Early fetal growth, PAPP-A and free β-hCG in relation to risk of delivering a small-for-gestational age infant

Authors

  • I. Kirkegaard,

    Corresponding author
    1. Department of Obstetrics and Gynecology, Aarhus University Hospital, Aarhus, Denmark
    2. Perinatal Epidemiology Research Unit, Aarhus University Hospital, Aarhus, Denmark
    • Department of Obstetrics and Gynecology, Aarhus University Hospital, Skejby, DK- 8200 Aarhus N, Denmark
    Search for more papers by this author
  • T. B. Henriksen,

    1. Department of Obstetrics and Gynecology, Aarhus University Hospital, Aarhus, Denmark
    2. Department of Pediatrics, Aarhus University Hospital, Aarhus, Denmark
    Search for more papers by this author
  • N. Uldbjerg

    1. Perinatal Epidemiology Research Unit, Aarhus University Hospital, Aarhus, Denmark
    Search for more papers by this author

Abstract

Objectives

To examine early fetal growth, pregnancy-associated plasma protein-A (PAPP-A) and free β-human chorionic gonadotropin (β-hCG) in relation to the risk of delivering a small-for-gestational age (SGA) infant.

Methods

Included in the study were 9450 singleton pregnant women who attended the prenatal screening program at Aarhus University Hospital, Denmark, between January 2005 and December 2007. Maternal serum levels of PAPP-A and free β-hCG were measured between gestational weeks 8 and 13. Two ultrasound examinations were performed, the first at 11–13 weeks and the second at 18–22 weeks, from which gestational age was estimated based on crown–rump length and biparietal diameter, respectively. Early fetal growth was expressed as an index: the ratio between the estimated number of days from the first to the second scan and the actual calendar time elapsed in days. SGA was defined as birth weight < 5th centile for gestational age, and the risk of SGA was evaluated according to different cut-offs of the early fetal growth index and the serum markers.

Results

PAPP-A < 0.4 MoM combined with an early fetal growth index < 10th centile resulted in an increased risk of SGA (odds ratio (OR), 5.8; 95% CI, 2.7–12.7). Low PAPP-A, low free β-hCG and slow early fetal growth were statistically, independently associated with SGA, and the association between free β-hCG < 0.3 MoM and SGA was as strong as that between PAPP-A < 0.3 MoM and SGA (OR, 3.1 and 3.0, respectively).

Conclusion

The combination of slow early fetal growth and low PAPP-A resulted in a nearly six-fold increased risk of delivery of an SGA infant. These findings might improve our chances of early identification of fetuses at increased risk of growth restriction. Copyright © 2011 ISUOG. Published by John Wiley & Sons, Ltd.

Introduction

Impaired growth of the fetus is a serious complication in pregnancy, being a major determinant of perinatal morbidity and mortality1. Early recognition of fetuses at increased risk of being growth-restricted enables more appropriate surveillance, thereby optimizing management, which has been shown to reduce the risk of adverse fetal outcome2.

Low first-trimester maternal serum levels of pregnancy-associated plasma protein-A (PAPP-A) has been shown to be associated with an increased risk of delivery of a small-for-gestational age (SGA) infant3–5. Free β-human chorionic gonadotropin (β-hCG) has also been suggested to be associated with SGA3, 4, as have smaller than expected biometries in the first and second trimesters6–10. Furthermore, it has been shown that pregnancies with early fetal growth (increase in biparietal diameter (BPD) between 11–14-week scan and 17–21-week scan) below the slowest 2.5% of the fetal population are associated with a 2.5-fold increased risk of SGA and a nearly five-fold increased risk of perinatal death11. Another study of some 3000 pregnancies showed that smaller-than-expected crown–rump length (CRL) and low PAPP-A were statistically independent predictors of birth weight < 10th centile for gestational age12. If this statistical independence between serology and biometry is confirmed, a combined test including both PAPP-A and early fetal growth for the identification of pregnancies at risk for SGA could be evaluated. Thus, the purpose of our study was to investigate the combination of measures of early fetal growth, PAPP-A and free β-hCG in relation to SGA.

