Role of Imprinting Disorders in Short Children Born SGA and Silver-Russell Syndrome Spectrum

Abstract Background (Epi)genetic disorders associated with small-for-gestational-age with short stature (SGA-SS) include imprinting disorders (IDs). Silver-Russell syndrome (SRS) is a representative ID in SGA-SS and has heterogenous (epi)genetic causes. Subjects and Methods To clarify the contribution of IDs to SGA-SS and the molecular and phenotypic spectrum of SRS, we recruited 269 patients with SGA-SS, consisting of 103 and 166 patients referred to us for genetic testing for SGA-SS and SRS, respectively. After excluding 20 patients with structural abnormalities detected by comparative genomic hybridization analysis using catalog array, 249 patients were classified into 3 subgroups based on the Netchine-Harbison clinical scoring system (NH-CSS), SRS diagnostic criteria. We screened various IDs by methylation analysis for differentially methylated regions (DMRs) related to known IDs. We also performed clinical analysis. Results These 249 patients with SGA-SS were classified into the “SRS-compatible group” (n = 148), the “non-SRS with normocephaly or relative macrocephaly at birth group” (non-SRS group) (n = 94), or the “non-SRS with relative microcephaly at birth group” (non-SRS with microcephaly group) (n = 7). The 44.6% of patients in the “SRS-compatible group,” 21.3% of patients in the “non-SRS group,” and 14.3% in the “non-SRS with microcephaly group” had various IDs. Loss of methylation of the H19/IGF2:intergenic-DMR and uniparental disomy chromosome 7, being major genetic causes of SRS, was detected in 30.4% of patients in the “SRS-compatible group” and in 13.8% of patients in the “non-SRS group.” Conclusion We clarified the contribution of IDs as (epi)genetic causes of SGA-SS and the molecular and phenotypic spectrum of SRS. Various IDs constitute underlying factors for SGA-SS, including SRS.

Approximately 10% of babies born small-for-gestationalage (SGA), a condition for babies with birth weight (BW) and/or birth length (BL) less than those expected for the gestational age and sex, do not have catch-up growth until 2 years of age (1,2). These children are categorized as short children born SGA (SGA-SS). SGA is mainly defined by the following criteria: (i) below the 10th percentile for both BW and BL (ICD10, ICD10data.com), and (ii) below −2 standard deviation score (SDS) for BW and/or BL for the gestational age and sex of the population (3)(4)(5). The causes of SGA are multifactorial, including environmental, maternal, paternal, placental, and fetal factors (1). Among them, multiple (epi)genetic disorders, which are considered as fetal factors, could lead to SGA-SS.
SRS is a representative ID associated with SGA-SS, and the diagnosis of SRS is based on clinical diagnostic criteria. The Netchine-Harbison clinical scoring system (NH-CSS) is the clinical diagnostic criteria for SRS having the most sensitive and the highest negative predictive value (12). A consensus statement recommends using NH-CSS as the clinical diagnostic criteria for SRS (13). Major genetic causes of SRS are loss of methylation of the H19/IGF2:intergenic (IG)-DMR (H19LOM) and maternal uniparental disomy chromosome 7 (UPD(7)mat) (7,13); however, some patients with H19LOM and UPD(7)mat did not satisfy NH-CSS criteria, and were diagnosed as SGA-SS (12). Moreover, some patients with etiologies other than H19LOM and UPD(7)mat met NH-CSS criteria and were diagnosed as SRS (12,14).
Previous work has examined the contribution of IDs to SGA-SS; however, the number of patients involved was not so large, and clinical assessment using clinical diagnostic criteria for SRS was not performed (15). To clarify the contribution of IDs to SGA-SS and the spectrum of molecular and phenotypic SRS, we carried out molecular and clinical analyses for 249 patients with SGA-SS. These patients were referred to us for genetic testing for SRS and SGA-SS. The following is a large comprehensive study investigating the etiologies related to IDs and the clinical features in patients with SGA-SS who were evaluated based on NH-CSS.

Ethics approval and consent to participate
This study was approved by the Institutional Review Board Committee at the National Center for Child Health and Development and performed after obtaining written informed consent from patients or patients' parents to participate in this study and for publication of the clinical and molecular information.

