Background
The most important functions of the placenta are adequate exchange of oxygen, nutrients and waste products between fetal and maternal circulation [
1]. At 12–13 weeks of gestation, trophoblast cells have invaded the decidual segments of maternal spiral arteries (SpA) and transformed these small, adrenergic-sensitive high-resistance vessels into wide, adrenergic-insensitive low-resistance vessels [
1,
2]. From 15 weeks onwards, a second, endovascular invasion starts to remodel the myometrial segments and is fully completed at mid-pregnancy [
3]. Downstream remodeling of the SpA decreases utero-placental resistance drastically and allows significant increase in volumetric blood flow to the placenta [
1,
4‐
7]. Interestingly, there are studies showing that trophoblast invasion is concentrated in the central area of the placental bed, whereas peripheral myometrial segments are much less changed [
8,
9].
The etiology of placenta syndrome (PS), a collective term for placenta-insufficiency related disorders, is poorly understood. Inadequate remodeling of SpA in PS-pregnancies potentially affects the development of the placenta as high velocity blood flow could cause mechanical stress on the tissue, or could influence development due to perfusion-reperfusion injury [
6,
7,
10].
The golden standard for the detection of absent or aberrant SpA remodeling is postpartum histopathological examination of the placenta, however this is behind time to affect clinical decision making in endangered gestation [
10]. Previous studies have shown that abnormal Doppler velocimetry of the more upstream uterine arteries in second trimester is associated with an increased risk on vascular pregnancy related complications (among which pre-eclampsia (PE), fetal growth restriction (FGR) and stillbirth) [
11,
12]. Abnormal uterine artery Doppler indices during earlier gestation may also predict PE and FGR [
13,
14]. As defective SpA remodeling ultimately associates with adverse outcome, capturing this incomplete SpA remodeling antedating complicated pregnancies during early gestation may timely indicate those at risk [
1,
15]. Our current knowledge on SpA remodeling is mostly based on previous literature discussing postpartum histological findings and cross-sectional ultrasound measurements. However, longitudinal ultrasound data throughout pregnancy would provide a more accurate insight in the SpA remodeling process during gestation. Therefore, we performed a systematic review and meta-analysis on Doppler measurements of the SpA, to antepartum visualize the remodeling process in human pregnancy by ultrasonography.
Methods
This systematic review and meta-analysis on the remodeling of SpA flow velocimetry throughout human pregnancy is part of a large series of meta-analysis on physiological and pathophysiological adaptation of relevant indices in pregnancy [
16‐
18]. The review was conducted in accordance to the “PRISMA Statement” for reporting systematic reviews and meta-analyses [
19].
Literature search
The electronic databases Cochrane library (1997–2019), Pubmed (1946–2019) and EMBASE (1974–2019), were searched for relevant articles evaluating SpA Doppler measurements. The keywords used in the literature search were “spiral artery” or “spiral remodeling” in combination with “Doppler” or “color Doppler”, or “Doppler velocimetry”, or “Doppler sonography” or “spiral artery Doppler”, see Table
1.
Table 1
Search strategy (Pubmed database)
Search (spiral artery) AND Doppler Search Pubmed (no limits or filters) 1. Spiral artery [All Fields] 2. Spiral artery remodeling [All Fields] 3. Search 1 OR 2 4. Color Doppler ultrasonography [Mesh] 5. Doppler [All Fields] 6. Doppler sonography [All Fields] 7. Doppler velocimetry [All Fields] 8. Pulse wave Doppler [All Fields] 9. Search 4 OR 5 OR 6 OR 7 OR 8 10. Search 3 AND 9 |
Additionally, the reference list of all selected primary articles were examined for potential citations not captured by the initial search. The search was limited to papers published in English until April 2019.
Study selection
The first selection, based on title and abstract, was performed independently by two investigators (VS and LE). In case of discrepancy, agreement was reached by consensus. The second selection was also performed by two investigators (VS and LE) independently, based on the full text. After checking the manuscripts and crosschecking their reference lists, the final selection of studies was made. Inclusion criteria were studies on singleton pregnancies in which absolute values for SpA Doppler measurements were documented at any gestational age during pregnancy. Studies with both nulliparous and multiparous women were included; no limitations were set on maternal characteristics. Moreover, both transvaginal and transabdominal measurements were included as well as measurements of central localized and peripheral localized SpA. We initially wanted to include only articles describing SpA measurements in healthy and hypertensive pregnancies. Given the limited amount of articles describing SpA measurements in hypertensive pregnancies, we decided to include all articles describing SpA measurements in complicated pregnancies. Therefore, complicated pregnancies in the included articles were represented by PE, preterm labor, FGR, missed abortion, miscarriage and placental abruption.
