Diagnostic Use of Genome Sequencing in Patients With 11p15.5 Imprinting Disorder Features: A Pilot Study

ABSTRACT

To assess the suitability of genome sequencing (GS) as the second step in the diagnostics of patients with the features of 11p15.5‐associated imprinting disorders (ImpDis: Silver–Russell syndrome [SRS], Beckwith–Wiedemann syndrome [BWS]), we performed short‐read GS in patients negatively tested for imprinting disturbances. Obtaining a genetic diagnosis for patients with the features of these syndromes is challenging due to the clinical and molecular heterogeneity and overlap, and many patients remain undiagnosed after the currently suggested stepwise diagnostic workup. GS was conducted in 48 patients (SRS features: n = 37 and BWS features: n = 11). The detection rate differed markedly between the ImpDis: although a genetic cause could be identified in 51% of patients referred with SRS features, no pathogenic variants were detected in patients with BWS features. Thus, GS substantially improves the diagnostic yield and broadens the spectrum of overlapping disorders with SRS features. Obtaining a precise molecular diagnosis provides the basis for a personalized clinical management. Our findings support the use of GS as a second‐tier diagnostic tool for patients with growth disturbances, as it addresses all currently known variant types and shortens the diagnostic odyssey.

Genome Sequencing should be considered as second‐tier diagnostic tool for patients with 11p15.5‐associated Imprinting Disorders, as it addresses all currently known variant types and shortens the diagnostic odyssey.

1. Introduction

Primordial growth disturbances are frequently caused genetically, and it is a common feature of a broad range of congenital syndromes. Until recently, genetic testing for these patients was conducted in a step‐wise fashion due to limitations of individual testing methods, but in recent years there has been a substantial improvement in diagnostic yields due to the introduction of next‐generation sequencing (NGS) into routine diagnostics [1]. Additionally, the precise molecular diagnosis in children with congenital disorders has become the basis for clinical decision making and personalized treatment.

Two prominent congenital syndromes mainly characterized by growth disturbances are Silver–Russell syndrome (SRS) and Beckwith–Wiedemann syndrome (BWS) [2, 3]. These disorders belong to the group of imprinting disorders (ImpDis), that is, a group of inborn disorders that are caused by dysfunction of the parent‐of‐origin specific regulation of monoallelically expressed genes [4]. BWS and SRS are commonly associated with alterations of the imprinting control regions (IC) in 11p15.5 (IC1:H19/IGF2:IG‐DMR; IC2:KCNQ1OT1:TSS‐DMR). For SRS, the major molecular causes are loss of methylation (LOM) of the IC1, structural variants (SVs) in 11p15.5 and maternal uniparental disomy (UPD) of chromosome 7 (upd(7)mat), and in rare cases pathogenic sequence variants affecting IGF2, CDKN1C, HMGA2, and PLAG1 cause the characteristic SRS phenotype [4]. In BWS, most patients show a LOM of IC2, paternal UPD of 11p15.5 or gain of methylation of the IC1. In addition, pathogenic single‐nucleotide variants (SNVs) in the imprinted gene CDKN1C and SVs encompassing the 11p15.5 region belong to the spectrum of pathogenic variants in BWS.

Pathogenic variants in genes regulated by the ICs in 11p15.5 together with functional downstream factors affect growth pathways and are involved in tumorigenesis, thereby leading to severe short stature, asymmetric growth and a relative macrocephaly in SRS, and vice versa to tall stature, lateralized overgrowth as well as increased risks for embryonal tumors (e.g., Wilms tumor) in BWS [5, 6]. Accordingly, an early identification of the underlying genetic alteration in patients with SRS and BWS facilitates initiation of a personalized treatment and surveillance regime.

Currently, stepwise testing procedures for patients with clinical signs of SRS or BWS have been consented, with methylation‐specific (MS) tests to identify the major imprinting disturbances on chromosomes 11 (in region p15.5) and 7, as the first‐line approach [5, 6]. When MS testing is negative for patients with suspected SRS, molecular karyotyping, and analysis of imprinted loci on chromosomes 14, 16, and 20, and mutation testing of IGF2 as well as CDKN1C have been suggested as next steps [5]. Due to the mosaic occurrence of some of the alterations, testing of further tissues should also be considered. In case of negative MS testing in BWS, Brioude et al. [6] suggested MS analysis of different tissue or locations, as overgrowth syndromes can also frequently be associated with mosaicisms as well [7], additionally, CDKN1C analysis is recommended. For both syndromes, testing for genetic differential diagnosis by DNA sequencing is advised as a last step. However, these stepwise procedures are time, resource and cost consuming, and ~40% of patients with the typical SRS phenotype and ~20% of patients with BWS remain without molecular diagnosis after the first two diagnostic steps, albeit fulfilling the clinical criteria according to the respective clinical scoring systems [5, 6]. In daily routine practice, the detection rate for the typical molecular causes of both ImpDis is much lower at ~20% for patients with SRS features and 25% for those referred for BWS testing, leaving the vast majority of patients without a molecular diagnosis [8].

In recent years, genome sequencing (GS) is increasingly used as a diagnostic tool, with the advantage of combining the diagnostic steps of molecular karyotyping, targeted gene analysis including intronic sequences, as well as of detection of pathogenic UPDs and repeat expansions in a single test [9]. Thus, application of GS in routine diagnostics now offers the potential to increase the diagnostic yield in patients with the features of ImpDis who were tested negatively for methylation disturbances.

