This study was part of a cross-sectional study of children who visited the Department of Pediatric Endocrinology, Hanyang University College Medicine. Written informed consent was obtained from the guardians of all participants prior to enrollment, and the study was conducted in accordance with the ethical principles of the Declaration of Helsinki, revised in 2013. A total of 30 children aged 4 to 6 years were recruited between July 2016 and January 2023, including 15 children diagnosed with GHD and 15 with ISS, matched for age and sex. Children presenting with short stature underwent GH stimulation testing to differentiate between GHD and ISS. The inclusion criteria for children with GHD were as follows: (I) GHD was diagnosed according to the guidelines for GHD (Cook and Rose, 2012); (II) peak serum GH level less than 10 ng/mL on at least two GH stimulation tests; (III) no other clinically significant neurological or psychiatric disorders, nor adrenocorticotropic hormone deficiency, hypoglycemia, thyroid-related, genetic, or metabolic diseases; (IV) in cases without radiological abnormalities on MRI; and (V) no prior history of GH therapy. Children with ISS met the same diagnostic criteria but exhibited a peak serum GH level exceeding 10 ng/mL in response to stimulation with at least one of the two agents—arginine or levodopa (L-dopa). Participants were excluded if they met any of the following criteria: (I) growth or developmental delay due to other causes, such as small for gestational age, intrauterine growth restriction, or prematurity; (II) a history of GH replacement therapy; or (III) brain lesions detected on MRI or a history of psychiatric or neurological disorders.

Clinical data were obtained from medical records and included age, sex, height, weight, body mass index, and serum IGF-1 and IGF-binding protein 3 (IGFBP-3) levels. Serum IGF-1 and IGFBP-3 levels were further converted to standard deviation scores (SDS) using age- and sex-specific reference intervals established for Korean children (21). The IGF-1/IGFBP-3 molar ratio was calculated according to the formula as previously described (22):

GH stimulation testing was performed to evaluate GHD and remains the gold standard for the clinical differentiation of short stature in children with either GHD or ISS (23). A clinical diagnosis of GHD was made when two or more GH provocation tests yielded insufficient peak GH levels. All children were instructed to fast overnight for 8 to 12 hours before the test. Participants underwent serial blood sampling at 0, 30, 60, 90, and 120 minutes following an intravenous bolus injection of arginine and L-dopa, and the peak GH level was determined based on the results of the provocation tests. Serum GH levels were measured using an ELISA-based immunoassay. Given the well-established inter-assay variability of GH immunoassays and the tendency of ELISA platforms to report higher GH concentrations than modern chemiluminescent assays (24, 25), an assay-appropriate diagnostic cutoff of<10 ng/mL was applied in accordance with our institutional protocol. As a peak GH level >10 ng/mL indicates the absence of GHD, a cutoff value of 10 ng/mL was employed in the diagnostic evaluation (26). Based on the peak GH level, participants were categorized as having either GHD or ISS.

MRI scans were acquired at the first hospital visit using a 3.0-T scanner (Philips Achieva, Best, Netherlands) equipped with a 16-channel phased array head coil and a magnetization-prepared rapid gradient echo sequence. All scans were performed without sedation, and an experienced pediatrician continuously monitored pulse oximetry to ensure stable cardiac and respiratory parameters throughout the procedure. The imaging protocol included whole-brain and standard pituitary MRI sequences, comprising sagittal and axial T1-weighted images. The parameters were: TE = 3.39 ms, TR = 2.10 ms, TI = 1 ms, field of view = 200 mm², voxel size = 0.9 × 0.9 mm², slice thickness = 1 mm, and slice number = 150.

Pituitary gland volume segmentation was performed using ITK-SNAP (version 3.6.0; University of Pennsylvania, PA, USA; http://www.itksnap.org), an open-source software offering manual and semi-automated segmentation functions. First, a DICOM image series was imported into ITK-SNAP. Segmentation of the pituitary gland was guided by anatomical landmarks, with the sphenoid sinus anteriorly, diaphragma sellae superiorly, cavernous sinuses laterally, and dorsum sellae posteriorly. Manual segmentation was performed on each sagittal slice using the paintbrush tool, carefully excluding Rathke’s cleft cyst if present. The volume of the segmented pituitary gland was automatically calculated in the “Volume and Statistics” module of ITK-SNAP, based on the total number of labeled voxels multiplied by the voxel volume.

