1 INTRODUCTION

Small for gestational age (SGA), defined as infants born with a weight and/or height that is two standard deviations (SDs) below the mean for their gestational age (at term or preterm),1 affects approximately 3%–10% of live births.2 Most children born SGA show catch-up growth during the first years of life, but 10% will continue with a pathologically short stature throughout childhood and adolescence.3

SGA syndrome affects glucose metabolism and insulin sensitivity (IS),4 which entails the subsequent risk of developing type-2 diabetes mellitus (DM) and other interrelated metabolic disorders, such as dyslipidemia, cardiovascular diseases and hypertension.5 SGA children with short stature show reduced levels of insulin-like growth factor-1 (IGF-1),3 a protein involved in foetal growth, development and metabolism regulation.5 IGF-1 levels are associated with IS, but this association seems to be complex6 and has not yet been well characterized. Several studies on children have shown cut-off values of insulin resistance (IR) between 2.5 and 3.2, according to the homeostatic model assessment of IR (HOMA-IR).7, 8

The only standard therapy approved for SGA syndrome is based on recombinant human growth hormone (rhGH), which leads to an increase in the growth rate and enables infants to grow according to the normal limits and have a normal adult height.9 The rhGH therapy has been shown to induce transient IR in children; therefore, there is a concern regarding the diabetogenic potential of rhGH therapy in children born with SGA.10 However, a recent review has revealed that, while rhGH treatment could pose a risk factor for the development of DM, family history could have more impact in its development.11

Although monitoring glucose homeostasis is recommended, there is no consensus on the appropriate method.10 A larger number of patients undergoing longer follow-up periods are needed to elucidate the relationship between IS patients and GH-treated SGA patients. Therefore, we carried out the first multicenter study in Spain to determine the long-term evolution of IR from the beginning of GH treatment in a larger population of SGA-children receiving rhGH treatment. The secondary endpoints of this study included the determination of the auxological and metabolic rhGH treatment effects to identify potential predictive factors and assess the treatment's safety profile.

2 METHODS 2.1 Study design and population

This prospective study was performed in 27 Spanish centres.

The patients included in the study were Spanish SGA-children (age ≥ 4 years) born at term and treated with rhGH (Saizen®; Merck-Serono. European authorization: September 2005)

All procedures performed during the study were in accordance with the Declaration of Helsinki and were approved by the Ethics Committee of Hospital Materno Infantil Carlos Haya of Málaga. The patients (≥12 years old) or their parents or legal representatives provided written informed consent.

The children were recruited between February 2007 and November 2012. Data from the patients' medical records were collected both prospectively and retrospectively, with the latter being collected at the start of the treatment and before the study authorisation (2005–2007).

The inclusion criteria were as follows: children with a current height < −2.5 SD; height adjusted to parental stature < −1 SD; born SGA at term (after Week 37 of gestation); weight and/or length at birth < −2 SD for their gestational age; aged ≥4 years old; and receiving rhGH treatment (daily dose: 0.035 mg/kg body weight; subcutaneously). The exclusion criteria were as follows: closed epiphysis, rhGH-hypersensitivity, active neoplasia, genetic or malformation syndromes and evidence of progression or recurrence of any underlying intracranial lesion.

The cessation treatment criteria were height velocity (HV) < 2 cm/year and bone age (BA) > 14 years in females and ≥15 years in males.

The sample size required for assessing the posttreatment safety profile (200 patients after 10 years of treatment, based on the SEPAGE study [NCT01082354] and our study [SER-GH-2005-01]) was determined based on the Saizen® Long Term Observational study (SALTO).12

The estimated sample size was 450 subjects, as calculated based on the SEPAGE (poor accrual and loss of 35% of patients who were to participate in SALTO) and SALTO (potential refusal of our patients to participate, and estimated loss of the participating patients; 12% during treatment and 35% during the follow-up) studies.

The safety population included all patients who received ≥1 rhGH dose. The evaluable population was defined as the safety population who had undergone ≥1-year of follow-up from the start of treatment.

Children were stratified according to the Tanner stage (Stage I or ≥I [II/III/IV, and adult]) and the age at treatment onset (6 years old [early start], or ≥6 years old [late start]). The second stratification was due to the differentiation and avoidance of children entering puberty during the first year of treatment, taking into account that some of them could have had advanced or precocious puberty. The stratification was established at 6 years of age to ensure a 3-year period before reaching puberty as the time for catch-up growth. Patients with Tanner stage >I were excluded from the analyses according to the start time of treatment to ensure the initial prepubertal situation.

