IGF-1 and IGF-binding proteins and bone mass, geometry, and strength: relation to metabolic control in adolescent girls with type 1 diabetes.

Abstract

Children and adolescents with poorly controlled type 1 diabetes mellitus (T1DM) are at risk for decreased bone mass. Growth hormone (GH) and its mediator, IGF-1, promote skeletal growth. Recent observations have suggested that children and adolescents with T1DM are at risk for decreased bone mineral acquisition. We examined the relationships between metabolic control, IGF-1 and its binding proteins (IGFBP-1, -3, -5), and bone mass in T1DM in adolescent girls 12-15 yr of age with T1DM (n = 11) and matched controls (n = 10). Subjects were admitted overnight and given a standardized diet. Periodic blood samples were obtained, and bone measurements were performed. Serum GH, IGFBP-1 and -5, glycosylated hemoglobin (HbA(1c)), glucose, and urine magnesium levels were higher and IGF-1 values were lower in T1DM compared with controls (p < 0.05). Whole body BMC/bone area (BA), femoral neck areal BMD (aBMD) and bone mineral apparent density (BMAD), and tibia cortical BMC were lower in T1DM (p < 0.05). Poor diabetes control predicted lower IGF-1 (r(2) = 0.21) and greater IGFBP-1 (r(2) = 0.39), IGFBP-5 (r(2) = 0.38), and bone-specific alkaline phosphatase (BALP; r(2) = 0.41, p < 0.05). Higher urine magnesium excretion predicted an overall shorter, lighter skeleton, and lower tibia cortical bone size, mineral, and density (r(2) = 0.44-0.75, p < 0.05). In the T1DM cohort, earlier age at diagnosis was predictive of lower IGF-1, higher urine magnesium excretion, and lighter, thinner cortical bone (r(2) >or=0.45, p < 0.01). We conclude that poor metabolic control alters the GH/IGF-1 axis, whereas greater urine magnesium excretion may reflect subtle changes in renal function and/or glucosuria leading to altered bone size and density in adolescent girls with T1DM.

