Obesity is related to left ventricular hypertrophy (LVH). Whether LVH on electrocardiography (ECG-LVH) is a result of increased cardiac electrical activity or due to increased left ventricular mass (LVM) remains to be determined. The aims of the present study were to investigate the relation between obesity and ECG-LVH and LVM by magnetic resonance imaging (MRI-LVM) in patients with hypertension and to investigate the relation of insulin resistance (IR) and LVH. Patients with hypertension (n = 421) were evaluated using Sokolow-Lyon voltage, Cornell voltage, and cardiac magnetic resonance imaging. Waist circumference was used as a measure of abdominal obesity. Linear regression analysis revealed an inverse relation (adjusted β = −0.02, 95% confidence interval −0.02 to −0.01) between waist circumference and Sokolow-Lyon voltage, indicating a decrease of 0.02 mV per 1-cm increase in waist circumference. There was a positive relation between waist circumference and MRI-LVM (β = 0.49, 95% confidence interval 0.32 to 0.67). Patients in the highest quartile of LVM had a worse metabolic profile than patients with the Sokolow-Lyon voltage criterion. The relations of IR with ECG-LVH and MRI-LVM were similar to those of waist circumference in relation to ECG-LVH and MRI-LVM. In conclusion, there is an inverse relation between waist circumference and ECG-LVH and a positive relation between waist circumference and MRI-LVM. This study indicates that obesity has a different relation to voltage criteria for LVH compared to anatomic criteria for LVH, supporting the hypothesis that IR decreases electrocardiographic voltages, despite an increase in MRI-LVM. The clinical implication is that especially in patients with IR, Sokolow-Lyon voltage is low in contrast to high MRI-LVM.
Insulin acts as a growth factor in many tissues, including the heart, leading to hypertrophy. In an insulin resistance (IR) state, plasma concentrations of insulin are usually elevated, together with changes in plasma adipokines, inflammation, and the thrombotic system, which may all influence cardiac tissue histology. Hypertension is more prevalent in obesity and in IR states and is a key component in the clustering of vascular risk factors closely associated with abdominal obesity, which is also known to be associated with left ventricular hypertrophy (LVH). Myocardial fatty acid oxidation, as measured using cardiac positron emission tomography, is decreased in patients with LVH, pointing to a changed myocardial fatty acid metabolism, as seen with IR in other organs. Therefore, it is likely that IR contributes to the changes in cardiac tissue seen in LVH. Our hypothesis is that abdominal obesity has a different relation to electrical criteria for LVH compared to anatomic criteria for LVH in patients with hypertension. In the present study, we evaluated the relation between electrical (electrocardiographic) and anatomic (magnetic resonance imaging [MRI]) LVH and obesity as measured by waist circumference, as well as the relations between LVH and visceral fat and between LVH and IR.
Methods
Patients originated from the Second Manifestations of Arterial Diseases (SMART) cohort. The SMART cohort consists of patients referred to the University Medical Center Utrecht with clinically manifest cardiovascular disease (cardiac, cerebral, and peripheral vascular disease or aneurysm of the abdominal aorta) or marked risk factors for cardiovascular disorders (e.g., hypertension, diabetes mellitus, dyslipidemia). All patients underwent extensive screening for cardiovascular risk factors at inclusion. The local medical ethics committee approved the SMART study, and all subjects gave their written informed consent. For this study, we randomly selected 1,273 subjects from the SMART population who had hypertension for ≥3 years and who were free from previous symptomatic coronary or valvular heart disease. Subjects with severe concomitant illness (n = 8) or contraindications to MRI (including claustrophobia; n = 34) were excluded. Five hundred thirty-six subjects agreed to participate.
For this analysis, patients were included in the period from January 1999 to July 2006 with routine measurement of waist circumference. Exclusion followed in case of missing crucial parameters (electrocardiography, n = 32; waist circumference, n = 61) or an electrocardiogram on which LVH could not be evaluated because of bundle branch block (QRS duration >120 ms; n = 22). This left a study population of 421 patients. Measurement of visceral fat was available from May 2000 onward (n = 346) and measurement of fasting insulin from July 2003. Patients with type 2 diabetes mellitus were excluded, because homeostasis model assessment of IR is not reliable in these patients. In this subset of patients with available data on fasting insulin levels, 117 patients were evaluated.
