Investigation of association of genetic variant rs3918242 of matrix metalloproteinase-9 with hypertension, myocardial infarction and progression of ventricular dysfunction in Irish Caucasian patients with diabetes: a report from the STOP-HF follow-up programme

The study cohort comprises Irish, Caucasian patients with diabetes and cardiovascular risk factors, but without symptomatic heart failure, from the St Vincent’s Screening TO Prevent Heart Failure (STOP-HF) follow-up programme [24]. A total of 498 patients with diabetes are included, some of whom had originally participated in the randomised, controlled prospective STOP-HF study between 2005 and 2011 [24]. Patients were referred by primary care physicians (PCPs) to the STOP-HF follow-up programme and included if they were aged over 40 years and had a history of one or more of: hypertension (medicated for ≥ 1 month); hypercholesterolemia, defined as total cholesterol (TC) greater than 5.0 mmol/L (4.5 mmol/L in high risk) and/or LDL-C 3.0 mmol/L (2.5 mmol/L in high risk) or on lipid-lowering therapy; obesity defined as body mass index (BMI) > 30 kg/m²; vascular disease including a history of angina or coronary artery disease (confirmed by angiography, or prior history of myocardial infarction), cerebrovascular disease, and peripheral vascular disease; diabetes mellitus; arrhythmia (mainly atrial fibrillation) requiring therapy; moderate to severe valvular disease. Excluded from the present analysis were those that refused to provide informed consent, those without a diagnosis of diabetes mellitus and those with a baseline diagnosis of heart failure.

The STOP-HF follow-up programme was approved by the St Vincent’s University Hospital Ethics Committee and conformed to the principles of the Declaration of Helsinki. The STOP-HF clinic nurse enrolled consecutive patients and obtained written, informed consent. Details regarding the STOP-HF programme are described elsewhere [24]. In brief, the STOP-HF follow-up programme baseline visit included clinical, cardiometabolic, medication and Doppler-echocardiography assessment (see protocol below). Cardiometabolic assessment included evaluation of blood pressure, lipids, glucose and creatinine. Biochemical analysis included evaluation of blood B-type natriuretic peptide (BNP), (Triage, Biosite). Participating patients consented to scheduled clinic follow up at intervals of 1–3 years depending on risk. At the final STOP-HF follow up programme visit, all patients underwent repeat clinical, cardiometabolic, biochemical and Doppler Echocardiography assessment. Patients and PCPs provided information on hospital visits, new diagnoses and medication changes since the previous visit. All patients underwent phlebotomy for assessment of genomic DNA and rs3918242 gene variant genotyping from leukocytes (details below). Patients, clinic personnel, investigators and primary care physicians were not informed of rs3918242 status.

Genomic DNA was extracted from venous blood (leukocytes) using a standard “spooling” method as described previously [25]. In brief, blood cells were pelleted by centrifugation (10 min, 2500 rpm), re-suspended in red cell lysis buffer (0.144 M NH₄Cl, 1 mM NaHCO₃) and incubated for 10 min at room temperature. Cells were pelleted again (10 min, 2500 rpm) and fresh red cell lysis buffer added. Following incubation and centrifugation the cell pellet was re-suspended in nuclei lysis buffer (10 mM Tris pH 7.6, 0.4 M NaCl, 2 mM EDTA Na₂ pH 8.0) supplemented with proteinase K (400 μl/mL) and SDS (0.625%) prior to incubation overnight at 37 °C. Protein was precipitated by the addition of saturated NaCl and removed by centrifugation (3000 rpm, 15 min). The supernatant was reserved and re-centrifuged a total of 3 times. Absolute ethanol was then added to the clarified supernatant to “spool” out the genomic DNA and it was dissolved in TE buffer (1 mM EDTA and 10 mM Tris–HCl, pH 8.0). Patients were genotyped for the MMP-9—1562 C/T polymorphism using a restriction fragment length polymorphism (RFLP) assay. Fidelity of the RFLP assay was verified using double stranded gBlock gene fragments (IDT, Integrated DNA Technologies) covering the sequence of interest. In brief genomic DNA (10 ng) was amplified by PCR using a Hot start QPCR master mix (SpeTaqular, Rovalab, Germany) and sequence specific primers (forward: 5′ GCCTGGCACATAGTAGGCCC 3′; reverse: 5′ CTTCCTAGCCAGCCGGCATC 3′). Amplicons were digested at 37 °C for 24 h with SphI (New England Biolabs), separated by electrophoresis on 2% agarose gels stained with GelRed, and the image captured on a Fusion Fx imaging system (Vilber Lourmat, Marne-la- Vallée Cedex, France). All ambiguous readings were repeated, and 10% of the samples were randomly selected for reanalysis to confirm genotype. All genotyping was undertaken with the operator blinded to phenotype.