Full text

Page 1

IGF-1 and IGF-Binding Proteins and Bone Mass, Geometry, and
Strength: Relation to Metabolic Control in Adolescent Girls With
Type 1 Diabetes
Laurie J Moyer-Mileur,1Hillarie Slater,1Kristine C Jordan,1,2and Mary A Murray3
ABSTRACT: Children and adolescents with poorly controlled type 1 diabetes mellitus (T1DM) are at risk for
decreased bone mass. Growth hormone (GH) and its mediator, IGF-1, promote skeletal growth. Recent
observations have suggested that children and adolescents with T1DM are at risk for decreased bone mineral
acquisition. We examined the relationships between metabolic control, IGF-1 and its binding proteins
(IGFBP-1, -3, -5), and bone mass in T1DM in adolescent girls 12–15 yr of age with T1DM (n ? 11) and
matched controls (n ? 10). Subjects were admitted overnight and given a standardized diet. Periodic blood
samples were obtained, and bone measurements were performed. Serum GH, IGFBP-1 and -5, glycosylated
hemoglobin (HbA1c), glucose, and urine magnesium levels were higher and IGF-1 values were lower in T1DM
compared with controls (p < 0.05). Whole body BMC/bone area (BA), femoral neck areal BMD (aBMD) and
bone mineral apparent density (BMAD), and tibia cortical BMC were lower in T1DM (p < 0.05). Poor
diabetes control predicted lower IGF-1 (r2? 0.21) and greater IGFBP-1 (r2? 0.39), IGFBP-5 (r2? 0.38),
and bone-specific alkaline phosphatase (BALP; r2? 0.41, p < 0.05). Higher urine magnesium excretion
predicted an overall shorter, lighter skeleton, and lower tibia cortical bone size, mineral, and density (r2?
0.44–0.75, p < 0.05). In the T1DM cohort, earlier age at diagnosis was predictive of lower IGF-1, higher urine
magnesium excretion, and lighter, thinner cortical bone (r2? 0.45, p < 0.01). We conclude that poor metabolic
control alters the GH/IGF-1 axis, whereas greater urine magnesium excretion may reflect subtle changes in
renal function and/or glucosuria leading to altered bone size and density in adolescent girls with T1DM.
J Bone Miner Res 2008;23:1884–1891. Published online on July 28, 2008; doi: 10.1359/JBMR.080713
Key words: metabolic control, bone, adolescence, diabetes
INTRODUCTION
G
mineral accretion is predictive of osteoporosis in aging
women.(1)Recent observations suggest that children and
adolescents with type 1 diabetes mellitus (T1DM) are at
risk for decreased bone mineral acquisition.(2–7)Our re-
search group has observed significantly lower bone mass in
adolescents with T1DM compared with a nondiabetic ref-
erence population.(8,9)Children and adolescents with
poorly controlled T1DM are at risk for decreased bone
mass. Disease duration,(7,10)poor metabolic con-
trol,(2,6,11,12)and diabetic complications such as retinopathy
or neuropathy(2,10)are reported to have a negative impact
on bone mass, although others have found no relation-
ship.(13,14)Poor blood glucose control has been associated
with lower bone mass in adults and adolescents with
T1DM(8–11)but not in children.(3,14)
The growth hormone (GH)/IGF axis is a major determi-
nant of bone mass acquisition. Circulating IGF-1 is pro-
OOD BONE HEALTH is important to the maintenance of
functionality in the aging population. Pubertal bone
duced by the liver, is structurally similar to insulin,(15)and
helps to mediate the skeletal growth promoting actions of
GH. IGF-1 is also produced locally by muscle and bone
tissue, where it is hypothesized to act in a paracrine manner.
Along with systemic GH and estradiol, local bone IGF-1
concentrations are regulated by PTH, 1,25-dihydroxy-
vitamin D3, and other cytokines and growth factors.(16)
IGF-1 functions as a key anabolic regulator of bone cell
activity by decreasing collagen degradation and increasing
bone matrix deposition and osteoblastic cell recruit-
ment.(15–17)In healthy children, serum IGF-1 concentra-
tions correlate well with BMC.(18)In adolescents with
poorly controlled T1DM, there is established evidence of
increased secretion of GH but low levels of IGF-1 com-
pared with matched controls.(13,17)Thus, the GH/IGF axis
has received considerable attention as a mechanism for in-
adequate bone formation in T1DM.(19–22)
IGF binding proteins (IGFBPs) control the tissue avail-
ability of IGF-1 and therefore are major regulators of
IGF-1 action. IGFBP-3 is the predominant circulating
IGFBP, binding >95% of circulating IGF. Production of
IGFBP-3 is stimulated by GH, whereas IGFBP-1 is in-
The authors state that they have no conflicts of interest.
1Center for Pediatric Nutrition Research, Department of Pediatrics, University of Utah, Salt Lake City, Utah, USA;2Division of
Nutrition, College of Health, University of Utah, Salt Lake City, Utah, USA;3Division of Endocrinology, Department of Pediatrics,
University of Utah, Salt Lake City, Utah, USA.
JOURNAL OF BONE AND MINERAL RESEARCH
Volume 23, Number 12, 2008
Published online on July 28, 2008; doi: 10.1359/JBMR.080713
© 2008 American Society for Bone and Mineral Research
1884

