The Association Between the Risk of Hypertensive Disorders of Pregnancy and Folic Acid: A Systematic Review and Meta-Analysis.

PURPOSE
Although folic acid (FA) supplementation has been shown to reduce general cardiovascular risks, its impact on hypertensive disorders of pregnancy (HDP) is unclear. We performed a systematic review and meta-analysis to clarify the association between FA and the risk of HDP (pre-eclampsia (PE) and gestational hypertension (GH)).


METHODS
PubMed, EmBase, and Cochrane Library were searched up to June 18, 2020, stratified by type of disease, initiation time of FA, form of FA and pre-conception Body Mass Index (BMI). The quality assessment of included studies was evaluated using Newcastle-Ottawa Scale (NOS) for cohort studies and Cochrane Collaboration's Risk of Bias Assessment Tool for randomized controlled trials (RCTs). Between-study heterogeneity was quantified using Cochran's Q-statistic and I2 statistics. Sensitivity analysis was performed by excluding the studies one by one, and publication bias was analyzed using funnel plots.


RESULTS
Twenty studies with 359041 patients were identified for inclusion in the meta-analysis which included 3 RCTs and 17 cohort studies. Pooled estimates showed RR of 0.83 (95%CI 0.74-0.93, P=0.0008) for association between low dose FA (LD-FA) and the risk of PE, but LD-FA was not associated with GH (RR 1.05, 95% CI 0.97-1.13, P=0.20). In addition, the results of subgroup analysis showed that post-conception LD-FA had a 31% decreased risk of PE (RR 0.69, 95% CI 0.59-0.80, P<0.00001), and LD-FA in patients with pre-conception BMI<25 kg/m2 had a 32% decreased risk of PE (RR 0.68, 95% CI 0.56-0.81, P<0.0001) Conclusions: LD-FA significantly decreased the risk of PE but not GH, and post-conception LD-FA and pre-conception BMI<25 kg/m2 were considered as protective factors to reduce the risk of PE.


INTRODUCTION
Hypertensive disorders of pregnancy (HDP) accounts for 16% of maternal deaths and is also associated with severe morbidity, long term disability and death among both mothers and their babies (1). This group of diseases includes preeclampsia (PE) and eclampsia, gestational hypertension (GH) and chronic hypertension. In addition, some evidence showed that women with HDP were at increased risk of developing chronic kidney disease, cardiovascular disease (CVD) and even cardiovascular mortality (2)(3). Therefore, effective prevention and treatment of HDP is critical. At present, several clinical practice guidelines have clearly recommended that aspirin could be used for the prevention of HDP in high-risk pregnant women (1)(2)(3). A meta-analysis which with focus on the general cardiovascular health indicated a 10% lower risk of stroke and a 4% lower risk of overall CVD with folic acid (FA) supplementation (4). However, studies investigating the use of FA to prevent HDP have yielded inconsistent results. Currently, the relationship between FA and HDP has been widely studied , and five meta-analyses with inconsistent results have been published (28)(29)(30)(31)(32). Two meta-analyses demonstrated that FA was not associated with the risk of GH or PE (28)(29). Two demonstrated that multivitamin containing FA (Vit-FA) could significantly lower GH or PE risk (30)(31). And one showed no significant difference between FA alone and Vit-FA in reducing PE (32). The latest retrieval deadline for these meta-analyses was Aug. 2017. Since then, six other studies have been published. In addition, FA initiation time, and preconception Body Mass Index (BMI) were used as grouping conditions in some cohort studies, so these factors are an important opportunity for subgroup analysis of meta-analysis.
The purpose of this study was to perform a systematic review and meta-analysis to clarify the association between FA and the risk of HDP.

Search Strategy
We conducted a comprehensive search in PubMed, EmBase, and Cochrane Library from the date of their inception up to June 18, 2020. The search terms were composed of the following: "folic acid", "vitamin b9", "preeclampsia", "pregnancy induced hypertension", "gestational hypertension", "hypertensive disorders of pregnancy". To identify potential publications, we also examined the references of relevant reviews and meta-analysis. The details of the search strategy were summarized in the Appendix 1.