In some studies, absolute values of Doppler indices could not be extracted as a result of data being reported in graphs or figures. We considered the use of, for example, a plot digitizer to extract the data out of graphs or plots. The accuracy of the assessment was in some cases questionable, and in other cases, not all graphs and plots showed 95% confidence interval (CI) lines, making it impossible to calculate a reliable standard deviation (SD). Therefore, adding a plot digitized estimated SpA value would have a great impact on the trustworthiness of the pooled mean. Therefore, we refrained from generating extra data using the plot digitizer as it may be less accurate compared to the other included articles.
Data extraction and quality assessment
Two reviewers (VS and LE) both extracted the data and assessed the risk of bias. In case of discrepancy in assessment of the risk of bias, agreement was reached by consensus. The following data was extracted from each included study: 1) Study design; 2) Sample size; 3) Probe; 4) Doppler method 5) MHz of the probe; 6) Gestational age during measurement; and 7) Doppler velocimetry outcomes (Table
2). The relevant outcomes included pulsatility index (PI), resistance index (RI) and peak systolic velocity (PSV, m/s).
Table 2
Baseline characteristics from included studies
| Prospective longitudinal | 49 | / | Transvaginal | Colour | 6 | 5–10 | PSV |
| Prospective longitudinal | 37 | / | Transvaginal | Colour | 6–9 | 8–14 | PI |
| Prospective longitudinal | 86 | 21a | Transabdominal | Colour | 3.5 | 11–13; 14–17; 18–24 | PI, RI |
| Prospective longitudinal | 97 | / | Transabdominal | Colour | 3.5 | 11–13; 14–17; 18–24 | PI |
| Prospective longitudinal | 94 | / | Transabdominal | Colour | 3.5 | 10–15; 15–20; 20–25; 25–30; 30–35; 35–40 | PI, RI, PSV |
| Cross-sectional | 115 | / | Transvaginal + transabdominal | Colour | 3.5 / 5 | 7–42 | PI, RI, PSV |
| Prospective cross-sectional | 60 | 54a | Transvaginal | Colour | 6 | 6–12 | PI, RI |
Mäkikallio, Jouppila ( ref. [ 26]) | Prospective cross-sectional | 31 | 10c | Transvaginal | Colour | 5 | 6; 8; 9; 11 | RI |
Mäkikallio, Tekay ( ref. [ 27]) | Prospective longitudinal | 16 | / | Transabdominal | Colour | 5 | 5; 7; 8; 10 | PI, PSV |
| Prospective cross-sectional | 64 | / | Transabdominal | Colour | 5 | 17–20 | PI, RI |
| Prospective longitudinal | 189 | 25b | Transvaginal | Colour | 12 | 5–12 | PI, RI |
| Prospective cross-sectional | 84 | 16c | Transvaginal | Colour | 6 | 6–12 | PI, RI |
In order to assess the quality and risk of bias of included studies, the Quality In Prognosis Studies (QUIPS) tool [
29] was modified for the purpose of this review. Quality assessment was performed on the following domains: 1) Study participation; 2) Study attrition; 3) Outcome measurement; 4) Data reporting; and 5) Study design (Table
3).