Page 2

creased in the absence of GH. In patients with osteoporosis,
IGFBP-1 and IGFBP-4 inhibit IGF bone cell proliferation
by sequestering IGF-1 and preventing binding to the IGF
receptor.(15)IGFBP-3 and IGFBP-5 facilitate IGF action in
bone cells; bone matrix proteoglycans do not bind IGFs in
the absence of IGFBP-3 and IGFs do not bind to hydroxy-
apatite in the absence of IGFBP-5.(15)There is also some
evidence that IGFBPs regulate bone formation indepen-
dent of IGF function. For example, IGFBP-5 has been
found to stimulate markers of bone formation in osteo-
blasts that lack functional IGF.(17)
There is a paucity of information on how IGF-1 and
IGFBPs regulate skeletal growth in children with T1DM
during puberty. The purpose of this study was to assess the
GH/IGF axis and bone health in adolescent girls with
T1DM. A secondary objective was to examine the relation-
ships between IGF-1 and its binding proteins and the char-
acteristics and markers of bone turnover. We tested the
central hypothesis that T1DM is associated with a state of
partial GH resistance resulting in decreased IGF action and
altered IGFBP activity, leading to diminished bone size,
density, and strength.
MATERIALS AND METHODS
Subjects
Girls 12–15 yr of age with T1DM for a minimum of 12 mo
(n ? 11) were recruited from the Utah Diabetes Center
Pediatric Program, Salt Lake City, UT, USA. Healthy girls
matched for race, age, and maturation were recruited as
controls (n ? 10). Exclusion criteria included poor meta-
bolic control (glycosylated hemoglobin [HbA1c] > 9.0%),
hypertension (diastolic blood pressure > 90th percentile for
age), microalbuminuria, hypo- or hyperthyroidism, GH de-
ficiency, celiac disease (tTG ?7.0 AU or symptoms) or
other health conditions or medication use known to alter
growth or bone mineral deposition. This study was ap-
proved by the University of Utah Institutional Review
Board for Human Subjects.
Protocol
The schema for the protocol is shown in Fig. 1. Potential
subjects were evaluated in a prestudy visit. Appropriate
subjects were admitted to the General Clinical Research
Center (GCRC) at 4:00 p.m. on the day of study. Written
informed consent and assent were obtained at admission.
An intravenous catheter was placed in a forearm vein for
blood access by 10:00 p.m., and blood samples were ob-
tained from 11:00 p.m. to 11:00 a.m. First and second morn-
ing voids of urine were collected at 6:00 and 8:00 a.m. Sub-
jects were given standardized meals and snacks with energy
based on the subject’s sex, age, and body weight and a
substrate distribution of 15% protein, 35% fat, and 50%
carbohydrate. T1DM subjects received their usual insulin
dose.
Each study participant completed a health history ques-
tionnaire, which included a family and personal medical
history and current medication use. Pubertal maturation
was determined by a pediatric endocrinologist using Tanner
stage criteria.(23)Calcium intake(24)and past-year physical
activity(25)were also assessed by questionnaire. Metabolic
hours of weight-bearing physical activity for the previous
year were calculated by assigning each weight-bearing ac-
tivity a number representing metabolic cost (MET).(26)
Height without shoes was measured to the nearest 0.1 cm
for each participant using a Height-Rite Stadiometer
(Model 225; Seca, Culver City, CA, USA), and weight was
measured to the nearest 0.1 kg by digital scale (Model 5002;
Scan-Tronix, Carol Stream, IL, USA). Body mass index
(BMI, kg/m2) was calculated for all subjects.
Bone measurements
Three cross-sectional slices of the nondominant tibia
were measured by pQCT (XCT-2000; Stratec/Orthometrix,
White Plains, NJ, USA) at relative distances of 4%, 38%,
and 66% from the distal tibia growth plate to assess trabec-
ular and cortical bone and the muscle cross-sectional area
(CSA), respectively. Dominance and nondominance were
determined by asking whether the subject was right or left
handed. Tibia metaphyseal bone properties, including tra-
becular volumetric BMD (vBMD; mg/cm3), were assessed
from the 4% CSA. The 38% CSA was used to assess tibia
diaphyseal bone properties, including cortical bone vBMD,
BMC (mg), and geometric bone properties: bone CSA
(mm2), cortical CSA (bone CSA less marrow CSA), mar-
row CSA (bone CSA less cortical CSA), and cortical thick-
ness (mm). Muscle CSA and the polar strength strain index
(pSSI, mm3) were determined from the 66% distal cross-
section to examine bone muscle relationship(27)and bone
strength.(28)Analysis parameters and modes were as pre-
viously described.(29)In addition, whole body bone area
(BA, cm2), BMC (g), lean body mass (LBM, kg), percent
body fat, femoral neck (FN) and lumbar spine (LS) BA,
BMC, and areal BMD (aBMD, g/cm2) were determined by
DXA (4500A; Hologic, Waltham, MA, USA). Height for
age, whole body BA to height, and whole body BMC to BA
FIG. 1. Protocol schema. Subjects were ad-
mitted to the General Clinical Research
Center at 5:00 p.m. and received standard-
ized meals at 6:00 p.m., 8:00 p.m., and 8:00
a.m. Hourly blood samples were obtained
from 11:00 p.m. to 11:00 a.m., and urine
samples were obtained at 6:00 and 8:00 a.m.
GLUCOSE CONTROL AND BONE MASS IN TYPE 1 DIABETES1885