Patients
From 2002 to 2019 we recruited 220 patients referred to us for investigation for SGA-SS and 431 patients referred to us for genetic testing for SRS (Fig. 1). Among them, 269 patients met inclusion criteria (Fig. 1). Subsequently, we excluded 20 patients with structural abnormalities detected by array comparative genomic hybridization (aCGH) using catalog array. Finally, we screened various IDs in 249 patients with SGA-SS, including 92 patients referred to us for investigation for SGA-SS and 157 patients referred to us for genetic testing for SRS. Clinical features of patients were evaluated by the attending physicians, consisting of general pediatricians, neonatologists, pediatric endocrinologists, and pediatric geneticists. We collected the clinical information from each attending physician using a questionnaire. When we required more clinical data related to NH-CSS and/or specific etiology in some patients, we contacted their attending physicians again. Inclusion criteria for this study are shown in Fig. 1. We assessed their BW, BL, and height at 24 months for evaluating whether they had SGA-SS. If we could not obtain height at 24 months, we evaluated the height at the closest age over 24 months before initiating growth hormone treatment. We included patients in this study with both their BW and BL less than the 10th percentile of those expected for the gestational age and sex and their height at 24 months less than −2 SDS. We classified these patients into 3 subgroups based on NH-CSS consisting of 6 key items, as follows: (i) SGA, BW and/or BL are ≤ −2 SDS for gestational age; (ii) postnatal growth failure, height at 24 ± 1 months ≤ −2 SDS or height ≤ −2 SDS below mid-parental target height; (iii) relative macrocephaly at birth, ie, SDS for BW or BL minus SDS for birth occipitofrontal circumference (BOFC) ≤ −1.5; (iv) prominent forehead during early childhood; (v) body asymmetry; and (vi) feeding difficulties (current use of a feeding tube or cyproheptadine for appetite stimulation) during early childhood and/or low body mass index (BMI) ≤ −2 SDS at 24 months. The 249 patients with SGA-SS were classified into 3 subgroups. Patients with 4 or more NH-CSS items were classified into the "SRS-compatible group." Patients with less than 4 NH-CSS items with normocephaly or relative macrocephaly at birth were classified into the "non-SRS with normocephaly or relative macrocephaly at birth group (non-SRS group)," and patients with less than 4 NH-CSS items with relative microcephaly at birth were classified into the "non-SRS with relative microcephaly at birth group (non-SRS with microcephaly group)." Based on the previous study, relative microcephaly was defined as the following: head circumference SDS below height SDS and weight SDS at birth (13).

Molecular analysis
Genomic DNA of patients was extracted from leukocytes of peripheral blood. First, we performed aCGH analysis using a 60K catalog array (catalog number G4827A, Agilent Technologies, Palo Alto, CA) to detect PCNVs. To access the PCNVs, we found copy number variants (CNV) that had 3 or more continuous probes with aberrant signals. Then we evaluated whether these CNVs were pathogenic or not by referring to the Database of Genomic Variants (http://dgv.tcag.ca/dgv/app/home) and previous reports. Eventually, when CNVs were detected in our study involved in the regions causing known microdeletion or microduplication syndromes, we defined those CNVs as PCNVs. Next, to detect IDs, we performed methylation analysis with pyrosequencing for 9 DMRs related to known IDs, namely, the H19/IGF2:IG-DMR on chromosome 11,  (7,8). Furthermore, to detect UPD(16)mat, we performed methylation analysis for the ZNF597:TSS-DMR on chromosome 16 (10). Patients with hypomethylation of the H19/IGF2:IG-DMR were diagnosed as H19LOM. For patients with hypermethylation of both the PEG10:TSS-DMR and MEST:alt-TSS-DMR, we performed microsatellite analysis for chromosome 7 in patients and their parents, and confirmed UPD(7)mat (7,16). For patients with the abnormal methylation levels of the DMR(s) related to IDs other than H19LOM and UPD(7)mat, such as TS14, PWS, UPD(6)mat, UPD(16)mat, and UPD(20)mat, we conducted structural analysis to detect micro-deletions and/or duplications involving the imprinted regions. For structural analysis, we used methylation-specific multiplex ligationdependent probe amplification (MS-MLPA, MRC Holland, Amsterdam, Netherlands) analysis for each imprinted region (ME030 for 11p15.5, ME028 for 15q11.3-5, ME031 for 20q13) and/or aCGH analysis using custom built array (Design ID 080262, Agilent Technologies, Palo Alto, CA) and/or genome-wide comparative genomic hybridization and single-nucleotide polymorphism array (catalog number G4890A, Agilent Technologies, Palo Alto, CA), according to the manufacturer's instructions. For patients with abnormal methylation levels of the DMRs related to known IDs other than H19LOM and UPD(7)mat, but not structural abnormalities, we carried out microsatellite analysis for chromosomes 6 (16), 14 (8), 15 (17), 16 (10), and 20 (11) using patients' and their parental genomic DNA samples. If patients had aberrant methylated DMR(s) other than their disease-related DMR(s), we considered these patients as having multilocus imprinting disturbance (MLID).