Herein, we report the results of a GS‐based diagnostic approach in MS‐test‐negative patients with suspected SRS or BWS, thereby assessing the suitability of a genome wide data analysis as a comprehensive second‐line diagnostic tool and comparing the results with the diagnostic yield of panel‐based approaches. Additionally, we highlight the spectrum of disorders that should be considered as differential diagnosis of 11p15.5‐associated ImpDis.

2. Methods

2.1. Cohort

The total cohort consisted of 48 children referred for genetic testing for either SRS (n = 37) or BWS (n = 11). From all patients and their parents, genomic DNA from peripheral lymphocytes was available.

Patients referred for SRS‐testing were clinically assessed by the Netchine‐Harbinson Clinical Scoring System (NH‐CSS). The NH‐CSS comprises six clinical features including SGA, postnatal growth retardation (PNGR), relative macrocephaly, asymmetry, feeding difficulties, and prominent forehead (Table 2) [10]. In case occipito‐frontal circumference (OFC) at birth was not available to determine relative macrocephaly, the NH‐CSS was adjusted by using the earliest documented OFC, height, and weight to determine relative macrocephaly.

TABLE 2.

Overview on the phenotypes and diagnosed disorders in the cohort of patients referred for SRS testing.

Genes	HMGA2	PLAG1	PCNT	BPTF	IGF1R	PSMD12	GNPTAB	TOP3A	TOP3A	GNAS
Individual	1	2	3	4	5	6	7	8	9	10
Sex	f	m	f	f	m	m	f	f	m	m
Diagnosis
Diagnosis	Silver–Russel syndrome	Silver–Russel syndrome	Microcephalic osteodysplasic primordial dwarfism, type II	Neurodevelopmental disorder with dysmorphic facies and distal limb anomalies	Insulin‐like growth factor I, restistance to	Stankiewitz–Isidor syndrome	Mucolipidosis II alpha/beta	Microcephaly, growth restriction, and increased sister chromatid exchange 2	Microcephaly, growth restriction, and increased sister chromatid exchange 2	Pseudopseudohypoparathyreoidismus
OMIM#	618908	618907	210720	617755	270450	617516	252500	618097	618097	612463
Birth measurements
Gestational week	28	37 + 0	38	37	40 + 1	NR	39 + 0	38	38	41
Birth weight (gr) (SD)	850 (−0.76)	2460 (−1.81)	1000 (−5.3)	2100 (−2.75)	2000 (−2.68)	2500	2250 (−2.61)	900 (−5.38)	2400 (−1.83)	1860 (−4.17)
Birth length (cm) (SD)	NR	NR	NR	< 3 PZ	48 (−1.51)	53	46 (−2.32)	NR	NR	46 (−3.09)
Birth OFC (cm) (SD)	NR	NR	NR	< 3PZ	30 (−4.72)	33	31 (−2.77)	NR	NR	NR
Latest measurements
Age at latest clinical assessment (in years)	4 (7/12)	1 (3/12)	1 (6/12)	9 (6/12)	10 (4/12)	2 (5/12)	0 (8/12)	5 (6/12)	2 (0/12)	4 (4/12)
Weight (kg) (SD)	9.5 (−5.51)	6.89 (−3.29)	4.2 (−7.16)	18 (−3.91)	16.5 (−5.88)	8.09 (−3.88)	4.9 (−3.38)	9 (−7.65)	7.5 (−3.93)	11.3 (−3.72)
Height (cm) (SD)	86.5 (−4.59)	73 (−2.18)	57 (7.78)	117 (−3.34)	124 (−2.75)	83 (−2.37)	63 (−2.45)	92 (−4.54)	77 (3.19)	90 (−3.65)
OFC (cm) (SD)	46.7 (−3.36)	—	37.7 (−10.42)	46.5 (−5.54)	48.5 (−3.88)	48 (−1.76)	40 (3.62)	43 (−7.79)	43 (−5.38)	—
NH‐CSS
NH‐CSS	4/5	3/5	4/5	3/6	4/6	4/6	5/6	5/6	4/6	2/5
SGA	No	No	Yes	Yes	Yes	No	Yes	Yes	No a	Yes
Relative macrocephaly	Yes a	NR	No a	No a	No	Yes a	No	No a	No a	NR
PNGR	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Asymmetry	—	No	—	No	Yes	No	Yes	Yes	Yes	No
Feeding difficulties	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	No
Prominent forehead	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	Yes	Yes
Other features
Cardiac features	NR	NR	Patent foramen ovale	NR	NR	NR	NR	Atrial septal defect	NR	NR
Developmental delay	NR	NR	Mild global developmental delay	Mild global developmental delay	NR	Mild global developmental delay	Mild global developmental delay	NR	NR	Delayed speech and languange dev.
Skeletal abnormalities	Delayed skeletal maturation	2–3 toe syndactyly	Short hand and feet	Delayed skeletal maturation	Delayed skeletal maturation	NR	NR	Pectus excavatum	Pectus excavatum	NR
Hypotonia	NR	NR	NR	NR	NR	NR	Yes	NR	NR	NR
Café au lait spots	Multiple	Multiple	NR	NR	NR	NR	NR	Multiple	Multiple	NR

