Efficacy and safety of folic acid on homocysteine and cardiovascular surrogate biomarkers in hyperhomocysteinemia: a systematic review and meta-analysis of RCTs Authors: Zhi-ping Long, Chang-wen Wang, Fan-fan Sun, Shi-bo Xu, Jun-zhen Yan, Cheng Zeng Published: 2026 ABSTRACT ObjectiveTo quantify the effect of folic acid supplementation on homocysteine (Hcy) levels and associated cardiovascular risk-related surrogate biomarkers in adults with hyperhomocysteinemia (Hhcy), and to summarise the available safety data.MethodsWe conducted a systematic review and meta-analysis (PROSPERO CRD42024617901) adhering to PRISMA 2020. randomized controlled trials (RCTs) that enrolled adults with Hhcy and compared folic acid monotherapy with placebo or usual care were identified in Web of Science, PubMed, Embase and Cochrane Library (inception–October 2024). Primary outcomes were change in plasma total Hcy and folate concentrations. Secondary outcomes included cardiovascular risk-related surrogate biomarkers: systolic/diastolic blood pressure (SBP/DBP) and lipid profiles.ResultsThirty-six RCTs were included. Folic acid supplementation was associated with dose-dependent reductions in tHcy levels (standardised mean difference [SMD] = − 0.99; 95% CI − 1.15 to − 0.83; I²= 90%). However, these results are statistically significant but highly heterogeneous and at substantial risk of bias, with likely overestimation of effect sizes due to small-study effects and publication bias. The largest SMDs were observed in the higher dose categories (5–15 mg/day) and after ≥ 6 months of treatment. Folate concentrations increased (SMD = 1.61; 95% CI 1.30–1.93). Statistically significant changes were observed in cardiovascular surrogate biomarkers: SBP (SMD = − 0.54), DBP (SMD = − 0.32), high-density lipoprotein cholesterol (HDL-C) (weighted mean difference [WMD] = + 1.37) and low-density lipoprotein cholesterol (LDL-C) (WMD = − 3.64). No cardiovascular adverse events were observed at doses ≤ 10 mg/day, whereas increased cardiovascular events occurred in trials using 15 mg/day (three studies; 35 events). Notably, no study provided adjudicated cardiovascular events (e.g., myocardial infarction, stroke, cardiovascular mortality), and all observed changes are limited to surrogate biomarkers rather than hard clinical endpoints. Therefore, no conclusions can be drawn regarding the reduction of clinical cardiovascular events based on the current evidence.ConclusionsThis meta-analysis indicates that folic acid is associated with low-certainty evidence of reduced total homocysteine (tHcy) and elevated folate levels—the efficacy of folic acid in this analysis is limited to these biomarker changes—along with modest changes in blood pressure and lipid cardiovascular surrogate biomarkers in adults with hyperhomocysteinemia (Hhcy), with a dose-dependent safety profile: no cardiovascular adverse events were observed at doses ≤ 10 mg/day, while increased events occurred exclusively at the 15 mg/day dose. Critically, no included studies reported adjudicated cardiovascular hard endpoints (myocardial infarction, stroke, cardiovascular mortality); target tHcy reduction remains an unvalidated surrogate for clinical cardiovascular disease (CVD) risk reduction, no optimal folic acid dose can be definitively defined from this analysis, and all dose-related considerations are strictly hypothesis-generating. No direct evidence supports that folic acid supplementation reduces clinical cardiovascular event risk. Interpretations of the clinical relevance of these findings must be extremely cautious, and large-scale, high-quality randomized controlled trials with adjudicated cardiovascular hard endpoints are urgently needed to validate the associations of folic acid supplementation with changes in clinical outcomes.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40795-026-01343-y. Introduction CVD has demonstrated a concerning epidemiological trajectory over the past three decades. Between 1990 and 2019, global CVD-related deaths surged from 5.6 million to 10.8 million, with its proportional mortality escalating from 23% to 35% of total deaths [1]. While ischemic heart disease and stroke persist as primary contributors, recent pathophysiological insights have identified hyperhomocysteinemia (Hhcy, defined as serum homocysteine > 15 µmol/L) as a significant modifiable risk factor [2, 3]. Epidemiological evidence reveals a dose-dependent relationship between homocysteine levels and cardiovascular risk, with each 5 µmol/L increment elevating CVD risk by 70% and coronary artery disease odds ratios by 1.6 (men) to 1.8 (women) [4, 5]. This association persists independent of methionine metabolism pathways [6], suggesting direct vascular toxicity through mechanisms including endothelial dysfunction, oxidative stress induction, and thrombotic predisposition [7–9]. Circulating homocysteine, blood pressure, and lipid profiles are widely recognized as surrogate biomarkers for cardiovascular risk, and modifications to these biomarkers are frequently used as intermediate endpoints in early-phase intervention studies. However, improvements in these surrogate markers do not reliably translate to reductions in clinical cardiovascular events. These observations underscore the imperative for systematically evaluating the impact of targeted Hhcy management on these CVD-related surrogate biomarkers, while clearly distinguishing between biomarker changes and actual clinical event reduction. The impact of folic acid supplementation on CVD-related surrogate biomarkers remains a subject of intense investigation. Mechanistically, folate’s capacity to enhance homocysteine remethylation and endothelial protection provides a biochemical rationale for modification of these surrogate biomarkers [10, 11]. Recent meta-analyses suggest particular benefit in renal-impaired populations without prior folate fortification exposure and with low baseline diabetes prevalence [12, 13]. However, this contrasts with data from 16 randomized controlled trials demonstrating no significant impact on cardiovascular events, cerebrovascular outcomes, or mortality [14]. These discrepancies may originate from critical methodological variations including: (1) differential dosing regimens (2) baseline folate status heterogeneity (3) population-level fortification practices [15, 16]. Current evidence thus presents a paradox: while biochemical models support folate’s therapeutic potential [17], clinical translation remains inconsistent [18]. To reconcile these contradictions, our investigation proposes a systematic evaluation of folate supplementation’s efficacy and safety profile in Hhcy management. Through rigorous meta-analytic methodology, we aim to: (1) quantify the magnitude of change in CVD-related surrogate biomarkers (2) explore the association between folic acid dose and Hcy-lowering efficacy (3) delineate population-specific response patterns in biomarker changes. This evidence synthesis seeks to inform clinical decision-making for Hhcy management and guide precision nutrition strategies for surrogate biomarker modification, with explicit caution against extrapolation to clinical cardiovascular event reduction without dedicated trial evidence [19, 20]. Methods The PROSPERO registry has been updated with the full details of the agreement. Its registration identification number is CRD42024617901. More information can be retrieved by clicking on the link (https://www.crd.york.ac.uk/PROSPERO/). This meta-analysis was executed in accordance with the PRISMA guidelines. For more comprehensive information, refer to the Additional file 2. Search strategy This meta-analysis was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for RCTs [21, 22]. A systematic literature search was conducted across four electronic databases: PubMed, Embase, Web of Science, and Cochrane Library, from database inception to October 2024. The search was structured around three core dimensions:Intervention: Folic acid (including all its pharmaceutical forms and synonyms, such as folate, vitamin B9);Population: Adults with hyperhomocysteinemia (Hhcy);Study design: Randomized controlled trials (RCTs) assessing the efficacy and safety of folic acid therapy. Intervention: Folic acid (including all its pharmaceutical forms and synonyms, such as folate, vitamin B9); Intervention: Folic acid (including all its pharmaceutical forms and synonyms, such as folate, vitamin B9); Population: Adults with hyperhomocysteinemia (Hhcy); Population: Adults with hyperhomocysteinemia (Hhcy); Study design: Randomized controlled trials (RCTs) assessing the efficacy and safety of folic acid therapy. Study design: Randomized controlled trials (RCTs) assessing the efficacy and safety of folic acid therapy. The full, standardized search string with complete MeSH terms, corrected Boolean operators, and detailed search syntax for each database is provided in Additional file 1. The reference lists of all included studies and relevant systematic reviews were hand-searched to identify additional eligible studies. Study selection Two investigators, namely Zhi-ping Long and Chang-wen Wang, were responsible for the literature selection. They began by scanning the titles, abstracts, and keywords to tentatively pick out the ones that seemed to meet the inclusion prerequisites. After this initial round of selection, they obtained the complete texts. Subsequently, a thorough assessment of these full texts was performed to determine if they qualified for inclusion. If any disagreements emerged during this evaluation, a meeting was held. At the meeting, the two would deliberate and work towards resolving the differences to reach an agreement. The standards for study inclusion were set based on the PICOS approach, which considers the following aspects: the particular population being studied, the intervention measures implemented, the comparison scenarios, the expected results, and the overall study architecture. The particulars of these PICOS elements are detailed as follows. Inclusion criteria The inclusion criteria for the study are as follows:Population (P): Patients diagnosed with Hhcy (homocysteine > 15 μmol/L), regardless of country, region, or ethnicity.Intervention (I): Intervention is only folic acid therapy.Comparison (C): conventional drug or placebo.Results (O): We pre-specified three tiers of outcomes for this review, with clear classification of biomarker attributes. Notably, none of the included trials provided data on adjudicated clinical cardiovascular hard endpoints (incident myocardial infarction, stroke, or cardiovascular mortality), so all outcomes are limited to changes in circulating biomarkers:Primary outcomes: Changes in plasma total