3D T1-weighted images were processed using the recon-all pipeline in FreeSurfer (version 7.2.0; Boston, MA, USA), which included surface reconstruction and cortical parcellation. Manual corrections, such as adding control points to adjust gray matter–white matter boundary errors, were applied and reprocessed as needed. Thalamic nuclei were segmented using an automated algorithm in FreeSurfer, based on a probabilistic atlas constructed from ex vivo MRI and histological data. The detailed methodology has been described previously (27). Segmentation accuracy was verified through visual inspection using Freeview. Total intracranial volume (tICV) was automatically extracted, and relative thalamic volume was calculated using the following formula:

All images were independently reviewed by two researchers, and only those that passed quality control were included in the final analysis. The pipeline of data pre-processing and thalamic nuclei volume analyses is shown in Figure 1. The thalamic nuclei were grouped into 10 categories based on their anatomical location and functional similarity: Anteroventral (AV), Ventral anterior (VA), Ventral lateral (VL), Ventromedial (VM), Ventral posterolateral (VPL), Lateral, Mediodorsal (MD), Lateral geniculate (LGN), Medial geniculate (MGN), and Pulvinar. Specifically, the VA group included the VA and ventral anterior magnocellular (VAmc) nuclei. The VL group comprised the ventral lateral anterior (VLa) and ventral lateral posterior (VLp) nuclei. The Lateral group consisted of the laterodorsal (LD) and lateral posterior (LP) nuclei. The MD group included the mediodorsal medial (MDm) and mediodorsal intermediate (MDi) nuclei. The Pulvinar group encompassed the pulvinar anterior (PUA), pulvinar medial (PuM), pulvinar lateral (PuL), and pulvinar inferior (PuI) nuclei (Figure 2).

All statistical analyses were performed using IBM SPSS Statistics (version 27.0; SPSS Inc., Chicago, IL, USA) and R (version 4.4.2). For pituitary gland volume analyses, absolute volumes were analyzed using analysis of covariance (ANCOVA), with age, sex, and tICV. For thalamic nuclei analyses, volumes were normalized to tICV and multiplied by 1,000 to reduce variability related to overall brain size. Group comparisons of thalamic nuclei were subsequently performed using ANCOVA with age and sex as covariates. Sex differences between the two groups were analyzed using the chi-squared test. Student’s t-tests or Fisher’s exact tests were used to compare clinical variables, as appropriate. Effect sizes (Cohen’s d) were calculated for all measures including both significant and non-significant comparisons to provide a comprehensive description of group differences. Post-hoc power estimates based on the observed effect sizes were obtained using G*Power (version 3.1.9.7). For multiple comparisons, false discovery rate (FDR) correction was applied to analyses of the thalamic nuclei volumes. Statistical significance was defined as a two-tailed p-value less than 0.05, after FDR correction. Partial correlation coefficients (R values) were calculated to assess the associations between biochemical parameters and individual thalamic nuclei volumes, controlling for age and sex. A lateralization index (LI) was calculated to characterize the asymmetry of thalamic development using the following formula:

Demographic and endocrine clinical characteristics are shown in Table 1. Fifteen children (age range 4–6 years, mean 5.07 ± 0.80 years) with GHD and fifteen children (age range 4–6 years, mean 4.87 ± 0.83 years) with ISS were included in this study. Mean height was 101.65 ± 4.77 cm in GHD group, which was significantly shorter than the mean of 105.77 ± 3.37 cm for the ISS group (p = 0.017). Mean height SDS was also smaller in the GHD group (GHD: -2.10 ± 0.54 vs ISS: -1.21 ± 1.26, p = 0.031). In the GHD group, peak GH levels during both arginine and L-dopa stimulation tests were below 10ng/mL, whereas in the ISS group, peak levels exceeded 10ng/mL in at least one of the two tests (p = 0.026 for arginine and p = 0.017 for L-dopa). While IGF-1 and IGFBP-3 levels tended to be higher in the ISS group, these differences were not statistically significant.