Waist circumference was measured halfway between the lower rib and iliac crest and was taken in standing position. A well-trained registered vascular technologist in a certified vascular laboratory performed ultrasonographic measurements (without bowel preparation) in the supine position using an ATL HDI 3000 (Philips Medical Systems, Eindhoven, The Netherlands) with a C 4-2 transducer. Intra-abdominal visceral fat was measured as the distance between the peritoneum and the lumbar spine or psoas muscles using electronic calipers at the end of a quiet inspiration, applying minimal pressure without displacement or compression of the abdominal cavity.
A 12-lead electrocardiogram was recorded at 25 mm/s and 1 mV/cm calibration. QRS duration was calculated by the electrocardiograph according to a standardized formula. Two independent readers manually obtained the voltage amplitude of the R waves and S deflections used for the LVH criteria, as well as the cardiac rhythm and T-top flattening or inversion. Differences were discussed and corrected. LVH on electrocardiography (ECG-LVH) was determined using Sokolow-Lyon voltage (S[V 1 ] + R[V 5 or V 6 , whichever is larger]) and the Sokolow-Lyon voltage criterion with a cut-off value of ≥3.5 mV ; the Sokolow-Lyon voltage product is calculated as the Sokolow-Lyon voltage times QRS duration (milliseconds). Cornell voltage was calculated as S(V 3 ) + R(aVL) and the Cornell voltage criterion with cut-off values of >2.8 mV in men and >2.0 mV in women. The Cornell voltage product was calculated as the Cornell voltage times QRS duration (milliseconds).
Plasma glucose, insulin, lipids, and creatinine, as well as urine albumin and creatinine, were obtained after an overnight fast. Laboratory assessment of insulin was performed using immunometric assay (Diagnostic Products Corporation, Los Angeles, California).
Cardiac MRI was done with a 1.5-T Philips Achieva MRI scanner (Philips Medical Systems, Best, The Netherlands) using a steady-state free precession pulse sequence for optimal blood myocardium contrast. These measurements were performed on serial short-axis views, using off-line dedicated software (ViewForum; Philips Medical Systems), as described previously. The reproducibility of the cardiac MRI readings was determined by calculating in a set of 50 duplicate readings the intraclass correlation coefficient. For intraobserver errors in left ventricular mass (LVM) the intraclass correlation coefficient was 99%, and for interobserver errors in LVM, it was 97%. For intraobserver errors in LVM, the mean difference was 1.1 ± 2.1 g, and for interobserver errors, it was 10.3 ± 5.8 g. With MRI, left ventricular end-diastolic volume was computed using a modified Simpson’s rule algorithm. LVM on MRI (MRI-LVM) was calculated as 1.05 × (epicardial volume − endocardial volume), as described earlier. MRI-LVM was used as a continuous variable. There is a body of research on partition values for LVH with MRI, and values are influenced by differences in technique. In studies using cardiac echocardiography, LVH was seen in 15% to 50% of different populations, with a prevalence of about 25% in a hypertensive population. Therefore, dichotomization was done by using the gender-pooled upper quartile as LVH on MRI. LVM index was calculated as LVM (grams)/height (meters) 2.7 , and homeostasis model assessment IR was calculated as fasting plasma insulin (milliunits per liter) × fasting plasma glucose (millimoles per liter)/22.5.
Analysis of the relation between waist circumference and LVH was done using linear regression analysis with SPSS version 18.0.2 for Windows (SPSS, Inc., Chicago, Illinois). Confounders in the relation between waist circumference and LVH and LVM were age, gender, blood pressure, and the use of β blockers and/or angiotensin-converting enzyme inhibitors and/or angiotensin II receptor blockers. Models were created with age and gender and with all confounders. A third explorative model was added with body mass index as confounder to determine the change in relation of abdominal obesity and LVH when adding another measurement of obesity.
Results
Most patients had hypertension (91%), with 69% using blood pressure–lowering medications, and 17% of the patients had albuminuria. Baseline measurements were divided in gender-pooled tertiles of waist circumference showing increasing glucose, triglycerides, and prevalence of type 2 diabetes mellitus but no change in blood pressure or medication use per increasing tertile of waist circumference ( Table 1 ). Per increasing tertile of waist circumference, Sokolow-Lyon voltage and the percentage of patients with positive Sokolow-Lyon voltage criterion were lower ( Table 2 ), whereas Cornell voltage and the percentage of positive Cornell voltage criterion were comparable per increasing tertile of waist circumference. MRI-LVM appeared to increase per increasing tertile of waist circumference.
