Lack of regression of left ventricular hypertrophy is associated with higher incidence of revascularization in hypertension: The LIFE Study

Abstract

Objective. Regression of left ventricular (LV) hypertrophy and albuminuria in hypertension has previously been shown to reduce clinical cardiovascular events and death. We aimed to investigate the associations of regression of electrocardiographic (ECG) LV hypertrophy and albuminuria with the incidence of revascularization. Methods. In 9193 hypertensive patients included in the Losartan Intervention For Endpoint reduction in hypertension (LIFE) study, we measured urine albumin/creatinine ratio (UACR), LV hypertrophy by electrocardiography, serum high-density lipoprotein (HDL) cholesterol, and blood pressure after 2 weeks of placebo treatment and yearly during 5 years of anti-hypertensive treatment with either an atenolol- or a losartan-based regimen. The incidence of coronary and peripheral revascularization was recorded. Results. In Cox regression analyses adjusted for treatment allocation and continent, high time-varying Sokolow–Lyon voltage (hazard ratio [HR]=1.01 [1.00–1.02], p=0.01), but not time-varying Cornell product or UACR, predicted coronary revascularization together with low time-varying HDL-cholesterol, low time-varying pulse pressure, high Framingham risk score and history of angina pectoris. Adjusted for treatment allocation and continent, high time-varying Sokolow–Lyon voltage (HR=1.01 [1.00–1.03], p=0.02), but not time-varying Cornell product or UACR, predicted peripheral revascularization together with high time-varying pulse pressure, high Framingham risk score, history of peripheral vascular disease and prior myocardial infarction. Conclusion. Higher Sokolow–Lyon voltage during antihypertensive treatment, but not UACR or the Cornell voltage–duration product, was independently associated with higher incidence of coronary as well as peripheral revascularization.

Key Words::

• Albuminuria
• hypertension
• left ventricular hypertrophy
• revascularization

Introduction

Left ventricular hypertrophy (LVH) is a strong predictor of cardiovascular (CV) morbidity and mortality (1–3). The underlying pathophysiological mechanisms are not fully understood, but may involve systemic inflammation and atherosclerosis as well as direct myocardial alterations. The resulting increase in LV mass is associated with increased myocardial oxygen demand and reduced tolerance of myocardial ischaemia (4). This creates a vicious circle, in which adaptive changes in the hypertrophied myocardium further increase the susceptibility for ischaemic episodes. A recent study showed that concentric LVH was independently associated with coronary artery calcium, supporting that LVH is a risk factor for subclinical atherosclerosis (5). LVH is also associated with atherosclerosis in other vascular territories (6). Regression of LVH is found to reduce the risk of CV events and sudden cardiac death in a hypertensive population (7,8). Likewise, one may speculate whether regression of LVH may represent lesser progression of atherosclerosis, or maybe even reductions in established atherosclerotic disease.

Urine albumin/creatinine ratio (UACR) is associated with LVH in hypertensive (9) and healthy persons (10). Increased UACR represents endothelial damage at the glomerulus, reflecting systemic vascular dysfunction. It has been related to atherosclerosis in healthy subjects (11) as well as in hypertensive patients (12). Cottone et al. demonstrated that increased UACR was associated with endothelial activation in patients with uncomplicated essential hypertension with no clinical evidence of atherosclerosis (13). This suggests that activation of the endothelium may precede the development of atherosclerosis.

The Losartan Intervention for Endpoint reduction in hypertension (LIFE) study is a prospective study of hypertensive patients with electrocardiographic LVH randomized to treatment with losartan or atenolol. We have previously demonstrated in this population that LVH defined by Sokolow–Lyon voltage criteria predicted both coronary and peripheral revascularization (14). However, to the best of our knowledge, no studies have investigated whether reduction of LVH or UACR per se reduces the rates of revascularization. Therefore, the present study was undertaken to investigate: (i) the effect of regression of LVH and UACR on the incidences of coronary and peripheral revascularization in hypertensive patients and (ii) whether regression of LVH after completed revascularization reduced CV mortality.