Page 3

were assessed to determine whether bone mass was re-
duced because of short, narrow, or lighter bones.(30)FN and
LS BMC and BA values were used to determine bone min-
eral apparent density (BMAD, g/cm3).(31)The CV for re-
positioning in adult volunteers was <2.5% for trabecular
and cortical bone vBMD using pQCT and <1.0% for
aBMD measured by DXA in our laboratory. The daily CVs
for calibration phantoms were 0.1% and 0.3% for pQCT
and DXA, respectively. The same experienced radiology
technician performed all measurements. Although radia-
tion exposure was minimal (≈21.5 ?SV total), pregnancy
tests were performed before densitometry on all girls with
Tanner stage ?2.
Biochemical measures
GH, IGF-1, insulin, and glucose levels were determined
from samples obtained hourly from 11:00 p.m. to 6:00 a.m.
and IGF-1 and IGFBP-1 from 6:00 a.m. to 10:00 a.m. The
area under the curve (AUC) was calculated for 11:00 p.m.
to 6:00 a.m. values to assess the GH/IGF axis activity and
between 9:00 and 10:00 a.m. to examine postprandial dif-
ferences between IGF-1 and IGFBP-1. IGFBP-3, -4, and -5
values. All other serum values were determined from the
fasting 8:00 a.m. blood sample. Pyridinoline (Pyd), deoxy-
pyridinoline (Dpd), and hydroxylysine-derived cross-links
of mature collagen degradation were selected to indirectly
assess bone resorption. Pyd and Dpd cross-links were mea-
sured from the 6:00 a.m. urine sample, whereas calcium,
phosphorus, magnesium, and creatinine assays were deter-
mined on the 8:00 a.m. sample. Serum samples (10 ml) were
collected using a standard technique from indwelling cath-
eters without anticoagulants and were processed to avoid
hemolysis. Serum was separated by centrifugation and
stored at –70°C until analysis. Urine samples were collected
without preservative and stored at –70°C. ELISA assays
were used to measure GH, IGF-1, IGFBP-1, -3, and -4,
insulin, 1,25(OH)2vitamin D (DSL), and bone-specific al-
kaline phosphatase (BALP; Metra); high performance liq-
uid chromatography (HPLC) was used to measure HbA1c;
and the spectrophotometric technique was used to measure
serum and urine minerals. IGFBP-5 levels were measured
by the Musculoskeletal Disease Center Laboratory, Loma
Linda University, using polyclonal guinea pig antiserum
and recombinant IGFBP-5 as standard and tracer, respec-
tively, Pyd and Dpd were measured by HPLC, with stan-
dards purchased from Quidel (San Diego, CA, USA). Uri-
nary Pyd cross-link concentration was normalized to
urinary creatinine (Beckman Creatinine Analyzer 2) and
calculated as ?mol/mol creatinine. Measurements were
made in duplicate, and the average values were entered into
the database.
Statistical methods
Statistical analyses included independent t-test and ?2to
identify differences in demographic variables of interest be-
tween T1DM and control subjects. Standard deviation
scores (SDSs) were generated for body weight and height
using EpiInfo.(32)The ratios of IGFBP-1, -3, -4, and -5 to
IGF-1 were calculated. Multivariate and repeated-measures
analysis of covariance (MANCOVA) and posthoc tests
were used to compare means between T1DM and control
groups for blood and urine results, bone measurements,
and body composition using pubertal stage and height SDS
as covariates. Bone results are reported as adjusted means
with the 5th and 95th CIs. Correlation analysis was run
between selected continuous variables. Stepwise linear re-
gression was performed to identify which variable predicted
biochemical and bone characteristics. Statistical analyses
for these data were performed using the SPSS-PC+ (Ver-
sion 13.1; SPSS) statistical software program with signifi-
cance set at p ? 0.05.
RESULTS
Demographic data are presented in Table 1. Age, puber-
tal maturation, body size and composition, reported cal-
cium intake, and physical activity levels were similar be-
tween groups.
Biochemical findings
GH/IGF-1 axis results were as expected (Table 2). The
rise in GH levels was greater at 12:00 p.m. and 1:00 a.m. and
the IGF-1 response was lower in T1DM girls compared
with controls (p < 0.05; Fig. 2). The mean AUC values for
GH and IGF-1 from 11:00 p.m. to 6:00 a.m., however, were
not statistically significant. The mean hourly values for se-
rum insulin and glucose at 12:00 p.m. to 6:00 a.m. were
significantly higher in T1DM girls (p < 0.05) than controls.
The mean hourly postprandial values for IGF-1 and
IGFBP-1 are plotted in Fig. 3. Differences are again evident
with lower mean IGF-1 and higher IGFBP-1 values found
in the T1DM cohort compared with controls at 8:00, 9:00,
TABLE 1. SUBJECT DEMOGRAPHICS
T1DM
(N = 11)
Control
(N = 10)
Age (yr)
Tanner stage (%)
II
III
IV
V
Menses (%)
Height (SDS)
BMI (kg/m2)
Lean mass (kg)
Body fat (%)
Calcium intake (mg/d)
Physical activity (MET h/wk)
Age at diagnosis (yr)
Disease duration (yr)
Insulin delivery mode
Pump
Injection
Insulin (U/kg/d)
12.9 ± 1.0 13.1 ± 1.1
9.1%
54.5%
18.2%
18.2%
45.5%
0.1 ± 1.0
20.9 ± 1.9
33.4 ± 5.4
29.8 ± 6.3
1811.8 ± 910.4
70.0 ± 57.3
7.0 ± 4.3
5.9 ± 3.7
10.0%
60.0%
10.0%
20.0%
50.0%
0.2 ± 0.6
20.3 ± 2.1
33.5 ± 5.4
29.1 ± 7.0
1752.4 ± 879.6
59.8 ± 39.2
7 (66.6%)
4 (33.4%)
1.0 ± 0.2
Values are mean ± SD.
No significant differences for any variable by group; p values not re-
ported.
SDS, standard deviation score; independent t-test and ?2.
MOYER-MILEUR ET AL.1886

