The Association Between ABO Blood Group and Preeclampsia: A Systematic Review and Meta-Analysis

Objective: This meta-analysis comprehensively evaluated the association between ABO blood group and the risk of preeclampsia (PE). Design: Systematic review and meta-analysis. Data sources: PubMed, Web of Science, and ScienceDirect databases from their inception to September 23, 2020. Methods: Pooled odds ratios (ORs) with 95% confidence intervals (CIs) were obtained through random-effects and fixed-effects models according to heterogeneity. Meta-regression analysis was applied to explore the source of heterogeneity. We conducted a subgroup analysis by the publication year, study design, state, and Newcastle-Ottawa Scale (NOS) score. In addition, we calculated the rate of each ABO blood group in PE by total pooled effects. Results: A total of 12 articles with 714,153 patients were included in our analysis. Compared with people without PE (control group), the O blood group presented a lower risk of PE (OR 0.95, 95% CI 0.93–0.97). The AB (OR 1.46, 95% CI 1.12–1.91) blood group presented a higher risk. However, the total pooled OR and 95% CI for the A (OR 1.02, 95% CI 0.90–1.16) and B (OR 1.02, 95% CI 0.98–1.05) blood groups were not significant. The funnel plot and linear regression equation showed that there was no publication bias for the O, A, or B blood groups (all P > 0.05). However, the funnel plot and linear regression equation for the AB blood group were obviously asymmetric (P < 0.05), and the publication bias persisted even after the trim-and-fill method was applied (P < 0.05). Multivariable meta-regression analysis did not find a specific source of heterogeneity. The A blood group showed an association with early-onset PE (OR 0.53, 95% CI 0.33–0.83), and the other blood groups showed no significant differences. In PE, the rates of the O, A, B, and AB blood groups decreased gradually (0.39, 0.33, 0.19, 0.07). Conclusion: These findings suggest that pregnant women with AB blood group are more likely to develop PE, and more attention should be paid to AB blood group whose blood pressure is high but not sufficient to diagnose PE. Systematic Review Registration: Prospero CRD42021227930.


INTRODUCTION
Preeclampsia (PE) is a common complication during pregnancy that affects 5-8% of pregnant women (1,2). PE is associated with a variety of short-term and long-term complications in mothers and infants, such as placental abruption, cardiovascular disease, renal disease, metabolic disorders, fetal growth restriction (FGR), and preterm birth (3)(4)(5). After 20 weeks of gestation, women with systolic BP ≥140 mmHg or diastolic BP ≥90 mmHg on two occasions with or without proteinuria and renal, liver, lung, or neurological organ dysfunction were considered to have PE (5,6). PE can be categorized as early onset (<34 weeks of gestation) or late onset (≥34 weeks of gestation) (7). In addition, PE can be classified as mild or severe PE depending on the severity of the condition (diagnostic criteria for severe PE are shown in Appendix S1 in Supplementary Material) (8,9). Risk factors associated with the development of PE have been reported, including previous history of PE, history of abnormal blood pressure, history of gestational diabetes mellitus, multiple pregnancy, and nulliparity (10,11). Additionally compared with pregnant women without PE, mounting evidence suggests that lower placental growth factor (PlGF) and higher soluble fmslike tyrosine kinase 1 (sFlt-1) levels of maternal blood during pregnancy are linked to PE. Thus, these factors and blood biomarkers may be used for risk prediction of PE before the appearance of the clinical syndrome (6,12,13).
In 1901, the ABO blood group system was first discovered and defined by Karl Landsteiner in Austria. It includes types A, B, AB, and O, which are defined according to the expression of agglutinins A and B (14). In recent years, an increasing number of studies on ABO blood groups have been conducted, and blood group has been reported to be associated with the development of many human diseases, such as thrombotic vascular diseases (15), gestational diabetes mellitus (GDM) (16), acute respiratory distress syndrome (ARDS) (17), cardiovascular disease (CVD) (18), gastric cancer (19), infectious diseases (20), and PE (21). Although, the ABO system has been studied for more than a century, its clinical biological significance remains ambiguous.
In 2008, one study evaluated the association between ABO blood group and vascular disease and indicated that non-O blood group was at higher risk for some vascular diseases compared with O blood group (21). In 2013, a systematic review and meta-analysis reported that the AB blood group was associated with the occurrence of PE (22). A systematic review from 2016 aiming to elucidate the association of ABO blood groups with pregnancy-related complications indicated that women with a non-O blood group have an increased risk of PE (23). However, the results of subsequent studies have been inconsistent. Two studies found that patients with blood group AB have a higher risk of PE (24,25), but another three studies considered that there was no distinct association between ABO blood group and PE (26)(27)(28). These five studies included four case-control studies and one cross-sectional study. The largest study population was 17,564 individuals, and the smallest study population was 147 individuals. Hence, we conducted this meta-analysis to comprehensively evaluate the association between ABO blood group and the risk of PE. In addition, we calculated the specific rate of each ABO blood group in PE.

