Vascular endothelial growth factor-D plasma levels and VEGFD genetic variants are independently associated with outcomes in patients with cardiovascular disease

VEGF-D, Biomarkers, Single nucleotide polymorphisms, Cardiovascular diseases

Time of primary review: 38 days

Translational perspective

Both plasma levels of VEGF-D and genetically determined VEGF-D independently predict cardiovascular (CV) outcome in patients with acute coronary syndrome (ACS) and chronic coronary syndrome (CCS). Mendelian randomization analysis implies a causal link between plasma VEGF-D levels and CV outcomes. Hence, measurements of plasma VEGF-D and/or VEGFD genetic variants may provide prognostic information in patients with acute and chronic CAD and will be useful for identifying individuals at high risk for CV disease. VEGF-D–related pathways might be interesting for development of novel treatment strategies in coronary artery disease (CAD).

1. Introduction

Cardiovascular (CV) diseases represent the major cause of mortality around the world.¹ The underlying pathological mechanisms of CV disease include inflammation, endothelial dysfunction, oxidative stress and subsequent atherosclerosis, fibrosis, dyslipidaemia, and thromboembolism.^2–5 Additionally, several studies have shown that agents promoting inflammation are associated with reduced cardiovascular risk in CV disease.^6,7 The study by Sut et al. suggests for example how dietary effects are highly associated with inflammation severity.⁸

The vascular endothelial growth factor (VEGF) family, regulators of blood and lymphatic vessel formation, consists of five proteins in humans, three of which regulate blood vessel growth [VEGF-A, VEGF-B, and PlGF (placental growth factor)], and two that mainly modulate lymphangiogenesis (VEGF-C and VEGF-D (also known as c-Fos-induced growth factor, FIGF)).⁹ There are three VEGF receptors: VEGFR1 (Flt-1, a scavenger receptor) and VEGFR2 (FLK1/KDR, promoting angiogenesis), mainly expressed in vascular endothelial cells, and VEGFR3 (Flt-4, promoting lymphopoiesis), primarily expressed in lymphatic endothelial cells. VEGF-A is able to bind VEGFR1 and VEGFR2; VEGFB and PIGF can only bind VEGFR1, whereas VEGF-C and VEGF-D bind to both VEGFR-2 and VEGFR-3. They control functions including promoting angiogenesis and lymphopoiesis, regulating inflammation, resisting oxidative stress, fibrosis, and regulating lipid metabolism and might therefore be involved in the pathophysiology of CV disease.¹⁰

The aim of this study was to investigate a potential link between circulating VEGF ligands and/or soluble receptors and CV outcome in patients with acute coronary syndrome (ACS) and chronic coronary syndrome (CCS). A panel of VEGF biomarkers, including VEGF-A, VEGF-C, VEGF-D, and PlGF, and the soluble receptors VEGFR1 (Flt-1) and VEGFR2 (KDR) was analysed in plasma samples from a sub-cohort of the PLATO trial of patients with ACS.¹¹ Secondly, a potential association between VEGF biomarker levels and CV outcomes was explored using VEGF-D-associated single nucleotide polymorphisms (SNPs) identified by genome-wide association study (GWAS) in the same patients and then applied as genetic instruments in Mendelian randomization (MR) analysis. Finally, the findings were confirmed in a cohort of patients with CCS from the STABILITY trial^12,13 and with ACS from the FRISC-II trial.^14,15

2. Methods

2.1 Study population and outcomes

2.1.1 Discovery cohort—PLATO

The global, randomized, placebo-controlled PLATelet inhibition and patient Outcomes (PLATO) trial enrolled 18 624 patients with either ST-elevation ACS or non-ST-elevation ACS (Clinical Trial Registration: http://ClinicalTrials.gov NCT00391872). The enrolled patients received optimal medical therapy including aspirin, and optional invasive strategy, and were randomized to either clopidogrel or ticagrelor treatment. The patients were recruited between October 2006 and July 2008 and were subject to blood sampling as a part of a predefined biomarker sub-study program. The overall aims and details of the biomarker sub-study¹⁶ and GWAS¹⁷ program have previously been presented.

Only a part of the PLATO biomarker sub-study was used for the biomarker analysis in the current study. For the selection of patients, both plasma samples and genetic data needed to be available, then a case enriched sampling procedure, including first a random subset followed by case enriching that also included all patients with the outcome modified primary event (i.e. CVD, spontaneous myocardial infarction (MI) or stroke) during follow-up. Samples from 2090 subjects were used to identify VEGF biomarkers associated with genetics and CV outcomes (see Figure 1). Based on the previously performed GWAS, 10 013 patients from PLATO were included in the Mendelian randomization (MR) association with outcome analysis.

Figure 1

Study cohorts and analysis flowchart.

2.1.2 Additional confirmation and Mendelian randomization cohorts—STABILITY and FRISC-II

For confirmation, we used one cohort of patients with CCS from the STabilization of Atherosclerotic plaque By Initiation of darapLadIb TherapY (STABILITY) trial, ClinicalTrials.gov ID NCT00799903^12,13 and one cohort with ACS from the Scandinavian multi-center trial FRagmin and fast revascularization during InStability in Coronary artery disease-II (FRISC-II), registration number ISRCTN82153174.¹⁸ We confirmed the association between the plasma VEGF-D level and outcomes using a case:cohort study (n = 4015) from the STABILITY trial.¹⁹ For Mendelian randomization analyses, we used the genetic sub-studies from STABILITY (n = 10 786)¹⁹ and from FRISC-II (n = 2952).^14,15

2.2 Ethical statement

All participants provided written informed consent and the studies complied with the Declaration of Helsinki. The studies were approved by all local Ethics Committees and Institutional Review Boards.

