- Email: [email protected]
Efficacy of Growth Hormone Treatment in Children with Type 1 Diabetes Mellitus and Growth Hormone Deficiency—An Analysis of KIGS Data
ARTICLE IN PRESS THE JOURNAL OF PEDIATRICS • www.jpeds.com
ORIGINAL ARTICLES
Efficacy of Growth Hormone Treatment in Children with Type 1 Diabetes Mellitus and Growth Hormone Deficiency—An Analysis of KIGS Data Walter Bonfig, MD1,2, Anders Lindberg3, Martin Carlsson3, Wayne Cutfield, MD, PhD4, David Dunger, MD, PhD5, Cecilia Camacho-Hübner, MD, PhD6, and Reinhard W. Holl, MD, PhD7,8 Objective To analyze first-year treatment growth response and growth hormone (GH) dosage in prepubertal patients with the combination of type 1 diabetes mellitus (T1DM) and growth hormone deficiency (GHD). Study design A total of 69 patients with T1DM and GHD treated with GH have been enrolled in KIGS (Pfizer International Growth Database). Of these, 24 prepubertal patients had developed T1DM before GHD and were included in this analysis. Of 30 570 patients with GHD without T1DM, 15 024 were prepubertal and served as controls. Values are expressed as mean ± SD. Results Patients with T1DM and GHD had similar characteristics compared with the GHD-alone group. Neither age (10.2 ± 3.13 vs 8.42 ± 3.46 years, P = .14), height SDS corrected for midparental height SDS at start of treatment (−1.62 ± 1.38 vs −1.61 ± 1.51, P = .80), nor GH dosage (0.24 ± 0.08 mg/kg/wk vs 0.20 ± 0.04 mg/kg/wk, P = .09) were different between those with and without T1DM. First-year catch-up growth was comparable between the 2 patient groups (first treatment year height velocity 7.54 ± 3.11 cm/year compared with 8.35 ± 2.54 cm/year in control patients, P = .38). Height SDS of children with T1DM and GHD improved from −2.62 ± 1.04 to −1.88 ± 1.11 over 1 year of GH treatment. Conclusion Short-term response to GH therapy appeared similar in subjects with T1DM who then developed GHD and in those with GHD alone. Thus, T1DM does not appear to compromise GH response in children with GHD and should not exclude GH treatment in these children. GH treatment was safe in both subgroups of patients. (J Pediatr 2018;■■:■■-■■). ith current criteria, the prevalence of growth hormone deficiency (GHD) is between 1:3500 and 1:8700.1 GHD already may be present in neonates, if caused by genetic disorders, but the average age at diagnosis is 6-8 years.1 Although rare, it is important to establish early diagnosis of GHD, as a missed or very late diagnosis may result in a poor height outcome. In most cases, GHD represents a relative lack of growth hormone (GH) secretion, leading to decreased growth velocity, retardation of bone maturation, and short stature. GH also plays an important role in glucose, lipid, and protein metabolism.2 Both GHD and GH excess are associated with disturbances of carbohydrate metabolism. GH decreases glucose oxidation and glucose uptake by muscle and increases gluconeogenesis, resulting in “insulin antagonist effects.”3 The growth-promoting effects of GH are mediated through the insulin-like growth factors (mainly insulin-like growth factor-I [IGF-I]), which are synthesized and secreted by the liver, as well as in target tissues. Insulin-like growth factors are bound to insulin-like growth factor binding proteins, with insulin-like growth factor-binding protein 3 (IGFBP-3) being the major one. IGF-I, IGFBP-3, and the acid labile subunit form a ternary complex extending the half-life of IGF-I.3 Adequate insulin secretion and portal insulin concentrations are needed to support normal serum concentrations of IGFs and IGFBPs, because insulin modulates hepatic GH receptor expression. Portal insulin deficiency leads to GH hypersecretion. Despite GH hypersecretion, circulating concentrations of IGF-I and IGFBP-3 are low, and concentrations of insulin-like growth From the 1Department of Pediatrics, Klinikum Welsfactor-binding protein 1—a major negative regulator of IGF-I bioactivity—is high Grieskirchen, Wels, Austria; 2Department of Pediatrics, 3 Technical University München, Munich, Germany; 3Pfizer in the state of insulin deficiency. Health AB, Sollentuna, Sweden; 4Liggins Institute, The incidence of type 1 diabetes mellitus (T1DM) is increasing, especially in University of Auckland, Auckland, New Zealand; 5Department of Pediatrics and the Wellcome Trust-MRC children aged <5 years.4 Early-diabetes onset and mean hemoglobin A1c >7.0% Institute of Metabolic Science, University of Cambridge, Cambridge, United Kingdom; 6Pfizer Inc, Endocrine Care, (>53 mmol/mol) correlate negatively with adult height.5 Therefore, workup of short New York, NY; 7Institute of Epidemiology and Medical
W
Biometry, University of Ulm, Ulm; and 8German Center for Diabetes Research (DZD), Munich-Neuherberg, Germany
AE DPV GH GHD IGFBP-3 IGF-I SAE T1DM
Adverse event Diabetes Prospektive Verlaufsdokumentation Growth hormone Growth hormone deficiency Insulin-like growth factor-binding protein 3 Insulin-like growth factor-I Serious adverse event Type 1 diabetes mellitus
W.C. received consultancy fees from Pfizer and is a member of the KIGS Steering Committee (SC). D.D. was a former member of the KIGS SC. A.L, M.C., and C.C.H. are employees of Pfizer. The other authors declare no conflicts of interest. Portions of this study were presented at the 55th Annual European Society of Paediatric Endocrinology Meeting, September 10-12, 2016, Paris, France. 0022-3476/$ - see front matter. © 2018 Elsevier Inc. All rights reserved. https://doi.org10.1016/j.jpeds.2018.02.035
1 FLA 5.5.0 DTD ■ YMPD9819_proof ■ April 12, 2018
THE JOURNAL OF PEDIATRICS • www.jpeds.com stature and diagnosis of GH deficiency in children with diabetes might become more frequent in the future. Very few data regarding GH treatment in children who are GH deficient with T1DM have been published.6,7 In a previous study, we showed that GH treatment is safe if the insulin dosage is adjusted accordingly.6 In that study, decreased efficacy of GH treatment was observed; no data on GH dosage and IGF-I concentrations were available. Therefore, our aim was to analyze first-year treatment growth response and GH dosage in prepubertal patients with T1DM and GHD and to compare these data with a large control cohort within the KIGS database (Pfizer International Growth Database).