Genes	TAB2	RIT1	EDC3	ZNF292	TRIO	EVC	CD44, RAG1, PDHX, SLC1A2, TRIM44, RAG2	NSD1, FGFR4, SCL34A1	NIPA2, CYFIP1,TUBGCP5, NIPA1,
Individual	11	12	13	14	15	16	17	18	19
Sex	F	M	M	M	F	F	F	F	F
Diagnosis
Diagnosis	Congenital heart defects, nonsyndromic, 2	Noonan syndrome	Intellectual developmental disorder, autosoma recessive 50	Intellectual developmental disorder, autosomal dominant 64	Intellectual developmental disorder, autosomal dominant 44, with microcephaly	Weyers acrofacial dysostosis	11p13 microduplication syndrome	Silver‐ Russel syndrome	Chromosome 15q11.2 deletion syndrome
OMIM#	614980	615355	616460	619188	617061	193530			615656
Birth measurements
Gestational week	40 + 3	38	39 + 2	38	31 + 2	39	39 + 0	38	38 + 0
Birth weight (g) (SD)	2940 (−0.8)	3370 (−0.05)	2600 (−1.55)	750 (−5.87)	480 (−3.13)	4500 (2.77)	1750 (−4.01)	2170 (−2.61)	2850 (−0.99)
Birth length (cm) (SD)	48 (−1.38)	53 (0.61)	NR	NR	31 (−3.07)	NR	NR	43 (−3.52)	49 (−0.93)
Birth OFC (cm) (SD)	36 (1.65)	36 (0.98)	NR	NR	22 (–3.64)	NR	NR	30 (−3.5)	NR
Latest measurements
Age at latest clinical assessment (in years)	5 (7/12)	3 (1/12)	4 (1/12)	6 (2/12)	3 (6/12)	11 (8/12)	4 (5/12)	4 (0/12)	8 (7/12)
Weight (kg) (SD)	14.3 (−2.77)	12.2 (−1.5)	9.5 (−5.04)	12 (−5.69)	10.2 (−3.29)	16 (−5.89)	11 (−3.78)	8.75 (−5.48)	19.5 (−2.68)
Height (cm) (SD)	102.7 (−2.55)	92 (−1.39)	89 (−3.56)	101 (−3.61)	89.6 (−2.41)	113 (−5.15)	96 (−2.31)	84 (−4.44)	112.5 (−3.47)
OFC (cm) (SD)	—	50.5 (−0.99)	46 (−4.13)	44.5 (−5.96)	42 (−7.48)	51.5 (−1.52)	46.6 (−3.25)	44 (−5.75)	50.9 (−1.03)
NH‐CSS
NH‐CSS	4/6	4/6	4/6	5/6	4/6	4/6	4/6	5/6	4/6
SGA	No	No	Yes	Yes	Yes	No	Yes	Yes	No
Relative macrocephaly	Yes	No	No a	No a	No	Yes a	No a	No	Yes a
PNGR	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Asymmetry	Yes	Yes	No	No	Yes	Yes	Yes	No	No
Feeding difficulties	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	Yes
Prominent forehead	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Other features
Cardiac features	Mitral valve reurgitation, right and left atrial enlargement, cardiac hypertrophy	NR	Patent foramen ovale	NR	NR	NR	NR	NR	NR
Developmental delay	NR	NR	Moderate global developmental delay	Moderate global developmental delay	Delayed speech and language dev. moderate global developmental delay	NR	Moderate global developmental delay	NR	Intellectual disability, moderate global developmental delay
Skeletal abnormalities	Short feet	NR	Delayed skeletal maturation, short hand and feet	Delayed skeletal maturation, short hand and feet	Delayed skeletal maturation	Delayed skeletal maturation, slender finger	Delayed skeletal maturation	NR	Delayed skeletal maturation
Hypotonia	Yes	NR	Yes	Yes	Yes	NR	NR	NR	NR
Café au lait spots	NR	NR	NR	NR	NR	NR	Multiple	NR	NR

Abbreviations: f: female, m: male, NH‐CSS: Netchine Harbinson Clinical Scoring System, NR: not reported, OFC: occipito‐frontal circumference, PNGR: postnatal growth restriction, SGA: small for gestational age.

^a Modified NH‐CSS.

According to the current SRS consensus recommendations, genetic testing for SRS is advised when a patient fulfills at least four of the six NH‐CSS criteria [5]. In case of a negative routine diagnostic testing result, patients are still diagnosed as SRS if their scoring consists of at least four of six criteria, including prominent forehead and relative macrocephaly [10]. If a strong clinical suspicion is raised, children scoring only three of five points may also be eligible for genetic testing [5].

In this group of 37 patients, clinical scoring resulted in three points in nine patients, and in 24 patients a score of four or more was achieved. Of these 24 patients, 6 patients fulfilled both relative macrocephaly as well as prominent forehead as the key features of SRS and were thus diagnosed with clinical SRS. Four of the 37 patients scored less than three points in the NH‐CSS, but they were included in the study because they were referred as SRS for genetic testing due to their growth disturbances. In summary, the cohort referred for SRS–testing can be split into three sub‐cohorts: suspected SRS (n = 27, NH‐CSS > 2 but not clinical SRS), clinical SRS (n = 6, NH‐CSS > 3 with both prominent forehead and relative macrocephaly), and patients with growth restrictions (n = 4, NH‐CSS < 3).

