Height and body mass index in molecularly confirmed Silver–Russell syndrome and the long‐term effects of growth hormone treatment

Abstract Objective Silver–Russell syndrome (SRS) causes short stature. Growth hormone (GH) treatment aims to increase adult height. However, data are limited on the long‐term outcomes of GH in patients with molecularly confirmed SRS. This study evaluated height, body mass index (BMI) and GH treatment in molecularly confirmed SRS. Design An observational study with retrospective data collection. Patients Individuals with molecularly confirmed SRS aged ≥13 years. Measurements Data were collected on height, height gain (change in height standard deviation score [SDS] from childhood to final or near‐final height), BMI and gain in BMI (from childhood to adulthood) and previous GH treatment. Results Seventy‐one individuals (40 female) were included. The median age was 22.0 years (range 13.2–69.7). The molecular diagnoses: H19/IGF2:IG‐DMR LOM in 80.3% (57/71); upd(7)mat in 16.9% (12/71) and IGF2 mutation in 2.8% (2/71). GH treatment occurred in 77.5% (55/71). Total height gain was greater in GH‐treated individuals (median 1.53 SDS vs. 0.53 SDS, p = .007), who were shorter at treatment initiation (−3.46 SDS vs. −2.91 SDS, p = .04) but reached comparable heights to GH‐untreated individuals (−2.22 SDS vs. −2.74 SDS, p = .7). In GH‐treated individuals, BMI SDS was lower at the most recent assessment (median −1.10 vs. 1.66, p = .002) with lower BMI gain (2.01 vs. 3.58, p = .006) despite similar early BMI SDS to GH‐untreated individuals (median −2.65 vs. −2.78, p = .3). Conclusions These results support the use of GH in SRS for increasing height SDS. GH treatment was associated with lower adult BMI which may reflect improved metabolic health even following discontinuation of therapy.


| INTRODUCTION
Silver-Russell syndrome (SRS) is a condition characterized by pre-and postnatal growth failure resulting in small-for-gestational-age (SGA) at birth, short stature, body asymmetry, relative macrocephaly at birth, a protruding or prominent forehead and feeding difficulties during childhood. The diagnosis of SRS can be made using the recently proposed Netchine-Harbison clinical scoring system (NHCSS). 1,2 In 50-60% of cases, loss of methylation at the intergenic H19/ IGF2 (H19/IGF2 LOM) differentially methylated region (DMR) at 11p15.5 has been identified. 3,4 In 5%-10% of cases maternal uniparental disomy of chromosome 7 (upd(7)mat) has been detected. 4,5 Mutations in CDKN1C, IGF2 and the PLAG1/HMGA2 pathway are sporadic causes of SRS. 2 Where there is a clinical diagnosis (i.e., at least the 4/6 of items of the NHCSS) without molecular confirmation and a differential diagnosis has been excluded, the term 'clinical SRS' is used. 2 There is considerable overlap with other imprinting disorders, such as Temple syndrome 1 and maternal uniparental disomy of chromosome 20. 1 Molecular genetic testing can therefore be useful to confirm a clinical diagnosis.
Children with SRS who are born SGA and remain short can be treated with growth hormone (GH) to increase adult height but there is variation in its use, nationally and internationally. Recent international consensus advocated early GH treatment in SRS for diminishing the risk of hypoglycemia, improving height and optimizing body composition. 2 GH treatment of SRS increases height velocity 6 and height standard deviation score (SDS). [7][8][9] Height gain was inversely related to height 8,10 and age 7,10 at treatment initiation. Height at the onset of puberty 8 and duration of treatment 6,9,10 were also positively related to height gain. A greater target height SDS positively affects the height gained. 9 GH treatment is associated with increased final height SDS, which also positively correlated with height at the start of treatment. [9][10][11] Two studies found that in GH-treated individuals, males reached a greater final height SDS than females. 8,10 However, all of these studies included individuals with both molecularly confirmed SRS and clinical SRS and many did not include details of treatment for bone age advancement during puberty, which untreated may compromise final height in SRS.
In association with feeding difficulties in SRS, an extremely lean appearance has been reported in clinical and molecularly confirmed SRS cases. 12 Evaluation of body mass index (BMI) has been reported in three studies of children with SRS: one study of molecularly and clinically diagnosed SRS reported a mean BMI SDS of −2.2 (SD 1.2). 4 Another study of molecularly confirmed SRS, which included GH treatment in 69%, reported an overall mean baseline BMI SDS of −2.4 (SD 0.8). 13 In the third study, examining the effect of appetite stimulation in molecularly and clinically diagnosed SRS, 8.8% had received GH and the median baseline BMI SDS was −2.8. 14 Some individuals were included in more than one study cohort, they reported BMI at a single time point or before and after a short-term intervention, and the effect of GH on BMI was not evaluated, but they demonstrate that BMIs in SRS are generally low in childhood.
There is less information in adults. In a recent case series of seven adults with molecularly confirmed SRS, the BMI SDS ranged from −2.8 to 2.5 (corresponding to BMI of 16.3-32.3 kg/m 2 ), providing some evidence that BMI increases considerably in adulthood in some individuals. Two of the seven (28.6%) had been treated with GH but there was no comparison of BMI between GH treatment groups. 15  at the start of GH treatment and increased both during treatment and 6 months after treatment discontinuation but subsequently stabilized. BMI SDS was not reported. 16 There is increasing interest in the long-term outcome of individuals with SRS in relation to height, 16 metabolic health, 9,15 'lived experience' 17 and recently the adult phenotype has been described. 18 We report the first study comparing long-term outcomes of GH treatment versus no treatment on height and BMI in a cohort of exclusively molecularly confirmed SRS cases.
Cohorts of individuals with molecularly confirmed SRS from the United Kingdom, France and Germany were identified. Data on the German cohort 10,19 and the UK cohort have previously been reported. 18