Methods Patient data were retrieved from KIGS (Pfizer International Growth Database). KIGS is a worldwide observational registry established in 1987 to monitor the outcome and safety of treatment with GH (Genotropin; Pfizer Inc, New York, New York). It is conducted in accordance with the Declaration of Helsinki. Patients enrolled in KIGS are classified according to the primary cause of short stature.8 Data collection and entry were performed by KIGS investigators. In this study, children with T1DM who then developed idiopathic GHD were compared with those with GHD alone during the first year of treatment. GHD was defined as a peak stimulated GH level <10 µg/ L. All subjects were prepubertal with Tanner breast stage B1 in girls and testes volume ≤3 mL in boys during the first year of GH therapy. In total, 69 patients with T1DM and idiopathic GH deficiency treated with GH are documented in KIGS. Of these 24, developed T1DM before GHD and were prepubertal during the first year of GH treatment and were included for analysis. Of 30 570 control patients with the diagnosis of GHD, 15 934 were prepubertal and served as controls. Of these, 13 010
Volume ■■
•
■■ 2018
control patients (80.5%) had isolated GHD and 2924 control patients (19.5%) had multiple pituitary hormone deficiency. Because treatment and control groups significantly differed in birth weight SDS, height, and weight SDS at start of treatment (Table I), we also performed a matched-pairs analysis. A propensity score–matched cohort of patients with T1DM and GHD and GHD alone was created by using 5 variables collected at start of GH therapy (height SDS, weight SDS, age, body mass index SDS, and start of GH therapy) to achieve matching balance between the 2 cohorts. Then, 1:2 matching was applied (2 controls per case) to minimize the selection bias and to increase the number of subjects in the control group from 24 to 48. To include as many patients with T1DM and GHD in the analysis, a second matched-pairs analysis was performed also including pubertal patients. Therefore, matching for pubertal status, GH peak, and midparental height SDS became necessary to achieve balanced matched populations. Complete matching confounder covariates were available in 45 patients with T1DM and GHD. Of these 45 matched patients, 1-year data were available in 33 subjects. Quality of diabetes control was not assessed in this analysis because of insufficient longitudinal data on hemoglobin A1c in both groups. Standing height was measured approximately 6 monthly with a wall-mounted stadiometer. Height SDS was calculated by using reference data from Prader et al.9 GH dosage was calculated as milligram per kilogram of body weight per week (mg/kg/wk). Statistical analyses (descriptive data analysis, calculation of SDS, and Wilcoxon rank-sum test) were carried out with SAS software (SAS, Version 9.2; SAS Institute, Cary, North Carolina). A 2-sided significance level of 5% was applied to all statistical tests. Growth velocity data were adjusted for age and sex. Values are expressed as mean with SDs unless otherwise stated.
Table I. Clinical characteristics and response to the first year of GH treatment in treatment and control group. T1DM and GHD Variables Background Birth weight SDS MPH SDS Max GH peak, µg/L Start of GH therapy Chronological age, y Height SDS Height—MPH SDS Weight SDS BMI SDS GH dose, mg/kg/wk 1 y on GH therapy Height velocity, cm/y Height SDS Delta height SDS Weight SDS BMI SDS GH dose, mg/kg/wk
GHD only
n
Median
Mean ± SD
n
Median
Mean ± SD
P
19 19 22
0.06 −0.23 6.93
−0.02 ± 1.15 −0.96 ± 1.32 6.80 ± 3.22
13 582 14 127 15 024
−0.80 −1.40 6.10
−0.81 ± 1.23 −1.36 ± 1.24 5.75 ± 2.75
.003 .111 .208
24 24 19 24 24 24
10.20 −2.62 −1.62 −1.55 0.04 0.24
9.39 ± 3.13 −2.58 ± 1.04 −1.59 ± 1.38 −1.45 ± 1.22 0.13 ± 1.09 0.23 ± 0.08
15 15 14 15 15 15
024 024 127 024 024 024
8.42 −3.01 −1.61 −2.18 −0.32 0.20
8.40 ± 3.46 −3.13 ± 1.15 −1.76 ± 1.51 −2.25 ± 1.47 −0.32 ± 1.27 0.22 ± 0.07
.144 .032 .802 .006 .084 .089
24 24 24 24 24 24
7.54 −1.88 0.57 −1.06 −0.05 0.23
8.16 ± 3.11 −1.90 ± 1.11 0.70 ± 0.55 −1.01 ± 1.27 0.11 ± 1.14 0.24 ± 0.08
15 15 15 14 14 15
024 024 024 935 935 024
8.35 −2.30 0.69 −1.69 −0.40 0.20
8.67 ± 2.54 −2.36 ± 1.10 0.78 ± 0.51 −1.73 ± 1.35 −0.39 ± 1.21 0.21 ± 0.07
.375 .058 .381 .017 .077 .041
BMI, body mass index; MPH, midparental height.
2
Bonfig et al FLA 5.5.0 DTD ■ YMPD9819_proof ■ April 12, 2018
■■ 2018
ORIGINAL ARTICLES
For the matching, we used logistic regression analysis to create a propensity score. We then used the algorithm developed by Parsons to perform greedy nearest neighbor matching without replacement. It is recommended that the optimal number of control subjects matched to case subject should not exceed 2; therefore, 1:2 matching was applied. The variables for the matched cohorts were compared at start of GH therapy and after 1 year on GH therapy. Background variables were evaluated as well. The quality of the propensity scores estimated in a sample were evaluated by using 2 types of comparisons: comparing the distributions of the propensity scores across the case and control groups and comparing the distributions of each covariate across the 2 groups. The distributions of the propensity scores was compared with the Kolmogorov–Smirnov statistical test applied at the 5% level of significance to test for any difference between the cumulative distribution functions for the 2 cohorts. The evaluation of balance was done by comparing the distribution of each covariate across the 2 groups at start of GH therapy and for the background variables by using a Wilcoxon test. The same test was applied on the variables collected after 1 year on GH therapy.