Children referred for BWS were assessed by the Beckwith–Wiedemann spectrum (BWSp) scoring system (BWSp‐SS) as proposed by Brioude et al. [6]. It is composed of six cardinal features rating two points each, as well as eight suggestive features scoring one point each. Patients scoring two or more points merit genetic testing, unless an alternative explanation is present for the clinical phenotypes (e.g., gestational diabetes/diabetes mellitus of the mother and macrosomia). Patients scoring four or more points are diagnosed as having classical BWS. This clinical diagnosis does not require molecular genetic confirmation; however, patients should still be evaluated for a molecular alteration [6].

All BWS patients included in the study showed a clinical scoring of at least two points, warranting genetic testing for BWS. Three of the 11 patients scored four points or more and were thus diagnosed with clinical BWS.

2.2. Genetic Pretesting

In the routine diagnostic workup, SRS and BWS specific alterations affecting the imprinting regions on chromosomes 7, 11, 14, and 20 were excluded by MS multiplex ligation probe dependent amplification (MS‐MLPA kits: ME030, ME032, ME034; MRC Holland, Amsterdam, NL). Patients were only included in the study in case of a negative testing result.

Seven patients had undergone a panel‐based exome sequencing (ES) evaluation, three of them belonged to the cohort of Meyer et al. [11], another three patients had undergone ES by routine diagnostics. However, they all remained unsolved at that time.

2.3. Genome Sequencing

GS was conducted by using the DNA PCR‐free kit (Illumina Inc. San Diego, CA, USA) and sequencing was performed on a NovaSeq 6000 System (S4 Reagent Kit v1.5) (Illumina Inc.), 2 × 158 cycles. Data analysis was performed on the Illumina DRAGEN pipeline (version: 07.021.645.4.0.3) using hg38 reference genome. Tertiary analysis was performed using both the emedgene software (Illumina Inc.) and an in‐house pipeline. In brief, the in‐house pipeline utilizes KGGSeq (v1.2, 06/Nov./2022) for variant filtering and annotation. Variants with a minor allele frequency higher than 0.75% in public databases (gnomAD) were discarded. Variant prioritization and evaluation of pathogenicity was based on different prediction tools (CADD, Polyphen, SIFT, Mutation Taster, Revel, SpliceAI) and variant frequency in public databases (gnomAD).

Copy‐number variants (CNV) were analyzed by an in‐house pipeline that implements the germline copy number variation pipeline provided by CNVkit [12]. The detected CNVs were filtered with in‐house cohort frequencies and visualized utilizing a streamlit web application. Variants were additionally analyzed with the emedgene software (Illumina Inc.) to further assess intronic variants, pathogenic repeat expansions and SVs.

SNVs were classified using the standard guidelines for the interpretation of sequence variants by the American College of Medical Genetics (ACMG) [13] with the updated point system [14]. Further updates proposed by the Sequence Variant Interpretation Working Group were integrated (see https://clinicalgenome.org/working‐groups/sequence‐variant‐interpretation/). CNVs were classified using the suggested guidelines by the ACMG and the Clinical Genome Resource (ClinGen) [15]. Both SNVs and CNVs were checked in ClinVar to assess whether they had already been reported (see https://www.ncbi.nlm.nih.gov/clinvar/). Filtered variants were then analyzed to exclude likely benign and benign variants as well as those that did not fit with the patient's phenotype.

Genomic visualization was performed with the Integrative Genomics Viewer [16].

For the determination of UPDs, the altAF tool was used which predicts isodisomies via runs of homozygosity on each chromosome and heterodisomies via the inheritance ratio of maternal and paternal SNPs per chromosome [17]. The altAF plots of chromosome 11 (BWS) and chromosome 7 (BWS, SRS) were additionally analyzed individually to determine whether a mosaic UPD could be present.

Detected variants were confirmed by a second method, either Sanger sequencing, molecular karyotyping or MLPA. UPDs were confirmed by short tandem repeat profiling.

3. Results

In the SRS cohort of 37 patients, 89% (33/37) scored at least three points in the NH‐CSS, and 16% (6/37) were diagnosed with clinical SRS. The most common documented features in the SRS cohort were PNGR and protruding forehead (95% (35/37) and 92% (34/37), respectively).

In the 11 patients referred for BWS–testing, 73% (8/11) patients scored two or three points, leading to the recommendation for genetic testing. The other 27% (3/11) scored between four and nine points, resulting in the diagnosis of classical BWS as well as leading to genetic testing for the typical genetic variants causing BWS. The most common features were lateralized overgrowth with 9/11 (82%) and birthweight of 2 SDS above the mean with 4/11 (36%) patients fulfilling the category.

In the cohort referred for SRS–testing, a genetic cause of the clinical features could be identified in 19 patients, corresponding to a detection rate of 51.35%.

In the sub‐cohort of suspected SRS, in 48.1% (13/27) of the patients a molecular diagnosis was achieved. One patient was diagnosed with molecular SRS (PLAG1, Patient 2), and another patient carried a duplication of NSD1 which has already been described in a patient with SRS features (Patient 18).

In the sub‐cohort of clinical SRS, 83.3% (5/6) patients gained a molecular diagnosis, one of them was diagnosed with molecular SRS (HMGA2, Patient 1).

In the sub cohort of growth restricted patients, 25% (1/4) patients could be molecularly diagnosed (GNAS, Patient 10).

In total, two of the 19 diagnosed patients referred for SRS testing were diagnosed with molecular SRS (Patients 1 and 2), the remaining 17 patients were diagnosed with already established differential diagnoses of SRS or other syndromes characterized by a growth disturbant phenotype. Disease‐causing SNVs (Table 1) were identified in 16 patients (43.2%), three patients carried large CNVs (8.1%) (Table 1). Further information about the patients is provided in Table 2 and the Supporting Information.

