Dose of aspirin to prevent preterm preeclampsia in women with moderate or high-risk factors: A systematic review and meta-analysis

Objective To evaluate the effect of aspirin dose on the incidence of all gestational age preeclampsia and preterm preeclampsia. Data sources Electronic databases (Cochrane, PubMed, Scopus, ClinicalTrials.gov and the Web of Science) were searched for articles published between January 1985 and March 2019 with no language restrictions. Methods We followed the PRIMSA guidelines and utilized Covidence software. Articles were screened by 2 independent reviewers, with discrepancies settled by an independent 3rd party. Study selection criteria were randomized trials comparing aspirin for prevention of all gestational age and preterm preeclampsia to placebo or no antiplatelet treatment in women aged 15–55 years with moderate or high-risk factors according to the list of risk factors from American College of Obstetricians and Gynecologists and United States Preventive Services Task Force guidelines. The quality of trials was assessed using the Cochrane risk of bias tool. The data were pooled using a random-effects meta-analysis comparing aspirin at doses of <81, 81, 100, and 150 mg. Pre-specified outcomes were all gestational age and preterm preeclampsia. Results Of 1,609 articles screened, 23 randomized trials, which included 32,370 women, fulfilled the inclusion criteria. In preterm preeclampsia, women assigned at random to 150 mg experienced a significant 62% reduction in risk of preterm preeclampsia (RR = 0.38; 95% CI: 0.20–0.72; P = 0.011). Aspirin doses <150 mg produced no significant reductions. The number needed to treat with 150 mg of aspirin was 39 (95% CI: 23–100). There was a maximum 30% reduction in risk of all gestational age preeclampsia at all aspirin doses. Conclusions In this meta-analysis, based on indirect comparisons, aspirin at a dose greater than the current, recommended 81 mg was associated with the highest reduction in preterm preeclampsia. Our meta-analysis is limited due to the deficiency of homogeneous high evidence data available in the literature to date; however, it may be prudent for clinicians to consider that the optimal aspirin dose may be higher than the current guidelines advise. Future research to compare the efficacy aspirin doses greater than 81 mg is recommended. Study registration PROSPERO, CRD42019127951 (University of York, UK; http://www.crd.york.ac.uk/PROSPERO/).


Introduction
Hypertensive disorders of pregnancy are a leading cause of maternal morbidity and mortality worldwide, accounting for more than 70,000 maternal deaths each year [1]. Of all maternal deaths, 10-15% are directly associated with preeclampsia and eclampsia [2]. Aspirin is an inexpensive and widely available drug that has the potential to safely help pregnant women and neonates. While randomized trials have investigated the prophylactic use of aspirin in preventing preeclampsia, the optimal dose remains unclear. Current guidelines endorsed by American College of Obstetricians and Gynecologists (ACOG) and United States Preventive Services Task Force (USPSTF) recommend low-dose aspirin prophylaxis in women at high risk of preeclampsia as well as for women with more than one of several moderate risk factors for preeclampsia [3,4]. More recent trials have investigated the use of aspirin at a higher dose of 150 mg [5]. In one trial, 150 mg of daily aspirin in pregnant women deemed high-risk for developing preeclampsia, significantly decreased the incidence of preterm preeclampsia when compared to placebo [5]. In a recent systematic review, the authors concluded that aspirin beginning at or before 16 weeks gestation reduces the incidence of preeclampsia and its adverse outcomes among pregnant women and neonates [6]. We, therefore, performed a systematic review and meta-analysis of 23 randomized controlled trials to explore whether aspirin is effective and, if so, the optimal dose for women with moderate or high-risk factors of preeclampsia. Our primary objective was to determine the effect of aspirin dose on the incidence of all gestational age and preterm preeclampsia.  OR "Acetylsalicylic acid" OR aspirin)) AND (pregnan � OR gestation � ). Approval from an institutional review board was not required for this review.

