Abstract：
This study aimed to: (1) identify all relevant studies reporting on the diagnostic accuracy of maternal circulating placental growth factor) alone or as a ratio with soluble fms-like tyrosine kinase-1), and of placental growth factor-based models (placental growth factor combined with maternal factors±other biomarkers) in the second or third trimester to predict subsequent development of preeclampsia in asymptomatic women; (2) estimate a hierarchical summary receiver-operating characteristic curve for studies reporting on the same test but different thresholds, gestational ages, and populations; and (3) select the best method to screen for preeclampsia in asymptomatic women during the second and third trimester of pregnancy by comparing the diagnostic accuracy of each method.
A systematic search was performed through MEDLINE, Embase, CENTRAL, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform databases from January 1, 1985 to April 15, 2021.
Studies including asymptomatic singleton pregnant women at >18 weeks\' gestation with risk of developing preeclampsia were evaluated. We included only cohort or cross-sectional test accuracy studies reporting on preeclampsia outcome, allowing tabulation of 2×2 tables, with follow-up available for >85%, and evaluating performance of placental growth factor alone, soluble fms-like tyrosine kinase-1- placental growth factor ratio, or placental growth factor-based models. The study protocol was registered on the International Prospective Register Of Systematic Reviews (CRD 42020162460).
Because of considerable intra- and interstudy heterogeneity, we computed the hierarchical summary receiver-operating characteristic plots and derived diagnostic odds ratios, β, θi, and Λ for each method to compare performances. The quality of the included studies was evaluated by the QUADAS-2 tool.
The search identified 2028 citations, from which we selected 474 studies for detailed assessment of the full texts. Finally, 100 published studies met the eligibility criteria for qualitative and 32 for quantitative syntheses. Twenty-three studies reported on performance of placental growth factor testing for the prediction of preeclampsia in the second trimester, including 16 (with 27 entries) that reported on placental growth factor test alone, 9 (with 19 entries) that reported on the soluble fms-like tyrosine kinase-1-placental growth factor ratio, and 6 (16 entries) that reported on placental growth factor-based models. Fourteen studies reported on performance of placental growth factor testing for the prediction of preeclampsia in the third trimester, including 10 (with 18 entries) that reported on placental growth factor test alone, 8 (with 12 entries) that reported on soluble fms-like tyrosine kinase-1-placental growth factor ratio, and 7 (with 12 entries) that reported on placental growth factor-based models. For the second trimester, Placental growth factor-based models achieved the highest diagnostic odds ratio for the prediction of early preeclampsia in the total population compared with placental growth factor alone and soluble fms-like tyrosine kinase-1-placental growth factor ratio (placental growth factor-based models, 63.20; 95% confidence interval, 37.62-106.16 vs soluble fms-like tyrosine kinase-1-placental growth factor ratio, 6.96; 95% confidence interval, 1.76-27.61 vs placental growth factor alone, 5.62; 95% confidence interval, 3.04-10.38); placental growth factor-based models had higher diagnostic odds ratio than placental growth factor alone for the identification of any-onset preeclampsia in the unselected population (28.45; 95% confidence interval, 13.52-59.85 vs 7.09; 95% confidence interval, 3.74-13.41). For the third trimester, Placental growth factor-based models achieved prediction for any-onset preeclampsia that was significantly better than that of placental growth factor alone but similar to that of soluble fms-like tyrosine kinase-1-placental growth factor ratio (placental growth factor-based models, 27.12; 95% confidence interval, 21.67-33.94 vs placental growth factor alone, 10.31; 95% confidence interval, 7.41-14.35 vs soluble fms-like tyrosine kinase-1-placental growth factor ratio, 14.94; 95% confidence interval, 9.42-23.70).
