Investigating whether serum IGF-1 and IGFBP-3 levels reflect the height outcome in prepubertal children upon rhGH therapy: LG growth study database

Serum insulin-like growth factor-1 (IGF-I) and IGF binding protein-3 (IGFBP-3) levels can be used to monitor the safety of recombinant human growth hormone (rhGH) therapy. In this study, we evaluated the changes in serum IGF-I and IGFBP-3 levels during rhGH therapy as a marker of height outcome in prepubertal children. Totally, 705 prepubertal children with short stature were enrolled from the LG Growth Study Database. Data for three groups of subjects were obtained as follows: Idiopathic GH deficiency (IGHD; n = 486); idiopathic short stature (n = 66); small for gestational age (n = 153). Serum IGF-I and IGFBP-3 levels at the baseline and after the 1st and 2nd year of rhGH therapy, as well as the Δheight standard deviation score (SDS), were obtained. Δheight SDS after the 1st and 2nd year of rhGH therapy had notably increased compared to that at the baseline for all three groups. IGF-I and IGFBP-3 levels in all three groups were significantly increased compared to those at the baseline (p <0.001). Δheight SDS was positively correlated with ΔIGF-1 SDS after the 1st year of therapy, ΔIGFBP-3 SDS after the 2nd year of therapy in the IGHD group, and ΔIGF-I SDS and ΔIGFBP-3 SDS after the 2nd year of therapy (p < 0.05), regardless of whether the height at the baseline was a covariate. The increase in IGF-I and IGFBP-3 levels during rhGH therapy was related to the growth response in children with IGHD. Therefore, it may be valuable to measure the change in serum IGF-I and IGFBP-3 levels, especially the latter, during rhGH treatment to predict the growth response upon long-term treatment.


Introduction
Recombinant human growth hormone (rhGH) therapy is widely used for short stature due to various causes, such as chromosomal/genetic abnormalities, small for gestational age (SGA), idiopathic short stature (ISS), and idiopathic GH deficiency (IGHD) [1]. Depending

Patients
This study included data from the LG Growth Study (LGS) [15].
LGS is a multi-center, openlabel, and non-interventional study. Clinical and laboratory records of 3199 prepubertal children with IGHD (n = 2195), ISS (n = 523), and SGA (n = 481), who had received rhGH therapy (Eutropin 1 inj., and Eutropin 1 Pen inj., LG Chem Ltd., Korea) for � 2 years between February 2001 and February 2020, were initially enrolled. The inclusion criteria were as follows: 1) Height below the third percentile according to the 2017 Korean National Growth Charts [16] at the time of hospital visit for IGHD and ISS cases, and at birth and at the time of hospital visit for SGA cases; 2) Two separate GH stimulation test results using insulin, clonidine, L-arginine, L-dopa, or glucagon; 3) children with a peak GH level less than 5 ng/mL (complete IGHD, CGHD) and between 5 to 10 ng/mL (partial IGHD, PGHD) [17][18][19][20][21] were classified as growth hormone deficient and those with that more than 10 ng/mL as ISS and SGA. The exclusion criteria were as follows: 1) Inappropriate auxological data at diagnosis; 2) no rhGH therapy within 1 month after diagnosis; 3) no clinical data, for serum IGF-I and IGFBP-3 levels before and 1 year after rhGH therapy; 4) age at the beginning of the therapy � 10 years; 5) rhGH therapy length under 2 years; 6) weekly rhGH (Eutropin 1 Plus inj., LG Chem Ltd., Korea); 7) patients with a pubertal sign before or during the 2-year rhGH therapy; 8) children with pituitary or hypothalamic lesions, chromosomal and genetic anomalies, and chronic diseases, as well as endocrinological, skeletal or nutritional abnormalities. The total number of patients included in this study in three different groups were 468, 66, and 153 for IGHD, ISS, and SGA groups, respectively (Fig 1). A total of 705 children with prepubertal short stature that had received rhGH therapy over 24 months were enrolled from the LG Growth Study Database and separated into three groups as follows: IGHD (n = 486), ISS (n = 66), and SGA (n = 153). IGHD group included 119 CGHD patients and 367 PGHD patients.