Study Selection
Inclusion criteria: (1) population-based studies, including cohort studies, case-control studies, and randomized controlled trials (RCTs). (2) the exposure factor is FA or Vit-FA. (3) the outcome is GH or PE. (4) the written language is English. Exclusion criteria: (1) not original studies (e.g., reviews, meta-analysis, commentaries, case reports, editorials and letters). (2) conducted in animals. (3) studies without control group (comparison between different dose or FA alone and Vit-FA) (4) studies lack the necessary data (e.g., non-full text articles, unpublished trials).

Data Extraction
Two investigators (YHY, XMS) assessed the eligibility of retrieved articles and extracted relevant data from each eligible study using a standardized form independently. From included studies, the following information were extracted: study characteristics (first author's name, year of publication, study design, range of years in the studies, country), participants' characteristics (population size, gestational age at recruitment, age, pre-conception BMI, FA dose, the initiation time of FA (pre-conception or post-conception), type of disease (GH or PE), the HRs, RRs, ORs with 95% CIs of GH/PE adjusted for confounders.

Outcomes
The primary outcome was the risk of PE. The secondary outcome was the risk of GH.

Quality Assessment
The methodological quality assessment of included studies was independently assessed by two reviewers (YHY, XMS) according to the Newcastle-Ottawa Scale (NOS) and the Cochrane Collaboration's Risk of Bias Assessment Tool. Two reviewers (XRW, XF) resolved the disagreements, and the final consensus was reached by all four reviewers. For cohort studies, the NOS includes three aspects with a nine-point scale: selection of study population, compatibility of the study groups, and ascertainment of outcomes, and the scores of 0-3, 4-6, and 7-9 were interpreted as low, moderate, or high quality. For RCTs, the Cochrane Collaboration's Tool includes six aspects with seven items: random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other bias.

Statistical analysis
This meta-analysis was conducted by Review Manager for Windows (version 5.3) using the generic inverse variance method. Between-study heterogeneity was quantified using Cochran's Qstatistic and I 2 statistics and a value of P＜0.10 or an I 2 ＞50% was considered statistically significant for heterogeneity among studies. The random-effect model was used if it was significant; otherwise, the fixed-effect model was acceptable. Subgroup analyses were performed according to the type of disease (GH or PE), initiation time of FA (preconception or post-conception), form of FA (Vit-FA or FA alone) and pre-conception BMI (BMI<25 kg/m 2 or BMI ≥ 25 kg/m 2 ), respectively. The sensitivity analysis was conducted by excluding the studies one by one to evaluate the stability of results without estimation bias from individual study. The funnel plots were used to evaluate publication bias qualitatively.

Study Characteristics
This meta-analysis included 3 RCTs and 17 cohort studies. Characteristics of the 20 included studies in the meta-analysis were presented in Table 1 and   Table 2. These studies were published between 2002 and 2020. The study populations ranged from 215 to 193554 participants, with a total number of 359041. There were 12 studies that didn't clearly describe the initiation time of FA, 4 studies that began taking FA in post-conception period, and 1 study in preconception period. FA alone or Vit-FA was 0.2-1mg in most studies, except for 3 RCTs which used high dose (4-5mg) FA. The ascertainment of GH/PE were from medical records and self-report, and there are 14 studies evaluating the risk of PE alone and 6 studies evaluating the risk of both GH and PE. In addition, 5 articles performed a subgroup analysis based on the pre-conception BMI. The ORs, RRs and HRs in 18 studies were adjusted based on the potential confounders. The 20 studies were undertaken in different countries, including the USA (n=6), Canada (n=4), China (n=4), Denmark (n=2), Netherlands (n=1), the UK (n=1), South Korea (n=1) and Australia (n=1).