Results Of 24 prepubertal subjects (17 males, 70.8%) with T1DM and GHD, 6 had multiple pituitary hormone deficiencies and 1 child had celiac disease. All subjects were euthyroid, and none were treated with glucocorticoids (no corticotropin-releasing hormone/adrenocorticotropic hormone deficiency in patients with multiple pituitary hormone deficiency). All 24 children were diagnosed with T1DM first and subsequently with GHD. Median and mean maximum GH peak during GH stimulation testing was 6.9 µg/L and 6.8 µg/L, respectively. The median time interval between diagnosis of T1DM and start of GH therapy was 4.8 years (mean 5.2 years, SD 3.4 years). Of 15 024 prepubertal control subjects (10 646 males, 70.9%) without diabetes, 2924 subjects (19.5%) had multiple pituitary hormone deficiency and 678 children (4.5%) also were treated with glucocorticoids. All subjects were euthyroid. The clinical characteristics and response to the first year of GH treatment in prepubertal children with T1DM and GHD compared with those with GHD alone are summarized in Table I. Subjects with T1DM and GHD had a statistically significant greater birth weight than subjects with GHD alone (median birth weight SDS 0.06 [mean –0.02, SD 1.15] vs −0.80 [mean −0.81, SD 1.23], P = .003) and were also heavier at start of GH treatment (median weight SDS −1.55 [mean −1.45, SD 1.22] vs −2.18 [mean −2.25, SD 1.47], P = .006). Age at start of GH treatment was not statistically different between children with combined T1DM and GHD (median age at start of therapy 10.2 years [mean 9.39 years, SD 3.13]) and children with GHD alone (median age at start of therapy 8.42 years [mean 8.40 years, SD 3.46]), P = .14. There were no significant differences between the T1DM and GHD vs the GHD alone group at start of treatment with regard to median height SDS corrected for midparental height SDS
Table II. Matched pairs analysis for prepubertal patients: comparison of matched cohorts at background, start of GH therapy, and after 1 year on GH therapy with the Wilcoxon rank-sum test Matched-pair controls “prepubertal GHD only” Variables Background and at start of GH therapy Birth weight SDS MPH SDS Max GH peak, µg/L Chronological age Height SDS Height—MPH SDS Weight SDS BMI SDS GH dose, mg/kg/wk After 1 y on GH therapy Height velocity, cm/y Height SDS Delta height SDS Weight SDS BMI SDS GH dose, mg/kg/wk
Prepubertal T1DM and GHD
n
Median
n
Median
P
43 47 48 48 48 47 48 48 48
−0.48 −1.02 6.8 10.43 −2.61 −1.62 −1.73 −0.17 0.21
19 19 22 24 24 19 24 24 24
0.06 −0.23 6.93 10.2 −2.62 −1.62 −1.55 0.04 0.24
.1866 .5954 .6670 .5150 .8344 .8318 .8907 .7973 .3964
48 48 48 48 48 48
8.25 −2.01 0.65 −1.12 0.07 0.21
24 24 24 24 24 24
7.54 −1.88 0.57 −1.06 −0.05 0.23
.3244 .7698 .3303 .8344 .9002 .2393
(−1.62 vs −1.61, P = .80), mean GH dosage (0.23 mg/kg/wk vs 0.22 mg/kg/wk, P = .09), and first-year catch-up growth expressed as height velocity (8.2 cm/y compared with 8.7 cm/y, P = .38). After 1 year of treatment, GH dosage was greater in subjects with T1DM and GHD (0.24 mg/kg/wk vs 0.21 mg/ kg/wk, P = .04). Height SDS of children with T1DM and GHD improved from −2.62 (mean −2.58, SD 1.04) to −1.88 (mean −1.90, SD 1.11), P < .001. Accordingly, there was also no difference between the 2 groups in the change of height SDS after 1 year of GH treatment, P = .38. All data are shown in Table I. Because treatment and control group significantly differed in birth weight SDS and height and weight SDS at start of treatment (Table I), we also performed a matched pairs analysis with 1:2 matching (24 patients with T1DM and GHD and 48 patients with GHD alone). The distributions of the propensity scores was compared with Kolmogorov–Smirnov test, P > .90. The evaluation of balance was done by comparing the distribution of each covariate across the 2 groups at start of GH therapy and for the background variables. All P values were >.05, indicating that a proper balance had been achieved between the 2 cohorts, as displayed in Table II. The 2 groups were evaluated after 1 year on GH therapy. No significant differences between the 2 cohorts were identified (P > .05), demonstrating that there was no difference in first-year growth response between the 2 groups. To include the maximum number of patients with T1DM and GHD, in a second matched-pairs analysis, pubertal patients were included. Complete confounder covariates were available in 45 of the initially identified 69 subjects. Of these 45 subjects, 1-year data were only available in 33 subjects.
Efficacy of Growth Hormone Treatment in Children with Type 1 Diabetes Mellitus and Growth Hormone Deficiency—An Analysis of KIGS Data FLA 5.5.0 DTD ■ YMPD9819_proof ■ April 12, 2018
3
THE JOURNAL OF PEDIATRICS • www.jpeds.com Table III. Matched pairs analysis including pubertal patients: comparison of matched cohorts at background, start of GH therapy, and after 1 year on GH therapy with the Wilcoxon rank-sum test* Matched-pair controls including pubertal patients “GHD only” Variables Background and at start of GH therapy Birth weight SDS MPH SDS Max GH peak, µg/L Chronological age Height SDS Height—MPH SDS Weight SDS BMI SDS GH dose, mg/kg/wk After 1 y on GH therapy Height velocity, cm/y Height SDS Delta height SDS Weight SDS BMI SDS GH dose, mg/kg/wk
T1DM and GHD including pubertal patients
n
Median
n
Median
P
80 90 90 90 90 90 90 90 90
−0.42 −0.85 6.97 11.8 −2.42 −1.55 −1.2 0.29 0.21
38 45 45 45 45 45 45 45 45
−0.23 −0.52 6.8 11.6 −2.6 −1.94 −1.2 0.22 0.23
.4823 .6290 .6076 .8064 .5786 .4751 .9851 .6575 .3789
77 77 77 77 77 77
8.54 −1.87 0.59 −0.92 0.08 0.21
33 33 33 33 33 33
7.64 −1.9 0.55 −1.05 −0.19 0.23
.1687 .5838 .7691 .7394 .6763 .1269
*One-year data only available in 33 patients with T1DM and GHD.
Results of this second matched pairs analysis are shown in Table III; again, there was no statistical difference between the 2 cohorts, including comparable first-year growth response in subjects with T1DM and GHD and control patients (Table III). There were 11 adverse events (AEs) reported in the T1DM and GHD group (1 event each of headache, acute sinusitis, naevi, allergic reaction, contusion of the chest wall, knee injury, and foreign body on external eye), and they were not related to GH treatment as reported by the investigators. All patients recovered. In addition, there were 3 AEs related to T1DM, namely diabetic nephropathy and the beginning of diabetic retinopathy, hypoglycemia, and worsening of metabolic control. Only the latter was considered related to GH treatment; the GH dose was reduced. One serious adverse event (SAE) was reported in a patient who was admitted to the hospital with acute pancreatitis, most likely in the context of diabetic ketoacidosis. The patient fully recovered, and the SAE was reported not to be related to GH treatment by the investigator. In 28 258 control subjects, 6808 AEs and 684 SAEs were reported by the investigators. Increased blood glucose/ hyperglycemia were reported in 16 control subjects: 14 subjects were diagnosed with T1DM, and 2 subjects were diagnosed with type 2 diabetes on GH therapy.
