Left ventricular hypertrophy detection and body mass index in essential hypertension.

Background and aim. Weight and height are two major determinants of left ventricular mass (LVM); the anthropometric parameter to which LVM should be normalized remains, however, debated. In a population of hypertensives, we compared the prevalence of left ventricular hypertrophy (LVH) defined by two indexation criteria of LVM in different subgroups of body mass index (BMI). Methods. A total of 4468 essential hypertensives included in the Evaluation of Target Organ Damage in Hypertension (ETODH), were divided in four groups according to BMI thresholds: lean (BMI<20 kg/m², 4.5%), normal (20–24.9 kg/m², 36.5%), overweight (25–29.9 kg/m², 41.9%) and obese (≥30 kg/m², 17.1%). All patients underwent quantitative echocardiography; LVH was defined by two criteria of LVM indexation: (A) ≥116 g/m² in men and ≥96 g/m² in women; (B) ≥49 g/m^2.7 in men and ≥45 g/m^2.7 in women. Results. Overall, 44.9% of the patients were found to have LVH by criterion A, 48.2% by criterion B and 37.0% by both criteria. Prevalence rates of LVH in the four BMI groups were 34.3%, 40.5%, 47.3%, 53.9% by criterion A, 19.8%, 37.0%, 53.6%, 69.7% by criterion B, and 14.2%, 30.9%, 41.5%, 47.8% by both criteria, respectively (p at least <0.05 for all). Conclusions. Our findings show that LVH prevalence in both overweight and obese hypertensives is higher when LVM is normalized to height^2.7 compared with body surface area (BSA), whereas the opposite trend occurs in normal weight/lean hypertensives. Thus, the risk related to LVH is underestimated when the LVH/height^2.7 criterion is applied to lean/normal weight individuals and the LVH/BSA criterion in overweight/obese individuals.

Keywords: Body size; indexation methods; left ventricular hypertrophy

Left ventricular hypertrophy (LVH) is a powerful, independent risk factor for all-cause mortality and cardiovascular events in the general population ([1]) as well as in different clinical settings, including uncomplicated hypertension ([2]), coronary artery disease ([3],[4]) and chronic kidney disease ([[5]]).

LVH regression during long-term effective antihypertensive treatment ([8],[9]) and, more importantly, a strong association between LVH reversal and improved prognosis have been documented by numerous studies and meta-analyses ([10],[11]).

The search for LVH by the electrocardiographic or the more accurate echocardiographic approach is recommended by hypertension guidelines in the initial evaluation as well as in the follow-up of patients ([[12]]).

LVH phenotype, indeed, is useful to identify individuals particularly exposed to the adverse effects of high BP and to refine cardiovascular risk stratification ([15]).

In epidemiological studies and in clinical trials echocardiographic LVH has been defined by left ventricular mass (LVM) estimated according to validated algorithms ([16]) and indexed to various body size parameters [i.e. body surface area (BSA), height and height to allometric power of 2.7]. The best method, however, for normalizing LVM in adults are still debated ([17]).

In the present study, we sought to compare the prevalence of LVH defined by two gender-specific criteria, namely LVM indexed to BSA and to height2.7, according to the Lang's intersociety report ([18]) in a large cohort of hypertensive subjects divided into four groups according to BMI (lean, normal, overweight and obese). Our aim was to test the consistency of these criteria in detecting hypertensive LVH within a wide range of body size.

This analysis was performed on data from the Evaluation Target Organ Damage in Hypertension (ETODH) study, a cross-sectional observational registry providing detailed information on hypertension-related organ damage in untreated and treated subjects with uncomplicated essential hypertension. Details of the study have been previously reported ([19]). Briefly, high BP was defined as a systolic BP (SBP) ≥140 mmHg and/or diastolic BP (DBP) ≥90 mmHg in untreated subjects. Treated hypertensive subjects were included regardless BP values. Obesity was identified by the 1998 National Institutes of Health classification as BMI ≥30 kg/m2.

Entry criteria included: (i) good echocardiographic window; (ii) absence of previous clinically overt cardiovascular disease, secondary causes of hypertension and life-threatening conditions. After their informed consent had been obtained during the initial visit, all patients underwent the following procedures within 1–4 weeks: medical history and physical examination, clinic BP measurement, blood and urine sampling, standard 12-lead electrocardiogram, 24-h urine collection for microalbuminuria (MA), non-mydriatic retinography, cardiac, renal and carotid ultrasonography. The study protocol was approved by the ethics committee of one of the institutions involved.

