Pituitary magnetic resonance imaging abnormalities and 1-year growth response in children with idiopathic growth hormone deficiency
Article information
Abstract
Purpose
Brain magnetic resonance imaging (MRI) is used to assess anatomical abnormalities in children with idiopathic growth hormone deficiency (IGHD), but its prognostic relevance remains uncertain. We investigated whether structural abnormalities on brain MRI are associated with 1-year growth response to recombinant human growth hormone (rhGH) therapy in Korean children with IGHD.
Methods
This multicenter retrospective study included 457 IGHD patients from the LG Growth Study registry who had baseline brain MRI and received rhGH therapy for ≥1 year. MRI findings were categorized as normal or as showing pituitary or extrapituitary abnormalities. Growth outcomes—change in height SDS (ΔHeight SDS) and height velocity—were compared across groups. Multivariable regression was performed to adjust for clinical variables.
Results
Among 457 patients, 122 (26.7%) had structural abnormalities on brain MRI, while 335 (73.3%) had normal findings. In unadjusted analysis, those with abnormalities showed slightly lower ΔHeight SDS than those with normal brain MRI (p=0.028). However, subgroup differences were not significant when analyzed categorically. In multivariable regression, younger age at treatment (β=-0.050, p=0.004), higher baseline body mass index SDS (β=0.055, p=0.002), and greater rhGH dose (β=0.806, p=0.039) were significantly associated with greater ΔHeight SDS, while brain MRI abnormalities were not predictive of response.
Conclusions
While brain MRI is an important diagnostic tool in IGHD for identifying anatomical abnormalities, our findings suggest that such abnormalities, particularly when mild or isolated, have limited utility in predicting short-term growth response to rhGH therapy.
Highlights
· In 457 Korean children with idiopathic growth hormone deficiency, 26.7% exhibited structural brain magnetic resonance imaging abnormalities, predominantly pituitary hypoplasia. However, these structural findings did not independently predict 1-year growth outcomes following recombinant human growth hormone therapy. Instead, younger age, higher baseline body mass index, and greater recombinant human growth hormone dose were the significant predictors of early height gain.
Introduction
Growth hormone deficiency (GHD) in children can have heterogeneous causes and presentations across different populations [1,2]. Epidemiological studies from various regions have documented differences in the prevalence and etiologies of GHD [2]. International research efforts, such as the Pfizer International Growth Database (Kabi International Growth Study, KIGS) registry, have proposed classification systems for GHD etiologies [3], and current consensus guidelines recommend including brain magnetic resonance imaging (MRI) as part of the diagnostic work-up for childhood GHD [4]. MRI of the brain (particularly the hypothalamicpituitary [HP] region) is recommended to identify structural lesions that might be associated with GHD. However, most of these studies involved Western populations, leaving a gap in understanding the characteristics and clinical implications of brain MRI findings in Asian cohorts. This study uniquely evaluates a large Korean cohort, providing region-specific data that may complement international findings and inform clinical practice in East Asian populations.
GHD is broadly categorized into organic and idiopathic types. Organic GHD is typically caused by identifiable structural or acquired conditions such as brain tumors, cranial irradiation, trauma, or congenital malformations of the HP axis. In contrast, idiopathic GHD (IGHD) lacks a clearly defined etiology despite thorough clinical and radiological evaluation. According to previous epidemiologic studies, organic GHD accounts for approximately 20%–30% of childhood GHD cases, while the majority are classified as idiopathic [4,5]. A substantial minority of these idiopathic cases exhibit abnormalities on brain MRI, such as an ectopic posterior pituitary or a small anterior pituitary gland (pituitary hypoplasia) [6]. These radiological findings may indicate an underlying congenital or structural defect in patients clinically classified as idiopathic. However, whether such brain MRI abnormalities have predictive value for treatment response to growth hormone (GH) therapy remains uncertain. Moreover, the classification of patients with isolated pituitary abnormalities as having idiopathic GHD has been debated. International guidelines, including those by the GH Research Society, permit classifying mild or isolated pituitary anomalies without multiple pituitary hormone deficiencies (MPHDs) as idiopathic [7], but this remains a source of conceptual ambiguity in clinical practice.