Discussion Analyzing data from the KIGS database, we found that shortterm GH treatment and response to therapy over 1 year in prepubertal children with T1DM and GHD were similar to those
Volume ■■ with GHD alone. Children with T1DM and GHD were treated with adequate GH doses, and after 1 year of treatment, the GH doses of children with T1DM and GHD were even greater compared with subjects with GHD alone. With the “borderline” median GH peak in GH stimulation tests both in the treatment and in the control group, children with idiopathic short stature and not truly GHD may have been included but still had a comparable growth response to GH treatment. In a previous study, we evaluated the metabolic effect of GH treatment in children with GH deficiency with T1DM in the Diabetes Prospektive Verlaufsdokumentation (DPV) database.6 We were able to show that GH treatment leads to a significantly greater insulin demand, but with adequate adjustment of insulin doses, metabolic control did not worsen during GH treatment in patients with T1DM. In that study, however, we observed that the growth response to 2 years of GH treatment in children with T1DM and GHD appeared reduced. Metabolic control (hemoglobin A1c) was statistically associated with growth response to GH treatment6; however, there was no information on GH dosage. We speculated that children with T1DM and GHD might have been treated with low GH doses because of the impact of GH on carbohydrate metabolism. For this reason, we decided to perform a similar analysis in KIGS. In this analysis, prepubertal children with T1DM and GHD were treated with comparable start GH doses and showed a comparable initial growth response to GH treatment compared with those with GHD alone. There was a trend toward older age at start of GH treatment in children with T1DM and GHD, as observed in the DPV cohort. Physicians may be hesitant to perform GH stimulation testing in short children with T1DM, attributing poor growth to poor diabetes control. As a result, primarily optimal metabolic control is attempted in diabetic children with growth failure instead of testing for other etiologies. In an Italian survey of 42 pediatric endocrine centers, 69% of centers stated that they usually avoid combined treatment with insulin and GH.7 In the Italian cohort, only 17 patients with T1DM and GH treatment were identified. GH treatment was prescribed for various indications (GHD, Turner syndrome, short stature after intrauterine growth retardation), and GH therapy was effective in promoting growth in most patients. Insulin requirements rose from 0.7 to 1.0 IU/ kg/d during GH treatment without any deterioration of glycemic control.7 A similar rise in insulin requirements was observed in the DPV cohort.6 Unfortunately, in the KIGS database, data on insulin dosage and hemoglobin A1C were insufficient to analyze. In patients with diabetes, hepatic GH resistance has been described previously.3,10,11 This GH resistance is a risk factor for decreased GH efficacy. Therefore, IGF-I concentrations would be of great interest in subjects with diabetes treated with GH. Unfortunately, longitudinal IGF-I data were insufficient for analysis in both the DPV and the KIGS databases. There are 2 large studies published12,13 that report an increased incidence of type 2 diabetes in children and adolescents and also in adults treated with GH. Child et al compared the incidence of type 1 and type 2 diabetes in patients aged
4
Bonfig et al FLA 5.5.0 DTD ■ YMPD9819_proof ■ April 12, 2018
■■ 2018
ORIGINAL ARTICLES
<20 years in the Genetics and Neuroendocrinology of Short Stature International Study (GeNeSIS) observational research program in the US SEARCH for Diabetes in Youth study.13 An increased incidence was only found for type 2, but not for type 1, diabetes in children treated with GH. Therefore, monitoring of glucose metabolism is recommended during GH treatment, especially in patients with pre-existing risk factors such as obesity. In a study by Cutfield et al, GH treatment did not affect the incidence of type 1 diabetes in any age group.14 Prepubertal children with T1DM and GHD display a similar initial growth response to GH treatment compared with those with GHD alone. Therefore, taken together with previous data, GH treatment is safe and effective in these patients. Because KIGS and DPV are observational studies, reconfirmation of these findings in multicenter studies should be pursued. Preexisting T1DM should not be a reason to deny GH treatment to children with GHD. ■ We thank all participating investigators for data collection and entry into KIGS (Pfizer International Growth Database). Submitted for publication Jul 27, 2017; last revision received Jan 23, 2018; accepted Feb 13, 2018 Reprint requests: Walter Bonfig, MD, Department of Pediatrics, Klinikum WelsGrieskirchen, Grieskirchner Str 42, Wels A-4600. E-mail: walter.bonfig@ mri.tum.de
References 1. Ranke M, Reiter EO, Price DA. Idiopathic growth hormone deficiency in KIGS: selected aspects. In: Ranke M, Price DA, Reiter EO, eds. Growth hormone therapy in pediatrics–20 years of KIGS. 1st ed. Basel: Karger; 2007, p. 116-35. 2. Moller N, Jorgensen JO. Effects of growth hormone on glucose, lipid and protein metabolism in human subjects. Endocr Rev 2009;30:15277.
3. Chiarelli F, Giannini C, Mohn A. Growth, growth factors and diabetes. Eur J Endocrinol 2004;151:109-17. 4. Ehehalt S, Dietz K, Willasch AM, Neu A. for the DIARY-Group BadenWuerttemberg. Prediction model for the incidence and prevalence of type 1 diabetes in childhood and adolescence: evidence for a cohort-dependent increase within the next two decades in Germany. Pediatr Diabetes 2007;13:15-20. 5. Bonfig W, Kapellen T, Dost A, Fritsch M, Rohrer T, Wolf J, et al. Growth in children with type 1 diabetes. J Pediatr 2012;160:900-3. 6. Bonfig W, Molz K, Woelfle J, Hofer SE, Hauffa BP, Schoenau E, Diabetes Patienten Verlaufsdocumentationsystem Initiative of the German Working Group for Pediatric Diabetology, et al. Metabolic safety of growth hormone in type 1 diabetes and idiopathic growth hormone deficiency. J Pediatr 2013;163:1095-8. 7. Zucchini S, Iafusco D, Vannelli S, Rabbone I, Salzano G, Pozzobon G, et al. Combined therapy with insulin and growth hormone in 17 patients with type-1 diabetes and growth disorders. Horm Res Paediatr 2014;82:53-8. 8. Ranke MB. The KIGS aetiology classification system. In: Ranke MB, Price DA, Reiter EO, eds. Growth hormone therapy in pediatrics—20 years of KIGS. 1st ed. Basel: Karger; 2007, p. 29-37. 9. Prader A, Largo RH, Molinari L, Issler C. Physical growth of Swiss children from birth to 20 years of age. First Zurich longitudinal study of growth and development. Helv Paediatr Acta 1989;52:1-125. 10. Mercado M, Baumann G. Characteristics of the somatotropic axis in insulin dependent diabetes mellitus. Arch Med Res 1995;26:101-9. 11. Holl RW, Siegler B, Scherbaum WA, Heinze E. The serum growth hormone-binding protein is reduced in young patients with insulindependent diabetes mellitus. J Clin Endocrinol Metab 1993;76:165-7. 12. Luger A, Mattsson AF, Koitowska-Hagstrom M, Thunander M, Goth M, Verhelst J, et al. Incidence of diabetes mellitus and evolution of glucose parameters in growth-hormone deficient subjects during growth hormone replacement therapy: a long-term observational study. Diabetes Care 2012;35:57-62. 13. Child CJ, Zimmermann AG, Scott RS, Cutler GB, Battelino T, Blum WF, et al. Prevalence and incidence of diabetes mellitus in GH-treated children and adolescents: analysis from the GeNeSIS observational research program. J Clin Endocrinol Metab 2011;96:1025-34. 14. Cutfield WS, Wilton P, Bennmarker H, Albertsson-Wikland K, Chatelain P, Ranke MB, et al. Incidence of diabetes mellitus and impaired glucose tolerance in children and adolescents receiving growth-hormone treatment. Lancet 2000;355:610-3.