Methods

We studied a cohort of all women with singleton pregnancies who attended the prenatal screening program at Aarhus University Hospital, Denmark, between January 2005 and December 2007. The prenatal program is free of charge and includes a combined biochemistry and ultrasound first-trimester screening test for chromosomal abnormalities and a second-trimester anomaly scan. In the present analyses only women with complete information on the ultrasound measures of fetal growth at gestational weeks 12 and 20, biomarkers PAPP-A and free β-hCG, and birth weight at delivery were included. Pregnancies identified with abnormal karyotype through prenatal diagnosis (n = 17) or major abnormalities at the second-trimester anomaly scan (n = 62) were excluded from the study. The final study population consisted of 9450 pregnancies.

All pregnant women who accepted the first-trimester screening test had blood drawn at the first visit with their general practitioner between gestational weeks 8 + 0 and 13 + 613. The serum samples were analyzed at one central laboratory and levels of PAPP-A and free β-hCG were determined by the Brahm's Kryptor method and registered in the hospital's electronic database of biochemical test results by the unique patient identifier. The PAPP-A and free β-hCG serum values were converted to multiples of the median (MoM) values by expressing the absolute concentration relative to the median value for the gestational age at the day of blood sampling. MoM values were corrected for maternal weight (as a continuous variable). During analysis, PAPP-A and free β-hCG MoM values were included both as continuous variables and as dichotomous variables with cut-off levels at 0.3, 0.4 and 0.5 MoM.

An ultrasound examination was performed in gestational weeks 11 + 2 to 13 + 6, when the nuchal translucency thickness (NT) was measured, and again in weeks 18–22, when the fetus was examined for structural abnormalities; specific measures were recorded in the clinical database. The scans were all performed by sonographers certified to perform NT measurement by The Fetal Medical Foundation, London, UK. In gestational weeks 11–13 the fetal size was measured by CRL, and in gestational weeks 18–22 it was measured by BPD, obtained in the axial plane at the level of the thalami, the third ventricle and the cavum septi pellucidi, from the outer border to the inner border of the skull. The gestational age of the fetus was estimated by means of the CRL in the first trimester (GA12) using the formula of Robinson and Fleming14 and by the BPD in the second trimester (GA20) using the formula of Altman and Chitty15. The calendar time in days between the two ultrasound examinations was noted (Dayscalendar) and the fetal growth between the first and second trimesters was calculated as the ratio between the estimated number of days from the first to the second scan and the actual calendar time elapsed in days (GA20 − GA12)/Dayscalendar. In this way we achieve the observed versus expected increase in fetal size, expressed as a time interval (days/day), which we designated the ‘fetal growth index’. A value of one corresponds to no deviation in growth, a value below one indicates some degree of growth restriction and a value above indicates growth faster than expected. In the analyses this early fetal growth index was included both as a continuous variable and categorized into slow (< 2.5th, 5th and 10th centiles of the fetal population) and fast (> 90th, 95th and 97.5th centiles) fetal growth.

Data regarding all variables derived from the ultrasound measures in the antenatal screening program as well as the biomarkers were obtained from the Astraia Database (www.astraia.com), which is a database developed for clinical purposes such as documentation and individual risk assessment. Only certified sonographers and medical doctors have access to type in data in the Astraia Database.

Information about birth weight at delivery was obtained from the Aarhus Birth Cohort16. This cohort holds information on obstetric and medical complications during pregnancy and delivery, obtained through birth registration forms completed by the attending midwife immediately after delivery and manually checked and compared with the medical charts by a research midwife, before data entry. Expected birth weight for gestational age was estimated by fetal growth curves developed by Marsal et al.17. For the analyses of birth weight according to gestational age as a continuous measure, the z-score was calculated. SGA was defined as birth weight < 5th centile for gestational age and large-for-gestational age (LGA) as birth weight > 95th centile for gestational age.

