Association of Vascular Endothelial Growth Factors (VEGFs) with Recurrent Miscarriage: A Systematic Review of the Literature

Recurrent miscarriage (RM) can be defined as two or more consecutive miscarriages before 20 weeks’ gestation. Vascular endothelial growth factors (VEGFs) play an important role in endometrial angiogenesis and decidualization, prerequisites for successful pregnancy outcomes. We conducted a systematic review of the published literature investigating the role of VEGFs in RM. In particular, we explored the methodological inconsistencies between the published reports on this topic. To our knowledge, this is the first systematic literature review to examine the role of VEGFs in RM. Our systematic search followed PRISMA guidelines. Three databases, Medline (Ovid), PubMed, and Embase, were searched. Assessment-bias analyses were conducted using the Joanna Bigger Institute critical appraisal method for case-control studies. Thirteen papers were included in the final analyses. These studies included 677 cases with RM and 724 controls. Endometrial levels of VEGFs were consistently lower in RM cases compared to controls. There were no consistent significant findings with respect to VEGFs levels in decidua, fetoplacental tissues, and serum when RM cases were compared to controls. The interpretation of studies that explored the relationship between VEGFs and RM is hampered by inconsistencies in defining clinical, sampling, and analytical variables. To clarify the association between VEGF and RM in future studies, researchers ideally should use similarly defined clinical groups, similar samples collected in the same way, and laboratory analyses undertaken using the same methods.


Introduction
Recurrent miscarriage (RM) is defined as two or more consecutive miscarriages before 20 weeks' gestation recognized by ultrasound or histopathology [1]. It affects approximately 1-5% of women who conceive. Multiple factors contribute to miscarriages, including fetoplacental chromosomal abnormalities, uterine abnormalities, endocrine disorders, acquired thrombophilia, parental balanced translocations, and immune disorders. However, the underlying causes remain undetermined in 50-70% of patients who experience RM [2].
RM with unknown etiology is defined as unexplained RM or idiopathic RM [1]. Recent advances in cytogenetics and immunogenetics have expanded our knowledge of implantation and maternal-fetal interactions. The placenta at the maternal-fetal interface provides the fetus with the metabolic requirements necessary for development through the exchange of nutrients and wastes [3]. To achieve this, the placenta maintains its own circulation and metabolism via angiogenesis, the formation and remodeling of blood vessels in metabolism via angiogenesis, the formation and remodeling of blood vessels in a vascular network [4]. The three steps of angiogenesis-initiation, proliferation-invasion, and maturation-differentiation-are all critical for normal placental development and successful implantation [5].