Auxological and clinical data
Anthropometric measures and medical histories, including sex, chronological age (CA), bone age (BA), height, weight, body mass index (BMI) (kg/m 2 ), mid-parental height (MPH), pubertal state, rhGH dose (mg/kg body weight/week), peak GH serum levels in GH stimulation tests, and serum IGF-I and IGFBP-3 levels, were obtained from the LGS database. The data were collected from their respective data management systems, where all participating centers provided laboratory data generated following standard procedures [22]. The standard deviation score (SDS) values of height, weight, and BMI were converted using the 2017 Korean National Growth Charts for children and adolescents [16]. MPH (cm) was calculated as follows: For girls, (paternal height − 13 + maternal height) / 2; for boys, (paternal height + maternal height + 13) / 2. Then MPH was converted to SDS values. BA was assessed using radiography images of the left hand using the Greulich-Pyle method [14] by the treating physician. The pubertal stage was determined according to Marshall and Tanner [23]. Testicular volume was measured using the Prader orchidometer. Serum IGF-I and IGFBP-3 concentrations (ng/mL) were obtained by multiple researchers. SDS values were calculated according to age-and sex-specific references of Korean children [24]. The difference (Δ) of clinical and laboratory data before and after the treatment was calculated as the change between the two values. The growth response (Δheight, ΔIGF-I and ΔIGFBP-3 SDS) was measured 12 ± 3 and 24 ± 3 months after the treatment. We considered a Δheight SDS of < 0.4 and < 0.3 after the 1 st year of therapy as a poor response for children with IGHD, and ISS/SGA, respectively [3,16,17]. patients and their parents/legal guardians before enrolment at each institute as part of the LGS.

Statistical analysis
All statistical analyses were performed using SPSS (ver. 23.0, IBM Corp., Armonk, NY, USA). The values before and after 1 and 2 years of treatment were compared using analysis of variance (ANOVA) for repeated measures followed by a posteriori Tukey test or Wilcoxon's rank test. Among baseline, 1-year and 2-year rhGH therapy, Δheight SDS, as well as serum ΔIGF-I and ΔIGFBP-3, were analyzed using paired t-test. One-way ANOVA or Kruskal-Wallis test was performed to compare the clinical and laboratory findings for the three groups. Comparisons between poor and good growth response groups were assessed using an independent ttest or Wilcoxon's rank-sum test, depending on the data distribution. Pearson correlation analysis was used to evaluate the association among Δheight SDS and serum ΔIGF-I and ΔIGFBP-3. Analysis of covariance (ANCOVA) between Δheight SDS during 2 nd year of rhGH therapy and ΔIGF-I-ΔIGFBP-3 as fixed factor and height at baseline as a covariate was performed. Statistical significance was set at p < 0.05. Results are presented as mean ± standard deviation.

Comparison of baseline clinical characteristics and laboratory findings
The clinical and laboratory features of prepubertal children with IGHD, ISS, and SGA before treatment and are shown in Table 1. IGHD group comprised CGHD (n = 119) and PGHD (n = 367) sub-groups. Serum IGF-I and IGFBP-3 levels and peak GH level with the GH stimulation test were significantly decreased in CGHD group compared with PGHD (p < 0.05), whereas the sex ratio, age, BA, height, weight, BMI, initial rhGH dose did not differ among the groups. According to the results of comparison among IGHD, ISS and SGA; sex ratio, age, height and IGFBP-3 level were not different (p > 0.05). BA, IGF-I SDS, initial rhGH dose, and peak GH level in the GH stimulation test were significantly low, and weight and BMI were showed the opposite trend in the IGHD group (p < 0.05) compared with ISS and SGA groups.

Comparison of the change in height, serum IGF-I and IGFBP-3 levels
IGFBP-3 SDS was significantly reduced in CGHD group before rhGH treatment compared with the PGHD group; however, the height SDS, Δheight SDS, IGF-1 SDS, ΔIGF-1 SDS, and ΔIGFBP-3 SDS were not significantly different between them before and after rhGH treatment at 1 year and 2 years. A similar significant improvement in height SDS was observed throughout the treatment in the two groups (p < 0.0001) and the increase in IGF-I SDS and IGFBP-3 SDS also showed a similar pattern ( Table 2).
Before rhGH treatment, the mean ± SD for height SDS was -2.69 ± 0.70 (range: -8.10, -1.85) in the IGHD group, -2.75 ± 0.62 (range: -4.73, -1.91) in the ISS group, and -2.61 ± 0.47 (range: -4.03, -1.93) in the SGA group. A similar significant improvement in height SDS was detected throughout the treatment in all three groups (p < 0.0001). During the 2-year treatment, Δheight SDS was the highest in the IGHD group compared to that in the other two groups despite no statistical difference. IGF-I SDS was significantly different among the three groups (p = 0.0184 at the baseline, p = 0.0059 after 1 year of treatment, and p = 0.0156 after 2 years of treatment). However, ΔIGF-I, IGFBP-3 SDS and ΔIGFBP-3 SDS were not significantly different among the three groups at the baseline, and after 1 and 2 years of treatment. IGF-I SDS and IGFBP-3 SDS increased after rhGH therapy after 1 and 2 years of treatment in all three groups. ΔIGF-1, and ΔIGFBP-3 during 1 and 2 years of treatment, were significantly increased compared to those at the baseline (p < 0.05) while ΔIGF-1 and ΔIGFBP-3 were similar between 1 and 2 years of treatment time points (p > 0.05) ( Table 3).

