Introduction

Hypertrophic cardiomyopathy (HCM) is the most common inherited cardiomyopathy with significant morbidity and risk of sudden cardiac death.1 HCM shows a variable genotype–phenotype correlation and penetrance, as well as a remarkable phenotypic heterogeneity.2 Often, underlying sarcomeric disease-causing mutations do not correlate with the phenotype and genotypic markers that can predict disease severity have yet to be found. This has hampered the use of genetic information in patient risk stratification. In particular, there is little data correlating genetic data with systolic or diastolic myocardial performance in this myocardial disease. While most research has focused on the sarcomeric, disease-producing, genes, recently, non-synonymous single-nucleotide polymorphisms (SNPs) in modifier genes in the pro-fibrotic and pro-inflammatory pathways have been identified in pediatric patients with HCM and are shown to have an impact on age of presentation, severity of disease, and cardiac diastolic dysfunction that is independent of underlying disease-causing mutation in sarcomeric genes.3 In particular, allelic variation in these pro-fibrotic and pro-inflammatory genes have been shown to be a predictor of diastolic dysfunction in HCM as measured by two-dimensional (2-D) and tissue Doppler echocardiography.3 This is of importance as diastolic dysfunction is a hallmark of HCM and a significant cause for HCM-associated morbidity and mortality in adult and pediatric HCM.1,4 In HCM patients, the grade of diastolic dysfunction correlates to exercise capacity5 and is an outcome predictor.6,7

Strain imaging to assess myocardial deformation is a more recent sensitive8 tool in the assessment of systolic and diastolic myocardial function in HCM.9 Mean 2-D longitudinal strain is reduced in HCM and it is particularly useful in the assessment of regional wall motion abnormalities and dysfunction, a common feature of HCM.10 Importantly, as traditional indices of left ventricular (LV) systolic function such as ejection fraction (EF) are typically normal, or even increased, in HCM, myocardial deformation imaging using 2-D strain can detect regional dysfunction in the context of normal EF.11

Furthermore, segmental 2-D and 3-D speckle-tracking strain has been shown to correlate with fibrosis in patients with HCM.11,12 Thus, 2-D systolic and diastolic strain imaging is a promising technology to explore phenotype–genotype correlations in HCM, but its relationship to genotype/modifier genes has not been investigated. Fibrosis in HCM may be mediated through the hypoxia-inducible factor-1α (HIF1A) and vascular endothelial growth factor (VEGF) pathway, activated by a thickened, hypoperfused, and hypoxic myocardium.13 This leads to the activation of not only HIF1A, which in turn stimulates angiogenesis,14 but also endothelial–mesenchymal transition via activation of transforming growth factor β1 (TGFβ1) that can lead to myocardial fibrosis,15 which is an important predictor of adverse outcome in HCM.16 In addition, VEGF is an important angiogenic factor and variants that reduce its activity could impair angiogenesis and adversely impact adaptation to myocardial ischemia.

This study investigates the relationship between previously identified allelic variance in the pro-fibrotic and angiogenic gene VEGF and hypoxia-response gene HIF1A and the level of LV myocardial systolic and diastolic dysfunction as measured by 2-D strain. We hypothesize that functional SNP in the HIF1A-VEGF pathway can lead to reduced myocardial deformation.

Methods

In a prospective cohort study, unrelated HCM cases <18 years old were enrolled at the Hospital for Sick Children Toronto (2007–2010). Patients with significant valvar insufficiency, hypertension, congenital heart disease, and metabolic disorders were excluded. The study was approved by the Institutional Review Board of the Hospital for Sick Children and informed consent was obtained from all subjects or parents/legal guardians.

Echocardiography

All patients underwent 2-D echocardiography using IE-33 (Philips Ultrasound, Bothell, WA) or Vivid 7 (GE, Horten, Norway) systems. Measurements were made offline by a single investigator (J.A.) who was blinded to subject genotype using commercially available software (Syngo Dynamics, Siemens, Malvern, PA). Echocardiographic data was obtained at last follow-up or last investigation before myectomy or transplantation. Echocardiographic assessment included 2-D measurements of LV dimensions in parasternal long- or short-axis planes, LV fractional shortening (FS), bi-plane Simpson EF, maximal end-diastolic interventricular septal wall thickness (IVSd), and LV posterior wall thickness (LVPWd). Measurements were made offline from digitally stored images using the average from three consecutive cardiac cycles. IVSd and LVPWd z-scores were calculated to adjust for anthropometric differences.