There are several well-defined families of angiogenic growth factors that play roles in the development of mature functioning blood vessels. The best characterized is the Vascular Endothelial Growth Factor (VEGF) family ( Figure 1). VEGFs are glycoproteins with angiogenic properties [6,7] such as induction of vascular permeability [6,8,9], stimulation of endothelial cell division and migration [9,10] and in vivo angiogenesis [6]. A good endometrial blood supply is usually considered to be a marker for endometrial receptivity [11]. This suggests that placental angiodysplasia and associated vascular endothelial dysfunction may be an important etiology of unexplained RM [12,13]. VEGF-A is a gene found on chromosome 6p12-p21.1 that codes for six different isoforms of VEGF-A, including the predominant isoform VEGF-A165, which binds to the receptors FLT-1, KDR, NRP-1, and NRP-2 and promotes angiogenesis, endothelial cell growth, and vascular permeability. PlGF, which has 42% amino acid sequence identity with VEGF-A, is predominantly expressed in the placenta, heart, and lungs, and plays a role in regulating VEGF-dependent angiogenesis under pathological conditions. VEGF-B, which forms stable heterodimers with VEGF-A, binds to the receptors FLT-1 and NRP-1 and behaves as an endothelial cell mitogen. VEGF-C regulates the lymphatic system during embryogenesis and adult life, while VEGF-D plays a role in lymphangiogenesis and can also activate the receptors FLT-4 and KDR. FLT-1 and KDR are receptor tyrosine kinases expressed on vascular endothelial cells that bind to VEGF-A, VEGF-B, and PlGF, while NRP-1 and NRP-2 are co-receptors that enhance the binding of VEGF-A and PlGF to their respective receptors.
The direct relationship between VEGFs and RM remains debatable with inconsistent results across the studies. In view of the established crucial role of endometrial factors in facilitating a successful human pregnancy (through angiogenesis and vascular tone reactivity), we conducted a systematic review of the literature to evaluate the role of VEGFs in RM. We also explored the clinical, sampling, and analytical variables in this published literature. To our knowledge, this is the first systematic review of the literature to explore the role of VEGFs in RM. VEGF-A is a gene found on chromosome 6p12-p21.1 that codes for six different isoforms of VEGF-A, including the predominant isoform VEGF-A165, which binds to the receptors FLT-1, KDR, NRP-1, and NRP-2 and promotes angiogenesis, endothelial cell growth, and vascular permeability. PlGF, which has 42% amino acid sequence identity with VEGF-A, is predominantly expressed in the placenta, heart, and lungs, and plays a role in regulating VEGF-dependent angiogenesis under pathological conditions. VEGF-B, which forms stable heterodimers with VEGF-A, binds to the receptors FLT-1 and NRP-1 and behaves as an endothelial cell mitogen. VEGF-C regulates the lymphatic system during embryogenesis and adult life, while VEGF-D plays a role in lymphangiogenesis and can also activate the receptors FLT-4 and KDR. FLT-1 and KDR are receptor tyrosine kinases expressed on vascular endothelial cells that bind to VEGF-A, VEGF-B, and PlGF, while NRP-1 and NRP-2 are co-receptors that enhance the binding of VEGF-A and PlGF to their respective receptors.
The direct relationship between VEGFs and RM remains debatable with inconsistent results across the studies. In view of the established crucial role of endometrial factors in facilitating a successful human pregnancy (through angiogenesis and vascular tone reactivity), we conducted a systematic review of the literature to evaluate the role of VEGFs in RM. We also explored the clinical, sampling, and analytical variables in this published literature. To our knowledge, this is the first systematic review of the literature to explore the role of VEGFs in RM.