Study selection
We conducted a systematic review and meta-analysis of randomized controlled trials that studied the use of aspirin for the prevention of preeclampsia in women aged 15-55 years with any moderate or high-risk factors of preeclampsia according to the list of risk factors from USPSTF and ACOG guidelines. Studies that were quasi-randomized, cluster randomized, or not peerreviewed were excluded. Trials using a different treatment regimen were similarly excluded. The primary outcome was the effect of aspirin dose on the incidence of all gestational age preeclampsia and preterm preeclampsia. The titles and abstracts of the searched articles were independently screened by two reviewers (R.VD. and I.F.), with discrepancies addressed by the third reviewer (K.H.). The full text of the selected articles was individually assessed by two reviewers (R.VD. and N.M.) and any conflicts were resolved by the third reviewer (K.H.). Once the final 24 articles were selected for extraction, N.M. and R.VD. extracted relevant data using Covidence software. The software provides comparison of the two reviewers' extractions and a consensus spreadsheet which requires a specific selection of which extracted data between the two reviewers gets included into the final document. Any disagreement was resolved by the third reviewer K.H. Extracted data was principally focused on our primary outcome and if available, demographic data, adverse maternal, and neonatal outcomes were also extracted. A final collation of data was exported to Microsoft Excel and sent for statistical review and analysis. Of the 24 articles selected for data extraction, one article reported results as percentage values only. Upon contacting the author for raw data, we did not receive a response. This study was then excluded making our total 23 articles.
Each of the 23 articles was subjected to a quality and risk of bias assessment within Covidence. The Covidence software has a built-in Cochrane RoB 1.0 risk of bias assessment tool. 44 Studies were screened for the following quality measures and bias risks: sequence generation, allocation concealment, blinding of participants and personnel for all outcomes, blinding of outcome assessors for all outcomes, incomplete outcome data for all outcomes, selective outcome reporting, and other sources of bias. Each category previously listed was independently rated as "High", "Low", or "Unclear" by the two reviewers. Covidence's reports of any discrepancies were subsequently voted on for a final rating.
Risk ratios (RR) quantifying aspirin's effect relative to placebo were estimated using DerSimonian and Laird's random-effects model, as we anticipated heterogeneity (measured by the I 2 statistic) would be above 50%. Pooled RRs and supporting 95% confidence intervals (CI) were computed for overall effect of aspirin on two outcomes; all gestational age preeclampsia and preterm preeclampsia and for studies grouped by aspirin dose 1) at most 81 mg versus more than 81 mg and 2) less than 81 mg, exactly 81 mg, exactly 100 mg, and exactly 150 mg for each outcome as well. Publication bias was assessed graphically using contour-enhanced funnel plots [8]. We also conducted a risk difference analyses allowing for a simple additive reduction in risk of preeclampsia with aspirin dosing, which allowed for an estimation of the number needed to treat (NNT) to prevent one case of preeclampsia. Analyses were conducted using Stata version 15.1 (StataCorp LLC, College Station, TX) with the contributed 'metan' [9], 'metabias' [10], and 'confunnel' [11] packages.

Results
The search strategy yielded 1,609 articles. Preliminary screening by two independent reviewers (R.VD. and I.F.) excluded 24 duplicates. The remaining article titles and abstracts were screened as well as any new articles published during the time of the search. A total of 1,585 potentially eligible manuscripts were identified. Of these manuscripts, 1,423 were excluded for not being relevant to our review criteria. The abstract screening was independently performed by the same reviewers as the initial screen. Lastly, 162 articles went through full-text screening and of those, 138 were excluded for not meeting our inclusion criteria, not having reliable translations, or for being duplicates and commentaries. Any disagreements were settled by the third reviewer (K.H.). Twenty-four articles met the final requirements for inclusion and data was extracted. One of the final 24 studies was excluded for lack of adequate reported data metrics which prevented us from being able to run the proper statistical analyses on their results. One of the remaining 23 articles was only used in preterm preeclampsia analysis. The complete flow diagram of study inclusion is presented in Fig 1. Study characteristics are presented in Table 1. Our primary outcome measures were the development of preeclampsia as defined by new onset hypertension (systolic blood pressure �140 mmHg or diastolic blood pressure �90 mmHg on two occasions at least 4 hours apart) in pregnant women >20 weeks of gestation and proteinuria (�300 mg on 24-hour urine collection, or protein/creatine �0.3 mg/dL, or dipstick reading of 2+), or in absence of proteinuria, new onset hypertension accompanied by any of the following: thrombocytopenia, renal insufficiency, pulmonary edema, or impaired liver function [3] and preterm (diagnosed at <37 weeks of gestation) preeclampsia. All studies were randomized controlled trials and ranged from 30 to 9,309 total participants. Preeclampsia event was reported by 22/23 studies, however only 11/23 studies reported the specific incidence of preterm preeclampsia.
Results of the bias assessment via the Cochrane Risk of Bias Template are in S1 Fig. None of the included studies had risk of bias for "other bias"; five had risk of bias for selective reporting; eight had risk of bias for incomplete outcome data; eight had risk of bias for blinding of outcome assessment; eight had risk of bias for outcome of blinding participants and personnel; eight had risk of bias for allocation concealment; nine had risk of bias for random sequence generation.
In our primary analysis involving all gestational age preeclampsia outcome, there was noted heterogeneity among studies using at most 81 mg (I 2 = 70.5%; p<0.001) but not in the more than 81 mg group (I 2 �0.5%; p = 0.725, Fig 2). When the analysis of all gestational age preeclampsia outcome was grouped by the four aspirin dosing groups, there was heterogeneity among the studies using at most 81 mg (I 2 = 72.6%; p<0.001) but not in the more than 100 mg group (I 2 = 0.0%; p = 0.617, Fig 3). Among the 11 studies used for the preterm preeclampsia analysis there was noted heterogeneity among studies using the at most 81 mg (I 2 = 62.7%; p = 0.013) but not in the more than 100 mg group (I 2 = <0.0%; p = 0.973, Fig 4). Publication bias cannot be excluded based on the funnel plot (S2A and S2B Fig).