Doppler-echocardiography was performed by an experienced Cardiac echocardiographer in the specialist STOP-HF unit. The evaluation was blinded to rs3918242 status as well as clinical and biochemical assessment (including BNP). A Philips IE33 ultrasound scanner with standard adult probe was used for data acquisition. All studies were performed in accordance with the American Society of Echocardiography recommendations as previously outlined [26]. Echocardiographic evaluation included: the functional measures ejection fraction (EF, Teicholz method), ratio of transmitral Doppler early filling velocity to tissue Doppler early diastolic mitral annular velocity at the lateral wall (E/e’), lateral e’; cardiac structural measures (Left atrial volume (LAV) and left ventricular mass (LVM, Devereux formula) indexed to indexed to body surface area (LAVI, LVMI respectively). Left ventricular dysfunction (LVD) was defined as either ejection fraction (EF) < 50% and/or left ventricular diastolic dysfunction (LVDD) with lateral E/e’ > 13 and/or lateral wall e’ < 9. LVD progression over the follow up period was defined as LV systolic dysfunction (LVSD) progression (follow up EF < 50% with a reduction of at least 5%) and LVDD progression (follow-up lateral E/e’ > 13, with an increase from baseline of > 2 and/or follow-up lateral e’ < 9 with a decrease from baseline of > 2). Progression of cardiac structural abnormalities was defined as follow up LAVI > 34 mL/m² with an increase from baseline > 3.5 mL/m² and/or progression of left ventricular hypertrophy (follow up LVMI > 125 g/m² / > 110 g/m² in males/females, with an increase of > 20 g/m²).

The primary endpoint of this study is the association between rs3918242 genotype and hypertension and/or myocardial infarction. The secondary endpoints are the association of rs3918242 with cardiometabolic phenotype (BMI, lipids, blood pressure, glucose and creatinine) as well as Doppler-echocardiographic measures at baseline and over follow up.

As exploratory endpoints, we also report death, major adverse cardiovascular events (MACE) requiring hospitalisation, and all cause admissions. MACE included arrhythmia, transient ischemic attack (TIA), stroke, myocardial infarction, peripheral or pulmonary thrombosis/embolus, valvular heart disease or heart failure. All clinical end points were assessed by a member of the study team by reviewing combinations of the patient report and primary care physician referral letter with hospital discharge summary. Due to the low frequency of homozygotes for the T allele, all analyses were performed assuming a dominant model (CT and TT genotypes combined).

With 498 evaluable subjects and a minor allele frequency of approximately 30% for the rs3918242 T allele, we anticipate 149 subjects with minor allele. This sample size gives 80% power, with a two-sided α = 0.05, to detect a 13% difference in hypertension and a 10% difference in myocardial infarction. For continuous secondary endpoints, this sample size also gives 80% power to detect a 0.25 standard deviation difference between groups, or a 1.9% difference in ejection fraction, a 2.5 mL/m² difference in LAVI and a 0.85 difference in E/e’ and an approximate 13% difference in progression of LVD and cardiac structural abnormalities. The study is also powered to detect a difference of 1.4 kg/m² in BMI, 5.5 mmHg SBP, and 3 mmHg DBP, which are considered within the bounds of clinical significance for these measures.

All analyses were carried out on an intention-to-treat basis using R 3.6.1 (CRAN Project, 2019) statistical programming language. Analyses were performed by a statistician employed by the St Vincent’s University Hospital and the School of Medicine in University College Dublin. The analyses of the primary and secondary endpoints included all 498 patients. Demographics and clinical characteristics were summarized with descriptive statistics (counts and percentages for categorical variables, or medians and interquartile ranges for continuous variables). Non-normal variables including BNP, serum glucose and triglycerides levels were log transformed. To test for an association between cardiovascular phenotype and genotype, categorical cardiovascular and metabolic phenotypes were initially evaluated use Pearson’s chi squared test and for characteristics of interest later using unadjusted and then age and sex adjusted logistic regression analyses with rs3918242 genotype as the explanatory variable. Associations between genotype and continuous variables confirmed to be normally distributed (using Shapiro-Wilks test) were analysed using two-tailed t-tests. If continuous variables were transformable to normal, they were power-transformed and two-tailed t-tests used. If data were not transformable to normal, the non-parametric t-test equivalents (Wilcoxon signed rank and rank sum test) were used to test differences. Linear regression analyses were used to test for associations between genotype and continuous cardiovascular and metabolic phenotypes in age and sex adjusted analyses. A nominal 2-sided significance level of 0.05 was used with Bonferroni correction for multiple testing of the primary outcomes.