Literature Search and Data Extraction
A systematic literature search following PRIMSA guidelines was devised ( Figure 2). Three databases Medline (Ovid), PubMed, and Embase were searched for relevant studies published between the years 2011 and 2022 (Supplementary Material, Section S1). Studies were included in the systematic review if they were published in English and complied with the inclusion criteria defined below. The results from the initial search were combined and duplicates were removed using the Endnote X8 and Covidence referencing systems. Citation searching was utilized to augment the initial results. A sample of the search strategy and its result is provided in Supplementary Material, Section S1.

Literature Search and Data Extraction
A systematic literature search following PRIMSA guidelines was devised ( Figure 2). Three databases Medline (Ovid), PubMed, and Embase were searched for relevant studies published between the years 2011 and 2022 (Supplementary Material, Appendix S1). Studies were included in the systematic review if they were published in English and complied with the inclusion criteria defined below. The results from the initial search were combined and duplicates were removed using the Endnote X8 and Covidence referencing systems. Citation searching was utilized to augment the initial results. A sample of the search strategy and its result is provided in Supplementary Material, Appendix S1. Data from included studies were extracted, stored, and analyzed using spreadsheets. Population characteristics and clinicopathological details of the cases and controls were extracted and summarized on excel spreadsheets. Baseline data extracted included country, tissue type utilized, laboratory methodology, statistical analysis, definition and number of cases and controls, and gestational age ( Table 1). The aims and results of each study were extracted and are summarized in Table 2. Each study was reviewed by a primary reviewer (NA) and was independently checked for accuracy against the original publication by a second reviewer (VK).
Assessment of bias analyses was conducted using the critical appraisal checklist for studies reporting prevalence data of Joanna Bigger Institute (JBI) (Supplementary Material, Appendix S3) The Student's independent samples t-test was conducted using Jamovi Version 2.3.21.0, Sydney, Australia to test the null hypothesis that the serum VEGF level (pg/L) is the same between cases of RM and controls [15].
Serum VEGF data were exported into Review Manager 5.4.1 for quantitative data analysis. The meta-analysis command with random effects to account for heterogeneity was used to estimate the mean difference between serum VEGF in RM cases compared to controls. A quantification of heterogeneity across studies was presented as an I2 score [16].
Almawi et al. was excluded in the meta-analysis and the Student's independent samples t-test as median values were reported and no raw data were available to calculate the mean [17]. Data from included studies were extracted, stored, and analyzed using spreadsheets. Population characteristics and clinicopathological details of the cases and controls were extracted and summarized on excel spreadsheets. Baseline data extracted included country, tissue type utilized, laboratory methodology, statistical analysis, definition and number of cases and controls, and gestational age ( Table 1). The aims and results of each study were extracted and are summarized in Table 2. Each study was reviewed by a primary reviewer (NA) and was independently checked for accuracy against the original publication by a second reviewer (VK).
Assessment of bias analyses was conducted using the critical appraisal checklist for studies reporting prevalence data of Joanna Bigger Institute (JBI) (Supplementary Material, Section S3) The Student's independent samples t-test was conducted using Jamovi Version 2.3.21.0, Sydney, Australia to test the null hypothesis that the serum VEGF level (pg/L) is the same between cases of RM and controls [15].
Serum VEGF data were exported into Review Manager 5.4.1 for quantitative data analysis. The meta-analysis command with random effects to account for heterogeneity was used to estimate the mean difference between serum VEGF in RM cases compared to controls. A quantification of heterogeneity across studies was presented as an I2 score [16].
Almawi et al. was excluded in the meta-analysis and the Student's independent samples t-test as median values were reported and no raw data were available to calculate the mean [17].   Investigate the temporal and spatial expression of series of angiogenic growth factors (AGFs) and their receptors: vascular endothelial growth factor (VEGF)-A, VEGF-C, VEGF-D, VEGF-R1, VEGF-R2, VEGF-R3, platelet-derived growth factor (PDGF)-BB, PDGF-Ra, PDGF-Rb, transforming growth factor (TGF)-b1, TGF-bRI, TGF-bRII, angiopoietin (Ang)-1, Ang-2 and Tie-2, in the proliferative, early secretory and mid-late secretory phase endometrium from control women as well as in the mid-late secretory phase of women with a history of RM.
Four cell types were investigated, namely, glandular epithelium, stromal cells, vascular smooth muscle cells (VSMCs), epithelial cells (ECs).  Investigate the levels of sFlt-1 and VEGF in serum and chorionic villus of RM patients compared to control.
• The mean seum sFlt-1 level was higher in normal pregnancy group compared to the non-pregnant group.

•
The mean VEGF level was higher in the normal pregnancy compared to the non-pregnant group.

•
The mean sFlt1 concentration was higher in RM cases compared to women with early normal pregnancy.

•
The VEGF concentration was significantly higher in RM cases compared to the normal pregnancy group.

•
The sFlt-1/sFlt-1/VEGF ratio in serum was significantly increased in RM cases compared with normal pregnancy women. • VEGF mRNA transcription was decreased in endometrium of women with RM compared to controls.

•
In RM, VEGF mRNA expression in the LPD endometrium was lower than in the endometrium with normal maturity.

