低剂量阿司匹林预防后早产型子痫前期的决定因素：来自亚洲阶梯楔形整群随机试验的证据。
全文（开放获取） · 中文译文
摘要 目的：旨在识别与随后发展为早产型子痫前期相关的潜在风险因素——包括母体因素、子痫前期（PE）特异性生物标志物以及基于孕早期胎儿医学基金会（FMF）三联试验得出的估计风险——并采用时间-事件分析评估任意发病型子痫前期的预测因素，研究对象为接受低剂量阿司匹林预防治疗的高危女性。 方法：这是一项多中心阶梯楔形整群随机试验的二次分析，纳入了2019年8月至2022年2月期间来自亚洲10个地区的18个产科/诊断中心。研究人群包括根据孕早期FMF三联试验被确定为早产型子痫前期高风险（估计风险≥1:100）的单胎妊娠女性，这些女性随后接受了低剂量阿司匹林预防治疗。首先采用局部估计散点图平滑（LOESS）曲线探索早产型子痫前期发病率与估计风险之间的关系。根据LOESS曲线观察到的趋势，参与者被分为五个估计风险组：1:2至1:5、1:6至1:10、1:11至1:20、1:21至1:50以及1:51至1:100。分别采用Logistic回归和Cox回归分析来识别早产型子痫前期的独立预测因素，并评估发生子痫前期分娩的风险。 结果：在接受阿司匹林预防治疗的2919名高危孕妇中，280名（9.6%）发生了子痫前期（PE），其中107名（3.7%）发生了早产型子痫前期。慢性高血压（校正比值比（aOR），1.835（95% CI，1.070-3.146））、平均动脉压（MAP）中位数倍数（MoM）> 1.149（aOR，2.162（95% CI，1.434-3.261））以及更高估计风险与更高的早产型子痫前期发生几率相关。与估计风险为1：51至1：100的孕妇相比，估计风险为1：2至1：5、1：6至1：10和1：11至1：20的孕妇发生早产型子痫前期的几率分别高出16倍（aOR，15.845（95% CI，8.166-30.746））、8倍（aOR，7.674（95% CI，3.925-15.004））和3倍（aOR，3.454（95% CI，1.802-6.622）），而估计风险为1：21至1：50的孕妇则未观察到显著差异（aOR，1.343（95% CI，0.722-2.499））。在Cox回归分析中，慢性高血压与分娩时发生子痫前期的风险增加3倍相关（校正风险比（aHR），2.979（95% CI，2.154-4.120）），而MAP MoM > 1.149使风险增加66%（aHR，1.659（95% CI，1.304-2.111））。与参考组（估计风险为1：51至1：100）相比，分娩时发生子痫前期的风险在各估计风险组中逐渐增加，1：21至1：50、1：11至1：20、1：6至1：10和1：2至1：5估计风险组的aHR分别为1.435（95% CI，1.051-1.959）、2.408（95% CI，1.653-3.508）、4.058（95% CI，2.658-6.197）和7.053（95% CI，4.588-10.841）。 结论：慢性高血压、平均动脉压升高及估计风险值增高，与经孕早期FMF三联检测确定为高风险女性中先兆子痫（PE）的较早发生相关，尽管这些女性使用了低剂量阿司匹林预防。将这些因素纳入风险分层框架，可能有助于改善孕早期筛查和预防策略中残余风险的检出，从而支持制定更有效的预防措施。© 2026作者。Ultrasound in Obstetrics & Gynecology由John Wiley & Sons Ltd代表国际妇产科超声学会（International Society of Ultrasound in Obstetrics and Gynecology）出版。 引言 低剂量阿司匹林被国际专业组织广泛推荐，作为对有早产性子痫前期（PE）高风险孕妇的有效预防性干预措施，，，。然而，并非所有高风险女性都能从该策略中获得同等益处。联合多标志物筛查与阿司匹林随机治疗用于循证子痫前期预防（ASPRE）试验，纳入了混合的欧洲人群，报告称，在使用胎儿医学基金会（FMF）联合检测（包括母体因素、平均动脉压（MAP）、子宫动脉搏动指数（UtA‐PI）、妊娠相关血浆蛋白-A 和胎盘生长因子（PlGF））被确定为早产性 PE 高风险的女性中，每日 150 mg 阿司匹林预防治疗与早产性 PE 发病率降低 62% 相关（比值比（OR），0.38（95% CI，0.20–0.74））。ASPRE 试验的一项二次分析提示，使用 FMF 联合检测获得的极高风险结果（估计风险 ≥ 1:50）、慢性高血压和低 PlGF 可能与阿司匹林反应性降低有关，。在亚洲，我们的研究小组进行了一项多中心阶梯楔形整群随机试验（妊娠早期筛查与子痫前期预防实施试验（FORECAST）），该试验证明，在使用妊娠早期 FMF 三联检测（包括母体因素、MAP、UtA‐PI 和 PlGF）被确定为高风险的女性中，阿司匹林预防治疗与早产性 PE 发病率降低 41% 相关（校正后比值比（aOR），0.59（95% CI，0.37–0.92））。 tors、MAP、UtA‐PI 和 PlGF），，。在 FORECAST 和 ASPRE 试验中观察到的低剂量阿司匹林预防效果差异表明，个体对阿司匹林的反应可能存在地区性或生物学差异。除慢性高血压和低 PlGF 外，其他因素——如体重指数升高、MAP 升高以及对阿司匹林预防的依从性不佳——也已被报道可导致阿司匹林无反应性，，，。 这些研究之间的不一致之处表明，需要进一步调查在使用FMF三联检测被判定为高风险并接受低剂量阿司匹林预防的女性中，发生早产型子痫前期（PE）的残余风险。因此，这项对FORECAST研究的二次分析旨在识别潜在的危险因素——包括母体因素、PE特异性生物标志物以及由FMF三联检测得出的估计风险——这些因素与尽管使用低剂量阿司匹林预防仍随后发生早产型PE相关，并通过时间-事件分析评估任意发病型PE的预测因素。 方法 研究设计与参与者 这是对FORECAST研究（NCT03941886）的一项二次分析，涉及在干预期间接受阿司匹林预防治疗的高危女性。2019年8月1日至2022年2月28日期间，在亚洲10个地区（香港特别行政区、中国大陆、台湾、新加坡、泰国、菲律宾、印度尼西亚、马来西亚、越南、日本）的18个产科/诊断单位进行了使用FMF三联检测的早产子痫前期孕早期筛查。根据试验人群的同质性，这些产科/诊断单位被分为七个集群。阶梯楔形设计首先从一个时期开始，在该时期，所有招募中心均提供常规产前保健，不进行与研究相关的干预。随后，每隔6周，一个集群被随机分配从非干预阶段过渡到干预阶段。在干预阶段，使用FMF三联检测进行早产子痫前期的孕早期筛查。对于估计早产子痫前期风险≥1:100的女性，在孕16周前开始低剂量阿司匹林预防治疗，持续至孕36周或直至分娩或在36周前发生子痫前期。所有纳入FORECAST研究的女性均签署了书面知情同意书以参与研究。纳入标准为孕11+0至13+6周的单胎妊娠且 maternal age≥18岁。排除标准为多胎妊娠、在11+0至13+6周评估时发现重大胎儿缺陷、胎儿无法存活（错过性自然流产或死产）或接受其他临床…… 临床干预。试验获得了香港中文大学–新界东联网联合临床研究伦理委员会的批准（CREC 参考编号：2018.391），此外还获得了其他地区各参与医院伦理委员会的批准。 FMF三联检测与阿司匹林预防 FMF三联检测是一种贝叶斯算法，结合母体因素（包括母体特征和产科病史）、MAP、UtA‐PI和PlGF来估算早产型PE的个体风险。孕龄通过测量胎儿头臀长确定。记录母体特征及内科和产科病史。MAP按照标准化方案使用经过验证的自动设备（BP3AQ1；Microlife，台北，台湾）测量。采用经腹彩色多普勒超声成像识别UtA，并使用脉冲波多普勒超声测量左右UtA‐PI，记录平均值。所有执行多普勒检查的操作者均已获得FMF颁发的相应能力证书（https://fetalmedicine.org/）。血清PlGF浓度使用以下自动免疫分析仪之一测量：AutoDELFIA或DELFIA Xpress系统（PlGF 1‐2‐3试剂盒；Revvity Inc.，Waltham，MA，USA）、BRAHMS KRYPTOR分析仪（ThermoFisher Scientific，Hennigsdorf，Germany）或Cobas e411系统（Roche Diagnostics，Rotkreuz，Switzerland）。使用已发表的公式，将MAP、UtA‐PI和PlGF的测量值转换为中位数倍数（MoM），并针对母体特征、既往和当前产科病史以及所使用的免疫分析仪（就PlGF而言）进行调整。