Repeat Antenatal Betamethasone and Cardiometabolic Outcomes

The ACTORDS (international standard randomized controlled trial number 48656428) was a placebo-controlled, double-blind, randomized trial of repeat antenatal betamethasone treatment conducted at 23 collaborating hospitals across Australia and New Zealand.² Women with a single, twin, or triplet pregnancy were eligible if they were <32 weeks’ gestation, had ongoing risk of preterm birth, and had received a single course of antenatal corticosteroids ≥7 days before trial entry. Exclusion criteria included chorioamnionitis requiring urgent delivery, labor at the second stage, mature fetal lung development, or further corticosteroid therapy that was considered essential.

A total of 982 women (1146 fetuses) were randomly assigned, via a central telephone service, to either the repeat betamethasone (Celestone Chronodose, comprising 7.8 mg betamethasone sodium phosphate and 6 mg betamethasone acetate; Schering-Plough, Sydney, Australia) or the masked saline placebo. Treatment was repeated each week if a woman was judged to be at continued risk of preterm birth, up to 32 weeks’ gestation.

Repeat antenatal betamethasone therapy was associated with neonatal benefit, including clinically important reductions in respiratory distress syndrome (RDS), severe lung disease, and other combined serious morbidity.² At the 2 years’ corrected age, neurodevelopment, growth and general health were similar between groups.⁶ Before the completion of the 2-year follow-up, midchildhood follow-up was planned because of concerns of the potential for long-term adverse effects of fetal corticosteroid exposure.

All surviving children of mothers who had participated in and remained in the ACTORDS were invited to partake in a study of neurocognitive function, growth, and general health, including spirometry (EasyOne 2001; New Diagnostic Designs Technologies, Zurich, Switzerland), at 6 to 8 years’ corrected age.⁷ In addition, children who were assessed in New Zealand were also invited to participate in a more detailed study of cardiometabolic function, as previously described.^8,9 This included the measurement of body composition by whole-body dual energy radiograph absorptiometry (DXA), glucose and insulin function by a frequently sampled intravenous glucose tolerance test with Minimal Modeling (MinimalModel, version 6.02), renal function by estimated creatinine clearance, and 24-hour ambulatory blood pressure (AMBP) monitoring (Spacelabs 90217; Spacelabs Healthcare Co, Issaquah, WA). These assessments were conducted throughout New Zealand by a single team of investigators, blinded to treatment allocation. Children with severe disability or who could not assent to testing were excluded from the cardiometabolic studies. Assessment of body composition was limited to 2 centers (Auckland and Christchurch) so that DXA measurements could be performed on identical instruments (Lunar Prodigy; General Electric Healthcare, Madison, WI). Ethical approval was obtained from the New Zealand Multicentre Ethics Committee. Written informed consent was obtained from caregivers at follow-up, and assent was sought from the children.

In this study, we hypothesized that exposure to repeat antenatal corticosteroids, compared with a single dose, would have an adverse effect on growth, cardiometabolic risk factors, bone mass, and lung function at 6 to 8 years’ corrected age in children with FGR but not children with normal prenatal growth.

Children who underwent 1 or more tests of cardiometabolic function at 6 to 8 years’ corrected age were included in this secondary analysis of data from the New Zealand ACTORDS Midchildhood Outcomes Study. The prespecified outcomes for this analysis were age- and sex-specific z scores for height, weight, and BMI; whole-body fat and lean mass, regional fat mass (android and gynoid), and limb lean mass; whole-body bone mineral content and area; fasting plasma glucose and insulin concentrations, and Minimal Model estimates of insulin sensitivity, acute insulin release, and disposition index; height-specific z scores for daytime and nighttime systolic and diastolic AMBP and percentage diurnal dip; plasma creatinine concentration and estimated glomerular filtration rate; and height- and sex-specific z scores for forced vital capacity (FVC), forced expiratory volume in 1 second (FEV1), FEV1/FVC ratio, and mean forced expiratory flow at 25% to 75% of FVC (FEF25-75).