Quality Assessment
NOS was used to assess the risk of bias for 17 cohort studies. All cohort studies were given a score of 7-9 stars, representing the high quality of studies. The six cohort studies had scores of 7-8 for the reason that FA and HDP were identified only based on selfreports. The results of this quality assessment were summarized in Appendix 2- Table A. Cochrane Collaboration's Tool was used to assess the risk of bias for 3 RCTs, and only random sequence generation of Charles 2005 was unclear. 6 Appendix 3- Figure 1 and Figure 2 showed the results of the risk of bias assessment, representing that all RCTs were of high quality.

High Dose (HD)-FA and the Risk of PE
Three studies compared risk of PE with HD-FA. The results showed that HD-FA was not associated with PE (3 studies, RR 1.11, 95% CI 0.75-1.62, P=0.61; Figure 2). No obvious asymmetry was found in the funnel plot (Appendix 4- Figure 1), which indicated that publication bias was unlikely in the analysis. The results of sensitivity analysis showed no substantial modification of the estimates after exclusion of individual study one by one.

Low Dose (LD)-FA and the Risk of HDP(GH/PE)
Eighteen studies compared risk of HDP with LD-FA. Stratification by GH and PE showed that LD-FA significantly decreased the risk of PE (18 studies, RR 0.83, 95%CI 0.74-0.93, P=0.0008, Figure 3), but was not associated with GH (6 studies, RR 1.05, 95% CI 0.97-1.13, P=0.20; Figure 3). No obvious asymmetry was found in the funnel plot (Appendix 4- Figure 2), which indicated that publication bias was unlikely in the analysis. The results of sensitivity analysis showed that the association between LD-FA and decreased risk of PE in PE subgroup was reliable.

Pre-conception or Post-conception LD-FA and the Risk of PE
11 studies with 23 results compared the association between peri-conception LD-FA and the risk of PE. According to the initiation time (Pre-conception or Post-conception), a subgroup analysis was  performed. The results showed that pre-conception LD-FA was not associated with the PE (9 studies,RR 0.92, 95%CI 0.84-1.02, P=0.12, Figure 4), but postconception LD-FA had a 31% decreased risk of PE (9 studies, RR 0.69, 95%CI 0.59-0.80, P<0.00001; Figure 4). No obvious asymmetry was found in the funnel plot (Appendix 4- Figure 3), which indicated that publication bias was unlikely in the analysis. The results of sensitivity analysis showed that the association between post-conception LD-FA and the decreased risk of PE was reliable.

LD-FA in patients with pre-conception BMI<25 kg/m 2 or BMI≥25 kg/m 2 and the Risk of PE
According to the pre-conception BMI (BMI<25 kg/m 2 or BMI≥25 kg/m 2 ), a subgroup analysis was performed to compare the association between LD- The results showed that LD-FA in patients with preconception BMI<25 kg/m 2 had a 32% decreased risk of PE (4 studies, RR 0.68, 95%CI 0.56-0.81, P<0.0001, Figure 5), but no difference was found in patients with pre-conception BMI ≥ 25 kg/m 2 (4 studies, RR 0.87, 95%CI 0.71-1.05, P=0.14; Figure  5). No obvious asymmetry was found in the funnel plot (Appendix 4- Figure 4), which indicated that publication bias was unlikely in the analysis. The results of sensitivity analysis showed that the association between LD-FA and the decreased risk of PE in pre-conception BMI<25 kg/m 2 subgroup was reliable.