Scarpellini et al., 2019 [27]
Investigate the effects of G-CSF treatment on the maternal fetal interface using immunohistochemistry to assess the expression of G-CSF and its receptor, the VEGF and its receptor VEGFR-1, and Foxp3 in the trophoblast and decidua of first trimester miscarriages of RPL women treated with G-CSF that miscarried again despite the treatment, in no treated RPL and in normal first trimester pregnancies. Investigate whether the expression of VEGF, VEGFR-1, -2 or -3 or the Tie-1 or Tie-2 receptors in the placenta or decidua are altered in RM compared to control.

•
The absence of VEGFR-3 immunoreactivity in decidual vascular endothelium was noted in all study groups. • Placental villi from the BO group presented blood vessel-like structures positive for VEGF, VEGFR-1, -2, -3, Tie-1 and Tie-2 receptor.

Population
All case-control studies (regardless of sample size) that investigated the association of RM and VEGFs in the endometrium, decidua, fetoplacental tissue (including chorion and trophoblasts), and/or maternal serum levels were included. Cases were excluded if a history of recurrent idiopathic miscarriage was unclear. No limitations were placed on age, ethnicity, country of origin, family history of recurrent miscarriage, or any other patient demographics.

Laboratory Methods
All studies that used laboratory methods for identifying VEGFs' distribution, localization, concentration, or level in the endometrium, decidua, fetoplacental tissue, or maternal serum were included.
Laboratory methods across studies included the following: (1) Enzyme-Linked Immunosorbent Assay (ELISA) to measure serum levels of the VEGFs.
(2) Real-Time Quantitative Reverse Transcription PCR (qRT-PCR) to detect VEGF-gene expression. (3) Immunohistochemistry (IHC) of endometrial, decidua, and/or placental tissue to determine localization and distribution of VEGFs. (4) Western blot in conjunction with IHC and/or qRT-PCR, of endometrial, decidua, or fetoplacental tissue to determine localization, distribution, and levels of VEGFs.
The different laboratory methodologies and tissue types utilized across the studies are summarized in Figure 3. (2) Real-Time Quantitative Reverse Transcription PCR (qRT-PCR) to detect VEGF-ge expression.
(3) Immunohistochemistry (IHC) of endometrial, decidua, and/or placental tissue to d termine localization and distribution of VEGFs. (4) Western blot in conjunction with IHC and/or qRT-PCR, of endometrial, decidua, or toplacental tissue to determine localization, distribution, and levels of VEGFs.
The different laboratory methodologies and tissue types utilized across the studies a summarized in Figure 3.

Outcome
All studies with a primary outcome measure of identifying VEGFs in RM were cluded, provided the definition of RM was two or more miscarriages in the first 20 wee gestation.

Exclusion Criteria
If the definition of RM was not two or more miscarriages in the first 20 weeks' gestatio the study was excluded.
Studies that recruited cases with other types of miscarriages such as threatened, miss incomplete, or complete miscarriage were excluded if the cases were not diagnosed w RM.
Studies that used only Western blot investigation of endometrial, decidua and/or toplacental tissue were excluded.

Search Results
A total of 480 titles and abstracts were retrieved from the literature searches and additional 3 from searching reference lists and systematic reviews. Eighteen citations we fully assessed (n =18). From these, seven were excluded (n = 7) (Figure 2).

Outcome
All studies with a primary outcome measure of identifying VEGFs in RM were included, provided the definition of RM was two or more miscarriages in the first 20 weeks' gestation.

Exclusion Criteria
If the definition of RM was not two or more miscarriages in the first 20 weeks' gestation, the study was excluded.
Studies that recruited cases with other types of miscarriages such as threatened, missed, incomplete, or complete miscarriage were excluded if the cases were not diagnosed with RM.
Studies that used only Western blot investigation of endometrial, decidua and/or fetoplacental tissue were excluded.