The mean volume of the pituitary gland was 153.650 ± 24.102 mm³ in the GHD group and 200.454 ± 54.130 mm³ in the ISS group. Children with GHD exhibited a significantly smaller pituitary gland volume compared with those with ISS, and this difference remained statistically significant after adjustment for age, sex, and tICV using ANCOVA (adjusted p = 0.011; Table 2, Figure 3). This group was associated with a large effect size (Cohen’s d = 1.117) and a high achieved post-hoc power (1-β = 0.840), indicating a robust between-group differences. Correlations between pituitary gland volume and biochemical parameters in the GHD group are summarized in Supplementary Table 1.

Table 3 shows the volumes of the whole thalamus and individual thalamic nucleus. Whole and bilateral thalamic volumes did not differ between the groups (Figures 4, 5). However, a significant group difference was found exclusively in the left AV nucleus. The GHD group exhibited a significantly greater volume of the AV nucleus than the ISS group (0.159 ± 0.027 mm³ vs. 0.133 ± 0.031 mm³; p = 0.036), suggesting a localized structural alteration. The increased volume of the left AV nucleus in children with GHD was associated with a large effect size (Cohen’s d = 0.868), although the post-hoc power was moderate (1−β = 0.631), reflecting the modest sample size. To further investigate the relationship between thalamic structure, brain development, and functional characteristics, the LI of the thalamic nuclei were analyzed. The LI of the VA nucleus showed more pronounced leftward lateralization in children with GHD. This difference corresponded to a medium effect size (Cohen’s d = 0.746), with limited post-hoc power (1−β = 0.505). No other thalamic nuclei exhibited significant group differences (Table 4, Figure 6).

Figures 7, 8 illustrate the correlations between the LI of individual thalamic nuclei volumes and the serum IGF-1 and IGFBP-3 levels. Analysis revealed that higher IGF-1 and IGFBP-3 levels were significantly correlated with increased LI of the AV nucleus (FDR-corrected p = 0.022 for all). Additionally, the VA nucleus showed significant positive correlations with the IGF-1 (p = 0.022) and IGFBP-3 levels (p = 0.033). Consistent with these findings, IGF-1 SDS was also significantly associated with the VA nucleus (p = 0.033) and ventral lateral nuclei (p = 0.044). No other thalamic nuclei showed significant correlations following FDR-correction (Table 5). The correlations between thalamic nuclei volume and biochemical parameters are provided in Supplementary Tables 2, 3.

Consistent with previous studies, our results demonstrate that children with GHD exhibited significantly reduced pituitary gland volumes compared to children with ISS. Importantly, this volumetric reduction remained statistically significant after adjustment for age, sex, and tICV. These findings suggest that pituitary gland volume represents a robust morphological feature associated with GHD during early childhood. Previous studies using conventional 2D MRI have similarly reported decreased pituitary morphology in children with GHD, although those studies often relied on geometric approximations (e.g., the ellipsoid formula), which may underestimate the true pituitary gland volume (28). By contrast, the present study employed semi-automated segmentation techniques applied to high-resolution 3D MRI, which allowed for more anatomically accurate and reproducible volume estimates. These findings support the clinical utility of 3D MRI pituitary gland volume in the evaluation of pediatric GHD and underscore the importance of considering tICV as a covariate when interpreting pituitary morphology during periods of rapid brain development.

Previous neuroimaging studies have consistently reported global brain morphometric abnormalities in children with GHD, including reductions in total brain volume, cortical surface area, and gray matter thickness (14). Region-specific atrophy has also been observed in structures such as the corpus callosum, hippocampus, and thalamus (4), implicating disruptions in interhemispheric integration, memory processing, and sensory integration. Diffusion tensor imaging studies further support the presence of microstructural white matter abnormalities, such as increased mean diffusivity in the corticospinal tract (29), which may contribute to altered motor function. Functional MRI studies have additionally demonstrated reduced structure–function coupling in the primary sensory and motor cortices among children with GHD (30), suggesting a decoupling between anatomical integrity and neural activation patterns. The present study extends previous research by identifying, for the first time, alterations in thalamic subnuclear morphology in children with GHD, a region that has received limited attention. In contrast to prior studies that have primarily focused on global volumetric reductions and cortical thinning, our results reveal a more nuanced pattern of subcortical remodeling. Notably, we identified an increased volume of the left AV thalamic nucleus in children with GHD. The AV nucleus is a key component of the Papez circuit (31), linking the hippocampus and cingulate cortex and contributing to memory-related processes. This focal volumetric alteration may reflect adaptive neurodevelopmental changes associated with disrupted GH signaling during early brain maturation.