homocysteine (tHcy) and folate concentrations from baseline to final follow-up;Secondary outcomes: Pre-stratified into two categories: cardiovascular risk-related surrogate biomarkers: SBP, DBP, HDL-C, LDL-C, total cholesterol (TC), triglyceride (TG);Safety outcomes: Incidence of adverse events, with a focus on cardiovascular adverse events, plus mild non-cardiovascular adverse reactions.Study design (S) : This study only incorporated Randomized Controlled Trials (RCTS). Population (P): Patients diagnosed with Hhcy (homocysteine > 15 μmol/L), regardless of country, region, or ethnicity. Population (P): Patients diagnosed with Hhcy (homocysteine > 15 μmol/L), regardless of country, region, or ethnicity. Intervention (I): Intervention is only folic acid therapy. Intervention (I): Intervention is only folic acid therapy. Comparison (C): conventional drug or placebo. Comparison (C): conventional drug or placebo. Results (O): We pre-specified three tiers of outcomes for this review, with clear classification of biomarker attributes. Notably, none of the included trials provided data on adjudicated clinical cardiovascular hard endpoints (incident myocardial infarction, stroke, or cardiovascular mortality), so all outcomes are limited to changes in circulating biomarkers:Primary outcomes: Changes in plasma total homocysteine (tHcy) and folate concentrations from baseline to final follow-up;Secondary outcomes: Pre-stratified into two categories: cardiovascular risk-related surrogate biomarkers: SBP, DBP, HDL-C, LDL-C, total cholesterol (TC), triglyceride (TG);Safety outcomes: Incidence of adverse events, with a focus on cardiovascular adverse events, plus mild non-cardiovascular adverse reactions. Results (O): We pre-specified three tiers of outcomes for this review, with clear classification of biomarker attributes. Notably, none of the included trials provided data on adjudicated clinical cardiovascular hard endpoints (incident myocardial infarction, stroke, or cardiovascular mortality), so all outcomes are limited to changes in circulating biomarkers:Primary outcomes: Changes in plasma total homocysteine (tHcy) and folate concentrations from baseline to final follow-up;Secondary outcomes: Pre-stratified into two categories: cardiovascular risk-related surrogate biomarkers: SBP, DBP, HDL-C, LDL-C, total cholesterol (TC), triglyceride (TG);Safety outcomes: Incidence of adverse events, with a focus on cardiovascular adverse events, plus mild non-cardiovascular adverse reactions. Primary outcomes: Changes in plasma total homocysteine (tHcy) and folate concentrations from baseline to final follow-up; Primary outcomes: Changes in plasma total homocysteine (tHcy) and folate concentrations from baseline to final follow-up; Secondary outcomes: Pre-stratified into two categories: cardiovascular risk-related surrogate biomarkers: SBP, DBP, HDL-C, LDL-C, total cholesterol (TC), triglyceride (TG); Secondary outcomes: Pre-stratified into two categories: cardiovascular risk-related surrogate biomarkers: SBP, DBP, HDL-C, LDL-C, total cholesterol (TC), triglyceride (TG); Safety outcomes: Incidence of adverse events, with a focus on cardiovascular adverse events, plus mild non-cardiovascular adverse reactions. Safety outcomes: Incidence of adverse events, with a focus on cardiovascular adverse events, plus mild non-cardiovascular adverse reactions. Study design (S) : This study only incorporated Randomized Controlled Trials (RCTS). Study design (S) : This study only incorporated Randomized Controlled Trials (RCTS). Exclusion criteria The items excluded from this study included conference abstracts, duplicate letters, case reports, meta-analyses, reviews, and data that were incomplete or unavailable. Besides, any studies that were not associated with the topic of the article, such as those utilizing interventions aside from Folate or B vitamin, were not considered. Data collection Among the 36 potentially eligible studies, each abstract was independently reviewed by two investigators, namely Zhi-ping Long and Fan-fan Sun, to assess its conformity with the inclusion eligibility criteria. All data from eligible trials were independently extracted in duplicate by two independent investigators using the standard protocol. Two reviewers separately extracted the data from the included literature into a Microsoft Excel spreadsheet and then summarized the data into a table. If studies present outcomes at different points in time, we extract data from the last point in time. The following data were extracted: the name of the first author, year of publication, study design, baseline characteristics (age and research area), details of intervention (folate dosage), treatment duration and outcome indicators. When data were missing or incomplete, attempts were made to contact the corresponding authors of the studies for clarification or additional information. Assessment of the risk of bias in the included studies Zhi-ping Long and Fan-fan Sun, the two reviewers, independently evaluated the bias risk of each included study using the Cochrane Collaboration’s tools [23]. If they disagreed, they’d first discuss. If still unresolved, they’d consult Shi-bo Xu. The tool distinguished trials according to six domains: Random sequence generation (selection bias), Allocation concealment (selection bias), Blinding of participants and personnel (performance bias), Blinding of outcome assessment (detection bias), Incomplete outcome data (attrition bias), Selective reporting (reporting bias), and Other bias. Each item was classified as having a low risk of bias, high risk of bias, and unclear risk of bias. Statistical analysis In the research of folate interventions in Hhcy, Review Manager 5.4 software was applied to analyze the overall and subgroup treatment effects. The data required for meta-analysis were obtained either directly from the original literature or indirectly calculated from the original data by means of relevant conversion tools. Subsequently, the pre-extracted mean, standard deviation and sample size were entered into the analysis software. For meta-analysis, when the selected studies measured tHcy, a biomarker of CVD, with identical measurement methods and units, the mean difference (MD) was computed to compare the results. Conversely, for studies using different measurement methods and units but with comparable results, the SMD was used for comparison. Heterogeneity among studies was assessed using the I² statistic, with an I² value greater than 50% indicating significant heterogeneity and an I² value less than 50% indicating no significant heterogeneity. In line with Cochrane Handbook recommendations, the choice between random-effects and fixed-effects models was pre-specified rather than determined solely by I². Given the variability in folate doses and patient types across studies, a random-effects model was selected to account for potential heterogeneity. we systematically investigated potential sources of variability through three approaches: sensitivity analyses (iteratively excluding individual studies to assess their impact on heterogeneity estimates), subgroup analyses (conducting stratified comparisons across subgroups categorized by country, patient type, and intervention characteristics), and meta-regression (quantifying the influence of baseline variables including age and tHcy levels on effect sizes). To evaluate publication bias for the primary outcomes, funnel plots were constructed, and Egger’s and Begg’s tests were performed. If significant heterogeneity persisted and its sources could not be identified, a descriptive analysis was conducted. Statistical significance was determined by a p < 0.05. Results Search flow A total of 6045 articles were obtained following a methodical search across four databases, with 1067 being from PubMed, 3299 from Embase, 1469 from Web of Science, and 210 from the Cochrane Library. Having removed duplicates, 4292 articles were left. When the titles and abstracts were read, 3345 articles were excluded due to the absence of relevant topics. Meanwhile, 199 articles were identified as animal experiments, 235 articles were previous reviews and meta-analyses, and 361 studies were found not to be randomized controlled trials. Thus, 152 potentially eligible articles were included. Nevertheless, after reviewing the full texts, it was discovered that 105 articles were conference abstracts and case reports. Also, for 11 articles, it was impossible to obtain the full text or related data. As a result, 36 articles were included in our meta-analysis. The detailed screening and inclusion process is demonstrated in Fig. 1. Fig. 1Literature selection and inclusion process Literature selection and inclusion process Research characteristics Of the 36 studies included, 10 were from China [24–33], two from Germany [34, 35], one from Iceland [36], two from the United States [37, 38], one from India [39], one from the Netherlands [40], four from Italy [41–44], two from Brazi [15, 45], two from the United Kingdom [46, 47], three from Iran [48–50], two from Greece [51, 52], one from the Czech Republic [53], one from Argentina [54], and one from South Africa [55]. Two from Australia [56, 57], and one from Turkey [58]. All studies were randomized controlled trials. All sample sizes ranged from 20 to 16,413 people and were published from 1994 to 2024. presented detailed characteristics of the included studies (Table 1). Table 1Summary of clinical studies of folic acid therapy for HhcyIncluded studiesAge(years) (con/exp)countryintervention populationPatients(con/exp)Placebo groupFolic acid dose(mg/d)Study duration of Treatment(month)CVD events and risk factorsMain outcome measuresAkoglu 2008 [34]50±11/52±13Germanyliver transplant recipients15/15Yes12CHD, stroke, PVD, ITBL,SVTVitB12,VitB6,tHcyArnadottir 2000 [36]59±16IcelandHaemodialysis patients14,14,14/14No2.14,5,101.5atherosclerosis,CHDtHcy,tCys,tGSHBao 2023 [25]63.4±6.2,62.9±5.9/63.2±6.5Chinahypertensive116,118/117No0.4,0.82HBPtHcy,SBP,DBP,Folate, SOCBeaulieu 1999 [37]46±2,50±2/44±2AmericaRenal Transplant Recipients20,20/20Yes0.4,2.43ESRD,CHDtHcyBhosale 2019 [39]17.6±12.6/16.2±16.8IndiaAdolescent Epileptics14/28Yes51HBPtHcy,SBP,DBP,Random BSL,HDL,TG,TCBrouwer 1999 [40]18-40Netherlandsnon-pregnant women45,50/49Yes0.5,0.253CAD,cerebrovascular,PVDFolate,RBCF,tHcyCagnacci 2009 [41]55.8±1.1/54.5±1.2Italypostmenopausal women15/15Yes150.75hypertension,diabetes,thyroid disordersTC,HDL,TG,Glucose,Insulin,HOMA-IR,tHcy,SBP,DBPDelfino 2007 [45]51.61±10.77/52.30±15Brazilhemodialysis patients20/26Yes106Cardiovas cularmortality,End-stage kidney diseasetHcy,TRAP,LOOHHuang 2018 [26]59.8±8.0Chinahypertensive adults8195/8218No0.854HBP,strokeTC,TG,HDL-C,Fasting