Information about other factors that may be associated with SGA, such as maternal weight, height, age and parity and maternal lifestyle factors such as smoking during pregnancy and alcohol consumption were also obtained from the Aarhus Birth Cohort16. In addition to the information on obstetric and medical complications during pregnancy and delivery, the cohort contains information on alcohol intake, marital status, educational level and occupational status, obtained through questionnaires completed by pregnant women in the second trimester.

Logarithmic transformation of PAPP-A and free β-hCG was applied to achieve a Gaussian distribution of the data. The data were analyzed by both linear regression (with birth weight adjusted for gestational age as a continuous variable) and logistic regression (dichotomized into SGA vs. not SGA and LGA vs. not LGA). Early fetal growth index, PAPP-A and free β-hCG in relation to low and high birth weight for gestational age are presented as odds ratios (OR) with 95% CIs. Multiple logistic regression analyses were used to estimate a potential explanation of the association of other variables (maternal age, parity, maternal body mass index (BMI) or height, maternal smoking and alcohol intake during pregnancy and educational level). Missing values on these variables were assigned to a separate category for each variable. Because the MoM values of PAPP-A and free β-hCG were already corrected for maternal weight, the analyses including PAPP-A and free β-hCG were adjusted for maternal height instead of BMI. The STATA statistical software package, version IC10, was used for all analyses.

Results

Maternal characteristics and lifestyle factors in relation to SGA, slow early fetal growth, low PAPP-A and low free β-hCG are presented in Table 1. The mean gestational age at the time of the NT scan was 12 + 5 (range, 11 + 2 to 13 + 6) weeks; for the second-trimester scan it was 19 + 6 (range, 17 + 0 to 22 + 6) weeks. The median gestational age at the time of serum sampling was 9 + 6 (range, 8 + 0 to 13 + 6) weeks.

Table 1. Maternal characteristics in relation to small-for-gestational age (SGA) infant, low early fetal growth, low pregnancy-associated plasma protein-A (PAPP-A) and low free beta-human chorionic gonadotropin (β-hCG) in 9450 pregnancies
CharacteristicnSGA < 5th centile (n (%))Early fetal growth index < 5th centile (n (%))PAPP-A < 0.4 MoM (n (%))Free β-hCG < 0.4 MoM (n (%))
  • Early fetal growth index defined as ratio between estimated number of days from first to second scan and actual calendar time elapsed in days.

  • *

    Two-sided P-value < 0.05. MoM, multiples of the median.

Maternal age     
 < 35 years7697389 (5.1)363 (4.7)332 (4.3)223 (2.9)
 ≥ 35 years175383 (4.7)107 (6.1)*85 (4.9)40 (2.3)
 Unknown0    
Parity     
 Primiparous4414303 (6.9)132 (3.0)193 (4.4)123 (2.8)
 Multiparous5035169 (3.4)*338 (6.7)*224 (4.5)140 (2.8)
 Unknown1    
Maternal height (cm)     
 < 1601173109 (9.3)62 (5.3)45 (3.8)44 (3.8)
 160–1705006255 (5.1)267 (5.3)214 (4.3)138 (2.8)
 > 1703231107 (3.3)*138 (4.3)155 (4.8)80 (2.5)
 Unknown40    
Maternal weight (kg)     
 < 603309220 (6.7)158 (4.8)139 (4.2)81 (2.5)
 60–803396152 (4.5)182 (5.4)152 (4.5)93 (2.7)
 > 802736100 (3.7)*129 (4.7)125 (4.6)89 (3.3)
 Unknown9    
Smoking during pregnancy (cigarettes/day)     
 06687304 (4.6)315 (4.7)295 (4.4)183 (2.7)
 ≥ 148250 (10.4)*33 (6.9)*26 (5.4)14 (2.9)
 Unknown2281    
Alcohol during pregnancy (units/week)     
 < 16708320 (4.8)330 (4.9)291 (4.3)187 (2.8)
 ≥ 132517 (5.2)12 (3.7)17 (5.2)6 (1.9)
 Unknown2417    
Educational level     
 < 12 years99463 (6.3)53 (5.3)50 (5.0)34 (3.4)
 ≥ 12 years5851264 (4.5)*276 (4.7)246 (4.2)151 (2.6)
 Unknown2605    
Neonatal gender     
 Female4684298 (6.4)329 (7.0)189 (4.0)87 (1.9)
 Male4766174 (3.7)*141 (3.0)*228 (4.8)176 (3.7)*
 Unknown0    