Placental growth factor with maternal factors ± other biomarkers determined in the second trimester achieved the best predictive performance for early preeclampsia in the total population. However, in the third trimester, placental growth factor-based models had predictive performance for any-onset preeclampsia that was better than that of placental growth factor alone but similar to that of soluble fms-like tyrosine kinase-1-placental growth factor ratio. Through this meta-analysis, we have identified a large number of very heterogeneous studies. Therefore, there is an urgent need to develop standardized research using the same models that combine serum placental growth factor with maternal factors ± other biomarkers to accurately predict preeclampsia. Identification of patients at risk might be beneficial for intensive monitoring and timing delivery.
A systematic search was performed through MEDLINE, Embase, CENTRAL, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform databases from January 1, 1985 to April 15, 2021.
Studies including asymptomatic singleton pregnant women at >18 weeks\' gestation with risk of developing preeclampsia were evaluated. We included only cohort or cross-sectional test accuracy studies reporting on preeclampsia outcome, allowing tabulation of 2×2 tables, with follow-up available for >85%, and evaluating performance of placental growth factor alone, soluble fms-like tyrosine kinase-1- placental growth factor ratio, or placental growth factor-based models. The study protocol was registered on the International Prospective Register Of Systematic Reviews (CRD 42020162460).
Because of considerable intra- and interstudy heterogeneity, we computed the hierarchical summary receiver-operating characteristic plots and derived diagnostic odds ratios, β, θi, and Λ for each method to compare performances. The quality of the included studies was evaluated by the QUADAS-2 tool.
The search identified 2028 citations, from which we selected 474 studies for detailed assessment of the full texts. Finally, 100 published studies met the eligibility criteria for qualitative and 32 for quantitative syntheses. Twenty-three studies reported on performance of placental growth factor testing for the prediction of preeclampsia in the second trimester, including 16 (with 27 entries) that reported on placental growth factor test alone, 9 (with 19 entries) that reported on the soluble fms-like tyrosine kinase-1-placental growth factor ratio, and 6 (16 entries) that reported on placental growth factor-based models. Fourteen studies reported on performance of placental growth factor testing for the prediction of preeclampsia in the third trimester, including 10 (with 18 entries) that reported on placental growth factor test alone, 8 (with 12 entries) that reported on soluble fms-like tyrosine kinase-1-placental growth factor ratio, and 7 (with 12 entries) that reported on placental growth factor-based models. For the second trimester, Placental growth factor-based models achieved the highest diagnostic odds ratio for the prediction of early preeclampsia in the total population compared with placental growth factor alone and soluble fms-like tyrosine kinase-1-placental growth factor ratio (placental growth factor-based models, 63.20; 95% confidence interval, 37.62-106.16 vs soluble fms-like tyrosine kinase-1-placental growth factor ratio, 6.96; 95% confidence interval, 1.76-27.61 vs placental growth factor alone, 5.62; 95% confidence interval, 3.04-10.38); placental growth factor-based models had higher diagnostic odds ratio than placental growth factor alone for the identification of any-onset preeclampsia in the unselected population (28.45; 95% confidence interval, 13.52-59.85 vs 7.09; 95% confidence interval, 3.74-13.41). For the third trimester, Placental growth factor-based models achieved prediction for any-onset preeclampsia that was significantly better than that of placental growth factor alone but similar to that of soluble fms-like tyrosine kinase-1-placental growth factor ratio (placental growth factor-based models, 27.12; 95% confidence interval, 21.67-33.94 vs placental growth factor alone, 10.31; 95% confidence interval, 7.41-14.35 vs soluble fms-like tyrosine kinase-1-placental growth factor ratio, 14.94; 95% confidence interval, 9.42-23.70).
Placental growth factor with maternal factors ± other biomarkers determined in the second trimester achieved the best predictive performance for early preeclampsia in the total population. However, in the third trimester, placental growth factor-based models had predictive performance for any-onset preeclampsia that was better than that of placental growth factor alone but similar to that of soluble fms-like tyrosine kinase-1-placental growth factor ratio. Through this meta-analysis, we have identified a large number of very heterogeneous studies. Therefore, there is an urgent need to develop standardized research using the same models that combine serum placental growth factor with maternal factors ± other biomarkers to accurately predict preeclampsia. Identification of patients at risk might be beneficial for intensive monitoring and timing delivery.