Efficacy of Growth Hormone Treatment in Children with Type 1 Diabetes Mellitus and Growth Hormone Deficiency—An Analysis of KIGS Data FLA 5.5.0 DTD ■ YMPD9819_proof ■ April 12, 2018
5
ORIGINAL ARTICLES
Efficacy of Growth Hormone Treatment in Children with Type 1 Diabetes Mellitus and Growth Hormone Deficiency—An Analysis of KIGS Data Walter Bonfig, MD1,2, Anders Lindberg3, Martin Carlsson3, Wayne Cutfield, MD, PhD4, David Dunger, MD, PhD5, Cecilia Camacho-Hübner, MD, PhD6, and Reinhard W. Holl, MD, PhD7,8 Objective To analyze first-year treatment growth response and growth hormone (GH) dosage in prepubertal patients with the combination of type 1 diabetes mellitus (T1DM) and growth hormone deficiency (GHD). Study design A total of 69 patients with T1DM and GHD treated with GH have been enrolled in KIGS (Pfizer International Growth Database). Of these, 24 prepubertal patients had developed T1DM before GHD and were included in this analysis. Of 30 570 patients with GHD without T1DM, 15 024 were prepubertal and served as controls. Values are expressed as mean ± SD. Results Patients with T1DM and GHD had similar characteristics compared with the GHD-alone group. Neither age (10.2 ± 3.13 vs 8.42 ± 3.46 years, P = .14), height SDS corrected for midparental height SDS at start of treatment (−1.62 ± 1.38 vs −1.61 ± 1.51, P = .80), nor GH dosage (0.24 ± 0.08 mg/kg/wk vs 0.20 ± 0.04 mg/kg/wk, P = .09) were different between those with and without T1DM. First-year catch-up growth was comparable between the 2 patient groups (first treatment year height velocity 7.54 ± 3.11 cm/year compared with 8.35 ± 2.54 cm/year in control patients, P = .38). Height SDS of children with T1DM and GHD improved from −2.62 ± 1.04 to −1.88 ± 1.11 over 1 year of GH treatment. Conclusion Short-term response to GH therapy appeared similar in subjects with T1DM who then developed GHD and in those with GHD alone. Thus, T1DM does not appear to compromise GH response in children with GHD and should not exclude GH treatment in these children. GH treatment was safe in both subgroups of patients. (J Pediatr 2018;■■:■■-■■). ith current criteria, the prevalence of growth hormone deficiency (GHD) is between 1:3500 and 1:8700.1 GHD already may be present in neonates, if caused by genetic disorders, but the average age at diagnosis is 6-8 years.1 Although rare, it is important to establish early diagnosis of GHD, as a missed or very late diagnosis may result in a poor height outcome. In most cases, GHD represents a relative lack of growth hormone (GH) secretion, leading to decreased growth velocity, retardation of bone maturation, and short stature. GH also plays an important role in glucose, lipid, and protein metabolism.2 Both GHD and GH excess are associated with disturbances of carbohydrate metabolism. GH decreases glucose oxidation and glucose uptake by muscle and increases gluconeogenesis, resulting in “insulin antagonist effects.”3 The growth-promoting effects of GH are mediated through the insulin-like growth factors (mainly insulin-like growth factor-I [IGF-I]), which are synthesized and secreted by the liver, as well as in target tissues. Insulin-like growth factors are bound to insulin-like growth factor binding proteins, with insulin-like growth factor-binding protein 3 (IGFBP-3) being the major one. IGF-I, IGFBP-3, and the acid labile subunit form a ternary complex extending the half-life of IGF-I.3 Adequate insulin secretion and portal insulin concentrations are needed to support normal serum concentrations of IGFs and IGFBPs, because insulin modulates hepatic GH receptor expression. Portal insulin deficiency leads to GH hypersecretion. Despite GH hypersecretion, circulating concentrations of IGF-I and IGFBP-3 are low, and concentrations of insulin-like growth From the 1Department of Pediatrics, Klinikum Welsfactor-binding protein 1—a major negative regulator of IGF-I bioactivity—is high Grieskirchen, Wels, Austria; 2Department of Pediatrics, 3 Technical University München, Munich, Germany; 3Pfizer in the state of insulin deficiency. Health AB, Sollentuna, Sweden; 4Liggins Institute, The incidence of type 1 diabetes mellitus (T1DM) is increasing, especially in University of Auckland, Auckland, New Zealand; 5Department of Pediatrics and the Wellcome Trust-MRC children aged <5 years.4 Early-diabetes onset and mean hemoglobin A1c >7.0% Institute of Metabolic Science, University of Cambridge, Cambridge, United Kingdom; 6Pfizer Inc, Endocrine Care, (>53 mmol/mol) correlate negatively with adult height.5 Therefore, workup of short New York, NY; 7Institute of Epidemiology and Medical
W
Biometry, University of Ulm, Ulm; and 8German Center for Diabetes Research (DZD), Munich-Neuherberg, Germany
AE DPV GH GHD IGFBP-3 IGF-I SAE T1DM
Adverse event Diabetes Prospektive Verlaufsdokumentation Growth hormone Growth hormone deficiency Insulin-like growth factor-binding protein 3 Insulin-like growth factor-I Serious adverse event Type 1 diabetes mellitus
W.C. received consultancy fees from Pfizer and is a member of the KIGS Steering Committee (SC). D.D. was a former member of the KIGS SC. A.L, M.C., and C.C.H. are employees of Pfizer. The other authors declare no conflicts of interest. Portions of this study were presented at the 55th Annual European Society of Paediatric Endocrinology Meeting, September 10-12, 2016, Paris, France. 0022-3476/$ - see front matter. © 2018 Elsevier Inc. All rights reserved. https://doi.org10.1016/j.jpeds.2018.02.035
1 FLA 5.5.0 DTD ■ YMPD9819_proof ■ April 12, 2018
THE JOURNAL OF PEDIATRICS • www.jpeds.com stature and diagnosis of GH deficiency in children with diabetes might become more frequent in the future. Very few data regarding GH treatment in children who are GH deficient with T1DM have been published.6,7 In a previous study, we showed that GH treatment is safe if the insulin dosage is adjusted accordingly.6 In that study, decreased efficacy of GH treatment was observed; no data on GH dosage and IGF-I concentrations were available. Therefore, our aim was to analyze first-year treatment growth response and GH dosage in prepubertal patients with T1DM and GHD and to compare these data with a large control cohort within the KIGS database (Pfizer International Growth Database).