FGR was defined a priori as 1 or more of the following: obstetric diagnosis of FGR at trial entry, cesarean delivery for FGR, or having a customized birth weight less than the third centile (version 6.7.8.3; Perinatal Institute, Birmingham, United Kingdom). Although this definition includes postrandomization factors, we felt that this was important because antenatal diagnosis of FGR substantially underrepresents the true incidence of FGR in the preterm population.^17,18 Customized birth weight centiles, which are based on fetal growth curves, were used because these have been shown to improve detection of FGR.^19,20 Although infants exposed to repeat doses in the ACTORDS had slightly lower birth weight z scores, meta-analysis has revealed that repeat doses do not increase the risk of being small for gestational age.³ 

Analysis was performed by using SAS (version 9.4; SAS Institute, Inc, Cary, NC). Data are presented as number (percent), mean (SD), or median (interquartile range). Treatment groups were compared for the prespecified outcomes by using generalized linear models with adjustment for sex and gestational age at trial entry and clustering of children from multiple pregnancy by generalized estimating equations. Body composition data were additionally adjusted for current height. The influence of FGR on treatment effect was assessed by an interaction test. Treatment effects within subgroups (FGR and non-FGR) are reported as either mean difference (MD) for normally distributed data or ratio of geometric means (RGM) for positively skewed data, with 95% confidence intervals (CIs).

Of the 323 surviving children eligible for follow-up in New Zealand, 308 were assessed at 6 to 8 years’ corrected age, of whom 266 underwent 1 or more tests of cardiometabolic function (repeat betamethasone 127; placebo 139; Fig 1). Children who did not complete any tests of cardiometabolic function, compared with those who were assessed, had lower birth weight z scores, and their mothers were more likely to be multiparous but were similar for other baseline characteristics (Supplemental Table 5).

Of the 266 children included in this analysis, the rate of FGR was similar between those exposed to repeat betamethasone (34%) and placebo (32%). In the FGR subgroup, those randomly exposed to repeat betamethasone were more likely than those exposed to placebo to have received ≥4 trial treatments (repeat betamethasone 10 of 43 [27.0%]; placebo 1 of 44 [2.9%]; P = .007) and had a longer gestation (repeat betamethasone mean [SD]: 31.8 [2.7] weeks; placebo mean [SD]: 30.6 [2.7] weeks; P = .03). Repeat betamethasone therapy reduced the severity of neonatal lung disease and the need for surfactant therapy (Table 1). In the non-FGR subgroup, there were no differences in baseline characteristics between children randomly exposed to repeat betamethasone or placebo, and, again, repeat betamethasone therapy reduced the incidence of RDS and severity of neonatal lung disease (Table 1).

The FGR subgroup, compared with non-FGR, was characterized by older maternal age (FGR mean [SD]: 32.4 [5.4] years; non-FGR mean [SD]: 30.7 [5.6] years; P = .03), higher maternal BMI (FGR mean [SD]: 28.7 [6.9]; non-FGR mean [SD]: 25.8 [5.7]; P = .004), later gestational age at trial entry (FGR mean [SD]: 28.9 [2.0] weeks; non-FGR mean [SD]: 28.2 [2.3] weeks; P = .02), and increased rates of multiple pregnancy (FGR: 25 of 72 [35.0%]; non-FGR: 28 of 159 [18.0%]; P = .007) and preeclampsia (FGR: 31 of 72 [43.0%]; non-FGR: 18 of 159 [11.0%]; P ≤ .001) (Table 1). The FGR subgroup, compared with non-FGR, had a shorter gestation (FGR mean [SD]: 31.2 [2.7] weeks; non-FGR mean [SD]: 32.1 [3.9] weeks; P = .03), lower birth weight z score (FGR mean [SD]: −1.1 [0.8]; non-FGR mean [SD]: 0.1 [0.8]; P ≤ .001), and an increased rate of serious neonatal morbidity (FGR: 28 of 87 [32.0%]; non-FGR: 36 of 179 [20.0%]; P = .03) (Table 1).

For childhood height, there was a significant interaction between repeat antenatal betamethasone therapy and FGR. In the FGR subgroup, children exposed to repeat betamethasone had larger height z scores than children exposed to placebo; no significant difference was seen between treatment groups in the non-FGR subgroup (FGR height z score MD: 0.59; 95% CI: 0.01 to 1.17; non-FGR MD: −0.29; 95% CI: −0.69 to 0.10; P = .01 for interaction). In post hoc analysis, this interaction remained significant after adjustment for ethnicity and maternal height (Table 2).

For all other cardiometabolic outcomes, treatment groups were similar, both in the FGR and non-FGR subgroups, with no evidence of an interaction effect. This included fat mass (whole body, android, gynoid), lean mass (whole-body, limb), bone mass (whole body bone mineral content and bone area) (Table 2); fasting glucose and insulin concentrations, and Minimal Model estimates for glucose and insulin kinetics (insulin sensitivity, acute insulin release, and disposition index) (Table 3); plasma creatinine concentration, estimated glomerular filtration rate, and AMBP (daytime systolic and diastolic and nighttime systolic and diastolic z scores and diurnal systolic and diastolic percentage dip) (Table 3); and lung function (FEV1, FEV1/FVC, and FEF25-75 z scores) (Table 4).