2.3 Biomarker analysis

In the PLATO sub-cohort, biomarkers of VEGF biology [bFGF, PlGF, VEGF-A, VEGF-C, VEGF-D and the soluble receptors Flt-1, KDR (VEGFR2), and Tie-2] were analysed in EDTA plasma (in-house), collected within 24 h of ACS symptoms, using V-PLEX Angiogenesis Panel 1 Human Kit (K15190D) and Human KDR kit (K151BOC) from Meso Scale Diagnostics (Rockville, Maryland), according to the manufacturer’s instructions (both assays were qualified in-house). The plasma samples were diluted 4- and 50-fold for Angiogenesis panel and KDR, respectively, prior to analysis. To control for between plate variation, quality control (QC) samples at two concentrations were run on each plate in two replicates; on all 56 plates, the between plate coefficient of variation (CV%) were ≤13% for all eight markers (acceptance criteria ≤20%). The lower limit of quantification (LLOQ) was set for each biomarker based on the performance of the calibration curves on each plate [lowest calibration standard with recovery 75–125% and CV% of ≤25%; bFGF (2.5–9.8 pg/mL), Flt-1 (9.1–36 pg/mL), PlGF (1.1–4.4 pg/mL), Tie-2 (375–1500 pg/mL), VEGF-A (2.3–9.1 pg/mL), VEGF-C (26–417 pg/mL), and VEGF-D (27–108 pg/mL)]. Haemolyzed samples were excluded from the analysis; in addition, a number of samples had missing data due to limited sample volume or technical errors.

In the STABILITY cohort, 157 biomarkers were analysed in EDTA plasma samples using OLINK Proteomics Multiplex CVD I 96 × 96 panel and the OLINK Proteomics Multiplex Inflammation 96 × 96 panel at Clinical Biomarkers Facility, Science for Life laboratory, Uppsala University, as described previously.²⁰ Replication in STABILITY followed the same methods as in Wallentin et al.²⁰ with the addition of biomarkers of interest being modelled as restricted cubic splines. The analyses were run using the recommended internal control, and inter-plate variability was adjusted by intensity normalization. The resulting relative values were log2-transformed to provide normalized protein expression (NPX) data. In the logarithmic phase of the curve, one increase of the NPX value corresponds to a doubling of the protein content, and a high NPX value corresponds to a high protein concentration.

2.4 Statistical analysis

Analyses of biomarker associations with outcomes were performed using multiple Cox-regression analyses adjusting for a set of pre-specified and well-known cardiovascular risk factors [age, sex, race, smoking status, randomized treatment, history of PCI, CABG, hypertension, dyslipidaemia (including hypercholesterolaemia), diabetes mellitus, angina pectoris, previous MI, congestive heart failure, non-haemorrhagic stroke, peripheral arterial disease, chronic renal disease, index event ST-elevation myocardial infarction (STEMI)/non-ST-elevation myocardial infarction (NSTEMI), and log cystatin-C]. The Cox-regression analyses included the sampling weights according to the study design and were estimated using a robust sandwich estimator. To be able to detect non-linear associations, the biomarkers were modelled using four-knot restricted cubic splines with knots at the 5th, 35th, 65th, and 95th percentiles of each biomarker distribution. Test of non-linear effects was performed by an overall Wald test on the non-linear terms of the restricted cubic spline. The predicted association was presented to aid in the interpretation where non-linear association was present. Biomarker hazard ratios are presented for the upper quartile vs. the lower quartile. Missing values on covariates used in the multiple Cox-regression models were imputed using one round of imputation as implemented in the Multivariate Imputation via Chained Equations package in R statistics software (version 4.1.2). The C-statistic (Harrell’s C) was used to evaluate each model’s ability to discriminate between events and non-events. In addition to the previously named packages, the packages rms and Hmisc were used.

2.5 Genotyping in PLATO, STABILITY, and FRISC-II

Genotyping in PLATO was performed in 3998 individuals using the Illumina HumanOmni2.5–4v1 (Omni2.5) BeadChip (Illumina, San Diego, CA, USA) and 6015 individuals using Illumina InfiniumHumanOmniExpressExome-8v1 BeadChip (Illumina, San Diego, CA, USA) at the SNP&SEQ Platform of the Science for Life Laboratory at Uppsala University. Raw data were analysed with Illumina GenomeStudio 2011.1. Genotyping in STABILITY was performed on the HumanOmniExpressExome-8 v1 array by Expression Analysis (Durham, NC, USA). Genotyping in FRISC-II was performed on the Illumina Global Screening Array (Illumina, San Diego, CA, USA) by Uppsala University’s SNP&SEQ service and data analysed using Illumina GenomeStudio 2.0.3.

2.6 Genetic QC in PLATO, STABILITY, and FRISC-II

Quality controls were performed using the whole-genome association analysis toolset PLINK v1.9 (https://www.cog-genomics.org/plink/) [sex check, call rate single nucleotide polymorphism (SNP) and individual >98%, minor allele frequency (MAF) > 0.001, and Hardy–Weinberg equilibrium (HWE) P-value >1 × 10–8]. Imputation of genotypes was performed using the Haplotype reference consortium v1.1, utilizing Eagle v2.3.3 for phasing and PBWT for imputation. Post-imputation, the data were filtered (MAF >0.005 and IMPUTE2s Info metric >0.7) and converted to hard calls using Plink. Genetic results are reported on build 37 hg19.

2.7 GWAS analysis of biomarkers

The biomarkers (log transformed), from the discovery PLATO cohort, were analysed using multiple linear regression on a genome-wide scale adjusting for gender, age, and the first six genetic principal components using PLINK. The genome-wide significance level was set at P < 5 × 10⁻⁸. If genome-wide significant signals were present, additional analyses were performed adjusting for each genome-wide hit until no signals were left. Throughout the project, SNPs were modelled as additive (i.e. coded 0,1 and 2 per increase of one minor allele) and X-chromosome variants were coded ‘aa’, ‘aA’ and ‘AA’ as 0,1 and 2 for females and ‘a’, ‘A’ as 0 and 1 for males. The amount of variance explained in the outcome by a variant, R², and the partial R² (here defined as the increase in amount of variance explained when the variable is added last to a set of variables) were calculated using univariable and multiple linear regression models in R version 4.1.2. Locus plots of findings were plotted using coordinates of known canonical genes downloaded from the UCSC genome browser (hg19, build37, February 2009).