Methods Patient data were retrieved from KIGS (Pfizer International Growth Database). KIGS is a worldwide observational registry established in 1987 to monitor the outcome and safety of treatment with GH (Genotropin; Pfizer Inc, New York, New York). It is conducted in accordance with the Declaration of Helsinki. Patients enrolled in KIGS are classified according to the primary cause of short stature.8 Data collection and entry were performed by KIGS investigators. In this study, children with T1DM who then developed idiopathic GHD were compared with those with GHD alone during the first year of treatment. GHD was defined as a peak stimulated GH level <10 µg/ L. All subjects were prepubertal with Tanner breast stage B1 in girls and testes volume ≤3 mL in boys during the first year of GH therapy. In total, 69 patients with T1DM and idiopathic GH deficiency treated with GH are documented in KIGS. Of these 24, developed T1DM before GHD and were prepubertal during the first year of GH treatment and were included for analysis. Of 30 570 control patients with the diagnosis of GHD, 15 934 were prepubertal and served as controls. Of these, 13 010
Volume ■■
•
■■ 2018
control patients (80.5%) had isolated GHD and 2924 control patients (19.5%) had multiple pituitary hormone deficiency. Because treatment and control groups significantly differed in birth weight SDS, height, and weight SDS at start of treatment (Table I), we also performed a matched-pairs analysis. A propensity score–matched cohort of patients with T1DM and GHD and GHD alone was created by using 5 variables collected at start of GH therapy (height SDS, weight SDS, age, body mass index SDS, and start of GH therapy) to achieve matching balance between the 2 cohorts. Then, 1:2 matching was applied (2 controls per case) to minimize the selection bias and to increase the number of subjects in the control group from 24 to 48. To include as many patients with T1DM and GHD in the analysis, a second matched-pairs analysis was performed also including pubertal patients. Therefore, matching for pubertal status, GH peak, and midparental height SDS became necessary to achieve balanced matched populations. Complete matching confounder covariates were available in 45 patients with T1DM and GHD. Of these 45 matched patients, 1-year data were available in 33 subjects. Quality of diabetes control was not assessed in this analysis because of insufficient longitudinal data on hemoglobin A1c in both groups. Standing height was measured approximately 6 monthly with a wall-mounted stadiometer. Height SDS was calculated by using reference data from Prader et al.9 GH dosage was calculated as milligram per kilogram of body weight per week (mg/kg/wk). Statistical analyses (descriptive data analysis, calculation of SDS, and Wilcoxon rank-sum test) were carried out with SAS software (SAS, Version 9.2; SAS Institute, Cary, North Carolina). A 2-sided significance level of 5% was applied to all statistical tests. Growth velocity data were adjusted for age and sex. Values are expressed as mean with SDs unless otherwise stated.
Table I. Clinical characteristics and response to the first year of GH treatment in treatment and control group. T1DM and GHD Variables Background Birth weight SDS MPH SDS Max GH peak, µg/L Start of GH therapy Chronological age, y Height SDS Height—MPH SDS Weight SDS BMI SDS GH dose, mg/kg/wk 1 y on GH therapy Height velocity, cm/y Height SDS Delta height SDS Weight SDS BMI SDS GH dose, mg/kg/wk
GHD only
n
Median
Mean ± SD
n
Median
Mean ± SD
P
19 19 22
0.06 −0.23 6.93
−0.02 ± 1.15 −0.96 ± 1.32 6.80 ± 3.22
13 582 14 127 15 024
−0.80 −1.40 6.10
−0.81 ± 1.23 −1.36 ± 1.24 5.75 ± 2.75
.003 .111 .208
24 24 19 24 24 24
10.20 −2.62 −1.62 −1.55 0.04 0.24
9.39 ± 3.13 −2.58 ± 1.04 −1.59 ± 1.38 −1.45 ± 1.22 0.13 ± 1.09 0.23 ± 0.08
15 15 14 15 15 15
024 024 127 024 024 024
8.42 −3.01 −1.61 −2.18 −0.32 0.20
8.40 ± 3.46 −3.13 ± 1.15 −1.76 ± 1.51 −2.25 ± 1.47 −0.32 ± 1.27 0.22 ± 0.07
.144 .032 .802 .006 .084 .089
24 24 24 24 24 24
7.54 −1.88 0.57 −1.06 −0.05 0.23
8.16 ± 3.11 −1.90 ± 1.11 0.70 ± 0.55 −1.01 ± 1.27 0.11 ± 1.14 0.24 ± 0.08
15 15 15 14 14 15
024 024 024 935 935 024
8.35 −2.30 0.69 −1.69 −0.40 0.20
8.67 ± 2.54 −2.36 ± 1.10 0.78 ± 0.51 −1.73 ± 1.35 −0.39 ± 1.21 0.21 ± 0.07
.375 .058 .381 .017 .077 .041
BMI, body mass index; MPH, midparental height.
2
Bonfig et al FLA 5.5.0 DTD ■ YMPD9819_proof ■ April 12, 2018
■■ 2018
ORIGINAL ARTICLES
For the matching, we used logistic regression analysis to create a propensity score. We then used the algorithm developed by Parsons to perform greedy nearest neighbor matching without replacement. It is recommended that the optimal number of control subjects matched to case subject should not exceed 2; therefore, 1:2 matching was applied. The variables for the matched cohorts were compared at start of GH therapy and after 1 year on GH therapy. Background variables were evaluated as well. The quality of the propensity scores estimated in a sample were evaluated by using 2 types of comparisons: comparing the distributions of the propensity scores across the case and control groups and comparing the distributions of each covariate across the 2 groups. The distributions of the propensity scores was compared with the Kolmogorov–Smirnov statistical test applied at the 5% level of significance to test for any difference between the cumulative distribution functions for the 2 cohorts. The evaluation of balance was done by comparing the distribution of each covariate across the 2 groups at start of GH therapy and for the background variables by using a Wilcoxon test. The same test was applied on the variables collected after 1 year on GH therapy.
