Highlight
– The APOSTEL 8 trial randomized 755 women with threatened preterm birth at 30+0 to 33+6 weeks to intravenous atosiban versus placebo and found no superiority for the primary composite of perinatal mortality and six severe neonatal morbidities (RR 0.90, 95% CI 0.58–1.40).
– Neonatal death was rare in both arms (0.7% vs 0.9%); maternal adverse events and maternal mortality did not differ.
– These findings challenge routine, standardized use of atosiban for threatened preterm birth in this gestational window and support more selective, individualized application focused on scenarios where immediate benefit is plausible (e.g., need for corticosteroid course or in utero transfer).
Background and clinical context
Preterm birth remains the leading cause of neonatal morbidity and mortality worldwide. Management of threatened preterm labour seeks to reduce immediate birth and thereby allow interventions that improve neonatal outcomes: administration of antenatal corticosteroids to accelerate fetal lung maturity, magnesium sulfate for neuroprotection before very preterm birth, and transfer to an appropriate neonatal unit. Tocolytics are drugs used to suppress uterine contractions temporarily, and they are widely recommended by many international bodies for selected patients to delay delivery long enough for these interventions or for maternal transfer.
Atosiban is an oxytocin receptor antagonist approved in many countries specifically as a tocolytic. Prior trials and systematic reviews have shown that tocolytics, including atosiban, can prolong pregnancy short term; however, robust evidence that tocolysis improves neonatal morbidity or mortality has been lacking. The APOSTEL 8 trial was designed to test whether atosiban improves clinically meaningful neonatal outcomes when used for threatened preterm birth between 30+0 and 33+6 weeks of gestation.
Study design and methods
APOSTEL 8 was an international, multicentre, randomized, double-blind, placebo-controlled superiority trial conducted in 26 hospitals across the Netherlands, England, and Ireland. Eligible women were aged 18 years or older with a singleton or twin pregnancy presenting with threatened preterm birth between 30+0 and 33+6 weeks of gestation. After written informed consent, participants were randomized 1:1 (stratified by centre) to receive intravenous atosiban or placebo.
The predefined primary outcome was a composite of perinatal mortality (stillbirth and death up to 28 days postpartum) and six severe neonatal morbidities. Secondary outcomes included individual components of the composite, overall neonatal length of stay, need for respiratory support, and maternal adverse events. Analysis followed intention-to-treat principles, with treatment effect expressed as relative risk (RR) and 95% confidence intervals (CIs). The trial was prospectively registered (EudraCT 2017-001007-72; Netherlands Trial Registry NL-OMON54673) and funded by ZonMw.
Key findings
Between Dec 4, 2017, and July 24, 2023, 755 participants were randomized and 752 were included in the intention-to-treat analysis (atosiban n=375, placebo n=377). Primary-outcome data were available for 884 infants (including twins): 449 infants in the atosiban group and 435 in the placebo group.
The primary composite outcome occurred in 37 of 449 infants (8.2%) in the atosiban arm versus 40 of 435 infants (9.2%) in the placebo arm (RR 0.90; 95% CI 0.58–1.40). This effect estimate is compatible with a modest reduction or no effect; the confidence interval crosses unity and the trial did not meet its predefined superiority threshold. There were three infant deaths (0.7%) in the atosiban arm and four (0.9%) in the placebo arm (RR 0.73; 95% CI 0.16–3.23). Clinical adjudication attributed none of the deaths as probably related to the study drug.
Maternal adverse events were similar between trial arms, and there were no maternal deaths. The authors report no significant differences in individual neonatal morbidities included in the composite, and overall safety signals were neutral.
Interpretation and clinical implications
APOSTEL 8 provides high-quality randomized evidence that routine use of atosiban for threatened preterm birth at 30+0 to 33+6 weeks does not improve a clinically meaningful composite of perinatal mortality and severe neonatal morbidity when compared with placebo. The absence of a clear neonatal benefit calls into question standardized, across-the-board administration of atosiban in this gestational window.
Several pragmatic implications follow.
- Reconsider routine tocolysis in late preterm threatened labour: Clinicians and guideline panels should weigh the lack of demonstrated neonatal benefit in this trial when recommending atosiban for 30–34 weeks gestation. Where tocolysis is used, a clear rationale should be documented (e.g., to allow a full course of antenatal corticosteroids in women who have not yet received them, or to allow safe maternal transfer to a higher-level neonatal facility).