2.8 Mendelian randomization

SNPs affecting levels of log(VEGF-D) were selected as genetic instruments in the Mendelian randomization analyses. The effect allele was selected as the minor allele in PLATO and all other studies were aligned according to this effect allele. The effect (betaX) vs. log(VEGF-D) per increase of one effect allele was taken from the GWAS analysis in PLATO adjusting for age, sex, and the first four genetic principal components. The effect of each SNP per increase of one effect allele vs. clinical events (betaY) was estimated in the three studies PLATO, STABILITY, and FRISC-II using multiple Cox-regression adjusting for age and sex. The betaY estimate per outcome and SNP was meta-analysed across studies using a fixed effects model as implemented in the R package metafor. To calculate the combined instrumental variable effect, we used the inverse variance-weighted method accounting for genetic linkage-disequilibrium as implemented in the ‘MendelianRandomization’ package in R;²¹ the linkage-disequilibrium (D’, r, and r²) between genetic instruments was calculated in the PLATO material.

3. Results

3.1 Baseline characteristics and biomarker descriptive

The analysis flow chart for the current study, including the discovery patient cohort (PLATO sub-study) and the confirmation patient cohorts (STABILITY for biomarkers and genetics, and FRISC-II for genetics), is presented in Figure 1. The baseline characteristics of the PLATO sub-cohort used to identify VEGF biomarkers associated with outcome are presented in Supplementary material online, Table S1. The random subset of patients from the discovery patient cohort in PLATO (i.e. not case enriched) were similar, compared to the original PLATO trial, in respect to baseline characteristics except for race being predominantly White (98%) within our sample whereas the original PLATO trial had around 92% White, 1% Black, 6% Asian, and 1% other.¹¹ The baseline characteristics of the PLATO genetic sub-cohort, STABILITY, and FRISC-II sub-cohort have previously been published.^15,17,19

The median concentrations of the eight biomarkers [bFGF, Flt-1, KDR (VEGFR2), PlGF, Tie-2, VEGF-A, VEGF-C, and VEGF-D] measured in the PLATO discovery cohort within 24 h of ACS symptom are presented in Supplementary material online, Table S2. VEGF-C was excluded from any further analysis as less than 20% of the analysed samples had values above the LLOQ of the assay (see Supplementary material online, Table S3). A correlation between the remaining seven biomarkers was only observed between VEGF-A and Flt-1 (ρ= −0.66).

3.2 Evaluation of VEGF biomarkers in relation to CV outcome

Using a model adjusting for covariates including cystatin-C, the associations between the seven biomarkers to CV outcome were investigated in the PLATO discovery cohort (Figure 2). The median follow-up time varied between 0.77 and 0.99 years for the different CV outcomes. Flt-1, PlGF, KDR (VEGFR2), and VEGF-D were significantly associated with several outcomes. Non-linear effects (P < 0.05 for the non-linear terms) were present for Flt-1, PlGF, VEGF-D, and VEGF-A (see Supplementary material online, Figures S1A-D). VEGF-D was strongly associated with all four CV outcomes, with the most pronounced association being with CV death. The cumulative risk of CV death by quartile groups of VEGF-D is plotted in Figure 3. The baseline characteristics per quartile for VEGF-D in PLATO are listed in Supplementary material online, Table S4. The associations between VEGF-D and the CV outcomes were replicated, with similar HR, in the confirmation cohort from STABILITY (Figure 4).

Figure 2

Adjusted^† HR for all PLATO VEGF biomarkers vs. outcome (n = 2091). Levels are the Q3 vs. Q1 for each biomarker for which the estimates are presented and N shows the number of events/number of controls. For ease of interpretation, point estimates for significant effects (P < 0.05) are highlighted. Note that the biomarkers are modelled using a restricted four-knot spline and due to possible nonlinear effects, a significant overall effect may be present whereas the confidence interval of the point estimate for Q3 vs. Q1 overlaps 1. Biomarkers with significant non-linear terms vs. any outcome are Flt-1, PlGF, VEGF-D, and VEGF-A (see Supplementary material online, Figures S1A-D). ^†Adjusted by age, sex, race, smoking status, randomized treatment, history of PCI, CABG, hypertension, dyslipidaemia (including hypercholesterolaemia), diabetes mellitus, angina pectoris, previous MI, congestive heart failure, non-haemorrhagic stroke, peripheral arterial disease, chronic renal disease, index event STEMI/NSTEMI, and log cystatin-C. The outcomes MACE = CVD, MI, or stroke, MACE* = CVD, spontaneous MI, or stroke.

Figure 3

Kaplan–Meier plot of the cumulative risk of CV death by quartile group of VEGF-D. Estimates and number at risk were calculated by incorporating sampling weights; hence, the number of patients at risk are more than observed in the data N = 1966.

Figure 4

Adjusted HR for VEGF-D vs. outcome in PLATO (n = 1966) and replication cohort STABILITY (n = 2994). Levels are the Q3 vs. Q1 for VEGF-D in each cohort for which the estimates are presented and N shows the number of events/number of controls. PLATO result adjusted by age, sex, race, smoking status, randomized treatment, history of PCI, CABG, hypertension, dyslipidaemia (including hypercholesterolaemia), diabetes mellitus, angina pectoris, previous MI, congestive heart failure, non-haemorrhagic stroke, peripheral arterial disease, chronic renal disease, index event STEMI/NSTEMI, and log cystatin-C. STABILITY result adjusted by age, sex, BMI, hypertension, previous stroke or TIA, randomized treatment, previous PCI or CABG, diabetes, previous MI, smoker or previous smoker, previous PAD, and log Cystatin-C. The outcomes MACE = CVD, MI, or stroke, MACE* = CVD, spontaneous MI, or stroke.