Results Of 24 prepubertal subjects (17 males, 70.8%) with T1DM and GHD, 6 had multiple pituitary hormone deficiencies and 1 child had celiac disease. All subjects were euthyroid, and none were treated with glucocorticoids (no corticotropin-releasing hormone/adrenocorticotropic hormone deficiency in patients with multiple pituitary hormone deficiency). All 24 children were diagnosed with T1DM first and subsequently with GHD. Median and mean maximum GH peak during GH stimulation testing was 6.9 µg/L and 6.8 µg/L, respectively. The median time interval between diagnosis of T1DM and start of GH therapy was 4.8 years (mean 5.2 years, SD 3.4 years). Of 15 024 prepubertal control subjects (10 646 males, 70.9%) without diabetes, 2924 subjects (19.5%) had multiple pituitary hormone deficiency and 678 children (4.5%) also were treated with glucocorticoids. All subjects were euthyroid. The clinical characteristics and response to the first year of GH treatment in prepubertal children with T1DM and GHD compared with those with GHD alone are summarized in Table I. Subjects with T1DM and GHD had a statistically significant greater birth weight than subjects with GHD alone (median birth weight SDS 0.06 [mean –0.02, SD 1.15] vs −0.80 [mean −0.81, SD 1.23], P = .003) and were also heavier at start of GH treatment (median weight SDS −1.55 [mean −1.45, SD 1.22] vs −2.18 [mean −2.25, SD 1.47], P = .006). Age at start of GH treatment was not statistically different between children with combined T1DM and GHD (median age at start of therapy 10.2 years [mean 9.39 years, SD 3.13]) and children with GHD alone (median age at start of therapy 8.42 years [mean 8.40 years, SD 3.46]), P = .14. There were no significant differences between the T1DM and GHD vs the GHD alone group at start of treatment with regard to median height SDS corrected for midparental height SDS
Table II. Matched pairs analysis for prepubertal patients: comparison of matched cohorts at background, start of GH therapy, and after 1 year on GH therapy with the Wilcoxon rank-sum test Matched-pair controls “prepubertal GHD only” Variables Background and at start of GH therapy Birth weight SDS MPH SDS Max GH peak, µg/L Chronological age Height SDS Height—MPH SDS Weight SDS BMI SDS GH dose, mg/kg/wk After 1 y on GH therapy Height velocity, cm/y Height SDS Delta height SDS Weight SDS BMI SDS GH dose, mg/kg/wk
Prepubertal T1DM and GHD
n
Median
n
Median
P
43 47 48 48 48 47 48 48 48
−0.48 −1.02 6.8 10.43 −2.61 −1.62 −1.73 −0.17 0.21
19 19 22 24 24 19 24 24 24
0.06 −0.23 6.93 10.2 −2.62 −1.62 −1.55 0.04 0.24
.1866 .5954 .6670 .5150 .8344 .8318 .8907 .7973 .3964
48 48 48 48 48 48
8.25 −2.01 0.65 −1.12 0.07 0.21
24 24 24 24 24 24
7.54 −1.88 0.57 −1.06 −0.05 0.23
.3244 .7698 .3303 .8344 .9002 .2393
(−1.62 vs −1.61, P = .80), mean GH dosage (0.23 mg/kg/wk vs 0.22 mg/kg/wk, P = .09), and first-year catch-up growth expressed as height velocity (8.2 cm/y compared with 8.7 cm/y, P = .38). After 1 year of treatment, GH dosage was greater in subjects with T1DM and GHD (0.24 mg/kg/wk vs 0.21 mg/ kg/wk, P = .04). Height SDS of children with T1DM and GHD improved from −2.62 (mean −2.58, SD 1.04) to −1.88 (mean −1.90, SD 1.11), P < .001. Accordingly, there was also no difference between the 2 groups in the change of height SDS after 1 year of GH treatment, P = .38. All data are shown in Table I. Because treatment and control group significantly differed in birth weight SDS and height and weight SDS at start of treatment (Table I), we also performed a matched pairs analysis with 1:2 matching (24 patients with T1DM and GHD and 48 patients with GHD alone). The distributions of the propensity scores was compared with Kolmogorov–Smirnov test, P > .90. The evaluation of balance was done by comparing the distribution of each covariate across the 2 groups at start of GH therapy and for the background variables. All P values were >.05, indicating that a proper balance had been achieved between the 2 cohorts, as displayed in Table II. The 2 groups were evaluated after 1 year on GH therapy. No significant differences between the 2 cohorts were identified (P > .05), demonstrating that there was no difference in first-year growth response between the 2 groups. To include the maximum number of patients with T1DM and GHD, in a second matched-pairs analysis, pubertal patients were included. Complete confounder covariates were available in 45 of the initially identified 69 subjects. Of these 45 subjects, 1-year data were only available in 33 subjects.
Efficacy of Growth Hormone Treatment in Children with Type 1 Diabetes Mellitus and Growth Hormone Deficiency—An Analysis of KIGS Data FLA 5.5.0 DTD ■ YMPD9819_proof ■ April 12, 2018
3
THE JOURNAL OF PEDIATRICS • www.jpeds.com Table III. Matched pairs analysis including pubertal patients: comparison of matched cohorts at background, start of GH therapy, and after 1 year on GH therapy with the Wilcoxon rank-sum test* Matched-pair controls including pubertal patients “GHD only” Variables Background and at start of GH therapy Birth weight SDS MPH SDS Max GH peak, µg/L Chronological age Height SDS Height—MPH SDS Weight SDS BMI SDS GH dose, mg/kg/wk After 1 y on GH therapy Height velocity, cm/y Height SDS Delta height SDS Weight SDS BMI SDS GH dose, mg/kg/wk
T1DM and GHD including pubertal patients
n
Median
n
Median
P
80 90 90 90 90 90 90 90 90
−0.42 −0.85 6.97 11.8 −2.42 −1.55 −1.2 0.29 0.21
38 45 45 45 45 45 45 45 45
−0.23 −0.52 6.8 11.6 −2.6 −1.94 −1.2 0.22 0.23
.4823 .6290 .6076 .8064 .5786 .4751 .9851 .6575 .3789
77 77 77 77 77 77
8.54 −1.87 0.59 −0.92 0.08 0.21
33 33 33 33 33 33
7.64 −1.9 0.55 −1.05 −0.19 0.23
.1687 .5838 .7691 .7394 .6763 .1269
*One-year data only available in 33 patients with T1DM and GHD.
Results of this second matched pairs analysis are shown in Table III; again, there was no statistical difference between the 2 cohorts, including comparable first-year growth response in subjects with T1DM and GHD and control patients (Table III). There were 11 adverse events (AEs) reported in the T1DM and GHD group (1 event each of headache, acute sinusitis, naevi, allergic reaction, contusion of the chest wall, knee injury, and foreign body on external eye), and they were not related to GH treatment as reported by the investigators. All patients recovered. In addition, there were 3 AEs related to T1DM, namely diabetic nephropathy and the beginning of diabetic retinopathy, hypoglycemia, and worsening of metabolic control. Only the latter was considered related to GH treatment; the GH dose was reduced. One serious adverse event (SAE) was reported in a patient who was admitted to the hospital with acute pancreatitis, most likely in the context of diabetic ketoacidosis. The patient fully recovered, and the SAE was reported not to be related to GH treatment by the investigator. In 28 258 control subjects, 6808 AEs and 684 SAEs were reported by the investigators. Increased blood glucose/ hyperglycemia were reported in 16 control subjects: 14 subjects were diagnosed with T1DM, and 2 subjects were diagnosed with type 2 diabetes on GH therapy.