Table 3
Quality assessment of included studies
Study participation | Adequate description of participants’ characteristics | | | | | | | | | | | | |
• Parity or gravidity | – | – | – | + | + | – | – | + | – | – | – | + |
• Health or comorbidities of participants | – | – | + | + | + | – | + | + | – | – | – | – |
• Clear reporting of weeks amenorrhea | + | + | + | + | + | + | + | + | + | + | + | + |
• Ethnicity | – | – | – | – | – | – | – | – | – | – | – | – |
• Age | – | – | + | + | + | – | + | + | + | + | + | + |
• Non-pregnant weight/BMI | – | – | + | + | – | – | – | – | – | – | – | – |
• Use of medication or supplements | – | – | – | – | – | – | + | – | – | – | – | – |
Adequate description of participant recruitment | – | – | + | + | + | – | + | – | + | + | + | + |
Adequate description of inclusion and exclusion criteria | – | – | + | + | + | + | + | + | + | – | + | + |
Study attrition | Reasons for loss to follow-up/drop-out are provided | – | + | – | – | – | – | – | – | – | – | – | – |
Adequate description of participants lost to follow-up/ differences between participants who completed and drop-outs | – | – | – | – | – | – | – | – | – | – | – | – |
Outcome measurement | Method of measurement is adequately valid and reliable | + | + | + | + | + | + | + | + | + | + | + | + |
The methods and setting are the same for all study participants and throughout follow up | + | + | + | + | + | – | + | + | + | + | + | + |
Data reporting | Data is extractable as mean with standard deviation for analysis | + | + | + | + | + | + | + | + | + | + | + | + |
Study design | Study used a longitudinal study design | – | + | + | + | + | + | – | + | + | – | – | – |
Multiple (> 2) longitudinal pregnant measurements during pregnancy of the variable | ? | + | + | + | + | + | ? | + | + | ? | ? | ? |
Assessment of intra- and/or inter-observer variation is performed | – | – | + | + | – | – | – | + | + | – | – | – |
Count (+) | 4 | 7 | 12 | 13 | 11 | 5 | 9 | 11 | 10 | 6 | 7 | 8 |
Score % | 24 | 41 | 71 | 76 | 65 | 29 | 53 | 65 | 59 | 35 | 41 | 47 |
Quality | Low | Medium | High | High | High | Low | Medium | High | Medium | Low | Medium | Medium |
Scoring of each criteria occurred as insufficient [−] or sufficient [+]. In case an item was not applicable for the study, a question mark [?] was used. Based on the number of [+], the total score as a percentage was calculated. Articles scoring < 30% were defined as low quality, between 30 and 60% as moderate quality, and > 60% as high quality.
Statistical analysis
The obtained SpA measurements were categorized into three pregnancy trimesters and divided by healthy versus complicated pregnancies and central versus peripheral SpA. Some articles performed two or more SpA measurements in one pregnancy trimester in the same population, explaining the repeated presence of one article in the forest plots. The analysis was conducted by using the R Project for Statistical, R version 3.4.0 with ‘meta’package version 2.0–0 [
30]. In few instances, the continuous outcome measures were not presented with their corresponding SD but as a standard error or 95% CI. In these cases, a SD was calculated from the available mean and range according to the Cochrane handbook for Systematic Review of Interventions [
31]. SpA weighted means with 95% CI were calculated separately for the predefined trimesters using a random-effects model, as described by DerSimonian and Laird [
32]. This model allows for inter-study variation and was chosen because heterogenic populations (both healthy and complicated pregnancies) were used. Means and standard deviations were pooled into one combined measurements for studies reporting multiple measurements within a pregnancy trimester. Heterogeneity was explored as the ratio between total heterogeneity and total variability with the I
2 statistic. I
2 can differentiate between true heterogeneity and sampling variance [
33]. Differences in SpA measurements between trimesters were considered statistically significant at
p < 0.05.
Discussion
In healthy gestation, we observed a consistent decrease of PI and RI in SpA from the first to the second trimester, after which no relevant changes towards the third trimester were seen. Transformation of adrenergic-sensitive high-resistance into adrenergic-insensitive low-resistance SpA ensures unrestricted blood flow into the placenta [
34,
35]. During early gestation, interstitial and endovascular trophoblast cells invade into SpA, where they start to replace the muscular wall and endothelium of these feeding vessels. Consequently, diameters of SpA broaden and lose their ability to respond to vasoactive stimuli, enabling continuous increased low-velocity low-resistance blood flow into the placental intervillous space and avoiding damage to placental tissue [
1,
3,
6]. In parallel, total peripheral vascular resistance decreases and maternal cardiac output rises [
36].
The exact time at which high oxidative maternal blood starts to flow into the intervillous space ranges from 5 to 12 weeks of gestation and is prevented earlier by the formation of endovascular trophoblast ‘plugs’ in the distal segment of the SpA [
20,
21,
24]. These trophoblast plugs create a hypoxic placental environment that protects the foetus from oxidative stress and its damage [
37]. The controversial data about timing of onset of maternal blood flow into the placenta could possibly be explained by the limited ultra-sonographic possibilities when attempting to visualize small vessels with low velocity. Roberts et al. [
38] circumvented this limitation by using a contrast agent that enabled them to apprehend the vascular filling of the placenta. Although maternal blood flow into the placenta was detected at already 6 weeks of gestation, they concluded that microvascular flux into the intervillous space did not progressively increase until 13 weeks. After 13 weeks of gestation, significant changes in the nature of SpA blood flow parameters occur, as is illustrated by our results.
