Main

IUGR is one of the major causes of short stature, and, although most IUGR infants (approximately 80%) show early catch-up growth during the first 6-12 mo of life and attain a final stature appropriate for parental height, those who do not catch-up in the first months of life remain below -2 SD throughout childhood, attain puberty at an early age, and reach a final height lower than genetic potential(1,2). Moreover, IUGR has long-term effects on morbidity and mortality. An increased incidence of cardiovascular diseases and maturity-onset diabetes mellitus has been reported in adulthood, and an IUGR-dependent reprogramming of endocrine development has been suggested as the possible causal mechanism(3,4).

In the complex network of factors which control fetal growth, the IGFs seem to play a crucial role not only due to their stimulation of fetal cell proliferation and anabolism but also to their capacity to determine partitioning of nutrients between the placenta and fetus in favor of the fetus(5).

IGFs are present in extracellular fluids bound to specific, high affinity binding proteins (IGFBPs) that modulate IGF biologic activity(6). To date, six distinct IGFBPs (IGFBP-1 through IGFBP-6) have been characterized(7). The affinity of IGFBPs for IGFs may be reduced by the intervention of specific cation-dependent proteolytic enzymes (IGFBP proteases) that fragment IGFBPs, leading to increased free IGF fraction and, therefore, augmented IGF bioavailability. IGF-I and -II, type 1 and -2 IGF receptors, IGFBPs, and IGFBP proteases form, as a whole, the IGF system which is actively involved in pre- and postnatal growth.

Direct correlation of umbilical cord serum IGF-I levels with birth weight has been repeatedly reported(8–14). In IUGR children, IGF-I and IGFBP-3 levels are reduced, whereas IGFBP-1 and IGFBP-2 concentrations have been found high both in the cord serum and in samples obtained in utero(12,15–18). Direct evidence for the pivotal role of IGF system in fetal growth has been provided by the finding that targeted mutagenesis of the genes encoding IGF-I and -II and type 1 IGF receptor induces profound embryonic, fetal, and postnatal growth retardation in mice(19,20). In humans, a patient with severe IUGR due to the deletion of IGF-I gene has been recently described(21). Furthermore, transgenic mice overexpressing IGFBP-1 show fetal growth retardation and impaired brain development, thus suggesting an inhibitory effect of IGFBP-1 on IGF growth-promoting action in the fetus(22). Finally, the elevated IGFBP-3 proteolytic activity found in mothers with multiple fetuses or fetuses affected by uteroplacental insufficiency is probably finalized to increase IGF bioavailability in situations in which fetal growth is threatened(23).

Despite that most of the IUGR children normalize growth in early postnatal life, few data on the IGF system-related variables in the first months of life are available. The aim of the present study was to assess the IGF system status in IUGR children in the first 2 months of life, relating the endocrine variables with growth parameters.

METHODS

Experimental subjects. Thirty-seven IUGR children with a birth weight below the 10th centile for gestational age according to the standards of Lubchenko(24) and 25 children with birth weights appropriate for gestational age (above the 25th centile) were included in the study. All mothers of subjects had a full-term uneventful pregnancy (38-42 wk). All the control children were born by spontaneous vaginal delivery, whereas, among IUGR newborns, 14 were born by vaginal delivery and 23 by cesarean section. Newborns with malformations or genetic disorders were excluded.

Thirty-five out of the 37 IUGR children underwent anthropometric measurements, including weight and supine length at birth, and weight, length, and knee-heel length at 7, 14, 30, 60, and 90 d of life. Serum samples were collected from umbilical cord of IUGR and control groups at birth. At the age of 2 mo a second blood sample was collected from an antecubital vein in 27 out of the 37 IUGR children. The second blood sample was collected between 0800 and 1200 h, at least 1 h after meals. Serum was stored at -25°C until assay.

The investigation was approved by the Ethical Committee of the "Tor Vergata" University Medical School, and written informed consent was obtained from parents of all children.

Anthropometric measurements. Length was measured with a portable infantometer (Rollametre, Raven Equipment Ltd., UK). Two observers were trained over 2 mo. A pilot study on 66 term newborns showed that the within-observer technical error was 4 mm and the between-observer error was 6 mm.