Discussion Analyzing data from the KIGS database, we found that shortterm GH treatment and response to therapy over 1 year in prepubertal children with T1DM and GHD were similar to those
Volume ■■ with GHD alone. Children with T1DM and GHD were treated with adequate GH doses, and after 1 year of treatment, the GH doses of children with T1DM and GHD were even greater compared with subjects with GHD alone. With the “borderline” median GH peak in GH stimulation tests both in the treatment and in the control group, children with idiopathic short stature and not truly GHD may have been included but still had a comparable growth response to GH treatment. In a previous study, we evaluated the metabolic effect of GH treatment in children with GH deficiency with T1DM in the Diabetes Prospektive Verlaufsdokumentation (DPV) database.6 We were able to show that GH treatment leads to a significantly greater insulin demand, but with adequate adjustment of insulin doses, metabolic control did not worsen during GH treatment in patients with T1DM. In that study, however, we observed that the growth response to 2 years of GH treatment in children with T1DM and GHD appeared reduced. Metabolic control (hemoglobin A1c) was statistically associated with growth response to GH treatment6; however, there was no information on GH dosage. We speculated that children with T1DM and GHD might have been treated with low GH doses because of the impact of GH on carbohydrate metabolism. For this reason, we decided to perform a similar analysis in KIGS. In this analysis, prepubertal children with T1DM and GHD were treated with comparable start GH doses and showed a comparable initial growth response to GH treatment compared with those with GHD alone. There was a trend toward older age at start of GH treatment in children with T1DM and GHD, as observed in the DPV cohort. Physicians may be hesitant to perform GH stimulation testing in short children with T1DM, attributing poor growth to poor diabetes control. As a result, primarily optimal metabolic control is attempted in diabetic children with growth failure instead of testing for other etiologies. In an Italian survey of 42 pediatric endocrine centers, 69% of centers stated that they usually avoid combined treatment with insulin and GH.7 In the Italian cohort, only 17 patients with T1DM and GH treatment were identified. GH treatment was prescribed for various indications (GHD, Turner syndrome, short stature after intrauterine growth retardation), and GH therapy was effective in promoting growth in most patients. Insulin requirements rose from 0.7 to 1.0 IU/ kg/d during GH treatment without any deterioration of glycemic control.7 A similar rise in insulin requirements was observed in the DPV cohort.6 Unfortunately, in the KIGS database, data on insulin dosage and hemoglobin A1C were insufficient to analyze. In patients with diabetes, hepatic GH resistance has been described previously.3,10,11 This GH resistance is a risk factor for decreased GH efficacy. Therefore, IGF-I concentrations would be of great interest in subjects with diabetes treated with GH. Unfortunately, longitudinal IGF-I data were insufficient for analysis in both the DPV and the KIGS databases. There are 2 large studies published12,13 that report an increased incidence of type 2 diabetes in children and adolescents and also in adults treated with GH. Child et al compared the incidence of type 1 and type 2 diabetes in patients aged
4
Bonfig et al FLA 5.5.0 DTD ■ YMPD9819_proof ■ April 12, 2018
■■ 2018
ORIGINAL ARTICLES
<20 years in the Genetics and Neuroendocrinology of Short Stature International Study (GeNeSIS) observational research program in the US SEARCH for Diabetes in Youth study.13 An increased incidence was only found for type 2, but not for type 1, diabetes in children treated with GH. Therefore, monitoring of glucose metabolism is recommended during GH treatment, especially in patients with pre-existing risk factors such as obesity. In a study by Cutfield et al, GH treatment did not affect the incidence of type 1 diabetes in any age group.14 Prepubertal children with T1DM and GHD display a similar initial growth response to GH treatment compared with those with GHD alone. Therefore, taken together with previous data, GH treatment is safe and effective in these patients. Because KIGS and DPV are observational studies, reconfirmation of these findings in multicenter studies should be pursued. Preexisting T1DM should not be a reason to deny GH treatment to children with GHD. ■ We thank all participating investigators for data collection and entry into KIGS (Pfizer International Growth Database). Submitted for publication Jul 27, 2017; last revision received Jan 23, 2018; accepted Feb 13, 2018 Reprint requests: Walter Bonfig, MD, Department of Pediatrics, Klinikum WelsGrieskirchen, Grieskirchner Str 42, Wels A-4600. E-mail: walter.bonfig@ mri.tum.de
References 1. Ranke M, Reiter EO, Price DA. Idiopathic growth hormone deficiency in KIGS: selected aspects. In: Ranke M, Price DA, Reiter EO, eds. Growth hormone therapy in pediatrics–20 years of KIGS. 1st ed. Basel: Karger; 2007, p. 116-35. 2. Moller N, Jorgensen JO. Effects of growth hormone on glucose, lipid and protein metabolism in human subjects. Endocr Rev 2009;30:15277.
3. Chiarelli F, Giannini C, Mohn A. Growth, growth factors and diabetes. Eur J Endocrinol 2004;151:109-17. 4. Ehehalt S, Dietz K, Willasch AM, Neu A. for the DIARY-Group BadenWuerttemberg. Prediction model for the incidence and prevalence of type 1 diabetes in childhood and adolescence: evidence for a cohort-dependent increase within the next two decades in Germany. Pediatr Diabetes 2007;13:15-20. 5. Bonfig W, Kapellen T, Dost A, Fritsch M, Rohrer T, Wolf J, et al. Growth in children with type 1 diabetes. J Pediatr 2012;160:900-3. 6. Bonfig W, Molz K, Woelfle J, Hofer SE, Hauffa BP, Schoenau E, Diabetes Patienten Verlaufsdocumentationsystem Initiative of the German Working Group for Pediatric Diabetology, et al. Metabolic safety of growth hormone in type 1 diabetes and idiopathic growth hormone deficiency. J Pediatr 2013;163:1095-8. 7. Zucchini S, Iafusco D, Vannelli S, Rabbone I, Salzano G, Pozzobon G, et al. Combined therapy with insulin and growth hormone in 17 patients with type-1 diabetes and growth disorders. Horm Res Paediatr 2014;82:53-8. 8. Ranke MB. The KIGS aetiology classification system. In: Ranke MB, Price DA, Reiter EO, eds. Growth hormone therapy in pediatrics—20 years of KIGS. 1st ed. Basel: Karger; 2007, p. 29-37. 9. Prader A, Largo RH, Molinari L, Issler C. Physical growth of Swiss children from birth to 20 years of age. First Zurich longitudinal study of growth and development. Helv Paediatr Acta 1989;52:1-125. 10. Mercado M, Baumann G. Characteristics of the somatotropic axis in insulin dependent diabetes mellitus. Arch Med Res 1995;26:101-9. 11. Holl RW, Siegler B, Scherbaum WA, Heinze E. The serum growth hormone-binding protein is reduced in young patients with insulindependent diabetes mellitus. J Clin Endocrinol Metab 1993;76:165-7. 12. Luger A, Mattsson AF, Koitowska-Hagstrom M, Thunander M, Goth M, Verhelst J, et al. Incidence of diabetes mellitus and evolution of glucose parameters in growth-hormone deficient subjects during growth hormone replacement therapy: a long-term observational study. Diabetes Care 2012;35:57-62. 13. Child CJ, Zimmermann AG, Scott RS, Cutler GB, Battelino T, Blum WF, et al. Prevalence and incidence of diabetes mellitus in GH-treated children and adolescents: analysis from the GeNeSIS observational research program. J Clin Endocrinol Metab 2011;96:1025-34. 14. Cutfield WS, Wilton P, Bennmarker H, Albertsson-Wikland K, Chatelain P, Ranke MB, et al. Incidence of diabetes mellitus and impaired glucose tolerance in children and adolescents receiving growth-hormone treatment. Lancet 2000;355:610-3.
Efficacy of Growth Hormone Treatment in Children with Type 1 Diabetes Mellitus and Growth Hormone Deficiency—An Analysis of KIGS Data FLA 5.5.0 DTD ■ YMPD9819_proof ■ April 12, 2018
5