Materials and Methods

Patient population

A total of 9193 hypertensive patients aged 55–80 years with previously untreated or treated essential hypertension and electrocardiographic LVH were included in the prospective, double-blind LIFE study. As described in detail elsewhere (15), patients were randomized to a losartan- or atenolol-based regimen and treated to a target blood pressure of <140/90 mmHg. In all patients, UACR, LVH by electrocardiography, serum high-density lipoprotein (HDL)-cholesterol, plasma glucose and blood pressure were measured after 2 weeks of placebo treatment and yearly during the mean 4.8 years of anti-hypertensive treatment. Patients with angina pectoris requiring treatment with a beta-blocker or a calcium antagonist were excluded.

The revascularization analyses were a pre-specified part of the LIFE protocol with revascularization being a pre-specified secondary endpoint (15). Coronary revascularization included all coronary artery revascularization procedures (angioplasty, atherectomy and stent) and heart transplant (n< 5). Peripheral revascularization included all non-coronary artery vascular surgeries and revascularization procedures (aortic aneurysm repair, carotid and peripheral revascularizations and amputations because of arterial vascular insufficiency and diabetes mellitus). Each revascularization event was reported by the investigators and verified by an independent endpoint classification committee consisting of two cardiology experts and based on definitions in a pre-defined endpoint manual. All patients gave written informed consent, and the protocol was approved by regional ethical committees.

Electrocardiography

Electrocardiograms (ECGs) were taken at baseline, at 6 months and at yearly follow-up intervals. ECGs were interpreted by observers blinded to clinical data at the Core Laboratory at Sahlgrenska University Hospital/Östra in Gothenburg, Sweden. LVH was defined as the product of QRS duration multiplied by the Cornell voltage combination (R_aVL+S_V3, with 6 mm added in women) higher than 2440 mm×ms or Sokolow–Lyon voltage (S_V1+R_V5/6) higher than 38 mm.

Renal evaluation

Albuminuria was measured as the albumin/creatinine ratio by standard methods using a turbidimetric method (Hitachi 717 Analyzer; Boehringer Mannheim, Mannheim, Germany) in a single spot urine collection on the morning of the baseline ECG and after 12 months. Both serum and urine creatinine were analysed using the Jaffè reaction without deproteinizing and then quantified by a photometric method (Hitachi 717 Analyzer). The urine albumin concentration was expressed as a ratio to urinary creatinine concentration, to provide a composite measure of renal glomerular capillary permeability adjusting for urine dilution.

Statistical analysis

Statistical analyses were performed using SPSS 12.0 (SPSS, Chicago, IL, USA) software. Continuous variables are presented as means and were compared using one-way analysis of variance (ANOVA) and Student's t-test. Categorical variables are presented as proportions and were compared using Pearson's chi-square test. Based on the distribution of residuals from the models and test for linearity and proportional hazard assumption in Cox regression analyses only UACR required logarithmic transformation. Cox regression analysis with time-varying covariates was used in order to evaluate the importance of baseline as well as in-treatment values through year five of treatment. Multiple Cox regression analyses were used to adjust for randomized treatment, continent (USA vs European countries), Framingham risk score (including age, gender, total- and HDL-cholesterol, systolic blood pressure, smoking, diabetes and LVH) (16) and different known CV diseases (previous angina pectoris, myocardial infarction, cerebral vascular disease (stroke and transient ischemic attack) and peripheral vascular disease) as well as in-treatment pulse pressure and in-treatment HDL-cholesterol. Backward selection was used until all variables but randomized treatment had p < 0.1, and only these significant variables were included in the final models. In-treatment Sokolow–Lyon voltage and Cornell product were used both as continuous as well as dichotomous variables and did not interact with age or gender. Additionally, data were analysed using baseline and 1-year ECGs to predict revascularization after one year of treatment in Figure 1 (excluding revascularization during the first year to avoid bias). In the time-varying analyses, however, all revascularization events were included. In Pearson's chi-square analyses of CV mortality after revascularization, patients were divided according to LVH presence on baseline ECG (Figure 2). Two-tailed p< 0.05 gave statistical significance.

