Interpretation in light of other evidence
To improve the prediction of adverse outcomes related to PE different tools such as the combination of signs and symptoms of PE, the evaluation of fetal and maternal Doppler ultrasound and biochemical markers alone and in combination with clinical factors have been investigated. In 2017, Thangaratinam et al demonstrated that the PREP-model predicts maternal outcomes in patients with clinical early-onset PE, but the prediction of perinatal outcomes was not evaluated (9). In our study, the PREP-L score had a limited predictive value of the adverse neonatal outcomes in early-onset PE with severe features (AUC ROC 0.69 [95% CI 0.51-0.86]).
There is controversy regarding the role of fetal Doppler in PE in predicting adverse neonatal outcome. Rani et al reported that Doppler indices of MCA and UA have good specificity but low sensitivity for detecting adverse perinatal outcomes in PE with or without severe features (13). Two prospective studies, including respectively 100 and 60 patients with severe PE, support CPR as a tool for the prediction of adverse perinatal outcomes but the majority of cases were late-onset PE (mean gestational age at admission 37 weeks of gestation) (10,11). Similarly, Orabona et al in a cohort study on 168 women with PE diagnosed at a mean gestational age of 32+6 weeks found that CPR was more accurate than each of their components alone in predicting adverse neonatal outcomes, albeit only marginally (34). The heterogenicity of the women included in these studies (mixing early and late; and non-severe and severe PE) may account for the inconsistent results. In our population of early-onset PE with severe features, Doppler indices of MCA, DV and CPR were not significantly different between the groups with and without adverse neonatal outcomes, and only the composite proportion of fetuses with advanced Doppler findings (absent/reversed diastolic flow in the UA or pulsatile DV) showed differences between groups. This could be explained by the greater placental involvement in the early-onset cases and the higher association with FGR; and the stronger impact of prematurity in these cases.
In the last years, several studies have shown that angiogenic factors can increase the prediction of PE and its adverse outcomes in patients with impending signs and symptoms of the disease (18,19,35). However, the role of angiogenic factors is not similarly promising in women with established severe PE. In 2014, Pinheiro et al reported a correlation between angiogenic imbalance and poor neonatal outcome in early-onset PE (36). Simon et al demonstrated an association between sFlt-1/PlGF ratio >655 and risk of delivery in less than 48 hours, nevertheless none of the angiogenic factors evaluated were good predictors of adverse maternal or perinatal outcomes (37). In addition, because both the degree of angiogenic imbalance and the neonatal outcomes are highly correlated with the gestational age at onset of the disease (5,21), we propose that the predictive role of these markers should be evaluated as the added value over a baseline risk capturing the gestational age at onset, such as the PREP score.
In 2021, Droge et al found that integrating all available clinical and biochemical markers into a regression model yields the best predictive performance of PE-related adverse outcomes, including both maternal and perinatal (the AUC of blood pressure and proteinuria was 69%, the AUC of the sFlt-1/PlGF on its own was 85.7% and including all clinical information was 88.7%). The cohort were women with suspected disease (n=1117) and only 351 women (31.4%) had the final diagnosis of PE, most with late-onset disease (38). Gomez-Arriaga et al, in 2014, using a cohort of 51 singleton pregnancies with early-onset PE suggested that the sFlt-1/PlGF ratio in combination with gestational age is useful for the prognostic assessment of neonatal complications (AUC was 89% corresponding to sensitivity, specificity PPV and NPV of 64%, 83%, 57% and 97% respectively), but this combination has limited value for the prediction of maternal complications (12). In the present study, we found that the combination of maternal characteristics at admission (PREP-L score) and advanced Doppler or PlGF has a good predictive value (AUC ~ 90%) for the prediction of neonatal complications.
Delivery is the definitive treatment of PE but the optimal time of delivery in severe cases remains controversial because the net benefit between reducing maternal risks by planned delivery and the secondary neonatal risk associated with prematurity is unclear. Therefore, it is important to develop prognostic tools to counsel the trade-off between neonatal benefits versus maternal risks of expectant management. While patients and health professionals give a similar importance to maternal complications as core outcomes of PE, neonatal complications are seen as more relevant by patients than by professionals or researchers (39). Therefore, to advance towards a patient-centered care and shared decision-making, prediction models for adverse neonatal outcomes are needed in the management of PE. The combination of a maternal risk score (which includes gestational age at onset of PE) and fetal Doppler and/or PlGF predicts with good accuracy those cases at risk of adverse neonatal outcomes.