Meta-analysis for the relationship between circulating pregnancy-associated plasma protein A and placenta accreta spectrum

Abstract

Background:

Changes in circulating pregnancy-associated plasma protein A (PAPP-A) have been observed in women with a placenta accreta spectrum (PAS). However, no consensus has been reached according to the previous studies. Our study investigated the relationship between circulating PAPP-A and PAS risk through a systematic review and meta-analysis.

Methods:

Studies comparing the circulating level of PAPP-A between pregnant women with and without PAS were obtained by searching the Medline, Cochrane Library, Embase, CNKI, and Wanfang databases from the inception of the databases until February 12, 2023. Heterogeneity was considered in the pooling of results via a random-effects model.

Results:

Eight observational studies were obtained for the meta-analysis, which included 243 pregnant women with PAS and 1599 pregnant women without PAS. For all these women, the first-trimester circulating level of PAPP-A was measured by immunoassay and reported as multiples of the median (MoM) values. The pooled results showed that compared to those who did not develop PAS, women with PAS had significantly higher first-trimester serum level PAPP-A (mean difference: 0.43 MoM, 95% confidence interval [CI]: 0.30 to 0.56, P < .001; I² = 32%). Furthermore, a high first-trimester serum PAPP-A level was related to a high PAS risk (odds ratio: 2.89, 95% CI: 2.13 to 3.92, P < .001; I² = 0%). Sensitivity analysis which excluded one study at a time, also obtained similar results (p all < 0.05).

Conclusion:

Pregnant women with a high serum PAPP-A level in the first trimester may be at an increased risk for PAS.

1. Introduction

Placenta accreta spectrum (PAS) is a severe pregnancy disorder caused by the placenta adhering too deeply and too firmly to the uterus.^[1] Clinically, PAS is associated with a high risk of intrapartum bleeding and postpartum hemorrhage that can be life-threatening.^[2] Furthermore, women with PAS are also associated with a higher incidence of various adverse clinical outcomes, such as disseminated coagulopathy, acute respiratory and renal failure, and even death.^[2] Pathologically, the placenta attaches to the uterus at different depths in PAS, which could be referred to as placenta accreta, placenta increta, or placenta percreta.^[3] Previous studies showed that the prevalence of PAS varies due to heterogeneous definitions and variable prenatal detection rates in different countries.^[4] Earlier statistics from developing countries showed a prevalence of PAS at 1 in 500 pregnant women, which is reported to have increased compared to the data from the 1980s.^[5] Placenta previa, previous cesarean delivery, uterine surgery history, assisted reproductive technologies, and advanced maternal age are all known risk factors for PAS.^[6] Regarding the treatment of women with PAS, although strategies such as prophylactic balloon occlusion of internal and common iliac arteries effectively minimize the volume of blood loss,^[7] the key to successful treatment relies on early identification of high-risk women.

Preoperative management of PAS can be facilitated with antenatal and type diagnoses. However, radiological strategies such as ultrasound technology and magnetic resonance imaging (MRI) may not be adequate for diagnosing PAS.^[8] On the other hand, the growing body of evidence suggests that despite the availability of imaging modalities to confirm the diagnosis of PAS,^[9] biomarkers are also important in predicting the risk of PAS during pregnancy because these markers are convenient to test and compare with standard threshold, and could be repeated during pregnancy.^[10] Among them, the pregnancy-associated plasma protein A (PAPP-A), a zinc-binding metalloproteinase, has been associated with various adverse clinical outcomes in pregnancy, such as preeclampsia,^[11] gestational diabetes mellitus,^[12] fetal growth restriction,^[13] and even fetal loss.^[14] More importantly, preclinical studies suggest that PAPP-A may be involved in trophoblast invasion and placental growth, raising the hypothesis that PAPP-A changes may be a PAS biomarker.^[15] Interestingly, changes in circulating PAPP-A have been noticed in women with PAS.^[16] However, the results of previous studies were inconsistent.^[17–24] Therefore, our study investigated the relationship between circulating PAPP-A and PAS risk through a systematic review and meta-analysis.

2. Methods

This study followed the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA 2020) guidelines^[25,26] and the Cochrane Handbook for Systematic Reviews and Meta-analyses.^[27]

This study did not involve the examination or treatment of patients or review of patient records, so it was exempt from review and approval by our research ethics committee.