Page 4

and 10:00 a.m. (p < 0.05). This finding is substantiated by
lower IGF-1 and higher IGFBP-1 AUC values in T1DM
girls from 8:00 to 10:00 a.m. (778.6 ± 232.3 ng/dl IGF-1 and
169.7 ± 84.4 ng/dl IGFBP-1 T1DM versus 954.8 ± 198.9
ng/dl IGF-1 and 79.5 ± 46.3 ng/dl IGFBP-1 controls; p ?
0.05). The mean fasting levels for IGBP-3, -4, and -5 were
similar (Table 2). The ratios for IGFBP-1, -3, -4, and -5
relative to IGF-1, however, were higher in T1DM girls,
although only IGFBP-1 and -5/IGF-1 ratios were statisti-
cally significant (p ? 0.05; data not shown).
Additional serum and urine results are also presented in
Table 2. T1DM girls had higher fasting HbA1cand serum
glucose values versus controls (p ? 0.01). The mean values
for markers of bone turnover tended to be greater, as evi-
denced by higher serum BALP and urine Dpd/Pyd levels
and lower mean 25(OH) vitamin D levels in T1DM girls.
Urine magnesium levels were markedly higher in T1DM
girls compared with controls (p < 0.05). We did not detect
differences in serum PTH, calcium, magnesium, and phos-
phorus levels or urine calcium and phosphorus excretion.
With the exception of serum HbA1c, glucose, and urine
magnesium for T1DM, all values were within the normal
reference range.
Skeletal characteristics
The whole body, FN, LS, and tibial bone geometry, den-
sity, and strength findings for T1DM and control groups are
presented in Table 3. Whole body BMC/BA and FN aBMD
and BMAD values were significantly lower in T1DM girls
(p ? 0.05). The evaluation of tibial bone geometry and
density showed significantly lower cortical BMC and corti-
cal BMC/muscle CSA (p < 0.05), with a trend toward thin-
ner cortical bone thickness and vBMD (p ? 0.07 and 0.09,
respectively) in T1DM girls. All other skeletal findings
were similar between groups.
Predictors
Poor metabolic control predicted lower IGF-1 AUC val-
ues, contributing 50–94% of the total variability in fasting
and postprandial levels, respectively (p ? 0.01). Poor meta-
bolic control was also correlated with higher IGFBP-1 and
-5/IGF-1 ratios (R ? 0.65, p ? 0.01), but was not a signif-
icant predictor for either IGFBP-1 or -5 in the regression
model. Greater urine magnesium excretion was predictive
of a higher IGFBP-3/IGF-1 ratio, accounting for 26% of the
variability (p ? 0.001). The onset of menses was predictive
of a higher IGFBP-4/IGF-1 ratio (r2? 0.23, p ? 0.001).
The absence of menses, higher serum 25(OH) vitamin D