METHODS
This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (29).

Literature Search
We searched for the relevant literature through the PubMed, Web of Science, and ScienceDirect databases from inception to September 23, 2020. The Population, Intervention, Comparator, Outcomes and Study designs (PICOS) principle was used to identify articles in the various databases (Appendix S2 in Supplementary Material). We restricted the language to English. We also tracked references to relevant articles. The details of the search process are shown in Appendix S3 in Supplementary Material. Two authors independently collected and integrated the data.

Eligibility Criteria
We selected articles on the basis of the database searches and applied EndNote X9 to remove duplicate articles. Then, we browsed the titles and abstracts to exclude unrelated articles. This meta-analysis followed the following inclusion criteria: (1) included data on ABO blood group for pregnant women; (2) included pregnant women with and without preeclampsia; (3) prospective and retrospective studies. Reviews, meta-analysis, articles lacking relevant data, letters, and abstracts were excluded.

Data Extraction and Study Quality Assessment
Two authors independently reviewed each study and decided whether it was eligible for inclusion in our meta-analysis, and if there was any disagreement, the corresponding author joined the discussion. We extracted the following data from the articles: first author name, year of publication, study design, state, and conclusions. The extracted data provided sufficient information for the construction of 2 × 2 tables. The study quality assessment was based on the Newcastle-Ottawa Scale (NOS, Australia and Ottawa, Canada) (30). Using this protocol, the maximum score for each study was nine. Studies with a score ≥7 were regarded as high-quality articles (31). Subgroup analysis was based on the publication year (<2010, ≥2010), study design, state, and NOS score (<7, ≥7) to further evaluate the association between ABO blood group and the risk of PE. Meanwhile, we assessed the association of ABO blood group with mild or severe PE and early-onset or late-onset PE. In addition, we calculated the rate of ABO blood group in PE based on the included studies.

Statistical Analysis
All the data were analyzed via R version 3.6. Forest plots were constructed to obtain Pooled Odds Ratios (ORs) and 95% Confidence Intervals (95% CIs). If I 2 < 50% and P heterogeneity > 0.05, the fixed-effects model was applied to calculate pooled effect estimates. If I 2 ≥ 50% or P heterogeneity ≤ 0.05, the random-effects model was applied. We conducted leave-one-out sensitivity analysis by removing each study to explore the robustness of the included literature. Publication bias was evaluated by funnel plots and linear regression equations. If the funnel plots were obviously asymmetric, we further adjusted the data by the trim-and-fill method. In addition, the multivariable meta-regression analysis was conducted to explore the source of heterogeneity on the basis of publication year, NOS score, state, and study design. We pooled the rates of O, A, B, and AB blood group in PE as proportions with 95% CIs after log transformation. A cut-off value of P < 0.05 was defined as statistically significant.

Study Characteristics
The main characteristics of the included studies are shown in Table 1. This meta-analysis included 12 articles with 714,153 patients: nine case-control studies, one cross-sectional study, one retrospective cohort study, and one prospective cohort study. The publication dates of these articles ranged from 1976 to 2020. Among these articles, the study areas included Europe for four studies; Asia, four; North America, two; South America, one; and Africa, one. The smallest sample size was 90, and the largest sample size was 679,740. The conclusions of seven articles indicated that there was no effect of ABO blood group on the risk of PE. Four studies showed that AB blood group increased the risk of PE. One study indicated that non-O blood groups had significantly higher odds of PE. One study showed that the A blood group increased the risk of PE.