The early fetal growth index ranged from 0.66 to 1.42 days/day among all fetuses, and the mean (SD) index for all fetuses was 1.04 (SD, 0.09) days/day. The early fetal growth index was significantly lower for SGA than for non-SGA newborns (mean, 1.01 (SD, 0.09) days/day; P < 0.001, unpaired t-test); conversely, the index was significantly higher for LGA newborns (mean, 1.07 (SD, 0.10) days/day; P < 0.001; unpaired t-test). The early fetal growth index was significantly associated with the z-score corresponding to birth weight for gestational age (P < 0.001, linear regression), indicating that the lower the early fetal growth index, the lower the birth weight for gestational age at delivery.

The associations between early fetal growth index below the 2.5th, 5th and 10th centiles and SGA are shown in Table 2. There was a dose–response-like relationship between the three cut-offs of the early fetal growth index and SGA, with the highest associations observed for an early fetal growth index < 2.5th centile (adjusted OR, 1.8; 95% CI, 1.1–2.9). Stratifying for fetal gender revealed a more pronounced association for male fetuses compared with female fetuses (results not shown). There was a strong dose–response-like relationship between fast early fetal growth and LGA (Table 3): for an early fetal growth index ≥ 97.5th centile, the adjusted OR was 3.5 (95% CI, 2.3–5.1).

Table 2. Low early fetal growth index in relation to small-for-gestational age (SGA) infant in 9450 pregnancies
Early fetal growth indexNn (%)ORORadjusted* (95% CI)P
  • Early fetal growth index defined as ratio between estimated number of days from first to second scan and actual calendar time elapsed in days. SGA defined as birth weight<5th centile for gestational age. N is the number of fetuses with a particular early fetal growth index and n is the number of those that were SGA at delivery.

  • *

    Adjusted for maternal age, body mass index, parity, smoking and alcohol intake during pregnancy and educational level. OR, odds ratio.

<2.5th centile23619 (8.1)1.691.75 (1.07–2.85)0.025
≥2.5th centile9214453 (4.9)ReferenceReference
<5th centile47031 (6.6)1.371.41 (0.96–2.07)0.081
≥5th centile8980441 (4.9)ReferenceReference
<10th centile94060 (6.4)1.341.33 (1.00–1.77)0.052
≥10th centile8510412 (4.8)ReferenceReference
Table 3. High early fetal growth index in relation to large-for-gestational age (LGA) infant in 9450 pregnancies
Early fetal growth indexNn (%)ORORadjusted* (95% CI)P
  • Early fetal growth index defined as ratio between estimated number of days from first to second scan and actual calendar time elapsed in days. LGA defined as birth weight>95th centile for gestational age. N is the number of fetuses with a particular early fetal growth index and n is the number of those that were LGA at delivery.

  • *

    Adjusted for maternal age, body mass index, parity, smoking and alcohol intake during pregnancy and educational level. OR, odds ratio.

