Main

Prenatal and postnatal growth restriction is a challenging problem in perinatology and neonatology. Normally the last trimester of pregnancy is characterized by a significant increase in the velocity of fetal growth, regulated by IGFs and their binding proteins (IGFBPs). They act via both endocrine and paracrine routes in promotion of cell proliferation and differentiation and in induction of protein synthesis (1). Collagens are the major connective tissue proteins, type I collagen being the only collagen found in mineralized bone and the most abundant collagen together with type III collagen, e.g. in skin, muscles, internal organs, and blood vessels (2, 3). The turnover of collagen is reflected in blood by markers of its synthesis and degradation. The markers of synthesis include the aminoterminal and carboxyterminal propeptides of type I procollagen, PINP and PICP, and the aminoterminal propeptide of type III procollagen, PIIINP. The markers of degradation include the aminoterminal and carboxyterminal telopeptides of type I collagen, INTP and ICTP (4).

Data on collagen turnover in the fetus are scarce. Cord blood PICP and ICTP (5) and the first postnatal urine INTP decrease with increasing gestational age, but no relationship with birth weight has been observed in term or near-term infants (6–8). PIIINP decreases toward term, reflecting fetal somatic growth velocity (9), and the PIIINP concentration has been lower in SGA than in AGA infants (10). During childhood, PICP and PIIINP reflect growth velocity (11).

Extensive data are available on the relationship of the fetal IGF system with fetal growth and maturity. Fetal IGF-I and IGFBP-3 are lower and IGFBP-1 higher in smaller infants (12–17), but data on the growth-regulating role of circulating IGF-II in the fetus are contradictory (12–14, 16). The IGFBP-1 concentrations are high also in the maternal circulation and amniotic fluid when the fetus is growth-retarded (18, 19). This may indicate that IGFBP-1 limits availability of free IGF-I in the fetus or at the fetomaternal interface where transfer of oxygen and nutrients takes place. In preeclampsia, both fetal and placental growth are often retarded, and both maternal and fetal IGFBP-1 are increased and IGFBP-3 is decreased (19–22).

In clinical practice, it is difficult to assess growth of sick premature infants, and biochemical growth markers would be useful. The use of IGFs and IGFBPs as such indicators is limited because they may be subject to measurement variation (23, 24). IGFs and IGFBPs regulate collagen synthesis and degradation in vitro(25–28), IGF-I administration increases collagen synthesis in adults (29), and pregnant women show a close correlation between IGF-I and PICP and PIIINP (30). Therefore, it is logical to examine whether markers of collagen metabolism would reflect growth and IGF and IGFBP action in the growing fetus and premature infant. This point has been highlighted by recent studies reporting a correlation between postnatal growth and PICP (31) and PIIINP (32). However, little is known about the relationship between the IGF system and collagen metabolism in premature infants.

With this background, we addressed the following questions. First, how do markers of type I and III collagen turnover and the IGF system in cord blood of premature infants relate to fetal growth and gestational age? Second, is there an independent interrelationship between these markers of type I and III collagen turnover and members of the IGF family, as suggested by results obtained in vitro and in vivo in studies in children and adults?

METHODS

Study Population

The 98 preterm infants (Table 1) born before 32 wk of gestation at the Department of Obstetrics and Gynecology, Helsinki University Central Hospital, Helsinki, Finland, had their gestational age confirmed by ultrasound before 20 wk of gestation. The study was approved by the Institutional Review Board of the Department of Obstetrics and Gynecology, Helsinki University Central Hospital.

Table 1 Clinical data and factors associated with prematurity
Full size table

Birth Weight SDS and Placenta/Infant Weight Ratio

The premature infants and their placentas were weighed immediately after birth. Because newborn length could not be recorded for all sick premature infants, it could not serve as a variable. To describe intrauterine growth in units adjusted for gestational age, a birth weight SDS was determined with reference to a Finnish newborn population of 74,766 singletons born from 1978 to 1982 (33). With infant birth weight, gestational age, and sex, each newborn infant's relative birth weight was expressed in birth weight SDS units. Because placental weight was closely correlated with birth weight SDS (r = 0.62;p < 0.0001) and absolute birth weight (r = 0.55;p < 0.0001), a placenta/infant weight ratio was calculated to express relative placental size.