Figure 1. LVH status in relation to incidences of coronary and peripheral revascularization: (a) LVH defined by Sokolow–Lyon voltage criteria, (b) LVH defined by Cornell voltage–duration product criteria.

Figure 2. Baseline LVH status in relation to cardiovascular (CV) mortality after coronary, peripheral and total revascularization: (a) LVH defined by Sokolow–Lyon voltage criteria, (b) LVH defined by Cornell voltage–duration product criteria.

Results

During the mean 4.8 years of follow-up, 337 patients (3.7%) underwent coronary revascularization and 231 patients (2.5%) underwent peripheral revascularization. Background variables for these patients have been presented previously (14). We divided the LIFE population into three groups based on LVH by Sokolow–Lyon voltage criteria (Table I). The intermediate group (LVH at baseline or at 1 year) was included to differentiate clearly between the other two groups. Because of measurement variation, regression to the mean and treatment, some patients were moving back and forth between LVH groups during the first year of the study. Patients with LVH defined by Sokolow–Lyon voltage at baseline and 1 year had significantly higher systolic blood pressure, pulse pressure, creatinine and Framingham risk score compared with patients without LVH at either time. However, patients with LVH by this index also had less abnormal metabolic risk factors [higher HDL-cholesterol and lower body mass index (BMI), total cholesterol and serum glucose].

Table I. Baseline characteristics in patients according to LVH status by Sokolow–Lyon voltage criteria.

LVH by Sokolow– Lyon:	No LVH at baseline nor at 1 year	LVH at baseline or at 1 year	LVH at baseline and at 1 year	p-value (ANOVA)
Number of patients	5860	1028	945
Age (years)	67	67	67*	0.03
Systolic blood pressure (mmHg)	174	176	176***	<0.001
Diastolic blood pressure (mmHg)	98	98	97***	<0.001
Pulse pressure (mmHg)	76	79	79***	<0.001
Pulse (beats/min)	74	72	72***	<0.001
Glucose (mmol/l)	6.1	5.8	5.8**	<0.001
Total cholesterol (mmol/l)	6.1	5.9	5.9***	<0.001
HDL-cholesterol (mmol/l)	1.48	1.53	1.53**	0.001
Serum creatinine (μmol/l)	85	89	91***	<0.001
Body mass index (kg/m^2)	29	26	26***	<0.001
Framingham risk score	22	23	24***	<0.001
Cornell voltage–duration product (100 mm/ms)	29	26	25***	<0.001
Sokolow–Lyon voltage (mm)	25	40	47***	<0.001
UACR (mg/mmol)	6.3	6.9	9.8**	0.002
Log UACR	−0.16	−0.07	0.01	0.01**

Data presented are means. p-values for analysis of variance (ANOVA) comparing all three groups are presented in the far right column. p-values for Student's t-test between group 1 (no LVH at baseline or at 1 year) and group 3 (LVH at baseline and at 1 year) are presented with: *p<0.05, p<0.01, ***p<0.001. LVH, left ventricular hypertrophy; HDL, high-density lipoprotein; UACR, urine albumin/creatinine ratio.

Patients with LVH at baseline and/or year-1 defined by Sokolow–Lyon voltage showed trends to more frequent revascularization, as illustrated by Figure 1(a). Peripheral revascularization was significantly different between groups (p=0.01). Coronary revascularization tended to occur more frequently in patients with increased LVH by Sokolow–Lyon voltage, although this trend did not attain statistical significance (p=0.13). In contrast, there was a trend of less frequent revascularization in patients with LVH at baseline and/or year-1 defined by Cornell voltage–duration product (Figure 1b).