Search Results
A total of 480 titles and abstracts were retrieved from the literature searches and an additional 3 from searching reference lists and systematic reviews. Eighteen citations were fully assessed (n =18). From these, seven were excluded (n = 7) (Figure 2). Table S1 in the Supplementary Material, Section S2 provides detailed information on why each study was excluded.
Thirteen papers were included for the final analyses (n = 13). These studies included 677 cases with RM. The studies were across 10 countries. India (n = 3) followed by China (n = 2) had the largest number of eligible studies. There were many studies published in the European continent (n = 3) and Middle Eastern countries (n = 3).
The aim and results of each study are summarized in Table 2.

Terms and Definitions
After a full-text review of the included studies, the following presents the definitions of recurrent idiopathic miscarriage, healthy controls, and VEGFs as defined by their authors.

Recurrent Idiopathic Miscarriage
Recurrent miscarriage was defined as two or more miscarriages in the first 20 weeks' gestation [1]. Unexplained or idiopathic miscarriages were defined as pregnancy loss with 'known' causes being excluded. These causes included anatomical abnormalities of the genital tract, chromosome abnormalities of partners, and hematological, endocrinological or immunological risk factors for RM.
Recurrent idiopathic miscarriages also included keywords: pregnancy loss, spontaneous abortion, and vaginal expulsion of fetus.
The different definitions for RM cases are summarized in Table 1.

Healthy Controls
Depending on the study design, definitions of healthy controls included: (1) Pregnant women with no current and/or previous pregnancy complications request a surgical termination of pregnancy for unwanted pregnancy. The different definitions of healthy controls are summarized in Table 1.

VEGFs in Endometrial Tissue
Three studies investigated the expression of VEGF in endometrial tissue in cases of RM compared to controls. Banerjee et al. (2013) [13] found that VEGF is significantly downregulated in the endometrial of women with RM. The study also reported that VEGF is significantly associated with vascular dysfunction and blood flow impairment. Lash et al. (2011) found that only VEGF-A expression was reduced in women with RM compared to mid-late secretory controls, whereas there was no difference in staining intensity for VEGF-C and VEGF-D in RM cases compared to controls in all stages of the menstrual cycle. VEGF-R1 and VEGF-R3 were increased in women with RM compared with mid-late secretory phase endometrium from controls. However, stromal cell immunoreactivity for VEGF-R2 across the menstrual cycle was reduced in women with RM compared to control women. Sadekova et al. (2015) [26] reported that VEGF mRNA transcription was decreased in the endometrium of women with RM compared to control.

VEGFs in the Maternal Decidua and Placental Tissues
Five studies investigated the expression of VEGF in the placental tissues in RM compared with controls. He et al. [22] reported that VEGF expression in either chorionic villi or decidua was significantly reduced in the RM group compared with control as revealed by immunostaining, Western blot, and VEGF mRNA [22]. However, Pang et al. reported that RM patients had a higher expression of VEGF in chorionic villi compared with normal pregnancy controls [24]. Papamitsou et al. [25] reported a significantly increased level of VEGF in the decidua and in the chorionic villi in the RM group compared with control. Scarpellini et al. demonstrated no differences in the expression of VEGF and VEGFR-1 in the decidua, a significantly reduced staining for VEGF in the trophoblasts, and a significantly increased expression of VEGFR-1 in the trophoblasts of RM compared with control [27].

VEGFs in Serum
Six studies investigated serum levels of VEGF in RM cases compared with controls. Three studies reported that serum VEGF levels were significantly reduced in RM cases compared with controls and three studies reported that serum VEGF levels were significantly higher in RM cases compared with controls. Bagheri et al. (2017) demonstrated that the concentrations of VEGF-A and VEGF-C were correlated with the clinical characteristics of cases and controls [20]. Gupta et al. reported that serum VEGF level was lower in women with underlying etiology compared to women with unexplained RPL [21]. Atalay et al. reported that serum VEGF levels did not differ with gestational age, both in RM and control groups [19].
An independent t-test showed no evidence that the mean serum VEGF level (pg/mL) in cases with recurrent miscarriage is different from the control. The mean serum VEGF levels (pg/mL) in control was 160.94 pg/mL compared to the mean serum VEGF levels in cases of 148.66 (t(8) = 0.21, p = 0.84). (Figure 4).
in the decidua and in the chorionic villi in the RM group compared with control. Sc et al. demonstrated no differences in the expression of VEGF and VEGFR-1 in the a significantly reduced staining for VEGF in the trophoblasts, and a significantly in expression of VEGFR-1 in the trophoblasts of RM compared with control [27].