Clinical Data

Clinical data came from hospital records. Preeclampsia was diagnosed on the basis of repeatedly elevated maternal blood pressure (≥140/90 mm Hg or a 15 mm Hg increase in diastolic and 30 mm Hg increase in systolic blood pressure) and proteinuria (≥0.3 g/24 h). No mother received magnesium sulfate. Amnionitis was diagnosed on the basis of maternal or fetal tachycardia, fever, leukocytosis, and elevated serum C-reactive protein. Maternal labor was not used as a variable because the heterogeneous clinical characteristics of the nonlabor group (Table 1) would have been likely to mask the possible effects of labor itself. For the fetal blood gas analysis, a heparinized syringe was used to aspirate blood from a single artery of a double-clamped cord immediately after birth. Cord artery base excess rather than pH was used as a variable in the data analysis because it gives a more specific estimate of metabolic acidosis possibly associated with tissue hypoxia.

Betamethasone (12 mg intramuscularly twice at 12-h intervals, treatment repeated in 7 to 10 d) served as an antenatal glucocorticoid treatment when preterm birth was imminent between 24 and 32 wk of gestation (Table 1). The number of treatments (0 if no betamethasone) was used as a variable in the data analysis.

Biochemical Analyses

Blood sampling.

Blood samples from the umbilical vein were drawn into EDTA-containing tubes, which were spun at 1000 ×g for 5 min, with plasma stored at −20°C until analysis. Because of the small volume of some blood samples, not all assays could be performed for all infants (Table 2).

Table 2 Concentrations of biochemical factors
Full size table

PINP, ICTP, and PIIINP assays.

The intact PINP, ICTP, and intact PIIINP concentrations were determined by specific RIAs against human antigens (Orion Diagnostica Ltd, Espoo, Finland). Of the several assays developed to assess degradation of type I collagen (4), we chose the serum assay for ICTP. ICTP is destroyed by cathepsin K, which is active during normal bone resorption. This assay thus most likely reflects the matrix metalloproteinase pathway of type I collagen degradation (34). Inter- and intraassay coefficients of variation for the ICTP assay vary in six different concentrations between 4.1 and 7.9% and 2.8 and 6.2%, respectively. The same figures for four different concentrations in the intact PINP assay are 3.1–8.2% and 4.8–13.7%, and for the intact PIIINP assay, 4.1–18.0% and 4.4–6.4%, respectively. The detection limits of these assays are PINP, 2 μg/L; ICTP, 0.5 μg/L; and PIIINP, 0.2 μg/L.

IGF-I, IGF-II, IGFBP-1, and IGFBP-3 assays.

The IGF-I-concentrations were determined by ELISA (Diagnostic Systems Laboratories, Webster, TX, U.S.A.). Inter- and intraassay coefficients of variation are 8.8% and 7.1%, respectively. The detection limit of this kit is 6.5 μg/L. The IGF-II concentrations were measured by competitive immunofluorometric assay as described (35). Inter- and intraassay coefficients of variation were 4.9–13% and 7.2–8.6%, respectively, and the detection limit was 120 μg/L. The IGFBP-1-concentrations were determined by immunoenzymometric assay as described (36). Inter- and intraassay coefficients of variation were 7.4% and 3.4%, respectively, and the detection limit was 0.4 μg/L. The IGFBP-3 concentrations were determined with immunofluorometric assay as described (37). Inter- and intraassay coefficients of variation were 4.9–11% and 3.6–6.2%, respectively, and the detection limit was 0.3 μg/L.

Data Analysis

The results are mean values ± SEM. Logarithmic transformation was used to normalize the data when appropriate. Correlations between the variables were estimated with simple linear regression analyses, and multiple stepwise linear regression was calculated by use of all variables with significant univariate correlation (p < 0.05) to determine which of the determinants had predictive values for the dependent determinant. When two groups were studied, t test for unpaired samples was applied. The difference was considered significant at p < 0.05. The calculations were performed on SPSS for Windows 8.0.1 (SPSS Inc., Chicago, IL, U.S.A.).