levels, and poor metabolic control predicted higher levels of
BALP (r2? 0.69, p ? 0.001), accounting for 32%, 28%,
and 9% of the variance, respectively. Higher 25(OH) vita-
min D levels were predicted by higher BALP and lower
GH levels, which accounted for 28% and 13% of variabil-
ity, respectively (p < 0.01). Higher serum glucose was the
single predictor of greater bone resorption (r2? 0.48, p ?
0.001) and urine magnesium excretion (r2? 0.61, p < 0.01),
accounting for 48% and 61%, respectively, of the variability
observed in Dpd/Pyd and urine magnesium levels.
The relationships between T1DM disease–related factors
and biochemical findings were examined by removing con-
trol and adding age at diagnosis, disease duration, and in-
sulin dose (U/kg/d) to the regression model. A younger age
at diagnosis was predictive of lower IGF-1 and higher
IGFBP-1/IGF-1 ratios and greater urine magnesium excre-
tion, accounting for 79–97% of the variability. Poor meta-
bolic control strongly correlated with the ratios of IGFBP-4
TABLE 2. SERUM AND URINE RESULTS*
T1DM (N = 11)Control (N = 10)pReference range
Serum
IGFBP-1 (ng/ml)
IGFBP-3 (ng/ml)
IGFBP-4 (ng/ml)
IGFBP-5 (ng/ml)
HbA1c(%)
Glucose (mg/dl)
BALP (U/liter)
25(OH)Vitamin D (ng/ml)
PTH (pg/ml)
Calcium (mg/dl)
Magnesium (mg/dl)
Phosphorus (mg/dl)
Urine
Creatinine (mg/dl)
Dpyd (?mol/mol creat)
Pyd (?mol/mol creat)
Dpyd:Pyd (?mol/mol creat: ?mol/mol creat)10−2
Calcium (mg/mg creat)10−2
Magnesium (mg/mg creat) 10−2
Phosphorus (mg /mg creat)10−2
66.37 ± 30.61
4.28 ± 1.49
388.31 ± 275.06
289.77 ± 40.45
8.11 ± 1.04
148.28 ± 60.17
173.27 ± 97.23
14.67 ± 7.67
37.61 ± 6.23
9.01 ± 0.71
2.06 ± 0.49
4.27 ± 0.90
37.13 ± 24.18
4.74 ± 0.74
434.81 ± 288.29
261.00 ± 47.45
4.94 ± 0.34
89.65 ± 6.01
128.92 ± 99.44
29.54 ± 8.50
27.34 ± 7.02
8.95 ± 0.55
2.24 ± 0.55
3.78 ± 0.45
0.02
0.85
0.47
0.13
0.001
0.04
0.18
0.001
0.31
0.71
0.22
0.22
20–200
2.1–6.2
Not established
300–600
4–6%
80–110
125–275
20–57
15–75
8.0–10.0
1.58–2.55
2.7–4.5
63.82 ± 36.45
61.63 ± 21.94
221.05 ± 71.05
27.61 ± 3.25
7.72 ± 5.41
3.14 ± 1.69
45.07 ± 28.34
81.91 ± 47.45
46.38 ± 25.15
179.23 ± 82.83
25.39 ± 3.01
7.72 ± 4.88
2.21 ± 0.94
37.58 ± 22.91
0.54
0.37
0.44
0.09
0.51
0.02
0.86
22–89
13–61
81–267
17–27
10–24
0.2–2.6
Not established
Values are mean ± SD. Significance determined by Mann-Whitney.
*Blood samples obtained at 8:00 a.m. and urine samples at 6:00 p.m. (fasting).
GLUCOSE CONTROL AND BONE MASS IN TYPE 1 DIABETES1887

