Pertinence between risk of preeclampsia and the renin-angiotensin-aldosterone system (RAAS) gene polymorphisms: an updated meta-analysis based on 73 studies

Abstract The aetiological mechanism of preeclampsia (PE) is unclear exactly, so we attempted to investigate the association between susceptibility to preeclampsia and renin-angiotensin-aldosterone system (RAAS) gene polymorphisms to explore the aetiology in terms of genetics. A systematic search was performed in electronic databases to identify relevant studies. Eventually 73 studies were enrolled, odds ratios were generated by 5 genetic models. In overall analysis, significant associations were detected for AGT M235T, AT1R A1166C and CYP11B2 C344T whereas negative correlation was shown for AGT T174M. As stratified by race and geography, AGT 235T allele and AT1R 1166C allele increased preeclampsia risk and AGT T174M was justified uncorrelated with preeclampsia. Our meta-analysis illustrated that AGT 235T allele and AT1R 1166C allele increased and CYP11B2 344T allele decreased preeclampsia risk while AGT T174M polymorphism did not change preeclampsia risk. Hence, pregnant women carrying high-risk genotypes need strengthened management to prevent and early identification of preeclampsia.


Introduction
Preeclampsia (PE) is one of the most serious complications specifically occurring in pregnancies, which is characterised by hypertension, new on-set proteinuria after 20 weeks of fertilisation.The incidence is approximately 2%-8% globally and 2%-6% in China and affects 3%-5% of pregnant women (Steegers et al. 2010, Qiao et al. 2013).If they couldn't receive timely and effective treatment, severe complications such as placental abruption, foetal growth restriction, eclampsia, HELLP syndrome (namely haemolysis, elevated liver enzymes and low platelets) may occur, intimidating the health of both mothers and offspring.Moreover, patients with a history of preeclampsia will suffer significantly increased risk of cardiovascular and kidney disease over the next 10 to 15 years and their offspring are at the margin of developing hypertension in early childhood and adulthood (Bellamy et al. 2007).
The pathogenesis of preeclampsia has not been fully elucidated up to now, recent research have demonstrated possible mechanisms including placental or trophoblast ischaemia and hypoxia, vascular endothelial injury, oxidative stress and inflammatory immunomodulatory disorders (Wang et al. 2009).Some epidemiological and genetic studies have confirmed that preeclampsia was associated with susceptible genes and genetic polymorphisms, but the potential mechanisms remained opaque due to complex ethnic, geographic and other factors.Therefore, research focussing on genetic profiles could provide references for the prevention and diagnosis of preeclampsia.
All components of renin-angiotensin-aldosterone (RAAS) showed compensatory and protective alterations during pregnancy whereas the balance was broken when preeclampsia occurred (Yang et al. 2013).As aforementioned, preeclampsia is a multifactorial disease with obvious genetic inclination, hence constantly increasing single nucleotide polymorphisms (SNPs) belonged to RAAS system are being explored and attached importance to it (Gathiram and Moodley 2020).Among them, the genetic polymorphisms of angiotensinogen (AGT) gene, angiotensin II type 1 receptor(AT1R) gene and aldosterone synthase gene are always hot spots in the academic field.
AGT M235T polymorphism (rs699 or T704C) is the surrogate of methionine to threonine at residue 235 and AGT T174M polymorphism (rs4762 or C521T) is methionine instead of threonine at residue 174 (Gunda et al. 2016).In addition, AT1R A1166C polymorphism (rs5186) is part of the non-coding region at 3 0 end of it and mutation occurs from adenine (A) to cytosine (C), (Campbell et al. 2018).Lastly, CYP11B2 C344T polymorphism (rs179998, thymine instead of cytosine, namely T to C) is at cis-acting element transcription factor 1(SF-1) and affects the transcription and secretion of aldosterone (Li 2021).They were eventually involved in our meta-analysis in view of the out-of-balance of RAAS system caused by them and various studies focussed on them showing inconsistent results with the correlations of PE.
So far, although the correlations of preeclampsia with these gene polymorphisms were investigated in several previous in-comprehensive meta-analyses, but a portion of results run into opposite direction and with significant heterogeneity (Lin et al. 2012, Ni et al. 2012, Li et al. 2015, Zhang et al. 2016, Wang et al. 2020b).Moreover, besides environmental factors, it is well established that race and geographic regions play important roles in leveraging SNPs and diseases, presumably leading to the disparity simultaneously (Wang et al. 2020a).Therefore, in this article, we attempted to update them with more new literature and detailed hierarchy, especially stratified by race and regions, and comprehensive genetic models to draw more robust and convincing evidence for clinical screening.