3.3 VEGFD genetic variants are significantly associated to VEGF-D plasma levels

Genome-wide significant associations with log(VEGF-D) levels were identified at the VEGFD locus on chromosome Xp22 (Figure 5, Supplementary material online, Table S4). The top signal came from rs192812042 (beta = 0.2536 [95% CI 0.1998–0.3073, P = 5.82 × 10⁻¹², MAF = 0.126) located in the gene PIR close to FIGF (VEGFD, see locus plot in Supplementary material online, Figure S1). The follow-up analysis, adjusting for rs192812042, gave additional signals from the region (see Supplementary material online, Table S5) where rs234500 was the variant closest to VEGFD (beta = −0.1053 [95% CI −0.1434 - −0.06727], P = 6.51 × 10⁻⁰⁸, MAF = 0.439). The variants rs192812042 and rs234500 increased the explained variance (i.e. partial R²) in log(VEGF-D) with 0.041 and 0.029, respectively, when added to a model including age, sex, and the first six genetic principal components. The combination rs192812042 and rs234500 increased the R² to 0.055.

Manhattan plot for the GWAS of log(VEGF-D) in PLATO (n = 1966). The horizontal red line indicates the level of genome-wide significance (P = 5 × 10−8). The analyses are adjusted by age, sex, and the first six genetic principal components.

Figure 5

Manhattan plot for the GWAS of log(VEGF-D) in PLATO (n = 1966). The horizontal red line indicates the level of genome-wide significance (P = 5 × 10⁻⁸). The analyses are adjusted by age, sex, and the first six genetic principal components.

Replication of the association of the SNP rs192812042 and rs234500 to VEGF-D was performed in 2944 patients in the STABILITY confirmation cohort using VEGF-D levels measured with the Olink CVD I panel. The effects of rs192812042 and rs234500 vs. VEGF-D (unit NPX on log2 scale) were beta 0.263 (95% CI 0.0.233–0.294, P = 1.54 × 10⁻⁶¹) and −0.145 (95% CI −0.166 - −0.124, P = 8.00 × 10⁻⁴⁰), respectively.

3.4 Mendelian randomization analyses with CV outcomes

The variants rs192812042 and rs234500 were selected as instruments in the MR analyses. The beta values (standard error) vs. log(VEGF-D) of 0.2536 (0.027) and −0.1458 (0.019) were taken from the first GWAS analysis (see Supplementary material online, Table S5). The LD between the variants in PLATO were D’ = 0.90, r = −0.18, and r² = 0.03. For males (n = 6932) the LD was D’ = 0.99, r = −0.14, and r² = 0.02 and for females (n = 3047)) the LD was D’ = 0.87, r = −0.30, and r² = 0.09. The outcomes major adverse cardiovascular events (MACE) (CVD, MI, or stroke), MACE* (CVD, spontaneous MI, or stroke), MI, CV death, and stroke were analysed in the three genetic cohorts of PLATO, STABILITY, and FRISC-II where each study has a median follow-up time of 0.99, 3.74, and 5.0 years, respectively. The final MR analysis, which combines the effect of the two SNPs across studies, demonstrated a significant effect on CV mortality [P = 0.0257, HR 1.81 (1.07, 3.04) per increase of one unit in log (VEGF-D)] (Figure 6). Per study, SNP results are presented in the Supplementary material online, Figure S3. Due to the genetic instruments being located on chromosome-X, sensitivity analyses were performed by gender for the causal effects (Supplementary Figures 4A–D). The low proportion of females, 25%, renders female results uncertain while the overall result generalizes to males. In addition to the MR analyses, we performed a colocalization analysis for VEGF-D levels vs. the clinical outcome on CV mortality. Although this method requires high power vs. exposure and outcome, the analysis gave signs that rs192812042 is a shared causal variant between the two traits (Supplementary results).

Figure 6

VEGF-D Mendelian randomization results based on meta-analyses of the three cohorts PLATO (N = 10 013), STABILITY (N =10 786), and FRISC-II (N = 2952). The estimate marked with a triangle is the main MR result per outcome (weighted result using both instruments). N, number of observations; N(e), number of events; N(s), number of studies. The analyses are adjusted by age and sex. For individual study estimates, see Supplementary material online, Figure S3. The outcome MACE = CVD, MI, or stroke.

4. Discussion

In the present study of patients with ACS and CCS, the VEGF-D plasma level was consistently and independently associated with an increased risk of subsequent CV death. Genome-wide significant associations with plasma VEGF-D levels were identified in the region of the VEGFD locus on chromosome Xp22 with the top signals from rs192812042 and rs234500 in patients with ACS and these VEGFD genetic variants were replicated in patients with CCS. The combined effect of the two SNPs across three CV studies in ACS and CCS demonstrated a significant effect on CV mortality (P = 0.0257, HR 1.81 per increase of one unit in log VEGF-D). Thereby, for the first time, we demonstrated a relationship between VEGFD genetic variants, VEGF-D plasma levels, and CV outcome in both ACS and CCS. These findings are consistent with a causal role of VEGF-D in coronary artery disease diseases and its complications.

Previous GWAS studies of VEGF-D levels reported in GWAS Catalog (https://www.ebi.ac.uk/gwas/) have identified loci on chromosomes 3, 4, 5, 14, and X, close to the VEGFD locus (see Supplementary material online, Table S7). In total, five SNPs located near VEGFD have been associated with VEGF-D levels, including rs192812042 as identified in the current study. In a recent study by Gudjonsson A et al.,²² rs192812042-A was associated with a 0.47 unit increase [0.43–0.51] (P = 2 × 10–122) of VEGF-D in a GWAS of serum proteins in 5364 Icelandic subjects.²² These data are consistent with our findings in that rs192812042-A is the VEGF-D-raising allele.

