The first Australian Phase 3 iSEARCH RCT.
Primary Research Question
Does maternal oral Sildenafil in labour, compared with placebo, reduce the primary composite endpoint of ten adverse neonatal outcomes? (Table 1)
Secondary Research Questions
(a) Does maternal oral Sildenafil in labour reduce the rate of the secondary outcome of emergency operative birth for fetal distress?
(b) Is it more cost effective?
Study Design: This is a two-arm parallel, randomised (1:1), placebo-controlled, double-blind multicentre superiority trial of Sildenafil vs. placebo for women in labour at term (≥ 37+ 0 weeks gestation).
This study aims to test the hypotheses that up to 3 doses of oral 50 mg Sildenafil (vs. placebo) is safe and will improve perinatal and maternal outcomes related to intrapartum hypoxia.
To evaluate whether, compared with placebo, Sildenafil reduces the relative risk of the composite perinatal endpoint by 35%, from 7–4.55%. The composite endpoint comprises the ten components shown in Table 1.
1. To evaluate whether Sildenafil results in a reduction of the RR of each individual component of the composite primary outcome.
2. To evaluate whether Sildenafil results in a reduction of the RR of Caesarean section or instrumental vaginal birth for fetal distress by 25% (from 20–15%).
3. To evaluate whether Sildenafil is more cost-effective than placebo.
The incidence of the composite adverse outcome is estimated at 7% based on data from several of the participating hospitals, the ARRIVE RCT,25 and the Australian and New Zealand Neonatal Network.37 To detect a 35% reduction from 7–4.55% for an alpha of 0.05 and > 80% power with 2% drop-out and 5% loss to follow up in each arm, needs about 3,200 women (Table 2, scenario 1). This sample size also yields > 90% power to detect a 25% reduction for the secondary outcome of Caesarean section or instrumental vaginal birth for fetal distress.
1. Women with singleton or dichorionic twin pregnancies, planning vaginal birth at term (≥ 37+ 0 weeks gestation).
2. Age ≥ 18 years.
3. Willing and able to comply with all study requirements.
4. Signed, written informed consent.
1. A woman should not be enrolled if the responsible clinician or the woman are, for any medical or nonmedical reasons, reasonably certain that Sildenafil would be inappropriate for her in comparison with no treatment or some other treatment that could be offered outside the trial.38
2. Monochorionic twins, triplets or higher order multiple births, which are generally delivered electively before term.
3. Women who are taking any type of nitrate drug therapy or who utilize short-acting nitrate-containing medications during labour (such as sodium nitroprusside, bosentan, fosamprenavir and ritonavir combination, hepatic enzyme inhibitors CYP3A4 (including itraconazole, ketoconazole, ritonavir, cimetidine, erythromycin, saquinavir, darunavir), or hepatic enzyme substrates (CYP3A4), medications used to treat pulmonary arterial hypertension, and other phosphodiesterase type 5 inhibitors, due to the risk of potentially life-threatening hypotension.39
4. Severe hepatic or renal impairment.39
Screening, registration and randomisation
All women attending antenatal clinics in participating hospitals from ≥ 34+ 0 weeks will be screened for eligibility by study midwives. Women who wish to participate will be registered in an online trial registration database, as detailed in the Study Manual (e-supplement 1). Once registered, each woman is assigned a unique study number and receives routine obstetric care until the onset of spontaneous labour or induction of labour, when randomisation is undertaken. Individuals may only be registered and randomised once. Randomisation is undertaken after the responsible clinician (midwife or obstetrician) has again confirmed that the woman is eligible to participate. If an eligible woman presents for the first time in labour, screening, registration and randomisation are done together. Once registration and randomisation are completed, the participant is assigned a treatment arm. A log at each site will record the number of women screened for eligibility, those eligible, those approached and those consented. If a woman declines participation, a reason is recorded if she agrees to provide one.
At registration, women will be invited to consent to participate in childhood neurodevelopmental follow-up and for data to be extracted from Medicare Benefits Schedule and Pharmaceutical Benefits Schemes claims records for herself and her infant and from the Australian cerebral palsy register. If she consents, contact details will be recorded and follow-up assessment will be performed as described below. If consent is declined or withdrawn before follow-up takes place (2–3 years corrected age), no further contact will take place.
The study intervention is oral Sildenafil. The control intervention is placebo. Study treatment only begins after transfer to the labour ward either in spontaneous labour or for induction of labour (artificial rupture of membranes +/- oxytocin). Women are given the first dose by the attending midwife in the labour ward. Women receive Sildenafil 50mg or identical placebo orally every 8 hours to a maximum of three doses, from identical matched treatment packs supplied by pharmacy.