VEGFs in Serum
Six studies investigated serum levels of VEGF in RM cases compared with c Three studies reported that serum VEGF levels were significantly reduced in RM ca pared with controls and three studies reported that serum VEGF levels were sign higher in RM cases compared with controls. Bagheri et al. (2017) demonstrated that centrations of VEGF-A and VEGF-C were correlated with the clinical characteristics and controls [20]. Gupta et al. reported that serum VEGF level was lower in wom underlying etiology compared to women with unexplained RPL [21]. Atalay et al. r that serum VEGF levels did not differ with gestational age, both in RM and contro [19].
An independent t-test showed no evidence that the mean serum VEGF level in cases with recurrent miscarriage is different from the control. The mean serum levels (pg/mL) in control was 160.94 pg/mL compared to the mean serum VEGF cases of 148.66 (t(8) = 0.21, p = 0.84). (Figure 4). Upon meta-analysis, the overall mean difference between serum VEGF w pg/mL. Levels were higher in controls compared to cases of RM (confidence interva −58.7-33.27%, I2:93%). This data was not statistically significant and had a high leve erogeneity ( Figure 5).  Upon meta-analysis, the overall mean difference between serum VEGF was 12.71 pg/mL. Levels were higher in controls compared to cases of RM (confidence interval (CI) 95% − 58.7-33.27%, I2:93%). This data was not statistically significant and had a high level of heterogeneity ( Figure 5).

VEGFs in Serum
Six studies investigated serum levels of VEGF in RM cases compared with controls. Three studies reported that serum VEGF levels were significantly reduced in RM cases compared with controls and three studies reported that serum VEGF levels were significantly higher in RM cases compared with controls. Bagheri et al. (2017) demonstrated that the concentrations of VEGF-A and VEGF-C were correlated with the clinical characteristics of cases and controls [20]. Gupta et al. reported that serum VEGF level was lower in women with underlying etiology compared to women with unexplained RPL [21]. Atalay et al. reported that serum VEGF levels did not differ with gestational age, both in RM and control groups [19].
An independent t-test showed no evidence that the mean serum VEGF level (pg/mL) in cases with recurrent miscarriage is different from the control. The mean serum VEGF levels (pg/mL) in control was 160.94 pg/mL compared to the mean serum VEGF levels in cases of 148.66 (t(8) = 0.21, p = 0.84). (Figure 4). Upon meta-analysis, the overall mean difference between serum VEGF was 12.71 pg/mL. Levels were higher in controls compared to cases of RM (confidence interval (CI) 95% −58.7-33.27%, I2:93%). This data was not statistically significant and had a high level of heterogeneity ( Figure 5).

Quality Assessment
Upon quality assessment, a few sources of publication bias were identified based on the JBI criteria. Most studies were of high quality and estimated to have a low risk of bias. Hence, a decision was made not to exclude any studies following the quality assessment. Specifically, 38.46% of the studies (n = 5) were assessed as having a low risk of bias on all 10 criteria according to the JBI method, and 23.8% of the studies (n = 3) did not address 1 criterion. Areas of concern across the studies that may affect the cumulative evidence included not ensuring that the RM cases and control were comparable in patients' characteristics and matched appropriately, not providing clear definitions for RM cases and controls, and not identifying confounding factors and strategies used to deal with them. Other areas included not utilizing IHC appropriately, i.e., not using more than one researcher to interpret staining intensity results and the lack of interpretation of a sufficient number of fields. Since qualitative assessment for selection bias was conducted after the papers met the inclusion criteria, no further analysis of bias was warranted (Supplementary Material, Section S3, Table S2 and Figure S1).

