Association between transabdominal uterine artery Doppler and small-for-gestational-age: a systematic review and meta-analysis

Background The association between uterine artery Doppler (UtA) measurements and small for gestational age (SGA) has not been quantitatively analyzed throughout the whole pregnancy. This systematic review and meta-analysis aims to comprehensively explore the association between UtA measurements and SGA in the first, second, and third trimesters. Methods Studies were searched from Pubmed, Embase, Cochrane Library, and Web of Science. Weighted mean difference (WMD), odds ratio (OR), and relative risk (RR) with 95% confidence interval (CI) were used as the effect size. Heterogeneity of all effect sizes was tested and quantified using I2 statistics. Sensitivity analysis was conducted for all outcomes, and publication bias was evaluated using Begg’s test. Results A total of 41 studies were finally included in our meta-analysis. In the first trimester, mean PI was significantly higher in the SGA group than the non-SGA group (WMD: 0.31, 95%CI: 0.19–0.44). In the second trimester, odds of notch presence (OR: 2.54, 95%CI: 2.10–3.08), mean PI (WMD: 0.21, 95%CI: 0.12–0.30), and mean RI (WMD: 0.05, 95%CI: 0.05–0.06) were higher in the SGA group. Also, abnormal UtA measurements were associated with the increased odds of SGA (all P < 0.05). In the third trimester, PI z-score (WMD: 0.62, 95%CI: 0.33–0.91) and PI MoM (WMD: 0.08, 95%CI: 0.06–0.09) showed a significant increase in the SGA group. The odds of SGA were higher in the women with mean PI > 95% (OR: 6.03, 95%CI: 3.24–11.24). Conclusions Abnormal UtA measurements were associated with high odds of SGA, suggesting that UtA might be an adjunctive screening method for SGA in the whole pregnancy. Supplementary Information The online version contains supplementary material available at 10.1186/s12884-023-05968-w.


Background
Small for gestational age (SGA) refers to the fetal birth weight ≤ 10th percentile according to local standards [1].Fetuses with late-onset SGA have a high risk of adverse perinatal outcomes, such as high rate of surgical delivery, low Apgar and arterial cord blood pH values, and high frequency of neonatal unit (NNU) admission [1].Screening for SGA is a key element of prenatal care [2].
Uterine artery Doppler (UtA) has been used to assess the risk of SGA in pregnant women [3].Studies on the association between UtA and the risk of SGA have been previously reported.He et al. have found that mean UtA-pulsatility index (UtA-PI) and UtA-resistance index (UtA-RI) were higher in the SGA fetuses compared to non-SGA fetuses [4].Običan et al. have reported that abnormal UtA indices were significantly correlated with an increased risk of SGA [5].Left uterine artery notching and PI > 95th percentile increased 1.76-fold and 1.83-fold risk of SGA, respectively [5].However, several limitations existed in the separate original studies, including insufficient sample size or being limited to one region.
Meta-analysis is a powerful tool to combine results from two or more separate studies, which shows a good evidence strength and facilitates healthcare decisionmaking [6,7].A systematic review by Meler et al. have already reported the association between UtA and SGA, while the results are not quantitatively analyzed [8].Cnossen et al. performed a meta-analysis to explore the predictive accuracy of UtA for SGA in the first and second trimesters, while they did not focus on the third trimester [9].The persistent increase in the uterine artery impedance in the third trimesters increased the risk of SGA [8].UtA examination can be conducted in transvaginal and transabdominal approaches, and transabdominal approach is recommended because most of the studies evaluating the UtA in the third trimester used a transabdominal approach [10].Therefore, we performed a systematic review and meta-analysis based on the previously published studies to comprehensively explore the association between transabdominal UtA measurements and the risk of SGA in the first, second, and third trimesters.