Inclusion and exclusion criteria
The inclusion criteria were: 1. Case-control and cohort studies on the associations between PE risk and AGT M235T, AGT T174M, AT1R A1166C and CYP11B2 C344T in order; 2. Proper preeclampsia criteria: preeclampsia was defined as gestational hypertension (>140/90mmHg) and albuminuria (300mg/24h or �1þ in a random urine protein) after 20 weeks of pregnancy ( € Olmez et al. 2022); 3. The wild and mutant genotypes of cases and control groups were clearly given or could be calculated (the data needed to be calculated again if there were genotyping failures in the experiments); 4. Cases and control groups were from the same race and region.
The exclusion criteria were:

Data extraction and quality assessment
The following information was extracted from each study: the first author, year of publication, study population (race, country), number of genotypes available of cases and controls (genotypic analysis of a few specimens failed in some studies), study types.Then, quality assessment of eligible studies was executed by two independent researchers using the Newcastle Ottawa Quality Assessment Scale (NOS) (Stang 2010).Any incongruity was arbitrated by a senior investigator.An NOS score of �6 was considered as high quality.

Statistical analysis
All statistical analyses were performed with STATA 17.0 MP.Primarily, the Hardy-Weinberg equilibrium (HWE) was calculated to evaluate representativeness of each study and then the odds ratios (ORs) were generated to quantify the correlations between PE risk and SNPs, which were examined by five genetic models (allele, dominant, recessive, codominant and over-dominant).Heterogeneity across studies was determined by Q-statistic test as well as I2 value, I 2 < 50% and p > 0.01 for the Q test indicated no statistical significance, then the fixed model would be implemented, otherwise, random model would be carried out.Further on, subgroup analyses stratified by race and geography were conducted to find the source of heterogeneity and explore the relationships between SNPs and the aforementioned two factors.After subgroup analysis, if heterogeneity was still beyond 75%, Galbraith plot would be implemented to exert and remove the outlier studies to decrease the heterogeneity.Egger's tests were performed to quantify publication bias.p < 0.05 was considered to be statistically significant.

Literature selection
Eventually 51 studies on AGT M235T including 5925 cases and 10,418 controls were enrolled, which contained 47 case-control studies and 4 cohort studies.For AGT T174M, 11 studies involving 1533 cases and 2923 controls were finally eligible for the inclusion criteria, only one study was cohort study.23 studies on AT1R A1166C including 3348 cases and 6015 controls were enrolled, which contained 21 case-control studies and 2 cohort studies.At last, totally 7 case-control studies including 955 cases and 930 controls were involved for CYP11B2 C344T.The NOS scores of all literature were between 6 and 8.
Figure 1 shows the flow chart of selection of relevant literature, Tables 1-4 show the detailed characteristics of all eligible studies and Tables 5-8 show the results of overall and subgroups comparisons of PE risk and SNPs, respectively (Tables 1-4 were presented in supplementary materials).

The main findings from AGT M235T
Table 5 shows the main results of pooled ORs between PE risk and AGT M235T.Overall, significant associations were found in all except over-dominant models, and the correlations were stable even after removing studies violation of HWE.
In the subgroup analysis stratified by race, increased PE risk was observed in all except over-dominant genetic models (OR ¼ 0.85, 95%CI:0.76 to 0.96) for Caucasians while the significant associations were found under allele, dominant and codominant models (OR ranged from 1.33 to 1.75) in Mongoloids.For F, there were also positive correlations both in recessive model (OR ¼ 1.38, 95%CI:1.04 to 1.82) and overdominant model (OR ¼ 1.37, 95%CI:1.07 to 1.76).Finally, there was no evidence of association between PE risk and AGT M235T in all models among mixed race.
As stratified by geography, the T allele was associated with increased risk compared to the M allele in Africa, America, Europe, East Asia under allele model.Under dominant model, higher risk to develop PE was found for the MT and TT genotypes than MM genotype in Australia, Africa, Europe, East and Central Asia area, and unlike it, only Americans and Africans with TT genotype had increased risk of PE under recessive model.Compared to MM genotype, just East Asians with TT genotype had a higher risk of PE (OR ¼ 1.69, 95%CI:1.13 to 2.53) while MT genotype carriers had increased PE risk in five regions across Australia, Africa, Europe, East and Central Asia, which were consistent with the dominant model.At last, in an over-dominant model, negative association was observed invariably in all regions.