The minor alleles rs192812042-A and rs234500-G were significantly associated with elevated and reduced plasma VEGF-D levels, respectively, in patients with ACS in the present study. These findings are consistent with eQTL data from the Genotype-Tissue Expression (GTEx) project (gtexportal.org) in which rs192812042-A is associated with higher VEGF-D mRNA expression in several tissues (including heart and liver) and rs234500-G is associated with lower VEGF-D mRNA expression in several tissues (including muscle and arteries) (see Supplementary material online, Figure S5). Both SNPs are also associated with mRNA expression of additional genes at the Xp22 locus according to GTEx data, most notably PIGA (phosphatidylinositol glycan anchor biosynthesis class A) and PIR (pirin), both apparently unrelated to VEGF biology (see Supplementary material online, Table S8). Thus, although we cannot rule out genes other than VEGFD as drivers of the genetic association between rs192812042 and rs234500 and investigated outcomes, we believe that VEGFD being the causal gene is the most biologically plausible assumption.

VEGF-D is a secreted factor that regulates lymphangiogenesis, angiogenesis, and endothelial proliferation either through VEGFR2 (KDR), the core receptor for angiogenesis, or VEGFR3, the essential receptor for regulating lymphatic growth.^23,24 VEGF-D together with VEGF-C and their receptor VEGFR3 are main components of the central pathway for the development, growth, and maintenance of lymphatic vessels, leading to important functions in the maintenance of tissue fluid balance and myocardial function after ischaemic injury. Therefore, VEGF-D signalling via VEGFR3 plays an important role in pathological processes such as inflammation, wound healing, and lymphedema, and mediates the transport of signalling molecules, lipoproteins, and immune cells between injured tissues and regional lymph nodes that can induce lymphangiogenesis which may contribute to CV diseases.²⁵ Recently, it was shown in a mouse model that proinflammatory cytokines, such as IL-1β, activated the NF-kB pathway to induce VEGFR3 expression, resulting in enhanced production and functional effects of VEGF-A, VEGF-C, and VEGF-D.²⁶ In an another mouse study, it was demonstrated that down-regulated VEGF-D/VEGFR3 pathway resulted in reduced expression of genes involved in production of triglycerides and cholesterol, which supports the role of VEGF-D as a regulator of lipid metabolism.²⁷

Besides modulating the lymphangiogenesis process, VEGF-D also acts as a promotor for angiogenesis, supported by a study showing that the elevated plasma VEGF-D levels in heart failure (HF) patients were reduced to normal values after heart transplantation.²⁸ These data reflect the reversal of pulmonary congestion and the recovery of pulmonary artery compliance and pulmonary vascular resistance following transplantation. In a mouse study, increased IL-1β induced down-regulation of VEGF-D in cardiac microvascular endothelial cells through ERK1/2, JNK, and PKCα/β, which also supports a role of VEGF-D in angiogenesis.²⁹ In addition, VEGF-D has been shown to promote cardiac fibrosis by stimulating myofibroblast growth, migration, and collagen synthesis.²⁴VEGFD is broadly expressed across human tissues, with highest expression in the lung. In the adult heart, VEGFD is preferentially expressed by fibroblasts and cardiomyocytes, which also supports a role for VEGF-D in myofibroblast growth.

Elevated levels of plasma VEGF-D have previously been found in patients with heart failure, atrial fibrillation, ischaemic stroke, pulmonary arterial hypertension (PAH), chronic thromboembolic pulmonary hypertension (CTEPH), and lymphangioleiomyomatosis (LAM).^30–34 Recently, serum VEGF-D was also shown to be an independent predictor of CV death in a prospective study on patients with CCS.³⁵ In the present study, we demonstrated that plasma VEGF-D levels and VEGFD genotype are independent predictors of CV death in patients with both ACS and CCS. In the ACS cohort (PLATO), we investigated the dynamics of plasma VEGF-D levels at three time points (at admission, i.e. within 24 h of symptoms, at discharge 3–4 days after symptoms, and 1 month after ACS event) and found the levels to be stable over time (data not shown), indicating the VEGF-D plasma levels did not just increased in the acute setting and supporting the hypothesis of a wider role of VEGF-D in CV disease.

As previously described, it is possible that circulating VEGF-D levels presents an adaptation to the demands of the lymphatic system to increase the lymphatic capacity and remove excess fluid from the extravascular space of the lungs and peripheral tissues in patients with various CV diseases.³⁰ Due to the novel finding of VEGF-D´s role, plasma VEGF-D levels or VEGFD genetic variants may directly reflect the capacity of lymphangiogenesis, angiogenesis, endothelial proliferation, and cardiac fibrogenesis to promote repair and remodelling. All data together support plasma VEGF-D or VEGFD genetic variants as biomarkers for prediction of CV death in both acute and chronic conditions.

4.1 Strengths and limitations

The strength of the current study is the use of well-defined clinical cohorts using three separate studies: one ACS cohort for the discovery phase and two independent cohorts of ACS and CCS patients for the confirmation study. The VEGF-D data were consistently and independently associated with an increased risk of subsequent CV death in both acute and chronic clinical settings. All data together strengthen the results of VEGF-D as a prognostic biomarker and an early predictor of CV disease. Limitations in the study include lack of data for VEGF-B and VEGF-C, the other lymphangiogenic factor, as well as data for VEGFR3 the receptor for both VEGF-D and VEGF-C. The MR-analyses were affected by the sample size in the clinical outcome analyses resulting in limited power. There is also a lack of publicly available GWAS results on chromosome X for similar cohorts and clinical outcomes (e.g. CV death) for replication. The low proportion female patients in the MR analyses combined with the instruments being located on chromosome X renders the generalizability of the results to female sex uncertain. Although we believe VEGF-D being the causal exposure in our MR analyses, there is always a possibility for violations of MR assumptions, such as horizontal pleiotropy, which could bias the results. However, the use of cis-acting variants should lower this risk. In addition, although limited by low power in the clinical outcome analysis, there were signs of colocalization for our main variant.