Glu,tHcy,VitB12,Folate, eGFRKutschka 2001 [35]55.6±10.2/54.3±8.9GermanyHeart transplant patients35/34No53CAVDFolate,ERCF,tHcyLin 2006 [27]74.6±7.8/71.6±9.0Chinacoronary artery disease patients22/24Yes0.42CADtHcy,FolateLiu 2020 [28]59.91 ± 9.38,60.50 ± 9.79/62.46±9.98Chinacoronary heart disease patients34,34/99No0.4,0.83.25CHD,HBPFBG,tHcy,LP-PLA2,Folate, DBP,SBP,TC,TG,HDL-C,LDL-CMatsumoto 2021 [46]66.35±1.61/67.08±1.77UKoutpatients with CKD47/66No53HBP,atherosclerosiseGFR,Creatinine,Glucose,Insulin,HOMA-IR,Folate,tHcy,VitB12,AST,ALT,TC,HDL,LDL,TG,Albumin,tGSHMorgan 1998 [38]53.2±13.6AmericaPatients with rheumatoid arthritis26/26Yes512MIFolate,tHcyNafar 2009 [48]40.2±9/41.5±11.1IranKidney transplant patients with end-stage renal disease30/30Yes56atherosclerosisFolate,tHcyNtaios 2010 [51]73.2±4.6/73.9±4.3Greecepatients with cardiovascular risk53/50Yes512arterial hypertension,CAD,dyslipidaemia,ischemic strokeFolate,tHcy,IMTPapandreou 2010 [52]NDGreece children6/20No52HBPtHcy,Folate,VitB12,SBP,DBP,Albumin,CreatininePeng 2018 [29]69.8±1.6/69.4±1.5ChinaChronic Heart Failure Patients62/62No51CHF6MWD,NT-proBNP,LVEDD,LVPWT,IVST,LVEF,tHcyPotena 2002 [42]54±11ItalyHeart transplantation recipients31/17No0.2512AVDtHcy,Folate,VitB12,CreatinineQin 2012 [24]57±9.8ChinaHypertensive adults147,146/152No0.4,0.82HBPtHcyRighetti 2003 [43]63±3,65±3/64±2Italyhemodialysis patients26,25/30No5,1512atherosclerosistHcySharifi 2010 [49]42±3/41±2Iranhypertensiveadults18/19Yes51.5HBPtHcy,Folate,VitB12,SBP,DBPSheu 2005 [30]43±2/40±2Chinaobese women38/36Yes53gastroplasty surgery,atherosclerosisFPG,FPI,HOMA-IR,TC,LDL-C,HDL-C,TG,VitB12,Folate,tHcyShidfar 2009 [50]44±7.06/45±7.78IranHypercholesterolemic Adults20/20Yes52hypercholesteremiaTC,TG,LDL-C,HDL-C,Folate,tHcy, TACStulc 2001 [53]NDCzech Republicsubjects with mixed hyperlipidemia11/11No52.5atherosclerosistHcy,Folate,CreatinineThambyrajah 2001 [47]63.0 (60.6–65.5)/63.4 (61.2–65.5)UKPatients with unequivocal angiographic evidence of CAD43/43Yes53CHDFolate,tHcy,EDD,vWF,Nitrate/nitriteTrimarchi 2002 [54]62.2±10.6/ 50.7±18.6Argentinahemodialysis patients16/13No104ESRD, chronic hemodialysis,atherosclerosistHcy,VitB12,FolateUbbink 1994 [55]40.0±13.5/40.6±14.5South AfricaND19/17Yes0.651.5MI,atherosclerosis,stenocardia,CHFtHcy,Folate,VitB12,PLPVianna 2007 [15]49.3±13.5/47.6±12.3Brazilpatients with end-stage kidney disease93/93Yes524atherosclerosisFolate,tHcy,TC,TG,HDL-C,LDL-C,Albumin,Calcium,Parathormone,C-reactive protein,VitB12,IMTVilla 2005 [44]55.4±2.2/53.1±1.8Italyhealthy postmenopausal women10/10Yes7.52atherosclerosis,TVDSBP,DBP,Cretinine,Estradiol,Estrone,Androstenedione,Methionine,VitB12,tHcy,TC,TG,HDL-C,LDL-C,VLDL-C,Fasting glucose,Fasting insulin,Fasting C-peptideWoo 1999 [31]54±10ChinaND17/17Yes102atherosclerosisFolate,tHcy,Vitamin B12,TC,EDD,GTNWoodman 2004 [56]49±2AustraliaND13/13Yes52atherosclerosistHcy,RBCF,FMD,GTNXu 2016 [32]60.3±8.3/60.1±8.1ChinaParticipants With CKD867/804No0.83stroketHcy,Folate,eGFRYilmaz 2007 [58]52.2±11.9/65.5±7.6Turkeypatients with coronary artery disease20/20Yes52atherosclerosis,CHDtHcy,Folate,EDD,NEDDZhang 2022 [33]63.5±2.2/63.9±2.2ChinaH-type hypertension55/55Yes0.83HBP,atherosclerosisSBP,DBP,FIB,DD,PT,APTT,TT,HCT,PLT,tHcy,ET-1Zoungas 2006 [57]56±13/56±14AustraliaPatients with chronic renal failure156/159Yes1543.2atherosclerosisIMT,tHcyCon control group, Exp experimental group, ND non-determined, RCT randomized controlled trial, tHcy total plasma homocysteine, tCys total plasma cysteine, tGSH total plasma glutathione, VitB12 vitamin B12, SOC system organ classification, TC cholesterol, RBCF red blood cell folate, TG triglycerides, TC total cholesterol, TRAP total plasma antioxidant potential, LOOH lipid hydroperoxide, eGFR estimated glomerular filtration rate, ERCF erythrocyte folate, FBG fasting blood glucose, LP-PLA2 lipoprotein-associated phospholipase A2, DBP diastolic blood pressure, SBP systolic blood pressure, HDL-C high density lipoprotein cholesterol, LDL-C low density lipoprotein cholesterol, HOMA-IR homeostasis model of assessment in insulin resistance, AST aspartate transaminase, ALT alanine transaminase, IMT intima-media thickness, 6MWD 6-minute walk distance, NT-proBNP N-Terminal pro Brain Natriuretic Peptide, LVEDD left ventricular end-diastolic dimension, LVPWT left ventricular posterior wall thickness, IVST interventricular septal thickness, LVEF left ventricular ejection fraction, FPG fasting plasma Glucose, FPI fasting plasma Insulin, TAC total antioxidant capacity, PLP plasma pyridoxal 5'-phosphate, EDD endothelium-dependent dilation, GTN glyceryl trinitrate (nitroglycerin), FMD flow-mediated dilatation, NEDD endothelium-independent dilation, FIB fibrinogen, DD D-dimer turbidimetric immunoassay, PT prothrombin time, APTT activated partial thromboplastin time, TT thrombin time, HCT hematocrit, PLT platelet, ET-1 endothelin 1, CHD chronic heart disease, PVD peripheral vascular disease, ITBL ischemic type biliary leason, SVT sinus vein thrombosis, HBP high blood pressure, ESRD end-stage renal disease, CAD coronary artery disease, PVD peripheral vascular diseases, CAVD cardiac allograft vascular disease, MI myocardial infarction, CHF chronic heart failure, AVD allograft vascular disease, TVD thromboembolic vascular diseases, VitB6 vitamin B6 Table 1Summary of clinical studies of folic acid therapy for HhcyIncluded studiesAge(years) (con/exp)countryintervention populationPatients(con/exp)Placebo groupFolic acid dose(mg/d)Study duration of Treatment(month)CVD events and risk factorsMain outcome measuresAkoglu 2008 [34]50±11/52±13Germanyliver transplant recipients15/15Yes12CHD, stroke, PVD, ITBL,SVTVitB12,VitB6,tHcyArnadottir 2000 [36]59±16IcelandHaemodialysis patients14,14,14/14No2.14,5,101.5atherosclerosis,CHDtHcy,tCys,tGSHBao 2023 [25]63.4±6.2,62.9±5.9/63.2±6.5Chinahypertensive116,118/117No0.4,0.82HBPtHcy,SBP,DBP,Folate, SOCBeaulieu 1999 [37]46±2,50±2/44±2AmericaRenal Transplant Recipients20,20/20Yes0.4,2.43ESRD,CHDtHcyBhosale 2019 [39]17.6±12.6/16.2±16.8IndiaAdolescent Epileptics14/28Yes51HBPtHcy,SBP,DBP,Random BSL,HDL,TG,TCBrouwer 1999 [40]18-40Netherlandsnon-pregnant women45,50/49Yes0.5,0.253CAD,cerebrovascular,PVDFolate,RBCF,tHcyCagnacci 2009 [41]55.8±1.1/54.5±1.2Italypostmenopausal women15/15Yes150.75hypertension,diabetes,thyroid disordersTC,HDL,TG,Glucose,Insulin,HOMA-IR,tHcy,SBP,DBPDelfino 2007 [45]51.61±10.77/52.30±15Brazilhemodialysis patients20/26Yes106Cardiovas cularmortality,End-stage kidney diseasetHcy,TRAP,LOOHHuang 2018 [26]59.8±8.0Chinahypertensive adults8195/8218No0.854HBP,strokeTC,TG,HDL-C,Fasting Glu,tHcy,VitB12,Folate, eGFRKutschka 2001 [35]55.6±10.2/54.3±8.9GermanyHeart transplant patients35/34No53CAVDFolate,ERCF,tHcyLin 2006 [27]74.6±7.8/71.6±9.0Chinacoronary artery disease patients22/24Yes0.42CADtHcy,FolateLiu 2020 [28]59.91 ± 9.38,60.50 ± 9.79/62.46±9.98Chinacoronary heart disease patients34,34/99No0.4,0.83.25CHD,HBPFBG,tHcy,LP-PLA2,Folate, DBP,SBP,TC,TG,HDL-C,LDL-CMatsumoto 2021 [46]66.35±1.61/67.08±1.77UKoutpatients with CKD47/66No53HBP,atherosclerosiseGFR,Creatinine,Glucose,Insulin,HOMA-IR,Folate,tHcy,VitB12,AST,ALT,TC,HDL,LDL,TG,Albumin,tGSHMorgan 1998 [38]53.2±13.6AmericaPatients with rheumatoid arthritis26/26Yes512MIFolate,tHcyNafar 2009 [48]40.2±9/41.5±11.1IranKidney transplant patients with end-stage renal disease30/30Yes56atherosclerosisFolate,tHcyNtaios 2010 [51]73.2±4.6/73.9±4.3Greecepatients with cardiovascular risk53/50Yes512arterial hypertension,CAD,dyslipidaemia,ischemic strokeFolate,tHcy,IMTPapandreou 2010 [52]NDGreece children6/20No52HBPtHcy,Folate,VitB12,SBP,DBP,Albumin,CreatininePeng 2018 [29]69.8±1.6/69.4±1.5ChinaChronic Heart Failure Patients62/62No51CHF6MWD,NT-proBNP,LVEDD,LVPWT,IVST,LVEF,tHcyPotena 2002 [42]54±11ItalyHeart transplantation recipients31/17No0.2512AVDtHcy,Folate,VitB12,CreatinineQin 2012 [24]57±9.8ChinaHypertensive adults147,146/152No0.4,0.82HBPtHcyRighetti 2003 [43]63±3,65±3/64±2Italyhemodialysis patients26,25/30No5,1512atherosclerosistHcySharifi 2010 [49]42±3/41±2Iranhypertensiveadults18/19Yes51.5HBPtHcy,Folate,VitB12,SBP,DBPSheu 2005 [30]43±2/40±2Chinaobese women38/36Yes53gastroplasty surgery,atherosclerosisFPG,FPI,HOMA-IR,TC,LDL-C,HDL-C,TG,VitB12,Folate,tHcyShidfar 2009 [50]44±7.06/45±7.78IranHypercholesterolemic Adults20/20Yes52hypercholesteremiaTC,TG,LDL-C,HDL-C,Folate,tHcy, TACStulc 2001 [53]NDCzech Republicsubjects with mixed hyperlipidemia11/11No52.5atherosclerosistHcy,Folate,CreatinineThambyrajah 2001 [47]63.0 (60.6–65.5)/63.4 (61.2–65.5)UKPatients with unequivocal angiographic evidence of CAD43/43Yes53CHDFolate,tHcy,EDD,vWF,Nitrate/nitriteTrimarchi 2002 [54]62.2±10.6/ 50.7±18.6Argentinahemodialysis patients16/13No104ESRD, chronic hemodialysis,atherosclerosistHcy,VitB12,FolateUbbink 1994 [55]40.0±13.5/40.6±14.5South AfricaND19/17Yes0.651.5MI,atherosclerosis,stenocardia,CHFtHcy,Folate,VitB12,PLPVianna 2007 [15]49.3±13.5/47.6±12.3Brazilpatients with end-stage kidney disease93/93Yes524atherosclerosisFolate,tHcy,TC,TG,HDL-C,LDL-C,Albumin,Calcium,Parathormone,C-reactive protein,VitB12,IMTVilla 2005 [44]55.4±2.2/53.1±1.8Italyhealthy postmenopausal women10/10Yes7.52atherosclerosis,TVDSBP,DBP,Cretinine,Estradiol,Estrone,Androstenedione,Methionine,VitB12,tHcy,TC,TG,HDL-C,LDL-C,VLDL-C,Fasting glucose,Fasting insulin,Fasting C-peptideWoo 1999 [31]54±10ChinaND17/17Yes102atherosclerosisFolate,tHcy,Vitamin B12,TC,EDD,GTNWoodman 2004 [56]49±2AustraliaND13/13Yes52atherosclerosistHcy,RBCF,FMD,GTNXu 2016 [32]60.3±8.3/60.1±8.1ChinaParticipants With CKD867/804No0.83stroketHcy,Folate,eGFRYilmaz 2007 [58]52.2±11.9/65.5±7.6Turkeypatients with coronary artery disease20/20Yes52atherosclerosis,CHDtHcy,Folate,EDD,NEDDZhang 2022 [33]63.5±2.2/63.9±2.2ChinaH-type hypertension55/55Yes0.83HBP,atherosclerosisSBP,DBP,FIB,DD,PT,APTT,TT,HCT,PLT,tHcy,ET-1Zoungas 2006 [57]56±13/56±14AustraliaPatients with chronic renal failure156/159Yes1543.2atherosclerosisIMT,tHcyCon control group, Exp experimental group, ND non-determined, RCT randomized controlled trial, tHcy total plasma homocysteine, tCys total plasma cysteine, tGSH total plasma glutathione, VitB12 vitamin B12, SOC system organ classification, TC cholesterol, RBCF red blood cell folate, TG triglycerides, TC total cholesterol, TRAP total plasma antioxidant potential, LOOH lipid hydroperoxide, eGFR estimated glomerular filtration rate, ERCF erythrocyte folate, FBG fasting blood glucose, LP-PLA2 lipoprotein-associated phospholipase A2, DBP diastolic blood pressure, SBP systolic blood pressure, HDL-C high density lipoprotein cholesterol, LDL-C low density lipoprotein cholesterol, HOMA-IR homeostasis model of assessment in insulin resistance, AST aspartate transaminase, ALT alanine transaminase, IMT intima-media thickness, 6MWD 6-minute