在每个站点，通过应用站点特异性生物标志物校正因子，将PlGF MoM和MAP MoM的分布重新中心化，使其中位数为1 MoM。 体重＜40 kg的高危女性给予阿司匹林每日100 mg，而体重≥40 kg者则根据各招募中心可获得的制剂，给予阿司匹林150 mg、160 mg或162 mg。阿司匹林依从性以实际服用总片数与处方总片数的比值表示，并以百分比形式报告。 妊娠结局 主要结局为早产子痫前期（PE），定义为妊娠＜37周分娩的PE。根据国际妊娠期高血压研究学会（ISSHP）指南，PE定义为既往血压正常的女性，间隔4小时至少两次测量收缩压≥140 mmHg和/或舒张压≥90 mmHg，并在20周或之后伴有以下一种或多种新发情况：蛋白尿；其他母体器官功能障碍的证据，如急性肾损伤、肝脏受累、肝酶升高、神经系统并发症或血液系统并发症；或子宫胎盘功能障碍。慢性高血压叠加PE定义为患有慢性原发性高血压的女性出现上述任何与PE一致的母体器官功能障碍。 统计分析 连续数据的正态性采用Kolmogorov–Smirnov检验进行评估。呈非正态分布的连续变量以中位数（四分位距（IQR））表示，并采用Kruskal–Wallis检验进行比较。分类变量以n（%）表示，并采用卡方检验进行比较。采用局部估计散点图平滑（LOESS）法，对早产子痫前期（PE）发病率在不同估计风险水平下的趋势进行探索性可视化。随后将估计风险分层为若干组。 采用单因素logistic回归分析，在母体因素（产妇年龄、身高、体重、受孕方式、受孕时吸烟状况、种族、慢性高血压、系统性红斑狼疮/抗磷脂综合征、1型糖尿病、2型糖尿病、产次、既往子痫前期病史、子痫前期家族史）、子痫前期特异性生物标志物（MAP MoM、UtA‐PI MoM、PlGF MoM）、基于孕早期FMF三联试验得出的估计风险以及阿司匹林依从性中，识别早产性子痫前期的潜在预测因素。将P值< 0.1的变量纳入多因素logistic回归分析，采用向后逐步选择法识别独立预测因素，最终模型中仅保留P < 0.05的变量。采用方差膨胀因子（VIF）评估自变量之间的多重共线性。分别使用单因素和多因素logistic回归模型估计候选变量与早产性子痫前期之间关联的OR值和aOR值，并附95% CI。通过受试者工作特征（ROC）曲线分析，使用最大约登指数确定MAP MoM的最佳截断值。采用LOESS曲线比较慢性高血压女性与非慢性高血压女性之间，以及MAP MoM高于最佳截断值组与低于最佳截断值组之间，早产性子痫前期发病率随估计风险增加的变化模式。早产性子痫前期的预测概率来自最终多因素模型，针对临床相关的预测因素组合进行计算。在亚组中。 在分组分析中（针对估计风险组≥1:10与<1:10，以及患有与未患有慢性高血压的女性），仅将整体分析中确定为显著的预测因子纳入多变量回归模型。检验了产次和PE病史与关键预测因子的交互作用。描述了FORECAST研究及本二次分析中各聚类中产次和PE病史的分布。进行了敏感性分析以考虑聚类效应。 采用Cox比例风险模型进行时间-事件分析，以估计子痫前期（PE）分娩的调整后风险比（aHR）及其95%置信区间（CI）。使用Schoenfeld残差检验评估比例风险假设。 统计分析使用IBM SPSS 29.0版（IBM Corp.，Armonk，NY，USA）进行。双侧P值<0.05被认为具有统计学显著性。 结果 研究人群 总体而言，88.0%（42 897/48 725）符合条件的女性同意接受早产子痫前期的一期筛查。经过排除后，非干预阶段有10 440/38 484（27.1%）名女性，干预阶段有28 044/38 484（72.9%）名女性。在干预阶段，3543/28 044（12.6%）名女性被确定为早产子痫前期高风险，其中82.4%（2919/3543）接受了阿司匹林预防治疗，而其余女性因各种原因未服用阿司匹林。在干预阶段，观察到191/28 044（0.7%）名女性发生早产子痫前期，而在非干预阶段，79/10 440（0.8%）名女性发生早产子痫前期。在本二次分析中，纳入了2919名被确定为早产子痫前期高风险且接受阿司匹林预防治疗的孕妇。其中，280名（9.6%）发展为子痫前期（PE），包括107例（3.7%）早产子痫前期。未产妇女占FORECAST总队列的55.4%，占本次二次研究的70.9%（表格和）。在各聚类中，聚类2在FORECAST总队列和本次分析中均具有最高的未产妇女比例。 应用LOESS曲线探讨早产子痫前期（PE）发生率与孕早期FMF三联检测得出的早产PE估计风险之间的关系（图）。随着估计风险的升高，发生率随之上升，但当估计风险低于1:50时，曲线几乎保持平坦。基于LOESS曲线观察到的趋势，将估计风险分为若干组（1:2至1:5、1:6至1:10、1:11至1:20、1:21至1:50及1:51至1:100），以便比较各风险分层间早产PE的发生率。 局部估计散点图平滑（LOESS）平滑后的早产子痫前期（PE）发生率（），附95%置信区间（灰色阴影），根据孕早期胎儿医学基金会三联检测得出的早产PE估计风险绘制。直方图显示参与者在不同估计风险水平上的分布（，早产PE；，无早产PE）。 基线母体特征及妊娠结局见表。阿司匹林依从率中位数为99.3%（IQR，96.1–100%），各风险组之间差异无统计学意义（P = 0.721）。在妊娠结局方面，各风险组在任意发病PE及PE亚型的发生率、分娩时孕周及出生体重方面均观察到显著差异（均P < 0.001）。 2919名经孕早期胎儿医学基金会（FMF）三联检测确定为早产子痫前期（PE）高风险并接受阿司匹林预防治疗的女性，其基线母体特征及妊娠结局，按估计风险分组列出。 数据以中位数（四分位间距）或n（%）表示。组间比较采用卡方检验用于分类变量，Kruskal–Wallis检验用于连续变量。对于母体特征、病史或作为FMF风险评分的确定性成分的生物标志物，未计算P值，因为这些变量参与了估计风险组的定义，对其分布的比较不具有有意义的统计学推断意义。APS，抗磷脂综合征；CRL，头臀长；GA，孕龄；MAP，平均动脉压；MoM，中位数倍数；PlGF，胎盘生长因子；SLE，系统性红斑狼疮；UtA‐PI，子宫动脉搏动指数。 Logistic回归分析 所有变量的方差膨胀因子（VIF）值均＜2，表明不存在显著的多重共线性。在多变量逻辑回归模型（表）中，慢性高血压与早产性子痫前期（PE）风险增加独立相关，调整后优势比（aOR）为1.922（95% CI，1.118–3.304）（P = 0.018）。平均动脉压倍数中位数（MAP MoM）与早产性PE呈强相关（aOR，37.279（95% CI，6.770–205.287））（P < 0.001）。与估计风险最低（1:51至1:100）的孕妇相比，早产性PE的aOR如下：估计风险为1:2至1:5者，aOR为14.337（95% CI，7.325–28.058）（P < 0.001）；估计风险为1:6至1:10者，aOR为7.003（95% CI，3.561–13.770）（P < 0.001）；估计风险为1:11至1:20者，aOR为3.289（95% CI，1.712–6.318）（P < 0.001）；估计风险为1:21至1:50者，aOR为1.329（95% CI，0.715–2.473）（P = 0.369）。 对2919例经孕早期胎儿医学基金会（Fetal Medicine Foundation）三联筛查确定为高风险并接受阿司匹林预防治疗的孕妇中与早产性子痫前期（PE）相关因素进行逻辑回归分析 除非另有说明，数据以中位数（四分位距）或n（%）表示。多变量方程：logit(p) = −8.23703 + (0.65353 × (若为慢性高血压则为1；否则为0)) + (3.61842 × MAP MoM) + (1.19049 × (若估计风险为1:11至1:20则为1；否则为0)) + (1.94631 × (若估计风险为1:6至1:10则为1；否则为0)) + (2.66281 × (若估计风险为1:2至1:5则为1；否则为0)) (R2 = 0.186)。对于连续变量，比值比（ORs）和校正比值比（aORs）按预测变量每增加1个单位报告。APS，抗磷脂综合征；MAP，平均动脉压；MoM，中位数的倍数；PlGF，胎盘生长因子；ref，参考；SE，标准误；SLE，系统性红斑狼疮；UtA‐PI，子宫动脉搏动指数。 采用ROC曲线分析，MAP MoM的最佳截断值确定为1.149，高于该值时早产子痫前期的风险显著增加（校正后优势比[aOR]，2.162（95% CI，1.434–3.261））（P < 0.001）（见表）。当MAP作为分类变量以最佳截断值纳入logistic回归模型时，慢性高血压和估计风险与早产子痫前期的关联基本保持不变。