Sex- and growth-specific characteristics of small for gestational age infants: a prospective cohort study

Background Asymmetric fetal growth and male sex are both associated with adverse neonatal outcome. However, less is known about the influence of asymmetric growth and fetal sex within SGA neonates, a group of infants already at increased risk for adverse neonatal outcomes. The aim of the present study was to provide insight into variance in risk factors for SGA in a fetal sex- and growth symmetry-specific way. Methods For this prospective, multicenter cohort study, data from the Screening for Pregnancy Endpoints (SCOPE) study were used with 5628 nulliparous participants, of which 633 (11.3%) pregnancies were complicated with SGA and 3376 (60.0%) women had uncomplicated pregnancies. Association between risk factors for SGA, SGA subgroups, and uncomplicated pregnancies were assessed with multivariable analyses. Results Prevalence of asymmetric growth varied from 45.8% of SGA infants to 5.5% of infants with a customized birthweight > 90th percentile (p < 0.001). Significantly more SGA males had asymmetric growth compared to SGA female infants (51.2% vs 40.4%, p = 0.009). Maternal pre-pregnancy diet and BMI < 20 and ≥ 30 were significantly associated with symmetric SGA but not with asymmetric SGA. Asymmetric SGA infants had not only lower customized birthweight percentile (4.4 (SD 2.8) vs 5.0 (SD 3.0), p < 0.001), but also lower rates of stillbirth (p = 0.041) and less often Apgar scores < 7 (p = 0.060). Conclusions Among SGA infants, low customized birthweight percentiles and male sex are associated with asymmetric growth. Only symmetric SGA is significantly associated with maternal risk factors in early pregnancy. There is a substantial variance in risk factors and neonatal outcomes for SGA based on growth symmetry, implying a different pathogenesis. Trial registration ACTRN12607000551493


Background
Small for gestational age (SGA) can be defined as neonates with a birthweight below the < 10th percentile customized for maternal factors such as parity, weight, height, and ethnicity [1][2][3]. SGA is associated with increased rates of stillbirth and neonatal death as well as metabolic disease in later life [1,[4][5][6]. SGA has many different causes and the aetiology of 'being SGA' in this heterogeneous group of infants is not yet understood [7][8][9]. Current risk prediction for SGA, including maternal risk factors, biomarkers, and ultrasound measurements is insufficient to reliably predict SGA and in clinical practice less than half of SGA infants are usually recognized before birth [7,[9][10][11][12].
Fetal growth restriction (FGR) implies the failure of a fetus to achieve its growth potential by showing reduced growth on serial ultrasound evaluation. In The Lancet's Stillbirths Series, Bhutta et al. estimated that improved detection and management of FGR could reduce stillbirth rates by 20% [13]. Early detection of FGR may benefit from closer monitoring and early intervention, although methods of monitoring FGR are improving, current methods are not yet reliable [14]. Although both FGR and SGA are associated with increased rates of stillbirth and adverse perinatal outcome, not all FGR will result in a SGA infant as the birth weight may be restricted but not below the designated customized birthweight percentile [3,14,15]. Among growth restricted fetuses and subsequently neonates, a distinction can be made between infants with a birth length or head circumference that is either proportional (symmetric) versus disproportional (asymmetric) to the infant's weight [16][17][18][19]. Previous studies have shown that asymmetric infants are at increased risk for neonatal death, operative interventions and respiratory distress compared to symmetric infants [4,15,[17][18][19][20].
In addition to the type of growth restriction, fetal sex is also known to influence pregnancy and neonatal outcome. While male-bearing pregnancies are at increased risk for early preterm birth, (term) preeclampsia and acute fetal distress, and also have higher rates of caesarean sections, female infants are more likely to be growth restricted but have fewer complications during and after birth [21][22][23][24]. In light of these observations, Clifton et al. described differences in growth reduction between male and female fetuses in response to an adverse environment in utero [23]. Whereas female fetuses reduce growth during maternal stress, males continue to grow thereby placing themselves at increased risk for stillbirth and neonatal death [23,25].
Asymmetric fetal growth and fetal sex are both known to be associated with neonatal outcomes, less is known about the influence of asymmetric growth and fetal sex within SGA neonates, a group of infants already at increased risk for adverse neonatal outcome [1,[4][5][6]. New insights into these different SGA subgroups could contribute to an improved understanding of its aetiology and inform new methods for more reliable SGA risk prediction. Therefore, the aim of the present study was to provide insight into differences in risk factors for SGA in a fetal sex-and growth symmetry-specific way.