Total Pooled Effect
As shown in Figure 2A, the heterogeneity among the eligible articles was I 2 = 18% (P = 0.26), so we chose a fixed-effects model. The total pooled effect showed that the O blood group presented as a protective factor against PE (OR 0.95, 95% CI 0.93-0.97). At the same time, we calculated the outcomes for A, B, and AB blood groups. The AB blood group presented a high risk of PE in the random-effects model, respectively (OR 1.46, 95% CI 1.12-1.91, I 2 = 62%, P heterogeneity = 0.01, Figure 3A). However, the total pooled OR and 95% CI showed no significance of the A blood group in the random-effects model (OR A 1.02, 95% CI A 0.90-1.16, I 2 A = 49%, P heterogeneity = 0.05) and B blood group in the fixed-effect model (OR 1.02, 95% CI 0.98-1.05, I 2 = 0%, P heterogeneity = 0.81) (Figures S1A, S2A). Although, only some articles studied mild or severe PE and early-onset or late-onset PE, we analyzed them further. As Figure S3 shows, regardless of mild or severe PE, there was no association between the ABO blood group and PE. However, the A blood group showed an association with early-onset PE, and the other blood groups showed no significance (OR 0.53, 95% CI 0.33-0.83, I 2 = 0%, Figure 4).

Publication Bias and Sensitivity Analysis
The funnel plot and linear regression equation showed that there was no publication bias with respect to the effects of the O, A, and B blood group (P O = 0.59, P A = 0.67, P B = 0.90) (Figure 2B,  Figures S1B, S2B). The funnel plot of AB blood group was clearly asymmetrical (Figure 3B). We further conducted the trim-andfill method; the funnel plot was symmetric, but publication bias still existed (P < 0.05, Figure 3C). As Figure 2C shows, when omitting one of these studies (28), the sensitivity analysis of the O blood group showed an OR of 0.95 (95% CI 0.93-0.97), nearly the same outcome as the total pooled effect (OR 0.95, 95% CI 0.93-0.97). Similarly, when omitting any one of the other studies, the outcomes showed that O blood group was a protective factor against PE. The sensitivity analysis of the AB blood group showed similar outcomes after omitting any one study (Figure 3D). The sensitivity analysis of the A and B blood groups showed that after omitting any one study, the effect of the A and B blood groups was not significant (Figures S1C, S2C).

Multivariable Meta-Regression Analysis
In the total pooled effect, heterogeneity of the AB blood group was I 2 = 62.0% (P < 0.05, Figure 3A). Thus, we conducted multivariable meta-regression analysis on the basis of publication year, NOS score, state, and study design. The results confirmed that these factors showed no significant effect on the heterogeneity (all P > 0.05, Table 2).

Subgroup Analysis
Subgroup analysis was based on the study design, state, NOS score (<7, ≥7) and publication year (<2010, ≥2010) to further evaluate the association between ABO blood group and the risk of PE. As shown in Table 3, the outcomes in the O blood group were almost the same. When publication year <2010 and studies performed in America, the outcomes showed no significance (all P > 0.05). Subgroup analysis of cross-sectional study design and studies performed in Africa, which included only one study each, showed no significance. The outcome for European region of AB blood group was not significant for PE (N Europe = 2, OR Europe 1.45, 95% CI Europe 0.74-2.84) ( Table 4). The subgroup analysis of the A blood group was somewhat inconsistent and showed no significance (

Rates of ABO Blood Group in PE
The rates of the O, A, B, and AB blood groups were further analyzed by forest plots. As shown in Figure 5, we can see that the rate of the O blood group was 0.39 (95% CI 0.33-0.44, I 2 = 90%). In PE, the rates of the A, B, and AB blood group decreased gradually (0.33, 0.19, and 0.07).

