the open mouth were matched to the five genes on 9p, SMARCA2, JAK2, NFIB, KDM4C and FREM1 (Figure S7) and two genes on 18p, SMCHD1 and TGIF1 (Figure S8). Notable similarities in facial appearance with epicanthal folds, highly arched eyebrows, sparse eyelashes, broad nasal bridge, downturned nasal tip in individuals carrying SMARCA2 variants have been reported by Cappuccio et al.,²⁰ that indicates the clinical importance of small critical region encompassing SMARCA2 on 9p24.3 region (Figure 4).

4.3 Genital phenotypes

Genital abnormalities are estimated to be present in up to 70% of patients with partial 9p monosomy.²¹ The DMRT genes (DMRT1, DMRT2 and DMRT3) located at 9p24.3 as a critical interval for 46,XY complete gonadal dysgenesis and haploinsufficiency of them may cause gonadal dysgenesis as well as abnormal external genitalia, especially in karyotypic males.²² Cryptorchidism and large penis were present in patient 2, but sex reversal was not observed in her. Abnormal external genitalia were not present in patient 1. Genitourinary system findings in our cases may reflect variable penetrance and expressivity of DMRT1. However, the matching of DMRT1, DMRT2 to abnormality of the male external genitalia and large penis by PhenogramViz (Figure S9) is in line with phenotypic and functional consequences of haploinsufficiency of DMRT genes. In addition to DMRT genes, RLN1 and INSL6 are two other genes that have been matched to abnormality of male external genitalia (Figure S9). Co-occurring of these two genes with the cryptorchidism in abstracts of biomedical publications from the DISEASES Text-mining Gene-Disease Association Evidence Scores dataset have been reported by Harmonizome with relatively high scores.²³

In summary, the major clinical features of chromosome distal 9p deletion syndrome (OMIM 158170) were seen in patients harbouring deleted CNVs smaller than 1 Mb in size in the 9p24.3 region. This segment of 9p as a hot spot of copy number losses may be responsible for major clinical manifestation of this syndrome.^{16, 24} In support of this hypothesis are clinical implications of critical genes within this region such as 46,XY gonadal dysgenesis and 46,XY ovotesticular disorder of sexual development (haploinsufficiency of DMRT1),²⁵ autosomal dominant ID 2 (heterozygous disruption of DOCK8 either by a translocation breakpoint or a deletion),¹⁴ familial cerebral palsy (deletion of KANK1)¹⁷ and cerebellar hypoplasia and ID with or without quadrupedal locomotion 1 (Mutations of VLDLR).²⁶ Based on DECIPHER search, eight genes with a pLI score were in the 9p24.3 region including CBWD1, FOXD4, DOCK8, KANK1, DMRT1, DMRT2, DMRT3 and SMARCA2 that three of which, DOCK8, KANK1 and SMARCA2 are OMIM morbid genes. like SMARCA2 as a multiple phenotype matched gene, TGIF1 matched to 10 phenotypes through ancestor phenotypes (Figure S10). Among these phenotypes, ventricular septal defect is notable in that it is matched directly to RFX3 located at 9p24.2.

Interestingly, 34 of 80 DECIPHER cases have CNVs involving brain overexpressed gene, PTPRD. 18 of them harbour only this gene that is intolerant to loss of function variants (pLI = 1) and also to haploinsufficiency mechanisms (score = 0.14). Hypothetically, having an additive effect in causing patient's clinical manifestations especially autistic behaviour could be attributed to PTPRD. Recently implicated as a candidate gene for restless legs syndrome²⁷ and associated with neurodevelopmental disorders,²⁸ PTPRD may present with ADHD and autistic behaviours; however, it is not associated with a disorder in the OMIM database, indicating that its implication in human disease still required further clarification.²⁸ Report of patients with ID and trigonocephaly harbouring homozygous deletions of PTPRD by Choucair et al.²⁹ is another supporting finding for the role of PTPRD in neurodevelopmental disorders. In the output of PhenogramViz, this gene was not matched to any clinical manifestation of our patients.

In conclusion, this study presents an overview of cytogenetic aberrations, developmental features, and clinical characteristics in two patients presenting with 9p22.3-p24.3 and duplication of 18p11.31-p11.32. Furthermore, in order to identify the gene or genes located in detected CNV regions that could potentially act as genetic modifiers of developmental features and clinical characteristics of our patients, the PhenogramViz plugin of CytoScape was used. Finally, for finding specific gene–phenotype associations and narrowing the critical region for the deletion 9p syndrome, overlapping CNVs were sought using DECIPHER database. It should be noted that although aCGH has greatly advanced the identification of CNVs in the human genome, mechanisms of origin of rearrangements are often difficult to resolve, and predicting their effects on gene expression and phenotype remains a challenge. Beside gene–phenotype characterizations via bioinformatic analysis, an important message of this study is that aCGH is a useful approach for diagnosis of the patients with neurodevelopmental delay/intellectual disability and dysmorphic features born to non-consanguineous parents.