Detailed echocardiographic evaluation of LV systolic and diastolic myocardial deformation was assessed using 2-D speckle-tracking echocardiography. The endocardial border was traced in the apical four-chamber view for longitudinal, and the parasternal short-axis view for circumferential and radial strain. The region of interest was adjusted to wall thickness. Myocardial tracking by the software was verified visually and retraced if necessary until adequate tracking was achieved. From the deformation curves, the LV lateral and septal wall longitudinal peak systolic and diastolic, radial peak systolic and diastolic, and circumferential peak systolic and diastolic strain (S) and strain rate (SR) were measured in three lateral and three septal, as well as anterior, inferior, and posterior segments to reflect multiplane LV myocardial deformation.

Genotyping

Blood or saliva was obtained for DNA extraction. Genotyping was performed for SNPs chosen based on previous association studies, functional effects, allele frequency, and prevalence.17,18,19 The following SNPs were investigated: VEGFA 2578A/C (rs699947) and VEGFA 1154A/G (rs1570360) both SNPs in the VEGF promoter, VEGFA 634C/G (rs2010963) in the 5′-untranslated region (UTR), and HIF1A 145C/T (rs10873142), HIF1A 1326C/T (rs2057482), and HIF1A 1744C/T (rs11549465) in the 3′-UTR (for primer sequences please see ref. 3). Functioning alleles (major alleles) in HIF1A and loss-of-function alleles (minor alleles) in VEGFA were considered risk alleles. These alleles have shown previous associations with coronary artery disease and chronic heart failure.17,18,20,21,22,23 SNP assays were performed using Applied Biosystems Taqman SNP genotyping technology (Assays-by-Design, Burlington, Canada).

Myocardial VEGF expression analysis was performed as described previously3 on a subset of six samples, where myocardial tissue was available. In brief, myocardial VEGF expression was scaled semi-quantitatively from 1 to 3, where high is 1 and 3 weak.

Statistics

Patient characteristics (age, sex, and positive family history), genetic variants, and phenotypes were described by summary statistics. Continuous variables were summarized in terms of mean and standard deviation; dichotomous and polytomous variables were summarized in terms of the number and proportion of patients. The association between the SNPs and phenotypes was assessed separately for each echocardiographic parameter using t tests at the 5% significance level. We used a dominant model where the presence of at least one risk allele (heterozygous or homozygous) was defined as a risk genotype. Because of the large number of echocardiographic markers and the small number of patients, the reliability of any single significant pairwise association can be limited. However, we hypothesized that if a genotype was associated with a favorable or adverse HCM-related phenotype, we would be able to detect associations across multiple parameters beyond chance alone. We used binomial tests as a means to account for the large number of pairwise comparisons in phenotypes within each genotype. The binomial tests are used to assess if the observed number of “statistically significant” comparisons is significantly larger or smaller than the expected number of false-positive comparisons. A gene is considered associated with HCM phenotype if the p value of the binomial test is <5%. Standardized mean difference was calculated as the ratio between between-SNP difference in an echocardiographic parameter and the pooled standard deviation. For intra- and inter-observer variability, we generated scatter plots comparing the original to intra-observers’ and inter-observers’ assessments. Next, we quantified the intra- and inter-observer variability using the Bland–Altman agreement plot and intraclass correlation (ICC) as described.24 All statistical analysis was conducted using SAS v9.3.

Results

Twenty-five patients with phenotypic evidence of HCM defined as increased IVSd with a z-score > 2 were included in the cohort with an average age of 6.1 ± 4.5 (median: 5.8) years at echocardiographic assessment; 19 (69%) were male and 15 (60%) had a positive family history of HCM. Sixty percent of patients harbored a disease-causing mutation in the following genes: MYBPC3 (n = 8), MYH7 (n = 6), and PTPN11 (n = 1). PTPN11 in our cohort was associated with LEOPARD syndrome.