Some studies have proposed that specific pituitary anomalies may correlate with differential growth outcomes [8] and that patients with distinct structural lesions may have more severe GHD and thereby exhibit varied treatment responses [5,9]. Other studies, however, have reported that children with normal brain MRI scans can respond just as well to GH therapy as those with abnormal scans [5,9].
Despite these insights, no study has specifically examined this question in a Korean pediatric population. We therefore undertook this multicenter retrospective study to evaluate whether structural abnormalities on brain MRI are associated with differences in clinical characteristics or short-term growth response in Korean children with idiopathic GHD. In particular, we aimed to assess the independent prognostic value of MRI findings in predicting growth outcomes following 1 year of GH therapy.
Materials and methods
1. Study design and population
This retrospective cohort analysis used data from the LG Growth Study (LGS), a multicenter noninterventional registry designed to evaluate the long-term safety and effectiveness of recombinant human GH (rhGH) therapy in Korean children [10,11]. The design and protocol of the LGS have been described in detail elsewhere [10], and the 5-year outcome data have demonstrated its safety and efficacy in real-world clinical practice [11].
The study included 3 daily rhGH products (Eutropin, Eutropin AQ, and Eutropin Pen) and 1 weekly product (Eutropin Plus), all manufactured by LG Chem Ltd.
When GH stimulation testing was indicated, appropriate pharmacologic agents—such as insulin-induced hypoglycemia, L-dopa, clonidine, L-arginine, or glucagon— were selected at the discretion of the treating physician. After at least 2 GH stimulation tests (GHSTs), patients with a peak GH level below 10 ng/mL [8] were classified as having GHD in the LGS database. Patients with both GH peaks <5 ng/mL were classified as having complete GHD, while those with at least one GH peak between 5 and 10 ng/mL were classified as partial GHD, in accordance with established diagnostic criteria [12].
We included children diagnosed between February 2001 and February 2020. Idiopathic GHD was defined by the absence of an identifiable organic cause for growth failure and peak GH levels <10 ng/mL on stimulation testing. All patients included in this study had isolated GHD, defined as GH deficiency without concurrent deficiencies in other pituitary hormones, including a normal thyroid axis (thyroid-stimulating hormone and free T4), adrenal axis (morning cortisol or adrenocorticotropic hormone stimulation results when available), and gonadotropin axis (luteinizing hormone/follicle stimulating hormone and sex steroid levels appropriate for age and pubertal status). To focus exclusively on idiopathic cases, we excluded patients with known genetic syndromes, brain tumors, prior cranial irradiation, or other acquired causes of GHD.
Of the 3,813 pediatric patients initially enrolled in the LGS, 2,273 were diagnosed with GHD based on investigator assessments. To ensure the validity of clinical and imaging data, we excluded 888 patients without available brain MRI data, 4 with organic GHD, 896 with incomplete height and/or weight measurements, 6 receiving medications known to affect serum glucose levels, 2 with MPHDs, 4 with missing GHST results, and 16 whose bone age (BA) exceeded chronological age (CA) (BA > CA) (Fig. 1).
Study population flow diagram. Flowchart illustrating patient selection from the LG Growth Study registry. Among 3,813 enrolled children, 2,273 were diagnosed with growth hormone deficiency (GHD) and selected for analysis. After applying exclusion criteria—absence of MRI (n=888), organic GHD (n=4), missing height and/or weight data (n=896), use of glucoseaffecting medications (n=6), multiple pituitary hormone deficiencies (MPHDs, n=2), missing growth hormone stimulation test (GHST) results (n=4), and BA exceeding CA (BA > CA, n=16)—a total of 457 patients were included in the final analysis. Of these, 132 had complete IGHD and 325 had partial IGHD. MRI, magnetic resonance imaging; BA, bone age; CA, chronological age; GH, growth hormone; IGHD, idiopathic GHD.