2.1. Literature search

Studies were found by searching Medline, Cochrane Library, Embase, China National Knowledge Infrastructure (CNKI), and Wanfang with a combined keyword strategy: “pregnancy-associated plasma protein A” OR “pregnancy-associated plasma protein” OR “PAPP-A”; and “placenta accreta spectrum” OR “placenta increta” “placenta accreta” OR “placenta percreta” OR “abnormally invasive placenta” OR “morbidly adherent placenta” OR “placenta.” A comprehensive literature search strategy allowed us to avoid overlooking potentially relevant studies. Additionally, we complemented this process by handpicking citations from related original and review articles. The search was from the databases’ inception to February 12, 2023.

2.2. Selection of studies

The following criteria were used to include studies in the analysis: full articles; pregnant women with and without PAS were included as cases and controls; measured the circulating level of PAPP-A in cases and controls; and reported the difference in circulating PAPP-A levels between cases and controls or the association between PAPP-A and the odds of PAS. The diagnosis of PAS was made according to the criteria of the included studies. No restriction was applied to the publication language, and we did not apply limitations to the year of publication. Only clinical studies involving human subjects published as full-length articles in peer-reviewed journals were considered. Grey literature such as unpublished data, posters, and conference papers were not included because these data are usually not peer-reviewed, which may affect the reliability of the results. Preclinical studies, review articles, studies without PAS cases, studies with no PAS controls, or studies that failed to report the circulating level of PAPP-A were excluded. If studies with overlapped patients were obtained, the studies with the largest sample sizes were used for the subsequent meta-analysis.

2.3. Data extraction and study quality evaluation

Endnote bibliographic software was used to avoid duplication. Data were gathered, assessed, and analyzed by 2 independent authors. Disagreements were resolved through discussions with the corresponding author. We extracted data about: author, year and location of the study; characteristics of participants, such as the number of women with and without PAS, as well as their mean ages; methods for validating the diagnosis of PAS in each study; source and characteristics of controls; timing of blood sampling, type of sampling, and methods for measuring PAPP-A levels; cutoff values for defining high PAPP-A levels in the studies reported the association between PAPP-A and the odds of PAS; and variables matched or controlled between cases and controls. We assessed study quality using the Newcastle-Ottawa Scale (NOS)^[28] based on 3 broad criteria: selection of cases and controls, comparability between groups, and exposure measurement. Study quality was measured by a total score between 1 and 9, with a higher score indicating better study quality.

2.4. Statistical methods

The difference in circulating PAPP-A levels between pregnant women with and without PAS was presented as a mean difference (MD) and the corresponding 95% confidence interval (CI). For studies that reported PAPP-A levels as categorized variables, the relationship between high PAPP-A and the odds of PAS was analyzed as odds ratio (OR) and the corresponding 95% CI.^[29] Data with 95% CIs or P values were used to calculate ORs and standard errors (SEs), and then a logarithmic transformation was performed to maintain normalized variance and distribution.^[27]

As described previously, we tested for homogeneity between studies using the Cochrane Q test and the I² statistic.^[29,30] An I² > 50% reflects significant heterogeneity. A random-effects model was used to pool the results and account for heterogeneity.^[27] The results were assessed for stability by excluding one study at a time in a sensitivity analysis. An examination of funnel plots for symmetry determined whether there was publication bias.^[31] Additionally, Egger regression analysis was conducted to test for publication bias.^[31] RevMan (Version 5.1; Cochrane Collaboration, Oxford, UK) and Stata (Version 17.0; Stata Corporation, College Station, TX) were applied for these statistical analyses. A P value of over .05 indicates statistical significance.

3. Results

3.1. Literature review and study identification

As shown in Figure 1, the search of electronic databases led to the retrieval of 1192 articles; after removing duplicate publications, 995 remained. As a result of noncompliance with the meta-analysis criteria, 975 titles and abstracts were excluded from the meta-analysis. Among the remaining 20 studies, 12 articles were excluded based on the reasons outlined in Figure 1 after 2 independent reviewers read the full texts. Thus, the meta-analysis included 8 observational studies.^[17–24]

Figure 1.

Diagram illustrating the process of searching databases and identifying studies.