RESULTS

Markers of Collagen Turnover

PINP and ICTP.

The PINP concentration was 2956 ± 223 μg/L, and the ICTP concentration was 153 ± 7 μg/L (Table 2). Both concentrations were about 50-fold higher than the mean concentrations for healthy adults (4), with a positive correlation between PINP and ICTP (r = 0.69;p = 0.0001). A negative correlation existed between gestational age and PINP, and between gestational age and ICTP (Fig. 1), and positive correlations between birth weight SDS and PINP, and between birth weight SDS and ICTP (Fig. 1). In addition, the number of antenatal glucocorticoid treatments showed a negative correlation with ICTP (r = −0.23;p = 0.04;Fig. 2) and cord artery base excess with PINP (r = 0.24;p = 0.03). In preeclampsia, both the PINP and ICTP concentrations were lower (PINP 1895 ± 386 versus 3213 ± 251 μg/L, p = 0.005; ICTP 109 ± 14 versus 163 ± 7 μg/L, p = 0.002), and in infants of mothers with chorioamnionitis ICTP, but not PINP, was higher (187 ± 15 versus 144 ± 8 μg/L, p = 0.003). However, the associations with preeclampsia and chorioamnionitis did not remain significant in a multiple regression model (see Collagen Turnover and the IGF System). No correlation was observed with cord artery pH or placenta/infant weight ratio. A positive correlation existed between ICTP/PINP ratio and gestational age (r = 0.29;p = 0.008) but not between ICTP/PINP ratio and other clinical data.

Fig 1
figure 1

Correlation between markers of collagen turnover and gestational age and birth weight SDS. Correlation coefficients and p values are indicated.

Full size image
Fig 2
figure 2

Box plots (median, range, and interquartile values) of cord plasma IGF-I, IGF-II, IGFBP-1, IGFBP-3, and ICTP with regard to the number of maternal antenatal glucocorticoid treatments. p values for linear trend are indicated. All associations remained significant also when adjusted for birth weight SDS, gestational age, and preeclampsia.

Full size image

PIIINP.

The PIIINP concentration in cord blood was 215 ± 23 μg/L (Table 2), i.e. more than 50-fold higher than the mean concentration for healthy adults (4). A negative correlation appeared between PIIINP and gestational age (Fig. 1). The PIIINP concentration was significantly higher in male than in female infants (268 ± 40 versus 154 ± 14 μg/L, respectively, p = 0.003;Fig. 3). In multiple regression analysis, both gestational age (β = −0.35;p = 0.001) and sex (β = −0.25;p = 0.02) remained significantly associated with PIIINP (R2 = 0.22). No interrelationships existed between PIIINP and birth weight SDS (Fig. 1) or any other clinical data.

Fig 3
figure 3

Box plot (median, range, and interquartile values) of cord plasma PIIINP in male and female infants. p value is indicated.

Full size image

The IGF System

IGF-I.

The IGF-I concentration was 27.1 ± 2.4 μg/L (Table 2). Correlations were positive between IGF-I and IGFBP-3 (r = 0.79;p < 0.0001), and negative between IGF-I and IGFBP-1 (r = −0.51;p = 0.0002) and between IGF-I and IGF-II (r = −0.45;p = 0.001). In addition, a positive correlation existed between IGF-I and birth weight SDS (Fig. 4), but the association with gestational age was not statistically significant (Fig. 4). Moreover, there was a positive correlation between IGF-I and the number of antenatal glucocorticoid treatments (r = 0.38;p = 0.002;Fig. 2), as well as cord artery base excess (r = 0.35;p = 0.007). In infants of mothers with preeclampsia, IGF-I was lower (13.3 ± 4.5 versus 29.9 ± 2.7 μg/L, p = 0.002;Fig. 5). There was no association with cord artery pH, chorioamnionitis, or placenta/infant weight-ratio. When we performed multiple regression analysis to assess which clinical variables explained variation in IGF-I levels, birth weight SDS (β = 0.39;p < 0.0001), number of antenatal glucocorticoid treatments (β = 0.33;p = 0.001), preeclampsia (β = −0.28;p = 0.009), and cord artery base excess (r = 0.22;p = 0.02) all remained significantly associated with IGF-I (R2 = 0.54).