| Growth assessment
Height and weight measurements were documented at a single study visit (UK participants) or from case note review of the most recent follow-up appointment (participants from France and Germany).
Height was measured using a stadiometer (free-standing in the United Kingdom and France; wall-mounted in Germany). Measurements (including age, height and height SDS) and intervention for growth including GH, gonadotrophin-releasing hormone analogs (GnRHa) and aromatase inhibitors were obtained from medical records. Final height was defined as: (1) 23 and Rolland-Cachera et al. 24 for BMI; and, for the German cohort, Prader et al. 25 and Niklasson et al. 26 were used.
Where the age of the individual was greater than the upper age limit, the data for the maximum age available was used.
Early height and weight data were defined as before the start of GH treatment in the GH-treated group and as close to age 2-5 years as available in the GH-untreated group. These data were used to calculate early height SDS and early BMI SDS. Total height gain was calculated as the difference between current height SDS and early height SDS. Change in BMI SDS was calculated similarly as the difference between BMI SDS at the most recent evaluation and early BMI SDS. with single-nucleotide polymorphism microarray as previously described. 30

| Statistical analyses
The overall cohort was categorized on the basis of any prior GH treatment. Various characteristics of GH-untreated and GHtreated cases were compared, including early height SDS.
Mann-Whitney U tests were used to compare continuous variables between two groups (i.e., GH-treated vs. GHuntreated or GH pre-vs. post-age 4 years) and the Kruskal-Wallis test for comparison between three groups (i.e., United Kingdom, French and German cohorts or GH-/GnRHauntreated, GH-treated GnRHa-untreated and GH-/GnRHatreated). Associations between categorical variables were tested using Fisher's exact test. Statistical significance was initially set as p < .05. However, in line with recent discussion, p values were not considered purely dichotomously (i.e., significant vs. not significant). 31