DISCUSSION
Our systematic review and meta-analysis comprehensively explored the association between ABO blood group and PE. Twelve articles comprising 714,153 patients were included. On the basis of previous studies and the outcomes we obtained, the present study demonstrated that compared with the control group, the O blood group presented as a protective factor for PE. Conversely, the AB blood group aggravated the risk of PE, and the A and B blood groups showed no significant effect on the risk of PE. Notably, we found that the A blood group showed an association with early-onset PE. In addition, we further calculated the specific incidences of the ABO blood groups in PE. The rate of the O blood group in PE was 0.39 (95% CI 0.33-0.44, I 2 = 90%), and the rates of the A, B, and AB blood groups were 0.33, 0.19, and 0.07, respectively. ABO blood group antigens exist on many kinds of cells of the human body; in addition to common red blood cells, these antigens are also expressed on vascular endothelial cells and neuronal cells (39). Existing studies found that ABO blood group status is correlated with many diseases, such as CVD, ARDS, GDM, and PE (16-18, 27, 28). Nevertheless, the association between PE and ABO blood group has been controversial. In 1976, an article published in JAMA studied 23 patients with PE, 23 patients with eclampsia, and 4,494 controls and suggested that there was no association between ABO blood group and PE (controls vs. PE: O blood group, 47.6 vs. 47.8%; A blood group, 37.9 vs. 32.6%; B blood group, 11.4 vs. 15.2%; AB blood group, 3.1 vs. 4.3%; all P > 0.05) (32). Studies performed in 2005 and 2008 arrived at the same conclusion. A meta-analysis also concluded that non-O blood groups are more susceptible to certain vascular diseases than O blood group (21). In 2009, for the first time, a population-based nested case-control study indicated that AB   (37). Furthermore, in 2013, a systematic review and meta-analysis that included only two eligible articles reported that the AB blood group was associated with the occurrence of PE (OR 2.42, 95% CI 1.63-3.58, P < 0.0001) (22). However, the results of subsequent studies are also inconsistent. A systematic review from 2016 aimed to elucidate the association of ABO blood groups with   (24,25), but another three studies considered that there was no distinct association between ABO blood group and PE (26)(27)(28). The conclusion of our analysis indicated that the O blood group is a protective factor against PE. Conversely, the blood group AB aggravated the risk of PE, while the A and B blood groups showed no significant effect on the risk of PE. Notably, we found that the A blood group showed an association with early-onset PE. In addition, we further calculated the specific rate of each ABO blood group in PE, and the rates of the O, A, B, and AB blood groups decreased gradually (0.39, 0.33, 0.19, 0.07). The strengths of our study include that different blood groups were analyzed and that subgroup analysis was carried out in detail. Furthermore, we evaluated both mild and severe PE and both early-onset and late-onset PE. Both funnel plots and linear regression equations were used to calculate publication bias. Multivariable meta-regression analysis on the basis of subgroups was also conducted to explore the source of heterogeneity. In addition, the rates of O, non-O, A, B, and AB blood groups were further specifically analyzed though total pooled effects. Undoubtedly, there are also some limitations in this study. First, only 12 articles were included, and the limited number of studies may influence the outcomes. We restricted the language of studies to English. In addition, in the meta-analysis of 2016, we were unable to find the full text of all included studies. PE is a complex physiological and pathological process, and many factors will affect the occurrence and development of PE (e.g., genetic factors, diet, and environment). It is obviously insufficient to use only the ABO blood group as a factor to predict PE, and all the potential risk factors may act as confounding factors in research outcomes (40)(41)(42)(43)(44). Although, the relationship between the ABO blood group system and disease has been studied for a long time, the mechanism of how the ABO blood group system causes and affects disease is not clear. Studies have found that placental protein 13 produced by pregnant women may be associated with the onset of PE by binding to β-galactosides (such as N-acetylgalactosamine, galactose, and fucose) at the end of ABO blood group antigens. However, the study on placental protein 13 is not conclusive, but a potential possibility. ABO blood group system has a high degree of polymorphism and it is difficult to simulate ABO blood group antigen in animal models, making it difficult to explore the relationship between ABO blood group and the pathogenesis of PE (45,46). With the development of molecular biology techniques, transgenic techniques, and bioanalytical tools, we expect to discover how the ABO blood group system causes and affects disease.

CONCLUSION
In conclusion, the O blood group is a protective factor against PE. Conversely, the AB blood group aggravates the risk of PE, and the A and B blood groups have no significant effect on the risk of PE. In addition, the A blood group showed an association with early-onset PE. These findings suggested that clinicians should pay more attention to pregnant women with blood group AB whose blood pressure is high but not sufficient to diagnose PE.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
TL and YW: study design, data extration, statistical analysis, and manuscript writing. LW, ZL, CL, and WL: study design, data extraction, and verification. KX and HD: study design, statistical analysis, manuscript editing and reviewing, and funding. All authors contributed to the article and approved the submitted version.