The ETODH registry started in January 1999 and by the end of July 2008 had enrolled 4554 subjects with untreated (27%) and treated essential hypertension. For the present analysis, 4468 hypertensive subjects with a complete echocardiographic report have been selected and categorized in four groups according to BMI: lean (<20 kg/m2), normal (20–24.9 kg/m2), overweight (25–29.9 kg/m2) and obese (≥30 kg/m2).

Clinic BP was measured in the outpatient clinic at two different visits by attending physicians using a mercury sphygmomanometer and taking the first and fifth phases of Korotkoff sounds to identify SBP and DBP, respectively Measurements started after the subjects had rested for 5 min in the sitting position. Three measurements were taken at 1-min interval, and the average was used to define clinic SBP and DBP.

Echocardiography was performed according to standardized procedures, as reported elsewhere ([19]). Briefly, M-mode, two-dimensional, Doppler echocardiographic examinations were made with commercially available instruments. All echocardiographic procedures were performed by experienced cardiologists unaware of the clinical characteristics of the subjects. LVM was estimated from end-diastolic left ventricular (LV) internal diameter, interventricular septum and posterior wall thickness according to Devereux's formula ([20]) and normalized to BSA and height2.7 to obtain LVM indexes (LVMI). Relative wall thickness (RWT) was defined by the ratio of posterior wall plus septal wall thickness divided by LV radius. LV filling was assessed by standard pulsed Doppler technique of the mitral flow; the following parameters were considered: early diastolic peak flow velocity (E), late diastolic peak flow velocity (A) and their ratio (E/A). Details about reproducibility of LVMI measurements in our laboratory have been previously reported ([21]).

LVH was defined according to two gender-specific criteria: (A) LVMI ≥116 g/m2 in men and ≥96 g/m2 in women; (B) ≥49 g/m2.7 in men and ≥45 g/m2.7 in women ([18]). Patterns of LV geometry were defined as follows: LV concentric remodelling (normal LVMI and RWT ≥0.43) eccentric LVH (increased LVMI and RWT <0.43) and concentric LVH (increased LVMI and RWT ≥0.43) ([13],[18]).

Moreover, LVH prevalence was calculated according to the gender-specific criteria recommended by the ESH/ESC guidelines (i.e. LVMI ≥125 g/m2 in men and ≥110 g/m2 in women) ([13]).

Statistical analysis was performed using the SAS system (version 6.12; SAS Institute Inc., Cary, North Carolina, USA). Values were expressed as means±SD or as percentages. Continuous variables were compared by analysis of variance (ANOVA), using the Student's t-test for dual comparison. Analysis of categorical data was carried out with the χ2 test or Fischer's exact test when appropriate. The strength of correlation between variables was tested by linear correlation analysis and multiple regression analysis. The limit of statistical significance was set at p<0.05.

Demographic and clinical data of the whole study population are shown in Table I. Overall, 2006 patients (44.9%) were found to have LVH when LVM was indexed to BSA and 2269 (48.2%) when indexed to height2.7 according to the gender-specific thresholds indicated under Methods ([18]). A total of 1655 (37.0%) were positive to both criteria. Prevalence of LVH was lower (26.6%) when this phenotype was diagnosed according to the gender-specific criteria recommended by ESH/ESC guidelines ([13]).

Table I. Demographic and clinical characteristics of the study population (n=4468).

5 Data are shown as means±SD or per cent; LVH, left ventricular hypertrophy; LVMI, left ventricular mass index.

As for LV geometric patterns, eccentric LVH was more prevalent than concentric LVH regardless of the indexation criteria, namely 25.2% vs 19.7%, by BSA index (p<0.001) and 27.4% vs 20.8% by height2.7 index (p<0.001), respectively.

As for body weight, 4.5% of the subjects were lean (BMI<20 kg/m2), 36.5% normal weight (BMI 20–24.9 kg/m2), 42.0% overweight (BMI 25–29.9 kg/m2) and 17.0% obese (BMI ≥30 kg/m2).

Clinical characteristics of these groups are reported in Table II. Mean age was lower in lean hypertensives than in other groups. Prevalence of men, average clinic SBP/DBP, duration of hypertension, fasting blood glucose, triglycerides, uric acid concentrations, urinary albumin excretion tended to be higher in obese and overweight than in lean and normal weight hypertensives; the opposite trend was observed for high-density lipoprotein (HDL) cholesterol and current smoking. The differences were in most instances statistically significant.