摘要：
目的：本研究旨在：（1）确定所有报告母体循环胎盘生长因子）单独或与可溶性fms样酪氨酸激酶-1的比例的诊断准确性的相关研究，和胎盘生长因子为基础的模型（胎盘生长因子结合母体因素±其他生物标志物）在孕中期或晚期预测先兆子痫的后续发展无症状妇女;（2）估计分层汇总受试者-工作特征曲线报告相同的测试，但不同的阈值，胎龄,和人群；（3）通过比较每种方法的诊断准确性，选择在妊娠中期和妊娠中期无症状妇女中筛查先兆子痫的最佳方法。
方法：通过MEDLINE进行了系统搜索，Embase,中部,ClinicalTrials.gov,和世界卫生组织国际临床试验注册平台数据库，从1985年1月1日至2021年4月15日。
方法：对包括无症状单胎妊娠妇女在妊娠18周有先兆子痫风险的研究进行了评估。我们仅纳入报告先兆子痫结局的队列或横断面检验准确性研究，允许2×2表格的制表，随访时间>85%，并单独评估胎盘生长因子的性能，可溶性fms样酪氨酸激酶-1-胎盘生长因子比率，或基于胎盘生长因子的模型。研究方案已在国际前瞻性系统审查登记册（CRD42020162460）上注册。
方法：由于研究内和研究间相当大的异质性，我们计算了分层汇总接收器-操作特征图和得出的诊断赔率比，β,θi,和Λ用于比较每种方法的性能。通过QUADAS-2工具评估纳入研究的质量。
结果：搜索确定了2028条引文，我们从中选择了474项研究对全文进行详细评估。最后,100项已发表的研究符合定性标准，32项符合定量综合标准。23项研究报道了胎盘生长因子检测在妊娠中期预测先兆子痫的表现，包括仅在胎盘生长因子测试报告的16个（有27个条目），9（有19个条目）报道了可溶性fms样酪氨酸激酶-1-胎盘生长因子的比例，和6（16个条目）在基于胎盘生长因子的模型上报告。14项研究报道了胎盘生长因子检测在妊娠晚期预测先兆子痫的表现，包括10个（18个条目）仅在胎盘生长因子测试报告，8（有12个条目）报道了可溶性fms样酪氨酸激酶-1-胎盘生长因子的比例，和7（有12个条目）在基于胎盘生长因子的模型上报告。在妊娠中期，与单独使用胎盘生长因子和可溶性fms样酪氨酸激酶-1-胎盘生长因子比率相比，基于胎盘生长因子的模型在总人群中预测早期先兆子痫的诊断优势比最高（基于胎盘生长因子的模型，63.20；95%置信区间，37.62-106.16与可溶性fms样酪氨酸激酶-1-胎盘生长因子比率相比，6.96；95%置信区间，1.76-27.61与单独的胎盘生长因子相比，5.62；95%置信区间，3.04-10.38）；在未选择的人群中，基于胎盘生长因子的模型比单独使用胎盘生长因子的模型具有更高的诊断优势比（28.45；95％置信区间，13.52-59.85vs7.09；95%置信区间，3.74-13.41)。在妊娠晚期，基于胎盘生长因子的模型实现了对任何先兆子痫的预测，该预测明显优于单独的胎盘生长因子，但与可溶性fms样酪氨酸激酶-1-胎盘生长因子比率相似（基于胎盘生长因子的模型，27.12；95%置信区间，21.67-33.94与单独的胎盘生长因子相比，10.31；95%置信区间，7.41-14.35与可溶性fms样酪氨酸激酶-1-胎盘生长因子比率相比，14.94；95%置信区间，9.42-23.70)。
结论：在妊娠中期确定的胎盘生长因子与母体因子±其他生物标志物在整个人群中对早期先兆子痫的预测表现最好。然而,在妊娠晚期，基于胎盘生长因子的模型对任何先兆子痫的预测性能优于单独的胎盘生长因子，但与可溶性fms样酪氨酸激酶-1-胎盘生长因子比率相似。通过这个荟萃分析,我们已经确定了大量非常异质的研究。因此,迫切需要使用相同的模型开发标准化研究，该模型将血清胎盘生长因子与母体因子±其他生物标志物相结合，以准确预测先兆子痫。识别处于危险中的患者可能有利于强化监测和定时递送。
