First-trimester maternal serum PAPP-A levels and hyperemesis gravidarum: unraveling the link – a meta-analysis

Eligibility criteria, information sources, search strategy

For this review, two reviewers, A.V. and V.L., systematically searched databases, including PubMed/MEDLINE, ScienceDirect, the Cochrane Library, and Google Scholar in February 2025 to identify relevant studies investigating the association between first-trimester maternal serum pregnancy-associated plasma protein-A (PAPP-A) levels and hyperemesis gravidarum (HG). The authors predetermined the eligibility criteria. This systematic review encompassed all observational studies involving human patients, including prospective and retrospective cohort studies, case-control studies, as well as cross-sectional studies, that examined the association of hyperemesis gravidarum and PAPP-A levels, measured at the first-trimester screening. The included studies examined both spontaneous and in vitro fertilization (IVF) pregnancies. Restrictions were applied to enhance the relevance and validity of the findings. Studies that did not report direct comparisons between HG and non-HG groups were excluded. Studies such as case reports, small case series, letters to the editor, animal studies, and review articles were excluded. Additionally, conference abstracts and proceedings were not considered, as they often lack critical details necessary for evaluating study limitations and the quality of evidence. Additionally, studies with insufficient data on PAPP-A levels or those reporting only adjusted associations without raw biomarker values were omitted. Articles were screened for eligibility using title and abstract review, followed by full-text assessment. Furthermore, only studies published in English were considered. This review is based on previously published, aggregated data and does not constitute a randomized controlled trial (RCT). PICO criteria for this study involve: P (Population): Pregnant women undergoing first-trimester screening; I (Intervention/Exposure): Women diagnosed with hyperemesis gravidarum; C (Comparison): Pregnant women without hyperemesis gravidarum (control group); O (Outcome): Comparison of PAPP-A levels between the two groups.

Our search included the following strategy: (Pregnancy-associated Plasma Protein-A) OR (PAPP-A) AND (hyperemesis gravidarum) and is presented in Figure 1. The final studies included, ranged from 2005 to 2025. The two independent reviewers worked together to identify eligible studies and reached a consensus on the final selection. They individually evaluated the search results, and any discrepancies were addressed through discussions with a third author (V.P.) before making the final decisions.

Figure 1:

PRISMA guidelines.

Study selection

The study selection process was carried out in three sequential stages. First, duplicate articles were eliminated. Then, reviews, animal studies, case reports and conference abstracts were excluded. Subsequently, two authors (A.V. and V.L.) independently screened the abstracts of the remaining articles to evaluate their relevance. Finally, full-text reviews were conducted for studies deemed potentially eligible, and only those meeting the inclusion criteria were retained. Any disagreements in the final selection were resolved through consensus among all authors.

Data extraction

The main objective of this study was to evaluate the relationship between first-trimester PAPP-A levels and hyperemesis gravidarum. Studies that examined hyperemesis gravidarum in relation to other factors or lacked comparable data were excluded. Ultimately, a total of six studies were included in the meta-analysis. The outcome measures were defined at the outset during the study’s initial design.

Quality assessment

The methodological quality of all included studies was assessed using the Newcastle–Ottawa Scale (NOS), a widely recognized tool for evaluating the quality of non-randomized studies in systematic reviews and meta-analyses. This tool evaluates studies based on eight criteria, categorized into three main domains: selection of study groups, comparability of groups (considering maternal weight and gestational age), and assessment of either exposure (in case-control studies) or outcome (in cohort studies). A star-based rating system was applied to provide a quick visual representation of study quality, with the highest-quality studies receiving a maximum of nine stars. Two authors (A.V. and V.L.) independently conducted the quality assessment, and any disagreements were resolved through discussion with a third author (D.V.). The overall risk of bias was classified as good, fair, or poor (Figure 2).

Figure 2:

Newcastle–Ottawa assessment scale of the included studies. √: the assessed quality item was identified and confirmed within the study (applicable to selection and outcome categories); √: the study groups were adjusted for a single key factor; √√: the study groups were adjusted for two key factors.

Data synthesis, prediction intervals and trial sequential analysis

A statistical meta-analysis was conducted using RStudio with the meta function [RStudio Team (2015). RStudio: Integrated Development for R. RStudio, Inc., Boston, MA. URL http://www.rstudio.com/]. Given the substantial methodological heterogeneity among the included studies (Table 1), statistical heterogeneity was not factored into the selection of the appropriate analytical model (fixed-effects or random-effects), as assuming comparable effect sizes across studies was not feasible. Confidence intervals were set at 95 %. Pooled odds ratios (OR) and their corresponding 95 % confidence intervals (CI) were estimated using the Hartung–Knapp–Sidik–Jonkman method instead of the traditional Dersimonian–Laird random-effects model (REM). This approach was chosen based on recent evidence suggesting its superiority, particularly when dealing with studies of varying sample sizes and considerable between-study heterogeneity.

Prediction intervals (PI) were also determined using the meta function in RStudio to assess the anticipated effect size in future studies within this field. This estimation incorporates the variability between studies, representing the existing heterogeneity on the same scale as the outcome being analysed.