Management of Labour and Puerperium
Continuous electronic intrapartum FHR monitoring will be performed in all women. Umbilical artery cord pH will be measured in all women after birth. Classification of FHR abnormalities is based on the Royal Australian and New Zealand College of Obstetricians and Gynaecologists guidelines.46 In a subset of women, 20ml of blood will be collected for assay of soluble Fms-like tyrosine kinase-1 (sFlt-1) and Placental Growth Factor (PlGF) levels. These women will also have an ultrasound scan performed to assess the fetoplacental circulation and liquor volume before and after treatment. The ultrasound data and maternal sFlt-1 and PlGF levels will allow post-hoc subgroup analysis to identify cohorts at risk of fetal distress that might derive greater benefit from Sildenafil treatment. Otherwise, intrapartum management will be in accordance with local hospital guidelines.
To detect possible persistent pulmonary hypertension of the newborn, all infants will also receive routine oxygen saturation screening to detect hypoxemia, generally 24–48 hours after birth but, if necessary, four hours after birth. Infants with oxygen saturations of > 95% are very unlikely to have major congenital heart or significant pulmonary disease, including persistent pulmonary hypertension of the newborn. Infants with oxygen saturations ≤ 95% will receive further assessment by the paediatric team which may include echocardiography.
In-hospital maternal and neonatal outcome events occurring from randomisation to discharge home will be collected from medical records into electronic Case Report Forms. Assessments of outcome will be blinded to treatment allocation. Except for 28-day neonatal mortality, all primary outcome data will be routinely available before discharge and collected electronically. There are no formal study assessments visits. Research midwives will contact women 30 days after discharge to ascertain further relevant issues. Childhood follow up will be conducted using parent report questionnaires and by linkage with educational databases and cerebral palsy registers.40–42
A detailed Statistical Analysis Plan will be published before data analysis begins43, as will a detailed health economic evaluation protocol. Primary and secondary analyses will adhere to an intention-to-treat basis using generalised linear models (binary or normal). Intervention effect will be presented as relative risk or mean difference, as appropriate, with 95% confidence intervals. Primary analyses will be unadjusted. Numbers needed to treat to prevent one adverse outcome will be calculated. Where there are differences in baseline characteristics between the two treatment groups that might be associated with outcomes, secondary analyses of the primary outcome will be carried out using multiple (log-binomial) regression. Reporting will follow CONSORT44 and TRIPOD45 guidelines.
Secondary outcomes include each of the ten individual components of the primary composite outcome (Table 1). P values adjusted for these ten comparisons will be derived using the Benjamin-Hochberg procedure, limiting the false discovery rate to 5%.46 47 Results of other endpoint, subgroup and sensitivity analyses will be interpreted in proper context and with due consideration of the risk of Type I error. Interpretation of statistical evidence will also consider the recommendations of Pocock et al.48 49
The international STRIDER RCTs used daily maternal Sildenafil doses of 75 mg in pregnant women with severe fetal growth restriction between 20 and 30 weeks gestation for up to 10 weeks, i.e. cumulative doses of up to 5,250 mg.50 51 In 2019, The Dutch STRIDER trial was stopped on the advice of the Data Safety Monitoring Board due to an increase in persistent pulmonary hypertension of the newborn (PPHN) with a non-significant increase in neonatal deaths in the Sildenafil group.52 PPHN occurred in 16 neonates (18.8%) in the Sildenafil group vs. 4 neonates (5.1%) in the placebo group (relative risk, 3.67; 95% CI, 1.28–10.51; P = 0.008). A subsequent meta-analysis of 329 participating women in all available trials showed no difference in neonatal deaths.53 However it was recommended that Sildenafil not be prescribed outside of clinical trials.35 The indication and cumulative dosage of Sildenafil in the iSEARCH trial are very different from the STRIDER trials as we will only use a maximum 150mg in women at term, which is thirty five times lower. PPHN will be monitored as one of ten components of the primary composite endpoint by the independent data and safety monitoring board, as outlined below.
The Australian iSEARCH RCT Independent Data and Safety Monitoring Board
The IDSMC will review interim data on the primary outcome, specified adverse events and other evidence after 50% of recruitment or whenever they deem appropriate, as recommended by Peto, Pocock and others.54–56 There will be no adjustment to alpha for interim analyses.
Interim analyses of the primary composite outcome
The IDSMC will advise the TMC if in their view there is proof beyond reasonable doubt of net benefit or harm for the primary composite endpoint, for example employing a commonly used formal threshold of P < 0.001 for nominal significance, as recommended by Geller and Pocock.56
Interim analyses of mortality
The IDSMC will advise the TMC if in their view there is a difference in mortality due to intrapartum stillbirth and/or 28-day neonatal mortality identified as a deviation from the null indicated by a Haybittle-Peto boundary of 3 standard errors from the null, which is equivalent to P < 0.0027,54 55 which would be needed to justify recommending early stopping.