Detailed baseline echocardiographic data of our cohort is shown in Table 1; briefly, our HCM cohort had normal LV FS (44.8 ± 13.3%) and EF (67.4 ± 11.1%), normal LV LVIDd (38.5 ± 6.9 mm, z = −1.7 ± 2.3), increased IVSd (17.7 ± 7.7 mm, z = 7.1 ± 3.5), and upper limit LVPWd (9.1 ± 2.7, z = 2.0 ± 1.8), reflecting values found in a recent large cohort study with pediatric HCM.25 Impaired LV relaxation was seen by prolongation of isovolumic relaxation time and E-wave deceleration time as described previously.3

Table 1 Echocardiographic parameters of the LV in all patients investigated (left), patients with the VEGF1 963CC or CG non-risk alleles (middle) and VEGF1 963GG risk allele (right).
Full size table

Table 2 shows the frequency of SNPs for the six alleles considered, which were VEGF1 947, VEGF1 360, VEGF1 963, HIF1A.101, HIF1A.21, and HIF1A.51. Pairwise comparison using these alleles and LV global and regional systolic and diastolic peak strain phenotypes identified the allele VEGF1 963GG as a risk allele, present in 13 (52%) of our HCM patients. Table 3 shows a total of 212 pairwise comparisons in LV global and regional systolic and diastolic peak strain phenotypes using t tests for each allele. The number of significant t tests ranged from 8 to 26 depending on the genotypes. Only two of the genotypes met the binomial probability threshold for statistical significance. For the allele VEGF1 963, 26 significant t tests (CC/CG vs. GG) were statistically significant (p < 0.001), with reduced strain and strain rate values identifying VEGF1 963GG as a risk allele for systolic and diastolic mean and segmental strain and strain rate values. The difference of absolute myocardial deformation parameters between non-risk and risk allele (CC/CG vs. GG) in VEGF1 963 is shown in Table 4. For better illustration, Fig. 1 shows the significant differences between non-risk and risk allele as standardized difference.

Table 2 The frequency of the genotypes for the six alleles in the 25 patients included in the study.
Full size table
Table 3 Assessment of significance of associations between alleles and LV deformation.
Full size table
Table 4 Absolute 2-D strain and strain rate values of the LV in all patients (left) investigated, patients with the VEGF1 963CC or CG non-risk alleles (middle) and patients with VEGF1 963GG risk allele (right).
Full size table
Fig. 1: Mean and regional LV systolic and diastolic myocardial deformation parameters associated with the risk allele genotype.
figure 1

Absolute strain and strain rate values for the non-risk alleles VEGF1 963CC or CG vs. the risk allele VEGF1 963GG are presented on the right. Standardized mean differences and 95% confidence intervals in the phenotypes with significant t tests for the risk allele VEGF1 963GG are shown. The deformation parameters are classified as follows: red, mean systolic strain; blue, regional systolic strain and strain rate; green, mean diastolic strain and strain rate; purple, regional diastolic strain rate. MV mitral valve level, PM papillary muscle level, AL apical lateral, S septal, ML mid-lateral, MS mid-septal; BS basal–septal, AS antero-septal, L lateral, P posterior, I inferior, A anterior.

Full size image

In detail, between the non-risk alleles (Fig. 2) (VEGF1 963CC/CG) and the risk allele (VEGF1 963GG), the clinically most important global LV systolic myocardial function values such as mean LV longitudinal peak systolic strain (−17.7 ± 4.8% vs. −13.9 ± 4.0%), mean LV circumferential peak systolic strain, mitral valve level (−19.8 ± 6.2% vs. −15.2 ± 4.1%) and mean LV radial peak systolic strain, papillary muscle level (38.6 ± 11.5% vs. 24.9 ± 18.0%) were abnormal compared to published pediatric reference data26 and reduced to below two standardized difference in the risk allele VEGF1 963GG (Table 4 and Fig. 1). Myocardial deformation parameters of global LV diastolic function were also significantly reduced for mean LV peak diastolic longitudinal (1.6 ± 0.5% vs. 1.0 ± 0.4%) and mean LV peak diastolic circumferential, mitral level (1.6 ± 0.6% vs. 0.8 ± 0.4%), and papillary muscle level (2.0 ± 0.6% vs. 1.1 ± 0.6%) strain rates.