The main findings from AGT T174M
As summarised in Table 6, AGT T174M was justified to have no association with PE risk under any genetic model in the total population whenever stratified by race or not.However, significant heterogeneity existed among Caucasians in allele, dominant and codominant (MT vs TT) genetic models, Galbraith plot failed to reduce the heterogeneity to less than 75% since there were only 4 studies.Then in the stratified analysis by geography, there was only one study from Zitouni et al. (2018) based on North African showing disparity from others: it reduced the risk of developing preeclampsia under dominant model (OR ¼ 0.55, 95%CI: 0.32 to 0.96) and codominant model (TM vs TT, OR ¼ 0.51, 95%CI: 0.29 to 0.89), but what was paradoxical, over-dominant model indicated a higher risk of PE (OR ¼ 1.99, 95%CI:1.13-3.51).Lastly, in the rest regions, no correlation was detected between PE risk and AGT T174M.

The main findings from AT1R A1166C
As shown in

The main findings from CYP11B2 C344T
There were just 7 articles harvested for CYP11B2 C344T in

Heterogeneity analysis and publication bias
Varied degrees of heterogeneity among studies were present in the overall and subgroups comparisons, after removing the outliers in Galbraith plot, we found that pooled ORs resembled the former, indicating the stability and robustness of the results.Significant publication bias was only found in the allele, dominant, over-dominant and codominant (TT vs CC) models of susceptibility of PE with AT1R A1166C.Unfortunately, we failed to eliminate publication bias with usage of multiple imputation.a The results after removed outliers using Galbraith plot when heterogeneity(I 2 ) was >75%.