DISCUSSION
Our meta-analysis of 20 studies involving 359041 patients found that LD-FA was associated with the decreased risk of PE but not GH. In addition, postconception LD-FA and pre-conception BMI<25 kg/m 2 were considered as protective factors to reduce the risk of PE. Some studies of the association between FA and the risk of GH or PE have shown a potential protective effect (5,7,9,11,(15)(16)(17)19), and Wen   (22) has shown that maternal exposure to FA antagonists appears to increase the risk of placentamediated adverse outcomes of pregnancy (33). It has been proved that elevated plasma homocysteine (Hcy) level was associated with increased risk for various forms of CVD, and it was also significantly elevated in women with PE (34). The possible pathogenic mechanisms involve damage of endothelial cell, reduction of vascular flexibility, changes in the hemostatic process and promotion of the development of inflammation. The elevated Hcy may arise from genetic defect of methylenetetrahydrofolate reductase (MTHFR) or insufficient intake of FA (35)(36). In the prevention and treatment of CVD, in addition to reducing Hcy levels, FA could also increase nitric oxide synthase (NOS) coupling and nitric oxide (NO) bioavailability to improve vascular health and function which are independent of the Hcy-lowering effect (37). However, two studies have shown that FA supplementation might be associated with the increased risk of GH or PE (14,24). There are some possible explanations, including the increased red cell blood mass which is correlated with elevated blood pressure, the altered process of DNA and protein synthesis of placental growth and development, and the imbalance between FA and vitamin B 12 (24).
In the late first trimester or early second trimester, placental growth and development reach the peak. Thus, FA supplementation in this period     FA, folic acid; Vit-FA, multivitamin containing folic acid; HD-FA, high dose folic acid; LD-FA, low dose folic acid; GH, gestational hypertension; PE, pre-eclampsia; BMI, Body Mass Index; HR, hazard ratio; OR, odds ratio; RR, relative risk; CI, confidence interval; NR, not reported may be most effective to prevent PE which was consistent with our finding that post-conception FA could reduce the risk of PE rather than preconception FA (22). It should be noted that preconception FA is a necessary measure to prevent neural tube defects (NTDs). Therefore, it is a more clinically appropriate choice to extend the use of FA into the later trimesters rather than starting it after pregnancy.
Another finding from our analysis was that FA supplementation could have a significant protective effect among women with pre-conception BMI<25 kg/m 2 but not statistically significant among overweight women (pre-conception BMI≥25 kg/m 2 ). Firstly, the pathogenic mechanisms of PE may be different among women with BMI<25 kg/m 2 and BMI≥25 kg/m 2 . The excess risk of PE might be associated with overweight or even obese status. Additionally, most pregnant women take low dose of FA, but both FA alone and multivitamin may not be adequate to overcome the metabolic disturbances of overweight or obese status (7,11).
Our meta-analysis had several advantages. First, RCTs of HD-FA supplementation were analyzed separately from LD-FA to avoid bias, as the relationship between HD-FA and PE is still controversial. In addition, we did the subgroup analysis and the results demonstrated that postconception LD-FA and pre-conception BMI<25 kg/m 2 were protective factors for pregnant women. We also acknowledged some limitations in this study. First, it has been recommended for women planning pregnancy to use FA until the third month of pregnancy to prevent NTDs. However, the discontinued time of FA is unclear for the pregnant women included in this study, particularly preconception FA. Therefore, the association of postconception FA and the decreased risk of PE need to be further confirmed. Secondly, most studies included in this study did not consider the effects of serum FA and serum Hcy. Meanwhile, the genotype of MTHFR was not considered in most studies but it was also an important factor because the studies included came from different countries. And it was difficult to accurately evaluate the dose of dietary FA, which could cause the bias of final results. In addition, although the result indicated that Vit-FA might be more meaningful than FA alone to reduce the risk of PE, it was hard to further compare the effects of FA with Vit-FA due to the few clinical studies. Lastly, only few studies have been conducted in pregnant women with high-risk (20,(22)(23). Thus, we need more high-quality clinical studies to confirm the findings of this study, such as detailed documentation for FA dose, start and end time, form of FA, high-risk factors, and detection of genotype of MTHFR and serum FA.

CONCLUSIONS
Our meta-analysis found that LD-FA tended to decrease the risk of PE. In addition, post-conception LD-FA and pre-conception BMI<25 kg/m 2 were considered as protective factors to reduce the risk of PE. Therefore, we recommend that pregnant women use LD-FA throughout the pregnancy period to avoid the occurrence of PE. More importantly, a more convincing result require the inclusion of more rigorous RCTs.

FUNDING
This study was not funded by any organizations.