Fig. 2: Representative strain and strain rate curves.
figure 2

Longitudinal (top panels) and radial and circumferential strain (lower panels) showing regional differences in strain and strain rate values.

Full size image

Besides global LV function impairment, regional LV longitudinal and circumferential strain and strain rates were reduced in multiple segments in the risk allele VEGF1 963GG (Fig. 1 and Table 4). Importantly, there was no difference in standard baseline echocardiographic LV function parameters (Table 1), and although indices of diastolic function such as septal pulse wave tissue Doppler (LV med E′) and E/E′ ratio were abnormal in our HCM cohort, there was no statistical difference between patients harboring the different alleles (Table 1). For the allele HIF1A.21, four t tests (CC/CT vs. TT) were statistically significant (p = 0.035). However, given that the number of statistically significant tests is lower than what would be expected by chance alone, this association was not considered further (Table 3). Intra- and inter-observer variability for myocardial deformation analysis was good, with intra-observer ICC values between 0.82 and 0.99 and inter-observer ICC values between 0.72 and 0.95, Bland–Altman agreement assessment showed that intra- and inter-observer assessments were within the 95% confidence interval (Supplementary Table 1 and Supplementary Figs. 1 and 2). Semi-quantitative myocardial VEGF expression analysis in a subset of six samples showed weaker expression in the VEGF1 963GG risk allele, graded as 1.83 ± 0.41 in non-risk alleles vs. 2.67 ± 0.52 in the risk allele (p = 0.01).

Discussion

HCM is a disease with marked phenotypic regional heterogeneity where myocardial performance may be influenced by hypoxia-related stress signaling pathways. This study investigated if non-synonymous SNPs in the angiogenic gene VEGF and the hypoxia-response gene HIF1A are associated with altered regional systolic and diastolic myocardial performance of the LV as measured by 2-D strain. In patients with HCM, the risk allele VEGF1 963G/G identified in our previous study3 was associated in the current study with statistically significant reduced mean peak systolic longitudinal, circumferential, and radial strain and strain rate compared to non-risk alleles (p < 0.001 and p < 0.05) (Fig. 1). Whereas longitudinal strain reduction in hypertrophied regions has been described in many HCM studies, circumferential strain often remains normal or is indeed increased.10,27,28 Our study shows that the risk allele VEGF1 963G/G has a negative association also with circumferential myocardial dynamics, leading to reduced global (mean) and regional circumferential strain. Longitudinal systolic and diastolic strain impairment in the risk phenotype reflects the hypertrophy in basal and mid part of the septum as shown previously.11 However, comprehensive assessment of longitudinal, circumferential, and radial strain and strain rates in all LV segments are few, and to the best of our knowledge, no study has previously associated this detailed regional myocardial functional analysis with genes that modify the hypoxia-stress response. In our study, both longitudinal and circumferential strain impairment associated with the risk phenotype were concentrated in the basal and mid portions of the LV, with diastolic circumferential strain impairment in particular at the papillary muscle level.

In pediatric patients with HCM, 2-D deformation imaging has been shown to detect early myocardial abnormalities not only in genotype-positive patients but also in phenotype-negative patients.29 Together with risk allele stratification as shown in this study, this could evolve into a sensitive tool to detect the impact of non-sarcomeric-modifying genes on disease phenotype and inform decisions on follow-up and management in this patient group. This requires further study.

