- Top of page
- Supporting Information
Pre-eclampsia (PE) affects about 2–8% of pregnancies. Early PE, resulting in delivery before 34 weeks of gestation, accounts for fewer than 25% of all cases of PE, thus complicating only about 0.5–2% of pregnancies. However, these cases tend to have the worst outcomes1.
About 5–10% of all pregnant women have risk factors for developing PE, such as prior PE and/or intrauterine growth restriction (IUGR), chronic hypertension, diabetes mellitus, chronic renal disease, certain autoimmune diseases and thrombophilias, high body mass index (BMI) and multiple pregnancy, increasing the incidence of early PE to 3–5% in these women2. Therefore, about a third of cases of early PE occur in women with such high-risk factors. However, most of these high-risk patients, who usually undergo intensified surveillance during their pregnancies, never develop PE. Hence, the application of a screening test for PE in a population with a priori high-risk factors for its development may be of particular interest in order to achieve a more accurate stratification of risk.
Recently, various screening tests have been published, some of them reporting a sensitivity and specificity as high as > 90% for the detection of early PE3, 4. However, most of them have been described in general populations and have not been validated in high-risk groups. Among these tests, one of the most promising and simple approaches results from evaluation of the sequential changes of uterine artery resistance between the first and second trimesters of pregnancy5, 6. The aim of our study was to analyze the value of this strategy for the prediction and exclusion of PE in a high-risk population.
- Top of page
- Supporting Information
This analysis included 135 high-risk women with complete follow-up. Twenty percent (27/135) of women developed PE; 21/135 (15.6%) developed late PE and 4.4% (6/135) developed early PE (Figure 1). Table 1 summarizes the demographic characteristics of the study population, comparing unaffected, late PE and early PE pregnancies. Women who subsequently developed early PE were more likely to have had previous PE/IUGR or to have more than one high-risk factor, but there were no significant differences for the other variables. Pregnancy outcomes are shown in Table 2.
Figure 1. Outcome of our study population of 135 women with singleton pregnancies at high risk for pre-eclampsia, according to uterine artery Doppler in first and second trimesters of pregnancy. Group 4 vs Group 1 for pre-eclampsia, P < 0.05 (chi-square test). Group 2 and Group 4 vs Group 1 for early pre-eclampsia, both P < 0.05 (chi-square test). Group 1 was the reference group for calculation of odds ratios (OR). MoM, multiples of the median.
Download figure to PowerPoint
Table 1. Characteristics of pregnancies included in the study according to occurrence of late (delivery ≥ 34 weeks) and early (delivery < 34 weeks) pre-eclampsia (PE)
|Characteristic||Unaffected (n = 108)||Late PE (n = 21)||Early PE (n = 6)|
|Maternal age at recruitment (years)||33.1 ± 4.3||33.3 ± 3.9||36.7 ± 4.6|
|Ethnic group|| || || |
|Body mass index (kg/m2)||27.0 ± 5.9||29.5 ± 9.0||27.4 ± 6.1|
|Family history of PE||4.9||5.0||0|
|Aspirin intake during pregnancy||39.8||57.1||83.3|
|Presenting risk factor|| || || |
| Previous PE/IUGR*||25.9||38.1||66.7|
| Chronic hypertension||25.0||33.3||50.0|
| Pregestational DM||11.1||19.0||16.7|
| Pregestational body mass index > 30 kg/m2||31.5||33.3||33.3|
| Renal disease||2.8||4.8||16.7|
| Autoimmune disease||15.7||14.3||0|
| Two or more factors involved*||25.9||47.6||83.3|
Table 2. Pregnancy outcome of the study cohort (n = 135) according to occurrence of late (delivery ≥ 34 weeks) and early (delivery < 34 weeks) pre-eclampsia (PE)
|Characteristic||Unaffected (n = 108)||Late PE (n = 21)||Early PE (n = 6)|
|GA at delivery (weeks)*†||38.1 ± 2.6||37.4 ± 2.1||31.2 ± 2.5|
|Birth weight (g)*†||3045 ± 516||2922 ± 482||1644 ± 599|
|Birth weight percentile||53 ± 30||60 ± 38||43 ± 27|
|Umbilical artery pH||7.26 ± 0.12||7.23 ± 0.12||7.13 ± 0.20|
|Preterm delivery (< 34 weeks)*†||4 (3.7)||0 (0)||6 (100)|
|Small-for-gestational age||8 (7.3)||4 (19.0)||1 (16.7)|
|Placental abruption||1 (0.9)||0 (0)||1 (16.7)|
|Perinatal death||3 (2.8)||0 (0)||1 (16.7)|
The distribution of mUtA-PI (MoM) values in the first and second trimesters in unaffected, late PE and early PE pregnancies is shown in Figure S1. In the unaffected group, the first-trimester mUtA-PI (MoM) median (interquartile range (IQR)) was 1.04 (0.79–1.25) and the 90th percentile was 1.47, and in the second trimester these values were 1.00 (0.82–1.25) and 1.49, respectively. These median values for the unaffected group were not significantly different in the first trimester from those observed in late (0.99 (0.85–1.29)) and early (1.08 (0.57–1.55)) PE groups; nor were they different in the second trimester from those observed in the late PE group (1.20 (0.83–1.47)). However, the median value in the second trimester of early PE cases (2.00 (0.97–2.20)) was significantly increased relative to the unaffected group (P < 0.01).