Discussion
The development of an embryo needs extensive and systematic blood vessels to support implantation and placentation. It has been hypothesized that this is impaired in RM patients. VEGF plays a significant role during implantation by stimulating endothelial cell proliferation and increasing vascular permeability [30][31][32][33]. Although evidence supports an association between VEGF and angiogenesis, the underlying role of VEGF in RM is unclear. Significantly lower vascular, stromal, and glandular expression patterns of endometrial VEGF were observed in RM compared to controls [13,23,26]. However, there were VEGF result inconsistencies when investigating the relationship using decidua, fetoplacental tissues, and serum.
The inconsistencies in the results are potentially attributed to dissimilar methodologies and study designs. To explore the relationship between VEGF and RM, studies should have a consistent design, similar clinical definitions and study populations, and comparable methodologies, together with sufficiently large sample sizes for statistically valid conclusions to be drawn. Tables 3-5 identify the clinical, sampling, and analytical variables that must be clearly accounted for when designing a case-control study investigating the role of placental, decidua, endometrial, and serum VEGF and RM.

Clinical Variables
When exploring the relationship between VEGF in RM cases compared to controls, it is essential that both groups are similar, barring the presence of RM in cases and its absence in controls. Despite some studies matching cases to controls based on multiple parameters, the majority neglected essential parameters including BMI, ethnicity, gestation age, menstrual phase, and parity and gravity. Multiple confounding factors including sperm quality, concomitant gynecological diseases, and medical predisposition for RM were not thoroughly explored in the included papers.
Serum, endometrial, placental, and decidual VEGF levels are dependent on multiple factors that might have confounded the results and led to different findings in RM compared to controls. For example, VEGF-gene polymorphisms, which are influenced by ethnicity, might be a factor contributing to the discrepancy in the results [17,34]. Almawi et al. reported a differential association of VEGF variants with RM based on the VEGF genotype, highlighting the contribution of ethnicity and the need for genetic association studies [17]. Bagheri et al.'s univariate analysis also demonstrated that clinical characteristics of both cases and controls were significantly associated with the concentrations of VEGF-A and VEGF-C, which are known to play a relevant role in angiogenesis regulation [20]. Moreover, supporting the influence of patient medical comorbidities on VEGF serum levels, Gupta et al. reported that VEGF levels were lowest in patients with an underlying etiology of miscarriage including hypothyroidism, uterine anatomical abnormalities, and antiphospholipid syndrome compared to patients with unexplained RM [21]. Hence, large sample-size studies, with well-documented clinical variables of cases and controls, are needed to explore the differential expression of VEGF and its relationship with RM (Table 3).

Sampling Variables
When exploring sampling variables, multiple factors could have impacted the perceived association between VEGF and RM (Table 4). A minority of the studies specified the histological layer of uterine, placental, and decidua tissue and the region of sampling. The location of samples within the placental tissue should also be clearly stated as gene expression may vary in central and peripheral placental tissue. The sample timing following the miscarriage/termination of pregnancy should also be well documented. No effect has been reported, in the literature, of this delay in tissue evacuation or "retention time" and the vascular parameters; however, it might influence interpretations of the histopathological observations.