Study protocol
Data from the Screening for Pregnancy Endpoints (SCOPE) study were used. In short, the SCOPE study was a prospective, multicenter cohort study with the main aim to develop screening tests to predict preeclampsia, spontaneous preterm birth and SGA infants. The SCOPE study had recruitment sites in Auckland (New Zealand), Adelaide (Australia), Manchester, Leeds, London (UK), and Cork (Ireland) and recruited participants between 2004 and 2011. Nulliparous women with a singleton pregnancy less than 16 weeks of gestation were eligible for the study. Women with major risk factors for preeclampsia, SGA and spontaneous pre-term birth were excluded from the study (e.g., chronic hypertension requiring antihypertensive drugs, pre-existing diabetes, antiphospholipid syndrome, ≥ 3 abortions or miscarriages, cervical suture, known fetal anomaly). Detailed information about in-and exclusion criteria are described elsewhere [26]. Ethical approval was obtained from the local institutional ethics committees and all participants gave written informed consent.
Participants were interviewed and examined by a research midwife at 15 ± 1 weeks' gestation. This interview included information about demographics, medical history of both participant and family, as well as information about the current pregnancy including: vaginal bleeding, diet, use of supplements and medication, smoking, alcohol and recreational drug use for both the 3 months before and after becoming pregnant. Weight, height and blood pressure were measured. Maternal socio-economic index (SEI) score was estimated [27]. At the appointments at 15 ± 1 weeks' and 20 ± 1 weeks' gestation, participants completed the Edinburgh Postnatal Depression Scale, the Short form State-Trait Anxiety Inventory and Perceived Stress Scale [28][29][30]. Morphology ultrasonography, including uterine and umbilical Doppler flow scans, was performed at 20 weeks' gestation. Each participant and her newborn were seen by a research midwife in the early post-partum periods; neonatal length (centimeters (cm), n = 5289), weight (grams, n = 5609) and head circumference (cm, n = 5464) were measured within 72 h of birth. Neonatal length was measured using the neonatometer (n = 3171, 60.0%) or using tape measures to 0.1 of a centimeter [31]. Head circumference was measured with tape measures to 0.1 of a centimeter. Participants were asked about vaginal bleeding, infections, medication and supplement use during the 3rd trimester. Additional details of late pregnancy and delivery were collected from clinical case notes.
Outcome SGA was defined as birthweight less than the 10th customized birthweight percentile. Customized birthweight percentiles are adjusted for maternal booking weight, height, ethnicity, parity, gestational age, and sex of the infant using the Gestation Related Optimal Weight (GROW) software on www.gestation.net [2]. This software has been studied and found to be reliable in the detection of SGA with an increased risk of adverse perinatal outcome within multi ethnic populations and maternal under-and overweighted populations [32,33]. SGA infants were grouped based on sex (males/females) and growth symmetry (symmetric/asymmetric). Asymmetric growth was defined as a Ponderal index < 10th percentile, corrected for gestational age based on reference values of Roje et al. [19]. Ponderal Index was calculated as (weight (grams) × 100)/(length (cm)) 3 . Pregnancies were classified as uncomplicated in the absence of SGA, spontaneous and iatrogenic preterm birth, stillbirth, preeclampsia, gestational hypertension or gestational diabetes [34,35].

Statistical methods
Univariate analyses were performed for maternal demographics, pregnancy characteristics and neonatal outcome. For continuous variables, mean and median were compared using the Students t test and Mann-Whitney U test. Categorical variables were compared using chisquare test. Overall, less than 2% of the data was missing, 3 variables had > 5% missing data: maternal birthweight (5.2%), mean uterine Doppler resistance index (RI) (6.1%) and Ponderal Index (6.0%). For multivariable analysis, missing data were imputed using multiple imputation [36]. Multivariable analysis was performed using backward stepwise logistic regression to compare pregnancies complicated with SGA to uncomplicated pregnancies. SPSS default values (PIN = 0.05 and OUT = 0.1) were selected for the backward stepwise logistic regression.
Twenty-nine variables that were found to be associated with SGA in prior SCOPE publications by McCowan et al. and Khashan et al. were included in the multivariable analysis [7,8,37]. These variables are reported in Table S1. Following the stepwise procedure, 16 of the 29 variables were significantly associated with SGA and were included in the final model. The same 16 variables were included in the sex-and growth-specific multinominal multivariable analysis, with uncomplicated pregnancies as reference group. Percentages of missing data for each of these variables are shown in supplementary data (Table S2). The reported odds ratios (OR) of the multivariable analysis are pooled effects of the multiple imputation data, and were compared between the SGA subgroups. The threshold for significance was set at p < 0.05. Statistical analyses were performed using SPSS version 24 (SPSS Inc. Chicago, IL, USA).