Comparison of the growth response after 1 year of rhGH therapy with clinical parameters and IGF-I/IGFBP-3 serum levels at the baseline
When biochemical data were analyzed according to good or poor 1-year Δheight, baseline BMI was lower in poor response than in good response CGHD group (p = 0.0327). Age (p = 0.0003) and BA (p = 0.0004) at the beginning of the rhGH treatment were higher in poor response group than in good response PGHD group. Sex, MPH, height SDS, IGF-I, and IGFBP-3 level/SDS were similar in both the groups (Table 4).
In comparison IGHD, SGA and SGA groups, we found that the age (p = 0.0010) and BA (p = 0.0002) at the beginning of the rhGH therapy were significantly higher in children with a poor response than in those with a good response in the IGHD group. There was no difference in sex, MPH SDS, BMI SDS, serum IGF-1 and IGFBP-3 levels in the IGHD group. ISS and SGA groups showed no significant difference in sex, MPH SDS, height SDS, BMI SDS, age, BA, IGF-1 and IGFBP-3 between poor and good responder groups (Table 5). BMI, body mass index; IGF-I, insulin-like growth factor-I; IGFBP-3, insulin-like growth factor-binding protein-3; SDS, standard deviation score; IGHD, idiopathic growth hormone deficiency; ISS, idiopathic short stature; SGA, small for gestational age.
Data are expressed as the mean ± standard deviation (SD) or number (%). �, †, ‡ , The same superscript indicates that there is no significant difference between the groups based on the Tukey multiple comparison test.

Correlation between Δheight SDS and serum ΔIGF-I and ΔIGFBP-3 levels
Data are expressed as the mean ± standard deviation (SD) or number (%). �, †, ‡ , The same superscript indicates that there is no significant difference between the groups based on the Tukey multiple comparison test. 1 , 1 year versus baseline; 2 , 2 years versus baseline; groups showed no correlation between Δheight and serum ΔIGF-I and ΔIGFBP-3 levels during 1 and 2 years of therapy (Table 6). ΔIGF-I SDS (beta = 0.0413, p = 0.0322) and ΔIGFBP-3 SDS (beta = 0.0352, p = 0.0467) were significantly positively related to Δheight SDS during 2 years of therapy, revealed using ANCONA analysis, after adjusting for baseline height as covariates in the IGHD group, which was not observed in ISS and SGA groups (Table 7).