Fig 4
figure 4

Correlation between IGF-I, IGF-II, IGFBP-1, and IGFBP-3 and gestational age and birth weight SDS. Correlation coefficients and p values are indicated.

Full size image
Fig 5
figure 5

Box plots (median, range, and interquartile values) of cord plasma IGF-I and IGFBP-1 in infants of mothers with and without preeclampsia. p values are indicated.

Full size image

IGF-II.

The IGF-II concentration was 523 ± 40 μg/L (Table 2). We found a negative correlation with birth weight SDS (Fig. 4) and the number of antenatal glucocorticoid treatments (r = −0.28;p = 0.02). No significant correlation appeared between IGF-II and IGFBP-1, IGFBP-3, gestational age (Fig. 4), or other clinical data. In multiple regression analysis, both birth weight SDS (r = −0.34;p = 0.003) and the number of glucocorticoid treatments (r = −0.26;p = 0.02) remained significantly associated with IGF-II (R2 = 0.19).

IGFBP-1.

The IGFBP-1 concentration was 297 ± 61 μg/L (Table 2), with no interrelationship with gestational age (Fig. 4) but a negative correlation with birth weight SDS (Fig. 4) and the number of antenatal glucocorticoid treatments (r = −0.26;p = 0.02;Fig. 2). The IGFBP-1 concentration was higher in infants of mothers with preeclampsia (551 ± 180 versus 255 ± 63 μg/L;p < 0.0001;Fig. 5), and in boys than in girls (415 ± 112 versus 182 ± 42 μg/L;p = 0.009). No correlation was apparent with cord artery base excess, chorioamnionitis, or placenta/infant weight-ratio. Four infants had very high IGFBP-1 concentrations (1500–3100 μg/L), the remaining IGFBP-1 concentrations being <640 μg/L. All these outliers were boys; three had a birth weight SDS below −2 SD, one's mother had preeclampsia, and none of them had a cord pH < 7.15 (one pH was unavailable). Considering that IGFBP-1 secretion is pulsatile with high amplitude (23, 24), we performed another analysis after exclusion of these outliers. The significant inverse correlations remained between IGFBP-1 and birth weight SDS (r = −0.35;p = 0.002) and between IGFBP-1 and the number of antenatal glucocorticoid treatments (r = −0.23;p = 0.05). Moreover, IGFBP-1 remained significantly higher in infants of mothers with preeclampsia (377 ± 49 versus 152 ± 18 μg/L;p < 0.0001) and in boys compared with girls (222 ± 28 versus 147 ± 25 μg/L;p = 0.02). In multiple regression analysis, preeclampsia (β = 0.42;p = 0.0001) and the number of antenatal glucocorticoid treatments (β = −0.24;p = 0.02) remained significantly associated with IGFBP-1 (R2 = 0.23).

IGFBP-3.

The IGFBP-3 concentration was 1110 ± 42 μg/L (Table 2). A positive correlation with gestational age (Fig. 4), birth weight SDS (Fig. 4), and the number of antenatal glucocorticoid treatments (r = 0.33;p = 0.004;Fig. 2) were noted, and there was a negative correlation with placenta/infant weight-ratio (r = −0.40;p = 0.002). No significant association was found with any other clinical data. Multiple regression analysis indicated birth weight SDS (β = 0.48;p < 0.0001), gestational age (β = 0.33;p = 0.002), and the number of antenatal glucocorticoid treatments (r = 0.30;p = 0.004) as determinants that were significantly associated with IGFBP-3 (R2 = 0.49).

