Introduction

Preeclampsia complicates around 2 to 8% of pregnancies world wild [1, 2]. It can lead to maternal mortality and morbidity, which in turn lead to perinatal morbidity and mortality due to fetal growth restriction (FGR) and medically indicated prematurity [3, 4]. Antiplatelet agents are known to prevent preeclampsia and its consequences when they are administered before 16 weeks of gestation [5, 6]. The difficulty, however, is early identification of pregnancies at high risk of preeclampsia that could benefit from this preventive treatment [7, 8]. Parous women with a history of preeclampsia are candidates, but what is challenging is identifying nulliparous women who should receive this preventive treatment as they have roughly twice the risk to develop preeclampsia when compared to parous women [9, 10].

Preeclampsia and growth restriction are the consequence of abnormal placental implantation and inadequate utero-placental blood flow. Normal placentation comprises trophoblast cell invasion of the decidual and myometrial segments of spiral arteries, which induces reversible changes in the normal arterial wall architecture. Trophoblastic invasion starts from eight weeks’ gestation and studies have shown that abnormal uterine artery Doppler as early as the first trimester can identify women at high risk of preeclampsia and fetal growth restriction [11]. The pulsatility index, alone or combined with notching, are the most predictive uterine artery Doppler indices in the first trimester [12].

As women who develop preeclampsia have different uterine artery Doppler patterns on their first-trimester ultrasound [13, 14], we decided to conduct a trial to test the efficacy of low-dose aspirin to reduce the incidence of preeclampsia or growth restriction in nulliparous pregnant women identified at "high-risk" by their first-trimester ultrasound Doppler findings.

Material and methods

Results

Between June 2012 and June 2016, 1104 women were randomized. The trial was stopped after 4 years due to recruiting difficulties resulting in a failure to conduct any of the four initially planned intermediate analyses. Because two women withdrew their consent and two terminations of pregnancy occurred after randomization, 1100 women were finally analysed: 550 in the intervention (low-dose aspirin) and 550 in the control (placebo) group (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1.

https://doi.org/10.1371/journal.pone.0275129.g001

The mean age of the women were 28.3 ± 4.9 years and 28.7 ± 4.7 years in the aspirin and the placebo group, respectively. Eight percent of women were of white ethnicity, S1 Table. The median gestational age at delivery were 39.7 [38.5–40.6] and 39.7 [38.5–40.8] in the aspirin and the placebo group, respectively. The number (rate) of cases of preeclampsia or infants with a birthweight ≤ 5th percentile was 88 (16.0%) in the low-dose aspirin group versus 79 (14.4%) in the placebo group (proportion difference 1.6 [-2.6; 5.9], p = 0.45 OR 1.14 95%CI (0.82–1.58)), S2 Table. These findings were supported by our complete-case analysis, which showed 76 (14.1%) and 72 (13.3%) with preeclampsia or infants with birthweight ≤ 5th percentile in the low-dose aspirin and the placebo groups, respectively (P = 0.68). The two groups did not differ for secondary outcomes. Also in each group, during treatment, around 25% of the women reported bleeding (epistaxis or gingival bleeding (80%) and metrorrhagia (20%)) as an adverse effect. Overall, 112 (20%) and 120 (22%) women, respectively, experienced at least one serious adverse event (S3 Table). In all, around 25% of the women in each group stopped their treatment before their 34^th week.

Discussion

In this trial conducted among women identified at high risk of preeclampsia by their first-trimester uterine artery Doppler examination, the rate of preeclampsia or birthweight<5^th percentile in women who received low-dose aspirin before 16 weeks gestation was not statistically different from that of women receiving placebo.

This trial was planned before the results from the ASPRE trial which also evaluated an intervention among women at high risk of preeclampsia [19]. They identified women by using an algorithm including demographic, clinical, biological, and Doppler findings, which produced a population very different from ours, with 30% of the women parous and 10% with a history of preeclampsia. In the ASPRE trial, approximately forty percent of the primary outcomes occurred in parous women. Being limited to nulliparas is the main strength of our trial, as parous women differ substantially in terms of risk assessment for preeclampsia and are not the women for whom the decision about preventive treatment is challenging. Other strengths include that all sonographers were certified for uterine artery Dopplers and also that all women were administered their treatment before 16 weeks of gestation which is the only way the preventive treatment can be effective.