The P.I. (weight (g)/length (cm)3) was used to discriminate between children with proportionate and nonproportionate intrauterine growth retardation. Children with P.I. ≥ 10th centile of reference values(25) were defined as proportionate, whereas those with P.I. < 10th centile were defined as nonproportionate. Catch-up growth was defined as length and weight increments greater than the 50th centile of the references values(26).

Knee-heel length was measured with a portable knemometer (Force Institute, Denmark). The instrument was described by Michaelsen et al.(27) and is made of a fixed graduated rod ending with a knee cap and a sliding rod, fitted with a heel cap and connected to a digital caliper that can be read with an accuracy of 0.01 mm. A series of five measurements was taken, each time removing and repositioning the knemometer. If the SD exceeded 0.8 mm, the entire series of five assessments was repeated. The two observers were trained over 2 mo, and a pilot study on 97 term newborns showed that within-observer technical error was 1 mm, and between-observer error was 1.6 mm.

Reagents. Polyacrylamide, SDS, ammonium persulfate, tetramethylethylendiamine, Tween 20, and SDS-PAGE standards were from Bio-Rad Labs (Hercules, CA). ECL protein biotinylation system and Hyperfilm-ECL were from Amersham International.

Antibodies. Rabbit polyclonal antibody αIGFBP-3 g1 was a kind gift from Prof. R. G. Rosenfeld (Oregon Health Science University, Portland, OR)(28). Horseradish peroxidase-conjugated donkey anti-rabbit IgG was from Amersham. Recombinant human glycosylated IGFBP-3 (45 kD) was kindly provided by Prof. R. G. Rosenfeld.

Assays. GH was measured by RIA (HGH kit liso-phase, Technogenetics, Italy). The intraassay CV was 5.7%, the interassay CV 7.2% and the limit of sensitivity 0.2 µg/L. IGF-I was determined by RIA(Nichols Institute, San Juan Capistrano, CA) after both acid-ethanol extraction and acid chromatography extraction by C18 Sep-Pak columns (Waters Assoc., Milford, MA). For the acid-ethanol extraction method, the intraassay CV was 3.0%, the interassay CV 8.4%, and the limit of sensitivity 13.5µg/L. For the C18 Sep-Pak column extraction method, the intraassay CV was 2.9%, the interassay CV 11.4% and the limit of sensitivity 12.86%µg/L. According to the recommendations for the valid measurement of total IGF-I concentrations in biologic fluids(29), in 33 samples IGF-I was also measured by RIA after acid-acetone extraction(30), and potential residual IGFBP binding sites were blocked by adding 10 ng of IGF-II in each tube in the IGF-I assay. IGFBP-3 was assessed by RIA (Diagnostic Systems Laboratories Inc., Webster, TX). The intraassay CV was 6.7%, the interassay CV 10.1%, and the limit of sensitivity 0.2 mg/L. IGFBP-1 was measured by an imunoenzymatic monoclonal assay(IEMA-test, Medix Biochemica, Finland). The intraassay CV was 3.4%, the interassay CV was 7.4%, and the limit of sensitivity 0.4 µg/L. C-peptide was determined by a commercial RIA kit (Byk-Sangtec Diagnostica, Germany). The intraassay CV was 5.0%, the interassay CV 10.0% and the limit of sensitivity 0.03 µg/L.

Western immunoblot analysis. Immunoblotting was performed as described by Liu et al.(31) except that BSA was used instead of nonfat dry milk for blocking nonspecific binding. Briefly, after addition of nonreducing SDS sample buffer, serum samples (3 µL) were processed by SDS-PAGE (12% gel) at 25 mA for stacking gel and 35 mA for separating gel for 5 h. Prestained molecular weight standards were diluted in nonreducing SDS sample buffer and processed in parallel. Separated proteins were electroblotted onto nitrocellulose filters in a Hoefer Semi-dry Transphor unit (San Francisco, CA) at 200 mA. Filters were sequentially treated with 1% BSA in 0.1 M TBS (Tris-HCl, 0.05 M NaCl). Then filters were sequentially incubated with anti-IGFBP-3 antibody (αIGFBP-3 g1 antibody was used at 1:1000 dilution in TBS) overnight at 4°C and with goat anti-rabbit IgG conjugated with horseradish peroxidase (Amersham) for 2 h at room temperature, with washes in 0.1% Tween 20 in TBS at 20°C in between. Filters were exposed to enhanced chemiluminescence reagents (Amersham) for 1 min at 20°C and exposed to hyperfilm ECL for 1 min to 1 h at 20°C. Densitometric analysis of bands was performed using a Bio-Rad GS 700 imaging densitometer (Bio-Rad, Richmond, CA). All the experiments were performed in duplicate.