Collagen Turnover and the IGF System

Positive correlations were found between PINP and IGFBP-3, ICTP and IGFBP-3, PINP and IGF-I, and ICTP and IGF-I (Fig. 6). Negative correlations were found between IGFBP-1 and PINP (r = −0.31;p = 0.01), and IGFBP-1 and ICTP (r = −0.30;p = 0.02). To study whether members of the IGF family may have been involved in the regulation of type I collagen turnover, we performed two separate stepwise multiple regression analyses with either PINP or ICTP as dependent determinants; and IGF-I, IGFBP-1, IGFBP-3, and all clinical determinants with significant univariate correlation as independent determinants. Determinants that remained significantly associated with PINP (R2 = 0.47) were gestational age (β = −0.66;p < 0.0001) and IGF-I (β = 0.52;p = 0.0001). Determinants that remained significantly associated with ICTP (R2 = 0.54) were IGF-I (β = 0.61;p < 0.0001), gestational age (β = −0.58;p < 0.0001), and the number of antenatal glucocorticoid treatments (β = −0.30;p = 0.02). Moreover, an inverse correlation appeared between the ICTP/PINP ratio and IGF-I (r = −0.32;p = 0.02), and this correlation was shown by multiple regression analysis to be independent of gestational age (β = −0.41;p = 0.002). No correlation existed between PIIINP and either IGF-I or IGF-II or IGFBP-1 or IGFBP-3.

Fig 6
figure 6

Correlation between cord blood IGF-I and PINP and ICTP. Correlation coefficients and p values are indicated. In multiple regression analysis, 47% of the variation in PINP was explained by gestational age and IGF-I (R2 = 0.47), and 54% of the variation in ICTP was explained by IGF-I, the number of maternal antenatal glucocorticoid treatments, and gestational age (R2 = 0.54).

Full size image

DISCUSSION

We studied whether markers of type I and III collagen turnover in cord blood of premature infants are indicative of fetal growth and gestational age and whether they are related to changes in the fetal circulating IGF system during intrauterine growth. In addition, the study design enabled us to provide data on fetal collagen turnover and the IGF system in preeclampsia and after exposure to maternal antenatal glucocorticoids.

We found that cord plasma concentrations of PINP and ICTP were related to fetal growth, the levels being lower in infants smaller for their gestational age. We also demonstrated a trend toward a decrease in fetal PINP, ICTP, and PIIINP concentrations between 24 and 32 wk of gestation. As hypothesized, fetal IGF-I is related to type I collagen turnover, reflected by concentrations of PINP and ICTP. In contrast, no such relationship emerged for type III collagen.

Our results show considerable turnover of type I collagen during fetal growth in late pregnancy. Disruptions of this turnover may take place in intrauterine growth restriction and impaired bone modeling in preterm infants, as recently suggested by another study (31). Reduced bone mineral content, resulting both from impaired bone modeling and dietary deficiency of minerals (38), is a well-recognized feature of intrauterine growth restriction and prematurity (39). In term infants, however, no difference in cord serum PICP and ICTP has been observed between SGA and AGA groups (7, 8). This difference may be related to different causes of growth retardation in preterm and full-term infants. In keeping with previous studies (6, 8, 31, 38), our results suggest that both type I collagen synthesis and degradation decrease with increasing gestational age, and the decrease in synthesis is even more rapid. It is of interest that type I collagen is also synthesized in the placenta (40), suggesting that placental collagen turnover may contribute to circulating PINP and ICTP. However, we found no association between PINP or ICTP and relative placental weight.

During childhood, one of the best single biochemical indicators of growth velocity is PIIINP (11, 41). During the second half of pregnancy, fetal PIIINP concentration mirrors the fetal growth velocity curve (9) and is lower in SGA than in AGA infants (10). Our data confirmed the inverse relationship between PIIINP and gestational age but, unexpectedly, we found no correlation between PIIINP and birth weight SDS. This may be because of differences in patient selection as the mean gestational age in our study was lower than in a previous study (10). Another reason may be that we used an intact PIIINP assay, which is more specific for the synthesis of type III procollagen than were the assays applied in previous studies (3). Therefore, in the preterm infants <32 wk of gestation, intact PIIINP may rather reflect developmental maturity, as indicated by gestational age.

In keeping with a previous study (9), the PIIINP concentration was found to be higher in male than in female infants. This finding could be explained by a possible sex-related difference in the body composition of the fetus, although in childhood the difference in body composition between boys and girls (42) is not accompanied by a similar difference in PIIINP concentrations (11).