Effect of aspirin on all gestational age preeclampsia prevention
The meta-analysis of the 22 studies revealed that aspirin regardless of dose brought about a 27% (RR = 0.73; 95% CI: 0.61-0.87; p<0.001) reduction in risk of all gestational age preeclampsia diagnosis compared to placebo (Fig 5).
Of the final 22 studies included, 15 used at most 81 mg of aspirin, one study used exactly 81 mg (the current recommendation according to ACOG and USPSFT guidelines), and 14 other studies used between 36 and 80 mg of aspirin. Six used exactly 100 mg and one study used 150 mg, the highest studied dose for preeclampsia prevention to date. The two groups (at most 81 mg vs higher) each showed significant reductions in risk of preeclampsia due to aspirin.  Aspirin at most 81 mg demonstrated a 29% reduction (RR = 0.71; 95% CI: 0.56-0.89; p = 0.003) while at more than 81 mg, a 25% reduction (RR = 0.75; 95% CI: 0.60-0.93, p = 0.009) in preeclampsia diagnosis (Fig 2). Analyses were also conducted using the risk difference to quantify the effect of aspirin on reduction of preeclampsia (S3 Fig). On this scale, the lower dose of aspirin showed a significant reduction in preeclampsia risk (p = 0.007) while more than 81 mg doses produced a nonsignificant effect (p = 0.115). Those given at most 81 mg saw a 2.6 percentage point (ppt) reduction in risk (95% CI: 0.7-4.4 ppt reduction). This translated to having to treat 38 (95% CI: 23-143) women with lower-dose aspirin to avoid one case of preeclampsia.
The studies accounting for all gestational age preeclampsia outcomes were grouped into the four levels determined by the aspirin doses less than 81 mg, 81 mg, 100 mg, and 150 mg. There was no evidence that exactly 81 mg of aspirin had an effect (RR = 0.88; 95% CI: 0.

Effect of aspirin on prevention of preterm preeclampsia
Eleven of the 23 studies presented data of preterm preeclampsia outcome. The meta-analysis of the overall effect of aspirin on preterm preeclampsia prevention revealed that aspirin is related to a 30% (RR = 0.70; 95% CI: 0.53-0.92; p = 0.011) reduction in risk of preterm preeclampsia compared to placebo (Fig 6).
When grouped according to aspirin dose at most 81 mg (eight studies) and more that 81 mg (three studies), the analysis revealed an insignificant effect of aspirin at most 81 mg (RR = 0.78; 95% CI: 0.58-1.05; p = 0.107), while at more than 81 mg (above currently recommended according to ACOG and USPSFT guidelines) the effect of aspirin was significant and

PLOS ONE
Aspirin dose and preeclampsia prevention: Meta-analysis demonstrated 54% (RR = 0.46; 95% CI: 0.28-0.74) reduction in risk of preterm preeclampsia compared to placebo (Fig 7). Furthermore, studies were grouped according to the four levels of aspirin dose: seven at the low end (36-80 mg), one at exactly 81 mg, two at exactly 100 mg, and one at exactly 150 mg. None of the doses/dose groups that were below 150 mg showed a significant reduction in risk of preterm preeclampsia. The single  study demonstrated a convincingly significant risk reduction of 62% (RR = 0.38; 95% CI: 0.20-0.72; p = 0.011) for aspirin compared to placebo (Fig 4). These results should be interpreted with caution considering that the Rolnik study at 150 mg is only a single value providing data for statistical analysis. These same 4 groups of studies, when expressed as risk differences (S5 Fig), showed evidence of protection at the low end (36-80 mg; p = 0.045; 2.4 (95% CI: 0.1-4.6 ppt reduction) and at the highest dose (150 mg; p = 0.002; 2.6 (95% CI: 1.0-4.3 ppt reduction), but not for 81 mg or 100 mg (p = 0.855 and 0.187, respectively). The NNT for 36-80 mg is estimated to be 42 (95% CI: 22-1000) and 39 (95% CI: 23-100) for the 150 mg dose.