In addition to volumetric alterations, children with GHD exhibited significantly more pronounced leftward lateralization of the VA nucleus compared to children with ISS. Hemispheric lateralization is a fundamental organizational principle of the developing brain, supporting functional specialization and neural efficiency (32, 33). This process follows a tightly regulated developmental trajectory during early childhood and is influenced by both genetic and hormonal factors. Structural asymmetries in subcortical regions, including the thalamus, have been associated with typical cognitive and motor development (34), and previous studies have reported pronounced leftward thalamic asymmetry during the preschool period (35). The VA nucleus occupies a strategic position within cortico–basal ganglia–thalamic loops and is extensively interconnected with the prefrontal and premotor cortices, as well as the basal ganglia (36–38). Through these connections, the VA nucleus plays a critical role in motor planning, initiation of voluntary movement, and higher-order cognitive processes such as executive control and response selection. The more pronounced leftward lateralization of the VA nucleus observed in children with GHD may therefore reflect asymmetric maturation of motor and cognitive circuits, potentially driven by region-specific sensitivity to GH/IGF signaling. Whether this pattern represents an adaptive compensatory mechanism or an early marker of altered hemispheric organization secondary to endocrine disruption remains to be clarified.

A key and novel contribution of the present study is the identification of significant positive correlations between thalamic asymmetry indices and biochemical markers of GH activity, specifically the serum IGF-1 and IGFBP-3 levels. Despite the absence of significant between-group differences in circulating IGF-1 and IGFBP-3 levels, the positive correlations observed within the children with GHD underscore an important distinction between group-level comparisons and individual variability. Prior neuroimaging studies in children with GHD have demonstrated that growth-related hormone levels, including IGF-1 and IGFBP-3, show significant associations with volumes of subcortical structures within the GHD group, even in the absence of significant between-group differences, supporting the notion that relative IGF signaling may be associated with subcortical neuroanatomy within affected populations (17). Whereas between-group analyses primarily reflect average hormonal status, within-group associations capture how inter-individual differences in residual GH/IGF axis activity may influence neurodevelopment under conditions of hormonal insufficiency. Thus, in children with GHD, even modest variability in IGF-1 and IGFBP-3 levels may represent biologically meaningful differences in residual IGF signaling, which could exert a disproportionate influence on the maturation and organization of subcortical structures, including hemispheric specialization. On this basis, thalamic lateralization in children with GHD may relate to disease severity along a continuous biological spectrum rather than a dichotomous diagnostic category. Greater alterations in thalamic LI may reflect more pronounced disruption of GH/IGF signaling, whereas relatively preserved IGF signaling may be associated with more typical hemispheric organization. By contrast, children with ISS exhibit largely intact GH/IGF axis function and may therefore lack a comparable biological gradient linking hormone levels to thalamic lateralization, potentially accounting for the absence of significant correlations within the ISS group.

IGF signaling plays a critical role in CNS development by promoting neural proliferation, maturation, survival, and growth (19). Consequently, even modest variations in circulating IGF-1 and IGFBP-3 levels may exert disproportionate effects on subcortical maturation during sensitive developmental windows. In the present study, this heightened sensitivity was reflected in the observed associations between IGF-related biomarkers and thalamic lateralization indices, suggesting that thalamic development may be especially responsive to relative differences in GH/IGF axis activity. In line with these observations, more pronounced leftward lateralization in the AV and VA thalamic nuclei was associated with higher circulating IGF-1 and IGFBP-3 levels. IGF-1 is known to exert neurotrophic effects by promoting synaptogenesis and axonal growth in regions such as the thalamus and hippocampus (10, 39), while IGFBP-3 modulates IGF-1 bioavailability by regulating its stability and tissue distribution (40). Given that the AV and VA nuclei are embedded within motor and cognitive relay circuits, including connections with the secondary motor cortex and cortico–basal ganglia–thalamic loops (41, 42), the restriction of these associations to specific subregions suggests preferential sensitivity of sensorimotor-related thalamic nuclei to GH/IGF signaling. Collectively, these findings indicate that inter-individual variability in GH/IGF activity may contribute to differences in thalamic lateralization through region-specific neurodevelopmental mechanisms.