The beneficial effect of maternal antenatal glucocorticoid treatment is well documented in reducing the neonatal morbidity and mortality of infants born before 32 wk of gestation (43). However, the use of multiple antenatal glucocorticoid treatments has raised concern about both short- and long-term side effects (44). In children and adults, markers of type I and III collagen turnover decrease during glucocorticoid treatment (41, 45). Our finding of a decreased ICTP after antenatal glucocorticoid treatment underlines the concerns of risks of repeated antenatal glucocorticoid treatments in respect of the possible effects on fetal collagen metabolism that may result in, for example, deviant bone modeling.

A novel finding of this study is an altered IGF and IGFBP profile associated with exposure to antenatal glucocorticoid treatment, especially with multiple treatments. The increase in IGF-I and IGFBP-3, as well as the decrease in IGFBP-1, may seem to be at variance with the role of glucocorticoids in reducing growth and increasing IGFBP-1 synthesis (46). Accordingly, healthy men receiving glucocorticoids show a similar pattern of changes, which is, however, associated with decreased IGF bioactivity (47). A point that makes our results especially intriguing is that glucocorticoids and the IGF system have been proposed to be key mediators of the fetal programming of adult cardiovascular disease (48). Further studies are required to elucidate whether the altered IGF and IGFBP profile indeed predicts long-term metabolic effects, but this finding emphasizes the importance of careful follow-up of children exposed to multiple antenatal glucocorticoid treatment in utero.

One of the major causes of intrauterine growth restriction is preeclampsia, characterized by chronic fetal hypoxia and inadequate nutrient supply (49). We found that even after adjustment for fetal size, infants of preeclamptic mothers showed lower cord plasma IGF-I and, in keeping with a previous study (22), higher cord plasma IGFBP-1. Chronic, rather then acute, fetal hypoxia has been shown to increase IGFBP-1 gene expression (46). Interestingly, in severe preeclampsia, IGFBP-1 at the fetomaternal interface is markedly elevated, supporting the hypothesis that elevated decidual levels of IGFBP-1 may control placental invasion by restricting the effects of placental IGFs (20, 50). Given this background, it is possible that the elevated fetal circulating IGFBP-1 levels reflect hypoxic regulation of IGFBP-1, a mechanism operating in the human fetus to restrict IGF-mediated intrauterine growth when nutritional conditions are inadequate. However, in infants with a low cord artery base excess, we found lower IGF-I but no difference in IGFBP-1. Low base excess nevertheless reflects fetal hypoxia during a few hours before birth, and the duration of hypoxia required to induce increased synthesis of IGFBP-1 may be longer (46). The origin of the fetal IGFBP-1 remains to be identified, the most likely source being the fetal liver (46).

The negative correlation between cord blood IGF-II and birth weight SDS was unexpected, because there are reports to the opposite (12, 16) or of no association at all (13, 14). However, our study includes the largest number of small premature infants so far reported using the same variables. IGF-II-encoding mRNA is abundant in the fetal adrenals and is stimulated by ACTH in fetal adrenal cortical cells (51). Because cord blood ACTH is higher in SGA infants (52), a link may exist between the IGF system and the fetomaternal-placental corticotropin-ACTH-adrenal cortex axis, which is activated also in preeclampsia (53). Given this background, our findings that IGF-II was lower in infants exposed to antenatal glucocorticoids and was not associated with preeclampsia show that, if it exists, the association between the IGF system and ACTH is very complex.

We found that preterm infants with higher IGF-I concentrations also had abundant type I collagen turnover, as indicated by elevated PINP and ICTP concentrations. Furthermore, these infants showed an increased rate of collagen synthesis compared with its degradation (low ICTP/PINP ratio). These results are in accordance with in vitro studies demonstrating the role of IGF-I in regulating type I collagen turnover and a recent study showing a similar relationship between IGF-I and PICP in preterm infants (31).

We thus conclude that in preterm infants, cord blood PINP and ICTP reflect both fetal growth pattern and maturity, and that fetal IGF-I may regulate fetal growth in part by influencing type I collagen turnover. Therefore, PINP and ICTP deserve being evaluated as potential indicators of postnatal growth velocity in small preterm infants. Intact PIIINP in these infants seems to be dependent on fetal maturity alone.