Overarching International Data Monitoring Committee
The Chair or a representative of the Australian iSEARCH trial will serve on an overarching international data monitoring committee which will receive regular primary outcome data from all participating trials, which may also recommend early stopping if there is evidence from pooled data with high confidence, e.g. a deviation in the primary outcome equivalent to P < 0.00156, or a deviation in mortality due to intrapartum stillbirth and/or 28 day neonatal mortality of more than 3 standard errors from the null, equivalent to P < 0.0027,54 55 which in conjunction with other available evidence would be likely to change the practice of most clinicians worldwide.57
Ethics and regulatory compliance
The study will be conducted according to the International Conference on Harmonization Guidelines for Good Clinical Practice. The study will comply with all applicable laws and regulations and the principles laid down by the World Medical Association in the Declaration of Helsinki 2013. It is registered in the Australian and New Zealand Clinical Trial Registry and ethical and regulatory approvals will be obtained before it begins.
Rationale for an IPD PMA to be coordinated by an international iSEARCH Consortium
While the first Australian iSEARCH Phase 3 RCT in 3200 women has 80% power to show a reduction of 35% in the relative risk of its primary composite endpoint, it lacks adequate power to show (a) a more moderate, but clinically important reduction of 20% in the relative risk of this outcome;57 (b) whether Sildenafil is similarly effective in important subgroups of participants in different settings; and (c) clinically relevant differences in rarer, but critically important outcomes such as intrapartum stillbirth, or neonatal or maternal death. We therefore outline the rationale for (i) an individual participant data prospective meta-analysis in countries with low perinatal mortality rates and (ii) a mega-trial in countries with moderate or high perinatal mortality rates, to be undertaken by a global alliance to be known as the international iSEARCH Collaboration.
What is an Individual Participant Data Prospective Meta-Analysis (IPD PMA)?
In an IPD PMA, studies are included in a meta-analysis before their results are known.36 An IPD PMA is like an international mega-trial, but less resource-intense for individual countries and funders. IPD PMAs have been described as next generation systematic reviews because they reduce key problems of traditional retrospective aggregate data meta-analyses such as publication bias, selective reporting bias and incompatible definitions of study outcomes which may otherwise lead to missing values, inability to synthesise all relevant data and consequent research waste.36 Because data is centrally collected and checked followed by re-analysis of the original, line-by-line data of all randomised patients from each of the trials, IPD PMAs offer many advantages over traditional retrospective aggregate data meta-analysis. The resulting analyses in IPD PMAs can (i) increase statistical power to detect meaningful effects and interactions; (ii) more reliably determine how benefits and harms vary according to individual risk factors; (iii) harmonise data collection and outcomes across the collaboration and agree on common core outcomes to be collected by all trials to provide more consistency and standardisation of results thus minimising research waste; and (iv) undertake robust subgroup analyses to examine potential differential intervention effects. Conducting an IPD PMA rather than a large multi-centre RCT also allows each contributing RCT to be funded independently, circumventing the need to fund a single, often financially prohibitive mega-trial, while achieving comparable statistical power. It also allows each contributing RCT the flexibility to answer additional local questions or modify procedures to reflect local practice. Ensuring that future RCTs to test questions addressing disability-free survival and/or mortality are planned, executed, and interpreted through IPD PMAs represents a major conceptual advance in perinatal medicine.5859
The IPD PMA outlined below will be conducted by an alliance known as the international iSEARCH Collaboration, which will include a central steering committee responsible for leading the PMA and managing the collaboration, a group responsible for data management, processing and synthesis, and representatives from each RCT, who will be involved in decisions on the protocol, analysis, and interpretation of the results. The iSEARCH Collaboration will establish a joint, overarching international Data Monitoring Committee, which will include the Chair or a representative of the IDSMCs of each contributing RCT, to receive synthesised data from all included studies at pre-specified times. Detailed protocols and statistical analysis plans for the two PMAs will be published with input from all participating trialists and collaborators, developing the outlines presented here.
I: IPD PMA of iSEARCH RCTs in high income countries with low perinatal mortality rates.