Molecular analysis
The results of molecular analyses are summarized in Fig. 2 (see also Table 1). Of 249 patients with SGA-SS, H19LOM and UPD(7)mat, which are major genetic causes of SRS,  were detected in 58 patients (23.3%), and IDs other than H19LOM and UPD(7)mat were in 29 patients (11.6%). Twenty patients with IDs other than H19LOM and UPD(7) mat were previously reported (see Table 1 and Table 2) (8,10,11,(18)(19)(20). These 249 patients with SGA-SS were classified into the SRS-compatible group (n = 148), non-SRS group (n = 94), and non-SRS with microcephaly group (n = 7). H19LOM and UPD(7)mat were identified in 30.4%, 13.8%, and none of patients in the SRS-compatible group, the non-SRS group, and the non-SRS with microcephaly group, respectively. IDs other than H19LOM and UPD (7) A single patient showed abnormal methylation levels in all examined DMRs due to parthenogenesis.

Clinical characteristics for each etiology
Clinical data for each subgroup in patients with H19LOM, UPD(7)mat, IDs other than H19LOM and UPD(7)mat, and unknown etiology are summarized in Table 3. A total of 157 patients out of 249 patients were suspected to have SRS, and the remaining patients were diagnosed with SGA-SS ( Fig. 1, Table 3). Fifty-two patients with H19LOM and UPD(7)mat, and 20 patients with IDs other than H19LOM and UPD(7)mat had clinically suspected SRS. Clinical manifestations in patients with each etiology are summarized in Table 3. Phenotypic comparison between H19LOM and UPD (7)mat showed that patients with H19LOM had typical SRS phenotypes, such as severe prenatal and postnatal growth failure and higher frequency of the remaining clinical features in NH-CSS. BOFC of patients with H19LOM were more preserved than those of patients with other etiologies. Patients with UPD(7)mat showed less serious prenatal growth failure compared to patients with H19LOM, showed severe postnatal growth failure similar to patients with H19LOM, and had a high frequency of relative macrocephaly at birth (Table 3). Patients with IDs other than H19LOM and UPD(7)mat showed feeding difficulties and/ or low BMI in over 70%. Of clinical features frequently observed in SRS patients, triangular face was observed in approximately 90% of patients with H19LOM and in 70% of patients with UPD(7)mat. Fifth finger clinodactyly was detected in more than 70% of patients with H19LOM and in about 45% of patients with IDs other than H19LOM and UPD(7)mat. Muscular hypotonia was observed at relatively high frequency in patients with IDs other than H19LOM and UPD (7)mat, compared to those with other etiologies. Speech delay was detected in 40% and 58.6% of patients with UPD(7)mat and IDs other than H19LOM and UPD (7) mat, respectively. Congenital heart diseases (CHDs) were observed in 16% of patients with IDs other than H19LOM and UPD (7)mat. Genitourinary anomalies were observed in approximately 30% of patients with H19LOM.
All 3 patients with MLID were classified into the SRScompatible group. Thus, the frequency of MLID was 2.0% of patients in the SRS-compatible group and 4.5% of patients with various IDs in the SRS-compatible group. MLID 1 had 5 NH-CSS items other than body asymmetry, and also showed triangular face, fifth finger clinodactyly, cryptorchidism, hypospadias, and pelvic bone defect. His BW, BL, and present height were −4.2, −3.8, and −5.8 SDS, respectively. MLID 2 had 5 NH-CSS items other than feeding difficulties together with fifth finger clinodactyly. Her BW, BL, and present height were −3.5, −3.4, and −3.3     a Birth weight, length, and head circumference were evaluated by sex-matched and gestational-age-matched Japanese reference data (http://jspe.umin.jp/medical/chart_dl.html). We adopted "head circumference SDS below height and weight SDS" as the definition of relative microcephaly based on the consensus statement (13).
c If we did not get information at 24 ± 1 months, we used the data at the nearest measure available older than 25 months before initiating growth hormone treatment.
e Head circumference at birth ≥ 1.5 SDS above birth length and/or weight SDS. Head circumference at birth was evaluated by sex-matched and gestational-age-matched Japanese reference data (http://jspe.umin.jp/medical/ chart_dl.html).
a Birth weight, length, and occipitofrontal circumference at birth were evaluated by sex-matched and gestational-age-matched Japanese reference data (http://jspe.umin.jp/medical/chart_dl.html). b Postnatal height, BMI, and weight were evaluated by sex-matched and age-matched Japanese reference data (http://jspe.umin.jp/medical/chart_dl.html). c If we did not get information at 24 ± 1 months, we used the data at the nearest measure available older than 25 months before initiating growth hormone treatment.