≥97.5th centile23734 (14.4)3.333.45 (2.34–5.10)<0.001
<97.5th centile9213441 (4.8)ReferenceReference
≥95th centile47559 (12.4)2.923.01 (2.23–4.07)<0.001
<95th centile8975416 (4.6)ReferenceReference
≥90th centile94695 (10.0)2.392.49 (1.95–3.18)<0.001
<90th centile8504380 (4.5)ReferenceReference

Both PAPP-A and free β-hCG were significantly associated with birth weight adjusted for gestational age (both P<0.001, linear regression). The 5th centile of PAPP-A corresponded to a MoM of 0.41, and the 5th centile of free β-hCG corresponded to a MoM of 0.46. The mean log10 PAPP-A MoM was significantly lower in SGA pregnancies compared with non-SGA pregnancies (SGA: mean, − 0.27 (SD, 0.27); non-SGA: mean, 0.03 (SD, 0.25); P < 0.001, unpaired t-test). PAPP-A below each of 0.3, 0.4 and 0.5 MoM was significantly associated with SGA in a dose–response-like manner, with the strongest associations being for PAPP-A < 0.3 MoM (adjusted OR, 3.0 (95% CI, 1.8–5.0)) (Table 4). The mean log10 free β-hCG MoM was also significantly lower in SGA pregnancies compared with non-SGA pregnancies (SGA: mean, 0.02 (SD, 0.26); non-SGA: mean, 0.06 (SD, 0.24); P < 0.001, unpaired t-test). The association between low levels of free β-hCG and SGA was as strong as that between PAPP-A and SGA (free β-hCG < 0.3 MoM: adjusted OR, 3.1 (95% CI, 1.6–6.1)) (Table 5). We found no association between high levels of PAPP-A and free β-hCG (cut-offs at 2, 2.5 and 3 MoM) and SGA or LGA (results not shown).

Table 4. Low pregnancy-associated plasma protein-A (PAPP-A) in relation to small-for-gestational age infant (SGA) in 9450 pregnancies
PAPP-ANn (%)ORORadjusted* (95% CI)P
  • SGA defined as birth weight < 5th centile for gestational age. N is the number of fetuses with a particular early fetal growth index and n is the number of those that were SGA at delivery.

  • *

    Adjusted for maternal age, height, parity, smoking and alcohol intake during pregnancy and educational level. MoM, multiples of the median; OR, odds ratio.

< 0.3 MoM14319 (13.3)2.993.02 (1.82–4.99)< 0.001
≥ 0.3 MoM9307453 (4.9)ReferenceReference
< 0.4 MoM41744 (10.6)2.372.51 (1.80–3.52)< 0.001
≥ 0.4 MoM9033428 (4.7)ReferenceReference
< 0.5 MoM87766 (7.5)1.641.75 (1.33–2.30)< 0.001
≥ 0.5 MoM8573406 (4.7)ReferenceReference
Table 5. Low free beta-human chorionic gonadotropin (β-hCG) in relation to small-for-gestational age infant (SGA) in 9450 pregnancies
Free β-hCGNn (%)ORORadjusted* (95% CI)P
  • SGA defined as birth weight < 5th centile for gestational age. N is the number of fetuses with a particular early fetal growth index and n is the number of those that were SGA at delivery.

  • *

    Adjusted for maternal age, height, parity, smoking and alcohol intake during pregnancy and educational level. MoM, multiples of the median; OR, odds ratio.

< 0.3 MoM7510 (13.3)2.973.08 (1.55–6.12)0.001
≥ 0.3 MoM9375462 (4.9)ReferenceReference
< 0.4 MoM26322 (8.4)1.771.91 (1.21–3.01)0.006
≥ 0.4 MoM9187450 (4.9)ReferenceReference
< 0.5 MoM61647 (7.6)1.631.81 (1.31–2.49)< 0.001
≥ 0.5 MoM8834425 (4.8)ReferenceReference

