The discriminatory performance of insulin-like growth factor 1 (IGF-1) and insulin-like growth factor binding protein-3 (IGFBP-3) was investigated by correlating their values with chronological age (CA), bone age (BA), and pubertal status (PS) for diagnosis of isolated growth hormone deficiency (IGHD).
We evaluated IGF-1 and IGFBP-3 levels in 310 short-stature subjects subdivided into 2 groups: IGHD (n=31) and non-IGHD (n=279). IGF-1 and IGFBP-3 were assayed using immune-radiometric assay and transformed into standard deviation score (SDS) according to CA, BA, and PS.
The highest sensitivity was found in IGF-1-SDS for CA and IGFBP-3-SDS for CA (22.6% and 30.0%, respectively). The highest specificity was found in IGF-1-SDS for PS and IGFBP-3-SDS for PS (98.2% and 94.4%, respectively). Groups with the highest positive predictive values were IGF-1-SDS for BA and IGFBP-3-SDS for BA (10.9% and 5.1%, respectively). Highest negative predictive values were seen in IGF-1-SDS for CA and IGFBP-3-SDS for CA (98.4% and 98.4%, respectively).
IGF-1-SDS for CA, instead of IGF-1-SDS for BA or PS, could be used as a standard variable for IGHD screening. The sufficiently high specificity of IGF-1-SDS for PS suggests that this value is a useful tool for identification of IGHD.
Short stature is defined as height below the 3rd percentile for age and sex. According to Korean Health Analytics Data from the KCDC (Korean Center for Disease Control), 11,938 of 519,062 children (2.3%) were categorized as short stature [
A hormonal cause of short stature termed isolated growth hormone deficiency (IGHD) is caused by abnormalities in the growth hormone (GH)-releasing hormone receptor. IGHD occurs in 1 of 4,000–10,000 individuals [
There is much debate on the validity of the provocative GH test due to its low reproducibility and high false positive rate. Variability of the results among provocative agents and peak GH cutoff value also are under debate. Despite these drawbacks, it is the most widely used diagnostic modality [
This study was a multi-institutional, 10-year retrospective study comprising 310 short-stature patients older than 4 years who visited our institutions from March 2009 to March 2019.
We applied the following inclusion criteria: (1) short stature defined as height below the 3rd percentile for age and sex; (2) BA radiography within 6 months from the provocative GH tests using the Greulich and Pyle method; BA was evaluated by averaging the values reported by 2 pediatric endocrinologists and one radiologist; (3) pubertal staging was established by pediatric endocrinologists at each hospital according to Tanner stage criteria based on breast development in girls and testicular volume in boys; (4) measurement of IGF-1 and IGFBP-3 levels was carried out using immune-radiometric assay (IRMA, Immunotech, Marseilles, France).
We applied the following exclusion criteria: (1) medical history of use of recombinant human GH, gonadotropin releasing hormone analogues, sex steroids or glucocorticoids; (2) endocrine disorders including precocious puberty, hypothyroidism, and Cushing disease; (3) chronic illnesses such as inflammatory bowel disease, celiac disease, occult renal disease, and anemia; or (4) genetic conditions such as Turner syndrome, Noonan syndrome, and short-stature homeoboxcontaining gene (
Two GH stimulation tests were conducted. Patients with a GH level less than 5 ng/mL were classified as GHD. All patients with GH deficiency underwent brain magnetic resonance imaging (MRI) to rule out intracranial cause of GH deficiency. Patients with no abnormal findings on brain MRI were classified as IGHD. L-dopa (49%), arginine (21%), glucagon (14%), insulin (10%), and clonidine (7%) were used as provocative agents.
Measurement of IGF-1 and IGFBP-3 levels was carried out using immune-radiometric assay (IRMA, Immunotech). IRMA is an assay based on reversible and noncovalent binding of an antigen by a specific antibody and labeled with a radioactive nuclide 125I as a tracer. The intra-assay and interassay coefficients of variation were 6.3% and 6.8% for IGF-1, respectively, and 6.0% and 9.5% for IGFBP-3.