Proteolytic activity. To test proteolytic activity in IUGR samples, 7 µL of IUGR serum were incubated with 3 µL of normal adult serum for 4 h at 37°C. As a control, third trimester pregnancy serum (5 µL) was incubated with normal adult serum (5 µL) for 4 h at 37°C. To inhibit protease action EDTA (at a final concentration of 25 mM) was added.

Statistics. Results are reported as the mean ± SEM. Differences between means were assessed using an unpaired two-tailed t test and one-way ANOVA. A paired t test was used to compare intra-IUGR group mean differences at birth and at 2 mo. Significance was assigned for p < 0.05. The relationship between parameters was evaluated by Pearson correlation and forward stepwise regression analysis. All independent variables were assessed step by step, and only those with a significant t value (p < 0.05) were included in the final regression model. Natural logged IGF-I and GH values were used for parametric statistical methods because these data resulted nonnormally distributed. A computer program was used for all statistical calculations (BMPD Statistical Software, SOLO 3.0, BMPD, Los Angeles, CA).

RESULTS

Anthropometry. Mean birth weight of control subjects (3.49± 0.08 kg) was significantly higher than that of the IUGR group (2.32± 0.03 kg, p < 0.0001). Mean P.I. of the IUGR newborns was 2.29 ± 0.04. According to the P.I., 21 IUGR children were classified as proportionate and 16 nonproportionate. In IUGR subjects, the mean weight gain (ΔW) from 0 to 90 d was 2.90 ± 0.08 kg, the mean length increase (ΔL) was 10.41 ± 0.30 cm, and the mean knee-heel growth (ΔKH) was 32.65 ± 1.09 mm. A significant relationship was seen between ΔW and ΔL (r = 0.65,p = 0.0001), ΔW and ΔKH (r = 0.63,p = 0.0001), and ΔL and ΔKH (r = 0.41,p = 0.02) at 3 mo of life. Among the IUGR children, in the first trimester only three did not recover either weight or length. Thirteen did not catch-up weight and six did not recover length.

Endocrine variables. A high degree of correlation was observed between IGF-I levels after acid-ethanol extraction and those after C18 Sep-Pak column extraction method (r = 0.76, p < 0.0001), although the latter method resulted in IGF-I levels systematically higher. In addition, IGF-I levels assessed after acid-ethanol extraction closely correlated with those obtained after acid-acetone extraction and addition of an excess of IGF-II to block potential residual IGFBP binding sites (r = 0.80, p < 0.0001) that resulted slightly but not significantly lower (mean ± SEM: 6.4 ± 0.65versus 5.6 ± 0.61 nmol/L). Here, IGF-I concentrations determined after acid-ethanol extraction are reported.

IUGR newborns had significantly higher concentrations of GH and IGFBP-1, and lower IGF-I, IGFBP-3, and C-peptide levels than did control subjects(Table 1). No significance difference in any of the endocrine variables was observed between IUGR children born by vaginal delivery and those by cesarean section.

Table 1 Endocrine variables at birth (umbilical cord serum) and 2 mo of life
Full size table

In IUGR infants, the endocrine assessment performed at the age of 2 mo showed, in comparison with cord blood, a significant reduction of GH and IGFBP-1, and a significant increase of IGF-I, IGFBP-3, and C-peptide(Table 1).

Relationships between anthropometric and endocrine variables. No relationship was observed between endocrine variables and anthropometric measurements in the IUGR group either at baseline or at 2 mo. No significant difference was seen in endocrine data between IUGR children who showed a catch-up growth in weight and/or length at 3 mo and those who did not. Finally, no significant difference in anthropometric and endocrine variables was observed between proportionate and nonproportionate IUGR newborns. IUGR subjects (n = 6) who did not recover length at 3 mo showed significantly higher birth weight (p < 0.005) and P.I.(p = 0.05) than children who caught-up. In the control group, C-peptide was related to birth weight (r = 0.40, p < 0.05).