Page 5

and -5 to IGF-1 (R ? 0.78 and 0.65, respectively, p < 0.01),
but was only predictive of IGFBP-5/IGF in the regression
model (r2? 0.91, p ? 0.001), accounting for 72% of the
variance.
The relationships between T1DM disease–related factors
and skeletal findings were examined by removing control
and adding age at diagnosis, disease duration, and insulin
dose (U/kg/d) to the regression model. Longer disease du-
ration was predictive of smaller whole body bone size and
lower tibia total and cortical CSA, cortical BMC, and thick-
ness (r2? 0.65, p ? 0.001), accounting for 6.9–80% of the
variance.
Higher urine magnesium excretion was predictive of a
greater height deficit, lighter bones, decreased FN BMAD,
lower tibia cortical bone CSA, thickness, BMC, and tibia
marrow CSA (r2? 0.26–0.84, p ? 0.05) and accounted for
26–60% of the variance. A larger ratio of IGFBP-1 relative
to IGF-1 was predictive of smaller bones and accounted for
36% of the variability observed in the BA/height ratio (p ?
0.001). Trabecular vBMD was inversely related to IGFBP-1
and -5/IGF-1 ratios (r2? 0.40, p ? 0.001), accounting for
11% and 29% of the variance, respectively. Greater tibia
bone strength (pSSI) was predicted by higher IGF-1 levels
(r2? 0.23, p ? 0.001).
DISCUSSION
To our knowledge, this is the first study directed at the
relationship between metabolic control, GH/IGF-1 axis ac-
tivity, and bone health in adolescent girls with T1DM. The
altered nighttime GH/IGF axis activity observed in our
study confirms the work of others. We also found signifi-
FIG. 2. (A and B) Mean serum GH and IGF-1 levels between
11:00 p.m. and 6:00 a.m. for T1DM (?) and control (?) groups.
Absence of IGF-1 rise in response to elevated GH confirms an
altered GH/IGF-1 axis response in T1DM.ap ? 0.05.
FIG. 3. (A and B) Mean postprandial serum IGF-1 and IGFBP-1
levels for T1DM (?) and control (?) groups. Lower IGF-1 and
higher IGFBP-1 values confirm sustained alteration in GH/IGF-1
axis in T1DM.ap ? 0.05.
MOYER-MILEUR ET AL.1888