Discussion
To the best of our knowledge, this was the first meta-analysis with the largest sample size up to now based on correlations between susceptibility of PE and gene polymorphisms in RAAS system, hence we conduct more meticulous hierarchy analyses by means of comprehensive genetic models to get more prudent, reliable and stable results.
Our conclusion illustrated AGT 235 T allele, AT1R 1166 C allele increased and CYP11B2 344 T allele decrease PE risk in certain models while AGT T174M would not alter the risk of preeclampsia in any model.In total population, several studies with unbalanced HWE were removed and the OR values were calculated again whereas the results showed that pooled ORs decreased slightly along with a narrower CIs without changes in statistical significance, which meant that studies unqualified for HWE affected the significance of results inconspicuously, which was unprecedented in the previous sages' contributions.
Compared with our findings, two out of seven previous meta-analyses focussing on AGT M235T failed to discover the correlation with PE whereas the rest five meta-analyses obtained a positive conclusion in overall analyses but subgroup analyses were not detailed as ours due to the limited sample size (Medica et al. 2007, Zafarmand et al. 2008, Lin et al. 2012, Ni et al. 2012, Buurma et al. 2013, Zhang et al. 2016, Wang et al. 2020a).For AT1R A1166C, no correlation was detected by Li et al. (2015) and the rest two meta-analyses reported positive association with PE risk but lacked inclusion criteria (mingled patients with gestational hypertension) yielded the results controversial (Zhang et al. 2014, Gong et al. 2015).At last, a meta-analysis from Wang (Wang et al. 2020b) reported that CYP11B2 C344T was not associated with preeclampsia, however, a reduced risk was explored in our study with latest studies accession.Therefore, despite that recent metaanalyses had investigated similar topic, our work made amendment of these drawbacks in some extent with more detailed classification, stringent inclusion criteria and new literature.
For AGT M235T, as stratified by race, the 235 T allele increased a greater risk of PE in Caucasians than Mongoloids and Negroids under partial genetic models while no association was detected in mixed race under any model.The overdominant model was never performed in previous meta-analyses, what puzzled us most was that Caucasians with 235 T allele had 15% reduction (OR ¼ 0.85, 95%CI: 0.76-0.96)risk of PE under over-dominant genetic model, that was to say heterozygous advantage was validated in this model.In stratified analysis by geography, unlike the study from Wang (Wang et al. 2020a) showing no association in various regions, our results varied across different genetic models, however, the risk towards of PE was approximate in various regions under dominant and codominant models (MT vs MM), which demonstrated that the MT genotype was a risk factor for PE over MM genotype in these regions.Among the 7 regions, the three genotypes predisposition to PE in East, South and Southeast Asia was highly in accordance with local race (Mongoloids in East Asia and mixed race in South and Southeast Asia) under different genetic models while the outcomes were inconsistent in America and Africa perhaps due to the diversity of race.Hence, the evaluation of AGT M235T polymorphisms and PE risk in these two regions should take race and geography into the consideration collectively.
For AGT T174M, 6 original studies were enrolled at most in previous meta-analyses, then we updated it with 11 available studies, a unanimous conclusion was reached: PE risk was negative correlated with AGT T174M polymorphism either in overall or subgroup analysis by race.Moreover, heterogeneity in Caucasians was extreme and Galbraith plot failed to reduce the heterogeneity to less than 75% given the merely 4 studies.We did a comprehensive literature review of the four studies: two studies violated HWE balance, one study based on the population of Arab origin, and the subjects of the last study were French-Canadians, collectively, these factors might contribute to the heterogeneity which was difficult to investigate (L� evesque et al. 2004, Knyrim et al. 2008, Zitouni et al. 2018, Procopciuc et al. 2019).As stratified by geography, the study of Arabs from North Africa showed appreciably differences from others: it seemed that MT genotype could reduce the risk of developing PE in this area.Nevertheless, the result should be interpreted discreetly as well considering the barely African study (Zitouni et al. 2018).
Similarly, both HWE-violating studies were located in European regions, which might account for the high heterogeneity of the European subgroup, hence, more studies on this area are still needed to be incorporate to validate this conclusion.
For AT1R A1166C, forest plots indicated the 1166 C allele elevated the PE risk in total population under allele and dominant genetic models, especially in Mongoloids of race and Asia area.However, the same outcome performance was shown in Mongoloids and Asian subgroups as AGT M235T in Caucasians under the over-dominant model: heterozygous AC genotype was more of a risk factor than homozygous AA and CC genotypes.As aforementioned, negative partial dominant and heterozygous dominant effects were also speculated to be the decisive factor.Nevertheless, publication bias was detected under several models which suggested that more articles are still necessary to enrich and validate the findings.
Surprisingly, our meta-analysis demonstrated CYP11B2 344 T allele was a protective factor in certain genetic models, unfortunately, subgroup analyses were not performed owing to too few available studies.After reviewing previous metaanalysis from Wang (Wang et al. 2020b), we found that it included an article composed by Ramirez-Szlazar (Ram� ırez-Salazar et al. 2011) in which the subjects were gestational hypertension but not PE, then the insufficiently rigorous inclusion criteria might result in unconvincing conclusion.Hence higher credibility was displayed in our work since this study was dropped and a more recent literature was added.
Up to date, exact reasons for the dysfunction of RAAS in preeclampsia is still obscure, in light of the present findings, the RAAS system may be activated during incipient and early stages of PE and subsequently suppressed (Aung et al. 2017).The limited evidence from Morgan (Morgan et al. 1999) suggested that the angiotensinogen expression level of pregnant women with AGT 235 T allele was elevated in decidual spiral arteries, which led to abnormal spiral artery remodelling such as smaller calibre lumens, thicker walls and inflated wall/diameter ratio, simultaneously mediators that regulated placental vasculogenesis changed including the declining placental growth factor (PIGF) and up-regulated expression of soluble FMS-like tyrosine (sFlt1), eventually cascaded events of failed trophoblast invasion, shallow placenta implantation, uteroplacental ischaemia and hypoxia, and then PE occurred.
It was a pity that no study addressed the impact of AGT T174M genotypes either on plasma angiotensin levels or other biological markers, so although we found that this polymorphism didn't alter the risk of PE, but no evidence revealed the correlation between the biological products of transcription as well as expression of these two alleles and preeclampsia.Therefore, correlations of PE with this SNP still necessitate more in-depth molecular mechanism studies.
Despite that the function of AT1R A1166C appeared to be silent presumably owing to the transversion was located at 5 0 end of the 3 0 untranslated region (UTR), the study from Abelson (Abelson et al. 2005) demonstrated SNPs in the 3 0 -UTR could affect gene regulation by interfering with micro RNA (miRNA) binding.Subsequently, further laboratory experiments showed that miR-155 binding to the 3 0 -UTR of AT1R mRNA could repress the expression of AT1R (Zhu et al. 2011).Moreover, overlapping effects were explored between AT1R polymorphisms and miR-155 target site and the ability of miR-155 to interact with the cis-regulatory site was proved by Sethupathy et al. (2007) decreased in 1166 C allele carriers, leading to attenuation of miR-155 silencing function and enhancement of AT1R expression.However, these experiments were not conducted on pregnant women, hence whether the pathophysiology could be extended to them remained to be determined.
Our meta-analysis demonstrated CYP11B2 344 T allele might be a protective factor of PE, we speculated that the possible pathological process was that C allele showed about four-fold higher affinity with the steroidogenic factor 1 (SF1) protein than T allele, this loss-of-function mutation resulted in upregulation of the serum aldosterone level (Azimi-Nezhad et al. 2020), leading to elevation of blood pressure.Conversely, aldosterone was verified to compensatory increase the placental perfusion in normotensive pregnancy, lack of aldosterone gave rise to deficiency in fluid and subsequent placental ischaemia in the development of PE (Roberts andCooper 2001, Shojaati et al. 2004).Hence, the mutated T allele somehow benefitted pregnant women and the opposite conclusion from previous meta-analysis might just on grounds of one study involved subjects of gestational hypertension.
According to the results of our meta-analysis and above pathological mechanisms, special attention is necessary to pregnancies carrying high-risk genotypes to prevent and early identification of PE.For pregnant women in plan or first trimester who have economic capability, we recommend genetic tests based on three gene polymorphisms correlated with PE risk aforementioned to identify high-risk genotypes to materialise earlier prevention and diagnosis and intervention.Moreover, further research on the molecular mechanism of polymorphisms of RASS system is expected to evaluate our findings.