In Cox regression analyses adjusted for treatment allocation and continent, coronary revascularization was predicted by high time-varying Sokolow–Lyon voltage and low time-varying pulse pressure, together with low time-varying HDL-cholesterol, high Framingham risk score and history of angina pectoris (Table II). Time-varying Cornell product and time-varying UACR did not reach statistical significance. Peripheral revascularization was, after adjustment for treatment allocation and continent, predicted by high time-varying Sokolow–Lyon voltage, but not time-varying UACR or Cornell product, together with high time-varying pulse pressure, high Framingham risk score, history of peripheral vascular disease and prior myocardial infarction (Table II).

Table II. Multivariate Cox regression models predicting revascularization in the whole group.

	HR	95% CI	p-value	HR	95% CI	p-value
a) Coronary revascularization
Time-varying Sokolow–Lyon voltage (mm)	1.01	1.00–1.02	0.01
LVH by Sokolow–Lyon voltage (categorical)				1.32	0.99–1.75	0.06
Time-varying HDL-cholesterol (mmol/l)	0.55	0.37–0.81	0.003	0.56	0.38–0.82	0.003
Time-varying pulse pressure (mmHg)	0.93	0.86–1.00	0.04	0.93	0.87–1.00	0.06
Treatment, losartan vs atenolol	1.02	0.82–1.26	0.89	1.00	0.81–1.24	0.97
Angina pectoris	2.67	2.06–3.46	<0.001	2.70	2.09–3.51	<0.001
Framingham risk score	1.04	1.03–1.05	<0.001	1.04	1.03–1.05	<0.001
Myocardial infarction	1.35	0.97–1.87	0.07	1.33	0.96–1.85	0.08
Continent, USA vs Europe	1.89	1.49–2.39	<0.001	1.89	1.49–2.39	<0.001
b) Peripheral revascularization
Time-varying Sokolow–Lyon voltage (mm)	1.01	1.00–1.03	0.02
LVH by Sokolow–Lyon voltage (categorical)				1.18	0.84–1.65	0.33
Time-varying pulse pressure (mmHg)	1.14	1.05–1.22	<0.001	1.15	1.07–1.24	<0.001
Treatment, losartan vs atenolol	0.82	0.63–1.07	0.14	0.81	0.63–1.06	0.12
Framingham risk score	1.05	1.04–1.07	<0.001	1.05	1.04–1.07	<0.001
Myocardial infarction	1.50	1.03–2.18	0.03	1.48	1.02–2.15	0.04
Peripheral vascular disease	5.05	3.74–6.83	<0.001	5.06	3.74–6.85	<0.001
Continent, USA vs Europe	2.11	1.61–2.78	<0.001	2.11	1.60–2.78	<0.001

Cox regression models with time-varying Sokolow–Lyon as a continuous variable are presented in the left column, and the same models with LVH defined by Sokolow–Lyon voltage as a categorical variable (yes/no) are presented in the right column. LVH, left ventricular hypertrophy; HDL, high-density lipoprotein; HR, hazard ratio; 95% CI, 95% confidence interval. Only significant variables are included in the models.

In the right column of Table II, the same multivariate analyses are presented using LVH defined by Sokolow–Lyon voltage as a dichotomous variable. LVH by Sokolow–Lyon voltage was a borderline significant predictor of coronary revascularization together with low time-varying HDL-cholesterol, high Framingham risk score and history of angina pectoris. Regarding peripheral revascularization, LVH by Sokolow–Lyon voltage was not a significant predictor in multivariate analysis (Table II).

Of the 568 patients in the LIFE study who underwent revascularization, 46 patients died of CV disease within the study period. Figure 2 (a) and (b) show the relations between CV mortality after revascularization and LVH at baseline by Sokolow–Lyon voltage and Cornell voltage–duration product, respectively. Using Sokolow–Lyon voltage criteria, there was no significant association between LVH status and CV mortality after coronary or peripheral revascularization. However, LVH by Cornell product was associated with higher CV mortality after coronary or peripheral revascularization (p=0.04). In Cox regression analyses that adjusted for treatment allocation and continent, LVH by Cornell voltage–duration product, but not LVH by Sokolow–Lyon voltage predicted CV mortality, together with Framingham risk score and prevalent cerebral vascular disease (Table III).