A total of 457 patients met the inclusion criteria and were included in the final analysis. Of these, 132 (28.9%) were classified as having complete GHD and 325 (71.1%) as partial GHD.
2. MRI classification
In the LGS registry, brain MRI information became more consistently available after 2018 as the data collection process was standardized. Brain MRI findings were obtained according to the study protocol and categorized as either normal or abnormal. Abnormalities included intrinsic defects within the pituitary region—such as ectopic posterior pituitary, an interrupted or thinned pituitary stalk, and anterior pituitary hypoplasia—as well as incidental extrapituitary lesions, including arachnoid cysts or midline cerebral anomalies. Pituitary hypoplasia in the LGS registry was defined based on the local radiologist’s impression of a small anterior pituitary gland. Quantitative pituitary height measurements (e.g., <mean -2 standard deviation [SD]) were not uniformly applied across centers. No formal interrater reliability assessment or standardized pituitary protocol was implemented across centers. For analysis, abnormal findings were further classified into pituitary and extrapituitary abnormalities according to lesion location. For the primary 3-group comparison (pituitary, extrapituitary, and normal) used in Tables 1–2 and Fig. 2, the 2 patients with combined pituitary and extrapituitary lesions were included in the pituitary abnormality group (n=108 isolated+n=2 combined=110), whereas the 4-group breakdown reported in Supplementary Fig. 1 (108 pituitary-only, 12 extrapituitary-only, 2 combined) presents these categories separately.
Baseline clinical characteristics by magnetic resonance imaging classification
One-year growth outcomes by magnetic resonance imaging classification
Change in height SDS by MRI group. Mean height SDS values at baseline and after 1 year of recombinant human growth hormone therapy, stratified by MRI classification: pituitary abnormalities, extrapituitary abnormalities, and normal findings. No statistically significant differences in growth response were observed among the 3 groups after adjustment for covariates. SDS, standard deviation score; MRI, magnetic resonance imaging.
3. Treatment and outcome measures
All patients received rhGH at standard pediatric dosing, administered via daily subcutaneous injection for at least 1 year. Baseline clinical and biochemical data collected at treatment initiation included CA, BA (assessed by the Greulich-Pyle method), height, weight, and body mass index (BMI) (all expressed as standard deviation scores [SDSs] calculated using the Lambda-Mu-Sigma (LMS) method with L, M, and S parameters derived from the 2017 Korean National Growth Charts for children and adolescents [13]). Pubertal status, serum insulin-like growth factor-1 (IGF-1) and insulin-like growth factor-binding protein 3 levels (including their SDS, calculated using Korean pediatric reference values [14]) were also recorded. Additional variables included CA minus BA (CA–BA) and the average GH dose (mg/kg/wk).
The primary outcome measure was the change in height SDS (ΔHeight SDS) after 1 year of treatment, calculated as the difference between 12-month and baseline values. Secondary outcomes included absolute height SDS at 1 year, and changes in weight SDS and BMI SDS over the same period.
4. Statistical analysis
Baseline characteristics and growth outcomes were compared between the MRI-normal and MRI-abnormal groups, as well as among the 3 MRI subgroups (pituitary-axis abnormality, extrapituitary abnormality, normal). Continuous variables were summarized as mean± SD or median [interquartile range], as appropriate. Categorical variables were summarized as counts and percentages. Group comparisons for continuous variables were performed using analysis of variance when assumptions of normality were satisfied, and using the Kruskal-Wallis test when not. Categorical variables were compared using the chi-square test or Fisher exact test, as appropriate. A P-value <0.05 was considered statistically significant. All statistical analyses were performed using SAS 9.4 (SAS Institute Inc., USA).