TABLE 1.

Overview on the disease‐causing SNVs and CNVs in the cohort of patients referred for SRS testing.

Individual	1	2	3	4	5	6	7		8
Sex	f	m	f	f	m	m	f		f
Gene	HMGA2	PLAG1	PCNT	BPTF	IGF1R	PSMD12	GNPTAB		TOP3A
Associated phenotype	Silver–Russel syndrome	Silver–Russel syndrome	Microcephalic osteodysplasic primordial dwarfism, type II	Neurodevelopmental disorder with dysmorphic facies and distal limb anomalies	IGF I, restistance to	Stankiewitz–Isidor syndrome	Mucolipidosis II alpha/beta		Microcephaly, growth restriction, increased sister chromatid exchange 2
Mendelian trait/zygosity	ad, het	ad, het	ar, hom	ad, het	ad, het	ad, het	ar, comp het	ar, comp het	ar, hom
Inheritance	mat	NA	mat/pat	de novo	pat	de novo	biallelic	biallelic	mat/pat
gDNA (hg38)	Chr12:g.65828088G > A	Chr8:g.56167201T > A	Chr21:g.46349143_46349144del	Chr17:g.67948156dup	Chr15:g.98943043C > T	Chr17:g.67350335G > A	Chr12:g.101753472_101753473del	Chr12:g.101768103T > A	Chr17:g.18278237dup
cDNA	NM_003483.6:c.198+1G > A	NM_002655.3:c.545A > T	NM_006031.6:c.1164_1165del	NM_182641.4:c.7776dup	NM_000875.5:c.3578C > T	NM_002816.5:c.299C > T	NM_024312.5:c.3503_3504del	NM_024312.5:c.1342A > T	NM_004618.5:c.2271dup
Protein	p.?	p.(Glu182Val)	p.(Arg388Serfs*2)	p.(Arg2593Thrfs*4)	p.(Ser1193Leu)	p.(Ala100Val)	p.(Leu1168Glnfs*5)	p.(Lys448*)	p.(Arg758Glnfs*3)
Effect	splicing	missense	frameshift	frameshift	missense	missense	frameshift	premature stop	frameshift
CADD		28.4			29.4	32
ACMG	VUS (5)	VUS (3)	path (10)	path (11)	lpath (7)	lpath (6)	path (15)	path (12)	path (15)
ClinVar	ID3068509	ID3068518	ID3068508	ID3068507	ID3068521	ID2683747	ID2271	ID3068517	ID560203

Individual	9	10	11	12	13	14	15	16
Sex	m	m	f	m	m	m	f	f
Gene	TOP3A	GNAS	TAB2	RIT1	EDC3	ZNF292	TRIO	EVC
Associated phenotype	Microcephaly, growth restriction, increased sister chromatid exchange 2	Pseudopseudo‐hypoparathyreoidismus	Congenital heart defects, nonsyndromic, 2	Noonan syndrome	Intellectual developmental disorder, autosomal recessive 50	Intellectual developmental disorder, autosomal dominant 64	Intellectual developmental disorder, autosomal dominant 44, with microcephaly	Weyers acrofacialdysostosis
Mendelian trait/zygosity	ar, hom	ad, het	ad, het	ad, het	ar,hom	ad, het	ad, het	ad, het
Inheritance	mat/pat	pat	de novo	de novo	mat/pat	de novo	mat	mat
gDNA (hg38)	Chr17:g.18278237dup	Chr20:g.58905468_58905471del	Chr6:g.149378935_149378936del	Chr1:g.155904475A > G	Chr15:g.74656041del	Chr6:g.87258862C > T	Chr5:g.14471469A > G	Chr4:g.5809560C > T
cDNA	NM_004618.5:c.2271dup	NM_000516.7:c.518_521del	NM_001292034.3:c.1020_1021del	NM_006912.6:c.265T > C	NM_025083.5:c.512del	NM_015021.3:c.5233C > T	NM_007118.4:c.5912+3A > G	NM_153717.3:c.2731C > T
Protein	p.(Arg758Glnfs*3)	p.(Asp173Valfs*11)	p.(Lys340Asnfs*17)	p.(Tyr89His)	p.(Gln171Argfs*9)	p.(Gln1745*)	p.?	p.(Arg911*)
Effect	frameshift	frameshift	frameshift	missense	frameshift	premature stop	splicing	premature stop
CADD				28.6
ACMG	path (15)	path (13)	path (11)	path (13)	path (10)	path (11)	lpath (9)	path (13)
ClinVar	ID560203	ID1191792	ID3068516	ID183411	ID3068515	ID3068522	ID3068520	ID850722

Individual	17*	18	19
Sex	f	f	f
Genes	CD44, RAG1, PDHX, SLC1A2, TRIM44, RAG2	NSD1, FGFR4, SCL34A1	NIPA2, CYFIP1, TUBGCP5, NIPA1
Associated phenotype	11p13 Microduplication syndrome	Silver–Russel syndrome	Burnside–Butler Syndrome
Mendelian trait/zygosity	ad, het	ad, het	ad, het
Inheritance	de‐novo	NA	mat
gDNA (hg38)	Chr11:(34759059_37524365)dup	Chr5:(177041167_177411815)dup	Chr15:(22648948_23124083)del
Band	11p12‐p13	5q35.5	15q11.2
Size	2.7 Mb	370.1 kb	475.1 kb
ClinGen	VUS (0.3)	path (1.3)	VUS (0.85)
ClinVar	ID3338423	ID3338422	ID3338424

Note: For classification of pathogenecity of SNVs see ACMG [15], for CNVs see ClinGen [16].