Risk alleles in VEGF, including VEGF 963G/G are associated with increased IVS thickness.30 However, the reduced strain values in the risk allele VEGF 963G/G cannot be explained by a difference in myocardial thickness of the LV alone as our previous study did not find any associations between IVS thickness and the VEGF 963G/G risk allele.3 Based on the known biology of VEGF-HIF1 signaling, the cause may be impaired angiogenesis and the associated increased fibrosis in the risk allele genotype individuals as shown in our previous study.3 The findings of our current study suggest that this pro-fibrotic phenotype subsequently results in earlier subclinical myocardial dysfunction of the LV. While knowledge on the pathophysiological mechanisms of the angiogenesis–fibrosis network is still incomplete, recent model animal studies in related disease processes suggest that perturbations in the pro-angiogenetic VEGF pathways are closely linked to fibrosis via TGFβ1,31 a key player in cardiac fibrosis.32 While it was out of the scope of this study to provide extensive cell and protein level functional data, we showed that the myocardial VEGF expression is reduced in the VEGF 963G/G risk allele. In the animal model, decreased VEGF expression is also associated with myocardial dysfunction in acquired cardiomyopathies.33 Treatment with VEGF protein has also been shown to ameliorate LV hypertrophy and decrease myocyte apoptosis in an aortic banding model,34 and this could be a further mechanism why the loss-of-function VEGF risk allele leads to reduced myocardial deformation. The findings of our study could thus be explained by increased fibrosis and reduced angiogenesis in the loss-of-function risk allele VEGF 963G/G. This is of importance as fibrosis is a risk factor for poor outcome in HCM16,35 and requires further study as we could not correlate our echo and genetic findings with tissue or magnetic resonance imaging (MRI) analysis of fibrosis.

While our study is novel in linking modifier genes to myocardial systolic and diastolic performance in relatively young patients, further longitudinal studies are needed to determine if these genetic variants are associated with progressive myocardial dysfunction and to link these findings to clinical outcomes such as arrhythmia or sudden death in HCM, to date the identified risk allele is known to be associated only to coronary artery disease36 and hypertrophy.30 Whereas our previous study showed an association of all risk alleles tested here with diastolic dysfunction, systolic and diastolic myocardial deformation impairment was only associated with the risk allele VEGF1 963G/G. Our previous analysis concentrated on parameters of global diastolic dysfunction and did not investigate myocardial deformation. While VEGF protein serum levels were recently shown to be associated with the morphological and functional HCM phenotype,37 it is yet unknown to what extent an unfavorable VEGF modifier genotype as described here can influence myocardial mechanics and our results are hence not contradictory. Future molecular functional studies will have to better quantify the effect of a loss-of-function VEGF allele on myocardial angiogenic capacity and its subsequent effect on systolic and diastolic dysfunction.

Limitations

We did not separately investigate endo-, mid-, and epicardial strain—although recent studies have identified myocardial layer-specific deformation patterns with preservation of endocardial deformation in patients with HCM,38 and modifier genes as investigated here might show a myocardial layer-specific interaction. Myocardial deformation imaging has been used in several pediatric studies to provide detailed description of myocardial function in pediatric cardiomyopathies, such as HCM,29 arrhythmogenic right ventricular cardiomyopathy,39 dilated cardiomyopathy,40 and left ventricular non-compaction cardiomyopathy41 with good intra- and inter-observer variability, and this corresponds to the satisfactory reliability data of our study. The role of strain imaging in pediatric HCM as an outcome predictor has, however, not been sufficiently explored, and at present our findings have therefore limited predictive value. While we have included VEGF expression data, a more detailed molecular functional characterization of the risk modifier allele might help explain the cause for the observed reduction in myocardial deformation, but this was beyond the scope of this study. It would have been informative to quantify myocardial extra-cellular volume and fibrosis burden using cardiac MRI, but this information was unavailable in our cohort also due to ethical concerns (e.g., general anesthesia for cardiac MRI). Additional studies are required to investigate this aspect, as well as correlation with tissue samples, although availability of tissues is limited in the pediatric population. Further studies to investigate the association of our findings with clinical outcomes, such as ventricular tachycardia and sudden death, are needed.

Conclusion

This is the first study to investigate the association between modifier gene risk alleles and LV myocardial deformation in children with HCM. It shows that early subclinical deficiencies in myocardial performance in children with HCM are associated with a previously identified risk allele in the angiogenic transcription factor VEGF. Our results describe new modifiers of myocardial function in HCM. This knowledge can potentially aid in patient risk stratification and prediction for myocardial dysfunction in patients with HCM.