walk distance, NT-proBNP N-Terminal pro Brain Natriuretic Peptide, LVEDD left ventricular end-diastolic dimension, LVPWT left ventricular posterior wall thickness, IVST interventricular septal thickness, LVEF left ventricular ejection fraction, FPG fasting plasma Glucose, FPI fasting plasma Insulin, TAC total antioxidant capacity, PLP plasma pyridoxal 5'-phosphate, EDD endothelium-dependent dilation, GTN glyceryl trinitrate (nitroglycerin), FMD flow-mediated dilatation, NEDD endothelium-independent dilation, FIB fibrinogen, DD D-dimer turbidimetric immunoassay, PT prothrombin time, APTT activated partial thromboplastin time, TT thrombin time, HCT hematocrit, PLT platelet, ET-1 endothelin 1, CHD chronic heart disease, PVD peripheral vascular disease, ITBL ischemic type biliary leason, SVT sinus vein thrombosis, HBP high blood pressure, ESRD end-stage renal disease, CAD coronary artery disease, PVD peripheral vascular diseases, CAVD cardiac allograft vascular disease, MI myocardial infarction, CHF chronic heart failure, AVD allograft vascular disease, TVD thromboembolic vascular diseases, VitB6 vitamin B6 Summary of clinical studies of folic acid therapy for Hhcy Con control group, Exp experimental group, ND non-determined, RCT randomized controlled trial, tHcy total plasma homocysteine, tCys total plasma cysteine, tGSH total plasma glutathione, VitB12 vitamin B12, SOC system organ classification, TC cholesterol, RBCF red blood cell folate, TG triglycerides, TC total cholesterol, TRAP total plasma antioxidant potential, LOOH lipid hydroperoxide, eGFR estimated glomerular filtration rate, ERCF erythrocyte folate, FBG fasting blood glucose, LP-PLA2 lipoprotein-associated phospholipase A2, DBP diastolic blood pressure, SBP systolic blood pressure, HDL-C high density lipoprotein cholesterol, LDL-C low density lipoprotein cholesterol, HOMA-IR homeostasis model of assessment in insulin resistance, AST aspartate transaminase, ALT alanine transaminase, IMT intima-media thickness, 6MWD 6-minute walk distance, NT-proBNP N-Terminal pro Brain Natriuretic Peptide, LVEDD left ventricular end-diastolic dimension, LVPWT left ventricular posterior wall thickness, IVST interventricular septal thickness, LVEF left ventricular ejection fraction, FPG fasting plasma Glucose, FPI fasting plasma Insulin, TAC total antioxidant capacity, PLP plasma pyridoxal 5'-phosphate, EDD endothelium-dependent dilation, GTN glyceryl trinitrate (nitroglycerin), FMD flow-mediated dilatation, NEDD endothelium-independent dilation, FIB fibrinogen, DD D-dimer turbidimetric immunoassay, PT prothrombin time, APTT activated partial thromboplastin time, TT thrombin time, HCT hematocrit, PLT platelet, ET-1 endothelin 1, CHD chronic heart disease, PVD peripheral vascular disease, ITBL ischemic type biliary leason, SVT sinus vein thrombosis, HBP high blood pressure, ESRD end-stage renal disease, CAD coronary artery disease, PVD peripheral vascular diseases, CAVD cardiac allograft vascular disease, MI myocardial infarction, CHF chronic heart failure, AVD allograft vascular disease, TVD thromboembolic vascular diseases, VitB6 vitamin B6 Risk assessment of bias Overall, most studies (32/36) had unclear allocation concealment, and approximately half (18/36) had high risk of detection bias due to inadequate blinding of outcome assessment. The outcomes of the assessment concerning the risk of bias and methodological suitability for the included studies were depicted in Figs. 2 and 3. We know from the figure that the first and last two items were low risk of bias [15, 24–58]. Allocation concealment (selection bias) 3 studies were low risk of bias [25, 38, 41], 1 study were high risk of bias [45], and 32 studies had unclear risk of bias [15, 24, 26, 27, 39, 40, 42–44, 46–58]. Blinding of participants and personnel (performance bias) 4 studies were high risk of bias [35, 44, 46, 53], and 32 studies were low risk of bias [15, 24–34, 36–43, 45, 47–52, 54–58]. Blinding of outcome assessment (detection bias) 17 studies were low risk of bias [24, 26, 28, 31, 34, 36–38, 40, 41, 43–45, 49, 51, 52, 54], 18 studies were high risk of bias [15, 25, 27, 30, 32, 33, 35, 39, 42, 46–48, 50, 53, 55–58], and 1 study were unclear risk of bias [29]. Incomplete outcome data (attrition bias) 30 studies were low risk of bias [15, 25–31, 33–48, 51–53, 56–58], and 6 studies were high risk of bias [24, 32, 49, 50, 54, 55]. Fig. 2Risk of bias graph: review authors’ judgments about each risk of bias item presented as percentages across all included studies Risk of bias graph: review authors’ judgments about each risk of bias item presented as percentages across all included studies Fig. 3Risk of bias summary: review authors’ judgments about each risk of bias item for each included studies Risk of bias summary: review authors’ judgments about each risk of bias item for each included studies Meta-analysis Random effects model and sensitivity analysis were used to include 36 eligible articles for meta-analysis. tHcy level and folate level were used as the primary outcomes, and changes in systolic blood pressure, diastolic blood pressure, HDL-C, LDL-C, TG and TC (all defined as CVD-related surrogate biomarkers) were used as the secondary outcomes to evaluate the efficacy of folic acid treatment for Hhcy. Adverse events were assessed as safety outcomes. Primary indicators tHcy levels In 36 studies with 10,851 patients in the folate group and 10,876 patients in the placebo group, tHcy levels were reported. Due to the significant difference in mean values among studies, the random effects model was employed as the effect indicator. The results of the forest plot indicated that there were statistically significant differences between the two groups (SMD=-0.99; 95% CI [-1.15, -0.83]; p < 0.00001; heterogeneity test p < 0.00001; I² = 91%). However, in order to investigate the influence of various factors on the efficacy of folic acid for Hhcy, we performed a subgroup analysis of tHcy. (Fig. 4). Fig. 4Forest plot of primary indicator: pooled results of tHcy levels Forest plot of primary indicator: pooled results of tHcy levels Folate levels For 23 studies involving 9962 patients in the folate group and 9984 patients in the placebo group, folate levels were reported. Due to the substantial difference in mean values among the studies, the random effects model was utilized as the effect indicator. The results of the forest plot indicated that there were statistically significant differences between the two groups (SMD = 1.61; 95% CI [1.30, 1.93]; p < 0.00001; heterogeneity test p < 0.00001; I²= 96%). (Fig. 5). Fig. 5Forest plot of primary indicator: pooled results of folic acid levels Forest plot of primary indicator: pooled results of folic acid levels Folic acid dose subgroup Folic acid dose subgroup of tHcy levels To investigate the impact of folate supplementation on Hcy at various doses after administration, a dose subgroup analysis for Hcy levels was conducted. The pooled analysis indicated that the Hcy levels in the folate group (SMD=-0.99; 95% CI [-1.15, -0.83]; heterogeneity test p < 0.00001; I²= 91%) were compared with those in the placebo group. The subgroup analysis using the random effects model showed that the Hcy level in the folic acid group within the dose range of [0 to 0.4) mg was (SMD=-0.64; 95% CI [-0.98, -0.31]; p = 0.0002), for the folic acid group within the dose range of [0.4 to 0.8) mg was (SMD=-0.47; 95% CI [-0.69, -0.25]; p < 0.0001), for the folic acid group within the dose range of [0.8 to 1) mg was (SMD=-0.43; 95% CI [-0.60, -0.27]; p < 0.00001), for the folic acid group within the dose range of [1 to 5) mg was (SMD=-0.96; 95% CI [-1.61, -0.31]; p = 0.004), for the folic acid group within the dose range of [5 to 10) mg was (SMD=-1.42; 95% CI [-1.75, -1.09]; p < 0.00001), and for the folic acid group within the dose range of [10 to 15] mg was (SMD=-1.80; 95% CI [-2.80, -0.79]; p = 0.0005). In these exploratory subgroup analyses, larger SMDs were observed at higher doses. Notably, these subgroup findings are hypothesis-generating only; no formal dose-response meta-regression was performed in this meta-analysis, and no optimal folic acid dose can be definitively determined from these results. (Fig. 6). Fig. 6Folic acid dose subgroup Folic acid dose subgroup Study duration of treatment Time subgroup of tHcy levels In order to study the effects of different treatment times on Hcy after administration, a subgroup analysis of Hcy levels was performed.The pooled analysis indicated that the Hcy levels in the folate group (SMD=-0.99; 95% CI [-1.15, -0.83]; heterogeneity test p < 0.00001; I²= 91%) were compared with those in the placebo group. The subgroup analysis using the random effects model showed that the Hcy level in the folic acid group within the time range of [0 to 3) month was (SMD=-1.05; 95% CI [-1.30, -0.79]; p < 0.00001), for the folic acid group within the time range of [3 to 6) month was (SMD=-0.66; 95% CI [-0.94, -0.38]; p < 0.00001), for the folic acid group within the time range of ≥ 6 month was (SMD=-1.59; 95% CI [-2.12, -1.06]; p < 0.0001). (Fig. 7). Fig. 7Study duration of treatment subgroup Study duration of treatment subgroup Subgroup analysis and meta-regression analysis of patient characteristics This meta-analysis found a significant overall tHcy-lowering effect of folic acid (pooled SMD = − 0.89, 95% CI − 1.13 to − 0.66; I²=91%), consistent with the primary forest plot (Fig. 4). Pre-specified subgroup analyses were performed to explore sources of heterogeneity, with full results detailed in Table 2:Geographic region: The largest tHcy reduction was observed in Italy (SMD = − 3.24, I²=95.3%), followed by Greece (SMD = − 0.87, I²=90.7%), with near-zero heterogeneity in folate-fortified regions including Germany (SMD = − 0.50, I²=0%) and Iceland (SMD = − 1.15, I²=0%);Population type: Renal patients had the greatest tHcy reduction (SMD = − 1.60, I²=95.1%), while non-renal transplant recipients showed homogeneous, consistent responses (SMD = − 0.54, I²=0%);Study design: Non-placebo-controlled trials had substantially higher heterogeneity (I²=90.6%) than placebo-controlled trials (I²=68.5%);Genetic background: High heterogeneity was observed in studies from China (I²=86.3%), a region with high MTHFR C677T variant prevalence. Geographic region: The largest tHcy reduction was observed in Italy (SMD = − 3.24, I²=95.3%), followed by Greece (SMD = − 0.87, I²=90.7%), with near-zero heterogeneity in folate-fortified regions including Germany (SMD = − 0.50, I²=0%) and Iceland (SMD = − 1.15, I²=0%); Geographic region: The largest tHcy reduction was observed in Italy (SMD = − 3.24, I²=95.3%), followed by Greece (SMD = − 0.87, I²=90.7%), with near-zero