2.2 Measurements; analysis/statistics

All variables were measured at least once a year, following the standard procedures of each centre. Only children with Tanner stage-I underwent analyses, thus ensuring their prepubertal state at the onset of treatment. Standardized (standard deviation score [SDS]) values were used to avoid variability in measurements among the centres.

The IR development/progression (primary endpoint) was calculated using the HOMA-IR index (mass units): fasting insulin (µu/ml) × fasting glucose (mmol/L)/22.5.

The auxological effects of treatment were assessed according to the following parameters: HV (cm/year), weight (kg) and height (cm) as SDS (chronological age [CA] and sex reference values obtained from Carrascosa),13 BA (assessed using the Greulich and Pyle atlas)14 and Tanner-stage (testicular size or breast development). The potential predictor value of body mass index (BMI) (SDS) for IR was calculated.

The metabolic effects of treatment were assessed according to the following parameters: fasting plasma IGF-I (ng/ml) (SDS) (reference values from Elmlinger et al.),15 plasma triglycerides (mg/dl), high-density lipoprotein cholesterol (mg/dl), HbA1c (% of total haemoglobin) and thyroid hormones. Assessments were carried out following different methods and at different local laboratories, and measures (reference values) were standardized for the comparative analyses.

Nonresponders to rhGH treatment were defined as a growth rate < +1 SDS during the first year of treatment.

Safety variables included adverse events (AEs; Medical Dictionary for Regulatory Activities), physical examination, vital signs and blood and urine tests.

Categorical variables are expressed as absolute and relative frequencies (%), while continuous variables are expressed as mean, SD, 95% confidence interval (95 % CI) and SDS. Continuous variables were analysed using the t-test or Wilcoxon test.

Regression analysis was performed on the direct endpoints. A correlation analysis was performed to determine the association between changes in HOMA-IR and HV (Spearman's correlation coefficient). Statistical significance was set at p ≤ .05. If the CI did not include the zero-effect value, it could be assumed that there was a statistically significant result. All statistical procedures were performed using the SAS 9.2 statistical software (SAS Institute).

3 RESULTS

At the time of the final analysis in October 2018, a total of 410 patients from 27 centres constituted the safety population, of which 393 were considered to be in the evaluable population (Table 1).

Table 1. Characteristics of the evaluable population (n = 393) Patients (N = 393) Male, n (%) 198 (50.4) Age, mean (years ± SD) At study inclusion 7.8 ± 3.1 At treatment initiation 7.2 ± 2.8 Gestational age at birth, mean (weeks ± SD) 37.6 ± 3.2 Missing (n) = 3 Genetic target height size, mean (cm ± SD); [SDS] 163.1 ± 7.9 Male ≤ 6 years 169.3 ± 4.8; [1.3] Female ≤ 6 years 156.9 ± 5.1; [−1.2] Male > 6 years and Tanner I 169.3 ± 4.2; [1.5] Female > 6 years and Tanner I 156.0 ± 4.3; [−1.6] Missing (n) = 29 Birth height, mean (cm ± SD) 43.6 ± 4.0 Missing (n) = 21 Birth weight, mean (kg ± SD) 2.2 ± 0.6 Missing (n) = 2 Relevant medical history at baseline, n (%) 50 (12.7) Past medical history 12 (24.0) Current ongoinga 38 (76.0) Mild–moderatea 35 (92.1) Familiar clinical history, n (%) (brother/sister diagnoses SGA) Missing (n) = 13 Yes 34 (9.0) No 277 (72.9) NA 56 (14.7) Abbreviations: NA, not applicable; SD, standard deviation; SDS, standard deviation score; SGA, small for gestational age. aCalculated on the number of patients with current relevant medical history at baseline.

The sample was sex-balanced (Table 1), with a mean age at treatment onset of 7.2 ± 2.8 years old.

Treatment was completed in 150 patients (38.17%). The reasons for discontinuation among the 200 (50.89%) patients with early withdrawal included the following: lack of efficacy (at the investigator's discretion), n = 15 (7.5%); inability to follow treatment, n = 10 (5.0%); AEs, n = 6 (3.0%); and administrative causes (follow-up data not reported), n = 137 (68.5%). Missing data occurred in 43 patients (10.94%). A total of 8.95% of the patients had an SGA sibling. Arterial hypertension and type-II DM were the most common familiar clinical histories (11.32% and 12.1%, respectively). Other types of familiar clinical history included cancer (10.0%), familial hyperlipidemia (9.47%), obesity (5.79%), gestational diabetes (2.37%), type-I DM (1.05%) and myocardial infarction or stroke before the age of 40 years (0.53% for each case).