Literature search strategy
This meta-analysis was performed based on the Preferred Reporting Items for Systemic Reviews and Meta-Analyses (PRISMA) guideline [11].Pubmed, Embase, Cochrane Library, and Web of Science were searched by two researchers (RJZ and XPT) for relevant studies up to July 28, 2022.The search terms used were "Uterine Artery" OR "Arteries, Uterine" OR "Artery, Uterine" OR "Uterine Arteries" AND "Ultrasonography, Doppler" OR "Doppler Ultrasound" OR "Doppler Ultrasounds" OR "Ultrasound, Doppler" OR "Ultrasounds, Doppler" OR "Doppler Ultrasonography" OR "Doppler Ultrasound Imaging" OR "Doppler Ultrasound Imagings" OR "Imaging, Doppler Ultrasound" OR "Imagings, Doppler Ultrasound" OR "Ultrasound Imaging, Doppler" OR "Ultrasound Imagings, Doppler" OR "PI" OR "pulsatility index" OR "RI" OR "resistance index" OR "blood flow index" OR "diastolic notch" OR "blood flow score" OR "ratio of systolic and diastolic blood flow velocity" OR "the ratio of systolic peak value and end diastolic velocity of blood flow" OR "S/D" OR "systolic maximum flow velocity" OR "Systolic low velocity" OR "diastolic minimum flow velocity" OR "Diastolic flow velocity" AND "Infant, Small for Gestational Age" OR "Small for Gestational Age" OR "SGA" OR "Fetal Growth Retardation" OR "Intrauterine Growth Retardation" OR "Growth Retardation, Intrauterine" OR "Intrauterine Growth Restriction" OR "Fetal Growth Restriction".We have registered this systematic review and meta-analysis with PROSPERO (registration number: CRD42023447101).

Inclusion and exclusion criteria
Only studies meeting the following criteria were included: (1)  The UtA parameters we observed in this study were mean UtA-PI, mean UtA-RI, multiple of median (MoM) values of UtA-PI (UtA-PI MoM), UtA-PI z-score, and notch presence.PI was calculated as (peak systolic velocity-end diastolic velocity)/average velocity, and RI was calculated as (peak systolic velocity-end diastolic velocity)/peak systolic velocity.Mean PI and RI were the average from the left and right uterine arteries [12].Notch presence meant the unilateral or bilateral notch in the diastolic notch [13].Abnormal UtA includes: the presence of diastolic notch, high RI (RI > 95%, RI > 75%, RI > 90%), or high PI (PI > 95%) [13][14][15][16].SGA was defined as the fetal birth weight ≤ 10th percentile according to local standards [1].
The following exclusion criteria were adopted: (1) animal studies; (2) studies irrelevant to the topic (studies not on transabdominal UtA or SGA definition not conformed); (3) reviews, meta-analyses, case reports, protocols, conference abstracts, guidelines, and expert consensus; (4) not published in English.

Data extraction
Two researchers (RJZ and XPT) independently evaluated the data suitable for this meta-analysis, and extracted the following information: the first author, publication year, country, study design, group, sample size, age, body mass index (BMI), birth weight, gestational age, complications, smoking, and Doppler time.If conflicts existed, a third researcher (HEL) provided the consultation.

Methodological quality appraisal
The quality of case-control studies and cohort studies was assessed using Newcastle-Ottawa Scale (NOS) [17].
For case-control studies, three items (selection, comparability, and exposure) were assessed.For cohort studies, three items (selection, comparability, and outcome) were evaluated.The total score of this scale was 9 points, and study quality was regarded as poor (0-3 points), fair (4-6 points), and good (7-9 points).

Statistical analysis
Weighted mean difference (WMD) with 95% confidence interval (CI) was used as the effect size for measurement data, and odds ratio (OR) with 95%CI was used as effect size for counting data.If relative risk (RR) was provided in the publications, RR was combined for analysis.Heterogeneity was tested for all effect sizes and quantified using the I 2 statistics.If the heterogeneity statistic I 2 ≥ 50%, random effect model was used for analysis; otherwise, fixed effect model was used for analysis.Sensitivity analysis was performed to assess the effect of a single study on the whole estimate by removing studies one by one.Publication bias was assessed using Begg's test for the outcomes included in more than nine studies [18].All statistical analysis was performed using Stata15.1 software (StataCorp, College Station, TX, USA), and P < 0.05 was considered to be statistically significant.