Clinical features of patients with IDs other than H19LOM and UPD(7)mat
We summarized clinical features of 29 patients with IDs other than H19LOM and UPD(7)mat in Table 2. Of 29 patients, 21 patients were classified into the SRS-compatible group, 7 patients were classified into the non-SRS group, and 1 patient was classified into the non-SRS with microcephaly. BOFC was relatively preserved in patients with UPD(16)mat and lower in patients with PWS than those with other etiologies. TS14, which was most frequently detected in IDs other than H19LOM and UPD (7)

Discussion
In this study, to clarify the contribution of IDs to SGA-SS and the molecular and phenotypic spectrum of SRS, we performed methylation analysis for the DMRs related to known IDs in 249 patients with SGA-SS, including 92 patients referred to us for investigation for SGA-SS and 157 patients referred to us for genetic testing for SRS who were classified into 3 subgroups based on NH-CSS. A total of 30.4% of patients in the SRS-compatible group, 13.8% in the non-SRS group, and none of patients in the non-SRS with microcephaly group had major genetic causes of SRS such as H19LOM and UPD (7)mat. These results suggest that H19LOM and UPD(7)mat do not contribute to the development of SGA-SS with relative microcephaly. The frequency of H19LOM and UPD(7)mat in the SRScompatible group in our study was lower than those (76.7%) in a previous study, which examined 60 patients with 4 or more NH-CSS items (12). Interestingly, IDs other than H19LOM and UPD (7) Clinical analysis among patients with H19LOM, UPD(7) mat, IDs other than H19LOM and UPD(7)mat, and unknown etiology produced several notable clinical findings. First, 21 patients satisfying NH-CSS had etiologies other than H19LOM and UPD(7)mat, and meanwhile, 13 patients with H19LOM and UPD(7)mat did not satisfy NH-CSS. Consistent with this, Azzi et al reported a single patient with UPD(7)mat who did not satisfy NH-CSS (12), and Eggermann et al reported that H19LOM and UPD(7) mat carriers do not always show the unambiguous SRS phenotype (21). The diagnosis of SRS based on NH-CSS items consisting of nonspecific clinical features may result in a wide molecular spectrum of SRS. With regard to patients with H19LOM and UPD(7)mat having SGA-SS without typical SRS phenotype, some other factors such as environmental, maternal, paternal, placental, and other genetic factors may affect their phenotypes. Second, most patients with H19LOM, two-thirds of patients with UPD(7)mat, and two-thirds of patients patients with IDs other than H19LOM and UPD(7)mat satisfied NH-CSS. Patients with H19LOM showed the most severe prenatal and postnatal growth failure. In addition, relative macrocephaly, protruding forehead, and body asymmetry were frequently detected in patients with H19LOM. Patients with UPD(7)mat had heavier BW and taller BL than patients with H19LOM; however, patients with UPD(7)mat showed severe postnatal growth failure as well as those with H19LOM. Furthermore, 90% of patients with UPD(7)mat had relative macrocephaly, whereas only 9% had body asymmetry. These results indicate the differences in clinical features related to NH-CSS between patients with H19LOM and patients with UPD(7)mat, as previously reported (7,12). Third, regarding clinical features other than NH-CSS items, muscular hypotonia was found in about 40% of patients with IDs other than H19LOM and UPD (7)mat. This is probably because patients with TS14 and PWS, who have muscular hypotonia with high frequency (8,9), occupied half of the patients with IDs other than H19LOM and UPD(7)mat. Speech delay was observed in higher frequency in patients with UPD (7) mat, and those with IDs other than H19LOM and UPD (7) mat, than in those with H19LOM. In this regard, abnormal gene expression of FOXP2 on chromosome 7 associated with language development and dysfunction of the genes related to language development caused by IDs may lead to speech delay (22,23). CHDs were detected in 16% of IDs other than H19LOM and UPD (7)mat. Previous studies also showed that some DNA methylation abnormalities were associated with CHD (24,25). For SGA-SS patients with CHD, we need to consider the examination of IDs other than H19LOM and UPD (7)mat. Genitourinary anomalies were observed in about 30% of patients with H19LOM. Several studies have reported that male genitourinary anomalies were frequently identified in SRS (13,26,27), and lower birth weight was associated with urogenital anomalies (28). However, it remains to be clarified whether SGA caused by SRS or genetic causes of SRS result in male genitourinary anomalies; our result shows the possibility that H19LOM itself leads to male genitourinary anomalies.