慢性高血压仍与早产子痫前期风险增加独立相关（aOR，1.835（95% CI，1.070–3.146））（P = 0.027）。同样，与最低风险组（1:51至1:100）的孕妇相比，估计风险分层逐级升高与早产子痫前期风险增加相关，其中1:11至1:20风险组的aOR为3.454（95% CI，1.802–6.622）（P < 0.001），1:6至1:10风险组的aOR为7.674（95% CI，3.925–15.004）（P < 0.001），1:2至1:5风险组的aOR为15.845（95% CI，8.166–30.746）（P < 0.001）。相比之下，1:21至1:50风险组与早产子痫前期无显著相关性（aOR，1.343（95% CI，0.722–2.499））（P = 0.352）。表列出了早产子痫前期的预测绝对概率，使临床医生能够解读高危女性在阿司匹林预防后的个体风险。未发现产次和既往子痫前期史与关键预测因子之间存在任何显著交互作用（见表）。敏感性分析（包括排除Cluster 2的参与者以及将cluster作为混杂因素进行校正）显示结果与主要分析一致，统计学显著性基本保持不变（见表和表）。 ）。 最终多变量逻辑回归模型，包含对2919名经孕早期胎儿医学基金会三重检测确定为高风险并接受阿司匹林预防治疗的女性中早产子痫前期（PE）的关键预测因子 多变量方程：logit(p) = −4.47196 + (0.60717 × （若合并慢性高血压则为1；否则为0）) + (0.77115 × （若MAP MoM > 1.149则为1；否则为0）) + (1.23957 × （若估计风险为1:11至1:20则为1；否则为0）) + (2.03789 × （若估计风险为1:6至1:10则为1；否则为0）)
全文（开放获取） · 英文原文
【Abstract】 OBJECTIVES: To identify potential risk factors - among maternal factors, pre-eclampsia (PE)-specific biomarkers and estimated risks derived from the first-trimester Fetal Medicine Foundation (FMF) triple test - that are associated with the subsequent development of preterm PE, and to evaluate predictors of any-onset PE using time-to-event analysis, in high-risk women receiving low-dose aspirin prophylaxis. METHODS: This was a secondary analysis of a multicenter stepped-wedge cluster randomized trial, including 18 maternity/diagnostic units from 10 regions in Asia between August 2019 and February 2022. The study population comprised women with a singleton pregnancy identified as being at high risk for preterm PE (estimated risk ≥ 1:100) according to the first-trimester FMF triple test, who subsequently received low-dose aspirin prophylaxis. A locally estimated scatterplot smoothing (LOESS) curve was applied initially to explore the relationship between the incidence of preterm PE and the estimated risk. Based on the trend observed in the LOESS curve, participants were categorized into five estimated risk groups: 1:2 to 1:5, 1:6 to 1:10, 1:11 to 1:20, 1:21 to 1:50 and 1:51 to 1:100. Logistic and Cox regression analyses were performed to identify independent predictors of preterm PE and to evaluate the risk of delivery with PE, respectively. RESULTS: Among 2919 high-risk pregnant women receiving aspirin prophylaxis, 280 (9.6%) developed PE, including 107 (3.7%) who developed preterm PE. Chronic hypertension (adjusted odds ratio (aOR), 1.835 (95% CI, 1.070-3.146)), mean arterial pressure (MAP) multiples of the median (MoM) > 1.149 (aOR, 2.162 (95% CI, 1.434-3.261)) and higher estimated risk were associated with higher odds of preterm PE. Compared to women with an estimated risk of 1:51 to 1:100, those with estimated risks of 1:2 to 1:5, 1:6 to 1:10 and 1:11 to 1:20 had a 16-fold (aOR, 15.845 (95% CI, 8.166-30.746)), eight-fold (aOR, 7.674 (95% CI, 3.925-15.004)) and three-fold (aOR, 3.454 (95% CI, 1.802-6.622)) higher odds of developing preterm PE, respectively, whereas no significant difference was observed in those with an estimated risk of 1:21 to 1:50 (aOR, 1.343 (95% CI, 0.722-2.499)). On Cox regression analysis, chronic hypertension was associated with a three-fold higher risk of delivering with PE (adjusted hazard ratio (aHR), 2.979 (95% CI, 2.154-4.120)), while MAP MoM > 1.149 increased the risk by 66% (aHR, 1.659 (95% CI, 1.304-2.111)). Compared with the reference group (estimated risk of 1:51 to 1:100), the risk of delivering with PE increased progressively across estimated risk groups, with aHRs of 1.435 (95% CI, 1.051-1.959), 2.408 (95% CI, 1.653-3.508), 4.058 (95% CI, 2.658-6.197) and 7.053 (95% CI, 4.588-10.841) for the 1:21 to 1:50, 1:11 to 1:20, 1:6 to 1:10, and 1:2 to 1:5 estimated risk groups, respectively. CONCLUSIONS: Chronic hypertension, elevated MAP and higher estimated risk are associated with the earlier development of PE in women identified as high risk by the first-trimester FMF triple test, despite low-dose aspirin prophylaxis. Incorporating these factors into the risk stratification framework may improve the detection of residual risk in first-trimester screening and prevention strategies, thereby supporting the development of more effective preventive measures. © 2026 The Author(s). Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology. 【INTRODUCTION】 Low‐dose aspirin is recommended widely by international professional organizations as an effective prophylactic intervention for pregnant women at high risk of preterm pre‐eclampsia (PE), , , . However, not all high‐risk women gain the same benefit from this strategy. The Combined Multimarker Screening and Randomised Patient Treatment with Aspirin for Evidence‐Based Preeclampsia Prevention (ASPRE) trial, which involved a mixed European population, reported that, among women identified as being at high risk for preterm PE using the Fetal Medicine Foundation (FMF) combined test (incorporating maternal factors, mean arterial pressure (MAP), uterine artery pulsatility index (UtA‐PI), pregnancy‐associated plasma protein‐A and placental growth factor (PlGF)), aspirin prophylaxis at 150 mg daily was associated with a 62% reduction (odds ratio (OR), 0.38 (95% CI, 0.20–0.74)) in the incidence of preterm PE. A secondary analysis of the ASPRE trial suggested that a very‐high‐risk result obtained using the FMF combined test (estimated risk ≥ 1:50), chronic hypertension and low PlGF may be associated with reduced responsiveness to aspirin, . In Asia, our research group conducted a multicenter stepped‐wedge cluster randomized trial (Implementation of First‐Trimester Screening and Prevention of Preeclampsia Trial (FORECAST)), which demonstrated that aspirin prophylaxis was associated with a 41% reduction (adjusted OR (aOR), 0.59 (95% CI, 0.37–0.92)) in the incidence of preterm PE among women identified as high risk using the first‐trimester FMF triple test (which incorporates maternal factors, MAP, UtA‐PI and PlGF), , . The observed discrepancy in the effectiveness of low‐dose aspirin prophylaxis between the FORECAST and ASPRE trials suggests potential regional or biological differences in how individuals respond to aspirin. Besides chronic hypertension and low PlGF, other factors – such as increased body mass index, increased MAP and suboptimal adherence to aspirin prophylaxis – have also been reported to contribute to aspirin non‐responsiveness, , , . These inconsistencies across studies call for further investigation into the residual risk for developing preterm PE among women identified as high risk using the FMF triple test who receive low‐dose aspirin prophylaxis. Therefore, this secondary analysis of the FORECAST study aimed to identify potential risk factors – among maternal factors, PE‐specific biomarkers and estimated risks derived from the FMF triple test – that were associated with the subsequent development of preterm PE despite low‐dose aspirin prophylaxis, and to assess the predictors of any‐onset PE using time‐to‐event analysis. 