Limitation
Firstly, given that the genetic models of the aforementioned gene polymorphisms with PE were not clear completely, a novel genetic model-free approach proposed by Cosetta Minelli (Minelli et al. 2005) could be implemented to salvage this dilemma, for uncertain genetic patterns, this advanced method could simplified computation without trying the five genetic models and was proved not inferior to the traditional pairwise comparison model.However, it required specialised programming capability and we look forward to trying this approach in the future.
Secondly, most of studies on AGT M235T and T174M were examined respectively, if studies of haplotypes could be accumulated for meta-analysis, an interaction of gene polymorphisms and PE risk might be understood deeply (Dimou et al. 2018).We hope that more similar studies would emerge in this area in the near future for further meta-analysis and deepen research.
Thirdly, existing research demonstrated that early-onset and late-onset PE shared different aetiology, and there were merely three studies on AGT M235T polymorphism and PE risk segregated the two groups into discussion in our metaanalysis (Roberts et al. 2004, Procopciuc et al. 2019, Loskutova et al. 2020), we expect more similar studies would enrol in the future to enrich our results.
Lastly, considering the publication bias of AT1R A1166C and few articles of CYP11B2 C344T, further studies based on them are still needed to validate our findings.

Conclusion
AGT 235 T allele, AT1R 1166 C allele increased and CYP11B2 344 T allele decreased PE risk in certain models while AGT T174M polymorphism would not alter the risk of preeclampsia in any model.Hence, for pregnant women carrying the three high-risk genotypes (AGT 235 T allele, AT1R 1166 C allele, CYP11B2 C344 allele), special attention and intensive prenatal management should be given to them by race and geography to prevent and early identification of PE.

Figure 1 .
Figure 1.The flow chart of selection and exclusion of relevant literature.

Table 1 .
Characteristics of the included studies of AGT M235T.
Note: S and SE Asia: south and southeast Asia. a For the controls.95%CI:1.09to 1.62) and the results remained statistically significant after excluding three studies unconformity with HWE.For subgroup analysis stratified by race, we observed that correlations with susceptibility of PE merely existed among Mongoloids: the risk of developing PE increased under allele, dominant and codominant (AC vs AA) models whereas it decreased under over-dominant model.As stratified by geography, similar results in line with Mongoloids in Asia were detected and negative pertinence was found in other areas.

Table 2 .
Characteristics of the included studies of AGT T174M.

Table 3 .
Characteristics of the included studies of AT1R A1166C.

Table 4 .
Characteristics of the included studies of CYP11B2 C344T.

Table 5 .
Overall and subgroups analysis of AGT M235T polymorphism and preeclampsia risk.

Table 6 .
Overall and subgroups analysis of AGT T174M polymorphism and preeclampsia risk.
Note: Numbers in bold type are significant.‹Overall analysis.›Overall analysis without control groups which deviation from HWE balance.

Table 7 .
Overall and subgroups analysis of AT1R A1166C polymorphism and preeclampsia risk.

Table 8 .
Analysis of CYP11B2 C344T polymorphism and preeclampsia risk.