In IGF-1 assay, patient sample, dissociation buffer, and tracer are added to the monoclonal antibody-coated tube. After incubation for 1 hour at 18℃–25℃ with shaking, 2 monoclonal antibodies highly specific for IGF-1 but not for IGF-2 bind to 2 epitopes of IGF-1 protein, producing a sandwich-like antibody complex. After incubation, the contents of the tube are removed, and bound radioactivity is measured and is directly proportional to IGF-1 concentration in the sample. The IGF-1 assay requires a 1:100 predilution of serum with an assayspecific diluting agent. Prediluted sample and tracer are added to the coated tubes and incubated for 3 hours at 18℃–25℃ with shaking. An identical procedure was followed for IGFBP-3.
The standard deviation scores (SDSs) for IGF-1 and IGFBP-3 were the corresponding reference values from 2012 data from Korean children and adolescents [
Measurements and calculations were presented as mean±SD. Statistical analysis was performed using IBM SPSS Statistics ver. 21.0 (IBM Co., Armonk, NY, USA).
Comparisons among groups were made using chi-square test or Wilcoxon rank sum test. The variables were tested for normality using the Shapiro-Wilk test. When calculating the sensitivity and specificity, we set the cutoff values of IGF-1-SDS and IGFBP-3-SDS as -2 for diagnosis of IGHD [
A total of 310 subjects with short stature who met the inclusion criteria was included in the study. Mean age was 7.8±3.1 years (males/females, 51%/49%). Average delay of BA was 17.7±13.4 months. The cohort was subdivided into 2 groups of IGHD (n=31) and non-IGHD (n=279).
The clinical characteristics and auxological data of the subjects are described in
Serum IGFBP-3 level was not measured in 11 patients of the non-IGHD group and 1 patient of the IGHD group. Thus, when analyzing IGFBP-3 values, 268 patients of the non-IGHD group and 30 patients of the IGHD group were analyzed. The IGF-1-SDS for CA and IGF-1-SDS for BA were lower among the IGHD subjects (-1.32±0.72 vs. -0.92±0.69 and -0.97±0.87 vs. -0.56±0.75, respectively,
In the IGHD group, IGF-1-SDS for CA, IGF-1-SDS for BA, and IGF-1-SDS for PS were normal in 24 (77.4%), 27 (87.1%), and 29 of the 31 subjects (93.5%), respectively. In the non-IGHD group, they were low in 12 (4.3%), 6 (2.1%), and 5 of the 279 subjects (1.8%), respectively.
The highest sensitivity was observed when IGF-1-SDS for CA and IGFBP-3-SDS for CA were applied (22.6% and 30.0%, respectively). The highest specificity and accuracy were found in IGF-1-SDS for PS (98.2% and 96.4%, respectively) and IGFBP-3-SDS for PS (94.4% and 92.6%, respectively). Groups with the highest PPV were IGF-1-SDS for BA and IGFBP-3-SDS for BA (10.9% and 5.1%, respectively). The highest NPV was measured in IGF-1-SDS for CA and IGFBP-3-SDS for CA (98.4% and 98.4%, respectively). Sensitivity increased in the order of PS<BA<CA for IGF-1-SDS and IGFBP-3-SDS for diagnosis of IGHD. Specificity increased in the order of CA<BA<PS for IGF1-SDS and IGFBP-3-SDS (
When P-values were considered together, growth factors showing statistically significant diagnostic performance were IGF1-SDS for CA and IGF1-SDS for BA (AUC of 0.647 and 0.652, respectively,
We also analyzed the ROC curves of height SDS, BA delay, and BMI SDS. The AUC values of ROC curves for height SDS, BA delay, and BMI SDS were 0.605, 0.518, and 0.625, respectively (P for AUC=0.062, P for AUC=0.738, and P for AUC=0.028, respectively). We constructed multivariable diagnostic models including height SDS, BMI SDS, BA delay, and growth factors. These models showed a higher AUC compared with all other previous models (
This study aimed to compare the discriminatory performance of growth factor levels investigated by CA, BA, and PS for diagnosis of IGHD. The greatest sum of specificity and sensitivity and the highest AUC were exhibited in IGF-1-SDS for CA. The IGF-1-SDS for CA and IGFBP-3-SDS for CA had the highest sensitivity, suggesting their usefulness as a screening test. All IGF-1-SDS and IGFBP-3-SDS values showed high specificity (>85%) but low sensitivity (≤30%). This indicates that the serum values of IGF-1 and IGFBP-3 are more powerful in diagnosing IGHD when the SDS values are low. The specificity of IGF-1-SDS for PS was high at 98.2%. This strongly suggests that short children whose IGF-1-SDS and/or IGFBP-3-SDS are less than -2 have a high probability of having IGHD. Although not completely substitutable for the GH stimulation test, IGF-1 testing might aid in clinical confirmation of IGHD.