【METHODS】 【Study design and participants】 This was a secondary analysis of the FORECAST study (NCT03941886), involving high‐risk women receiving aspirin prophylaxis during the intervention period. First‐trimester screening for preterm PE using the FMF triple test was performed in 18 maternity/diagnostic units across 10 regions in Asia (Hong Kong SAR, Mainland China, Taiwan, Singapore, Thailand, The Philippines, Indonesia, Malaysia, Vietnam, Japan) between 1 August 2019 and 28 February 2022. The maternity/diagnostic units were organized into seven clusters based on the homogeneity of the trial populations. The stepped‐wedge design started with a period during which routine antenatal care was provided without study‐related intervention at all recruiting centers. Subsequently, at regular 6‐week intervals, one cluster was randomized to transition from the non‐intervention phase to the intervention phase. During the intervention phase, first‐trimester screening for preterm PE using the FMF triple test was performed. For women with an estimated risk for preterm PE of ≥ 1:100, low‐dose aspirin prophylaxis was initiated before 16 weeks' gestation, until 36 weeks or until delivery or the onset of PE before 36 weeks. All women included in the FORECAST study provided written informed consent to participate. Inclusion criteria were a singleton pregnancy at 11 + 0 to 13 + 6 weeks' gestation and maternal age ≥ 18 years. Exclusion criteria were multiple pregnancy, major fetal defect identified at the 11 + 0 to 13 + 6‐week assessment, non‐viable fetus (missed spontaneous miscarriage or stillbirth) or pregnancy receiving other clinical intervention. Approval for the trial was obtained from the Joint Chinese University of Hong Kong–New Territories East Cluster Clinical Research Ethics Committee (CREC Ref. No. 2018.391) in Hong Kong, in addition to the ethics committee of each participating hospital in the other regions. 【FMF triple test and aspirin prophylaxis】 The FMF triple test is a Bayesian algorithm that combines maternal factors (including maternal characteristics and obstetric history), MAP, UtA‐PI and PlGF to estimate the individual risk of preterm PE, . Gestational age was determined from the measurement of fetal crown–rump length. Maternal characteristics and medical and obstetric history were recorded. MAP was measured using a validated automated device (BP3AQ1; Microlife, Taipei, Taiwan) according to a standardized protocol. Transabdominal color Doppler ultrasound imaging was used to identify the UtAs and pulsed‐wave Doppler ultrasound was used to measure left and right UtA‐PI, with the average value recorded. All operators performing Doppler studies had received the appropriate Certificate of Competence from the FMF (https://fetalmedicine.org/). Serum PlGF concentrations were measured using one of the following automated immunoanalyzers: AutoDELFIA or DELFIA Xpress system (PlGF 1‐2‐3 kits; Revvity Inc., Waltham, MA, USA), BRAHMS KRYPTOR analyzer (ThermoFisher Scientific, Hennigsdorf, Germany) or Cobas e411 system (Roche Diagnostics, Rotkreuz, Switzerland). Measured values of MAP, UtA‐PI and PlGF were converted into multiples of the median (MoM), adjusting for maternal characteristics, past and current obstetric history and immunoanalyzer utilized (in the case of PlGF), using published formulae, , , . At each site, the distributions of PlGF MoM and MAP MoM were recentered to have a median of 1 MoM by applying a site‐specific biomarker correction factor. High‐risk women weighing < 40 kg were offered aspirin at a daily dose of 100 mg, while those weighing ≥ 40 kg were offered aspirin at doses of 150 mg, 160 mg or 162 mg, according to the preparations available at each recruiting center. Aspirin compliance was reported as the ratio of the total number of tablets consumed to the total number of tablets prescribed, expressed as a percentage. 