Certainty of evidence
The certainty of the evidence was assessed using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE).The GRADE system categorized the certainty of the pooled estimate of effect as high, moderate, low, or very low according to the following criteria: study design, risk of bias, inconsistency, indirectness, imprecision, and other considerations.An evidence profile was produced to summarize the results using the GRADEpro GDT (https:// gdt.grade pro.org/).

Study selection and study characteristics
A total of 7,079 studies were identified from the abovementioned databases.After removing 2,513 duplicates, 4,566 studies remained.After screening title and abstract, animal studies (n = 258), studies irrelevant to the topic (n = 3,234), and reviews, meta-analyses, case reports, protocols, conference abstracts, guidelines and expert consensus (n = 856) were eliminated.After screening the full texts, 179 studies were removed due to irrelevant to the topic (n = 153) or not published in English (n = 24).Finally, 41 studies were included in our metaanalysis (Fig. 1) [1,4,5,[12][13][14][15][16].Table 1 displays the characteristics of the included studies.There were 38    cohort studies and 3 case-control studies.According to the Newcastle Ottawa scale, 32 studies were assessed as good quality and 9 studies were assessed as fair quality (Supplementary table S1-S2).

Comparison of UtA measurements between SGA group and non-SGA group
In the first trimester, mean PI and PI z-score were significantly higher in the SGA group than in the non-SGA group (WMD: 0.31, 95%CI: 0.19-0.44;WMD: 0.30, 95%CI: 0.18-0.42)(Fig. 2A-B).There was no significant difference in mean RI between the SGA group and non-SGA group (WMD: 0.06, 95%CI: -0.04-0.16).A study by Arakaki et al. reported that RI z-score was significantly higher in the SGA group compared to non-SGA group [12].In the second-trimester, we found that the risk of notch presence in the SGA group was higher than in the non-SGA group (OR: 2.54, 95%CI: 2.10-3.08)(Fig. 2C).

Sensitivity analysis and publication bias
Sensitivity analysis showed that no study displayed an important effect on the final pooled UtA measurements and SGA incidence (Tables 1 and 2).There was no evidence of publication bias in the reporting of notch presence in the second-trimester across studies (Z = 0.36, P = 0.974).

Certainty of evidence
We used GRADE to assess the level of evidence.The results showed very low level of evidence for all outcomes (Supplementary table S3).