Pearson's correlation coefficient (r) was calculated to evaluate if there was any association between the continuous variables of the three markers. The correlation coefficients were: 0.2715 (P < 0.001) for PAPP-A and free β-hCG, 0.0319 (P = 0.003) for PAPP-A and early fetal growth index and − 0.0701 (P < 0.001) for free β-hCG and early fetal growth index; the fact that all three coefficients were low indicates very weak correlation between the markers. When PAPP-A, free β-hCG and early fetal growth index were included in the logistic regression model, the association of each of these three variables with SGA remained unchanged as compared with the three models including only one variable at a time. Furthermore, we included interaction terms between the variables in the logistic regression model and none of these interaction terms was statistically significant (results not shown). Thus, each of the three variables was independently associated statistically with SGA. Various combinations of low PAPP-A and low early fetal growth index in association with SGA are shown in Table 6. We found that PAPP-A < 0.4 MoM combined with an early fetal growth index < 10th centile resulted in an increased association with SGA, with an adjusted OR of 5.8 (95% CI, 2.7–12.7). Both low PAPP-A in combination with low free β-hCG and low early fetal growth index in combination with low free β-hCG resulted in slightly lower associations with SGA than did low PAPP-A in combination with low early fetal growth index (results not shown). Combinations of lower cut-offs were also analyzed, but too few numbers in the different combinations resulted in very large confidence intervals as did the combination of all three markers.

Table 6. Combinations of low pregnancy-associated plasma protein-A (PAPP-A) and low early fetal growth index in relation to small-for-gestational age infant (SGA) in 9450 pregnancies
PAPP-A/early fetal growth index combinationNn (%)ORORadjusted* (95% CI)P
  • Early fetal growth index defined as ratio between estimated number of days from first to second scan and actual calendar time elapsed in days. SGA defined as birth weight < 5th centile for gestational age. N is the number of fetuses with a particular early fetal growth index and n is the number of those that were SGA at delivery.

  • *

    Adjusted for maternal age, height, parity, smoking and alcohol intake during pregnancy and educational level. MoM, multiples of the median; OR, odds ratio.

PAPP-A < 0.4 MoM and early fetal growth index < 10th centile449 (20.5)5.295.82 (2.67–12.70)< 0.001
PAPP-A < 0.4 MoM and early fetal growth index ≥ 10th centile37335 (9.4)2.132.25 (1.55–3.26)< 0.001
PAPP-A ≥ 0.4 MoM and early fetal growth index < 10th centile89651 (5.7)1.241.22 (0.90–1.67)0.20
PAPP-A ≥ 0.4 MoM and early fetal growth index ≥ 10th centile8137372 (4.6)ReferenceReference

Discussion

In this cohort study of 9450 pregnancies, we found a nearly six-fold increased risk of delivery of an SGA infant when low PAPP-A was present together with slow early fetal growth. Furthermore, we found slow early fetal growth, low PAPP-A and low free β-hCG to be statistically, independently associated with SGA and fast early fetal growth to be highly associated with LGA.

This six-fold increased risk of SGA is highly relevant clinically. It is in accordance with results of Fox et al.10, who studied 239 pregnancies with low PAPP-A (< 5th centile) and found that, compared with normal second-trimester growth, second-trimester growth restriction (measured at 18–24 weeks as either (a) estimated fetal weight < 25th centile, (b) sonographic gestational age > 7 days smaller than the gestational age established by first-trimester CRL or last menstrual period or (c) head circumference (HC):abdominal circumference ratio > 90th centile for gestational age) was associated with significantly higher rates of birth weight < 10th centile. Compared with the study of Fox et al., our population was larger (470 pregnancies with PAPP-A < 5th centile) and unselected; Fox et al. studied only high-risk pregnancies with low PAPP-A. Furthermore, our definition of an early fetal growth index was based only on ultrasound measures, whereas the definition of Fox et al. included a comparison with gestational age established by either CRL or last menstrual period.