Abbreviations: ad: autosomal dominant, ar: autosomal recessive, comp het: compound heterozygote, f: female, het: hterozygote, hom: homozygote, lpath: likely pathogenic (6–9 points), m: male, mat: maternal, NA: not assessed, pat: paternal, path: pathogenic (> 9 points), VUS: variant of unknown significance (1–5 points), VUS: Variant of unknown significance.

^*This case has originally been published in [22].

In the BWS cohort, we could not identify any pathogenic variant. One patient carried a maternal uniparental heterodisomy of chromosome 21 (upd(21)mat), and another a heterozygous repeat expansion in the FXN gene, but both variants did not explain their clinical features.

The disease causing SNVs were identified in 15 different genes (Table 1). Six of these SNVs had already been reported in (ClinVar: https://www.ncbi.nlm.nih.gov/clinvar/), the remaining 10 SNVs had not yet been reported (Table 1).

Eleven of the SNVs followed an autosomal dominant mode of inheritance, with five being inherited from one parent (HMGA2, IGF1R, GNAS, TRIO, EVC, Table 1) and five occurring de‐novo (BPTF, PSMD12, TAB2, RIT1, ZNF292, Table 1). In one (Patient 2, PLAG1), no parental data was available. Five variants were associated with autosomal recessive inheritance: Patient 7 was compound heterozygous for two variants in the GNPTB gene (c.3503_3504del; c.1342A > T). The other four cases carried homozygous recessive variants (PCNT, TOP3A, EDC3).

Two of the SNVs were classified as variants of unknown significance (VUS), the other 14 SNVs were categorized as pathogenic or likely pathogenic according to the ACMG criteria [13] (Table 1). The two VUS (HMGA2, c.198+1G > A, Patient 1; PLAG1, c.545A > T, Patient 2) were identified in genes which are regarded as SRS genes (SRS type 5, OMIM# 618908; SRS type 4, OMIM# 618907) [18, 19].

The remaining variants were identified in genes that have not yet been described as differential diagnoses of SRS. Interestingly, two unrelated patients (Patients 8 and 9) from Egypt carried the same homozygous frameshift variant in TOP3A (Table 1, c.2271dup) that had already been reported in the literature [20]. Variants in TOP3A have recently been described to cause a Bloom‐like syndrome characterized by microcephaly, growth restriction and increased sister chromatid exchange [21].

By analyzing the GS data for SVs, three presumably disease causing CNVs were detected. Patient 17 (Figure 1) carried a de‐novo 2.7 Mb duplication in 11p13 (Table 1) and has already been suggested as a new syndrome with a phenotype similar to patients with SRS [22], but in addition with developmental delay and intellectual disability. Patient 18 (Figure 1) carried a heterozygous 370 kb duplication within chromosome 5q35.5 (Table 1) encompassing the NSD1 gene (Table 1). A duplication of NSD1 has been identified already in a patient with SRS features [23]. A 475 kb deletion on chromosome 15 (Table 1) was detected in Patient 19 by single GS. MLPA analysis confirmed the 15q11.2 microdeletion in the mother who did not exhibit clinical features. The phenotype described in literature is consistent with that in our patient (Table 2), and up to 50% of the patients with 15q11.2 microdeletion syndrome inherit the CNV from an apparently unaffected parent, more often from the mother than the father [24].

FIGURE 1.

Patients with clinical suspicion of SRS, now molecularly diagnosed by WGS (y: years).

Testing for UPD is inevitable in phenotypes suspicious for ImpDis. In fact, nearly all clinically relevant UPDs are detectable by the respective MS testing methods used in routine molecular diagnostics, with the exception of upd(16)mat which has also been reported in patients with SRS features [5]. We therefore searched for upd(16)mat in the SRS cohort by using the altAF plotter [17], but an upd(16)mat could not be detected. However, in a patient referred as BWS, chromosome 21 was flagged for potential heterodisomy, and maternal UPD of chromosome 21 (upd(21)mat) could be confirmed by microsatellite analysis. There was no evidence for a mosaic trisomy 21 in fluorescence‐in situ hybridization analysis of buccal swabs. However, this constitution is not associated with an ImpDis phenotype and therefore does not explain the clinical features of our patient [25].

4. Discussion

To identify genomic alterations in patients with SRS and BWS features and to address the molecular heterogeneity in these patients, stepwise diagnostic algorithms have been consented and are widely used in clinical practice [5, 6]. However, these stepwise procedures are time, resource and cost intensive, and in the daily routine diagnostic workup up to ~80% of patients referred for SRS testing and up to ~75% of patients with clinical suspicion of BWS remain without molecular diagnosis [8].

Until now, studies exploring the suitability of GS as a comprehensive diagnostic tool have not yet been performed in patients suspicious for 11p15.5‐associated ImpDis but negatively tested by the first‐step MS assays. In fact, a panel‐based approach based on GS data as a first‐tier diagnostic tool in SRS patients was recently applied by Alhendi et al., resulting in a diagnostic yield of 27% including upd(7)mat. However, this approach did not address IC1 LOM as the major molecular finding in SRS [26]. In 2016, a study on the suitability of ES as a second‐tier diagnostic method in patients referred as SRS has been published with a detection rate for disease‐causing genetic variants of 28% [11]. For patients with BWS features, ES or GS based studies to identify the genetic cause after first‐line methylation testing have not yet been published.

