The Genetic Landscape of Children Born Small for Gestational Age with Persistent Short Stature

Introduction: Among children born small for gestational age, 10–15% fail to catch up and remain short (SGA-SS). The underlying mechanisms are mostly unknown. We aimed to decipher genetic aetiologies of SGA-SS within a large single-centre cohort. Methods: Out of 820 patients treated with growth hormone (GH), 256 were classified as SGA-SS (birth length and/or birth weight <−2 SD for gestational age and life-minimum height <−2.5 SD). Those with the DNA triplet available (child and both parents) were included in the study (176/256). Targeted testing (karyotype/FISH/MLPA/specific Sanger sequencing) was performed if a specific genetic disorder was clinically suggestive. All remaining patients underwent MS-MLPA to identify Silver-Russell syndrome, and those with unknown genetic aetiology were subsequently examined using whole-exome sequencing or targeted panel of 398 growth-related genes. Genetic variants were classified using ACMG guidelines. Results: The genetic aetiology was elucidated in 74/176 (42%) children. Of these, 12/74 (16%) had pathogenic or likely pathogenic (P/LP) gene variants affecting pituitary development (LHX4, OTX2, PROKR2, PTCH1, POU1F1), the GH-IGF-1 or IGF-2 axis (GHSR, IGFALS, IGF1R, STAT3, HMGA2), 2/74 (3%) the thyroid axis (TRHR, THRA), 17/74 (23%) the cartilaginous matrix (ACAN, various collagens, FLNB, MATN3), and 7/74 (9%) the paracrine chondrocyte regulation (FGFR3, FGFR2, NPR2). In 12/74 (16%), we revealed P/LP affecting fundamental intracellular/intranuclear processes (CDC42, KMT2D, LMNA, NSD1, PTPN11, SRCAP, SON, SOS1, SOX9, TLK2). SHOX deficiency was found in 7/74 (9%), Silver-Russell syndrome in 12/74 (16%) (11p15, UPD7), and miscellaneous chromosomal aberrations in 5/74 (7%) children. Conclusions: The high diagnostic yield sheds a new light on the genetic landscape of SGA-SS, with a central role for the growth plate with substantial contributions from the GH-IGF-1 and thyroid axes and intracellular regulation and signalling.


Introduction
Approximately 5% of children are born small for gestational age (SGA)with a birth weight and/or length below −2 SD compared to normative values for their gestational age [1].The aetiology of SGA is heterogeneous (environmental, maternal, placental, and endogenous factors, including defined gene variants [2]).Up to 90% of SGA children develop catch-up growth during the first 2 years of life, while the remaining fail to catch up and are referred to as 'small for gestational ageshort stature' (SGA-SS).These children are known to remain small throughout childhood and reach a substantially reduced adult body height [3,4].They are therefore indicated for treatment with growth hormone (GH) [1,5,6].Nevertheless, the response to GH administration is variable among individual SGA-SS children, which may reflect the heterogeneous aetiology of their growth failure [7,8].
In SGA-SS, several genetic mechanisms should be taken into consideration: imprinting disorders and abnormal methylation patterns such as Silver-Russell syndrome (SRS), Temple syndrome, IMAGe syndrome, and others [9][10][11].In addition, a long list of single gene conditions has been associated with the regulation of human growth and thus impact on final height, albeit not necessarily associated with prenatal growth restriction [12,13].Some of these genes regulate the structural development of the cerebral midline and pituitary and functional components of the GH-IGF-1 axis (hormones, their receptors, and post-receptor signalisation).Moreover, new genes have been discovered which code for important growth plate paracrine factors, proteins of cartilage extracellular matrix, components of intracellular regulating cascades, and proteins involved in fundamental intranuclear processes [2].
The elucidation of the genetic background of SGA-SS was initiated no more than 2 decades ago [2].In some cases, a child might present with typical features, leading to targeted genetic testing.A typical example is the genetic diagnosis of SRS in individuals fulfilling the Netchine-Harbison clinical criteria [9].However, most SGA-SS children present with no apparent syndromic features; therefore, genetic diagnosis is challenging.
New possibilities of genetic testing such as nextgeneration sequencing (NGS) allowed new advancements in discovering the genetic aetiology of short stature within the past decade [14].Knowledge of the genetic basis of growth disorders in these children not only helps in better understanding the pathophysiology of growth but may have important consequences for their treatment and follow-up as well.The aim of this study was to decipher genetic aetiologies among a large single-centre cohort of SGA-SS children treated with GH and to stratify them according to molecular mechanisms leading to combined pre-and postnatal growth failure.