3.1 Insulin resistance

The mean HOMA-IR increased significantly during the first year of treatment: overall, 0.62 ± 1.47 (Table 2; Figure 1A); Tanner-I ≤ 6 years old, 0.40 ± 0.8 and >6 years old, 0.80 ± 1.65 (Table 3); and Tanner-II, 0.99 ± 3.16 (Table 3). Overall, no significant differences were observed thereafter, although an increasing trend was observed until visit 6 (Figure 1A). The HOMA-IR values were maintained within normal ranges and were similar in both age groups (≤6 years old and >6 years old; Table 3). The increase was higher in children ≤6 years old because their values from baseline to Visit 3 were significantly lower than those in the >6 years old group. Nevertheless, the values were similar in both groups at Visit 6 (normal range).

Table 2. Annual evolution by the visit of HOMA-IR and auxologic and metabolic variables, up to the 10th year of rhGH-treatment N mean ± SD (95% CI) HOMA-IR (mass units) HV (cm/year) HV (SDS) Height (SDS) Weight (SDS) BMI (SDS) BA/CA Ratio IGF-1 (SDS) Basal 1.19 ± 1.14 (1.05, 1.32) – – −3.00 ± 0.61 (−3.06, −2.93) −1.72 ± 0.56 (−1.78, −1.67) −0.70 ± 0.80 (−0.78, −0.62) 0.71 ± 0.17 (0.70, 0.73) −0.31 ± 1.06 (−0.41, −0.20) V1

282

1.91 ± 2.58 (1.60, 2.21)

8.59 ± 2.33 (8.35, 8.83) 2.75 ± 2.39 (2.50, 2.99) −2.35 ± 0.71 (−2.42, −2.28) −1.47 ± 0.55 (−1.53, −1.42) −0.76 ± 0.70 (−0.83, −0.69) 0.79 ± 0.15 (0.78, 0.81) 0.99 ± 1.20 (0.87, 1.12) V2

235

2.09 ± 1.61 (1.88, 2.30)

6.80 ± 1.48 (6.64, 6.97) 1.09 ± 1.40 (0.94, 1.25) −2.02 ± 0.73 (−2.10, −1.94) −1.35 ± 0.60 (−1.42, 1.29) −0.77 ± 0.74 (-0.85 ± −0.69) 0.85 ± 0.14 (0.83, 0.87) 1.21 ± 1.21 (1.07, 1.35) V3

192

2.21 ± 1.40 (2.01, 2.41)

6.51 ± 1.66 (6.30, 6.72) 0.88 ± 1.22 (0.73, 1.04) −1.60 ± 2.74 (−1.93, −1.26) −1.06 ± 3.01 (−1.43, −0.69) −0.75 ± 0.70 (−0.84, −0.67) 0.89 ± 0.13 (0.88, 0.91) 1.37 ± 1.40 (1.20, 1.55) V4

149

2.40 ± 1.33 (2.19, 2.62)

5.97 ± 2.66 (5.59, 6.35) 0.82 ± 3.08 (0.38, 1.27) −1.59 ± 0.81 (−1.70, −1.47) −1.18 ± 0.58 (−1.26, −1.10) −0.76 ± 0.72 (−0.86, −0.66) 0.92 ± 0.11 (0.91, 0.94) 1.14 ± 1.35 (0.94, 1.35) V5

97

2.67 ± 2.28 (2.21, 3.13)

5.42 ± 4.26 (4.68, 6.16) 0.30 ± 45.40 (−0.64, 1.24) −1.66 ± −1.37 (−1.70, −1.47) −1.17 ± 0.63 (−1.28, −1.06) −0.77 ± 0.72 (−0.90, −0.65) 0.95 ± 0.10 (0.94,0.97) 1.09 ± 1.40 (0.84, 1.34) V6

61

2.74 ± 1.57 (2.34, 3.15)

5.42 ± 2.10 (4.95, 5.89) 0.68 ± 3.97 (−0.21, 1.57) −1.50 ± 0.88 (−1.68, −1.31) −1.22 ± 0.62 (−1.35, −1.08) −0.83 ± 0.67 (−0.97, −0.68) 0.95 ± 0.09 (0.93, 0.97) 1.00 ± 1.42 (0.67, 1.33) V7