Table III. Multivariate Cox regression model predicting cardiovascular mortality after revascularization.

	HR	95% CI	p-value
Cornell voltage–duration product (100 mm/ms)	1.02	1.01–1.03	<0.001
Treatment, losartan vs atenolol	0.88	0.73–1.06	0.19
Framingham risk score	1.05	1.04–1.06	<0.001
Cerebral vascular disease	2.48	1.95–3.15	<0.001
Continent, USA vs Europe	1.47	1.18–1.83	<0.001

HR, hazard ratio; 95% CI, 95% confidence interval. Only significant variables are included in the model.

Discussion

The present study demonstrates that lack of regression of LVH, but not UACR, during anti-hypertensive treatment is independently associated with increased incidence of coronary and peripheral revascularization in a large hypertensive population. We can only speculate about the causal relationships, but the associations between revascularization and time-varying Sokolow–Lyon voltage were independent of time-varying pulse pressure indicating that the associations were not just related to less blood pressure reduction in patients with lack of LVH regression. A likely possibility could be that patients with persisting LVH are more likely to experience angina because of more myocardial tissue per vessel leading to angiography and in some cases coronary revascularization. Further, it may be likely that patients with persistent high LVH in parallel have more persistent peripheral structural vascular hypertrophy leading to more peripheral revascularization.

It was surprising that regression of UACR had no effect on coronary revascularization. Several data link albuminuria to CV complications and a few studies have specifically examined its relationship to atherosclerosis, concluding that albuminuria is directly correlated with angiographic evidence of coronary artery disease (17,18). The lack of association between UACR and coronary revascularization in our study might reflect that patients without reduction in UACR have more generalized atherosclerotic disease and are therefore not candidates for revascularization. Yet, we cannot exclude the possibility of a potential statistical type II error, even though our material includes over 500 events. Few studies have addressed the risk of peripheral artery disease in patients with microalbuminuria. The PREVEND study showed that microalbuminuria was independently associated with coronary heart disease, but not peripheral artery disease in the general population (19). In contrast, C-reactive protein (CRP) was related to damage in all vascular beds, suggesting that different risk factors may play different roles in various types of vasculature.

In the present study, we found that persistently high Sokolow–Lyon voltage predicted coronary and peripheral revascularization, whereas time-varying Cornell voltage–duration product did not. This may be related to different patient characteristics at baseline according to LVH classification by Sokolow–Lyon voltage or Cornell product. Patients with LVH defined by Sokolow–Lyon voltage were younger and leaner and had higher pulse pressure and HDL-cholesterol compared with patients without LVH by this method (20). In addition, they were more likely to be men and to smoke, both of which would tend to lead to greater clinical suspicion of peripheral arterial disease. Patients with LVH defined by Cornell product were older and had higher BMI and total cholesterol and a trend toward higher serum glucose compared with patients without LVH by this method. Consequently, one may speculate that patients recruited by the Cornell criteria were more likely to develop metabolic disorders with increased risk for surgical complications, possibly making them less suitable for surgery. Patients with Sokolow–Lyon with less co-morbidity may on the other hand be more likely to be offered revascularization.

One of the most important determinants of LV mass is body weight (21), as increased body weight may increase heart size because of the need for higher cardiac output. Weight loss has been associated with a decrease in LV mass among hypertensives in several studies (22,23). During the mean 4.8 years of follow-up in the LIFE study, the study population experienced a small increase in weight (unpublished data). Consequently, our findings of LVH regression cannot be explained by changes in body weight.