Relationships between endocrine variables. In IUGR children, a close inverse correlation between IGF-I and IGFBP-1 was shown at birth(r = -0.44, p = 0.01, Fig. 1), and this relationship was confirmed at 2 mo (r = -0.41, p< 0.05). Furthermore, ΔIGF-I was related to ΔIGFBP-1(r = -0.57, p = 0.01). IGFBP-1 did not correlate with C-peptide. The relationship between IGF-I, GH, C-peptide (independent variables), and IGFBP-1 (dependent variable) was evaluated by regression analysis. Only IGF-I (t = -2.61, p = 0.01) showed a significant relationship and was included in the final regression equation:y = 189.93 - 28.58x1, R2 = 0.19, F = 6.82, p = 0.01; where y = IGFBP-1 and x1 = In IGF-I. Also IGFBP-3 was closely related to IGF-I at birth (r = 0.45, p < 0.005,Fig. 2), but this relationship was lost at 2 mo. The relationship between IGF-I, GH, and C-peptide (independent variables) and IGFBP-3 (dependent variable) was assessed. Only IGF-I was significantly related (t = 3.02, p = 0.005) and was included in the final regression equation: y = 0.216 + 0.193x,R2 = 0.20, F = 9.1, p = 0.005; where y = IGFBP-3 and x = In IGF-I. IGFBP-3 and IGF-I did not show any correlation with GH levels. In control subjects, no correlation among endocrine variables was demonstrated.

Figure 1
figure 1

The relationship between IGF-I concentrations (expressed as loge IGF-I) and IGFBP-1 levels in serum samples of IUGR children at birth. IGF-I was determined by specific RIA and IGFBP-1 by IEMA in serum obtained from the umbilical cord; r= -0.44, p = 0.01.

Full size image
Figure 2
figure 2

The relationship between IGF-I concentrations (expressed as loge IGF-I) and IGFBP-3 levels in serum samples of IUGR children at birth. IGF-I and IGFBP-3 were determined by specific RA in serum obtained from the umbilical cord; r = 0.45, p < 0.005.

Full size image

IGFBP-3 immunoblotting. Control subjects showed the normal intact 42-39-kD IGFBP-3 doublet and the major ∼29-kD fragment both at birth and 60 d (Fig. 3). In addition, a ∼18-kD band was seen in 4/6 control subjects at birth (Fig. 3). Serum samples from 19/27 IUGR children were tested at birth and 60 d. Birth samples of 14/19 IUGR infants showed the presence of the 42-39-kD doublet and the ∼29-kD IGFBP-3 form (Fig. 4,A and B). The∼18-kD IGFBP-3 fragment was present at birth in four, at 60 d in six, and completely absent at birth and 60 days in nine IUGR samples. It is noteworthy that all children who showed the appearance of the ∼18-kD fragment at 60 d were proportionate IUGR, whereas 3/4 subjects showing the fragment at birth and 7/9 infants with the total absence of the fragment were nonproportionate IUGR.

Figure 3
figure 3

Detection of IGFBP-3 forms by Western immunoblotting in sera from control children at birth and 60 d. Serum samples(3 µL), after addition of nonreducing SDS sample buffer, were processed by SDS-PAGE (12% gels), and electroblotted onto nitrocellulose filters. Nitrocellulose was incubated with αIGFBP-3 g1 antiserum and binding, after incubation with goat anti-rabbit IgG conjugated horseradish peroxidase, was detected using the enhanced chemiluminescence detection system. Autoradiographs were developed after 1-60-min exposure. Molecular mass markers are indicated on the left.

Full size image
Figure 4
figure 4

(A and B) Detection of IGFBP-3 forms by Western immunoblotting in sera from IUGR children at birth and 60 d.