Discussion
In this study, we demonstrated the efficacy of serum IGF-I and IGFBP-3 levels in adequately reflecting the height outcome of the first 2 years of rhGH therapy in prepubertal short stature children in IGHD, ISS, and SGA groups. We found that the response to rhGH therapy depended on ΔIGF-I and ΔIGFBP-3 levels in the IGHD group. Indicating that the continuous increase in height for GHD groups in the 2 nd year of treatment was related to the increased concentration of IGF-I and IGFBP-3 in the serum.
Growth response in the first year of rhGH therapy is one of the best indicators of long-term stature growth [25,26]. Initial growth rate, bone remodeling marker levels after 3 months of rhGH therapy, and height SDS have been used to evaluate response to treatment [27,28]. We used height SDS as an indicator and found that children presented with a good Δheight SDS after 1 year of therapy in the IGHD group. The PGHD group had a lower CA and BA at the beginning of the rhGH therapy, showing the importance of early diagnosis and rhGH therapy in IGHD. The patterns of growth and increase in IGF-I and IGFBP-3 levels were similar between CGHD and PGHD subjects. In this study, MPH was not different between CGHD and PGHD and among IGHD, ISS, and SGA groups. Thus, judging the growth responsiveness by clinical data such as MPH alone is insufficient; instead, genetic testing results are more BMI, body mass index; IGF-I, insulin-like growth factor-I; IGFBP-3, insulin-like growth factor-binding protein-3; SDS, standard deviation score; IGHD, idiopathic growth hormone deficiency; ISS, idiopathic short stature; SGA, small for gestational age.
Data are expressed as the mean ± standard deviation (SD) or number (%). �, †, ‡ , The same superscript indicates that there is no significant difference between the groups based on the Tukey multiple comparison test. 1 , 1 year versus baseline; 2 , 2 years versus baseline; efficient in the prediction of rhGH therapy response in short stature children, especially in nogrowth hormone deficiency [29]. The GH-IGF-1 axis is the primary key to the endocrine system that controls linear growth during childhood [30]. GH regulates IGF-I release by acting on the liver, which modulates bone length during childhood [7,31]. IGF-I secretion does not depend on diurnal variation or pulsatility [6]. IGF-I binds to specific IGFBPs in the circulation. IGFBP-3 is the main abundant circulating IGFBP and plays an important role in the metabolic effect of IGF-1 [32]. IGFBP-3 inhibits the bioactivity of IGF-I through binding, thereby reducing the concentration of free IGF-I in the circulation [32]. IGFBP-3 forms a ternary complex with IGF-I through its acid- labile subunits, and the formation of this complex extends their half-life in the circulation [30].
The half-life of IGF-I is about 10 min in free form, and 12-15 h in the ternary complex form [33]. There are continuing efforts to find markers that reflect growth effects. In recent years, references by age, gender, and diagnosis have been derived from many rhGH-treated prepubertal children registries such as the National Cooperative Growth Study (n = 7000) and Pfizer International Growth Database (KIGS) (n = 8500) [21,34]. Although the criteria for patients with IGHD are slightly different in each registry, it is commonly reported that the growth rate in  43) 0.7049 BMI, body mass index; IGF-I, insulin-like growth factor-I; IGFBP-3, insulin-like growth factor-binding protein-3; MPH, mid-parental height; SDS, standard deviation score; IGHD, idiopathic growth hormone deficiency; ISS, idiopathic short stature; SGA, small for gestational age.
the first year is a future growth response parameter, and the growth rate decreases during the second year. Similar results have been reported from 162 prepubertal Swedish children with IGHD and SGA [35]. As a related study, recently Cho, et al. [36] reported a prediction analysis of first-year growth in response to rhGH therapy in 345 prepubertal Korean children with IGHD in the same LGS registry as this study. Therefore, in this study, we focused on the relationship between IGF-I and IGFBP-3 based on growth rates in the first and second years. We presented IGF-I and IGFBP-3 as growth markers. IGF-1 receptor (IGF-IR) dysfunctions and a reduced number of IGF-IR via gene mutations could lead to IGF-I insensitivity [37]. Single nucleotide polymorphisms in the IGFBP-3-coding region are associated with height variation [37]. Especially, the A allele in the IGFBP-3 promoter region reportedly causes an increase in IGFBP-3 expression, which reflects the growth rate in rhGH therapy in prepubertal children with GHD [38] and Turner syndrome [39], but not ISS [40].
GH binds to the GH receptor and stimulates STAT5b activation in the hepatocytes [27]. Subsequently, it stimulates the production and secretion of not only IGF-I but also IGFBP-3 from the liver [7]. Serum levels of IGF-I and IGFBP-3 are even more important in GH/IGF-I assessments for older children and children younger than 3 years of age [7]. However, many