Generally, the clinical judgments behind revascularization procedures are based on three factors: patient symptoms, patients’ co-morbidities and lastly, local medical traditions. As expected, we found that the frequency of both coronary and peripheral revascularization was higher in USA independently of traditional CV risk factors. To account for the latter, we adjusted our analyses for continent (USA vs European countries). If there hypothetically indeed were done revascularization procedures based on non-medical decisions in USA, this would only possibly dilute our results, and not strengthen the associations.

We have previously demonstrated that the incidence of peripheral revascularization in patients without peripheral vascular disease or albuminuria at baseline was lower in patients randomized to losartan vs atenolol-based antihypertensive treatment (14). The mechanism for this beneficial effect of losartan was not clear. Studies of hypercholesterolaemic monkeys have shown that losartan was associated with regression of atherosclerotic lesions in the aorta, independent of effects on blood pressure (24,25). However, there is yet no direct evidence that inhibition of the renin–angiotensin–aldosterone system (RAAS) slows or reverses the progression of atherosclerosis in humans. Theoretically, in our population, atenolol might improve angina pectoris and worsen peripheral artery disease symptoms compared with losartan. As the present study was undertaken to evaluate the effect of regression of LVH and UACR per se, independent of medication effect, we have adjusted the analyses for study drug treatment.

After revascularization (either coronary or peripheral), we found that continuing LVH by Cornell voltage–duration product was a significant predictor of CV mortality. This would suggest, if persistently complete coronary revascularization had been produced by the interventions, that LVH assessed by Cornell product is associated with increased CV mortality by other mechanisms than the coronary atherosclerosis known to be associated with LVH. Whether this is related to LVH by itself or to the association of LVH with systemic atherosclerosis (6) or other associated risk factors is unclear. LVH by Sokolow–Lyon criteria did not reach statistical significance, potentially related to loss of power because of the lower prevalence of LVH by this criterion than by Cornell product criteria, or other differences in risk profile.

Several studies have shown that patients with LVH have increased in-hospital mortality after coronary artery bypass graft (CABG) surgery (26,27). Recently, the first data assessing long-term mortality risk postoperatively were published, revealing that ECG LVH was a risk factor for all-cause late mortality, especially after the third postoperative year (28). Lauer et al. followed 8166 patients for a mean 6 years after CABG surgery and evaluated quantitative ECG measures. One of the independent predictors of long-term all-cause mortality was Sokolow–Lyon voltage in the first or fourth quartile (29). Cornell voltage showed a continuous risk pattern with increasing risk at increasing values. From these data together with our study, we can conclude that LVH regression should be an independent goal in the future therapy and follow-up after revascularization procedures in order to reduce the CV risk for these patients.

Limitations

Some limitations of our study must be taken into account. UACR was calculated based on a single spot urine collection. However, other studies have shown a close correlation between this ratio and measures of albuminuria from 24-h collection (30). Patients with angina pectoris requiring treatment with a beta-blocker or a calcium antagonist, heart failure, known left ventricular ejection fraction <40% or recent myocardial infarction within the past 6 months were excluded from the LIFE study; hence our results may not be representative of the typical hypertensive population undergoing revascularization. Finally, our analyses cannot exclude the effect of potential, yet unknown, confounders that may cause microalbuminuria or LVH and require revascularization. As the LIFE study was not designed to investigate causal relationships between baseline predictors and revascularization, we can only draw correlational conclusions.

Conclusions

Continuing high Sokolow–Lyon voltage was, in contrast to UACR, associated with higher incidence of coronary as well as peripheral revascularization independent of prior CV disease and traditional CV risk factors including time-varying pulse pressure and HDL-cholesterol. After revascularization, LVH by Cornell voltage–duration product predicted CV mortality together with Framingham risk score and cerebral vascular disease.

Acknowledgements

This work was supported by Merck & Co., Inc. Drs Wachtell, Okin, Dahlöf, Devereux, Kjeldsen and Olsen have received grant support from Merck & Co, Inc., the sponsor of the LIFE study.

Declaration of interest: The authors alone are responsible for the content and writing of the paper.