3.2. Study characteristics

Table 1 summarizes the characteristics of the studies included in the meta-analysis. In general, 7 retrospective studies^{[17,18,20–24]} and one prospective study,^[19] consisting of 243 pregnant women with PAS and 1599 pregnant women without PAS, were available for the subsequent meta-analysis. These studies were conducted in the United States, the United Kingdom, Turkey, Russia, and China between 2014 and 2022. The mean ages of the included women were 30 to 35 years. The diagnosis of PAS was histologically validated in 7 studies^[17,19–24] and confirmed by the medical records in one study.^[18] First-trimester serum PAPP-A levels were measured with an immunoassay for all included studies and reported as multiples of median (MoM) values. All included studies reported differences in the serum PAPP-A levels between women with and without PAS.^[17–24] Five studies also reported the association between high first-trimester serum PAPP-A levels and the odds of PAS.^{[18,20,22–24]} The cutoffs for defining high first-trimester serum PAPP-A levels varied among the included studies. Gestational age (GA) was matched or adjusted between women with and without PAS among included studies, and other factors, such as maternal age, body weight, smoking, diabetes, and history of previous cesarean section, were also adjusted to varying degrees in some of the included studies.^[17–21,23] Generally, the included studies received 7 to 9 stars, which indicates good quality (see details in Table 2).

Table 1.

Characteristics of the included studies.

Study	Country	Study design	Maternal age (yr)	Diagnosis of PAS	Source of control	No. of women with PAS	No. of control	Timing of sampling	Type of sampling	Methods for PAPP-A measuring	Cutoff for PAPP-A	Variables matched or adjusted
Desai, 2014[17]	USA	Retrospective	33.5	Histologically validated	Pregnant women without PAS	16	66	First trimester	Serum	Immunoassay	NA	Maternal age and GA
Thompson, 2015[19]	UK	Prospective	34	Histologically validated	Pregnant women without PAS	17	344	First trimester	Serum	Immunoassay	NA	Ethnicity, maternal cigarette smoking, maternal weight, GA, and conception use
Lyell, 2015[18]	USA	Retrospective	NR	Medical records validated	Pregnant women without PAS	37	699	First trimester	Serum	Immunoassay	≥2.63 (95th %)	GA, maternal weight, race/ethnicity, smoking status, and preexisting DM
Büke, 2018[20]	Turkey	Retrospective	30.7	Histologically validated	Pregnant women without PAS	19	69	First trimester	Serum	Immunoassay	≥1	GA, maternal weight, DM, and smoking status
Penzhoyan, 2019[21]	Russia	Retrospective	34.9	Histologically validated	Pregnant women without PAS	25	39	First trimester	Serum	Immunoassay	NA	Maternal age, GA, and BMI
Borovkov, 2020[22]	Russia	Retrospective	NR	Histologically validated	Pregnant women without PAS	75	150	First trimester	Serum	Immunoassay	≥1.41 (ROC derived)	GA
Wang, 2021[23]	China	Retrospective	30.2	Histologically validated	Pregnant women without PAS	35	122	First trimester	Serum	Immunoassay	≥1	Maternal age, GA, BMI, smoking, and previous cesarean section history
Wu, 2022[24]	China	Retrospective	31.2	Histologically validated	Pregnant women without PAS	19	120	First trimester	Serum	Immunoassay	≥1.21 (ROC derived)	GA

BMI = body mass index, DM = diabetes mellitus, GA = gestational age, NA = not applicable, NR = not reported, PAPP-A = pregnancy-associated plasma protein A, PAS = placenta accreta spectrum, ROC = receiver-operating characteristic analysis.

Table 2.

Evaluation of study quality via the Newcastle-Ottawa Scale.

	Adequate definition of the cases	Representativeness of the cases	Selection of Controls	Definition of Controls	Controlled for GA	Controlled for other factors	Ascertainment of the exposure	The exact method of ascertainment of exposure for cases and controls	Non-response rate	Overall
Desai, 2014[17]	0	1	1	1	1	0	1	1	1	7
Thompson, 2015[19]	1	1	1	1	1	1	1	1	1	9
Lyell, 2015[18]	0	1	1	1	1	1	1	1	1	8
Büke, 2018[20]	0	1	1	1	1	1	1	1	1	8
Penzhoyan, 2019[21]	0	1	1	1	1	0	1	1	1	7
Borovkov, 2020[22]	0	1	1	1	1	0	1	1	1	7
Wang, 2021[23]	0	1	1	1	1	0	1	1	1	7
Wu, 2022[24]	0	1	1	1	1	0	1	1	1	7

GA = gestational age.