38

2.39 ± 1.26 (1.98, 2.81)

5.67 ± 2.23 (5.05, 6.28) 0.33 ± 1.26 (−0.02, 0.68) −1.39 ± 0.88 (−1.62, −1.17) −1.14 ± 0.70 (−1.32, −0.97) −0.76 ± 0.72 (−0.95, −0.58) 0.96 ± 0.09 (0.94, 0.99) 0.86 ± 1.23 (0.51, 1.21) V8

31

2.37 ± 1.40 (2.21, 3.24)

5.54 ± 2.82 (4.54, 6.39) 0.39 ± 1.67 (−0.16, 0.94) −1.38 ± 0.79 (−1.62, −1.15) −1.25 ± 0.55 (−1.42, −1.09) −0.89 ± 0.60 (−1.07, −0.71) 0.97±0.08 (0.94,1.00) 0.79 ± 1.05 (0.45, 1.12) V9

14

2.46 ± 0.98 1.89, 3.032

4.38 ± 2.97 (3.18, 5.58) −0.01 ± 1.24 (−0.51, 0.49) −1.25 ± 0.73 (−1.55, 0.96) −1.19 ± 0.62 (−1.44, −0.94) −0.84 ± 0.72 (−1.13, −0.55 0.97 ± 0.08 (0.94, 1.00) 0.60 ± 0.57 (0.37, 0.84) V10

9

2.11 ± 0.51 1.72, 2.50

4.44 ± 3.01 (2.29, 6.60) 0.44 ± 1.17 (−0.39, 1.27) −1.13 ± 0.64 (−1.54, −0.72) −1.00 ± 0.79 (−1.50, −0.50) −0.64 ± 0.88 (−1.20, −0.08) 0.97 ± 0.06 (0.92, 1.02) 0.64 ± 0.54 (0.26, 1.03) Note: Evaluable population (n = 393). aThe bold values are intended to indicate the N of patients. Abbreviations: BA, bone age; BMI, body mass index; CA, chronological age; CI, confidence interval; HOMA-IR, homeostatic model assessment of insulin resistance; HV, height velocity; IGF-I, insulin-like growth factor-I; rhGH, recombinant human growth hormone; SD, standard deviation; SDS, standard deviation score. aThe difference between the 393 patient (evaluable population size) and the number of patients with data in each visit are ‘missing’ patients. It includes those patients that had complete/discontinued the treatment. (A) HOMA-IR index by visit (from baseline to Visit 10) and by age at start of treatment—(Tanner I and Tanner II) (evaluable population). (B) IGF index by visit (from baseline to Visit 10) and by age at start of treatment (evaluable population). (C) htSDS index by visit (from baseline to Visit 10) and by age at start of treatment (evaluable population). HOMA-IR, homeostatic model assessment of insulin resistance; IGF, insulin-like growth factor [Color figure can be viewed at wileyonlinelibrary.com] Table 3. Change from baseline to HOMA-IR and height measured every year versus previous year by age and Tanner stage at start of treatment Change HOMA-IR Change height N Early start (≤6 years) Late start (>6 years) Mean ± SD (95% CI) Tanner I Tanner I Tanner II Change in HSDS Gain in HSDS V1 − basal 0.40 ± 0.84 (0.23, 0.57)

100

0.80 ± 1.65 (0.47, 1.12)

11

0.99 ± 3.16 (−1.13, 3.12)

0.65 (0.48) (0.60, 0.70)

387

3.53 (2.75) (3.26, 3.81)

V2 − V1

76

0.41 ± 1.28 (0.12, 0.70)

98

0.07 ± 2.00 (−0.33, 0.47)

10

−0.02 ± 1.21 (−0.89, 0.84)

0.30 (0.34) (0.27, 0.34)

323

1.77 (2.06) (1.54, 2.00)

V3 − V2

63

0.15 ± 1.30 (−0.18, 0.47)

85

0.03 ± 1.66 (−0.33, 0.38)

8

−0.05 ± 0.76 (−0.69, 0.58)

0.42 (2.67) (0.09, 0.75)

256

1.88 (4.88) (1.28, 2.48)

V4 − V3

41

0.31 ± 1.32 (−0.11, 0.73)

71

0.04 ± 1.45 (−0.31, 0.38)

2

−0.08 ± 1.16 (−10.52, 10.35)

0.18 (0.43) (0.12, 0.24)

195

1.08 (2.62) (0.70, 1.45)

V5 − V4

27

0.02 ± 1.16 (−0.44, 0.48)