Figure 1 depicts the semi-quantitative evaluation of mUtA-PI (MoM) in the first and second trimesters of gestation, following the previously described classification into Groups 1–4. Pregnant women in Group 4 had a significantly higher risk of developing PE and early PE when compared with those in Group 1 (P < 0.05). Women in Group 2 also showed a trend towards a higher risk of developing PE (P = 0.16) but the difference only reached statistical significance for early PE (P < 0.05).
The quantitative evaluation of the sequential changes in mUtA-PI (MoM) between the first and second trimesters using log ratio 2T-1T is illustrated in Figure S2. The median (IQR) of log ratio 2T-1T in unaffected pregnancies was 0.01 (–0.08 to 0.08), in late PE cases it was 0.07 (−0.06 to 0.14) and in early PE cases it was 0.16 (0.09 to 0.34). Compared with in unaffected pregnancies, log ratio 2T-1T was significantly higher only in early PE cases (P < 0.01). Figure 2 shows the ROC curves for the prediction of late PE and early PE by means of mUtA-PI at first and second-trimester scans and log ratio 2T-1T, and the screening characteristics derived from these ROC curves are summarized in Table 3. According to the AUCs, the best accuracy for prediction of early PE was obtained by log ratio 2T-1T, although significant differences were not found. Although mUtA-PI (MoM) in the second trimester was slightly superior when the false-positive rate was fixed at 10%, log 2T-1T was the only screening test to detect all early PE cases when the false-positive rate threshold was raised to 25%.
Table 3. Comparison of screening performance for late (delivery ≥ 34 weeks) and early (delivery < 34 weeks) pre-eclampsia (PE) in high-risk pregnancies, by mean uterine artery pulsatility index (mUtA-PI) at first- and second-trimester scans and log ratio of second-to-first-trimester mUtA-PI in multiples of the median (MoM) (log ratio 2T-1T)
| ||AUC (mean (95% CI))||Detection rate (%) for 10% FPR||Detection rate (%) for 25% FPR|
|Screening test||Late PE||Early PE||Late PE||Early PE||Late PE||Early PE|
|1st trimester mUtA-PI (MoM)||0.528 (0.391–0.665)||0.517 (0.200–0.833)||14.3||33.3||28.6||50.0|
|2nd trimester mUtA-PI (MoM)||0.590 (0.441–0.740)||0.767 (0.484–1.000)||19.0||66.7||47.6||66.7|
|Log ratio 2T-1T||0.599 (0.458–0.741)||0.853 (0.749–0.956)||19.0||50.0||47.6||100.0|
- Top of page
- Supporting Information
This research further develops the concept that uterine artery Doppler performs differently in women at high risk for PE compared with the general population, and this should be taken into account in the prediction of PE13.
The reason for observed differences between these high-risk women and the general population might be related to (1) the tool itself (uterine artery Doppler), (2) the timing of its application (first or second trimester) and/or (3) the intrinsic characteristics of PE in high-risk pregnancies. Regarding the first point (the tool itself: uterine artery Doppler), it has been argued that in women with constitutional predisposition, the vascular remodeling of the spiral arteries could be hampered14, resulting in higher resistance in the uterine arteries. However, although a significantly higher prevalence of notching has been found in women with prior PE, resistance was not increased when compared with controls15. This was further confirmed in our study, in which the distributions of mUtA-PI in the first and second trimesters corresponded well with those described in unselected pregnancies5, 6. Therefore, mUtA-PI may be applied in high-risk women as described in the general population6.
Regarding the second point (timing), when exploring the feasibility of using uterine artery Doppler in the prediction of PE, the available information reveals that its performance is poorer in high-risk than in low-risk pregnancies, regardless of whether it is used in the first or the second trimester13, 16, 17. Remarkably, first-trimester values were very similar in normal and PE pregnancies. Our results correlate well with previous reports, showing that late PE cannot be predicted in high-risk women using uterine artery Doppler and finding the sensitivity for the prediction of early PE at 90% specificity to be 33% in the first trimester and 66% in the second trimester, below the 50% and 81%, respectively, reported in unselected pregnancies7.
There is a consensus that impairment of deep invasion of the myometrial arterial segments after the steep rise in placental oxygen at 10–12 weeks is the most likely mechanism involved in early PE18. Therefore, the predictive value of sequential strategies assessing changes in uterine artery resistance through the first half of pregnancy has been explored on the basis that these changes should reflect the evolution of this process of deep placentation19. These models have reported excellent results for the prediction of early PE in the general population5, 6 but, as far as we know, their performance in high-risk pregnancies has never before been examined.