Nevertheless, our trial has limitations, including that we were not able to reach the number of inclusions initially planned, although we extended our inclusion period by one year. This may be explained partly by an inaccurate initial assessment of feasibility and the lower than expected number of eligible women. Recruitment may also have been difficult as healthy nulliparous women may not have understood how their participation could improve outcomes for them or the baby as, by definition, nulliparas have never experienced adverse pregnancy outcomes. Another limitation was the lower than expected rate for the main outcome in the control group (14% of events versus 21% expected). Our hypothesis and sample size were built on numbers of outcomes which were not specific from the French population. A French study conducted when the trial was designed found that preeclampsia affected only 1.7% of nulliparous women whereas we used an expected rate of 6% [10]. It is likely that the lower than expected rate for the main outcome may be explained, in part, by this erroneous assessment of the rate of preeclampsia in nulliparous women in France [9]. Thus, we can speculate that if an intermediate analysis would have been conducted, it is likely that the trial would have been stopped for futility. One in four woman interrupted their treatment before their 34^th week of gestation which is concordant with compliance findings from metanalysis evaluating the effect of aspirin to prevent preeclampsia in the pregnant population [5]. The reported rates of women who were fully compliant in the studies included in the metanalysis varied between 60 and 90%. Detailed compliance was not assessed precisely or regularly through the follow up of the pregnancy so we are unable to explain these adherence rates. Also, the screening took place in 17 participating units and we were not able to assess how many first-trimester ultrasound examinations of nulliparous women were conducted by sonographers during the inclusion period. We are therefore unable to estimate how many women were screened. Finally, we suspect some of the adverse events may have been under-reported. Indeed, some of the adverse events were also secondary outcomes (metrorrhagia and preterm delivery) and there are some discrepancies between the numbers of outcomes and adverse events which is why we are cautious in the interpretation our adverse event data.

The observed rate of preterm preeclampsia in our control group was half that in the ASPRE trial control group, which means that their algorithm was more successful in identifying women at risk of preeclampsia. We found another study also focusing on the evaluation of first trimester uterine Doppler among a population of nulliparas who were not particularly at high risk [23]. They found a rate of preeclampsia in their population (4.9%) which was close to what we observed in our study thereby confirming that we were unable to identify a high-risk population. Bujold et al.’s meta-analysis which evaluated acetylsalicylic acid for the prevention of preeclampsia and intra-uterine growth restriction in women with abnormal uterine artery doppler found that this treatment, when initiated early, could reduce the incidence of preeclampsia and in particular severe preeclampsia [24]. It turns out that most of the observed cases of term preeclampsia and was mild which may explain partly the lack effect of the treatment in our trial.

Algorithms that aim to identify women at high risk of preeclampsia by clinical, biological, and ultrasound parameters have been published and have sparked enthusiasm, but the cost effectiveness of their implementation at a population level is still debated [25–28]. The trial was designed in 2008 and at the time, none of those algorithms were recommended which is why we selected our population differently. Until now, those algorithms are still not recommended by national guidelines even if recommended by the International Federation of Gynecology and Obstetrics, the International Society of ultrasound in obstetrics and Gynecology and the International Society for the Study of Hypertension in Pregnancy [29, 30]. Many studies were published during the elapsed time between de design of the trial (2008) and the publication (2022). This delay was mainly due to motivation issues and organisation difficulties. Inclusion criteria to select women at high risk of preeclampsia which seemed appropriate at the time are now obsolete. Because women in most high-income countries are advised to undergo a first-trimester ultrasound, screening for preeclampsia through clinical and Doppler findings would not require a major increase in cost [31]. Accordingly, a combination of clinical and Doppler findings should be used in further studies to evaluate interventions to reduce preeclampsia.

Conclusion

In our study, low-dose aspirin was not associated with a lower rate than placebo of either preeclampsia or FGR in women identified as at high risk of preeclampsia during a first-trimester uterine artery Doppler examination. We interpret our results cautiously because of the lack of power related to insufficient number of patients recruited.

Supporting information

Acknowledgments

We thank all the women who agreed to participate in the study. We would like to thank Anne Rebion for her analysis, Hélène Bourgoin for her help with the pharmacy. We also thank the research assistants who helped to conduct this trial Yoann Desvignes and Catherine Fermont and the participating members of the “Groupe de Recherche en Obstétrique et Gynécologie (GROG)”, Thomas Schmitz, Elie Azria, Catherine Deneux-tharaux, Anne Ego, Francois Goffinet, Cyril Huissoud, Gilles Kayem, Bruno Langer, Camille Le Ray, Olivier Morel, Marie-Victoire Senat and Damien Subtil.