54

0.66 ± 2.56 (−0.04, 1.36)

1

0.07 (−)

0.12 (0.51) (0.03, 0.21)

131

0.87 (3.20) (0.32, 1.42)

V6 − V5

15

0.54 ± 1.80 (−0.46, 1.54)

33

0.11 ± 1.35 (−0.37, 0.59

0 0.09 (0.30) (0.03, 0.16)

86

0.68 (2.09) (0.23, 1.13)

V7 − V6

11

−0.27 ± 2.39 (−1.88, 1.33)

21

−0.03 ± 1.36 (−0.65, 0.59)

0 0.10 (0.34) (0.02, 0.19)

60

0.75 (2.41) (0.13, 1.37)

V8 − V7

10

0.63 ± 1.25 (−0.27, 1.52)

12

0.05 ± 0.96 (−0.56, 0.66)

0 0.08 (0.40) (−0.04, 0.20)

44

0.56 (2.95) (−0.34, 1.46)

V9 − V8

6

0.17 ± 1.06 (−.94, 1.28)

4

−0.61 ± 2.49 (−4.57, 3.34)

0 0.02 (0.28) (−0.09, 0.14)

25

0.22 (2.13) (−0.67, 1.10)

V10 − V9

5

−0.38 ± 0.97 (−1.58, 0.82)

2

−1.62 ± 0.07 (−2.27, −0.98)

0 0.08 (0.27) (−0.10, 0.26)

11

0.59 (1.99) (−0.74, 1.93)

Note: Descriptive statistics. Evaluable population. The bold values are intended to indicate the N of patients. Abbreviations: CI, confidence interval; HOMA-IR, homeostatic model assessment of insulin resistance; HSDS, height standard deviation score; SD, standard deviation.

The changes in height velocity, height and weight, BMI, ratio of BA and CA, and IGF-1 can be seen in Tables 2 and 3 and in Figure 1B,C.

3.2 HOMA-IR relations

No relationship was found between the HOMA-IR index and HV, except for the second year (Spearman's correlation coefficient = 0.19; p < .05).

3.3 Multiple correlation analysis

The HOMA-IR values increased proportionally with baseline age and BMI, and inversely with baseline HOMA-IR and weight values. These four variables explained 21% of the change in the HOMA-IR values (Table 4).

Table 4. Factors influencing the change of HOMA-IR Parameter Estimate Standard error t Pr > |t|  Intercept −5.1530 1.2605 −4.09 <.0001 Basal insuline resistance (HOMA-IR) −0.5003 0.1034 −4.84 <.0001 Age at start of treatment 0.5659 0.1329 4.26 <.0001 Basal weight −0.2170 0.0646 −3.36 0.0009 Basal BMI 0.4066 0.0987 4.12 <.0001 R2 0.2138 Note: Multiple linear regression model. Evaluable population. Abbreviations: BMI, body mass index; HOMA-IR, homeostatic model assessment of insulin resistance. 3.4 Metabolic parameters

No significant changes in plasma triglycerides, high-density lipoprotein cholesterol, HbA1c and thyroid hormones were observed during rhGH treatment.

3.5 Safety

A total of 411 patients were included in the safety population, of which 16 (3.9%) reported AEs. In 14 of them, the event was due to the treatment. Furthermore, six (2.9%) patients discontinued treatment because of AEs. The AEs were musculoskeletal and connective tissue disorders (back pain, osteochondrosis, osteonecrosis and scoliosis), which were considered to be moderate, except for one mild AE (osteochondrosis; Table 5).

Table 5. Treatment emergence adverse events according to organ/system classification and preferred terms Adverse event Events (n) Severity Serious Related to treatment Congenital, familial and genetic disorders Congenital hypothyroidism 1 Mild No Not related Endocrine disorders Autoimmune thyroiditis 1 Mild No Not related General disorders and administration site conditions Oedema peripheral 1 Mild No Unlikely Hepatobiliary disorders Cholelithiasis 1 Mild No Not related Hypertransaminasaemia 1 Mild No Unlikely Investigations IGF 1 Mild No Probable IGF increase 4 Mild No Probable Metabolism and nutrition disorders Hyperglycaemia 1 Moderate Yes Possible Type 2 diabetes mellitus 1 Moderate Yes Probable Musculoskeletal and connective tissue disorders Back pain 1 Moderate No Probable Osteochondrosis 2 Mild/moderate No Unlikely Osteonecrosis 1 Moderate Yes Possible Scoliosis 1 Moderate