Analytical Variables
Analyses of the serum, endometrial, placental, and decidual tissue should be similar between cases and controls (Table 5). Tissue samples should be correlated with hysteroscopic and/or ultrasonic findings, and samples should be taken from well-defined menstrual cycle phases or gestation ages. Isoforms of VEGFs sequenced, and antibody types used in ELISAs, Western blot, and/or IHC should also be well described in the study reports. The analysis of immunostaining should control for any systemic errors by the utilization of blinding, analysis of the intensity of staining, and analysis of the percentage of cells for each staining intensity.
To explore the association between VEGFs and RM, it is worth noting that multiple angiogenic and immunogenic factors influence endometrial and placental tissue. There are key angiogenesis pathways that could be contributing to this process and overall angiogenic balance of the placenta including TGF-beta (Tumor Growth Factor Beta), FKBPL (FK506binding protein like), CD44, HSP90 (Heat shock protein 90), and Notch. In the context of placenta development, CD44 and VEGF may be involved in a complex signaling network that regulates placental development. Specifically, CD44 has been shown to modulate the expression and activity of VEGFR1 and its ligands, such as VEGF, in trophoblasts. CD44 has also been shown to regulate trophoblast migration and invasion by modulating VEGF signaling. Moreover, CD44 has been suggested to function as a co-receptor for VEGFR2, another VEGF receptor, and to enhance VEGF-mediated signaling in endothelial cells. This suggests that CD44 may play a role in the formation and maintenance of the placental vasculature [35]. FK506 inhibits migration of human microvascular endothelial cells through binding to the CD44 receptor, leading to downstream effects on the actin cytoskeleton and alterations in expression of CD44 and related proteins, suggesting a co-regulatory pathway between FKBPL and CD44 [14,36]. In placenta development, HSP90 has been shown to interact with and stabilize VEGF, allowing for proper VEGF signaling and angiogenesis. HSP90 also plays a role in regulating the expression of VEGF receptors, such as VEGFR2, on endothelial cells. This interaction between HSP90 and VEGF is important for the development and maintenance of the fetal vasculature and ensuring adequate blood flow [37]. Studies have shown that VEGF induces the expression of Notch signaling components in endothelial cells, which, in turn, promotes the formation of new blood vessels in the placenta. In addition, Notch signaling has been shown to regulate the expression of VEGF receptors in endothelial cells, which is required for the proper response of endothelial cells to VEGF stimulation [38,39]. During placenta development, TGF-beta signaling regulates the differentiation and migration of trophoblast cells, which are specialized cells that form the placenta. It also regulates the remodeling of maternal blood vessels that supply nutrients and oxygen to the developing fetus [38,39]. TGF-beta works in conjunction with VEGF to regulate angiogenesis and blood vessel formation in the placenta [38,39]. The exact interactions of these cytokines, immune cells, and angiogenic factors are not well-defined in the literature. Hence, it is simplistic to attribute the cause of idiopathic RM to one angiogenic factor family, VEGFs. Future studies could attempt to explore the role of VEGF amongst other angiogenic and immunogenic factors that have established or proposed associations with RM.
Finally, this systematic review had limitations in conducting a meta-analysis. One study calculated the median serum VEGF levels as opposed to the mean and hence was excluded from the quantitative analysis [17]. Some studies investigated VEGF and others specifically investigated VEGF-A or VEGF-C and hence a quantitative comparison is not completely accurate. In terms of measuring levels of VEGF using IHC, most studies did not include raw numbers of cases and controls that fell into each category of strength and intensity of staining, which made it difficult to quantitively analyze the data. The heterogeneity was high; therefore, some other confounders such as population characteristics within each study need to be assessed to potentially explain the variability found in this meta-analysis. Ideally, individual participant data (IPD) meta-analysis is needed to obtain more accurate estimates. Despite these limitations, our results yield important conclusions and approaches to assess VEGF levels in future studies.

Conclusions
Although there is evidence supporting the role of VEGFs in implantation and placentation, the underlying VEGF role in RM is unclear. The overall interpretation of studies that explored the relationship between VEGFs and RM is hampered by inconsistencies in defining clinical characteristics of cases and controls, tissue sampling methods, and technical variations in VEGF identification and measurement. Clinical, sampling, and analytical variables must be well-defined and further multi-ethnic population-based studies of the association between VEGFs and RM are warranted. In this paper, we identified potential study variables that would impact the observed relationship between VEGF and RM, and accordingly, provide recommendations for case-control studies attempting to explore this association.