3.3. Meta-analysis results

The pooled results of 8 studies^[17–24] showed that compared to those who did not develop PAS, women with PAS had significantly higher first-trimester serum PAPP-A levels (MD: 0.43 MoM, 95% CI: 0.30 to 0.56, P < .001; Fig. 2A) with mild heterogeneity (p for Cochrane Q test = 0.17; I² = 32%). The sensitivity analysis, performed by excluding one study at a time, produced consistent results (p all < 0.05; Table 3), which suggested the robustness of the finding. Furthermore, 5 studies also reported the association between high first-trimester serum PAPP-A and the odds of PAS.^{[18,20,22–24]} The pooled results of these studies showed that a high first-trimester serum PAPP-A level was significantly associated with a higher risk of PAS (OR: 2.89, 95% CI: 2.13 to 3.92, P < .001; Fig. 2B) without significant heterogeneity (p for Cochrane Q test = 0.95; I² = 0%). A consistent result was also obtained by excluding one study at a time in the sensitivity analysis (p all < 0.05; Table 4).

Figure 2.

Forest plots representing the association between first-trimester serum PAPP-A levels and the risk of PAS. (A) forest plots for the meta-analysis of the difference in first-trimester serum PAPP-A levels between pregnant women with and without PAS; and (B) forest plots for the meta-analysis of the association between high first-trimester serum PAPP-A levels and the risk of PAS. PAPP-A = pregnancy-associated plasma protein A, PAS = placenta accreta spectrum.

Table 3.

Results of the sensitivity analysis for the difference of PAPP-A between women with and without PAS.

Study excluded	MD (95% CI)	I2	P
Desai, 2014[17]	0.40 [0.28, 0.53]	21%	<.001
Thompson, 2015[19]	0.46 [0.31, 0.60]	35%	<.001
Lyell, 2015[18]	0.42 [0.26, 0.57]	38%	<.001
Büke, 2018[20]	0.44 [0.30, 0.58]	41%	<.001
Penzhoyan, 2019[21]	0.46 [0.32, 0.59]	33%	<.001
Borovkov, 2020[22]	0.40 [0.28, 0.52]	16%	<.001
Wang, 2021[23]	0.44 [0.29, 0.58]	42%	<.001
Wu, 2022[24]	0.47 [0.34, 0.60]	22%	<.001

CI = confidence interval, MD = mean difference, PAPP-A = pregnancy-associated plasma protein A, PAS = placenta accreta spectrum.

Table 4.

Results of the sensitivity analysis for the association between high PAPP-A and the risk of PAS.

Study excluded	OR (95% CI)	I2	P
Lyell, 2015[18]	2.84 [2.06, 3.92]	0%	<.001
Büke, 2018[20]	2.93 [2.13, 4.02]	0%	<.001
Borovkov, 2020[22]	3.23 [2.04, 5.10]	0%	<.001
Wang, 2021[23]	2.76 [1.96, 3.88]	0%	<.001
Wu, 2022[24]	2.88 [2.10, 3.94]	0%	<.001

CI = confidence interval, OR = odds ratio, PAPP-A = pregnancy-associated plasma protein A, PAS = placenta accreta spectrum.

3.4. Publication bias

The funnel plots for the meta-analyses indicating the difference in serum PAPP-A levels and the association between PAPP-A and odds of PAS are shown in Supplemental Figures 1A and 1B, http://links.lww.com/MD/J380. Considering the symmetry of these plots, there is little risk of publication bias. Additionally, Egger regression tests did not indicate publication bias (P = .48 and 0.71, respectively).

4. Discussion

In this systematic review and meta-analysis, after combining and analyzing the results of 8 eligible observational studies, it was shown that compared to controls (pregnant women without PAS), the women who developed PAS had significantly higher first-trimester serum PAPP-A levels. In addition, an elevated serum PAPP-A level was related to higher odds of PAS in pregnant women. These results were consistent with the sensitivity analyses performed by excluding one study at a time. These findings suggest that a high first-trimester serum PAPP-A level may be associated with the risk of PAS in pregnant women.