The association between low PAPP-A and SGA has been well established previously in large cohort studies3–5, 18–21, and our results regarding early fetal growth are in accordance with those of Pedersen et al.11, who found early fetal growth (defined as millimeters of growth in BPD per day) < 2.5th centile to be significantly associated with SGA. However, they did not relate the results to PAPP-A and free β-hCG. Our study adds that the association with SGA grows considerably stronger when both low levels of PAPP-A and low early fetal growth index are present. We found the association between low free β-hCG and SGA to be as strong as that between low PAPP-A and SGA. This finding is in contrast to that of previous publications3–5, 18, 19, three of which failed to find any association with β-hCG5, 18, 19, while two found only a minor association3, 4. A possible explanation for this discrepancy could be that we determined free β-hCG at a lower gestational age (average, 9 + 6 weeks). This timing may influence the association with SGA since the level of free β-hCG increases until 10 weeks of gestation and then declines, reaching a constant level at around 20 gestational weeks22. Our results indicate that measuring free β-hCG earlier could result in a more pronounced association between low free β-hCG and SGA than might otherwise be found. Unexpectedly, we found that low free β-hCG correlated negatively with early fetal growth index. We have no obvious biological explanation for this finding, although the correlation was very weak. Further investigation is required before any firm conclusion can be reached.

Our prenatal screening program covers the entire population in the area and is free of charge; the first- and second-trimester scan attendance rates in 2007 were 96% and 99%, respectively. Accordingly, our study population is highly representative, without oversampling of high-risk pregnancies. All data, both in the Astraia database and in the Aarhus Birth Cohort, were collected prospectively without knowledge of the study in the clinical settings, and any misclassification may therefore be non-differential. The data from the Aarhus Birth Cohort made it possible for us to adjust for a wide range of potential explanatory variables, which further strengthens our study. We estimated an early fetal growth index between two separate ultrasound scans; with this method we obtained a more accurate measure as compared with a single measurement evaluated in relation to menstrual history or a growth chart.

A limitation of our study is that fetuses with growth restriction prior to the first scan, but with subsequently normal growth, would have been missed. However, growth restriction prior to the 12th gestational week is rare and was beyond the focus of this study. Potential bias could have been caused by errors in the estimation of gestational age by CRL at weeks 11–14. If CRL is overestimated it will result in a falsely low PAPP-A MoM, a reduced early fetal growth index and a lower birth weight for gestational age. This may explain some of the associations we observed. However, we have no reason to believe that CRL measurements are systematically overestimated and random measurement errors would produce unbiased results. Another cause of statistically unbiased imprecision in our results was the use of BPD instead of the HC, as HC is considered as a more accurate measure of fetal size than is BPD15, 23.

There seems to be increasing evidence that fetal growth restriction might be established in early pregnancy24 and that increased surveillance of high-risk fetuses reduces the risk of adverse fetal outcome for SGA fetuses that are detected, in comparison with unidentified SGA fetuses2. It might prove more cost-effective to implement methods for the early identification of fetuses at high risk of fetal growth restriction than to offer third-trimester ultrasound scans with fetal weight estimation to the entire population. Since PAPP-A, free β-hCG and measures of early fetal growth are already assessed in pregnant women participating in prenatal screening programs, the use of these measures in early identification would be at no extra cost. Early risk assessment might be based on PAPP-A, free β-hCG and early fetal growth index, perhaps combined with other serum markers, uterine artery Doppler flow assessment and possibly even three-dimensional ultrasound measures of placental volume25–28. Prospective studies are needed to test such early prediction in clinical settings. Furthermore, prediction might be better if all biomarker levels are used instead of only the extreme values, but as we obtained fairly low ORs using the extreme values, we found it inappropriate to attempt to develop an algorithm for risk assessment using all levels of the markers.

In conclusion, we found that low PAPP-A and low free β-hCG as well as slow early fetal growth are independently associated with SGA. A combination of low PAPP-A and slow early fetal growth increased the risk further—by almost six-fold. Using this combination in screening might improve our chances of early identification of fetuses at increased risk for growth restriction, although the clinical utility of these findings has yet to be investigated.

Ancillary