Patients
The study cohort was selected from 820 children treated with GH in our centre between May 2008 and December 2018 using a stepwise selection process as displayed in Figure 1.Other causes of growth failure were considered and appropriately evaluated before starting GH therapy.Extremely preterm children (gestational age <28 weeks) were excluded due to missing relevant normative values for their size at birth.After exclusion of children treated with GH for other causes (chronic kidney disease, acquired GH deficiency (GHD), Turner syndrome, Prader-Willi syndrome, and primary GHD born either appropriate for gestational age or SGA but with life-minimum height >−2.5 SD), 256 children with SGA-SS (birth weight or length <−2 SD and body height <−2.5 SD after 3 years of life) remained for further evaluation.Out of them, 176/256 (69%) families agreed to genetic testing; therefore, the child and both of his/her parents were enrolled in the study (Fig. 1).The clinical assessment of all children included measurements of weight (using an electronic scale) and height (mean of three measurements using a calibrated stadiometre to the nearest 1 mm).These results were converted to the SDS using age-and sex-specific normative values [15].The height of the parents was either obtained during the patients' visit using the same method or referred from their medical records.Birth parameters were obtained from medical records.
All SGA-SS children underwent long-term GH treatment with a dosage of 35 μg/kg/day as suggested in the consensus from Clayton et al. [5].If the child was also found to have GHD, the dose was in the range of 25-35 μg/kg/day in accordance with summary of product characteristics, and in the case of SHOX deficiency, 50 μg/kg/ day as recommended in previous studies [16].

Genetic Testing
Genetic Testing Prior to the Study All children with a clinical suspicion of a specific genetic disorder underwent genetic examination with an appropriate method (karyotype, FISH, MLPA, targeted Sanger sequencing) prior to the study.The remaining children were examined for SRS.After its' exclusion, patients were examined by NGS methods.
Examination of SRS Methylation-specific multiplex ligation-dependent Probe amplification (MS-MLPA) was done in all patients.MS-MLPA (probe mixes ME030 and ME032 examining regions of 11p15, 7q32, 7p12, and 14q32, respectively) and subsequent data analyses by software Coffalyser were performed according to the manufacturer's instructions (MRC Holland, Amsterdam, The Netherlands).
Targeted NGS Genomic DNA was extracted from peripheral blood using QIAmp DNA Blood Mini (Qiagen, Hilden, Germany) or from saliva (collected into Oragene OG-500) according to the manufacturer's instructions (DNA Genotek, Ontario, Canada).DNA of patients without a verified genetic cause of their growth failure was analysed using a custom-targeted NGS panel of 398 genes with a known or potential association with growth [17] using SureSelect Custom Kit (Agilent Technologies, Santa Clara, CA, USA), and the indexed products were sequenced by synthesis on an Illumina MiSeq platform (San Diego, CA, USA) with ×100 average coverage.Altogether 6 DNA samples from probands underwent the whole-exome sequencing using SureSelect Human All Exon v6+UTR Kit (Agilent Technologies).The indexed products were sequenced by synthesis on an Illumina MiSeq or NextSeq platform (San Diego, CA, USA) with ×100 average coverage.Obtained sequences were annotated and mapped to reference genome followed by variant calling as described previously [17].Detected variants were filtered using software Variant Annotation and Filter Tool [18] with filter settings described previously [17].

Evaluation of Genetic Findings
Confirmation of all variants of interest in the patient and segregation analyses in available family members were performed by direct Sanger sequencing [19].Subsequently, variants were scored according to the American College of Medical Genetics and Genomics (ACMG) standards and guidelines [20] implemented in the VarSome software [21] as pathogenic (P), likely pathogenic (LP), benign (B), likely benign (LB), or as variants of uncertain significance (VUS).Consideration of co-segregation in the pathogenicity classification of variants (criterion PP1 in the ACMG guidelines) was applied based on recommendations by Jarvik and Browning [22].

Ethics Statement
This study protocol was reviewed and approved by the Institutional Ethics Committees of the 2nd Faculty of Medicine, Charles University in Prague, and University Hospital Motol, Czech Republic (date of approval: June 30, 2017; not numbered).Written informed consent was obtained from the parents/legal guardians of the patients for publication of the details of their medical cases and any accompanying images.Flowchart of the study.GH, growth hormone; CKD, chronic kidney disease; GHD, growth hormone deficit; IGHD, idiopathic growth hormone deficit; AGA, appropriate for gestational age; MS-MLPA, methylation-specific multiplex ligation-dependent probe amplification; WES, whole-exome sequencing; SRS, Silver-Russell syndrome; VUS, variant of uncertain significance; LB, likely benign; B, benign; LP, likely pathogenic; P, pathogenic.Yes BA-CA, difference between bone age and chronological age; BL (cm), birth length (cm); BW (g), birth weight (grams); GW, gestational week; GH, growth hormone; GHD, growth hormone deficiency; MRI, magnetic resonance imaging; N/A, not available; SDS, standard deviation score.
Genetic results in children born small for gestational age with persistent short stature (SGA-SS)