We now report on the first study on GS basis as the second‐line test in patients referred for testing of 11p15.5‐associated ImpDis due to their clinical features, and achieved a detection rate of 51% in the cohort with SRS features, which is considerably higher than the diagnostic yield from the aforementioned publications [11, 26]. This substantial improvement in the diagnostic rate by GS (including genome wide data interpretation in patients with SRS features) confirms its diagnostic utility as a second‐line test for this group of patients.

However, many laboratories apply a virtual filtering and focus on a panel‐based interpretation of NGS data to avoid incidental findings and tailor the diagnostic process to specific clinical inquiries, a strategy which is generally recommended by the EuroGentest for GS [27]. We therefore compared the genes altered in our cohort of patients with SRS features with the list of genes currently covered by the “growth failure in early childhood” and “Silver–Russell syndrome” panels, suggested by PanelApp UK (see https://panelapp.genomicsengland.co.uk/panels/473/ and https://panelapp.genomicsengland.co.uk/panels/199/). Interestingly, only 31.6% (6/19) of the pathogenic SNVs and CNVs in our cohort would have been detected by data filtering based on these panels. Therefore, we propose refraining from using panel‐based data filtering GS approaches in clinically heterogeneous phenotypes like SRS. This suggestion is further corroborated by comparison of our overall detection rate of 51% with the diagnostic yield of 27% in the panel‐based GS approach recently published [26].

To improve the diagnostic yield by GS, EuroGentest suggests the use of trio GS analysis for disorders suspected to be associated with de‐novo variants [27]. This recommendation is supported by our study in which 50% (15/30) of the trio GS patients were successfully diagnosed, whereas only 23.5% (4/17) of single GS cases could be solved.

Among the patients exhibiting features of SRS, we detected pathogenic SNVs in three genes already known to be associated with SRS (HMGA2, PLAG1, NSD1) [11, 18, 19, 23]. The remaining 16 disease‐causing SNVs and CNVs affect genes and genomic regions that are associated with other congenital syndromes. Three of them (PCNT, IGF1R, BPFT) have already been suggested as differential diagnosis of SRS [11, 28]. With the identification of the likely pathogenic missense variant c.299C > T in PSMD12 (Patient 6) and the recent report on a pathogenic CNV encompassing the same gene [26], Stankiewizc‐Isidor syndrome is further established as another differential diagnosis of SRS. The other affected genes have not yet been reported in patients with the clinical signs of SRS. However, our data show that all these syndromes should be regarded as differential diagnosis of SRS, as nearly all these patients were positively scored by the NH‐CSS.

In contrast to the results in the patients with SRS features, we did not identify any disease‐causing variants in blood samples of the probands with clinical signs reminiscent for BWS. A reason for this lack of detection is the mosaic occurrence of genetic alterations in many patients with (asymmetric) overgrowth syndromes, where alterations are often not present in blood samples, but in other tissues. Furthermore, the methodological sensitivity for mosaicism detection will improve in the future with the progress in analytical and bioinformatic tools [3].

Analysis of pathogenic variant types and their genomic localization detected by our GS study revealed that they supposably would all have been detected by ES, therefore the advantages of GS are not obvious at first glance but become obvious by considering the following issues:

A great benefit of GS is its homogeneous coverage, for example, in GC‐rich regions, as these regions can only be reliably addressed by PCR‐free assays [29]. Furthermore, the wet lab workflow is substantially faster and less resource consuming in comparison to ES. Additionally, deep intronic variants and most of the repeat expansions are not covered by ES, and SVs are better resolved with GS [30]. In fact, a recent study about the use of GS for diagnosing rare diseases showed that 28% of the variants identified by GS were not detectable by ES [31], and thereby demonstrates that GS overcomes the limitations of ES. The future diagnostic utility of GS will further increase as methylation signatures will be addressable in parallel, and bioinformatic pipelines permanently advance. Furthermore, the knowledge and databases, especially for deep intronic variants as well as variants in regulatory DNA domains, their impact and their interpretation will improve. Thus, we would like to emphasize the value of GS as the second diagnostic step of ImpDis in the future.

The increased detection rate in our study also fits with the general observation that the periodic re‐analysis of NGS data leads to an increase of the detection rate up to 10%–20% [32, 33]. This is illustrated by three patients from our cohort who remained unsolved after first ES analysis several years ago, but could now be solved due to the following reasons:

1. The occurrence of additional clinical features in the course of the disease which were not present at the first time of clinical evaluation, as well as the broadening of the phenotype in already known monogenetic disorders. An example is Patient 11, where the first ES data were evaluated in respect to growth restriction and SRS‐associated features. In the meantime, cardiac features (Table 2) evolved, and trio GS now revealed the pathogenic de‐novo c.1020_1021del TAB2 variant (Table 1). TAB2 was initially reported to be associated with non‐syndromic congenital heart defects (OMIM# 614980), but recently growth retardation as well as characteristic facial features have been added as further features of the disease [34].

2. Second, there is an ongoing identification of new disease‐causing genes. Examples are Patients 8 and 9 who had undergone ES analysis before the identification of TOP3A as a disease‐causing gene. In the meantime, pathogenic variants in TOP3A have been associated with microcephaly, growth restriction and increased sister chromatid exchange 2 (OMIM# 618097) [21].