【Pregnancy outcome】 The primary outcome was preterm PE, defined as PE with delivery < 37 weeks' gestation. According to the guideline of the International Society for the Study of Hypertension in Pregnancy (ISSHP), PE was defined as systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg on at least two occasions measured 4 h apart in a previously normotensive woman, accompanied by one or more of the following new‐onset conditions at or after 20 weeks: proteinuria; evidence of other maternal organ dysfunction, such as acute kidney injury, liver involvement, elevated liver enzymes, neurological complications or hematological complications; or uteroplacental dysfunction. PE superimposed on chronic hypertension was defined when a woman with chronic essential hypertension developed any abovementioned maternal organ dysfunction consistent with PE. 【Statistical analysis】 The normality of continuous data was assessed using the Kolmogorov–Smirnov test. Continuous non‐normally distributed variables are presented as median (interquartile range (IQR)) and were compared using the Kruskal–Wallis test. Categorical variables are presented as n (%) and were compared using the chi‐square test. Locally estimated scatterplot smoothing (LOESS) was applied for the exploratory visualization of trends observed in the incidence of preterm PE across estimated risk levels. Estimated risks were subsequently stratified into groups. Univariable logistic analysis was performed to identify potential predictors of preterm PE among maternal factors (maternal age, height, weight, mode of conception, smoking status at conception, ethnicity, chronic hypertension, systemic lupus erythematosus/antiphospholipid syndrome, Type‐1 diabetes mellitus, Type‐2 diabetes mellitus, parity, personal history of PE, family history of PE), PE‐specific biomarkers (MAP MoM, UtA‐PI MoM, PlGF MoM), estimated risks derived from the first‐trimester FMF triple test and aspirin compliance. Variables with a P‐value of < 0.1 were included in the multivariable logistic regression analysis to identify independent predictors using backward stepwise selection, retaining only those with P < 0.05 in the final model. Multicollinearity among independent variables was assessed using the variance inflation factor (VIF). Univariable and multivariable logistic regression models were used to estimate ORs and aORs, respectively, with 95% CI, for the associations between candidate variables and preterm PE. The optimal cut‐off for MAP MoM was determined by receiver‐operating‐characteristics (ROC)‐curve analysis using the maximum Youden index. LOESS curves were used to compare the patterns of preterm PE incidence with increasing estimated risk between women with vs those without chronic hypertension, and between women with MAP MoM above the optimal cut‐off vs those with MAP MoM below the optimal cut‐off. Predicted probabilities of preterm PE were derived from the final multivariable model for clinically relevant combinations of predictors. In the subgroup analyses (for estimated risk groups of ≥ 1:10 vs < 1:10, and for women with vs those without chronic hypertension), only the predictors identified as significant in the overall analysis were included in the multivariable regression models. The interaction of parity and PE history with key predictors was tested. The distribution of parity and PE history across clusters in the FORECAST study and this secondary analysis were described. Sensitivity analyses were conducted to account for clustering effects. Time‐to‐event analysis was conducted using the Cox proportional hazards model to estimate adjusted hazard ratios (aHRs) with 95% CI for delivery with PE. The proportional hazards assumption was evaluated using the Schoenfeld residuals test. Statistical analysis was performed using IBM SPSS version 29.0 (IBM Corp., Armonk, NY, USA). A two‐tailed P‐value of < 0.05 was considered statistically significant. 