In post hoc analysis, additional adjustment for maternal smoking during pregnancy and the number of trial treatments administered did not alter results.

In this secondary analysis of data from the New Zealand ACTORDS Midchildhood Outcomes Study, we found that among children with FGR, exposure to repeat antenatal betamethasone was associated with increased height at midchildhood by ∼0.5 SD, an effect not seen in children without FGR. Exposure to repeat antenatal betamethasone did not alter any other measures of cardiometabolic function in those with and without FGR.

FGR is associated with reduced height in childhood and adulthood, both in preterm- and term-born subjects.²¹ This was particularly evident in the placebo group, in which children with FGR were ˃1 SD shorter in midchildhood than children without FGR. Contrary to our hypothesis, repeat betamethasone treatment appeared to have a positive effect on midchildhood height in children with FGR. Although we cannot exclude the possibility of a type 1 error, authors of previous studies have suggested that antenatal corticosteroid treatment may influence skeletal growth.^22,23 In the Auckland Steroid Trial, a single course of antenatal betamethasone compared with placebo was associated with a greater proportion of adult stature in the lower body segment,²⁴ although, in this trial and in a similar observational study of moderate-to-late preterm birth, antenatal corticosteroid exposure was not associated with altered final height.²⁵ Nevertheless, in a cohort study of adolescents with very low birth weight, adolescents exposed to antenatal corticosteroids were ∼0.3 SD taller, even after adjustment for confounding factors.²⁶ 

Delayed onset of puberty, leading to a prolonged period of accelerated appendicular growth, has been proposed as 1 possible explanation for these changes,⁸ as suggested in 1 trial.²⁷ However, with our study, we aimed to assess children before pubarche.⁹ Premature adrenarche, which is more common in children born small for gestation age, could lead to an increase in linear growth velocity, but all children were carefully examined, and none had confirmatory signs, apart from several who had also commenced puberty (Tanner stage 1).⁹ Childhood obesity may also accelerate linear growth,²⁸ but, in this cohort, FGR was associated with lower rather than increased BMI and fat mass.

There is ongoing debate about the long-term health effects of catch-up growth in preterm infants. Infants with growth restriction who do not exhibit catch-up growth during the neonatal period are more likely to have permanent short stature²¹ and subnormal cognitive function in adulthood.²⁹ Conversely, human and animal studies have revealed an association between rapid catch-up growth during infancy and increased risk of cardiometabolic disease in later life.^30,–34 Infants with FGR are predisposed to insulin resistance and central obesity, and these changes may be exacerbated by rapid catch-up growth, possibly because of altered epigenetic profiles.^35,36 However, in this study, FGR did not appear to be associated with increased adiposity or insulin resistance, either in children exposed to repeat betamethasone or those exposed to placebo. In addition, in the FGR subgroup, repeat antenatal betamethasone therapy was not associated with any adverse effects on cardiometabolic function. Thus, it seems unlikely that the catch-up in height seen with repeat betamethasone in the FGR subgroup, if true, poses any risk to cardiometabolic health. We do not know the mechanism underlying this catch-up growth, although authors of an ovine study found that repeat doses of betamethasone reduced postnatal concentrations of insulin-like growth factor binding protein 3, which may potentiate the effects of insulin-like growth factor 1 on linear growth.³⁷ Investigation of the effect of repeat antenatal corticosteroids on linear growth in meta-analysis of individual patient data, including assessment of dose-response effects, may be used to help quantify this association and interaction with FGR and elucidate possible mechanisms.³⁸ 

A challenging issue for this study was the definition of FGR, which is conceptually straightforward but much more difficult to identify in practice. Although subgroup analysis should strictly only involve factors identified before randomization, we were concerned that the rate of FGR reported at trial entry (10%) underrepresented the actual degree of FGR in this high-risk cohort and could potentially obscure the influence of FGR on outcomes. Authors of studies have estimated that <25% of infants who are small for gestational age are identified antenatally as having a potential growth problem, even in pregnancies known to be at risk, and detection rates are even lower in those not considered at risk.^39,40 We took a conservative approach, defining FGR as a birth weight less than the third centile, rather than the more commonly used 10th centile. Customized rather than population centiles were used because of the strong association between preterm birth and FGR.^18,41 Despite this narrower definition, FGR was identified in a third of the cohort, which is consistent with other reports.¹⁸ As expected, the rate of serious neonatal morbidity was 60% higher in the FGR subgroup, although the rate of severe neonatal lung disease did not differ between infants with and without FGR.