According to Inoue-Lima et al. [
There are some factors that explain the low sensitivity of IGF-1-SDS and IGFBP-3-SDS in our study. First, an appropriate cutoff value for SDS is a crucial factor in determining the diagnostic validity. The highest AUCs for IGF-1-SDS for CA, IGF-1-SDS for BA, and IGF-1-SDS for PS were found at SDS cutoff values of -1.23, -1.03, and -1.17, respectively. The sensitivity and specificity of each changed to 61.3% and 68.5%, 51.6% and 77.5%, and 41.9% and 79.6%, respectively. Due to the reduced specificity mentioned above, using the optimal cutoff value of each growth factor is less beneficial. Since each optimal cutoff value is different among 6 growth factor levels, it is less practical when applied in clinical practice.
Second, in addition to GH, there is a multitude of other hormones such as insulin, thyroid hormones, glucocorticoids, and sex steroids that influence production of IGF-1 and IGFBP-3. Also, serum IGF-1 and IGFBP-3 levels are affected by multiple environmental and epidemiologic factors such sex, age, ethnicity, PS, nutritional status, liver insufficiency, hypothyroidism, diabetes mellitus, and systemic illness [
Last, we have to consider the poor diagnostic performance of the provocative GH test. A high false positive rate for the provocative GH test is a known limitation of the assay [
Measurement of spontaneously secreted GH is the most accurate assessment of GH physiology or pathophysiology. However, this method usually is restricted to research use. Therefore, various auxological parameters including height velocity, height SDS, BMI SDS [
We could not include height velocity as a clinical factor for diagnosis of IGHD, though it has long been considered an important factor in evaluation of short stature. However, according to Hintz [
Our result of BMI SDS in the IGHD group was higher than that of the non-IGHD group. We think this is a valid result because IGHD group patients are prone to weight gain compared to the normal population. According to Savastano at el. [
In conclusion, IGF-1-SDS for CA, instead of IGF-1-SDS for BA or PS, could be used as a standard variable for IGHD screening. The sufficiently high specificity of IGF-1-SDS for PS suggests that this value is a useful tool to facilitate identification of IGHD. Further studies are warranted to improve accuracy in diagnosis of IGHD.
This study was approved by the Institutional Review Board of Yeouido St. Mary's Hospital, College of Medicine, The Catholic University of Korea (No. XC20RIDI0004). Need for written informed consent by the patients was waived due to the retrospective nature of this study
No potential conflict of interest relevant to this article was reported.
Comparison of diagnostic value of IGF-1 and IGFBP-3 for isolated growth hormone deficiency. (A) Box-plot chart of IGF-1. (B) Box-plot chart of IGFBP-3. IGF-1, insulin-like growth factor 1; IGFBP-3, insulin-like growth factor binding protein 3; CA, chronological age; BA, bone age; PS, pubertal status; SDS, standard deviation score.
Diagnostic performance of IGF-1 and IGFBP-3 for isolated growth hormone deficiency. (A) ROC plot of IGF-1. (B) ROC plot of IGFBP-3. IGF-1, insulin-like growth factor 1; IGFBP-3, insulin-like growth factor binding protein 3; CA, chronological age; BA, bone age; PS, pubertal status; SDS, standard deviation score; ROC, receiver operator characteristic; AUC, area under the curve.