Discussion
All aspirin doses less than 150 mg were ineffective in reducing preterm preeclampsia; however, there was a 62% reduction in risk using 150 mg (RR = 0.38; 95% CI: 0.20-0.72; P = 0.011). The NNT using 150 mg for preterm preeclampsia prevention is 39 (95% CI: 23-100). Dividing the aspirin doses into four groups demonstrated that only less than 81 mg and the 150 mg groups had a significant reduction in risk of all gestational age preeclampsia which was approximately 28% at each end. The NNT using less than 81 mg and 150 mg was 38 (95% CI: 22-143) and 31 (95%CI: 16-333), respectively for preeclampsia prevention at all gestational ages. Again, with the 150 mg being only a single study, it is important to acknowledge that there is a need for more RCTs using 150 mg dosing to increase the statistical power of these conclusions.
Strengths of our review include the number of studies we analyzed compared to other systematic reviews performed on this topic. The last systematic review in February 2018 included 10 studies; we added 13 more randomized controlled trials [6]. We aimed to define a specific dose of aspirin to employ as prevention for preeclampsia versus the current standard of 81 mg, or the generally used and non-specific term: "low-dose aspirin". Limitations to our review include the potential risk of bias among the articles included, despite our risk assessment and quality measure of each study. We cannot confidently know that all biases were excluded in the methodologic approach to each study that we included in our review. Our percentage point (ppt) risk reduction analysis with the following NNT calculation is limited based on the heterogeneity of the pooled data and should be interpreted cautiously. Our analysis is limited to the congruency among the studies included. Differing outcome measures, cohorts of women, and how each study reported results defined how and what amount of data we could extract. Especially the difference in the gestational age of aspirin therapy initiation in the studies included into this meta-analysis needs to be acknowledged. There is limited data on the optimal aspirin dosing, specifically, of 81 mg compared to 150 mg. This warrants further randomized controlled trials to validate our findings. We recommend future conduction of randomized control trials comparing the efficacy of the daily dosage of 81 mg vs 150 mg aspirin. Analysis of the efficacy of increasing the clinically used dose of aspirin from 81 mg to 150 mg on short and long-term maternal and neonatal outcomes is needed and also recommended.
Since the first evidence of aspirin for reducing the risk of preeclampsia in 1985, many studies have attempted to determine the effect with controversial results. Aspirin dosage remains a topic of debate [12,13]; however, it is believed there is a dose dependent response.

PLOS ONE
Aspirin dose and preeclampsia prevention: Meta-analysis Currently in the United States, it is recommended in high-risk and considered in moderate-risk women to initiate aspirin prior to 16 weeks and take 81 mg daily until delivery [3,4,14]. The USPSTF states that universal aspirin prophylaxis for preeclampsia would save $365 million dollars in direct healthcare costs annually [15]. At best, we interpret all women with moderate or high-risk factors may have approximately a 30% reduction in preeclampsia if they initiate 81 mg of aspirin. When acknowledging the impact of a diagnosis of preterm preeclampsia, this reduction is not sufficient. Our analysis suggests that aspirin doses less than 150 mg may not be reducing the risk of preterm preeclampsia sufficiently. Our results indicate a dose of 150 mg may provide up to 62% reduction of preterm preeclampsia.
The effect of aspirin on COX-dependent prostaglandin synthesis is dose dependent. At lower dosages (less than 100 mg/day) aspirin irreversibly acetylates COX-1, resulting in decreased platelet synthesis of thromboxane-A2 without affecting vascular wall production of prostacyclin, which improves the thromboxane-A2/prostacyclin balance in favor of prostacyclin [12,13]. At higher doses, aspirin inhibits both COX-1 and COX-2; 1) effectively blocking all prostaglandin production and 2) blocking COX-2 decreases sensitivity to angiotensin II, decreases the activation of the immune system, and decreases oxidative stress. High dose aspirin may be the best option allowing for restoration of the angiogenic balance and improvement in vascular function associated with preeclampsia [16]. Aspirin at low doses has good maternal and fetal safety profiles; however, doses over 100 mg have not been adequately studied. Based on our study, increasing the standard dose of aspirin to 150 mg would provide a substantial reduction in preterm preeclampsia. However, future studies are needed to confirm the mechanism leading to an increased reduction of preterm preeclampsia by use of higher doses of aspirin. In addition, while a higher dosage appears to be the most effective dose for preterm preeclampsia prevention, future studies should validate we are not compromising safety of the mother and baby.
Despite this being based on a single study, our review demonstrated a significant reduction in preterm preeclampsia with 150 mg aspirin compared to all other dosages. We did not find significant differences in risk reduction of preeclampsia at 81 mg in comparison to the other dosages. Randomized trials that compare higher doses of aspirin to 81mg are necessary to complete the totality of evidence.