| Variable | Waist Circumference Tertile ⁎ | ||
|---|---|---|---|
| 1 (n = 139) | 2 (n = 145) | 3 (n = 137) | |
| Waist circumference (cm) | 82 ± 8 | 93 ± 7 | 109 ± 9 |
| Waist circumference range (cm) | 65–93 | 80–103 | 93–139 |
| Age (years) | 52 ± 13 | 53 ± 12 | 52 ± 11 |
| Men | 83 (60%) | 81 (56%) | 80 (58%) |
| Systolic blood pressure (mm Hg) | 150 ± 19 | 149 ± 21 | 149 ± 20 |
| Diastolic blood pressure (mm Hg) | 89 ± 11 | 90 ± 12 | 92 ± 11 |
| Body mass index (kg/m 2 ) | 23.9 ± 2.9 | 26.6 ± 2.6 | 32.1 ± 4.7 |
| Visceral fat (cm) | 7.2 ± 1.9 | 8.4 ± 2.2 | 11.2 ± 2.8 |
| Total cholesterol (mmol/L) | 5.5 ± 1.2 | 5.6 ± 1.3 | 5.4 ± 1.2 |
| High-density lipoprotein cholesterol (mmol/L) | 1.5 ± 0.5 | 1.3 ± 0.4 | 1.2 ± 0.4 |
| Triglycerides (mmol/L) | 1.4 (1.0–2.0) | 1.5 (1.1–2.6) | 2.0 (1.4–2.9) |
| Low-density lipoprotein cholesterol (mmol/L) | 3.3 ± 1.2 | 3.4 ± 1.0 | 3.1 ± 1.1 |
| Glucose (mmol/L) | 5.4 (5.1–5.9) | 5.7 (5.2–6.2) | 5.9 (5.5–7.5) |
| Insulin (mU/L) | 9 ± 8 | 11 ± 8 | 19 ± 17 |
| Homeostasis model assessment of IR | 2.1 ± 1.8 | 3.1 ± 3.0 | 6.0 ± 7.4 |
| Estimated glomerular filtration rate (ml/min/1.73 m 2 ) | 79 ± 17 | 77 ± 17 | 82 ± 19 |
| Albuminuria | |||
| Microalbuminuria | 14 (10%) | 16 (11%) | 27 (20%) |
| Macroalbuminuria | 4 (3%) | 2 (1%) | 8 (6%) |
| Cerebrovascular disease | 35 (25%) | 48 (33%) | 42 (31%) |
| Peripheral arterial disease | 9 (6%) | 10 (7%) | 11 (8%) |
| Type 2 diabetes mellitus | 11 (8%) | 25 (17%) | 36 (26%) |
| Hypertension | 122 (88%) | 131 (90%) | 130 (95%) |
| Smoking (current or previous) | 92 (66%) | 106 (73%) | 97 (71%) |
| β blockers | 40 (29%) | 45 (31%) | 42 (31%) |
| Angiotensin-converting enzyme inhibitors or angiotensin II receptor blocker | 60 (43%) | 73 (50%) | 72 (53%) |
| Diuretics | 19 (14%) | 18 (12%) | 25 (18%) |
| Other blood pressure–lowering medication | 41 (29%) | 36 (25%) | 51 (37%) |
| Lipid-lowering medication | 47 (34%) | 46 (32%) | 54 (39%) |
| Platelet aggregation inhibitors | 47 (34%) | 53 (37%) | 50 (36%) |
| Variable | Waist Circumference Tertile ⁎ | ||
|---|---|---|---|
| 1 | 2 | 3 | |
| Electrocardiography | |||
| Sokolow-Lyon voltage criterion | 23 (17%) | 11 (8%) | 7 (5%) |
| Sokolow-Lyon voltage (mV) | 2.7 ± 0.9 | 2.5 ± 0.7 | 2.2 ± 0.8 |
| Cornell voltage criterion | 8 (6%) | 13 (9%) | 10 (7%) |
| Cornell voltage (mV) | 1.5 ± 0.6 | 1.6 ± 0.6 | 1.5 ± 0.5 |
| MRI | |||
| LVM (g) | 101 ± 27 | 104 ± 28 | 114 ± 29 |
| Upper quartile of LVM ⁎ | 21 (15%) | 35 (24%) | 54 (39%) |
| LVM index (g/m 2.7 ) | 22 ± 5 | 23 ± 5 | 26 ± 6 |
Waist circumference showed an inverse relation with Sokolow-Lyon voltage ( Table 3 ). A 1-cm increase in waist circumference was associated with a decrease in Sokolow-Lyon voltage of 0.02 mV (0.2 mm) (95% confidence interval [CI] −0.02 to −0.01 mV). The relation of waist circumference with Sokolow-Lyon voltage product was inverse as well. In contrast, the relation between waist circumference and MRI-LVM was positive. Every 1-cm increase in waist circumference was associated with an increase of 0.49 g in MRI-LVM (95% CI 0.32 to 0.67). No clear relation was found for waist circumference with Cornell voltage and with Cornell voltage product. When adding body mass index to the model, the point estimates did not change. The relation between visceral fat and ECG-LVH criteria and MRI-LVM were in general comparable to the results of the analyses with waist circumference ( Table 3 ).