Discussion
In this systematic review and meta-analysis, we explored the association between transabdominal UtA and SGA in the first, second, and third trimesters.The results showed that UtA measurements in the SGA group were significantly higher than the non-SGA group during the whole pregnancy.Also, SGA group had a higher odds of notch presence than the non-SGA group.In addition, we found that abnormal UtA was associated with the higher odds of SGA compared to normal UtA in the second and third trimesters.
Transabdominal UtA is a noninvasive test of the uteroplacental circulation, and has been applied to predict the risk of SGA in the clinical practice [4,5,44].PI and RI are common observation indices in the UtA [52].UtA-PI reflects total resistance distal to the measurement point, and UtA-RI reflects the vascular resistance at the measurement point [52].A study showed a stable decrease in UtA-PI values until the late stages of pregnancy [53], whereas Cavoretto et al. found that UtA-PI showed a progressive non-linear decrease throughout the pregnancy by using fractional polynomial [54].In this meta-analysis, we observed higher levels of UtA-PI and UtA-RI in women with SGA compared to those without SGA during the whole pregnancy.Previous studies have reported the similar findings [13,31].Borna et al. performed a study to identify patients at the risk of SGA using UtA, and they observed that mean UtA-PI in women with SGA newborns was significantly higher than those without SGA newborns [13].Similarly, Maged et al. showed that UtA-RI was significantly higher in women who developed SGA compared to controls [31].Also, we found that the odds of SGA were higher in the abnormal UtA group compared to normal UtA in the second and third trimesters.This finding was consisted with the studies by Običan et al. and Groom et al. [5,15] Običan et al. suggested that the risk of SGA was significantly higher when PI > 95% [5].Groom et al. indicated that the incidence of SGA was higher in women with UtA-RI > 90% than those with normal UtA-RI in the second trimester [15].Evidence showed that trophoblastic invasion may be the reason for the increase of uterine vascular impedance; subsequently, changes in the uteroplacental circulation was detected by UtA [55,56].
Diastolic notch is the characteristic of vessels with resistance, and depends on the compliance of vessel wall [57].Dugoff et al. reported that 34.2% of 1067 American pregnant women had diastolic notches in the uterine artery; however, there was no significant association between diastolic notch and SGA [22].He et al. also reported no significant association between notch and SGA although they found notching in the SGA fetuses was 40% higher than in the non-SGA fetuses [4].One potential reason for this is that diastolic notch is dichotomous rather than numeric variables, which might introduce misclassification bias [4].In this meta-analysis, we found a significant association between the notch presence and SGA and that notch presence was significantly associated with the increased odds of SGA.The similar finding was reported in former studies [5,13,35].Borna et al. found that the incidence of SGA in women with notch was significantly greater than women without notch in ultrasonography [13].In the study by Mitsui et al., a higher incidence of SGA was found in pregnant women with notch than those without (29.2% vs. 7.7%) [35].This was consistent with the finding from the study of Običan et al. that UtA notch was significantly associated with SGA [5].
This meta-analysis explored the association between transabdominal UtA measurements and SGA in the first, second, and third trimesters, and found that abnormal UtA measurements were significantly associated with the high odds of SGA in the whole pregnancy.However, there are some limitations in this meta-analysis.First, judgement of normality or abnormality and classification of centiles in UtA measurements relies upon different curves and charts for uterine arteries, which may affect the reliability of the pooled results.In the future, a uniform judgement for abnormality needs to be explored.Second, there is heterogeneity in some results.Pregnancy complications (such as gestational hypertension, preeclampsia, gestational diabetes), maternal smoking, and history of SGA may be the sources of heterogeneity.However, we were unable to perform the subgroup analysis to explore the sources of heterogeneity because the above factors could not be analyzed based on the included studies.Third, SGA is defined as fetal birth weight ≤ 10th percentile of the standard weight of the fetus at the same gestational age.
In included studies, standard weight varies from region to region, which may cause some bias on the results.Fourth, the method of conception of the included pregnancies is likely heterogeneous.UtA-PI values are significantly different in pregnancies after different conception method [58].Fifth, due to the limitation of the included studies, we failed to explore the influence of the changes of UtA measurements on SGA in different pregnancy periods.

Conclusion
In conclusion, our meta-analysis found a significant association between abnormal UtA measurements and increased odds of SGA in the whole pregnancy, indicating that UtA might be an adjunctive screening method for SGA in the whole pregnancy.

Fig. 2
Fig. 2 Forest plots regarding to mean PI (A) and PI Z-score (B) in the first trimester; notch presence (C), mean PI (D), and mean RI (E) in the second trimester; PI z-score (F) and PI MoM (G) in the third trimester

Table 1
Characteristics of included studies

Table 1 (
continued) NA Abbreviation: BMI body mass index, GA gestational age, QA quality assessment, SGA small-for-gestational age, PIH pregnancy-induced hypertension, NA not applicable, UtA uterine artery Doppler, GH gestational hypertension, PE pre-eclampsia a Data are presented as the median (range) b Data are presented as the median (IQR)

Table 2
Comparison of UtA parameters between SGA groups and non-SGA groups Abbreviation: UtA uterine artery Doppler, SGA small for gestational age, PI pulsatility index, RI resistance index, MoM multiple of median, WMD weighted mean difference, RR relative risk, OR odds ratio, CI confidence interval a presented WMD b presented OR

Table 3
Comparison of SGA incidence between abnormal and normal UtA groupsAbbreviation: UtA uterine artery Doppler, SGA small for gestational age, PI pulsatility index, RI resistance index, WMD weighted mean difference, RR relative risk, OR odds ratio, CI confidence interval a presented OR