Key words
vitamin D - small-for-gestational-age - insulin resistance - β-cell function - in
utero growth restriction
Introduction
Recent studies have extended the activity of vitamin D [25(OH)D] well beyond that
in calcium homeostasis and bone metabolism [1]. Clinical and experimental evidence supports a role of 25(OH)D in the pathogenesis
of type 2 diabetes, with effects on either insulin sensitivity or β-cell function,
or both [2]
[3]. Patients at risk for type 2 diabetes may have lower serum levels of 25(OH)D [3]. Positive correlation between levels of 25(OH)D and insulin sensitivity, and a negative
effect of hypovitaminosis D on β-cell function have also been observed in persons
without glucose intolerance [4]
[5].
Subjects born small-for-gestational age (SGA) because of intrauterine growth restriction
constitute a population at risk of early development of chronic diseases, including
type 2 diabetes [6]. A series of children born SGA had increased indices of insulin resistance early
in life, correlated with both in utero growth restriction and accelerated postnatal
growth [7]
[8]
[9]. 25(OH)D levels in correlation to insulin resistance have not been studied in this
population. This case-control study was designed to investigate the relationships
between serum levels of vitamin 25(OH)D, indices of insulin resistance and β-cell
function, and birth weight in prepubertal children with severe in utero growth restriction.
Research Design and Methods
Research Design and Methods
The study aimed to include all children born SGA at the University Hospital of Ioannina
after a term pregnancy (GA 37–41 weeks) during the 18-month period January 2003–June
2004. Most of the births (>80%) in a well-defined area of northwest Greece took place
in this hospital. SGA children were defined as having birth weight <2 standard deviations
(SDs) below the mean based on local growth charts. Of the children born SGA during
the 18-month period, 44 were eligible; for 27 (15 males) of which the parents gave
consent for participation in the study. Of the other children, the parents of 12 did
not wish their children to participate and for 5 the contact details were not found.
The clinical and socio-economic characteristics of the nonparticipating children did
not differ from those of the study group.
The control group comprised 38 children (19 males) born in the hospital during the
same period, appropriate for gestational age (AGA), defined as having birth weight
between the 20th and 80th percentiles for gestational age on local growth charts specific for age and gender.
They were matched with the SGA children for age, gender, height, weight, body mass
index (BMI), and pubertal status (stage 1 according to the Tanner criteria for puberty
development). A total of 170 AGA children were initially contacted to indentify the
38 matched controls. All the study children were healthy, were not receiving drugs
for any cause, and had no history of liver or renal disease or malabsorption. Exclusion
criteria for both groups were: congenital malformations or genetic disorders, known
metabolic disorder or chronic disease, obesity (BMI ≥95) at the time of the study,
and a positive family history of diabetes or gestational diabetes.
The study protocol was approved by the Research Ethics Committee of Ioannina University
Hospital. The parents of the eligible children were contacted, their written informed
consent was secured and the children were evaluated at between 5 and 7.5 years of
age. The following anthropometric data were recorded: birth weight, crown to heel
length, and head circumference, obtained from the birth records, and body weight,
body height, BMI and waist circumference at the time of the study, measured by standard
methods. z-Scores for birth weight, BMI and waist circumference were derived from
appropriate reference population standards. A morning venous blood sample for biochemical
determinations was collected from each child after a rigorous 12 h fast, during the
winter months (December–February); recruitment was balanced between SGA and AGA children
across the months.
Fasting serum levels of glucose, insulin, 25(OH)D, and parathyroid hormone (PTH) were measured. Insulin was determined using
an immunoenzymatic method (analyzer AXSYM, Abbott) and glucose by the glucose oxidase
method. The homeostasis model assessment for insulin resistance (HOMA-IR) index and
β-cell function (HOMA-β%) was used to detect the degree of insulin resistance and
β-cell function, respectively [10]. HOMA-IR was assessed by the formula: [(insulin (mU/l)×glucose (mmol/I)]/22.5, and
HOMA-β% by [20×insulin (mU/l)/(glucose (mmol/l)–3.5].