To assess the required information size, trial sequential analysis (TSA) was conducted. This method allows for the evaluation of type I error in the pooled results of meta-analyses focusing on the primary outcomes predefined in this study. A minimum of three studies was considered necessary to perform the analysis. Since repeated significance testing in meta-analyses increases the likelihood of type I errors, TSA adjusts the significance threshold using the O’Brien–Fleming α-spending function. Through sequential interim analyses, TSA evaluates the influence of each individual study on the overall meta-analysis findings. The risk of type I error was set at 5 %, while the type II error threshold was established at 20 %. TSA was conducted using TSA version 0.9.5.10 Beta software (http://www.ctu.dk/tsa/).

Study selection and methodological characteristics

The extensive literature search resulted in 1,520 records: three from PubMed, 908 from ScienceDirect, 0 from the Cochrane Library, and 609 from Google Scholar. After eliminating 500 duplicate entries, 650 animal studies, 300 reviews, 24 conference abstracts, editorials and case reports, we screened 46 original studies based on their titles and abstracts. Out of those, 38 were excluded as they were not relevant to our study and two could not be retrieved. Lastly, six studies, published between 2005 and 2025, were included in our meta-analysis that involved 1,039 pregnant women, 501 with hyperemesis gravidarum and 538 controls, that underwent the first-trimester screening.

In Table 1, methodological characteristics of the included studies are depicted. The timing of sample collection for the assessment of PAPP-A occurred during the first-trimester screening, between 10 and 14 weeks of gestation, in all the included studies. The mean maternal age varied across studies but was comparable between HG and control groups. Three studies [11], 14], 16] found significantly higher PAPP-A levels in the hyperemesis gravidarum group, whereas others reported no significant difference.

Synthesis of results

This meta-analysis supports the hypothesis that PAPP-A levels are elevated in pregnancies affected by hyperemesis gravidarum. The forest plot in Figure 3 summarizes the findings from the six included studies, subgrouped as prospective and retrospective studies, displaying individual mean differences (MD) with 95 % confidence intervals (CI), along with the overall pooled estimate. The pooled mean difference random effects models was 0.12 (95 % CI: 0.00–0.24) for the prospective studies and 0.22 (95 % CI: 0.11–0.34) for the retrospective studies, indicating that PAPP-A levels were significantly higher in women with hyperemesis gravidarum compared to the control group. Notably, individual study estimates varied, with Abdel-Dayem et al. [13] reporting a non-significant mean difference (0.02, 95 % CI: −0.12–0.16), while Çakır et al. [14] reported the largest effect (0.36, 95 % CI: 0.07–0.65). The heterogeneity statistic (I²=46 % – I²=52 % for the prospective studies and I²=7 % for the retrospective studies) suggests moderate variability among studies, but it is not high enough to indicate strong inconsistency. The prediction interval (−0.08–0.4) suggests that while most future studies are expected to show a positive association, some may yield non-significant or even negative results.

Figure 3:

Forest plots of odds ratio for PAPP-A levels in the first-trimester screening in women with hyperemesis gravidarum compared to controls with 95 % confidence intervals (CI) and weighted pooled summary statistics using bivariate random-effects model. The studies are subgrouped into prospective and retrospective. Vertical line = “no difference” point between the two groups. Green squares=risk ratios; diamonds=pooled risk ratios and 95 confidence intervals for all studies; horizontal black lines=95 % CI; horizontal red line=pooled 95 % prediction intervals).

In order to evaluate the evidential value and potential bias, p-curve analysis was conducted, as shown in Figure 4. The observed p-curve (blue line) demonstrated a right-skewed distribution, with all statistically significant p-values (p<0.05) falling below 0.025. This right-skewness test was statistically significant (p=0.033, p=0.0462), suggesting that the results are likely due to true effects rather than selective reporting or p-hacking. The estimated statistical power of the included studies was 69 % (95 % CI: 20–94 %), indicating moderate power, though with some uncertainty. The lack of an excess of p-values between 0.03 and 0.05 reduces concerns about selective reporting bias. However, two additional non-significant results (p>0.05) were excluded from the p-curve, indicating that some studies did not find a significant association. Overall, the p-curve analysis supports the presence of a genuine association between PAPP-A levels and hyperemesis while highlighting the need for further studies with larger sample sizes to improve statistical power.

Figure 4:

p-Curve analysis demonstrating the distribution of significant p-Values from the included studies. The right-skewed curve suggests evidential value for a real effect, while power is estimated at 69 % (95 % CI: 20–94 %).

A contour-enhanced funnel plot, depicted in Figure 5, was used to assess publication bias in the meta-analysis of PAPP-A levels and hyperemesis gravidarum. The plot showed a reasonably symmetric distribution of studies around the pooled mean difference (∼0.16), suggesting minimal publication bias. Most studies fell within regions indicating statistical significance (p<0.05), further supporting the robustness of the findings. However, a slight asymmetry, with fewer studies reporting negative or null results, suggests the potential underrepresentation of small, non-significant studies. Despite this, the overall pattern does not indicate a strong small-study effect, reinforcing the reliability of the meta-analysis conclusions.

Figure 5:

Contour-enhanced funnel plot for publication bias assessment in the meta-analysis of PAPP-A levels and hyperemesis gravidarum. The X-axis represents the mean difference, and the Y-axis represents the standard error. The dashed triangle indicates expected study distribution in the absence of bias. Shaded regions represent significance thresholds (darker=more significant). The plot suggests minimal publication bias, though a slight asymmetry indicates potential underreporting of small, non-significant studies.