Defective remodeling of SpA can lead to a premature onset of maternal, high-flow and -oxidative blood into the placenta, generating mechanical and biochemical trophoblastic damage and increased apoptosis [
39]. In PS-pregnancies, when oxidative stress is not yet sufficient to influence the preservation of the conceptus, transformation of the SpA walls will occur partially in the center of the placental bed and is limited to the decidual segments of the vessels. This results in vessels with high resistance, diminished perfusion and as a possible result the development of clinical hypertensive disorders or FGR [
2,
6,
7].
Three of the included studies reported PI/ RI in healthy and complicated pregnancies [
15,
27,
28]. The observed differences are based on purely first trimester measurements, while the first to second trimester change in PI/RI as sign of adequate spiral artery remodeling may be the pivot between uncomplicated and complicated pregnancy. Additionally, not all reported complications are strictly associated with defective spiral artery remodeling, possibly confounding the results.
Pijnenborg et al. [
40] found that with enlargement of the placental site, the SpA in the peripheral parts of the placenta come to lie more obliquely and causes their distal segments to lie more parallel to the basal plate. Multiple openings are formed in the walls of the SpA that are in connection with the intervillous space, depriving the more distal segments of SpA of blood flow and leading to local decidual necrosis in the peripheral parts of the placenta. Jauniaux and Burton put another theory forward that there may be a developmental gradient in the extent of extravillous trophoblast plugging, being the greatest in the central part of the placenta. The onset of the maternal flow into the placenta starts with unplugging of the SpA, which is more extensive in the peripheral parts and leads to a local hyperoxic environment. Higher levels of oxygen cause oxidative stress, induce a down-regulation of angiogenic growth factors and inhibit the invasive and proliferative activity of the trophoblast [
41‐
43]. These changes in peripheral SpA could explain the previous described differences in Doppler measurements between central and peripheral parts in the placenta [
9,
23] and a single placental (sono-)biopsy may not be representative for the entire vascular placental bed given the dissimilitude in impedance [
44]. However, our meta-analysis could not confirm statistically significant differences between central and peripheral SpA measurements when pooling the data. Moreover, most of the included articles in our meta-analysis did not specify where in the placenta the measurement was made [
15,
21,
25,
28] or only measured central SpA [
2,
20,
22,
26,
27].
The use of Doppler ultrasonography offers a non-invasive technique for the measurement of SpA flow [
9]. Nonetheless, results on Doppler velocimetry of SpA are controversial, and possibly explained by the bloom artefact where the Doppler signal diverges beyond the vessels walls [
10,
45]. Measurement of SpA could be challenging due to their small diameter and torturous character, although modernistic improvement in ultrasound imaging and the validation of protocols may overcome this problem [
15,
46,
47]. Inter-observer reproducibility of uterine measurements is strongly correlated with the experience of the ultra-sonographer, so presumptively this also applies for SpA [
46]. Tekay et al. [
48] found that the main sources of intra-observer variance in uterine measurements include maternal heart rate, breathing, blood pressure and placenta localization. None of our included studies showed results for these variables. All used colour Doppler, while the use of Power Doppler could increase the precision and its clinical appliance when measuring SpA [
49].
Lost to follow up was not documented in the longitudinal designs, possibly influencing the constructed meta-regression curves. Most studies only performed measurements during first trimester, and those studies used transvaginal probes, whereas transabdominal probes were used in studies measuring during the second and third trimester. However, Marchi et al. [
46] showed that measurements of the uterine arteries during first trimester can be performed with similar results in both transabdominal and transvaginal approach. Obesity may influence transabdominal examinations quality [
15,
46]. Unfortunately, quality assessment revealed insufficient reporting of maternal weight or body mass index, which interferes with the interpretation of the measurements. Furthermore, no conformity of used ultrasound settings was observed between studies and publication bias could not be excluded given the small amount of studies that met the inclusion criteria in this review. Last, we only included studies that were written English language, which could possibly induce a source of bias. However, previous studies have concluded there is no evidence of a systematic bias from the use of language restriction in systematic review-based meta-analyses in conventional medicine [
50‐
52]. As all authors must agree with the content of the paper, all authors must at least be able to judge the content of included studies for systematic reviewing. If we would not use language restriction, it would influence the applicability and reliability of our results. Besides, out of 93 articles that were not published in English language, only 5 articles (5%) were applicable for full text analysis based on title and abstract. Given these considerations, we only included papers written in a language that could be weighed on validity by all authors.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.