Patient
No.  *Based on American College of Medical Genetics and Genomics (ACMG) standards and guidelines [20] implemented to the VarSome software [20] (on the date December 16, 2021).a Probands who carried these variants have been reported previously in our studies.
Overall, in our cohort, 40 out of 74 patients (54%) had positive genetic findings and no dysmorphic features.Part of these results were published in our previous reports on children from families with vertical transmission of short stature ("familiar short stature") [17] and/or in a paper summarising the effect of GH therapy in children with pathogenic NPR2 variants [23] and nonsyndromic collagenopathies [24].The principal clinical and growth data are summarised in Table 1.All the genetic findings are presented in Table 2.The singlegene conditions (and SRS) and their significance at three levels of growth regulation are displayed in Figure 2a-c.

Discussion
In our study, we examined a unique large single-centre cohort of SGA-SS children by NGS methods.We elucidated the genetic cause of growth disorder in 42% (74/ 176) of them.The results demonstrate a multifarious genetic landscape of SGA-SS and further contributed to the understanding of its aetiology.
Advances in genetic diagnostics led to better knowledge regarding the mechanisms causing short stature.Depending on the study cohort, NGS methods elucidated the aetiology of a growth disorder in 14.5-52% of cases [17,[26][27][28][29][30].Two of these studies focused on SGA-SS children.Freire et al. identified monogenic SGA-SS in 8/55 (15%) children with no apparent syndromic features.The aetiologies of their short stature were mostly primary growth plate disorders accompanied by the disruption of the RAS/MAPK signalling pathway [29].Li et al. [26] demonstrated that including syndromic SGA-SS children can substantially increase the detection rate of genetic variants elucidating the aetiology of SGA-SS (17% in non-syndromic SGA-SS vs. 31% in syndromic SGA-SS).In our study, we have managed to find the genetic cause of SGA-SS in a higher number of children (42%), even in the case of non-syndromic SGA-SS (54%).
The aetiology of SGA-SS in our study cohort was rather heterogeneous.Not surprisingly, 42% of children with genetic aetiology elucidated carried a causal variant in the gene that is essential for correct growth plate function.This finding is in line with previous studies [26,29] and corresponds with the new paradigm with the growth plate playing a key role in short stature pathogenesis [31].As expected, other relatively frequent genetic diagnoses in our SGA-SS study cohort were causal variants in genes affecting fundamental intracellular Fig. 2. Three levels of the genetic growth regulation in children born small for gestational age with persistent short stature (SGA-SS).The numbers in brackets in blue boxes show the numbers of patients identified with P/LP variants of the entire gene (if more than one).The numbers in closed circles refer to patient numbering in Table 1. a Genes involved in cerebral midline and pituitary development, and in the GH-IGF-1 and thyroid axes.b Genes encoding growth plate matrix components and elements of chondrocyte paracrine regulation.c Genes involved in intracellular signalling, in the stability of nuclear membrane, and in the fundamental intranuclear processes.A brief description of selected patients' facial phenotypes: processes including RASopathies which corresponds with the results of previous studies as well [26,29].
Another condition typically associated with SGA-SS is SRS [9].Not surprisingly, SRS was another frequent diagnosis in our study cohort (16% of cases with genetic aetiology elucidated).Importantly, SRS is diagnosed clinically using the Netchine-Harbison scoring system (NHS).Genetic examination may consequently provide useful confirmation of the clinical diagnosis [9].In our study, we took a different approachgenetic examination of SRS was performed in all SGA-SS children.Surprisingly, we have genetically diagnosed SRS in 2 children who do not fulfil NHS criteria.Genetic examination of SRS can therefore be considered in all SGA-SS children, regardless of the presence of its typical clinical features.
In our study, we also had several less expected findings.GH is essential for normal growth, and children with GHD may have severe short stature [32].However, GH is generally considered to affect mainly the postnatal phase of growth, and children with GHD should therefore be born with normal birth parameters [32].In contrast with this concept, we have found causative genetic variants in genes affecting pituitary development or directly influencing GH production in 6/74 (8%) of SGA-SS children with genetic diagnosis in our study cohort.Some of them might have different causes of prenatal growth failure (e.g., patient no. 1 with LP GHSR variant whose mother suffered from HELLP syndrome during pregnancy); however, in other children, we found no additional explanation of prenatal growth impairment.On the other hand, other studies also have discovered genetic findings typical for GHD in SGA children [2].We can speculate that GH might play a role in prenatal growth in some children or genetic variants found might affect growth on other levels besides affecting GH.
Another interesting result was the pathogenic variant found in the gene THRA encoding thyroid hormone receptor type A, which has also been previously associated with short stature [33].In one patient, we diagnosed a homozygous variant in the gene TRHR, leading to central hypothyroidism.Since its first observation in 1997 [34], short stature has been recognised as one of the consistent features of central congenital hypothyroidism due to TRHR defects and their bi-allelic pathogenic variants.However, perinatal data were not displayed in patients published so far.Thus, we are adding the TRHR gene as a novel causative gene for SGA.These two findings in the thyroid axis increase the diagnostic yield by additional 3% and clearly show that pre-and postnatal growth is affected by the thyroid axis far beyond the classical hypothyroidism.The other variants found in our study were genes involved in the regulation of growth plates and other fundamental processes of intracellular signalling.
The clinical response to GH treatment has not been systematically studied in the sub-cohorts of SGA-SS with defined genetic aetiology, with the exception of children with SRS [35].Thus, the currently available reports have their origin in retrospective analysis of children genetically diagnosed at a late stage of their therapy.The currently available data are scarce, as summarised in the latest consensus [36].In our cohort, the long-term growth data are available only in a minority of children; therefore, we present shorttime growth data following 1 year and 3 years of GH administration and, when available, final height SDS.Continuing the observation of the study cohort might bring important new data on the impact on the individual of the new genetic finding.
Some genetic studies in short stature tend to suffer from selection bias as the study population originates mostly from tertiary centres [26,37,38].This questions the extrapolation of the genetic spectrum of growth disorders to the general population.The strength of our study is its population-based principle: all newborns in the Czech Republic have their birth weight and length measured and carefully recorded together with their gestational week.All children subsequently undergo regular, mandatory body height examinations that enable the identification of growth failure, an early referral to a paediatric endocrinologist, and the start of GH treatment if indicated.Our centre provides GH to about 30% of children in our country [39]; their selection depends mainly on their residence.Thus, our study should be relatively free of selection bias with the exception of those who neglected regular body height checkups (which is rare) or refused either GH therapy or genetic testing.
Our study had some limitations as well.Chromosomal analysis including microarrays was performed only in children whose phenotype led to the initial referral to the department of clinical genetics.Moreover, non-coding variants (with the exception of the disruption in exonintron boundaries), epigenetic, and somatic changes were not captured by NGS.Finally, our study lacks the functional studies to help evaluate the pathogenicity of the discovered variants.
To conclude, our study elucidated the genetic aetiology of 42% of SGA-SS children from a genetically relatively homogenous, non-consanguineous population.The results demonstrate a complex aetiology of short stature affecting all the three key levels of growth regulation including the endocrine system, growth plate function, and fundamental processes of intracellular regulation and signalling.A conclusive genetic finding not only provides a clear explanation of the growth disorder but also enables focussing on possible associated hidden comorbidities and genetic consulting [40].In our opinion, routine genetic testing may therefore become a standard of diagnostic care in resource-rich countries for all SGA-SS children after other causes of growth failure are ruled out.