25(OH)D was determined by an enzyme-immunoassay (EIA) method using the kit of IDS
Systems Ltd, UK. The sensitivity of the method was 5.0 nmol/l, and the intra- and
interassay CVs were 5.3% and 4.6%, respectively. The biologically intact molecule
of PTH (iPTH) was measured by a 2-site enzyme linked immunosorbent assay (ELISA) using
the kit of BIOMERICA Inc. (USA). The sensitivity of the method was 0.09 pmol/l, and
the intra- and interassay CVs were 3.2% and 7.7%, respectively. The sample volume
required for each assay was 25 μl. A detailed questionnaire was completed by the parents
of each child at the time of the study, concerning outdoor activities and food consumption,
with emphasis on oral vitamin D preparations and foods containing or enriched with
vitamin D (i.e., oily fish, eggs, and fortified breakfast cereals). Dietary vitamin
D intake was estimated, based on the USDA food composition data (http://www.ars.usda.gov/nutrientdata), and on the labels in the case of the fortified breakfast cereals.
Statistical analysis
Students’ t-test was used after examination of parameters for normal distribution. Simple and
multiple regression analyses were conducted to define relationships among the examined
parameters, namely 25(OH)D level, birth weight z-scores, insulin resistance and β-cell
function indices, BMI, waist circumference z-scores and age. A sample size of 65 children
could define a 25% difference among parameters with a power of 80% at the p<0.05 level
[11]. Statistical analyses were performed using the Stat View software package of SAS
Institute Inc. (Cary, USA).
Results
According to the questionnaire data, the AGA and SGA children showed no differences
in their diet (caloric content, food type, etc.) or time spent on outdoor activity.
The mean dietary vitamin D intake estimated on a weekly basis was no different between
the 2 groups: 990±250 (SGA) and 950±235 (AGA) IU/week.
As shown in [Table 1], the mean serum level of 25(OH)D was higher in the SGA group. In 2 SGA and 20 AGA
children, the level of 25(OH)D was below a cutoff value of 15 ng/ml, while in 3 of
the AGA children it was below 10 ng/ml. Only 6 SGA and 2 AGA children had 25(OH)D
levels above 30 ng/ml, which has recently been proposed as the cutoff of adequacy.
Table 1 Anthropometric characteristics and metabolic variables in pre-pubertal children,
born small (SGA) or appropriate (AGA) for gestational age.
|
SGA (n=27)
|
AGA (n=38)
|
|
*p<0.05; **p<0.01; ***p<0.001
|
|
At birth
|
|
Gestational age (weeks)
|
38.4±1.3
|
38.3±1.3
|
|
Birth weight (g)
|
1 894±440***
|
3 338±390
|
|
Birth weight (g) z-score
|
− 3.42±92***
|
− 0.13±0.98
|
|
Birth length (cm)
|
45.1±3.7**
|
51.3±2.6
|
|
At the time of study
|
|
Age (years)
|
5.9±1.6
|
6.2±1.1
|
|
Body weight (kg)
|
25.2±7.1
|
26.7±6.1
|
|
Body height (cm)
|
119±9
|
120.8±9
|
|
BMI (kg/m2)
|
17.4±3.1
|
17.8±2.1
|
|
BMI-SDS
|
− 0.08±1.0
|
0.29±0.7
|
|
Waist circumference-SDS
|
0.04±1.2
|
0.53±1.2
|
|
25(OH)D (ng/ml)
|
26.2±10***
|
17.2±7
|
|
iPTH (pg/ml)
|
21.8±14
|
23.2±9.2
|
|
Fasting glucose (mg/dl)
|
80.2±9.2
|
82.3±6.5
|
|
Fasting insulin (μIU/ml)
|
6.35±3.4*
|
4.62±2.21
|
|
HOMA-IR
|
1.34±0.67*
|
0.99±0.53
|
|
HOMA-β%
|
135±56*
|
97.0±60
|
The insulin resistance indices for fasting insulin and HOMA-IR were higher in the
SGA group, while HOMA-β%, which represents β-cell function, was favorable in this
group. PTH level did not differ between the 2 groups. By comparing children with 25(OH)D
levels below and above 15 ng/ml no differences in insulin resistance and β-cell function
indices were found (data not shown).