This IPD PMA will include the first iSEARCH Phase 3 RCT whose protocol is described in this paper. Additional Phase 3 iSEARCH RCTs are being planned in Australia and other countries with low perinatal mortality rates and, if funded by NHMRC in Australia and other agencies, are likely to begin before the first iSEARCH RCT completes recruitment. This would fulfil the scenario of a de novo PMA of RCTs beginning at different stages,36 like that conducted by the five oxygen saturation targeting trials of the NeOProM Collaboration,60 61 62 which was formed after the first RCT63 was about to begin and the others were in the early planning stages.64–66 Thus, new trials will be eligible to join the IPD PMA up until the point when the first Australian iSEARCH trial has published its primary outcome results, currently expected to be in 2025. Because we estimate that the incidence of the primary composite endpoint of this PMA in high income countries is 7%, to demonstrate a 20% reduction in the relative risk of this outcome from 7–5.6% with 90% power would require all iSEARCH RCTs in Australia and other HICs to enrol ~ 14,000 participants (Table 2, scenario 2). Under the same assumptions, if the incidence of the primary composite endpoint in HICs varied between 5% and 9%, the total sample size required in HICs might vary between about 20,000 and 10,000 respectively (Table 2, scenarios 3 and 4).
Should we wait for the results of the first iSEARCH RCT before seeking further funding for a second Australian RCT and additional Phase 3 RCTs? The sample size of 3200 in the current iSEARCH RCT yields less than 50% power to detect a moderate and realistic 20% reduction in its primary composite outcome. An inconclusive result therefore may well represent a false negative conclusion. Such uncertainty could only be resolved by achieving a larger sample. Waiting for the results of the current iSEARCH RCT could thus delay obtaining a definitive result by several years. On the other hand, if it or the IPD PMA to which it contributes, clearly demonstrates a relative reduction in their primary composite outcome in countries with low perinatal mortality rates, it could be stopped early, and Sildenafil introduced into routine intrapartum care.
II: Rationale for an iSEARCH mega-trial in countries with moderate or high perinatal mortality rates.
As the combined incidence of intrapartum stillbirth and 28-day neonatal mortality is rare (< 1%) in high-income countries with low perinatal mortality rates, a multi-centre RCT or IPD PMA might require hundreds of thousands of women for sufficient statistical power to detect a moderate reduction in this endpoint. This question is thus better answered by studies in low or middle-income countries with moderate or high perinatal mortality rates, where the incidence of intrapartum stillbirth or neonatal mortality is much higher.
Collecting all ten component outcomes of the primary composite endpoint of the Australian iSEARCH RCT would be problematic in countries with moderate or high perinatal mortality rates, where the infrastructure and expertise to determine umbilical artery pH, diagnose persistent pulmonary hypertension or provide neonatal respiratory support or admission to a neonatal intensive care unit are not consistently available. Conversely, the combined incidence of intrapartum stillbirth67 68 or 7-day neonatal mortality69 70 is considerably higher in low resource settings, making it logical and feasible to adopt this as the primary composite endpoint of a mega-trial in these countries.71 Assuming a control rate of 2% for a primary composite endpoint of intrapartum stillbirth and 7-day neonatal mortality, a RCT of Sildenafil vs placebo contributing to the IPD PMA in countries with moderate or high perinatal mortality rates might need 14,000 participants to demonstrate a 35% reduction in the relative risk of this outcome with two tailed statistical significance of 5% and 90% power. (Table 2, scenario 5). Assuming the same rates of drop out and loss to follow up, detecting a more moderate, but still clinically important relative risk reduction of 20% with similar precision might require about 50,000 women. (Table 2, scenario 6). Under the same assumptions, if the incidence of the composite outcome of intrapartum stillbirth or 7-day neonatal mortality varied between 1.2% and 2.8%, the total sample size required in countries with moderate or high perinatal mortality rates would vary between about 85,000 and 36,000 respectively, (Table 2, scenarios 7 and 8). Achieving such numbers will require highly streamlined processes to ensure fast recruitment with minimal data collection,5 72 (Table 3) as in the international mega-trials which each enrolled tens of thousands of patients with suspected myocardial infarction,6–9 the MAGPIE trial72 of magnesium sulphate in severe pre-eclampsia, which enrolled 10,141 women in 33 countries in 3.5 years, and the UK RECOVERY5 and WHO SOLIDARITY31 trials.
III: Role of the ALPHA Collaboration in advancing randomised studies of mortality and disability
A key challenge for future perinatal trials is how to enrol the numbers needed to find moderate, but clinically relevant, improvements in healthy survival.73 The ALPHA Collaboration74–77 (www.alphacollaboration.com) will work with other organisations to identify questions of high priority to stakeholders worldwide for perinatal trials addressing mortality as the primary outcome and focus globally collaborative efforts on rapidly answering those questions in a new generation of low-cost, mega-trials and prospective meta-analyses of trials enrolling from 5,000 to 50,000 or more participants, as is planned by the iSEARCH Trials. Randomised perinatal mega-studies of this size will also yield considerably more precise and reliable estimates of disability in survivors than has previously been typical - a major benefit.