Results
A total of 5690 participants were enrolled in the SCOPE study of whom 62 (1.1%) participants were lost to follow-up or had a miscarriage or termination before 20 weeks' gestation. Of the remaining 5628 pregnancies, 3376 (60.0%) were uncomplicated and 633 (11.2%) were complicated by SGA. Participant distribution per study site is shown in supplementary data (Table S3). Maternal demographics, pregnancy characteristics and outcome for the whole SCOPE cohort, uncomplicated and SGA pregnancies are presented in Table 1. Compared to uncomplicated pregnancies, women with a SGA pregnancy more frequently had a low birthweight themselves (p < 0.001), BMI < 20 or ≥ 30 (p < 0.001), a lower SEI score (p < 0.001) and higher systolic and diastolic blood pressure (p < 0.001) at 15± weeks'. Women with a SGA pregnancy were less likely to be Caucasian (p = 0.028) and less likely to have a Rhesus negative blood group (p = 0.022) compared with women with uncomplicated pregnancies. At 20 weeks' gestation, women with a SGA pregnancy were more likely to smoke cigarettes and had higher uterine and umbilical Doppler flow RI compared to women with uncomplicated pregnancies (p < 0.001). The prevalence of asymmetric and symmetric growth by customized birthweight deciles for the SCOPE cohort are presented in Fig. 1. Of the SGA infants, 45.8% were asymmetric, compared to 5.5% of the infants with a birthweight > 90th percentile. The prevalence of symmetric and asymmetric growth was significantly different between customized birthweight deciles (p < 0.001). Within the whole SCOPE cohort, 606 males (22.5%) had asymmetric growth compared to 478 (18.4%) females (p < 0.001), these numbers include both SGA and non-SGA infants.
Maternal demographics, pregnancy characteristics, and outcome for SGA by fetal sex and growth symmetry are presented in Table 2. Between male and female SGA infants, there were no significant differences in maternal demographics or clinical risk factors at 15 weeks' gestation. However, maternal SEI was on average lower for women bearing a female SGA infant compared to those bearing a male SGA infant (37  vs 45 , p = 0.054). At 20 Compared to asymmetric SGA, women bearing a symmetric SGA infant were more often Caucasian (p = 0.001), more often had a BMI < 20 or ≥ 30 (p = 0.030) and had lower SEI scores (36  for symmetric and 45  for asymmetric SGA, p = 0.010). There were no significant differences between symmetric and asymmetric SGA infants in mean umbilical and uterine Doppler RI at 20 weeks' gestation. Regarding neonatal outcome, asymmetric SGA infants had a lower customized birthweight percentile compared to symmetric SGA (mean of 4.4 (2.8) and 5.0(3.0) respectively, p = 0.017). Symmetric SGA infants were more often born spontaneously pre-term (< 37 weeks) than asymmetric SGA infants (5.5% vs 2.3%, p = 0.053). Table 3 shows the OR of clinical risk factors with a significant independent association with SGA, compared to uncomplicated pregnancies. Separate analyses were performed for the SGA subgroups of interest. Daily vigorous exercise was significantly associated with both SGA males (4.2 (1.8-10.0)) and SGA females (2.7 (1.1-7.1)). The OR per unit increase for Uterine Doppler RI was higher in SGA males (1.7 (1.5-1.9)) than females (1. 5 (1.3-1.7)). Whereas for Umbilical Doppler RI this was only significantly associated with SGA females (1.6 (1.3-1.9) vs 1.0 (0.8-

1.3)).
In sensitivity analyses, we restricted multivariate testing to unimputed data excluding missing data (Table S4). These showed similar results to multivariate testing with imputed data.

Main findings
The data from this large prospective cohort demonstrate that there is a substantial variance in risk factors and neonatal outcome for SGA based on fetal sex and growth symmetry. Low birthweight percentiles and male sex are associated with higher rates of asymmetric growth.
In the present study, we did not find significant sex-specific differences in pregnancy outcome, regarding stillbirth, low Apgar scores, and preeclampsia. SGA males were generally longer and had a relatively larger head circumference but were not heavier than SGA females. Asymmetric growth was predominantly seen in SGA males, while symmetric growth was more commonly seen in females, implying that growth trajectory, specifically growth symmetry, is sex-specific.
Previous research showed that the predictive value of HC/AC ratio is low and poorly correlated with Ponderal Index and should therefore be rejected as a measurement for asymmetric growth in utero [14,18,38].  This is consistent with the present study, where the rates of infants with a HC/AC ratio > 95th percentile at the time of the 20 weeks' morphology scan were not significantly different between symmetric and asymmetric SGA infants. One might speculate that the fetus demonstrating HC/AC discordance is more easily recognised by ultrasound compared to the symmetrically growing fetus. However, the rate of induction of labour or emergency caesarean section was not different between the two SGA groups. Compared to symmetric SGA, asymmetric SGA infants had lower customized birthweight percentiles, but were longer and had a relatively larger head circumference, suggesting potential brain sparing.  Results are expressed as OR (95%CI) with uncomplicated pregnancies as the referent group. Bold indicate that the OR is significant