heterogeneity in folate-fortified regions including Germany (SMD = − 0.50, I²=0%) and Iceland (SMD = − 1.15, I²=0%); Population type: Renal patients had the greatest tHcy reduction (SMD = − 1.60, I²=95.1%), while non-renal transplant recipients showed homogeneous, consistent responses (SMD = − 0.54, I²=0%); Population type: Renal patients had the greatest tHcy reduction (SMD = − 1.60, I²=95.1%), while non-renal transplant recipients showed homogeneous, consistent responses (SMD = − 0.54, I²=0%); Study design: Non-placebo-controlled trials had substantially higher heterogeneity (I²=90.6%) than placebo-controlled trials (I²=68.5%); Study design: Non-placebo-controlled trials had substantially higher heterogeneity (I²=90.6%) than placebo-controlled trials (I²=68.5%); Genetic background: High heterogeneity was observed in studies from China (I²=86.3%), a region with high MTHFR C677T variant prevalence. Genetic background: High heterogeneity was observed in studies from China (I²=86.3%), a region with high MTHFR C677T variant prevalence. Table 2Subgroup analysis and meta-regression analysis of patient characteristicsSMD95%CIdfp-value I 2 Tau-squared Z P Country Germany-0.497-0.904,-0.09010.6730.00%02.330.02 Iceland-1.149-1.647,-0.65120.9530.00%05.80 China-0.481-0.632,-0.33111086.30%0.04966.410 America-0.623-1.090,-0.15620.14548.30%0.09262.730.006 Netherlands-0.596-0.891,-0.30110.680.00%03.920 Italy-3.235-5.315,-1.1554095.30%5.1943.160.002 Brazil-1.381-1.862,-0.90110.8260.00%09.490 UK-1.511-2.908,-0.11410.07668.20%0.10564.360 Iran-1.026-1.406,-0.64520.11453.90%0.12834.460 Greece-0.869-1.578,-0.16010.00190.70%2.26331.610.107Placebo Control Yes-0.898-1.135,-0.66120068.50%0.19367.420 No-0.759-0.959,-0.55819090.60%0.12867.410Patient Type Non-Renal Transplant Recipients-0.538-0.873, -0.20420.8090.00%03.160.002 Renal Disease Patients-1.596-2.159,-1.03313095.10%1.02635.560 Cardiovascular Disease Patients-0.540-0.733,-0.34710084.30%0.07555.470 Other Population Groups-1.459-2.053,-0.8659086.40%0.72964.810 Patients with Metabolic Disorders-1.115-1.783,-0.44720.0566.50%0.22973.270.001standardised mean difference, SMD confidence interval, CI degrees of freedom, df heterogeneity statistic, I2; p < 0.05 was considered statistically significant Table 2Subgroup analysis and meta-regression analysis of patient characteristicsSMD95%CIdfp-value I 2 Tau-squared Z P Country Germany-0.497-0.904,-0.09010.6730.00%02.330.02 Iceland-1.149-1.647,-0.65120.9530.00%05.80 China-0.481-0.632,-0.33111086.30%0.04966.410 America-0.623-1.090,-0.15620.14548.30%0.09262.730.006 Netherlands-0.596-0.891,-0.30110.680.00%03.920 Italy-3.235-5.315,-1.1554095.30%5.1943.160.002 Brazil-1.381-1.862,-0.90110.8260.00%09.490 UK-1.511-2.908,-0.11410.07668.20%0.10564.360 Iran-1.026-1.406,-0.64520.11453.90%0.12834.460 Greece-0.869-1.578,-0.16010.00190.70%2.26331.610.107Placebo Control Yes-0.898-1.135,-0.66120068.50%0.19367.420 No-0.759-0.959,-0.55819090.60%0.12867.410Patient Type Non-Renal Transplant Recipients-0.538-0.873, -0.20420.8090.00%03.160.002 Renal Disease Patients-1.596-2.159,-1.03313095.10%1.02635.560 Cardiovascular Disease Patients-0.540-0.733,-0.34710084.30%0.07555.470 Other Population Groups-1.459-2.053,-0.8659086.40%0.72964.810 Patients with Metabolic Disorders-1.115-1.783,-0.44720.0566.50%0.22973.270.001standardised mean difference, SMD confidence interval, CI degrees of freedom, df heterogeneity statistic, I2; p < 0.05 was considered statistically significant Subgroup analysis and meta-regression analysis of patient characteristics standardised mean difference, SMD confidence interval, CI degrees of freedom, df heterogeneity statistic, I2; p < 0.05 was considered statistically significant Meta-regression analyses revealed no significant association between patient age and the effect of folate intervention (β = 0.01, 95% CI [-0.13, 0.14], p = 0.344), nor between baseline homocysteine (tHcy) levels and the effect of folate intervention (β = 0.15, 95% CI [-0.08, 0.38], p = 0.207). This suggests that differences in the responses to folate treatment among study subjects of varying ages and baseline tHcy levels are not statistically significant. Secondary indicators SBP and DBP levels In this systematic review, SBP levels were reported in six studies of 383 patients in the Folate group and 486 patients in the placebo group. The meta-analysis showed a significant pooled effect size between the two groups (SMD=-0.54; 95% CI [− 0.89, -0.18]; p = 0.003; heterogeneity test p < 0.0001; I2 = 80%) .(Fig. 8) DBP levels were reported in 6 studies of 383 patients in the Folate group and 486 patients in the placebo group. The meta-analysis showed a significant pooled effect size between the two groups (SMD=-0.32; 95% CI [− 0.58, -0.05]; p = 0.02; heterogeneity test p = 0.008; I2 = 63%).(Fig. 9). Fig. 8Forest plot of secondary outcome: pooled results of SBP levels Forest plot of secondary outcome: pooled results of SBP levels Fig. 9Forest plot of secondary outcome: pooled results of DBP levels Forest plot of secondary outcome: pooled results of DBP levels HDL-C and LDL-C levels In this systematic review, HDL levels were reported in 6 studies of 177 patients in the Folate group and 295 patients in the placebo group. The meta-analysis showed a significant pooled effect size between the two groups, with a WMD of + 1.37 (95% CI [0.71, 2.03]; p<0.0001; heterogeneity test p = 0.06; I2 = 50%).(Fig. 10) LDL levels were reported in 4 studies of 134 patients in the Folate group and 266 patients in the placebo group. The meta-analysis showed a significant pooled effect size between the two groups, with a WMD of − 3.64 (95% CI [-5.91, -1.36]; p = 0.002; heterogeneity test p = 0.25; I2 = 26%). (Fig. 11). Fig. 10Forest plot of secondary outcome: pooled results of HDL-C levels Forest plot of secondary outcome: pooled results of HDL-C levels Fig. 11Forest plot of secondary outcome: pooled results of LDL-C levels Forest plot of secondary outcome: pooled results of LDL-C levels TG and TC levels In this systematic review, TG levels were reported in 5 studies of 177 patients in the Folate group and 295 patients in the placebo group. The meta-analysis showed a significant pooled effect size between the two groups (SMD=-0.06; 95% CI [-0.39, -0.27]; p = 0.72; heterogeneity test p = 0.02; I2 = 61%).(Fig. 12) TC levels were reported in 7 studies of 194 patients in the Folate group and 312 patients in the placebo group. The meta-analysis showed a significant pooled effect size between the two groups (SMD=-0.15; 95% CI [-0.72, 0.42]; p = 0.61; heterogeneity test p < 0.00001; I2 = 88%).(Fig. 13). Fig. 12Forest plot of secondary outcome: pooled results of TG levels Forest plot of secondary outcome: pooled results of TG levels Fig. 13Forest plot of secondary outcome: pooled results of TC levels Forest plot of secondary outcome: pooled results of TC levels Sensitivity analysis To test the stability and reliability of the meta-analysis results, the sensitivity analysis of the main outcome indicators was carried out. Forest plots of tHcy levels showed heterogeneity test p < 0.0001; I2 = 91% (Fig. 4). Righetti et al. The influence of hemodialysis patients on tHcy levels may be the source of heterogeneity [43]. We found that when studies were deleted, heterogeneity was reduced and the results for tHcy levels were meaningful(SMD=-0.86; 95% CI[− 1.00,-0.71]; p < 0.00001; heterogeneity test p < 0.00001; I2 = 88%) (Fig. 14), which may be related to inconsistent inclusion of participants in the study, leading to publication bias in the conclusions. Fig. 14Forest plot of sensitivity analysis: pooled results of tHcy levels after excluding heterogeneous studies Forest plot of sensitivity analysis: pooled results of tHcy levels after excluding heterogeneous studies Adverse events assessment Safety remains a key concern in folic acid supplementation for hyperhomocysteinemia. We systematically evaluated adverse reactions across all included studies, with a focus on cardiovascular safety. Routine non-cardiovascular adverse reactions were mild and predominantly gastrointestinal, including anorexia, nausea, and abdominal distension, which were self-limiting and did not require treatment discontinuation. Regarding cardiovascular adverse events, these were exclusively reported in three trials administering 15 mg/day folic acid [41, 43, 58] (Table 3): Zoungas et al. documented 77 cardiovascular events (14.9 per 100 patient-years) including myocardial infarction, stroke, cardiovascular death, unstable angina, revascularization, and peripheral vascular disease; Righetti et al. reported 24 new cardiovascular events (angina in 9 cases, myocardial infarction in 10 cases, carotid artery stenosis in 1 case, thrombotic stroke in 4 cases); Cagnacci et al. recorded 3 cardiovascular events (2 cases of angina, 1 case of stroke). Notably, no cardiovascular adverse events were observed in any trial using folic acid doses ≤ 10 mg/day, confirming the dose-dependent safety profile of folic acid supplementation [41, 43, 57]. (Table 3). Table 3Information on adverse reactions after folic acid supplementationFirst AuthorNumber and Specific Types of Cardiovascular Adverse EventsCriteria for Judging Cardiovascular Adverse EventsMarco RighettiA total of 24 new cardiovascular events: 9 cases of angina pectoris, 10 cases of myocardial infarction, 1 case of carotid artery stenosis, and 4 cases of thrombotic stroke1. Angina pectoris: Typical medical history combined with abnormal myocardial scintigraphy or coronary angiography results; 2. Myocardial infarction: Abnormal serial 12-lead electrocardiogram accompanied by elevated myocardial enzymes; 3. Symptomatic extracranial carotid artery stenosis: Definite stenotic lesions requiring carotid endarterectomy; 4. Stroke: Neurological examination confirming new-onset focal neurological deficits, with or without computed tomography evidence of cerebral infarction; Event analysis was conducted in a blinded manner (without knowledge of patients' homocysteine levels)A Cagnacci3 cases: 2 cases of angina pectoris and 1 case of strokeClinical routine diagnostic criteria (consistent with recognized cardiovascular event judgment standards in the field)Sophia ZoungasFolic acid group (15 mg/d): 77 cardiovascular events (including myocardial infarction, stroke, cardiovascular death, unstable angina, revascularization, and peripheral vascular disease), incidence rate of 14.9 per 100 patient-years;1. Primary composite endpoint: Definitively diagnosed myocardial infarction, stroke, and cardiovascular-related death; 2. Secondary endpoints: All fatal and non-fatal cardiovascular events (including the aforementioned types); 3. All clinical endpoints were adjudicated blindly by an independent endpoint