Table II. Clinical characteristics of the study population categorized in four groups by body mass index (BMI).

6 Data are shown as means±SD or per cent; BMI, body mass index; BSA, body surface area; SBP, systolic blood pressure; DBP, diastolic blood pressure; HTN, hypertension; UAE, urinary albumin excretion. aLVMI≥116/96 g/m2; bLVMI≥49/45 g/m2.7; IMT, intima-media thickness.

Echocardiographic data, shown in Table III, indicated that LV internal diameter, interventricular septum and posterior wall thickness, RWT, absolute and indexed LV mass, left atrial and aortic root diameter progressively increased across the four groups, whereas E/A ratio showed the opposite trend. As for LV geometric patterns, eccentric LVH was more prevalent than the concentric one in all groups, regardless of the criteria used (Figure 1).

Graph: Figure 1. Prevalence rates of eccentric and concentric left ventricular hypertrophy (LVH), defined by two different gender-specific criteria (A=left ventricular mass index ≥116/96 g/m2; B=left ventricular mass index ≥49/45 g/m2.7) in hypertensive patients categorized according to four body mass index categories: lean, normal, overweight and obese. *p at least <0.05 eccentric LVH vs concentric LVH.

Table III. Echocardiographic parameters of the study population categorized in four groups by body mass index (BMI).

7 Data are shown as means±SD or per cent; LVIDd, left ventricular internal diameter diastole; LVIDs, left ventricular internal diameter systole; IVSTd, interventricular septum thickness diastole; PWTd, posterior wall thickness diastole; RWT, relative wall thickness; LA, left atrium; AR, aortic root; E, early diastolic mitral flow; A, late diastolic mitral flow; LVM, left ventricular mass; LVMI, left ventricular mass index; LVH, left ventricular hypertrophy; aLVMI≥116/96 g/m2; bLVMI ≥49/45 g/m2.7.

As for the discordance between the indexation criteria in defining LVH, about 19% of the total study population had LVH by only one criterion (11.2% by LVM/height2.7 and 7.6% by LVM/ BSA, respectively). Overall, the prevalence of the discordant groups was lowest in normal weight (15.2%), intermediate in overweight (17.5%) and lean (25.8%), and highest in obese hypertensives (26.9%).

The risk attributable to LVH was significantly greater in lean and normal weight when LV mass was indexed to BSA than to height2.7 (39.8% vs 35.0%, p<0.001); the opposed trend occurred in overweight and obese hypertensives (48.8% vs 58.1%, p<0.0001).

Table IV shows the clinical characteristics of the patients classified as having LVH by the gender-specific criterion of LV mass/indexed to BSA or to height2.7 and by both criteria. Absolute LVM, BMI, prevalence of obesity and clinic SBP were highest in the group with LVH according to both sets of criteria, lowest in the group with LVH by height2.7 indexation criterion, and intermediate in that with LVH by BSA criterion. Interestingly, extra-cardiac target organ damage, as assessed by urinary albumin excretion and intima-media thickness of the common carotid artery, was greatest in the group with LVH defined by two concordant criteria, intermediate in that with LVH by height2.7 criterion only and lowest in the group with LVH by BSA criterion.

Table IV. Demographic and clinical findings of essential hypertensives with left ventricular hypertrophy categorized according to the following criteria: (A) LVMI ≥ 116/96 g/m2, (B) LVMI ≥ 49/45 g/m2.7.

8 Data are shown as means ± SD or per cent; for abbreviations, see Tables II and III.

In the whole population, LVM showed, in ranking order, significant and positive correlations (p < 0.0001 for all independent variables) with BSA (r = 0.60), BMI (r = 0.35), serum creatinine (r = 0.25), log-transformed MA (r = 0.24), uric acid (r = 0.23), fasting blood glucose (r = 0.22), clinic SBP (r = 0.20), DBP (r = 0.20), carotid intima-media thickness (r= 0.18), triglycerides (r = 0.16) and age (r=0.09); an inverse correlation was found with HDL cholesterol (r = 0.27).

When these variables were tested in multiple regression analyses, in the first model, BSA (β = 0.564, p < 0.0001), SBP (β = 0.190, p < 0.0001) and log-transformed MA (β = 0.158, p < 0.001) turned out to be the best correlates of LV mass; this was the case for BMI (β = 0.285, p<0.0001), SBP (β = 0.164, p < 0.0001) and HDL cholesterol (β = −0.147, p < 0.0001) in the second model.