In simple regression analysis in the SGA group, the 25-OH D level was positively correlated
with HOMA-β% (R=0.38, p<0.05, [Fig. 1]), but not with HOMA-IR or fasting insulin (R=0.16, 0.14, respectively, p=ns). In
the AGA group 25-OH D was not correlated with either HOMA-β% or HOMA-IR or insulin
or (R=0.17, − 0.08, − 0.07, respectively). PTH showed no correlation with vthe metabolic
indices in any group or the total cohort.
Fig. 1 Relationship between serum 25(OH)D level and HOMA-β% in the small for gestational
age (SGA) children of the study (n=27).
Multiple regression analysis was conducted for the total cohort of children to identify
possible relationships between the birth weight z-score (dependent variable) and 25(OH)D
levels, insulin resistance indices, and obesity indices after appropriate adjustments.
The birth weight z-score, was negatively associated with 25(OH)D (β= − 0.31, p=0.02)
after adjusting for waist circumference, BMI, HOMA-IR, and age ([Fig. 2]). The birth weight z-score was negatively associated with HOMA-IR (β= − 0.36, p=0.003)
after adjusting for BMI, waist circumference z-scores, 25(OH)D, and age. This relationship
was stronger without 25(OH)D in the model (β= − 0.39, p<0.0004) but continued to be
strong after entering 25(OH)D. Finally the birth weight z-score was positively associated
with waist circumference z-score (β=+0.28 p=0.03) after adjusting for HOMA-IR, insulin,
25(OH)D, and age.
Fig. 2 Relationship between serum 25(OH)D level and birth weight (BW) z-scores in the study
children (n=65).
Discussion
This study showed that children born severely SGA had higher levels of 25(OH)D than
those of AGA in prepuberty, affected indices of insulin resistance (insulin and HOMA-IR)
in accordance with the findings of other researchers [7]
[8]
[9] but improved insulin secretion (HOMA-β%). No correlation between 25(OH)D and insulin
resistance was found but a positive relationship between 25(OH)D and HOMA-β% was demonstrated
in the SGA group.
Studies conducted mainly in adults have shown inverse relationships between 25(OH)D
levels and indices of insulin resistance or risk of type 2 diabetes or several other
components of the metabolic syndrome and positive relationships between 25(OH)D levels
and β-cell function [4]
[5]
[12]
[13]
[14], although these findings were not confirmed in some reports [15]. Conversely, in agreement with the present study, several studies in children have
failed to show association between vitamin D and insulin resistance [16]
[17]
[18]
[19]
[20]. Even in studies in children where correlation between vitamin D and several components
of the metabolic syndrome was found, independent relationships between insulin resistance
indices and 25(OH)D were not confirmed [21]
[22]
[23]. Pacifico et al. [23] found that after making appropriate adjustments (including BMI) only blood pressure
and the metabolic syndrome remained significantly correlated with vitamin D, but not
with IR indices (insulin or HOMA). Similarly, Gannagé-Yared et al. [22] showed that vitamin D was an independent predictor of systolic blood pressure and
fasting glucose, but not of insulin or HOMA-IR.
It could be speculated that at this early period of life other factors may influence
vitamin D and insulin resistance. One such factor could be PTH, as vitamin D deficiency
results in hyperparathyroidism [24]
[25], through which glucose metabolism may be affected. Higher PTH, but not lower vitamin
D, was shown to increase the risk of metabolic syndrome in a recent study [26]. In the present study PTH levels were similar in the 2 groups, implying that the
observed difference in vitamin D may not be sufficient to affect the PTH level. Furthermore,
no clear association between PTH level and insulin resistance indices was found.