The distribution of rs3918242 genotypes was CC (358, 71.9%), CT (129, 25.9%) and TT (11, 2.2%), with genotype distribution in Hardy Weinberg equilibrium. Patients who carried the minor allele (CT/TT) accounted for 28.1% of the total cohort. This is similar to several reports on rs3918242 in other European populations, but not to that of a Chinese population with subarachnoid haemorrhage, where minor allele carriers accounted for almost half of the population (Additional file 1: Table S1). The baseline demographics and medical history of the population are described in Table 1 and blood biochemistry and Doppler echocardiography at baseline are presented in Table 2. Medication usage was similar in both genotype groups (Additional file 1: Table S2) at baseline and follow up. It was also typical of the patient population with 333 (66.9%) taking RAAS modifying therapies, 332 (66.7%) taking antiplatelet agents (predominantly aspirin) and 348 (69.9%) taking statins.

The minor T allele was associated with a significantly higher baseline prevalence of myocardial infarction (MI), but not hypertension, in our diabetes population (Table 1). Using logistic regression, carriers of CT/TT genotype (CT/TT) were more than twice as likely to have suffered MI than carriers of CC genotype (CC) (OR = 2.37, [IQR 1.34, 4.21], p = 0.002) and, in adjusted analyses, this relationship remains significant (OR = 2.02, [IQR 1.13, 3.63], p = 0.018). There was no evidence of any association between the genotype and the risk of hypertension (OR = 0.94, [IQR 0.61,1.50], p = 0.77).

There were no differences in the cardiometabolic measures (SBP, DBP, heart rate, lipid, glucose, creatine and BMI measures) between both groups. While Doppler-echocardiographic analyses show no difference in the baseline prevalence of LVSD or LVDD between the genotypes, there was a small but significant reduction in left ventricular ejection fraction associated with the minor T allele (Table 2).

Furthermore, over a median follow-up of 3.5 years [IQR 2.6, 4.9 years] there was a significant difference in LVSD progression (Table 3) suggesting that diabetes patients with the minor T allele are also more at risk of progression of systolic abnormalities over the follow up period [unadjusted OR = 2.98 (1.19, 7.51), p = 0.015; adjusted OR = 2.56 (1.09, 6.01), p = 0.026]. There were no differences between the two groups in terms of LVDD, LVD or LVD and structural cardiac abnormality progression. There was also a significant difference between both groups in the proportion of patients demonstrating progression of LAVI [unadjusted OR = 1.92 (1.0, 3.70), p = 0.047], although this did not remain significant when adjusted for age and sex.

There were no differences between the groups in terms of the following clinical event endpoints: death and all-cause emergency hospitalization (CC n = 88, 24.6%, vs. CT/TT n = 36, 25.7%, OR = 1.06 [0.68, 1.66], p = NS); death or MACE or LVD progression (CC n = 104, 29.1%, vs. CT/TT n = 50, 35.7%, OR = 1.36 [0.89, 2.05], p = NS); death or MACE or LVD or structural cardiac abnormality progression (CC n = 114, 31.8%, vs. CT/TT n = 52, 37.1%, OR = 1.24 [0.82, 1.87], p = NS).

Although myocardial infarction occurred in only 11% of the total population, the majority of patients with LVSD at baseline (54%) or with LVSD progression (58%) had a history of MI. Due to the strong association between the minor T allele, myocardial infarction and reduced LVEF in our study, we carried out a post-hoc analysis of patients with a history of myocardial infarction to see if severity of cardiac damage, denoted by reduced LVEF, was greater in this subset. We compared clinical characteristics, medications and Doppler echocardiography of the patients with myocardial infarction according to CC or CT/TT genotype (Additional file 1: Tables S3–S5). Carriers of the minor allele had significantly reduced EF and the magnitude of the difference in EF was greater than the wider study population, as was baseline LVSD (n = 10, 32% vs. n = 4, 16%) and progression of LVSD (n = 8, 26% vs. n = 4, 16%), although there was insufficient power to demonstrate statistical significance. As with the overall cohort, there were no baseline differences in terms of other clinical characteristics, blood biochemistry, medications and other measures of cardiovascular remodelling and diastolic dysfunction (e.g. LAVI, LVMI, e’, E\e’).