This study addresses the impact of two gender-specific criteria for indexing LVM, namely BSA and height to allometric power of 2.7, on LVH prevalence in a large population of uncomplicated hypertensive subjects attending our outpatient hospital clinic. Major new findings and potential clinical implications are the following: (i) the prevalence of LVH was much higher in overweight and obese hypertensives when LVM was indexed to height2.7, compared with BSA; the opposite trend occurred in normal weight and lean hypertensives; (ii) LVM normalized to height2.7 is more appropriate for detecting LVH in overweight and obese subjects, whereas LVM normalized to BSA is indicated in lean and normal weight hypertensive individuals.

Several aspects of our study deserve to be commented. First, the risk attributable to echocardiographic LVH in a population depends on a variety of factors including clinical characteristics, partition values and indexation methods used to define this phenotype.

Extensive evidence indicates that body size accounts for up to 50% of adult LV dimensions and that BMI increase is associated to a greater likelihood of LVH ([22],[23]). According to these observations, in our series LVH increased from lean to obese subjects by approximately 1.6-fold when defined by LVM/BSA, by 3.2-fold when defined by LVM/height2.7 and by 3.3-fold by both sets of criteria.

In the Losartan Intervention for Endpoint Reduction in Hypertension (LIFE) study, prevalence rate of LVH ranged from 42% to 78% and that of normal LV geometry from 15% to 40% depending on the criteria used ([24]). In our study population, the gender-specific LVH/BSA criteria recommended by the ESH/ESC guidelines reduced LVH prevalence to 27%.

It is worth noting that not only diagnostic thresholds, but also the methods for scaling LVM to body size represent an important source of variability in identifying LVH. Indexation to height2.7, indeed, has been shown to provide the highest LVH prevalence rates, since both obesity- and BP-related LVH are identified by this index. Indeed, findings from population-based studies ([17],[25]) and hypertensive cohorts ([19]) have shown that LVM normalized to height to allometric signals identifies a higher portion of patients with LVH, compared with BSA. In a previous analysis of 2213 treated hypertensives, we found that LVH defined by LVM/height2.7 (i.e. >51/47 g/m2.7 in males and females, respectively) was more prevalent (46% vs 31%) than that defined by LVM/BSA (i.e. >125/110 g/m2); the corresponding figures in obese were 71% and 40%, respectively ([26]). It should be pointed out that no substantial differences in echocardiographic LVH between the two indexation methods have been reported among patients enrolled in the LIFE study ([24]). This apparent inconsistency may be ascribed to the fact that LIFE cohort was likely to have a more pronounced cardiac damage having been selected on the basis of electrocardiographic LVH.

The present study offers a new piece of information by showing that the difference between the two indexation methods in the risk attributable to LVH was about 3% in the entire population. Our data also indicate that: (i) LVH was identified by only one criterion in approximately 19% of the patients; (ii) the risk attributable to LVH was significantly greater in lean (34.5% vs 19.7%, p<0.001) and normal weight (40.4% vs 36.7%, p<0.01) when LVM was indexed to BSA compared with that indexed to height2.7. The opposite trend occurred in overweight (47.1% vs 53.4%, p<0.001) and in obese hypertensives (52.2% vs 69.6%, p<0.0001).

Our results add further information on clinical correlates of patients with LVH defined by a single criterion. LVH phenotype identified by LVM/ height2.7 was associated with a higher prevalence of obese, male gender, higher levels of fasting blood glucose, total cholesterol, triglycerides, uric acid, MA and intima-media thickness but lower clinic BP than those identified by LVM/BSA.

Finally, one interesting result of the present study is that eccentric hypertrophy resulted the most frequent abnormal LV pattern in the total population as well as in various BMI categories regardless the type of indexation.

In conclusion, our findings suggest that the risk attributable to LVH is lower when LVM is indexed to BSA in overweight and obese hypertensives; the same occurs when LVM is normalized for height to allometric signals in lean and normal weight subjects.

Indexation of LVM to lean body mass assessed by bioelectric impedance is now regarded as the best option to differentiate physiological cardiac adaptation from pathological alterations related to obesity and hypertension; lean body mass, however, is not routinely estimated in clinical practice ([27]). Thus, in order to minimize the shortcomings related to the systematic use of a single indexation method in patients attending ultrasound laboratories, we recommend that LVM should be normalized by height2.7 in overweight/obese and by BSA in lean/normal weight hypertensives.

Conflict of interest: None.