Interpretation
Our findings are consistent with the theories reported by Resnik et al. and Clifton et al. that symmetric growth restriction occurs earlier in pregnancy than asymmetric growth restriction and that there are sex-specific strategies by which males and females cope with adverse in utero environments [15,23]. Symmetric growth restriction is hypothesized to be caused by early whole body impairment of fetal growth, for example, by maternal drug use, infection or chromosomal abnormalities [15]. In contrast, asymmetric growth restriction may arise later in gestation, due to inadequate availability of substrates for fetal growth possibly caused by maternal vascular disease and decreased uteroplacental perfusion [15]. In the SCOPE cohort, clinical risk factors for SGA, such as low and high maternal BMI, low SEI, and pre-pregnancy diet, seem to be more strongly associated with symmetric SGA. However, importantly in the present study increased uterine artery Doppler RI and preeclampsia were not more prevalent within asymmetric SGA pregnancies. In contrast with previous findings, asymmetric SGA infants had lower rates of stillbirth, spontaneous preterm birth, and higher Apgar scores compared to symmetric SGA infants [20].
Most studies report no negative consequences of (vigorous) exercise during pregnancy on fetal well-being [39,40]. Clapp et al. reported improved fetoplacental growth in women who begin or maintain exercise in early pregnancy and decrease their exercise in mid and late pregnancy [41,42]. However, McCowan et al. found daily vigorous exercise as a major risk factor for SGA [7]. The present study can add to this that the association between vigorous exercise in early pregnancy and SGA may be stronger for male SGA than female SGA infants.
Zhou et al. reported a gene environment interaction for the maternal angiotensin-converting enzyme (ACE) A11860G gene variant and low SEI or low leafy vegetable intake as a risk factor for SGA in female-bearing pregnancies. ACE gene encodes a potent zinc metalloenzyme involved in renin-angiotensin system (RAS) activity which is also involved in the trophoblast function [43]. Myatt et al. studied trophoblast function in placentas of obese women and reported reduced mitochondrial respiration and adenosine triphosphate (ATP) generation [44]. Findings of both studies suggest compromised placental function. In the present study, female and symmetric SGA were both associated with low leafy vegetable intake and increased maternal BMI. The increased umbilical Doppler RI in these SGA subgroups suggests impaired growth of the placenta over the first 20 weeks'. Maternal BMI, leafy vegetable intake and umbilical Doppler RI were not significantly associated with male and asymmetric SGA, suggesting that these SGA subgroups may have a different pathogenesis.

Strengths and limitations
To our knowledge, the present study is the first to report on growth symmetry and sex differences in SGA infants. The strength of this prospective study is the extensive amount of detailed information before and during pregnancy. The limitation is that, while this is a large prospective cohort study, the number of SGA infants (n = 633) is insufficient to investigate stillbirth and neonatal death rates and thus these findings should be interpreted with caution in a clinical context. Furthermore, the neonatometer was used for 60.0% of the neonates, the remaining 40.0% of the neonates were measured with a tape measure. This may have introduced variability in length measurements. Maternal weight gain was not included in the study design of the SCOPE study; therefore, we are unable to provide any details regarding maternal weight gain. Although the GROW software has been found to be reliable in the detection of SGA with an increased risk of adverse perinatal outcome within multi ethnic populations and maternal under-and overweight populations, the use of customized birthweight centiles to determine SGA infants is not universally accepted [32,33].

Conclusion
Among SGA infants low customized birthweight percentiles and male sex are associated with asymmetric SGA. Poor maternal health in early pregnancy is associated with symmetric SGA, while increased uterine Doppler flow in later pregnancy is associated with both symmetric and asymmetric SGA. Further research regarding the biology of growth symmetry and the value of additional Doppler flow scans as predictors of growth symmetry may aid in a better insight in the pathophysiology of different SGA phenotypes.

Perspectives and significance
This manuscript contributes to an improved understanding of the aetiology of sex-specific strategies by which males and females cope with adverse in utero environments. We demonstrate that there is a substantial variance in risk factors and neonatal outcome for SGA based on fetal sex and growth symmetry. Among SGA infants, low birthweight percentiles and male sex are associated with higher rates of asymmetric growth which has different risk factors compared to symmetric fetal growth, indicating a different pathogenesis.