| Variable | β (95% Confidence Interval) | ||
|---|---|---|---|
| Model 1 | Model 2 | Model 3 | |
| Relation between waist circumference and LVH (n = 421) | |||
| Electrocardiography | |||
| Sokolow-Lyon voltage (mV) | −0.02 (−0.02 to −0.01) | −0.02 (−0.02 to −0.01) | −0.01 (−0.02 to 0.00) |
| Sokolow-Lyon product (mV · ms) | −1.66 (−2.26 to −1.06) | −1.69 (−2.27 to −1.12) | −1.13 (−2.08 to −0.19) |
| Cornell voltage (mV) | −0.001 (−0.005 to 0.003) | −0.001 (−0.005 to 0.003) | −0.004 (−0.010 to 0.003) |
| Cornell product (mV · ms) | −0.22 (−0.67 to 0.23) | −0.26 (−0.71 to 0.18) | −0.53 (−1.25 to 0.20) |
| MRI | |||
| LVM (g) | 0.49 (0.32 to 0.67) | 0.50 (0.33 to 0.67) | 0.17 (−0.11 to 0.45) |
| LVM index (g/m 2.7 ) | 0.11 (0.07 to 0.15) | 0.11 (0.07 to 0.15) | −0.04 (−0.10 to 0.02) |
| Relation between visceral fat ⁎ and LVH (n = 346) | |||
| Electrocardiography | |||
| Sokolow-Lyon voltage (mV) | −0.06 (−0.09 to −0.03) | −0.06 (−0.09 to −0.03) | −0.02 (−0.06 to 0.01) |
| Sokolow-Lyon product (mV · ms) | −6.27 (−9.57 to −2.97) | −6.50 (−9.68 to −3.32) | −3.15 (−7.14 to 0.85) |
| Cornell voltage (mV) | 0.007 (−0.016 to 0.030) | 0.007 (−0.016 to 0.029) | 0.018 (−0.010 to 0.047) |
| Cornell product (mV · ms) | −0.17 (−2.58 to 2.25) | −0.18 (−2.57 to 2.21) | 0.83 (−2.20 to 3.85) |
| MRI | |||
| LVM (g) | 1.57 (0.62 to 2.51) | 1.63 (0.68 to 2.58) | 0.00 (−1.17 to 1.17) |
| LVM index (g/m 2.7 ) | 0.45 (0.23 to 0.67) | 0.46 (0.24 to 0.69) | 0.01 (−0.26 to 0.28) |
⁎ Intra-abdominal visceral fat was measured in centimeters by abdominal echocardiography.
To better understand the discrepant findings with different ECG-LVH criteria, we evaluated the distribution of the ECG-LVH criteria ( Figure 1 ). The Venn diagram illustrate that there was only limited overlap between patients with positive Sokolow-Lyon voltage criterion and those with positive Cornell voltage criterion. Of the 62 patients with positive electrocardiographic criteria, only 10 patients had both criteria positive (15%). Similarly, the overlap between ECG-LVH criteria and high MRI-LVM (gender-pooled upper quartile LVM) was also limited. In this study, 141 patients had positive criteria on either electrocardiography or MRI, and only 4 patients (2%) had all criteria positive. When we calculated the correlation between the 2 ECG-LVH measures and MRI-LVM, we found Pearson’s correlation coefficients of 0.23 (Sokolow-Lyon voltage) and 0.17 (Cornell voltage). The differences in patient characteristics per ECG-LVH criterion are shown in Table 4 . Patients with high MRI-LVM compared to patients with positive Sokolow-Lyon voltage criterion had a worse metabolic profile, with higher body weight, higher triglycerides, and more organ damage shown in albuminuria and medical history. Patients with positive Cornell voltage criterion were between those extremes.