【RESULTS】 【Study population】 Overall, 88.0% (42 897/48 725) of eligible women agreed to undergo first‐trimester screening for preterm PE. After exclusions, there were 10 440/38 484 (27.1%) women in the non‐intervention phase and 28 044/38 484 (72.9%) women in the intervention phase. In the intervention phase, 3543/28 044 (12.6%) women were identified as high risk for preterm PE, of whom 82.4% (2919/3543) received aspirin prophylaxis, whereas the others did not take aspirin for various reasons. In the intervention phase, preterm PE was observed in 191/28 044 (0.7%) women, while in the non‐intervention phase, preterm PE occurred in 79/10 440 (0.8%) women. In this secondary analysis, 2919 pregnant individuals identified as high risk for preterm PE who received aspirin prophylaxis were included. Of those, 280 (9.6%) developed PE, including 107 (3.7%) cases of preterm PE. Nulliparous women accounted for 55.4% of the total FORECAST cohort and 70.9% of the present secondary study (Tables and ). Among the clusters, Cluster 2 had the highest proportion of nulliparous women in both the overall FORECAST cohort and the present analysis. A LOESS curve was applied to explore the association between the incidence of preterm PE and the estimated risk of preterm PE derived from the first‐trimester FMF triple test (Figure ). The incidence rose with increasing estimated risk, although the curve remained nearly flat when the estimated risk was below 1:50. Based on the trend observed in the LOESS curve, the estimated risks were stratified into groups (1:2 to 1:5, 1:6 to 1:10, 1:11 to 1:20, 1:21 to 1:50 and 1:51 to 1:100) to facilitate comparison of the incidence of preterm PE across risk strata. Locally estimated scatterplot smoothing (LOESS)‐smoothed incidence of preterm pre‐eclampsia (PE) () with 95% CI (gray shading), according to estimated risk of preterm PE derived from first‐trimester Fetal Medicine Foundation triple test. Histogram shows distribution of participants across estimated risk levels (, preterm PE; , no preterm PE). Baseline maternal characteristics and pregnancy outcomes are presented in Table . The median aspirin compliance rate was 99.3% (IQR, 96.1–100%) and did not differ significantly between risk groups (P = 0.721). Regarding pregnancy outcomes, significant differences were observed between risk groups in the incidence of any‐onset PE and PE subgroups, as well as in gestational age at delivery and birth weight (all P < 0.001). Baseline maternal characteristics and pregnancy outcome in 2919 women identified as high risk for preterm pre‐eclampsia (PE) by first‐trimester Fetal Medicine Foundation (FMF) triple test who received aspirin prophylaxis, according to estimated risk group Data are given as median (interquartile range) or n (%). Comparison between groups was performed using chi‐square test for categorical variables and Kruskal–Wallis test for continuous variables. P‐values were not calculated for maternal characteristics, medical history or biomarkers that are deterministic components of the FMF risk score, as these variables contributed to the definition of estimated risk groups and comparison of their distributions would not provide meaningful statistical inference. APS, antiphospholipid syndrome; CRL, crown–rump length; GA, gestational age; MAP, mean arterial pressure; MoM, multiples of the median; PlGF, placental growth factor; SLE, systemic lupus erythematosus; UtA‐PI, uterine artery pulsatility index. 