Full size image

Proteolytic activity assay. After incubation of serum samples from IUGR children at birth (n = 9) with normal adult serum, an increase of the intensity of the 29-kD band was observed (Fig. 5, lanes C and G). This increase disappeared after incubation with EDTA (Fig. 5,lanes D and H). The incubation of serum samples from IUGR subjects at 60 d of life (n = 6) with adult serum induced the appearance of the 18-kD band fragment (Fig. 5,lanes E and I), which was only slightly reduced after EDTA addition (Fig. 5, lanes F and L).

Figure 5
figure 5

Proteolytic activity assay. IUGR serum samples (7 µL) were incubated with normal adult serum (3 µL) for 4 h at 37°C. As a control, third trimester pregnancy serum (5 µL) was incubated with normal adult serum (5 µL) for 4 h at 37°C. To inhibit protease action EDTA (at final concentration of 25 mM) was added. Samples were then tested with Western immunoblotting for IGFBP-3 as described above. Lane A, recombinant human glycosylated IGFBP-3 (45 kD). Lane B, normal adult serum. Lanes C G, coincubation of normal adult serum with birth IUGR serum. Lanes D H, coincubation of normal adult serum with birth IUGR serum and addition of EDTA (25 mM). Lanes E I, coincubation of normal adult serum with 60-d IUGR serum. Lanes F L, coincubation of normal adult serum with 60-d IUGR serum and addition of EDTA (25 mM). Lane M, coincubation of normal adult serum with third trimester pregnancy serum. Lane N, coincubation of normal adult serum with third trimester pregnancy serum and addition of EDTA (25 mM).

Full size image

DISCUSSION

Our data from cord blood samples are in agreement with previous investigations showing high levels of GH and IGFBP-1 and low concentrations of IGF-I, IGFBP-3, and C-peptide in cord serum of IUGR newborns(12,15–18). At 2 mo, GH, IGF-I, IGFBP-3, IGFBP-1, and C-peptide levels overlapped those found in newborns with birth weight appropriate for gestational age. In normal children, IGF-I levels have been reported either to remain constant(32,33) or progressively decline from birth to 6 mo(34), whereas IGFBP-3 has been shown to physiologically increase during the first 3 mo of life(33,34). In normal infants, IGFBP-1 decreases and insulin does not vary significantly in early postnatal life(32). Finally, GH concentrations are high in cord blood from normal newborns and have been reported either to remain elevated(35) or to physiologically decrease(32,33) during the first 8-12 wk of life.

The reduced concentrations of IGF-I, IGFBP-3, and C-peptide found in cord blood from IUGR children might be due to chronic exposure to subnormal supply of nutrients during intrauterine life. The increased IGFBP-1 levels may signify an accumulation of this "shuttle" protein in the circulation, resulting in decreased transport of IGF to target tissues and reduced growth. The high GH concentrations might reflect partial GH receptor deficiency(32) as well as the lack of the negative feed-back exerted by IGF-I on hypothalamus-pituitary axis.

Although no relationship between C-peptide and IGFBP-1 levels was observed in IUGR children, a significant relationship between IGF-I and IGFBP-1 was found. A similar close relationship between IGF-I and IGFBP-1 has been also reported in small for gestational age rats(36) and normal human fetuses(37). These in vivo findings together with the in vitro observation that IGF-I and its potent analog Des(1-3)IGF-I inhibit the release of IGFBP-1 from the hepatoma cell line Hep G2(38), suggest that IGF-I may directly regulate the hepatic production of IGFBP-I during fetal and early postnatal life. We have also found a close relationship between IGF-I and IGFBP-3 in IUGR newborns. This finding is consistent with the increasing body of evidence suggesting that IGF-I mediates much of the GH dependency of IGFBP-3in vivo(39,40) and in vitro(41–43).

The question whether the IGF system-related endocrine pattern at birth may be predictive of the growth outcome of IUGR children remains to be addressed. No relationship was observed between the anthropometric parameters and the endocrine variables either considered as baseline values or as 2-mo variations. However, the apparent failure of the endocrine variables to predict growth in the first trimester of postnatal life might be due to the short follow-up. To overcome this bias we used knemometry, which has been shown to be an accurate technique for measuring linear growth velocity over short periods(27). However, even taking into account the variation of knee-hell length over 90 days, no relationship between growth outcome and endocrine variables was found. It is noteworthy that Leger et al.(33) have recently reported, after studying a large cohort of IUGR children, that GH, IGF-I, and IGFBP-3 levels are not predictive of the length gain in the first two years of life.