方法：通过MEDLINE进行了系统搜索，Embase,中部,ClinicalTrials.gov,和世界卫生组织国际临床试验注册平台数据库，从1985年1月1日至2021年4月15日。
方法：对包括无症状单胎妊娠妇女在妊娠18周有先兆子痫风险的研究进行了评估。我们仅纳入报告先兆子痫结局的队列或横断面检验准确性研究，允许2×2表格的制表，随访时间>85%，并单独评估胎盘生长因子的性能，可溶性fms样酪氨酸激酶-1-胎盘生长因子比率，或基于胎盘生长因子的模型。研究方案已在国际前瞻性系统审查登记册（CRD42020162460）上注册。
方法：由于研究内和研究间相当大的异质性，我们计算了分层汇总接收器-操作特征图和得出的诊断赔率比，β,θi,和Λ用于比较每种方法的性能。通过QUADAS-2工具评估纳入研究的质量。
结果：搜索确定了2028条引文，我们从中选择了474项研究对全文进行详细评估。最后,100项已发表的研究符合定性标准，32项符合定量综合标准。23项研究报道了胎盘生长因子检测在妊娠中期预测先兆子痫的表现，包括仅在胎盘生长因子测试报告的16个（有27个条目），9（有19个条目）报道了可溶性fms样酪氨酸激酶-1-胎盘生长因子的比例，和6（16个条目）在基于胎盘生长因子的模型上报告。14项研究报道了胎盘生长因子检测在妊娠晚期预测先兆子痫的表现，包括10个（18个条目）仅在胎盘生长因子测试报告，8（有12个条目）报道了可溶性fms样酪氨酸激酶-1-胎盘生长因子的比例，和7（有12个条目）在基于胎盘生长因子的模型上报告。在妊娠中期，与单独使用胎盘生长因子和可溶性fms样酪氨酸激酶-1-胎盘生长因子比率相比，基于胎盘生长因子的模型在总人群中预测早期先兆子痫的诊断优势比最高（基于胎盘生长因子的模型，63.20；95%置信区间，37.62-106.16与可溶性fms样酪氨酸激酶-1-胎盘生长因子比率相比，6.96；95%置信区间，1.76-27.61与单独的胎盘生长因子相比，5.62；95%置信区间，3.04-10.38）；在未选择的人群中，基于胎盘生长因子的模型比单独使用胎盘生长因子的模型具有更高的诊断优势比（28.45；95％置信区间，13.52-59.85vs7.09；95%置信区间，3.74-13.41)。在妊娠晚期，基于胎盘生长因子的模型实现了对任何先兆子痫的预测，该预测明显优于单独的胎盘生长因子，但与可溶性fms样酪氨酸激酶-1-胎盘生长因子比率相似（基于胎盘生长因子的模型，27.12；95%置信区间，21.67-33.94与单独的胎盘生长因子相比，10.31；95%置信区间，7.41-14.35与可溶性fms样酪氨酸激酶-1-胎盘生长因子比率相比，14.94；95%置信区间，9.42-23.70)。
结论：在妊娠中期确定的胎盘生长因子与母体因子±其他生物标志物在整个人群中对早期先兆子痫的预测表现最好。然而,在妊娠晚期，基于胎盘生长因子的模型对任何先兆子痫的预测性能优于单独的胎盘生长因子，但与可溶性fms样酪氨酸激酶-1-胎盘生长因子比率相似。通过这个荟萃分析,我们已经确定了大量非常异质的研究。因此,迫切需要使用相同的模型开发标准化研究，该模型将血清胎盘生长因子与母体因子±其他生物标志物相结合，以准确预测先兆子痫。识别处于危险中的患者可能有利于强化监测和定时递送。