5. Ethics statement
This study was conducted in accordance with the Declaration of Helsinki. The study protocol was reviewed and approved by the institutional review board of Seoul National University Bundang Hospital (X-2102-664-901), and written informed consent was obtained from all patients and their legal guardians.
Results
1. Patient characteristics
A total of 457 children with IGHD were included in the study (282 boys and 175 girls). At the start of rhGH treatment, the mean age was 7.56±2.97 years. The mean baseline height SDS and BMI SDS were -2.68±0.61 and -0.40±1.16, respectively. BA was delayed compared to CA, with a mean BA of 6.10±2.97 years and a mean CA–BA of 1.93±1.07 years (Table 1). Because only 457 of the 2,273 patients with GHD had complete MRI and growth data, the possibility of selection bias cannot be excluded. However, in a separate comparison, patients excluded due to missing MRI or incomplete growth data demonstrated broadly similar baseline characteristics—including age, sex distribution, height SDS, BMI SDS, and peak GH levels—compared with the analytic cohort (Supplementary Table 1).
2. MRI findings
Among the 457 patients included in the analysis, brain MRI was normal in 335 patients (73.3%), while structural abnormalities were observed in 122 patients (26.7%). The most common abnormal MRI finding was pituitary hypoplasia, defined as an unusually small anterior pituitary gland (n=77, 63.1% of abnormal cases). Other frequent abnormalities included Rathke cleft cysts (n=15), partial empty sella (n=6), arachnoid cysts (n=6), ectopic posterior pituitary (n=3), and pineal cysts (n=3).
Of the 122 patients with confirmed brain MRI abnormalities, 108 (88.5%) had lesions confined to the pituitary region, 12 (9.8%) had only extrapituitary abnormalities, and 2 (1.6%) had abnormalities in both the pituitary region and extrapituitary areas (Supplementary Fig. 1).
Baseline clinical characteristics were comparable among the 3 MRI-defined groups: pituitary, extrapituitary, and normal. In particular, there were no significant differences in initial height SDS, weight SDS, BMI SDS, or BA delay among the 3 groups (all P>0.05) (Table 1). Peak GH levels during stimulation testing were also comparable across MRI groups (Table 1), suggesting that structural abnormalities were not preferentially associated with either complete or partial GHD.
3. Growth response to rhGH therapy
After 1 year of rhGH treatment, patients showed substantial improvement in height SDS, with a mean increase of 0.77±0.36. Female patients exhibited a greater increase in height SDS compared with males in unadjusted analysis (P=0.038); however, this difference was not significant in multivariable regression analysis (Supplementary Fig. 2). No significant difference in ΔHeight SDS was observed between patients with complete and partial GHD (Supplementary Fig. 3). Similarly, baseline height SDS and 1-year height SDS were comparable across the 3 MRI groups (pituitary, extrapituitary, and normal), with mean ΔHeight SDS values of 0.72, 0.66, and 0.80, respectively (P=0.079) (Table 2; Fig. 2).
To further explore the impact of brain MRI classification, we applied multiple grouping strategies, including: (1) normal versus all abnormal findings; (2) normal, pituitary-only, combined pituitary and extrapituitary, and extrapituitary-only groups; (3) normal versus any pituitary involvement versus extrapituitary-only; and (4) normal, pituitary-only, and all others with extrapituitary lesions. Among these, only the first grouping (normal vs. all abnormal) demonstrated a statistically significant difference in ΔHeight SDS (P=0.028), with the abnormal group showing a slightly lower response; however, this difference disappeared after adjusting for clinical covariates. No significant differences were observed with the other grouping strategies.