4.2 Conclusion

This is the first large-scale cohort study to demonstrate that both VEGF-D plasma levels and VEGFD genetic variants are independently associated with CV outcomes in patients with ACS and CCS indicating a possible causal role of VEGF-D in CV diseases. Measurements of VEGF-D levels and/or VEGFD genetic variants may provide incremental prognostic information in patients with ACS and CCS and VEGF-D–related pathways might be interesting in the development of novel treatment strategies in ACS and CCS.

Supplementary material

Supplementary material is available at Cardiovascular Research online.

Authors’ contributions

P.D., S.E., N.E., and L.W. designed the study; S.E., A.B., and A.Ca. performed the biomarker analysis; N.E. performed the statistical analysis; P.D., S.E., N.E., A.W., and L.W. interpreted the data and wrote the manuscript; All authors revised and approved the manuscript.

Pia Davidsson (Ph.D.), Susanna Eketjäll, Niclas Eriksson, Anna Walentinsson, Richard C Becker, Anders Cavallin, Anna Bogstedt, Anna Collén, Claes Held, Stefan James, Agneta Siegbahn, Ralph Stewart, Robert F Storey, Harvey White, and Lars Wallentin.

Acknowledgements

The current work could not have been done without excellent support from colleagues at both AstraZeneca and Uppsala Clinical Research Center.

Funding

This work was supported by AstraZeneca; Swedish Heart-Lung Foundation; Swedish Foundation for Strategic Research; and Uppsala Clinical Research Center, Uppsala University, Sweden, for biochemical and statistical analyses. GlaxoSmithKline sponsored the main STABILITY trial, biobanking of the samples, and the GWAS analyses but provided no specific support for this sub-study. Pharmacia (Pfizer) sponsored the FRISC2 trial and the biobanking of the samples but provided no specific support for this sub-study.

Data availability

The data underlying this article are available in the article and in the Supplementary material.

References

1

Mc Namara

K

,

Alzubaidi

H

,

Jackson

JK

.

Cardiovascular disease as a leading cause of death: how are pharmacists getting involved?

.

Integr Pharm Res Pract Vol

2019

;

8

:

1

–

11

.

2

Libby

P

.

Inflammation and cardiovascular disease mechanisms

.

Am J Clin Nutr

2006

;

83

:

456S

–

460S

.

3

Alfaddagh

A

,

Martin

SS

,

Leucker

TM

,

Michos

ED

,

Blaha

MJ

,

Lowenstein

CJ

,

Jones

SR

,

Toth

PP

.

Inflammation and cardiovascular disease: from mechanisms to therapeutics

.

Am J Prev Cardiol

2020

;

4

:

100–130

.

4

Sun

HJ

,

Wu

ZY

,

Nie

XW

,

Bian

JS

.

Role of endothelial dysfunction in cardiovascular diseases: the link between inflammation and hydrogen sulfide

.

Front Pharmacol

2020

;

10

:

1568

–1583.

5

Cervantes Gracia

K

,

Llanas-Cornejo

D

,

Husi

H

.

CVD and oxidative stress

.

J Clin Med

2017

;

6

(

2

):

1–22

.

PMID: 28230726

.

6

Artiach

G

,

Sarajlic

P

,

Bäck

M

.

Inflammation and its resolution in coronary artery disease: a tightrope walk between omega-6 and omega-3 polyunsaturated fatty acids

.

Kardiol Pol

2020

;

78

(

2

):

93

–

95

.

PMID: 32108752

.

7

Sarajlic

P

,

Artiach

G

,

Larsson

S

,

Bäck

M

.

Dose-dependent risk reduction for myocardial infarction with eicosapentaenoic acid: a meta-analysis and meta-regression including STRENGTH trial

.

Cardiovasc Drugs Ther

2021

;

35

:

1079

–

1081

.

PMID: 34101063

.

8

Sut

A

,

Chizynski

K

,

Rozalski

M

,

Golanski

J

.

Dietary intake of omega fatty acids and polyphenols and its relationship with levels of inflammatory markers in men with chronic coronary syndrome after percutaneous coronary intervention

.

Kardiol Pol

2020

;

78

:

117

–

123

.

9

Apte

RS

,

Chen

DS

,

Ferrara

N

.

VEGF in signaling and disease: beyond discovery and development

.

Cell

2019

;

176

:

1248

–

1264

.

10

Zhou

Y

,

Zhu

X

,

Cui

H

,

Shi

J

,

Yuan

G

,

Shi

S

,

Hu

Y

.

The role of the VEGF family in coronary heart disease

.

Front Cardiovasc Med

2021

;

8

:

738325

.

11

Wallentin

L

,

Becker

RC

,

Budaj

A

,

Cannon

CP

,

Emanuelsson

H

,

Held

C

,

Horrow

J

,

Husted

S

,

James

S

,

Katus

H

,

Mahaffey

KW

,

Scirica

BM

,

Skene

A

,

Steg

PG

,

Storey

RF

,

Harrington

RA

.

Ticagrelor versus clopidogrel in patients with acute coronary syndromes

.

N Engl J Med

2009

;

361

:

1045

–

1057

.

12

White

H

,

Held

C

,

Stewart

R

,

Watson

D

,

Harrington

R

,

Budaj

A

,

Steg

PG

,

Cannon

CP

,

Krug-Gourley

S

,

Wittes

J

,

Trivedi

T

,

Tarka

E

,

Wallentin

L

.