Comprehensive methylation analysis for DMRs identified patients with various IDs. Patients with UPD(11) mat mosaic or 11p15 maternal duplication showed SRS phenotype as well as previous reports (29)(30)(31). UPD (11) mat mosaic results in decreased gene expression of IGF2 that functions as a growth promoter, increased gene expression of CDKN1C that functions as a negative regulator of cell proliferation (32,33), and 11p15 maternal duplication that results in increased CDKN1C gene expression (33). Consistent with this, gain-of-function gene mutation in CDKN1C was reported in patients with SRS (34). Over 70% of patients with TS14, UPD(20)mat, and UPD(16)mat were classified into the SRS-compatible group, as well as in the previous report (35). On the other hand, 1 out of 3 patients with UPD(6)mat in our study met NH-CSS criteria. UPD(6)mat is a rare ID with only 29 cases reported (http://molbiol.sci.am/ssmc/ssmc-tl.com/upd.html, accessed Jan 1, 2020). IUGR has been reported as one of the most common features of UPD(6)mat; however, it is unknown whether SS and the remaining NH-CSS items are common features or not due to lack of clinical information about patients with UPD(6)mat (36). In our study, speech delay was not observed, but genitourinary anomalies were detected in 2 patients with UPD(6)mat. Consistent with this, previous studies reported some patients having abnormal genitalia and mild speech delay (37,38). Further studies are required to clarify the clinical features of patients with UPD(6)mat.
In our study, the frequency of MLID was 3.4% in the patients with various IDs, and 4.5% in patients with various IDs classified into the SRS-compatible group. Recently, Eggermann et al reported that the frequency of MLID was 3.6% in SRS patients with aberrant findings (39). These findings suggest that MLID in patients with SRS is less frequent. All patients with MLID in our study fulfilled NH-CSS criteria (5-6 out of 6 items) and showed prenatal and postnatal growth failure. Only a single patient had additional congenital abnormalities, namely, cryptorchidism, hypospadias, and pelvic bone defect. Although Pooles et al reported that patients with MLID more frequently exhibit growth delay and additional congenital abnormalities (40), other reports showed that there were no significant differences in the clinical findings between SRS patients with MLID and "isolated" SRS patients (41,42). Our results were partially consistent with previous reports, specifically, severe postnatal growth failure in 2 patients. To clarify the clinical features of the patients with MLID, further accumulation of cases with MLID are required.
This study has some limitations. First, because our laboratory is one of the largest facilities that can perform genetic analyses of SRS in Japan, some selection bias for recruiting patients may occur, that is, patients with SGA-SS clinically diagnosed by attending physicians have some aspect of SRS phenotype. Second, many patients included in our study were preschool children under 3 years old. It is possible that developmental delay and characteristic features of each etiology were unclear at the time of examination. Lastly, we did not perform mutation screening of genes that cause genetic syndromes leading to SGA-SS. Patients with unknown etiology in our study may have had gene mutations of genetic syndromes resulting in SGA-SS.
In conclusion, we clarified the contribution of IDs to SGA-SS. This study also showed broad molecular and phenotypic spectrums of SRS. Various IDs constitute underlying factors for SGA-SS, including SRS.