4. Multivariable regression analysis
To further identify predictors of growth response to rhGH therapy, we conducted multivariable linear regression analyses using the change in height SDS (ΔHeight SDS) as the dependent variable (Table 3). Younger age at treatment initiation (β=-0.050, P=0.004), higher BMI SDS at screening (β=0.055, P=0.002), and higher average rhGH dose during treatment (β=0.806, P=0.039) were significantly associated with greater height gain. Higher peak GH levels during stimulation testing showed a modest but significant negative association (β=-0.022, P=0.005), while BA exhibited a nonsignificant inverse trend (P=0.604). Additional results using absolute height SDS at 1 year are provided in Supplementary Table 2.
Multivariable linear regression analysis predicting change in height SDS (ΔHeight SDS) after 1 year of rhGH therapy
Discussion
In this multicenter cohort of Korean children with IGHD, approximately 26.7% showed structural abnormalities on brain MRI, most commonly pituitary hypoplasia. While unadjusted comparisons suggested a significant difference in growth response between the normal and abnormal MRI groups, this association disappeared after adjusting for clinical covariates, consistent with previous reports that such anomalies, although common, do not consistently predict growth response [8,15,16]. Our study extends these findings to a large Korean population, complementing existing Western registry data.
Although CA–BA and IGF-1 SDS differed between MRI groups at 1 year, these parameters were not correlated with growth outcomes, suggesting hormonal or maturational variations rather than structural determinants. Jung et al. [17] proposed that serum IGF-I SDS and GH peak levels may help predict brain MRI abnormalities, but such approaches cannot replace direct imaging. Brain MRI remains essential for suspected organic GHD, particularly with low GH peaks or MPHDs [4,18,19].
Guidelines from the GH Research Society [4] and KIGS registry analyses [3,12] allow classification as IGHD when mild or isolated anomalies—such as pituitary hypoplasia—are present without multiple hormone deficiencies. Isolated brain MRI abnormalities may still reflect congenital HP axis defects [20]. Recent quantitative MRI studies [21] suggest that such anomalies could involve myelin alterations and increased vulnerability in GH–IGF-1–related neural networks. An ectopic posterior pituitary or stalk interruption is strongly linked to severe, often multiple deficiencies [5,18]. However, for mild anomalies without additional endocrine deficits, the structure–function relationship remains unclear [9,15]. Interpretation is further complicated by age- and puberty-related changes in pituitary size and interobserver variability [22].
In our study, none of the MRI grouping strategies predicted growth response in regression models. Notably, the extrapituitary group had lower IGF-1 SDS despite similar GH stimulation results, suggesting possible subtle hypothalamic–pituitary dysregulation.
Recent reports provide further context for our findings. In a meta-analysis of nonacquired GHD, Hwang et al. [8] summarized 15 observational studies and found that brain MRI abnormalities were present in approximately one-quarter of patients, but their prevalence and clinical implications varied substantially across cohorts. Importantly, MRI findings were not consistently predictive of growth outcomes, echoing our adjusted results. In Korea, Jung et al. [17] reported that pituitary abnormalities were frequent among IGHD patients in a single tertiary center, but these structural findings did not translate into reliable prognostic information for GH treatment response. Similarly, Oh et al. [23] evaluated pituitary volume using quantitative MRI and noted an association with growth response in GHD but not in idiopathic short stature, underscoring the limited generalizability of structural parameters. From a regional perspective, Zhang et al. [24] observed that smaller pituitary height correlated with greater catch-up growth in Chinese children with GHD, suggesting potential ethnic or methodological influences on outcome variability. Collectively, these studies support our conclusion that although structural anomalies are common and sometimes associated with hormonal profiles, they are not independently reliable predictors of short-term growth response once clinical and biochemical covariates are taken into account.
Previous studies have yielded mixed results: some show associations between specific anomalies and growth outcomes [6,24-26], whereas others report none [15,16]; large datasets demonstrate wide variability even with similar MRI features [3,8]. This likely reflects the multifactorial nature of GH response, influenced by receptor function, IGF-I bioactivity, genetics, nutrition, adherence, and treatment timing [4,16]. Even with uniform anatomical diagnoses such as stalk interruption, outcomes may differ substantially [18].