The suitability of GS to reach a better diagnostic yield has already been shown by Evans et al. [35], although the costs of ES are currently lower. Thus, the appropriateness of diagnostic strategies must be weighed against the cost‐effectiveness. Performing initial GS analysis after negative MS‐MLPA will maximize the diagnostic outcome, which is particularly relevant for young children waiting for a precise diagnosis as the prerequisite for a personalized treatment. This observation is in accordance with the cost‐effectiveness analysis of GS versus ES in children with suspected genetic disorders conducted by Nurgis et al. [36]. They showed that implementing GS as a first‐tier strategy is most cost‐effective, further highlighting the suggestion to perform GS as the second‐line step in patients with ImpDis features.

To determine the suitability of the NH‐CSS as a clinical diagnostic tool in the cases negative for IC1 LOM and upd(7)mat, individuals with both relative macrocephaly and prominent forehead as the key features of clinical SRS [5] were regarded separately. In our solved SRS cohort, 26.3% (5/19) of patients showed both features and scored at least four criteria of the NH‐CSS. Only one of them was diagnosed with molecular SRS (HMGA2, c.198+1G > A, Patient 1). In contrast, the other four patients received other molecular diagnoses (PSMD12, TAB2, EVC, 15q11.2 microdeletion), albeit fulfilling the criteria for clinical SRS. These data demonstrate that not all clinical SRS patients exhibit typical molecular SRS variants but might be affected by other disorders. This observation is in accordance with the finding from patients with the typical molecular SRS alterations (IC1 LOM, upd(7)mat) who do not generally exhibit the characteristic SRS phenotype [10]. This confirms that the exclusive use of the clinical data ascertained by the SRS and BWS clinical scoring systems for GS data interpretation might hamper the identification of the disease‐causing variant. An example is the identification of the c.1020_1021del variant in TAB2 in Patient 11 which was assessed as disease‐causing because of the reported cardiac symptoms. A precise and comprehensive documentation of clinical features in patients with suspected ImpDis is thus required to improve the detection rate of GS data.

Due to its basic role for genetic counseling and personalized clinical management strategies, the precise molecular diagnosis is crucial in disorders characterized by growth disturbances. For instance, in growth retarded patients, growth hormone treatment is a common therapy. Nevertheless, the decision on whether the patient might benefit or be harmed by growth hormone therapy depends on the molecular diagnosis, as growth hormone therapy might trigger tumorigenesis [37, 38]. In our Patients 8 and 9, the pathogenic c.2271dup TOP3A variant was identified and the associated increased number of sister chromatid exchanges is overlapping with Bloom syndrome, a tumor predisposition syndrome [39]. Only limited data about adolescent patients with TOP3A variants are currently available; therefore, the risk of neoplasia cannot yet be determined in these patients.

Another example for the relevance of a precise molecular diagnosis for clinical management is Patient 12 with molecular Noonan syndrome. The underlying pathogenic c.265T > C RIT1 variant is associated with an increased risk of developing hypertrophic cardiomyopathy, thus an early cardiological monitoring is recommended to detect early signs [40]. Furthermore, Kouz et al. [40] speculated that RIT1 variants within the functional switch II region (affected in Patient 12) might have a higher risk of neoplasia, but further research is needed to confirm this assumption.

An early and regular screening for Moyamoya disease and aneurysms is advised for patients with MOPDII based on biallelic pathogenic variants in PCNT (Patient 3, c.1164_1165del). These patients have an increased risk of developing global vascular diseases, thus they should frequently undergo MRI examinations as well as regular renal ultrasounds and laboratory tests to assess their renal function [41].

Finally, most of the detected variants in our study are associated with very rare syndromes and therefore only limited experiences about clinical management of these diseases are available, thus making a prognosis difficult.

Our study emphasizes the need for broad genetic testing in patients with growth disturbances and syndromic phenotypes. Additionally, we confirmed that a genome‐wide data analysis approach yields a higher diagnostic yield compared to a panel‐based evaluation of GS data in patients with 11p15 ImpDis features and first–tier negative MS testing.

Author Contributions

Luise Kessler: writing – original draft, validation, formal analysis, investigation, and visualization. Jeremias Krause: methodology, software, formal analysis, and data curation. Florian Kraft: methodology, software, and data curation. Asmaa K. Amin, Gyorgy Fekete, Anna Lengyel, Eva Pinti, and Arpad Kovacs: resources. Annette Lischka and Katja Eggermann: validation. Ingo Kurth: conceptualization, and funding acquisition. Cordula Knopp: validation and resources. Miriam Elbracht: conceptualization, resources, and funding acquisition. Matthias Begemann: methodology, and data curation. Thomas Eggermann: conceptualization, validation, formal analysis, resources, visualization, supervision, project administration, and funding acquisition. All authors: reviewing and editing.

Ethics Statement

The study was approved by the ethical committee of the Medical Faculty of the RWTH Aachen (EK303‐18). All patients and their families gave written informed consent. The authors affirm that the patients and their families provided informed consent for publication of the images in Figure 1.

Conflicts of Interest

The authors declare no conflicts of interest.

Peer Review

The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer‐review/10.1111/cge.14649.

Supporting information

Data S1. Clinical description of the patients with molecular diagnoses after GS.

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author, Thomas Eggermann. The data are not publicly available due to their containing information that could compromise the privacy of research participants.