( 8 )
Fig.2.Three levels of the genetic growth regulation in children born small for gestational age with persistent short stature (SGA-SS).The numbers in brackets in blue boxes show the numbers of patients identified with P/LP variants of the entire gene (if more than one).The numbers in closed circles refer to patient numbering in Table1.a Genes involved in cerebral midline and pituitary development, and in the GH-IGF-1 and thyroid axes.b Genes encoding growth plate matrix components and elements of chondrocyte paracrine regulation.c Genes involved in intracellular signalling, in the stability of nuclear membrane, and in the fundamental intranuclear processes.A brief description of selected patients' facial phenotypes: (8) OTX2: bilateral anophthalmia; (9) POU1F1: depressed nasal bridge and frontal bossing; (10) PROKR2: no apparent facial dysmorphism; (11) PTCH1: mild orbital hypotelorism, midface hypoplasia, and anteverted ears; (12) STAT3: no specific facial signs in a boy with severe immune dysregulation leading to early onset diabetes, hypothyroidism, cytopenia and lymphoproliferation, and short stature due to defective STAT3 signalling; (16) ACAN: mild facial dysmorphism similar to previously published cases; (30) FLNB: facial phenotype of atelosteogenesis type Iprominent forehead, depressed nasal bridge with a grooved tip, and micrognathia; (37) NPR2: the father and two daughters with vertical transmission of an NPR2 pathogenic variant, no facial phenotype; (40) KMT2D: Kabuki syndrome resembling the makeup in traditional Japanese theatre; (41) LMNA: gradually developing phenotype of Hutchinson-Gilford progeria syndrome; (43) PTPN11 and (47) SOS1: facial signs of Noonan syndrome; (48) SRCAP: typical face of Floating-Harbour syndrome.Image adapted from [25].

Table 1 .
Clinical findings in children born small for gestational age with persistent short stature (SGA-SS) with elucidated genetic diagnosis