Study design and rationale for the clinical outcomes of the STABILITY Trial (STabilization of Atherosclerotic plaque By Initiation of darapLadIb TherapY) comparing darapladib versus placebo in patients with coronary heart disease

.

Am Heart J

2010

;

160

:

655

–

661.e2

.

13

White

HD

,

Held

C

,

Stewart

R

,

Tarka

E

,

Brown

R

,

Davies

RY

,

Budaj

A

,

Harrington

RA

,

Steg

PG

,

Ardissino

D

,

Armstrong

PW

,

Avezum

A

,

Aylward

PE

,

Bryce

A

,

Chen

H

,

Chen

MF

,

Corbalan

R

,

Dalby

AJ

,

Danchin

N

,

De Winter

RJ

,

Denchev

S

,

Diaz

R

,

Elisaf

M

,

Flather

MD

,

Goudev

AR

,

Granger

CB

,

Grinfeld

L

,

Hochman

JS

,

Husted

S

,

Kim

HS

,

Koenig

W

,

Linhart

A

,

Lonn

E

,

López-Sendón

J

,

Manolis

AJ

,

Mohler

ER

III,

Nicolau

JC

,

Pais

P

,

Parkhomenko

A

,

Pedersen

TR

,

Pella

D

,

Ramos-Corrales

MA

,

Ruda

M

,

Sereg

M

,

Siddique

S

,

Sinnaeve

P

,

Smith

P

,

Sritara

P

,

Swart

HP

,

Sy

RG

,

Teramoto

T

,

Tse

HF

,

Watson

D

,

Weaver

WD

,

Weiss

R

,

Viigimaa

M

,

Vinereanu

D

,

Zhu

J

,

Cannon

CP

,

Wallentin

L

.

Darapladib for preventing ischemic events in stable coronary heart disease

.

N Engl J Med

2014

;

370

(

18

):

1702

–

1711

.

Epub 2014 Mar 30. PMID: 24678955

.

PubMed

14

Mälarstig

A

,

Eriksson

P

,

Hamsten

A

,

Lindahl

B

,

Wallentin

L

,

Siegbahn

A

.

Raised interleukin-10 is an indicator of poor outcome and enhanced systemic inflammation in patients with acute coronary syndrome

.

Heart

2008

;

94

:

724

–

729

.

15

Johansson

Å

,

Alfredsson

J

,

Eriksson

N

,

Wallentin

L

,

Siegbahn

A

.

Genome-wide association study identifies that the ABO blood group system influences interleukin-10 levels and the risk of clinical events in patients with acute coronary syndrome

.

PLoS ONE

2015

;

10

(

11

):

e0142518

.

16

Wallentin

L

,

Lindholm

D

,

Siegbahn

A

,

Wernroth

L

,

Becker

RC

,

Cannon

CP

,

Cornel

JH

,

Himmelmann

A

,

Giannitsis

E

,

Harrington

RA

,

Held

C

,

Husted

S

,

Katus

HA

,

Mahaffey

KW

,

Steg

PG

,

Storey

RF

,

James

SK

,

PLATO study group

.

Biomarkers in relation to the effects of ticagrelor in comparison with clopidogrel in non-ST-elevation acute coronary syndrome patients managed with or without in-hospital revascularization: a substudy from the Prospective Randomized Platelet Inhibition and Patient Outcomes (PLATO) trial

.

Circulation

2014

;

129

(

3

):

293

–

303

.

Epub 2013 Oct 29. PMID: 24170388

.

17

Varenhorst

C

,

Eriksson

N

,

Johansson

Å

,

Barratt

BJ

,

Hagström

E

,

Åkerblom

A

,

Syvänen

A-C

,

Becker

RC

,

James

SK

,

Katus

HA

,

Husted

S

,

Steg

PG

,

Siegbahn

A

,

Voora

D

,

Teng

R

,

Storey

RF

,

Wallentin

L

.

Effect of genetic variations on ticagrelor plasma levels and clinical outcomes

.

Eur Heart J

2015

;

36

:

1901

–

1912

.

18

Wallentin

L

,

Lagerqvist

B

,

Husted

S

,

Kontny

F

,

Ståhle

E

,

Swahn

E

.

Outcome at 1 year after an invasive compared with a non-invasive strategy in unstable coronary-artery disease: the FRISC II invasive randomised trial. FRISC II Investigators. Fast revascularisation during instability in coronary artery disease

.

Lancet

2000

;

356

(

9223

):

9

–

16

.

PMID: 10892758

.

19

Yeo

A

,

Li

L

,

Warren

L

,

Aponte

J

,

Fraser

D

,

King

K

,

Johansson

K

,

Barnes

A

,

MacPhee

C

,

Davies

R

,

Chissoe

S

,

Tarka

E

,

O'Donoghue

ML

,

White

HD

,

Wallentin

L

,

Waterworth

D

.

Pharmacogenetic meta-analysis of baseline risk factors, pharmacodynamic, efficacy and tolerability endpoints from two large global cardiovascular outcomes trials for darapladib

.

PLoS One

2017

;

12

(

7

):

e0182115

.

PMID: 28753643; PMCID: PMC5533343

.

20

Wallentin

L

,

Eriksson

N

,

Olszowka

M

,

Grammer

TB

,

Hagström

E

,

Held

C

,

Kleber

ME

,

Koenig

W

,

März

W

,

Stewart

RAH

,

White

HD

,

Åberg

M

,

Siegbahn

A

.

Plasma proteins associated with cardiovascular death in patients with chronic coronary heart disease: a retrospective study

.

PLoS Med

2021

;

18

:

e1003513

.

21

Yavorska

OO

,

Burgess

S

.

Mendelianrandomization: an R package for performing Mendelian randomization analyses using summarized data

.

Int J Epidemiol

2017

;

46

(

6

):

1734

–

1739

.

PMID: 28398548; PMCID: PMC5510723

.