Clinical characteristics and auxological data of the subjects
Characteristic | Overall (n=310) | Non-IGHD (n=279) | IGHD (n=31) | P-value |
---|---|---|---|---|
Sex | 0.97 | |||
Male | 159 (51.3) | 143 (51.3) | 16 (51.6) | |
Female | 151 (48.7) | 136 (48.7) | 15 (48.4) | |
MPH | 0.969 | |||
Mean±SD | 163.1±7.1 | 163.1±7.2 | 163.1±6.6 | |
Median (5th and 95th percentile range) | 163.5 (151.5–173.5) | 163.5 (151.0–173.5) | 163.0 (151.5–172.0) | |
CA–years | 0.435 | |||
Mean±SD | 7.8±3.1 | 7.8±3.5 | 8.2±2.9 | |
Median (5th and 95th percentile range) | 7.0 (4.0–13.0) | 7.0 (4.0–14.0) | 9.0 (4.0–12.0) | |
BA–years | 0.400 | |||
Mean±SD | 6.5±3.3 | 6.5±3.3 | 6.9±3.1 | |
Median (5th and 95th percentile range) | 6.0 (2.0–12.0) | 6.0 (2.0–12.0) | 7.0 (2.0–11.0) | |
BA delay–months | 0.737 | |||
Mean±SD | 17.7±13.4 | 17.8±13.7 | 16.9±11.4 | |
Median (5th and 95th percentile range) | 16.0 (-2.0 to 42.0) | 16.0 (-2.0 to 45.0) | 15.0 (1.0–39.0) | |
Height SDS | 0.056 | |||
Mean±SD | -2.6±0.5 | -2.5±0.5 | -2.7±0.7 | |
Median (5th and 95th percentile range) | -2.4 (-3.5 to -2.0) | -2.4 (-3.4 to -2.0) | -2.7 (-3.9 to -2.0) | |
BMI SDS | 0.023 | |||
Mean±SD | -0.5±1.5 | -0.6±1.5 | 0.2±1.8 | |
Median (5th and 95th percentile range) | -0.6 (-2.2 to 1.3) | -0.7 (-2.3 to 1.1) | -0.3 (-1.9 to 3.2) | |
Pubertal stages | 0.949 | |||
I | 246 (79.4) | 219 (78.5) | 27 (87.1) | |
II | 38 (12.3) | 35 (12.5) | 3 (9.7) | |
III | 17 (5.5) | 16 (5.7) | 1 (3.2) | |
IV | 6 (1.9) | 6 (2.2) | 0 (0.0) | |
V | 3 (1.0) | 3 (1.1) | 0 (0.0) |
Values are presented as number (%) for categorical variable unless otherwise indicated.
P-value for difference were determined by using chi-square or Wilcoxon rank sum test.
IGHD, isolated growth hormone deficiency; MPH, midparental height; CA, chronological age; BA, bone age; SDS, standard deviation score; BMI, body mass index; SD, standard deviation.
Comparison of the IGF-1 and IGFBP-3 levels
Variable | Overall | Non-IGHD | IGHD | |
---|---|---|---|---|
IGF-1 | (n=310) | (n=279) | (n=31) | |
IGF-1-SDS for CA | 0.007 | |||
Mean±SD | -0.96±0.70 | -0.92±0.69 | -1.32±0.72 | |
Median (5th and 95th percentile range) | -0.95 (-2.06 to 0.08) | -0.93 (-1.95 to 0.11) | -1.32 (-2.29 to -0.22) | |
IGF-1-SDS for BA | 0.006 | |||
Mean±SD | -0.60±0.77 | -0.56±0.75 | -0.97±0.87 | |
Median (5th and 95th percentile range) | -0.63 (-1.84 to 0.54) | -0.61 (-1.74 to 0.54) | -1.04 (-2.11 to 0.62) | |
IGF-1-SDS for PS | 0.134 | |||
Mean±SD | -0.59±0.82 | -0.56±0.82 | -0.80±0.87 | |
Median (5th and 95th percentile range) | -0.68 (-1.65 to 0.92) | -0.67 (-1.65 to 0.92) | -0.90 (-2.26 to 0.96) | |
IGFBP-3 | (n=298) | (n=268) | (n=30) | |
IGFBP-3-SDS for CA | 0.046 | |||
Mean±SD | -0.31±1.75 | -0.26±1.74 | -0.78±1.79 | |
Median (5th and 95th percentile range) | -0.69 (-2.54 to 3.41) | -0.64 (-2.49 to 3.41) | -1.36 (-3.07 to 2.32) | |
IGFBP-3-SDS for BA | 0.067 | |||
Mean±SD | 0.06±1.79 | 0.11±1.78 | -0.39±1.83 | |
Median (5th and 95th percentile range) | -0.29 (-2.22 to 3.70) | -0.26 (-2.14 to 3.70) | -0.99 (-2.32 to 2.81) | |
IGFBP-3-SDS for PS | 0.586 | |||
Mean±SD | -0.01±1.74 | 0.01±1.76 | -0.15±1.58 | |
Median (5th and 95th percentile range) | -0.42 (-2.28 to 3.25) | -0.41 (-2.28 to 3.17) | -0.63 (-1.78 to 3.53) |
P-value for difference were determined by using Wilcoxon rank sum test.