PLOS ONE
Relationship between serum IGF-I/IGFBP-3 and height in rhGH therapy hormones control circulating IGF-1 levels, such as insulin [41], thyroid hormone [42], and sex hormones [43]. Especially, low levels of estrogen stimulate IGF-1 secretion, but high levels of it decrease IGF-1 secretion [43]. In our study, the subjects did not have malnutrition, chronic endocrine/nutritional/inflammatory diseases, and prepuberty. Circulating IGF-I helps monitor compliance and efficacy in response to rhGH therapy despite the limitations of decreased IGF-I concentration in a catabolic state and within-individual and inter-assay variability [6], but there is no clarity on the effect of the IGF-I/IGFBP-3 molar ratio [2,5,25]. In our study, the increase in height continued to improve over 2 years of rhGH therapy in IGHD, ISS, and SGA groups, while IGF-I SDS and IGFBP-3 SDS reached a plateau for all three groups. Domene et al. [44] showed that an increase in serum IGF-I levels over IGFBP-3 levels in patients with GHD was more pronounced compared to that in the SGA group perhaps due to differences in IGF-I levels and/or GH susceptibility at the start of the treatment. Our results also supported their results.
Low ΔIGF-I levels during the first 3 months of therapy in non-IGHD short stature subjects can be an indicator of GH resistance and predict the response to rhGH therapy [27]. Our results confirmed this only for the IGHD group. Previous reports showed that ΔIGF-I is approximately 180 ng/mL in healthy prepubertal short children [45] and under 120 ng/mL in GH-resistant children during the 1 st year of therapy [46].
There is no consensus on the usefulness of serum IGFBP-3 levels due to the lack of related data in rhGH therapy. Furthermore, IGFBP-3 is fragmented by proteases after being secreted [2] under certain conditions, such as inflammation [47]. Moreover, fragmented IGFBP-3 has no function [48,49]. These increase the doubt of using serum IGFBP-3 levels as a tool to monitor rhGH therapy efficacy. Recently, Perez-Colon et al. [27] reported the baseline serum IGFBP-3 level as the key to determining not only the degree of GH resistance but also the growth response to rhGH and IGF-1 treatment in non-GH-deficient short stature and IGF-1 deficient children. Seino et al. [28] reported that baseline IGFBP-3 is significantly related to height velocity. Their data showed a major role of IGFBP-3 in response to rhGH treatment. In the present study, both baseline IGF-I SDS and IGFBP-3 SDS were not notably different between children who showed poor or good height response to rhGH therapy after 1 year in each group. Also, despite a significant increase in height, serum ΔIGFBP-3 levels among the three groups showed a different increasing rate. Therefore, we analyzed serum ΔIGFBP-3 during the 1 st and 2 nd years of thGH therapy and found that ΔIGFBP-3 levels were positively related with Δheight SDS in rhGH therapy in the IGHD group. To the best of our knowledge, this study is the first to report such results. Studies on the predictiveness of serum IGFBP-3 levels in rhGH therapy are limited. Following rhGH therapy, higher serum IGF-I levels, than IGFBP-3, have been reported [27], which suggests that GH has different effects on IGF-I and IGFBP-3 [50] despite the changes in serum IGFBP-3 levels, for example, serum IGF-I levels are correlated with GH concentration. Moreover, IGFBP-3 is reportedly involved in apoptosis and growth inhibition of cancer cells [6]. Therefore, more research is warranted to clarify the function of IGFBP-3 in response to GH.
Our study has several limitations. First, the number of subjects evaluated for the 2-year rhGH therapy was relatively small, especially for ISS and SGA groups, and the study period was short, with just two years. Retrospective studies after reaching adult height are needed to verify that adult height predictions based on growth and hormonal responses during the first two years are useful and accurate. Second, we did not assess serum insulin, IGF-II, free IGF-I, PAPP-A2, and intact IGFBP-3 levels. Third, although children who had already been diagnosed with genetic abnormalities were excluded, not all patients underwent genetic testing for short stature. As the height of children is inherited from the parents in a polygenic manner, a genetic analysis would be needed to find a good responder group, especially in ISS and SGA groups [29]. Fourth, we collected the clinical and laboratory data such as IGF-I and IGFBP3 levels from multiple centers (total 73 sites). As IGF-I and IGFBP-3 levels were measured at different institutions, this could contribute to a measurement bias. However, IGF-I and IGFBP-3 measurements were higher standardized, which were converted to SDS values by LGS [37]. Fifth, all the subjects were not tested using the same GH stimulation test. As mentioned, we determined the peak GH levels from two separate GH stimulation test results using insulin, clonidine, L-arginine, L-dopa, or glucagon. However, the provocation test using insulin was common among all of them.
Despite its limitation, this study has several important advantages. First, the large sample size with homogeneous subjects such as prepubertal children and a single ethnic group race in LGS data were analyzed in this study. Additionally, the LGS data are significant because they can reduce errors among researchers when the information is collected through multicenter databases [19]. Second, various clinical parameters were positively analyzed and subdivided among patients classified as IGHD, especially CGHD and PGHD, and ISS and SGA. Third, although a few previous studies about IGF-I/IGFBP-3 and growth response consisting of GHD, ISS, and SGA in patients exist, there has been no study involving a relatively large cohort of prepubertal GHD patients like ours using our rhGH therapy.
In summary, most prepubertal children included in this study, who received rhGH therapy, showed an increase in height after the 1 st and 2 nd year of rhGH treatment in all three groups. The subjects with IGHD had normalized serum IGF-I values within the 1st year of rhGH therapy; serum IGF-I and IGFBP-3 levels were associated with height improvement.
In conclusion, IGF-I and IGFBP-3 are the primary factors involved in growth; therefore, it is important to identify more factors that affect their levels in serum. We believe that serum ΔIGFBP-3 level may serve as a marker to predict the height response along with serum ΔIGF-I level in patients with IGHD. Also, our results may help improve the overall growth outcomes of children receiving personalized rhGH therapy.