As far as we know, this study is the first meta-analysis to summarize the relationship between PAPP-A and PAS. The methodological strengths of the meta-analysis included the following: First, 5 commonly used electronic databases were searched with extensive literature search strategies, which obtained up-to-date literature regarding the association between serum PAPP-A levels and PAS in pregnant women. Second, GA was matched or adjusted in all included studies when the association between serum PAPP-A level and PAS was analyzed, which, therefore, could minimize the possible confounding effect of different GA between cases and controls. This is important because the maternal serum PAPP-A level has been shown to be strongly correlated with GA despite the different immunoassays used.^[32] Furthermore, the potential association between first-trimester serum PAPP-A levels and the risk of PAS was evaluated with the PAPP-A analyzed both in the continuous and categorized variables. The consistent results of the 2 meta-analyses further validated the reliability of the findings. Finally, sensitivity analyses performed by omitting one study at a time produced similar results, further confirming the robustness of the data. These results suggest that a rise in serum PAPP-A during the first trimester of pregnancy correlates with the development of PAS. Further studies are needed to explore the optimal cutoff points of serum PAPP-A levels in the first trimester as a predictor of PAS. Besides, studies are also required to explore whether the combination of first-trimester serum PAPP-A levels with imaging modalities, such as ultrasound and magnetic resonance imaging, could improve the early diagnosis of PAS.

Whether increased serum PAPP-A in the first trimester of pregnancy plays a role in PAS pathogenesis or is simply a biomarker is still unknown. Physiologically, placental syncytiotrophoblasts secrete PAPP-A, a zinc metalloproteinase, into the maternal circulation in increasing concentrations until birth.^[33] Despite its poorly understood function, PAPP-A is known to participate in the proteolysis of IGF-binding protein 4 into insulin-like growth factor (IGF), suggesting the role of PAPP-A in placental growth.^[33] Moreover, PAPP-A has also been proposed as a potential marker of healthy placental trophoblasts.^[15] Consequently, it could be hypothesized that PAPP-A is overexpressed in the first trimester and may be involved in trophoblast invasion, which may lead to the pathogenesis of PAS.^[34] Future studies should aim to validate the above hypothesis and elucidate the potential molecular mechanisms underlying the association between high first-trimester PAPP-A levels and PAS.

Although radiological strategies such as ultrasound technology are still a first-line resource for diagnosing PAS, early prediction of PAS in pregnancy with ultrasound mainly relies on the position of the gestational sac,^[35] which may not provide a sufficient diagnosis for PAS. The use of MRI in patients with suspected PAS and as part of the assessment of preoperative implantation is still considered an essentia diagnostic tool. Still, its actual performance needs to be verified by independent studies.^[36] Accordingly, biochemical prediction of the risk of PAS using biomarkers has been proposed because of their advantages such as convenience, repeatability, and easiness for comparison, etc.^[10] However, the shortcoming of using biomarkers for the diagnosis of PAS is also evident because these biomarkers are mostly not specific to PAS but also associated with other comorbidities and adverse complications during pregnancy.^[16] Accordingly, a combined prediction and diagnostic strategy with biomarkers and imaging may be optimal for the early identification of PAS, and further studies are warranted.^[37]

There are limitations to our study. First, there were a limited number of included studies and participants; validating the findings in large-scale research is essential. Second, most of the included studies were retrospective. Large-scale prospective studies are necessary to determine whether a high first-trimester serum PAPP-A level is independently associated with PAS. Furthermore, as a meta-analysis of observational studies, this study could not confirm a causal relationship between a high PAPP-A level and the development of PAS during pregnancy. Finally, although some potential confounding factors were matched or controlled between women with and without PAS, there may still be residual factors that affect the association between serum PAPP-A and PAS.

In conclusion, women with PAS during pregnancy are associated with significantly high serum PAPP-A levels in the first trimester. It is necessary to conduct more studies to determine whether measuring serum PAPP-A in the first trimester is helpful for the early diagnosis of PAS during pregnancy.

Author contributions

Investigation: Yan Li.

Methodology: Yan Li.

Project administration: Jin He.

Resources: Yan Li.

Software: Yan Li, Yizi Meng, Ping Li.

Supervision: Jin He.

Visualization: Yan Li, Yizi Meng, Ping Li.

Validation: Jin He.

Writing – original draft: Yan Li.

Writing – review & editing: Yizi Meng, Yang Chi.

Supplementary Material

medi-102-e34473-s001.pdf ^{(262.3KB, pdf)}