Our findings are perhaps somewhat surprising, given the number of animal studies in which authors have linked fetal corticosteroid exposure to both reduced fetal growth and cardiometabolic disease.^12,42,–45 In addition, animal studies have suggested that maternal betamethasone administration may exacerbate fetal cardiovascular dysfunction in FGR.^12,46 However, human trials have consistently demonstrated that antenatal corticosteroid treatment is safe in the long-term.^1,3 The reason for this discrepancy has not been fully explained but may be due to the use of physiologic doses of corticosteroids, shorter durations of exposure in clinical studies, and interspecies differences in the timing of organ development.⁴⁷ 

Authors of some studies have suggested that antenatal corticosteroid treatment may be less effective in FGR because of increased endogenous corticosteroid production and reduced clearance of corticosteroids from the fetal compartment.^48,49 However, in preterm lambs with FGR induced by single uterine artery ligation, maternal antenatal betamethasone continued to induce lung structural maturation and expression of surfactant proteins, although fetuses with growth restriction had increased endogenous corticosteroid concentrations.⁵⁰ In a large population study of preterm infants who were small for gestational age, those exposed to a single course of antenatal corticosteroids had reduced rates of death and intraventricular hemorrhage but not respiratory morbidity.⁵¹ Our study has revealed that, in FGR, repeat doses of antenatal betamethasone reduced the severity of neonatal lung disease and use of surfactant, suggesting that repeat doses may be required for pulmonary maturation in FGR in humans.

This study has some important limitations, including the limited size of the New Zealand ACTORDS cohort and the inherent risk of bias in subgroup analysis. However, we felt that it was important to undertake this exploratory analysis given the high rates of FGR among preterm infants and the ongoing concerns around efficacy and safety of repeat corticosteroid exposure in this subgroup. In addition, the ACTORDS is the only trial of repeat antenatal corticosteroids in which researchers have undertaken detailed follow-up of participants in midchildhood, at an age when cardiometabolic parameters become predictive of later risk of chronic disease.⁹ We did not have access to paternal height, but given that maternal height was similar in randomized treatment groups, we would not expect any imbalance in midparental height.

Repeat antenatal betamethasone treatment had no adverse effects on later cardiometabolic function, even in the presence of FGR. It may have a positive effect on childhood height in FGR. Clinicians should use repeat doses of antenatal corticosteroids when indicated before preterm birth, regardless of FGR, in view of the associated neonatal benefits and absence of later adverse effects.

The following ACTORDS Follow-up Steering Group members are nonauthor contributors:

Ross R. Haslam (the Women’s and Children’s Hospital), Lex W. Doyle (Departments of Paediatrics and Obstetrics and Gynaecology, The Royal Women’s Hospital, University of Melbourne), Peter J. Anderson (Department of Paediatrics, University of Melbourne and Department of Clinical Sciences, Murdoch Children’s Research Institute), Jeffrey S. Robinson (Department of Obstetrics and Gynaecology, School of Medicine, The University of Adelaide) (including authors Caroline A. Crowther and Jane E. Harding); ACTORDS clinical trial coordinator: Pat J. Ashwood (Department of Obstetrics and Gynaecology, School of Medicine, The University of Adelaide); New Zealand ACTORDS clinical trial coordinator: Coila Bevan (Liggins Institute, University of Auckland); trial statistician: Kristyn Willson (Department of Obstetrics and Gynaecology, School of Medicine, The University of Adelaide); DXA scans on children in Christchurch were conducted by Ann Mansfield (St Georges Radiology, Christchurch Radiology Group); supervision of laboratory assays was conducted by Eric Thorstensen (Liggins Institute, University of Auckland).

ACTORDS

Australasian Collaborative Trial of Repeat Doses of Corticosteroids

AMBP

ambulatory blood pressure

CI

confidence interval

DXA

dual energy radiograph absorptiometry

FEF25-75

mean forced expiratory flow at 25% to 75% of forced vital capacity

FEV1

forced expiratory volume in 1 second

FGR

fetal growth restriction

FVC

forced vital capacity

MD

mean difference

RDS

respiratory distress syndrome

RGM

ratio of geometric means

We thank the mothers and children who participated in the ACTORDS trial and this follow-up study.