22

Gudjonsson

A

,

Gudmundsdottir

V

,

Axelsson

GT

,

Gudmundsson

EF

,

Jonsson

BG

,

Launer

LJ

,

Lamb

JR

,

Jennings

LL

,

Aspelund

T

,

Emilsson

V

,

Gudnason

V

.

A genome-wide association study of serum proteins reveals shared loci with common diseases

.

Nat Commun

2022

;

13

(

1

):

480

.

PMID: 35078996; PMCID: PMC8789779

.

23

Rissanen

TT

,

Markkanen

JE

,

Gruchala

M

,

Heikura

T

,

Puranen

A

,

Kettunen

MI

,

Kholová I, Kauppinen RA, Achen MG, Stacker SA, Alitalo K, Ylä-Herttuala S.

VEGF-D Is the strongest angiogenic and lymphangiogenic effector among VEGFs delivered into skeletal muscle via adenoviruses

.

Circul Res

2003

;

92

:

1098

–

1106

.

24

Zhao

T

,

Zhao

W

,

Meng

W

,

Liu

C

,

Chen

Y

,

Bhattacharya

SK

, Sun Y.

Vascular endothelial growth factor-D mediates fibrogenic response in myofibroblasts

.

Mol Cell Biochem

2016

;

413

:

127

–

135

.

25

Lammoglia

GM

, Van Zandt CE, Galvan DX, Orozco JL, Dellinger MT, Rutkowski JM.

Hyperplasia, de novo lymphangiogenesis, and lymphatic regression in mice with tissue-specific, inducible overexpression of murine VEGF-D

.

J Physiol Heart Circ Physiol

. 2016;

311

:

H384

–

H394

.

26

Meng

F

,

Jing

X

,

Song

G

,

Jie

L

,

Shen

F

.

Prox1 induces new lymphatic vessel formation and promotes nerve reconstruction in a mouse model of sciatic nerve crush injury

.

J Anat

2020

;

237

:

933

–

940

.

27

Tirronen

A

,

Vuorio

T

,

Kettunen

S

,

Hokkanen

K

,

Ramms

B

,

Niskanen

H

,

Laakso

H

,

Kaikkonen

NU

,

Jauhiainen

M

,

Gordts

PLSM

,

Ylä-Herttuala

S

.

Deletion of lymphangiogenic and angiogenic growth factor VEGF-D leads to severe hyperlipidemia and delayed clearance of chylomicron remnants

.

Arterioscler Thromb Vasc Biol

2018

;

38

(

10

):

2327

–

2337

.

28

Ahmed

S

,

Ahmed

A

,

Säleby

J

,

Bouzina

H

,

Lundgren

J

,

Rådegran

G

.

Elevated plasma tyrosine kinases VEGF-D and HER4 in heart failure patients decrease after heart transplantation in association with improved haemodynamics

.

Heart Vessels

2020

;

35

(

6

):

786

–

799

.

Epub 2020 Jan 20. PMID: 31960146

.

29

Mountain

DJ

,

Singh

M

,

Singh

K

.

Downregulation of VEGF-D expression by interleukin-1beta in cardiac microvascular endothelial cells is mediated by MAPKs and PKCalpha/beta1

.

J Cell Physiol

2008

;

215

:

337

–

343

.

30

Borné

Y

,

Gränsbo

K

,

Nilsson

J

,

Melander

O

,

Orho-Melander

M

,

Smith

JG

,

Engström

G

.

Vascular endothelial growth factor D, pulmonary congestion, and incidence of heart failure

.

J Am Coll Cardiol

2018

;

71

:

580

–

582

.

31

Berntsson

J

,

Smith

JG

,

Johnson

LSB

,

Söderholm

M

,

Borné

Y

,

Melander

O

,

Orho-Melander

M

,

Nilsson

J

,

Engström

G

.

Increased vascular endothelial growth factor D is associated with atrial fibrillation and ischaemic stroke

.

Heart

2019

;

105

:

553

–

558

.

32

Säleby

J

,

Bouzina

H

,

Lundgren

J

,

Rådegran

G

.

Angiogenic and inflammatory biomarkers in the differentiation of pulmonary hypertension

.

Scand Cardiovasc J

2017

;

51

:

261

–

270

.

33

Säleby

J

,

Bouzina

H

,

Ahmed

S

,

Lundgren

J

,

Rådegran

G

.

Plasma receptor tyrosine kinase RET in pulmonary arterial hypertension diagnosis and differentiation

.

ERJ Open Res

2019

;

5

:

00037

–

02019

.

34

Seyama

K

,

Kumasaka

T

,

Souma

S

,

Sato

T

,

Kurihara

M

,

Mitani

K

,

Tominaga

S

,

Fukuchi

Y

.

Vascular endothelial growth factor-D is increased in serum of patients with lymphangioleiomyomatosis

.

Lymphat Res Biol

2006

;

4

(

3

):

143

–

152

.

PMID: 17034294

.

35

Wada

H

,

Suzuki

M

,

Matsuda

M

,

Ajiro

Y

,

Shinozaki

T

,

Sakagami

S

,

Yonezawa

K

,

Shimizu

M

,

Funada

J

,

Takenaka

T

,

Morita

Y

,

Nakamura

T

,

Fujimoto

K

,

Matsubara

H

,

Kato

T

,

Unoki

T

,

Takagi

D

,

Wada

K

,

Wada

M

,

Iguchi

M

,

Masunaga

N

,

Ishii

M

,

Yamakage

H

,

Kusakabe

T

,

Yasoda

A

,

Shimatsu

A

,

Kotani

K

,

Satoh-Asahara

N

,

Abe

M

,

Akao

M

,

Hasegawa

K

.

Distinct characteristics of VEGF-D and VEGF-C to predict mortality in patients with suspected or known coronary artery disease

.

J Am Heart Assoc

2020

;

9

:1–12