| Effect of GH on height in the overall cohort
The characteristics of the GH-untreated and GH-treated groups are shown in Table 2. GH treatment was received in 77.5% for a median duration of 7.1 years (interquartile range [IQR]: 4.0-11.0). The median value of the mean GH dosage administered was 47.7 mcg/kg/ day (IQR: 34.9-56.0) or 1.57 mg/m 2 /day (IQR: 1.13-1.89). There was a suggestion of differences in GH dosages between countries (p = .08). The median age of starting GH treatment was 5.80 years (IQR: 4.5-9.0) and the median time since GH discontinuation was In the GH-treated group, those who started treatment before 4 years of age (n = 13), were shorter at the onset (median −5.26 SDS vs.  One individual in the French cohort had been treated with an aromatase inhibitor, however, details of treatment were not obtained.

| DISCUSSION
Our study is the first to compare growth outcomes in GH-treated versus GH-untreated molecularly confirmed SRS. Inclusion of only molecularly confirmed SRS cases is important as the clinical features of SRS overlap with other conditions and historical cohorts included those born SGA along with SRS. 11 Others have included SRS exclusively but included clinical SRS cases with different diagnostic criteria. [8][9][10] Previous findings may thus not truly reflect the growth outcomes of molecularly confirmed cases of SRS and as we move to an era of personalized medicine, epigenotype-phenotype correlations are increasingly relevant. We have, therefore, studied a group of older individuals with molecularly confirmed SRS and evaluated differences in height and BMI between those previously treated with GH and those untreated. Such a control group will be increasingly difficult to identify as GH becomes more widely used.
This study aimed to evaluate previous treatments to provide evidence to support health professionals' and families' decisions. Our novel findings show that in individuals with molecularly confirmed SRS, prior GH treatment is associated with greater height gain and lower BMI later in life. Greater adult height with GH treatment has been reported in SRS, including molecularly and clinically diagnosed SRS and including home height measurements. 10 In our study, GH treatment was associated with greater total height gain, providing further evidence in support of GH treatment to enhance height in SRS.
The GH-treated group was shorter in early life and may have been more severely affected by SRS. Shorter individuals may also be more likely to receive treatment. The final height attained in the GHtreatment group in our study (median SDS −2.22) was comparable to previous studies in GH-treated individuals. 9,10 However, those cohorts included patients with molecularly and clinically diagnosed SRS, whereas our study included exclusively molecularly confirmed SRS.
There was a negative association between age of onset of GH treatment and height gain, similar to a previous study, 10 suggesting that early treatment initiation should be considered in SRS. Some of the previously reported cases were included in the study reported here.
In our study, females with SRS attained a lower final height SDS than males, in agreement with previous results. 10 The reasons for this are unclear but it is possible that there is differential sensitivity of the growth plate to sex steroids in men and women with SRS. Sexdependent effects of estradiol have been shown on the mouse growth plate, 33 which would support this theory. There was no difference in the proportion of females and males treated to delay puberty. However, our study did not evaluate height or age at pubertal onset and completion, which would be interesting areas for further research.
We have demonstrated an association between prior GH treatment and lower BMI after treatment. Only 1.8% of the GHtreated group had a BMI SDS ≥ 2 compared with 25% in the untreated group. GH treatment in SGA has been shown to promote the development of lean mass and reduce fat mass during treatment, 34 and reduce fat mass but increase central fat distribution. 35 However, one group has shown increased fat mass SDS during and 2 years after GH treatment in SRS, which was similar to in individuals with non-SRS SGA 16 and another study showed a similar fat mass and fat distribution in adults who were born SGA both treated and untreated with GH. 36 As far we are aware, only one study has evaluated body composition exclusively in SRS but that study was limited to seven cases, the patients were younger than those presented here (mean age of 26.9 ± 9.1 years and range 18-46 years), GH had been received in 2/7 cases and treatment effects were not analyzed. 15  In conclusion, our data show that in individuals with molecularly confirmed SRS, prior GH treatment is associated with greater height gain and reduced BMI later in life (despite cessation of GH-treatment many years previously). Lower long-term BMI with GH treatment may, in turn, indicate an improved prognosis for metabolic health.