【Logistic regression analysis】 All variables had VIF values of < 2, indicating no significant multicollinearity. In the multivariable logistic regression model (Table ), chronic hypertension was associated independently with increased odds of preterm PE, with an aOR of 1.922 (95% CI, 1.118–3.304) (P = 0.018). MAP MoM demonstrated a strong association with preterm PE (aOR, 37.279 (95% CI, 6.770–205.287)) (P < 0.001). Compared with women with the lowest estimated risk (1:51 to 1:100), the aORs for preterm PE were as follows: 14.337 (95% CI, 7.325–28.058) (P < 0.001) for those with an estimated risk of 1:2 to 1:5, 7.003 (95% CI, 3.561–13.770) (P < 0.001) for those with an estimated risk of 1:6 to 1:10, 3.289 (95% CI, 1.712–6.318) (P < 0.001) for those with an estimated risk of 1:11 to 1:20 and 1.329 (95% CI, 0.715–2.473) (P = 0.369) for those with an estimated risk of 1:21 to 1:50. Logistic regression analysis of factors associated with preterm pre‐eclampsia (PE) among 2919 women identified as high risk by first‐trimester Fetal Medicine Foundation triple test who received aspirin prophylaxis Data are given as median (interquartile range) or n (%), unless indicated otherwise. Multivariable equation: logit(p) = −8.23703 + (0.65353 × (1 if chronic hypertension; else 0)) + (3.61842 × MAP MoM) + (1.19049 × (1 if estimated risk 1:11 to 1:20; else 0)) + (1.94631 × (1 if estimated risk 1:6 to 1:10; else 0)) + (2.66281 × (1 if estimated risk 1:2 to 1:5; else 0)) (R2 = 0.186). For continuous variables, odds ratios (ORs) and adjusted ORs (aORs) are reported per 1‐unit increase in predictor variable. APS, antiphospholipid syndrome; MAP, mean arterial pressure; MoM, multiples of the median; PlGF, placental growth factor; ref, reference; SE, standard error; SLE, systemic lupus erythematosus; UtA‐PI, uterine artery pulsatility index. Using ROC‐curve analysis, the optimal cut‐off for MAP MoM was identified as 1.149, above which the odds of preterm PE were significantly increased (aOR, 2.162 (95% CI, 1.434–3.261)) (P < 0.001) (Table ). When MAP was included in the logistic regression model as a categorical variable using the optimal cut‐off, the associations of chronic hypertension and estimated risk with preterm PE remained largely unchanged. Chronic hypertension remained associated independently with increased odds of preterm PE (aOR, 1.835 (95% CI, 1.070–3.146)) (P = 0.027). Similarly, compared with women in the lowest risk group (1:51 to 1:100), progressively higher estimated risk strata were associated with increasing odds of preterm PE, with aORs of 3.454 (95% CI, 1.802–6.622) (P < 0.001) for the 1:11 to 1:20 risk group, 7.674 (95% CI, 3.925–15.004) (P < 0.001) for the 1:6 to 1:10 risk group and 15.845 (95% CI, 8.166–30.746) (P < 0.001) for the 1:2 to 1:5 risk group. In contrast, the 1:21 to 1:50 risk group was not significantly associated with preterm PE (aOR, 1.343 (95% CI, 0.722–2.499)) (P = 0.352). Table presents the predicted absolute probabilities of preterm PE, allowing clinicians to interpret individual risk in high‐risk women after aspirin prophylaxis. Parity and previous PE were not found to have any significant interactions with the key predictors (Table ). Sensitivity analyses, including excluding participants from Cluster 2 and adjusting for cluster as a confounder, showed results consistent with the primary analysis, with statistical significance remaining largely unchanged (Tables and ). Final multivariable logistic regression model including key predictors for preterm pre‐eclampsia (PE) among 2919 women identified as high risk by first‐trimester Fetal Medicine Foundation triple test who received aspirin prophylaxis Multivariable equation: logit(p) = −4.47196 + (0.60717 × (1 if chronic hypertension; else 0)) + (0.77115 × (1 if MAP MoM > 1.149; else 0) + (1.23957 × (1 if estimated risk 1:11 to 1:20; else 0)) + (2.03789 × (1 if estimated risk 1:6 to 1:10;