The semi-quantitative strategy that we applied to our high-risk pregnancies showed that abnormal mUtA-PI in both examinations (Group 4) was associated with the highest risk of developing PE, but in those gestations in which abnormal mUtA-PI in the first trimester had become normal by the second (Group 3) the risk of PE was not increased compared with the group with normal mUtA-PI in both scans (Group 1). This disagrees with studies in low-risk women, in which the abnormal first-trimester but normal second-trimester mUtA-PI group was more predisposed to developing PE6. In contrast, in our Group 2 (normal first-trimester but abnormal second-trimester mUtA-PI) there was a substantial increase in risk of having PE, as has also been observed in low-risk women6. It should be noted that, given the high-risk nature of our study population, the overall incidence of PE was much higher in all groups when compared with low-risk women. However, when analyzing early PE cases separately, pregnant women with a presumably physiological second wave of trophoblast invasion (Groups 1 and 3) had a very low risk of developing early PE (< 2%), an incidence similar to that observed in the general population. However, in patients with presumably impaired trophoblast invasion (Groups 2 and 4), the incidence of early PE was very high (10–20-fold increase) compared with in the general population6.
The advantages of taking into account sequential changes in mUtA-PI are best evidenced by our quantitative strategy. With this approach, even in the two cases of early PE with normal mUtA-PI in both examinations, we detected a relative worsening in mUtA-PI (MoM) from the first to the second trimester, by means of the log ratio 2T-1T. Therefore, although measurement in the second trimester seems to be more discriminative in assessing the true risk of early PE in high-risk pregnancies, the first trimester scan also provides valuable information for its prediction. In fact, all cases of early PE were detected by means of log ratio 2T-1T when the false-positive rate was fixed at 25%. Although this cut-off may not seem sufficiently discerning, its utility for screening should be considered, taking into account that the source is already a high-risk population. Therefore, regarding the third point (intrinsic characteristics of PE in high-risk pregnancies), our study shows that the performance of log ratio 2T-1T enables a better reassessment of risk in women with a priori high-risk factors for developing early PE, and that defective deep placentation plays a critical role in the pathogenesis of early PE in high-risk women. This is further evidenced by the fact that daily low-dose aspirin intake is most likely to be beneficial for the prevention of PE when it is administered to women at high risk for PE and is started early in pregnancy20.
Despite there being a lack of evidence to demonstrate whether the use of uterine artery Doppler in high-risk pregnancies improves maternal–fetal outcome21, expert monitoring of PE patients has been shown to be associated with reduced maternal risk22. However, the limited resources for such monitoring require further selection of the true high-risk patients in order to improve our efficiency. In this sense, we believe that the improved reassignment of risk provided by log ratio 2T-1T may help to redirect resources rationally, focusing attention on the group of patients with increased log ratio 2T-1T but without exposing those with normal results to any increased risk.
We acknowledge two main limitations in our study. First, the sample size was relatively small, a problem common in previous series also. However, it is not easy to recruit pregnancies with true high-risk factors for PE (i.e. those associated with at least a three-fold increase in risk)1, 2. It is important to keep in mind the restrictive criteria that were used for inclusion of our patients, with all of them having at least one major risk condition, which were responsible for our high incidence (20%) of PE. Second, in this observational study, the influence of aspirin on the evolution of resistance in the uterine arteries and the incidence of PE cannot be quantified, as aspirin intake was not randomized. However, a clinical trial with aspirin is probably impossible due to ethical concerns derived from the widespread practice of giving aspirin to all pregnant women at high risk for PE.
In conclusion, the application of semi-quantitative and especially quantitative models to evaluate sequential changes in uterine artery Doppler findings between the first and second trimesters could be of additional value in assessing high-risk women regarding their true risk of developing early PE. This information may be useful to improve our prenatal resources and clinical efficiency.
SUPPORTING INFORMATION ON THE INTERNET
The following supporting information may be found in the online version of this article:
Figures S1 and S2 Box-and-whisker plots showing distribution of values expressed as multiples of the median (MoM) of mean uterine artery pulsatility index (mUtA-PI) at first-trimester (11–13 weeks' gestation) and second-trimester (19–22 weeks' gestation) scan (Figure S1) and showing distribution of log ratio of second-trimester mUtA-PI in MoM to first-trimester mUtA-PI in MoM (log ratio 2T-1T) (Figure S2) in unaffected pregnancies and in those complicated by late and early pre-eclampsia.
Table S1 Regression models used to calculate mean uterine artery pulsatility index (mUtA-PI) expressed as multiples of the median (MoM) adjusted by gestational age at first- and second-trimester scans and log ratio of second- to first-trimester mUtA-PI in MoM (log ratio 2T-1T)6.