- Individualize decision-making: The primary objective of tocolysis is neonatal benefit, not merely prolongation of pregnancy. Shared decision-making that considers gestational age, cervical status, fetal condition, availability of neonatal care, urgency of steroid administration, and maternal preferences is essential.
- Selective use remains reasonable in specific scenarios: The trial does not directly speak to very preterm gestations (<30 weeks), where the neonatal risk is higher and the potential absolute benefit of interventions enabled by tocolysis (e.g., steroids, magnesium, transfer) is greater. It also does not negate the utility of tocolysis when a clear, short-term objective exists (for example, to complete a steroid course or stabilize a maternal transfer).
Strengths and limitations of the trial
Strengths of APOSTEL 8 include rigorous randomization and blinding, multicentre international conduct, inclusion of twins and singletons, clinically meaningful composite primary outcome, and intention-to-treat analysis. The trial focused on a pragmatic and policy-relevant question in a gestational window where practice varies.
Limitations to consider:
- Event rate and power: The observed primary-event frequency was relatively low (≈8–9%), and although the trial enrolled a substantial sample (n=752 analysed), the confidence interval around the RR includes both a clinically meaningful benefit and no effect. The trial was designed for superiority; absence of superiority does not prove equivalence.
- Composite outcome interpretation: Composite endpoints combine events of variable severity and frequency. If one component dominates event counts, interpretation for specific clinically relevant outcomes (e.g., severe respiratory morbidity versus intraventricular haemorrhage) can be nuanced.
- Generalisability: The study was conducted in high-resource settings (Netherlands, England, Ireland) where neonatal care and use of antenatal corticosteroids are readily available. Outcomes might differ in lower-resource settings where the ability to deliver corticosteroids, magnesium sulfate, or neonatal intensive care is limited, and where delaying birth even briefly could be more consequential.
Expert commentary and guideline perspective
APOSTEL 8 aligns with the wider body of evidence showing that tocolytic drugs can delay delivery but have not consistently demonstrated improvements in hard neonatal outcomes. For clinicians, the trial underlines the importance of targeting tocolysis to scenarios with plausible downstream benefits. Guideline committees should integrate these data when updating recommendations for management of threatened preterm labour in the late preterm window.
For hospital protocols, the trial supports reducing reflexive or routine atosiban administration for all cases at 30–34 weeks and instead adopting an approach that documents a clear objective for tocolysis (e.g., to enable complete corticosteroid administration or safe transfer) and discusses risks and expected benefits with the patient.
Areas for further research
Remaining questions include whether tocolysis influences outcomes in earlier gestations with higher baseline neonatal risk, whether particular subgroups (by cervical dilatation, biomarkers, or fetal condition) benefit, and whether alternative tocolytic strategies or combination approaches provide net neonatal advantage. Research in lower-resource settings is also warranted because the balance of benefits and harms may differ.
Conclusion
The APOSTEL 8 randomized, double-blind trial shows that atosiban given for threatened preterm birth at 30+0–33+6 weeks did not reduce a composite of perinatal death and severe neonatal morbidity compared with placebo. These results argue against routine, standardized use of atosiban in this gestational window and support selective, individualized use when there is a clear, time-limited objective such as completing antenatal corticosteroids or securing maternal transfer. Guideline bodies and clinicians should consider these findings when making recommendations and counselling patients.
Funding, registration, and trial citation
Funding: ZonMw.
Trial registration: EudraCT 2017-001007-72; Netherlands Trial Registry NL-OMON54673.
Key citation: van der Windt LI, Klumper J, Duijnhoven RG, Kok M, Ris-Stalpers C, de Boer MA, van Kaam AH, Pajkrt E, Mol BW, Walker KF, McAuliffe FM, van der Post JA, Roos C, Oudijk MA; APOSTEL 8 Study Group. Atosiban versus placebo for threatened preterm birth (APOSTEL 8): a multicentre, randomised controlled trial. Lancet. 2025 Mar 22;405(10483):1004-1013. doi: 10.1016/S0140-6736(25)00295-8. Epub 2025 Mar 3. PMID: 40049187.
References
van der Windt LI, Klumper J, Duijnhoven RG, et al. Atosiban versus placebo for threatened preterm birth (APOSTEL 8): a multicentre, randomised controlled trial. Lancet. 2025;405(10483):1004–1013. doi:10.1016/S0140-6736(25)00295-8.