Another explanation is that the levels of vitamin D, although higher in the SGA group,
were below the recently proposed insufficiency threshold (32 ng/ml), thus lacking
a favorable effect on the insulin resistance, which could possibly confer a higher
25(OH)D serum level. A Canadian study in children showed that even a high increment
in 25(OH)D level (i.e., 10 ng/ml) correlated with only a slight decrease in the fasting
blood glucose level and HOMA- IR [18]. There is also some evidence for a lower threshold by which vitamin D deficiency
confers negative effects on insulin sensitivity [27]. In a previous study in adolescents, this threshold was estimated at 15 ng/ml [27], but application of this threshold to the present study generated no differences
in insulin resistance indices.
Other studies have demonstrated correlation of high 25(OH)D levels with β-cell function
rather than insulin resistance [28]. This was the case in the present study in SGA group where the higher, positively
correlated with 25(OH)D, HOMA-β% imply increased β-cell function. The higher HOMA-β%
may attenuate the risk for impaired glucose tolerance posed by the higher HOMA-IR
in this group, as the risk of the 2 parameters is regarded as additive [29].
The higher vitamin D levels in SGA children are difficult to interpret, but they could
be related to differences in adipose tissue between the 2 groups. Although the groups
were matched for BMI, waist circumference was about 0.5 z-score higher in AGA group.
This difference, although not statistically significant may be physiologically significant
with regard to vitamin D serum level. Higher adiposity values would provide a larger
distribution volume and lower circulating 25(OH)D in the AGA group. Higher 25(OH)D
level was shown to be independently correlated with low birth-weight in the present
study. This relationship may arise from alterations in adipose tissue due to in utero
growth restriction. It has been demonstrated that SGA children have a diminished capacity
to store fat subcutaneously (the metabolic equivalent of a partial lipodystrophy)
[30], thus the SGA children in this study may have had a limited overall storage capacity
for vitamin D, with the result that higher circulating vitamin D levels were measured.
This hypothesis, however, is difficult to prove, as fat partitioning and tissue vitamin
D were not studied in the present study.
Catch-up growth occurs in over 90% of SGA children. This was the case for the SGA
study children, who were all within the normal range of development at prepuberty,
and around the mean, despite their severe growth restriction at birth. The increased
insulin resistance indices may therefore be related with both their low birth-weight
and subsequent accelerated growth [7]
[8]. Obese children were excluded from the present study to avoid the well documented
effect of adiposity on both 25(OH)D level and insulin resistance status, thus the
results of the present study are limited in nonobese SGA children who had experienced
catch-up growth.
Other limitations of this study are the small sample size and the use of surrogates
of glucose homeostasis, although they correlate well with standard methods [10]
[31]. The small number of included children could lead to sample bias limiting the possibilities
of finding other associations beyond those already indentified. Moreover possible
selection bias could not be excluded since the final number of included SGA children
was considerably smaller than the original target population. The duration of sunshine
exposure was based on parental recall, so slight differences in time spent outdoors
could confound the estimations of vitamin D status. The differences in 25(OH)D levels
were too large, however, to be attributed to slight deviations in sun-exposure time
between the 2 groups belonging to the same cohort and living in the same geographical
area (latitude 39° N). Finally, as the study was observational, the alterations found
in the parameters measured may be physiologically unrelated or due to other unrecognized
factors.
To the best of authors’ knowledge, this is the first study to examine 25(OH)D in a
special population with severe in utero growth restriction and a tendency to become
insulin resistant early in life. Nonobese SGA born children had higher serum 25(OH)D
levels, which were correlated with insulin secretion, but not with insulin resistance
indices. Low birth weight appears to be a factor that is correlated independently
with both 25(OH)D levels and insulin resistance in this group of children.