monitoring committee to ensure judgment consistencyPlacebo group: 86 cardiovascular events (same types as the folic acid group), incidence rate of 16.3 per 100 patient-years;Primary composite endpoint (myocardial infarction, stroke, cardiovascular death): 46 cases in the folic acid group (8.9 per 100 patient-years), 55 cases in the placebo group (10.4 per 100 patient-years)CV Cardiovascular, MI myocardial infarction, PVD peripheral vascular disease Table 3Information on adverse reactions after folic acid supplementationFirst AuthorNumber and Specific Types of Cardiovascular Adverse EventsCriteria for Judging Cardiovascular Adverse EventsMarco RighettiA total of 24 new cardiovascular events: 9 cases of angina pectoris, 10 cases of myocardial infarction, 1 case of carotid artery stenosis, and 4 cases of thrombotic stroke1. Angina pectoris: Typical medical history combined with abnormal myocardial scintigraphy or coronary angiography results; 2. Myocardial infarction: Abnormal serial 12-lead electrocardiogram accompanied by elevated myocardial enzymes; 3. Symptomatic extracranial carotid artery stenosis: Definite stenotic lesions requiring carotid endarterectomy; 4. Stroke: Neurological examination confirming new-onset focal neurological deficits, with or without computed tomography evidence of cerebral infarction; Event analysis was conducted in a blinded manner (without knowledge of patients' homocysteine levels)A Cagnacci3 cases: 2 cases of angina pectoris and 1 case of strokeClinical routine diagnostic criteria (consistent with recognized cardiovascular event judgment standards in the field)Sophia ZoungasFolic acid group (15 mg/d): 77 cardiovascular events (including myocardial infarction, stroke, cardiovascular death, unstable angina, revascularization, and peripheral vascular disease), incidence rate of 14.9 per 100 patient-years;1. Primary composite endpoint: Definitively diagnosed myocardial infarction, stroke, and cardiovascular-related death; 2. Secondary endpoints: All fatal and non-fatal cardiovascular events (including the aforementioned types); 3. All clinical endpoints were adjudicated blindly by an independent endpoint monitoring committee to ensure judgment consistencyPlacebo group: 86 cardiovascular events (same types as the folic acid group), incidence rate of 16.3 per 100 patient-years;Primary composite endpoint (myocardial infarction, stroke, cardiovascular death): 46 cases in the folic acid group (8.9 per 100 patient-years), 55 cases in the placebo group (10.4 per 100 patient-years)CV Cardiovascular, MI myocardial infarction, PVD peripheral vascular disease Information on adverse reactions after folic acid supplementation CV Cardiovascular, MI myocardial infarction, PVD peripheral vascular disease Publication bias Quantitative assessment of publication bias using Egger’s linear regression and Begg’s rank-correlation tests revealed significant asymmetry for tHcy (Egger: t = − 7.13, P < 0.001; Begg: z = − 2.18, P = 0.029) and folate levels (Egger: t = 2.35, P = 0.027; Begg: z = 2.53, P = 0.011), suggesting potential small-study effects or selective outcome reporting. For LDL-C, the Begg test was borderline (z = − 1.96, P = 0.050), whereas the Egger test did not reach statistical significance (t = − 2.07, P = 0.130), indicating that the possibility of publication bias for this outcome requires further verification. No evidence of significant bias was detected for the remaining endpoints—SBP, DBP, TC, TG, and HDL-C (all P > 0.05) (Table 4). Table 4The results of Begg's and Egger's tests for publication bias included in the studyBegg's TestEgger's test z p t p tHcy-2.180.029-7.13<0.001folic acid levels2.530.0112.350.027SBP0.490.621-1.350.227DBP-1.240.216-1.470.192TC-0.740.458-1.290.245TG-0.150.881-1.020.356HDL-C1.350.1762.250.074LDL-C-1.960.050-2.070.130 Table 4The results of Begg's and Egger's tests for publication bias included in the studyBegg's TestEgger's test z p t p tHcy-2.180.029-7.13<0.001folic acid levels2.530.0112.350.027SBP0.490.621-1.350.227DBP-1.240.216-1.470.192TC-0.740.458-1.290.245TG-0.150.881-1.020.356HDL-C1.350.1762.250.074LDL-C-1.960.050-2.070.130 The results of Begg's and Egger's tests for publication bias included in the study GRADE certainty of evidence assessment We assessed the certainty of evidence using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) framework. The full Summary of Findings table is presented in Table 5. The certainty of evidence was rated as low for primary outcomes (tHcy and folate levels) due to serious concerns regarding risk of bias (unclear allocation concealment in 32/36 studies) and inconsistency (I² > 90%). For secondary surrogate biomarkers, certainty was rated as very low for SBP, DBP, TC, and TG, and low for HDL-C and LDL-C, reflecting additional concerns about indirectness (surrogate outcomes) and publication bias. The certainty of evidence for cardiovascular safety outcomes and clinical hard endpoints was rated as very low. Table 5Grade for each outcomeOutcome (measurement: change from baseline to final follow-up)No. of RCTs (Total participants, Intervention/Control)Pooled effect estimate (95% CI)Certainty of evidence (GRADE)Key reasons for downgradingPrimary outcomesPlasma total homocysteine (tHcy) level36 RCTs (21727 total; 10851/10876)SMD −0.99 (−1.15 to −0.83); P<0.00001Low① Serious risk of bias (32/36 studies had unclear allocation concealment; 18/36 studies had high risk of detection bias); ② Serious inconsistency (I²=91%, extreme heterogeneity); ③ Serious publication bias (Egger’s test P<0.001, Begg’s test P=0.029)Plasma folate concentration23 RCTs (19946 total; 9962/9984)SMD 1.61 (1.30 to 1.93); P<0.00001Low① Serious risk of bias (majority of studies had unclear allocation concealment and outcome assessment blinding); ② Serious inconsistency (I²=96%, extreme heterogeneity); ③ Serious publication bias (Egger’s test P=0.027, Begg’s test P=0.011)Secondary outcomes: cardiovascular risk-related surrogate biomarkersSystolic blood pressure (SBP)6 RCTs (869 total; 383/486)SMD −0.54 (−0.89 to −0.18); P=0.003Very low① Serious risk of bias; ② Serious inconsistency (I²=80%, high heterogeneity); ③ Serious imprecision (small sample size, limited number of studies)Diastolic blood pressure (DBP)6 RCTs (869 total; 383/486)SMD −0.32 (−0.58 to −0.05); P=0.02Very low① Serious risk of bias; ② Serious inconsistency (I²=63%, moderate-high heterogeneity); ③ Serious imprecision (small sample size, 95% CI close to the null effect line)High-density lipoprotein cholesterol (HDL-C)6 RCTs (472 total; 177/295)WMD 1.37 mg/dL (0.71 to 2.03); P<0.0001Low① Serious risk of bias; ② Serious imprecision (small total sample size)Low-density lipoprotein cholesterol (LDL-C)4 RCTs (400 total; 134/266)WMD −3.64 mg/dL (−5.91 to −1.36); P=0.002Low① Serious risk of bias; ② Serious imprecision (very limited number of studies and small sample size)Total cholesterol (TC)7 RCTs (506 total; 194/312)SMD −0.15 (−0.72 to 0.42); P=0.61Very low① Serious risk of bias; ② Serious inconsistency (I²=88%, extreme heterogeneity); ③ Serious imprecision (small sample size, 95% CI crosses the null line, no statistically significant effect)Triglyceride (TG)5 RCTs (472 total; 177/295)SMD −0.06 (−0.39 to 0.27); P=0.72Very low① Serious risk of bias; ② Serious inconsistency (I²=61%, moderate-high heterogeneity); ③ Serious imprecision (small sample size, 95% CI crosses the null line, no statistically significant effect)SMD ssscscscscsssccsscscsSMD standardized mean difference, WMD weighted mean difference, CI confidence interval, RCT randomized controlled trial, I2 heterogeneity statistic Table 5Grade for each outcomeOutcome (measurement: change from baseline to final follow-up)No. of RCTs (Total participants, Intervention/Control)Pooled effect estimate (95% CI)Certainty of evidence (GRADE)Key reasons for downgradingPrimary outcomesPlasma total homocysteine (tHcy) level36 RCTs (21727 total; 10851/10876)SMD −0.99 (−1.15 to −0.83); P<0.00001Low① Serious risk of bias (32/36 studies had unclear allocation concealment; 18/36 studies had high risk of detection bias); ② Serious inconsistency (I²=91%, extreme heterogeneity); ③ Serious publication bias (Egger’s test P<0.001, Begg’s test P=0.029)Plasma folate concentration23 RCTs (19946 total; 9962/9984)SMD 1.61 (1.30 to 1.93); P<0.00001Low① Serious risk of bias (majority of studies had unclear allocation concealment and outcome assessment blinding); ② Serious inconsistency (I²=96%, extreme heterogeneity); ③ Serious publication bias (Egger’s test P=0.027, Begg’s test P=0.011)Secondary outcomes: cardiovascular risk-related surrogate biomarkersSystolic blood pressure (SBP)6 RCTs (869 total; 383/486)SMD −0.54 (−0.89 to −0.18); P=0.003Very low① Serious risk of bias; ② Serious inconsistency (I²=80%, high heterogeneity); ③ Serious imprecision (small sample size, limited number of studies)Diastolic blood pressure (DBP)6 RCTs (869 total; 383/486)SMD −0.32 (−0.58 to −0.05); P=0.02Very low① Serious risk of bias; ② Serious inconsistency (I²=63%, moderate-high heterogeneity); ③ Serious imprecision (small sample size, 95% CI close to the null effect line)High-density lipoprotein cholesterol (HDL-C)6 RCTs (472 total; 177/295)WMD 1.37 mg/dL (0.71 to 2.03); P<0.0001Low① Serious risk of bias; ② Serious imprecision (small total sample size)Low-density lipoprotein cholesterol (LDL-C)4 RCTs (400 total; 134/266)WMD −3.64 mg/dL (−5.91 to −1.36); P=0.002Low① Serious risk of bias; ② Serious imprecision (very limited number of studies and small sample size)Total cholesterol (TC)7 RCTs (506 total; 194/312)SMD −0.15 (−0.72 to 0.42); P=0.61Very low① Serious risk of bias; ② Serious inconsistency (I²=88%, extreme heterogeneity); ③ Serious imprecision (small sample size, 95% CI crosses the null line, no statistically significant effect)Triglyceride (TG)5 RCTs (472 total; 177/295)SMD −0.06 (−0.39 to 0.27); P=0.72Very low① Serious risk of bias; ② Serious inconsistency (I²=61%, moderate-high heterogeneity); ③ Serious imprecision (small sample size, 95% CI crosses the null line, no statistically significant effect)SMD ssscscscscsssccsscscsSMD standardized mean difference, WMD weighted mean difference, CI confidence interval, RCT randomized controlled trial, I2 heterogeneity statistic Grade for each outcome SMD standardized mean difference, WMD weighted mean difference, CI confidence interval, RCT randomized controlled trial, I2 heterogeneity statistic Discussion This meta-analysis incorporated 36 randomized controlled trials assessing the impact of folic acid on Hhcy.Critical methodological limitations must be highlighted upfront: 32 of the 36 included RCTs had unclear allocation concealment, 18 of 36 RCTs were at high risk of detection bias due to inadequate blinding