Variability in brain MRI acquisition and interpretation across centers represents a major methodological limitation of this study. Because imaging protocols were not standardized and no centralized review was performed, interobserver variability may have influenced the classification of structural abnormalities and could partly explain the absence of a consistent association between MRI findings and growth outcomes. Additional limitations include the 1-year duration of follow-up—which precluded assessment of longer-term outcomes such as near-adult height or pubertal progression—lack of detailed information on rhGH dose adjustments or adherence, and the absence of genetic or GH/IGF-I bioactivity analyses that may affect treatment response. Selection bias is also possible because only 457 of 3,813 enrolled patients had complete MRI and growth data; although excluded patients showed broadly similar baseline characteristics, unmeasured differences cannot be ruled out.
With these considerations in mind, our findings suggest that while brain MRI has clear diagnostic value in IGHD, mild or isolated abnormalities should be interpreted cautiously and are unlikely to provide reliable prognostic information regarding short-term growth response. Despite these methodological constraints, this large multicenter cohort highlights that clinical and biochemical factors remain the primary determinants of early treatment outcomes.
Future work should examine whether integrating genetic and hormonal profiling with imaging can improve prognostic accuracy [27-29], given reports both structural and functional brain changes in GHD [20,30].
In conclusion, this large multicenter study provides region-specific evidence on brain MRI abnormalities in Korean children with IGHD. Although such findings did not independently predict short-term growth response, they remain important for diagnosis and should be interpreted in conjunction with clinical and biochemical data. Incorporating brain MRI results into a broader clinical context may help guide balanced treatment decisions and optimize patient care.
Supplementary materials
Supplementary Tables 1-2 and Supplementary Figs. 1-3 are available at https://doi.org/10.6065/apem.2550296.148.
Comparison of baseline characteristics between included and excluded subjects
Multivariable linear regression analysis predicting absolute height SDS after 1 year of rhGH therapy
Classification of patients by MRI findings. Flowchart illustrating the distribution of IGHD patients with available MRI (n=457). Among them, 335 (73.3%) had normal MRI findings and 122 (26.7%) had abnormal findings. The abnormal group was further categorized into pituitary abnormalities (n=108), extrapituitary abnormalities (n=12), and combined abnormalities (n=2). IGHD, idiopathic growth hormone deficiency; MRI, magnetic resonance imaging.
Change in height SDS from baseline to 1 year, stratified according to sex. Unadjusted analysis shows that the female patients exhibit a significantly greater increase in height SDS than male patients (P=0.038). However, this difference is not significant after adjusting for clinical covariates in the multivariate regression analysis. Bars represent mean±standard deviation. SDS, standard deviation score.
Change in height SDS according to IGHD subtype. Mean height SDS values at baseline and after 1 year of rhGH treatment, stratified according to IGHD subtype. The unadjusted analysis shows no significant differences in growth response between the complete and partial IGHD groups. SDS, standard deviation score; IGHD, idiopathic growth hormone deficiency; rhGH, recombinant human growth hormone.
Notes
Conflicts of interest
JC received research funding from LG Chem. JL received research funding from LG Chem and Dong-A ST. JK received research funding and honoraria from LG Chem, Dong-A ST and Novo Nordisk. Except for that, no potential conflict of interest relevant to this article was reported.
Funding
This study received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Data availability
The data that support the findings of this study can be provided by the corresponding author upon reasonable request.
Acknowledgments
We thank all physicians who contributed patient data to the LG Growth Study. The authors also thank LG Chem, Ltd. for providing statistical support.
Author contributions
Conceptualization: JK; Data curation: JC, JK; Formal analysis: JC, JK; Methodology: JK; Project administration: JK; Visualization: JK; Writing - original draft: JC; Writing - review & editing: JC, JL, JK