IGF-1, insulin-like growth factor 1; IGFBP-3, insulin-like growth factor binding protein 3; IGHD, isolated growth hormone deficiency; SDS, standard deviation score; SD, standard deviation; CA, chronological age; BA, bone age; PS, pubertal status.
Diagnostic value of IGF-1 and IGFBP-3 for the diagnosis of isolated growth hormone deficiency
Variable | Non-IGHD | IGHD | Sensitivity | Specificity | Accuracy | PPV | NPV |
---|---|---|---|---|---|---|---|
IGF-1 | (n=279) | (n=31) | |||||
IGF-1-SDS for CA | |||||||
≥-2 | 267 | 24 | 22.6 | 95.7 | 94.2 | 9.7 | 98.4 |
<-2 | 12 | 7 | |||||
IGF-1-SDS for BA | |||||||
≥-2 | 273 | 27 | 12.9 | 97.9 | 96.2 | 10.9 | 98.2 |
<-2 | 6 | 4 | |||||
IGF-1-SDS for PS | |||||||
≥-2 | 274 | 29 | 6.5 | 98.2 | 96.4 | 6.8 | 98.1 |
<-2 | 5 | 2 | |||||
IGFBP-3 | (n=268) | (n=30) | |||||
IGFBP-3-SDS for CA | |||||||
≥-2 | 233 | 21 | 30.0 | 86.9 | 85.8 | 4.5 | 98.4 |
<-2 | 35 | 9 | |||||
IGFBP-3-SDS for BA | |||||||
≥-2 | 251 | 25 | 16.7 | 93.7 | 92.1 | 5.1 | 98.2 |
<-2 | 17 | 5 | |||||
IGFBP-3-SDS for PS | |||||||
≥-2 | 253 | 29 | 3.3 | 94.4 | 92.6 | 1.2 | 98.0 |
<-2 | 15 | 1 |
IGF-1, insulin-like growth factor 1; IGFBP-3, insulin-like growth factor binding protein 3; IGHD, isolated growth hormone deficiency; PPV, positive predictive value; NPV, negative predictive value; SDS, standard deviation score; CA, chronological age; BA, bone age; PS, pubertal status.
*For the assessment of the PPV and NPV, we considered a maximum prevalence of IGHD of 2%.
Diagnostic performance of the multivariable models for the diagnosis of isolated growth hormone deficiency
Model | AUC | P-value | Model | AUC | |
---|---|---|---|---|---|
IGF-1 | |||||
IGF1-SDS for CA | 0.647 | 0.013 | IGF-1-SDS for CA + height-SDS + BMI-SDS + BA delay | 0.731 | 0.006 |
IGF1-SDS for BA | 0.652 | 0.012 | IGF-1-SDS for BA + height-SDS + BMI-SDS + BA delay | 0.713 | 0.030 |
IGF1-SDS for PS | 0.582 | 0.178 | IGF-1-SDS for PS + height-SDS + BMI-SDS + BA delay | 0.695 | 0.017 |
IGFBP-3 | |||||
IGFBP3-SDS for CA | 0.611 | 0.063 | IGFBP-3-SDS for CA + height-SDS + BMI-SDS + BA delay | 0.716 | 0.031 |
IGFBP3-SDS for BA | 0.602 | 0.094 | IGFBP-3-SDS for BA + height-SDS + BMI-SDS + BA delay | 0.713 | 0.027 |
IGFBP3-SDS for PS | 0.530 | 0.594 | IGFBP-3-SDS for PS + height-SDS + BMI-SDS + BA delay | 0.714 | 0.006 |
AUC, area under curve; IGF-1, insulin-like growth factor 1; IGFBP-3, insulin-like growth factor binding protein 3; IGHD, isolated growth hormone deficiency; SDS, standard deviation score; CA, chronological age; BA, bone age; PS, pubertal status; BMI, body mass index.
P-value indicates the significance level for AUC=0.5
P-value for difference in AUC were determined by using the DeLong test.