of outcome assessment, and significant publication bias with small-study effects was detected for the primary outcomes, all of which may overestimate the magnitude of the observed treatment effects of folic acid supplementation. The findings showed significant reductions in Hcy levels, along with statistically significant changes in the cardiovascular surrogate biomarkers SBP (SMD = − 0.54), DBP (SMD = − 0.32), HDL-C (WMD = + 1.37) and LDL-C (WMD = − 3.64). Given the presence of publication bias and small-study effects for tHcy and folate, the true biomarker effect is likely smaller than our pooled estimate. These primary results, while statistically significant, are derived from evidence with high heterogeneity (I²: 61–96%) and substantial risk of bias (32/36 studies with unclear allocation concealment; 18/36 at high risk of detection bias), with likely overestimation of effect sizes due to small-study effects and publication bias. The safety profile of folic acid for Hhcy management showed a clear dose-dependent characteristic: no serious cardiovascular adverse reactions were documented in trials with folic acid doses ≤ 10 mg/day, while a significant increase in cardiovascular adverse events was observed exclusively in the subgroup receiving a 15 mg/day dose. This indicates that folic acid has acceptable tolerability and safety within a specific low-to-moderate dosage range, and deeper analyses based on this research could offer valuable guidance for the design of future clinical trials. Significant publication bias and small-study effects were detected for our two primary outcomes (plasma tHcy and folate concentrations), with both Egger’s and Begg’s tests reaching statistical significance. Small-study effects typically arise when smaller trials with positive, statistically significant results are more likely to be published than those with null or negative findings. For this meta-analysis, this means the observed magnitude of tHcy-lowering efficacy and folate concentration elevation associated with folic acid supplementation is likely inflated: the true treatment effect in the broader population may be smaller than the pooled effect size we reported. In this study, folic acid supplementation significantly reduced tHcy levels and increased serum folate concentrations in adults with Hhcy, consistent with prior meta-analyses. Our findings on folic acid monotherapy and surrogate biomarker changes do not contradict these large RCTs that failed to show clinical event reduction. Instead, they highlight the persistent gap between biochemical improvements and meaningful clinical outcomes, underscoring the limitations of relying on surrogate endpoints in this field. As a core cofactor in one-carbon metabolism, folic acid facilitates the remethylation of Hcy to methionine, which is the primary biochemical mechanism underlying its tHcy-lowering effect [59]. Critically, no included studies reported adjudicated cardiovascular hard endpoints (myocardial infarction, stroke, cardiovascular mortality). Therefore, no conclusions can be drawn regarding a reduction in clinical cardiovascular events, and target tHcy reduction remains an unvalidated surrogate for clinical CVD risk reduction. Improvements in cardiovascular surrogate biomarkers cannot be directly equated to reduced clinical event risk, requiring caution when interpreting these findings. Given the low to very low certainty of evidence across most outcomes, clinicians should exercise extreme caution when considering folic acid supplementation for Hhcy management. Further analysis revealed that daily supplementation with 15 mg of folic acid had the most pronounced effect in reducing Hcy levels, followed by daily doses of 5–10 mg. This suggests that the Hcy-lowering effect of folic acid is dose-dependent. However, this dose-dependent effect on tHcy is accompanied by a dose-dependent increase in safety risks: while doses ≤ 10 mg/day achieved significant Hcy-lowering effects with no observed cardiovascular adverse events, the 15 mg/day dose, despite the greatest Hcy reduction, was associated with a significantly increased risk of cardiovascular adverse events in the included trials [60]. Some studies have indicated that high doses of folic acid may mask symptoms of vitamin B12 deficiency, leading to an increased risk of neurological damage [61]. Additionally, excessive folic acid intake may be associated with a higher risk of certain cancers [62]. Our results also showed that a supplementation duration exceeding six months yielded the most significant effects, possibly due to the time required for folic acid to accumulate and exert its effects in the body [63]. Therefore, the dose-dependent Hcy-lowering effect and dose-dependent safety risk identified in this analysis are hypothesis-generating and should inform the design of future prospective trials, rather than being interpreted as guidance for clinical practice [60]. No formal dose-response meta-regression was performed to quantify the dose-efficacy-safety relationship, and no included trials explicitly tested individualized dosing algorithms; thus, no definitive clinical recommendations for dose selection can be derived from this analysis. Based on the observed dose-response patterns in these exploratory subgroup analyses, which are strictly hypothesis-generating, future RCTs could consider investigating a range of doses (e.g., 0.8–1.2 mg/day) in non-fortified populations [64]. This analysis does not, however, provide a basis for clinical dosing recommendations. The cardiovascular benefits of folic acid beyond tHcy reduction are biologically plausible: its active metabolite tetrahydrofolate (THF) acts as a methyl donor, working synergistically with vitamin B12 to upregulate methionine synthase activity [65]. Folic acid may also modulate vascular health via regulation of DNA methylation and endothelial gene expression, which may explain the observed modest improvements in blood pressure and lipid biomarkers in this analysis [66]. Secondary indicators such as SBP, DBP, HDL, LDL, TG, and TC can serve as functional status indicators in the clinical management of CVD. In this review, analyses showed statistically significant differences between the folic acid group and the control group in SBP, DBP, HDL, and LDL, while differences in TC and TG were not statistically significant. These findings indicate that folic acid intervention is associated with favorable changes in these specific surrogate biomarkers. This may be because folic acid enhances the bioavailability of nitric oxide (NO), improving endothelial function and thereby lowering blood pressure [67]. In terms of lipid metabolism, folic acid may influence the distribution of lipoproteins, reduce the oxidation of LDL-C, and enhance the anti-atherosclerotic effects of HDL-C [68]. However, its impact on TG and TC may require a longer duration or higher doses to become evident [69]., or it may involve other metabolic pathways that warrant further research. This meta-analysis showed that folate supplementation in a specific population moderately reduced homocysteine levels, accompanied by slight improvements in blood pressure and lipids, but the effect size heterogeneity was high, suggesting that the efficacy may be affected by differences in population characteristics and intervention programs. Based on the results of the subgroup analysis, non-kidney transplant recipients and populations in areas without a folic acid fortification policy may be more likely to benefit from supplementation, while patients with kidney disease and metabolic abnormalities have more variable responses, which should be carefully evaluated individually [70–72]. Due to the wide range of folic acid doses in the included studies and the lack of direct comparison of dose-response relationship, the current evidence is insufficient to establish a uniform optimal dose; Despite the observed improvements in blood pressure and lipid biomarkers, the complete lack of data on adjudicated cardiovascular hard endpoints (such as myocardial infarction, stroke, and cardiovascular mortality) means that the clinical significance of these biomarker changes for CVD prevention remains unproven and must be verified by long-term, large-scale clinical trials; Additionally, performance bias may exist: due to the intervention nature, neither participants nor personnel were blinded, which may overestimate efficacy. High heterogeneity was a key feature of this meta-analysis, arising from three main sources: renal patient cohorts had the largest tHcy reduction (SMD=-1.60) but extreme variability (I²=95.1%), likely due to insufficient estimated glomerular filtration rate (eGFR) stratification, variable dialysis status, and vitamin B12 deficiency; populations with high MTHFR C677T polymorphism prevalence (e.g., China, Italy) and non-fortified regions (e.g., Greece) showed marked response heterogeneity (I²>90%), whereas fortified regions (e.g., Germany, Iceland) had homogeneous outcomes (I²=0%); non-placebo trials exhibited higher heterogeneity (I²=90.6%) than placebo-controlled studies (I²=68.5%), potentially confounded by co-interventions such as vitamin B12. These subgroup and meta-regression findings are hypothesis-generating rather than prescriptive for clinical practice, given the observational nature of post-hoc subgroup analyses, the extreme between-study heterogeneity, and the potential for residual confounding. These results may inform the design of future prospective, genotype-stratified and eGFR-based randomized controlled trials, rather than providing practice-ready clinical recommendations. Importantly, high heterogeneity precludes a universal optimal dosage, emphasizing that clinical practice should focus on target tHcy levels rather than fixed regimens. In non-fortified regions with high MTHFR C677T prevalence, biomarker responses were more pronounced but highly heterogeneous; in fortified regions, effects were smaller and more homogeneous, suggesting limited incremental benefit where dietary folate intake is already adequate. Notably, the generalizability of our findings is constrained by the geographic distribution of included studies, with most conducted in regions without mandatory dietary folate fortification (e.g., China, Italy, Greece, Iran) and a minority in fortified regions (e.g., Germany, Iceland, United States, Australia). Although formal subgroup analysis by fortification status was not performed, contextual evidence from geographic variations in heterogeneity (extreme variability in non-fortified regions vs. homogeneity in fortified regions) suggests that baseline folate status impacts treatment response. In populations with mandatory fortification, higher baseline