In the circulation more than 75% of the IGFs are complexed with IGFBP-3 and an acid-labile subunit(44) in a ∼150-kD molecular mass ternary complex that represents a storage pool for circulating IGFs(45). Proteolysis of IGFBP-3 was first described in pregnancy serum, and proteases have been shown to be calcium-dependent serine proteases(46–49). Our results on IGFBP-3 circulating forms show that the typical ∼42-39-kD doublet and the ∼29-kD forms are predominant in cord blood samples from both normal and IUGR children. The ∼29-kD IGFBP-3 fragment has been shown to be glycosylated, to bind IGFs(50) and to be the predominant form of IGFBP-3 in normal human fetal serum(51). More interestingly, we first demonstrated the presence of a ∼18-kD IGFBP-3 fragment in umbilical cord serum from normal newborns and the predominance of this fragment in one third of IUGR sera at 2 mo of life in concomitance with a dramatic reduction of the ∼42-39-kD and∼29-kD band intensity. The ∼18-kD form of IGFBP-3 has been recently characterized in urine and serum from healthy children and GH deficient patients and found to be glycosylated, able to bind IGFs and form ternary complex at a lower molecular mass(52). The ∼18-kD IGFBP-3 fragment was found to be more represented in specimens from healthy children than healthy adults, and in GH-deficient patients during GH therapy than GH-deficient subjects off therapy(52). The presence of the ∼18-kD IGFBP-3 form in sera from normal newborns, healthy children, GH-deficient patients on substitutive therapy, and our IUGR children at 2 mo, suggests a relation between the expression of this fragment and body growth rate. The predominance of the ∼18-kD IGFBP-3 form might reduce ternary complex formation determining the presence of a higher proportion of IGF-I in lower molecular mass, binary complexes, which have shorter half-lives than the ternary complex, and, therefore, increase IGF bioavailability.

Interestingly, all six IUGR infants with the appearance of the ∼18-kD band at 2 mo were proportionate at birth, thus suggesting an early onset of intrauterine growth retardation with consequent early adaptation to an unfavorable intrauterine environment. It is tempting to speculate that proportionate IUGR newborns might require a longer time to normalize their growth-promoting machinery.

The identity and regulation of IGFBP-3 proteases are still unknown. IGF-I itself and insulin have been postulated as possible modulators of IGFBP protease activity(53,54). In keeping with these previous reports, we have observed high IGFBP-3 proteolytic activity yielding the major ∼29-kD fragment in birth IUGR samples, together with subnormal levels of IGF-I and C-peptide. It is noteworthy that the appearance of the ∼18-kD fragment at 2 mo of life occurred in concomitance with a significant increase of IGF-I and C-peptide and partial or total disappearance of the other IGFBP-3 forms. In addition, our data on EDTA-mediated inhibition of proteolytic activity indicate none or only marginal effects on the∼18-kD band. These findings suggest the presence of at least two different proteases in IUGR children, one present at birth and able to yield the∼29-kD form as the major proteolytic fragment, and a second one, differently regulated, which is active at 2 mo and produces the ∼18-kD fragment. The intervention of this second protease in IUGR infants might be finalized to amplify the growth-promoting action of the IGF system, increasing IGF availability in tissues and thus cooperating in the postnatal catch-up growth which occurs in the first 6-12 mo of life. Alternatively, increasing concentrations of a single protease may lead to progressively smaller fragments of IGFBP-3, thus explaining the appearance of the∼18-kD band.

In conclusion, our findings suggest that the growth-promoting machinery, one of whose main gears is the IGF system, is frozen in IUGR fetuses probably to minimize the energy expenditure for growth and favor survival and development of vital organs, and seems to be fully operative at 2 mo of life, thus allowing the catch-up growth in the vast majority of IUGR children.

Acknowledgments. The authors thank Prof. R. G. Rosenfeld(Oregon Health Sciences University, Portland, OR) for his generous gift ofαIGFBP-3 g1 antibody and recombinant human glycosylated IGFBP-3 (45 kD). We also thank Dr. A. Spagnoli ("Tor Vergata" University, Rome, Italy) for helpful suggestions.