folate levels are expected to attenuate the tHcy-lowering effect of supplemental folic acid, as dietary fortification already addresses folate insufficiency—the primary driver of Hhcy in many regions. Clinically, this implies that in fortified settings, folic acid supplementation may offer limited incremental benefit in modifying cardiovascular surrogate biomarkers. In contrast, non-fortified populations—especially those with high prevalence of the MTHFR C677T polymorphism—showed larger effect estimates in exploratory analyses of targeted supplementation, as their baseline folate deficit amplifies the therapeutic response to exogenous folic acid. It is important to clarify that no formal subgroup analysis by fortification status was performed; these inferences are ecological and should be interpreted cautiously. The lack of formal subgroup analysis by fortification status is a limitation, and future research should prioritize dedicated trials in fortified populations to define optimal dosing strategies and validate residual cardiovascular risk reduction potential, ensuring clinical recommendations are tailored to regional nutrition policies. Given the low to very low certainty of evidence across most outcomes, clinicians should exercise extreme caution when considering folic acid supplementation for Hhcy management. Critically, target tHcy reduction remains unvalidated as a surrogate for clinical CVD event reduction in this population, and no optimal folic acid dose can be definitively defined from this meta-analysis.The observed benefits are limited to surrogate biomarkers with low certainty evidence, with no validated clinical cardiovascular benefit. The dose range of 0.8–1.2 mg/d may be considered for populations without mandatory folate fortification, with regular monitoring of tHcy levels and adverse reactions. For populations with high heterogeneity in treatment response (e.g., kidney disease patients), the potential benefits and risks of supplementation should be weighed on an individual basis, and the value of genetic testing (e.g., MTHFR C677T) requires validation in prospective trials before clinical application. Future large-scale RCTs in the general population are urgently needed to clarify the dose threshold for biomarker modification and, most importantly, to validate whether these biomarker changes translate to clinical cardiovascular hard endpoint benefits. Strengths and limitations This meta-analysis provides one of the most comprehensive syntheses to date on folic acid’s efficacy for Hhcy management and modification of associated CVD-related surrogate biomarkers. By systematically evaluating 36 RCTs across global populations, we identified consistent tHcy-lowering efficacy and modest blood pressure improvements with low adverse event rates overall. Subgroup analyses identified populations with more homogeneous responses (e.g., non-renal transplant recipients), suggesting groups where future research could be focused. These exploratory findings are hypothesis-generating only and should not be interpreted as providing practice-ready clinical recommendations.It is important to distinguish between two categories of findings in this meta-analysis. (a) Robust descriptive findings: the associations between folic acid supplementation and changes in tHcy, folate, and selected cardiovascular surrogate biomarkers (SBP, DBP, HDL-C, LDL-C), based on pooled analyses with transparent handling of heterogeneity. These findings are supported by low certainty evidence. (b) Exploratory, hypothesis-generating findings: all dose-related observations (including subgroup analyses by dose range and treatment duration) and subgroup analyses by geographic region, population type, and study design. These findings are presented solely to inform the design of future trials and should not be interpreted as definitive guidance for clinical practice. Despite rigorous methodology, inherent constraints persist: (1) The retrospective synthesis of heterogeneous RCTs limits causal inference regarding folic acid’s direct impact on clinical outcomes; (2) High heterogeneity (I² ranging from 61% to 96%)—particularly in renal cohorts (I²=95.1%) and genetically diverse populations—precludes unified dosing recommendations and restricts generalizability to multi-morbid groups; post-hoc subgroup and meta-regression analyses are observational in nature, with potential residual confounding, and the findings are only hypothesis-generating rather than clinically actionable; no formal dose-response meta-regression was conducted to quantify the dose-efficacy-safety relationship, and no included RCTs explicitly tested the individualized dosing algorithms referenced in the discussion, meaning no definitive clinical dosing recommendations or optimal dose can be derived from this analysis; (3) The complete absence of adjudicated hard CVD endpoints (e.g., myocardial infarction, stroke, cardiovascular mortality) across all included studies necessitates extreme caution when interpreting surrogate biomarker improvements as clinical benefit; target tHcy reduction remains unvalidated as a surrogate endpoint for clinical CVD event reduction in this population; (4) Critical methodological bias from inadequate trial design: 32 of 36 studies had unclear allocation concealment, and 18 studies carried high risk of detection bias due to inadequate blinding of outcome assessment, which may overestimate the magnitude of biomarker changes; formal GRADE assessment rated the certainty of evidence as low to very low for all outcomes, further limiting the strength of clinical recommendations that can be derived from this meta-analysis; (5) Significant publication bias and small-study effects: Both Egger’s and Begg’s tests confirmed significant publication bias for the two primary outcomes (tHcy and folate levels), which is likely driven by small-study effects. This bias may inflate the apparent magnitude of the tHcy-lowering and folate-elevating benefits of folic acid supplementation, meaning the true treatment effect may be smaller than the pooled effect size reported in this meta-analysis; (6) Most included studies were conducted in regions without mandatory folate fortification, limiting the external validity of findings to populations with fortified diets. While sensitivity analyses confirmed robustness for primary endpoints (p < 0.01), residual confounding from unmeasured variables (e.g., MTHFR C677T, MTR A2756G, and MTRR A66G polymorphisms) may bias complication-risk estimates. Future large-scale trials should validate our findings using genotype-stratified designs, eGFR-based stratification for renal patients, standardized tHcy monitoring protocols, and adjudicated cardiovascular hard endpoints-with dedicated studies in fortified regions to address current generalizability gaps. Conclusions In summary, the results of this meta-analysis provide two categories of findings: (a) Robust descriptive biomarker findings (low certainty evidence): Folic acid supplementation is associated with significant reductions in tHcy levels, a marked increase in folate levels, and statistically significant changes in blood pressure(SBP, DBP) and lipid (HDL-C, LDL-C) cardiovascular surrogate biomarkers. (b) Speculative, hypothesis-generating findings (for future trial design only): Dose-related observations and subgroup analyses suggesting differential responses based on population characteristics (e.g., geographic region, patient population, folate fortification status) are strictly exploratory. No optimal folic acid dose can be definitively defined from this analysis, and any dose-related considerations require validation in future dedicated RCTs. The safety profile showed a clear dose-dependent pattern: no cardiovascular adverse events were observed at doses ≤ 10 mg/day, while increased events occurred exclusively in trials using 15 mg/day. Critically, no included studies reported adjudicated cardiovascular hard endpoints; target tHcy reduction remains an unvalidated surrogate for clinical CVD risk reduction. Current evidence does not support the use of folic acid to reduce clinical cardiovascular events. A paramount limitation of this meta-analysis, which demands explicit emphasis, is the complete lack of adjudicated cardiovascular events (e.g., myocardial infarction, stroke, cardiovascular mortality) across all 36 included studies. All observed benefits are restricted to surrogate biomarkers, and no direct evidence supports that folic acid supplementation reduces the incidence of hard cardiovascular endpoints. This critical gap prevents any definitive conclusion about the clinical utility of folic acid for reducing clinical cardiovascular disease risk, as surrogate marker improvements do not reliably translate to reduced clinical events. Furthermore, due to high heterogeneity among studies and potential issues related to literature quality, no formal dose-response meta-regression was performed in this meta-analysis, and all subgroup and dose-related findings remain hypothesis-generating and exploratory, and cannot be used to guide clinical practice directly. In particular, the safety of folic acid supplementation is strictly dose-limited, and high-dose folic acid (≥ 15 mg/day) is not recommended for clinical Hhcy management due to the observed increased risk of cardiovascular adverse events, as well as the potential risks of vitamin B12 deficiency masking and cancer risk elevation reported in existing literature. Future research must prioritize: (1) large-scale, high-quality randomized controlled trials designed with adjudicated cardiovascular hard endpoints to validate whether the biochemical and surrogate marker benefits of folic acid supplementation translate to meaningful clinical outcomes; (2) robust dose-response meta-regression to define the dose-efficacy-safety relationship of folic acid for Hhcy management; (3) prospective trials to validate the clinical value of subgroup-specific dosing strategies; and (4) studies to confirm whether tHcy reduction can serve as a valid surrogate for CVD event reduction in this population. Until such evidence is available, interpretations of the current results for clinical decision-making regarding clinical cardiovascular risk reduction must be extremely cautious to avoid over-extrapolation beyond the observed surrogate biomarker changes, especially given the low to very low certainty of evidence across most outcomes, and the absence of definitive dosing recommendations or optimal dose definition from this analysis. Supplementary Information Additional file 1: Detailed search strategy. Additional file 1: Detailed search strategy. Additional file 2: PRISMA 2020 Checklist. Additional file 2: PRISMA 2020 Checklist.